ASTROBIOLOGYVolume 3 Number 4 2003copy Mary Ann Liebert Inc

Europa

The Production of Oxidants in Europarsquos Surface

RE JOHNSON1 TI QUICKENDEN2 PD COOPER2 AJ MCKINLEY2

and CG FREEMAN3

ABSTRACT

The oxidants produced by radiolysis and photolysis in the icy surface of Europa may be nec-essary to sustain carbon-based biochemistry in Europarsquos putative subsurface ocean Becausethe subduction of oxidants to the ocean presents considerable thermodynamic challenges weexamine the formation of oxygen and related species in Europarsquos surface ice with the goal ofcharacterizing the chemical state of the irradiated material Relevant spectral observations ofEuropa and the laboratory data on the production of oxygen and related species are first sum-marized Since the laboratory data are incomplete we examine the rate equations for forma-tion of oxygen and its chemical precursors by radiolysis and photolysis Measurements andsimple rate equations are suggested that can be used to characterize the production of oxi-dants in Europarsquos surface material and the chemical environment produced by radiolysis Pos-sible precursor molecules and the role of radical trapping are examined The possibility ofoxygen reactions on grain surfaces in Europarsquos regolith is discussed and the earlier estimatesof the supply of O2 to the atmosphere are increased Key Words EuropamdashOxygenmdashIcemdashChemical kinetics Astrobiology 3 823ndash850

823

INTRODUCTION

THE PUTATIVE SUBSURFACE OCEAN on Europa isof interest to astrobiologists as an environ-

ment that might be able to sustain life processesGiven the presence of oxygen carbon sulfurand possibly nitrogen organisms dependent onconventional carbon-based biochemistry may befeasible However the presence of aerobic lifeprocesses may be contingent on the transport ofoxidants or oxygenated species formed at the sur-face to an ocean that is covered by a substantialthickness of ice (Chyba 2000 Kargel et al 2000

Chyba and Hand 2001 Cooper et al 2001) Al-though oxygen has been detected in Europarsquosvery thin atmosphere and oxidants have been ob-served as trapped species in its icy surface theirtransport to a subsurface ocean presents consid-erable thermodynamic and kinetic challenges

It is generally agreed that the radiolysis of iceby energetic ions and electrons trapped in the Jov-ian magnetosphere produces the oxygen and per-oxide seen in Europarsquos icy surface (eg Johnsonet al 2003) This occurs because the bombardingparticles chemically alter the ice and decomposeit producing molecular hydrogen and oxygen

1Engineering Physics University of Virginia Charlottesville Virginia2School of Biomedical and Chemical Sciences University of Western Australia Crawley Australia3Department of Chemistry University of Canterbury Christchurch New Zealand

The hydrogen readily escapes from the surfacematerial and in turn from Europarsquos atmospherebecause of the low gravitational field The optionsfor oxygen are the release into the tenuous at-mosphere where it is retained because of its highmolecular mass and retention in the ice lattice ashomogeneously dissolved oxygen as trappedbubbles of oxygen or as an oxidized trace speciesBecause of the preferential loss of hydrogen oxygen-rich species are formed preferentially inEuroparsquos surface ice (Johnson and Quickenden1997 Madey et al 2002) Therefore oxygen per-oxide and oxidized sulfur and carbon have beendetected as trapped species (eg Johnson et al2003)

The production and trapping of such species inthe surface layers are of course not sufficient forour purposes as indicated schematically in Fig1 Radiolysis can occur to depths of the order oftens of centimeters because of the penetration ofthe energetic component of the incident radiation(Cooper et al 2001 Paranicas et al 2001) Mixingof these radiolytic products to greater depths oc-curs because of meteoroid bombardment (Cooperet al 2001) This bombardment also produces aporous regolith (Buratti 1995) composed of sin-tered grains (Grundy et al 2001) which increasesthe effective radiation penetration depth In ad-dition the atmospheric O2 permeates pore spacein the regolith Therefore oxygen gas in theporous regolith can be trapped by the collapse ofpores on meteoroid impact or can react on grain

surfaces well below the typical radiation pene-tration depth Finally macroscopic mass trans-port of trapped species by crustal subduction(Prockter and Pappalardo 2000) must occur Be-cause of the temperature gradient relevant speciesmust be stable in the ice at temperatures wellabove Europarsquos surface temperature (80ndash110K)and at higher pressures The subduction processeffectively is a macroscopic mass transport pumpwhich is needed to carry oxidants to Europarsquosocean

As a first step we focus in this paper on thechemical processes leading to the formation ofoxygen and related oxidants in Europarsquos regolithThe goal is to characterize the chemical state andstability of the irradiated surface material Wefirst summarize the spacecraft and telescopic ob-servations of oxygen and related species at Eu-ropa Next we summarize the laboratory data onthe radiolysis and photolysis of ice Because dataare not available for all incident species and inci-dent energies we concentrate on the descriptionsof the chemistry of formation of molecular oxy-gen The focus on oxygen is not because this isthe species delivered to a subsurface ocean butbecause understanding the chemistry of its for-mation in ice is key to describing the chemicalstate and stability of Europarsquos irradiated surfacematerial We then use the data to correct previ-ous estimates of the production rate of oxygen atEuropa and suggest experiments in order to char-acterize the chemical state of the irradiated sur-face that leads to the production of O2

OBSERVATIONS

Molecular oxygen

Tenuous oxygen atmospheres have been re-ported at both Ganymede (Hall et al 1998) andEuropa (Hall et al 1995) based on Hubble SpaceTelescope (HST) observations The detection ofmolecular oxygen was indirect That is two lines(1356 and 1304 nm) due to excited O were ob-served in emission and their ratio (19) sug-gested that the excited atoms were produced byelectron impact dissociation of molecular oxygenIn addition two well-known transitions for in-teracting pairs of O2 molecules 1Dg 1 1Dg r3^g

2 1 3^g2 were observed in the reflectance

spectra of Ganymede (Calvin et al 1995 Spenceret al 1995) and Europa (Spencer and Klesman

JOHNSON ET AL824

FIG 1 Diagram of processes occurring on Europarsquossurface as discussed in the text

2001 Spencer and Calvin 2002) using ground-based telescopes and the HST These absorptionbands cannot be due to atmospheric O2 becausevery high pressures or very long line-of-sightcolumns of O2 would be required

The positions of the two signature absorptionsfor dimer O2 at 5772 nm and 6275 nm respec-tively (Landau et al 1962 Cooper et al 2003a)suggest that the O2 is locally quite dense (Calvinet al 1996 2003) Therefore these bands weresuggested to be due to inclusions of O2 trappedwithin Ganymedersquos icy surface (Calvin et al 1996Johnson and Jesser 1997) It is seen from Table 1that although the atmospheric and surface reser-voirs of O2 are coupled the oxygen content in thesurface ice is the largest

The observed O2-dimer band shapes and rela-tive band depths are distorted from solid or liq-uid O2 (Johnson 1999) At Ganymede where spa-tial resolution was obtained using the HST boththe position of the band minimum and its shapevary with latitude (Calvin and Spencer 1997)Therefore care must be taken when comparingthe peak positions of the bands seen in the dis-covery spectrum at Ganymede with those of solidor liquid oxygen This spectrum is a hemispher-ical average of spectral features that vary in theirpeak positions and band shapes It was obtainedby ratioing Ganymedersquos reflectance to that of Cal-listo (Spencer et al 1995) which we now knowexhibits a weak solid-like O2 feature It was re-cently shown that O2 is also trapped in Europarsquosicy surface (Spencer and Klesman 2001 Spencerand Calvin 2002) However the solid-like O2 fea-ture at Europa is much weaker than that atGanymede (5772 nm band depth 03 vs18) This could be due to the fact that Eu-roparsquos surface scatters light more efficiently in thevisible leading to shorter path lengths (Johnsonand Jesser 1997) It could also be due to compet-

ing chemical processes (Calvin et al 1996 John-son and Quickenden 1997)

Ozone and peroxide

Related to the oxygen observations the Hartleyband of O3 has been tentatively identified in the ul-traviolet (UV) reflectance spectra of Ganymede andin the icy satellites of Saturn (Noll et al 1996 1997)O3 trapped within the icy surface would be con-sistent with the presence of trapped oxygen ex-posed to radiolysis (Johnson and Quickenden1997) The presence of trapped O3 supports the in-terpretation that the dimer-O2 observations are dueto inclusions of oxygen trapped in the ice How-ever the shape of the O3-like feature also varies sig-nificantly with latitude at Ganymede (Hendrix etal 1999a) and is highly distorted Based on theband shape the presence of an additional absorb-ing species was inferred This was initially sug-gested to be trapped OH (028 mm) (Noll et al1996) but may more likely be due to organic mol-ecules (Johnson 2001) This O3-like feature is su-perimposed on a very broad UV absorption ex-tending from about 04 mm to shorter wavelengthsseen on all of the icy satellites The broad UV ab-sorption has been attributed to another radiolyti-cally produced oxidant H2O2 (Carlson et al 1999aHendrix et al 1999b) but may also include ab-sorption by trapped HO2 and OH (Taub and Eiben1968 Johnson and Quickenden 1997) Identifica-tion of specific radiolytic products in the UV is dif-ficult because the bands are very broad and over-lap The presence of H2O2 trapped in Europarsquos icewas confirmed by a band in the infrared observedusing the Near Infrared Mapping Spectrometer(NIMS) on the Galileo spacecraft (Carlson et al1999a) Remarkably the amount seen in the in-frared is roughly consistent with the amount sug-gested by the decrease in reflected light in the UV

OXIDANTS AT EUROPA 825

TABLE 1 GLOBAL OXIDANT INVENTORY

Reservoir Oxidant Column density (1016 molcm2)

Atmosphere1 O2 010Regolith (gas)2 O2 1025

Trapped in ice (1 mm at unit density)34 O2 50ndash500H2O2 400

1Based on Hall et al (1995 1998)2Estimated using average regolith depth of 1 m3Europa band depth approximately one-sixth that in Calvin et al (1996) who give 01ndash1 concentration on

Ganymede4For comparison sulfur species on the trailing hemisphere [SO2] 300 [sulfate] or 1000 3 1016 molcm2

Whereas H2O2 is observed in ice at Europa at013 level the O3-like UV feature seen atGanymede is not obvious at Europa (eg Fanaleet al 1999) This again may be due to the fact thatless trapped O2 is seen in reflectance at Europathan at Ganymede That is since O2 is a precur-sor of O3 the factor of 6 reduction in O2 meansthat the identification of this broad feature su-perimposed on other broad UV bands is prob-lematic However the absence of the O3-like fea-ture could also be due to the presence of sulfurspecies that strongly compete for excess oxygen(Johnson and Quickenden 1997)

Summary space observations

Although each of the individual identificationsof suggested products of water radiolysis or pho-tolysis (OH H2O2 O2 O3) has some uncertaintythe reflectance spectra taken as a whole give afairly clear self-consistent picture The irradiatedsurfaces of the icy satellites contain trapped mol-ecules that would be expected from radiolysis orphotolysis of an icy surface (Johnson and Quick-enden 1997) Indeed OH is a suggested precur-sor of H2O2 which is a suggested precursor ofO2 which in turn is a precursor of O3 Howeverthe separation into amounts of individual speciesbased on the absorption in the UV identificationsremains difficult Therefore the observed band oftrapped H2O2 in the infrared and the pair ofbands in the visible of a condensed form of O2are critical as they confirm the presence of radi-olytically produced oxidizing molecules in Eu-roparsquos icy surface

Consistent with the above oxygen-rich mole-cules derived from the radiation processing of anice containing sulfur (SO2 and a sulfate) and car-bon (CO2 and a carbonate) have been identified(see eg Johnson et al 2003) The sulfate hasbeen suggested to be a hydrated salt mineral (Mc-Cord et al 1999) or a hydrated sulfuric acid (Carl-son et al 1999b 2002) The latter is an efficientoxidant that is also refractory Together these ob-servations indicate that volatile and refractory ox-idizing molecules are readily formed by radioly-sis and photolysis in Europarsquos surface material Ifindeed that material can be delivered to Europarsquosputative subsurface ocean a source of free energymay be available for potential biological activityHere we consider the formation of molecular ox-idants by radiolysis with emphasis on O2

RADIOLYTIC PRODUCTIONLABORATORY DATA

Introduction

In this section we review the laboratory dataleading to the production of oxygen and relatedspecies by radiolysis and photolysis of ice Wefirst compare the data for production of oxygenby incident ions electrons and UV photons Thisis followed by a review of the data on the loss ofhydrogen which is the other principal product ofradiation-induced decomposition of ice We thenconsider the evidence that certain proposedspecies such as peroxide or trapped O might beprecursors to molecular oxygen formation in iceSome of these data have been summarized ear-lier but with a very different emphasis Here thegoal is to motivate the development of chemicalmodels for describing irradiated ice models thatcan be used to suggest new experiments and tointerpret data Individuals familiar with the datacan go directly to the modeling section in whichrate equations are developed and used to inter-pret aspects of the diverse database

Molecular oxygen

Gas-phase O2 has been produced by irradiat-ing low temperature ice using UV photons low-energy (eV) electrons keV ions and electrons andMeV ions (see eg reviews by Johnson andQuickenden 1997 Delitsky and Lane 1998 John-son 1990 1998 2001 Strazzulla 1998 Baragiola2003 Johnson et al 2003) In spite of the consid-erable differences in excitation density and pen-etration depth for these incident particles the re-sults exhibit important similarities Here wesummarize the principal findings We first notethat the efficiency of this process in pure ice islikely to be lower than estimates based on veryearly measurements of the production of O2 in ice(see eg Hart and Platzman 1961) because oflow levels of contaminants in those early experi-ments (Baragiola and Bahr 1998 Moore andHudson 2000) Therefore trace species can be im-portant in determining the O2 production ratesfrom Europarsquos icy surface and will be discussedshortly

In studying the ability of energetic plasma par-ticles to eject (sputter) molecules from ice aprocess occurring on small Outer Solar Systembodies Brown Lanzerotti and co-workers (Brown

JOHNSON ET AL826

et al 1982 1984) found that at temperatures rel-evant to the surface of Europa H2 and O2 wereimportant components of the ejecta They thenused those data to predict the presence of a radi-olytically generated O2 atmosphere on Europa(Johnson et al 1982) The production efficiencyfor gas-phase O2 was found to be strongly tem-perature dependent as seen in Fig 2a for MeVHe1 ions incident on ice and Fig 2b for keV Ar1

ions on ice In Fig 2 the production rate is givenas a yield which is the number of moleculesejected per incident ion (ie O2 production ratedivided by the flux of incident particles photonselectrons or ions) Yields are typically measuredat a low incident particle flux so that sample heat-ing does not occur The yield versus temperatureshown in Fig 2 appears to increase exponen-tially with T over a range of temperatures Al-though this is not an Arrhenius dependencerough activation energies can be extracted overnarrow ranges of T (eg Fig 2a inset) Theseare of the order of 001ndash007 eV (Table 2) Suchvalues are much smaller than those for diffusionof vacancies or trapped species in ice (Livingstonet al 2002) but are comparable to the reportedactivation energies for H1 or H diffusion in iceas noted earlier (Johnson and Quickenden1997) The lowest energies are also comparableto the enthalpy difference between amorphoussolid water and crystalline ice (Ayotte et al2001) Therefore the estimated energies are likely

related to H1 and H transport or to hydrogenbond re-orientation It has also been shown thatas the temperature is increased above 80ndash100Kmorphological changes occur that can affect thereaction kinetics and give incorrect activation en-ergies This is especially so when using measure-ments of average quantities such as yields ratherthan directly extracting rate constants by mea-suring transients (Selby 2003) In addition the ac-tivation energies can be representative of the com-peting (destruction) processes rather than theformation processes (eg Matich et al 1993) asdiscussed in the model section

The measured O2 production rates (yields)were also found to be fluence dependent wherefluence is the radiation flux 3 the time or the to-tal number of particles per unit area onto the iceMultiplying the fluence by the incident particleenergy and dividing by the average penetrationdepth gives the energy per unit volume oftencalled an average dose As seen in Fig 3 the O2yield versus fluence rises from zero implyingthat a single ion incident on a fresh surface doesnot efficiently produce and eject oxygen This isthe case even for an incident 15 MeV Ne1 forwhich the excitation density along the ion path ishigh However it is seen in Fig 3 that with in-creasing irradiation time individual impactingions produce and eject O2 with increasing effi-ciency as the material is chemically altered Atlarge fluences the yield appears to saturate indi-

OXIDANTS AT EUROPA 827

FIG 2 a Yield for D2 and O2 at saturation (steady state) from 15 MeV He1 on D2O ice versus the temperature ofthe ice (adapted from Brown et al 1982) Inset D2 yield minus the constant yield at low T versus 1kT giving an av-erage activation energy of 003 eV b Yield of O2 at saturation versus temperature for incident 30 keV Ar1 ions(adapted from Baragiola et al 2003)

a b

cating that destruction processes are in a roughsteady state with formation processes The yieldsshown in Fig 2 are purportedly those after satu-ration although full saturation is often ap-proached slowly

The slope of the O2 yield versus fluence shownin Fig 3 increases with increasing temperaturegiving the activation energies in Table 2 Thesevalues are similar to those obtained from Fig 2For energetic ions that fully penetrate the film asin Fig 2a and 3a the yield versus fluence can ex-hibit a second rise at higher fluences for whichthe radiation track cores begin to overlap This in-crease also depends on temperature and exhibitssimilar activation energies (Reimann et al 1984)The strong dependence on temperature is at firstsurprising since the energy density deposited bya 15 MeV Ne1 produces transient temperaturesin the track core that are much higher than theambient material temperature Therefore tracktemperature (local excitation density) alone doesnot determine the O2 yield

For all incident particle types and energies themeasured fluence dependence is found to be sta-ble over laboratory time scales That is for ex-periments carried out at high vacuum the yieldreturns to the value it had when the beam wasturned off This indicates that the irradiated ice isldquopermanentlyrdquo altered (Reimann et al 1984)Thus a relatively stable chemical state is createdthat is favorable for O2 production This could bethe formation of a particular trapped radiationproduct in ice that acts as a stable precursor or isdue to the altered chemical and physical state ofthe irradiated ice Trapped O H2O2 and HO2(Matich et al 1993 Johnson and Quickenden1997 Sieger et al 1998) have been proposed asstable precursor species

Using isotopically labeled H2O Benit andBrown (1990) showed that considerable transportof O occurs along the ion track when an energeticheavy ion penetrates an ice sample Because ofthis the O2 yield was found to depend on sam-ple thickness Therefore for thin ice samples the

JOHNSON ET AL828

TABLE 2 OXYGEN AND PEROXIDE PRODUCTION

Radiationa Product T (K) Ea (eV) Percent at saturation G (product100 eV)b

He1(1) O2 015g(23) O2 77 (300) 0003 (0)Ne1 (30 MeV)(3) O2 300 0031He21 (6 MeV)(3) O2 300 0008He21 (29 MeV)(3) O2 300 00016H1 (80 keV)(4) O2 70 00002H1 (200 keV)(5) O2 120 00007He21 (15 MeV)(6) O2 50 (100) 003 (07ndash4) 3 1024

Ne1 (15 MeV)(7) O2 7 10ndash15c 003(8) 70ndash140 005ndash007 01

Ar1 (01 MeV)(9) O2 30ndash140 001ndash004 0006ndash008160 001

e (30ndash100 eV)(10) O2 120 002ndash003 0005ndash0007hn (102 eV)(11) O2 50 (100) 003 00001 (00003)H1 (1 MeV)(12) H2O2 16 (77) 01 (001)

1 10 CO2 77 011 10 O2 77 04

H1 (30 keV)(13) H2O2 16 (77) 022 (038) 0098 (0086)C1 (30 keV) H2O2 16 (77) 17 (22) 023 (034)N1 (30 keV) H2O2 16 (77) 16 (14) 021 (011)O1 (30 keV) H2O2 16 (77) 14 (40) 026 (015)Ar1 (30 keV) H2O2 16 (77) 22 (20) 011 (006)

aReferences are as follows 1Lefort (1995) 2Ghormley and Stewart (1956) 3Baverstock and Burns (1976) 4Baragiolaet al (1999) 5Baragiola and Bahr (1998) 6Brown et al (1982) (ion does not stop in sample) 7Benit and Brown (1990)(ion does not stop in sample) 8Reimann et al (1984) (ion does not stop in sample) 9Baragiola (2003) 10Sieger et al(1998) (steady-state value) 11Westley et al (1995) (steady-state value as no fluence dependence seen) 12Moore andHudson (2000) 13Gomis et al (2003)

bG values are for escaping O2 For H2O2 and O2 formation both of which require multiple events yields after sat-uration are used In a few cases yields are obtained by melting the sample Values in parentheses are those at anothertemperature

cPercent of film eroded as O2 molecules for long-term irradiation

ldquosaturatedrdquo yields change slowly with fluenceMore importantly O from water molecules at aconsiderable depth into the sample contributedto the O2 ejected from the surface Since there wasno observed diffusion of the O2 in ice at the am-bient temperature (7K) over the time for their experiment they concluded that the oxygentransport and production are stimulated by theincident ions The transport was suggested as be-ing due to percolationdiffusion along the dam-aged and transiently heated tracks (Reimann etal 1984 Benit and Brown 1990) They also foundthat in fully eroding (sputtering) thin ice filmsthe total amount of O2 produced required about10ndash15 of the D2O molecules in the films at T 57K On warming irradiated samples some O2

formed at depth is found to remain trapped forexperimentally relevant times (see eg Hart andPlatzman 1961 Benit and Brown 1990 Baragiolaand Bahr 1998)

Matich et al (1993) detected luminescence in77K H2O and D2O ice from newly formed and ex-cited O2 They suggested that the O2 is formed

from the interaction of a freshly produced mo-bile O interacting with a previously producedand trapped O or H2O2 They favored the formerIn matrix isolation studies of peroxide irradiatedby UV photons luminescence from newly formedexcited O2 was seen However the oxygen wasformed from mobile O (greater than 30K in Krlattice) interacting with O trapped as H2O-O(Khriachtchev et al 2000) These studies also sug-gest that trapped O possibly in the form H2O-Omay be an important precursor

In Fig 4a the O2 yield for low-energy electronsplotted as a function of electron energy exhibits anapparent ldquothresholdrdquo at 10 eV (Orlando andSieger 2003) Although the uncertainties are largethe authors suggested there may be a lower thresh-old 6 eV corresponding to a very small crosssection for O2 production (10220 cm2) This couldbe associated with an excitation at the ice surfaceBulk ice exhibits very weak absorption at photonenergies below 6 eV but exhibits a strong ab-sorption that depending on the ice structure be-gins at about 73 eV with a peak at 86 eV This

OXIDANTS AT EUROPA 829

b

a

FIG 3 a Relative yields versus fluence (ion flux 3 time) for 15 MeV Ne1 incident on D2O ice at 7K (adapted fromReimann et al 1984) Top panel D2O yield Middle panel D2 yield Bottom panel O2 yield Yields at other tem-peratures exhibited similar dependences with the slope of O2 yield versus fluence varying with T Fits give activa-tion energies between 002 eV at the lowest fluences and lowest T to 007 eV at the highest T and highest fluencesb Oxygen yield versus fluence of low-energy electrons on D2O ice (from Sieger et al 1998)

is followed by a peak in absorption at 10 eVThe results in Fig 4a show that there is a dra-matic increase in the production of O2 at 10 eVwhich is the onset of secondary electron produc-tion in the bulk (Orlando and Sieger 2003)

For incident fast ions the yield at a given tem-perature was found after a small fluence to beindependent of whether the sample was crys-talline or amorphous ice (Brown et al 1982)However using low-energy electrons the yieldswere found to be smaller for an initially crys-talline ice (Orlando and Sieger 2003) Since verylow doses of ion radiation (a few eV per mole-cule) can convert crystalline ice to amorphous iceat low temperatures (Leto and Baratta 2003) thedamage produced by the penetrating ions usedby Brown et al (1982) affected the O2 yield Thestructural damage produced may create trappingsites cause changes in the electronic relaxationpathways or enhance percolation

The temperature dependence of the O2 yield isshown in Fig 5 for electrons incident on initiallyamorphous (Fig 5a) and initially crystalline (Fig5b) ice A small decrease in efficiency is seen inboth cases at 120ndash130K where Ic forms fromamorphous ice and pore collapse occurs (Row-land et al 1991) This is followed by an increas-ing yield with increasing T and then by a more

dramatic drop near 160K where Ih forms Dur-ing the transition to Ih trapped molecules havebeen shown to escape efficiently from ice samples (Ayotte et al 2001) A large drop in the yield isalso seen in Fig 2b for incident Ar1 ions

In modeling the dramatic decreases in the yieldin Figs 2b and 5 care must be taken That is thetime for crystallization of an amorphous sample(Baragiola 2003) even allowing for the large un-certainties is 30 min at 145K and becomes 1min at 160K Therefore pulsed radiolysis exper-iments should be carried out at these tempera-tures and the transients observed Since the timefor conversion from amorphous to crystalline icecan be shorter than the experimental times at145K the results are not comparable to thosefor which the chemical state of the ice at a givenfluence is stable over experimental times Belowwe focus on the data below 140K

We have emphasized processes initiated by theelectronic excitations and ionizations producedHowever momentum transfer collisions whichare the dominant energy loss process for low-en-ergy ions also produce dissociations defects andO2 in ice (Bar-Nun et al 1985) The relative effi-ciency of momentum transfer and electronicprocesses for producing O2 needs further studyAs the penetration depths are much smaller when

JOHNSON ET AL830

FIG 4 a Relative yield of O2 from D2O ice versus incident electron energy E at T 5 120K (from Orlando et al1998 Orlando and Sieger 2003) b Relative yield of D2 from D2O ice versus incident energy E at T 5 88K (fromKimmel et al 1994)

a

b

the former dominates the produced O2 likely es-capes readily

H2 formation and loss correlation with O2

The production of O2 from a low temperatureice by radiolysis correlates with the formationand loss of H2 (or D2 from D2O ice) (Reimann etal 1984) D2O is used to reduce the backgroundin mass spectrometry studies The yields of O2from D2O and H2O ices were shown to be com-parable However the total yields from D2O wereabout 10 less (Reimann et al 1984) consistentwith H (D) transport being important It is seenin Fig 3 that D2 is promptly produced and lostfrom the D2O ice At higher fluence the D2 yieldis seen to exhibit a dependence like that for O2For the 15 MeV Ne1 ions which fully penetratethe ice sample the D2 yield eventually decreasessince D2 is produced and escapes from the fullthickness of the film

Because of H2 (D2) loss the surface region of

the irradiated ice becomes depleted in H (D) rel-ative to O favoring the formation of oxygen-richspecies The change seen in the ratio of H (D) toO in the irradiated ice is small An irradiated film(100 nm) remains stoichiometric to better than2 (Brown et al 1980) and in the near surface re-gion Auger spectroscopy suggests only a slightoxygen enhancement due to irradiation by keVelectrons (Rye et al 1978) Sieger et al (1998) sug-gested a precursor concentration of the order ofa percent Such concentrations are an order ofmagnitude smaller than the steady-state concen-tration of dissociated H2O (Table 2)

Using infrared spectroscopy the damaged H2Oconcentrations at low temperatures in the satu-ration regime are about 10 over the irradiatedvolume (Benit et al 1987 Watanabe et al 2000Gomis et al 2003) This fraction is the ratio of thedissociation cross section to the effective crosssection for recombination to H2O as discussedlater At a 10 concentration most H2O will havea neighbor that is dissociated

OXIDANTS AT EUROPA 831

FIG 5 a Relative steady-state yields of O2 from a D2O ice sample versus T (upper panel) The ice samples wereformed and irradiated at these temperatures Film erosion as a yield (lower panel) (from Orlando and Sieger 2003)b Relative steady-state yields of O2 versus T as in (a) (upper panel) Here the D2O ice samples are formed at 160Kpresumably becoming crystalline and then irradiated at the T shown A new sample is used for each data point in(a) and (b) Film erosion as a yield (lower panel) (from Orlando and Sieger 2003)

At temperatures above 40K it is seen in Fig2 that the D2 and O2 yields exhibit the same tem-perature dependence suggesting a correlationbetween D2 and O2 loss H2 (D2) can be formedby reactions between diffusing hydrogen fromdissociated water or directly [ie H2O R H2 1O (eg Kimmel and Orlando 1995) H2O R (H1 OH)cage R H2 1 O (eg Matich et al 1993)H3O1 1 e R H2 1 OH (eg Williams et al1996)] Subtraction of the nearly constant D2 yieldin Fig 2a gives the Arrhenius plot in the insetSince molecular hydrogen readily escapes atthese temperature it is the H2 (D2) formationprocess that has an activation energy of 003eVmolecule The fact that slopes of the yields forH2 and O2 above 40K are similar suggests thatH2 (D2) formation and loss is the rate-limitingstep in determining the production and loss of O2at these temperatures

Although the data at low T in Fig 2a have con-siderable uncertainties below about 20ndash40K theD2 production is roughly independent of tem-perature but the O2 yield continues to decreasewith decreasing temperature Radiolytically pro-duced D2 is mobile and can escape above15ndash20K (Watanabe et al 2000) but O2 is im-mobile at such temperatures and the incidentHe1 ions are much less efficient at causing per-colation than the incident Ne1 discussed earlier

The production of H2 (D2) from ice has alsobeen studied using low-energy electrons and UVphotons at very low temperatures In the formerstudies absolute yields were not obtained butthere appeared to be a threshold at 63 eV (Fig4b) This is roughly the threshold for excitationof a surface state (Khusnatdinov and Petrenko1992) and the threshold for peroxide productionis close to this value The efficiency for H2 (D2)production increases around 11 eV with a sec-ond steep rise at 17 eV (Kimmel et al 1994) TheO2 and D2 yields versus electron energy in Fig 4exhibit some similarity in the threshold regionalthough the uncertainties are large and the sam-ple temperatures different Measuring the H2 (D2)and the O2 ejecta simultaneously over a range oftemperatures as in the Ne1 experiments wouldbe instructive

Using 98 eV photons at low T (12K) Watanabeet al (2000) measured the D2 ejected and trappedin ice They obtained a cross section for produc-tion of D2 (Table 2) at low fluences that is10ndash20 of the total absorption cross sectionwith significant uncertainties They also found

that the equilibrium damage fraction in the sam-ple was 10ndash20 of the D2O molecules but theequilibrium production of D2 was about an orderof magnitude smaller 1ndash2 at their tempera-tures They suggested that the D2 is formed atgrain interfaces and cracks and on the surfaces ofpores In a mixed ice D2O 1 CO exposed to UVphotolysis D2 and CO2 formation are correlated(Watanabe and Kouchi 2002) The D2O dissoci-ates to produce D2 plus O and this transientlymobile O reacts with CO to produce CO2

Precursor state and peroxide

In order to produce O2 from low temperatureice it appears that the ice sample must lose hy-drogen and form trapped oxygen-rich reactivespecies The efficiency of production also appearsto be affected by the level of defects and damagein the ice consistent with the suggestion that thehydrogen-bonding network is important in theproduction of O2 via availability of trapping sitesnew electronic relaxation pathways or the natureof the bonding of precursors Here we briefly ex-amine the possible role of the suggested precur-sor molecules H2O2 and HO2 We note that H2O2might not readily dissolve in ice possibly ac-counting for its low abundance in some experi-ments It appears to be heterogeneously distrib-uted in the form of microcrystals (Gurman et al1967) If H2O2 does act as a precursor the for-mation of aggregates could affect the dimer ab-sorption bands (Cooper et al 2003a) and affectthe production and stability of O2 (Cooper et al2003b)

The results of Gerakines et al (1996) in whichice at 10K is irradiated by UV photons confirmthat H2O2 forms in ice from two OH after a build-up of trapped OH They monitored the band as-sociated with H2O-HO (Langford et al 2000) inwhich the HO is bound to water with 024 eVTherefore they might only be monitoring a sub-set of the trapped OH seen in earlier experiments(eg Taub and Eiben 1968) The H2O2 concen-tration in ice reaches a steady-state level at flu-ences of about 1018 photonscm2 (Gerakines et al1996) As seen in the model section this gives anestimate of the H2O2 photodestruction cross sec-tion of the order of 10218 cm2 Similarly HO2 be-gins to be seen in absorption at about a few times1017cm2 shortly after they begin to detect H2O2The absolute yields were not measured

Westley et al (1995) detected H2 and O2 pro-

JOHNSON ET AL832

duced by Lyman-a (98 eV) radiation of ice butfound no fluence dependence for fluences greaterthan 1017cm2 They also saw small amounts ofH2O2 by programmed thermal desorption of thesample but made no quantitative evaluationComparing their lowest fluence for the produc-tion of O2 with the result of Gerakines et al (1996)for peroxide suggests that H2O2 and HO2 areprobably not important precursors of O2 In UVphotolysis of H2O2 using 64 eV photons the pro-duction of O2 was not observed (Vaghjiani et al1992) and the excitation of an irradiated icewhich presumably contained some peroxide by37 eV photons did not produce O2 (Baragiola etal 2003) Direct excitation of matrix-isolatedH2O2 at low temperature also did not yield O2but the production of a mobile O that reacted withtrapped O (H2O-O) did give O2 (Khriachtchev etal 2000)

Irradiating with 200 keV protons at tempera-tures of 40ndash129K Bahr et al (2001) found onlyabout 0001ndash002 H2O2 using programmed ther-mal desorption as compared with earlier experi-ments that suggested larger values (eg Table 2)Using infrared absorption Moore and Hudson(2000) found a G value of 01 for incident MeVprotons at low temperature but the signal in theobserved band became negligible at 80K sug-gesting that H2O2 was not being formed Theyalso found that adding O2 significantly increasedthe production efficiency of H2O2 That is thatthe amount of peroxide was affected by theamount of trapped oxygen Based on the matrixisolation studies (Khriachtchev et al 2000) dis-sociated mobile O might trap as H2O-O whichon excitation converts to H2O2 Therefore even ifH2O2 is not the precursor the peroxide andtrapped oxygen concentrations will be related

Although molecular oxygen was not a productof photolysis of matrix isolated H2O2 it was aproduct of photolysis at 266 nm (466 eV) of per-oxide dimers in an argon matrix 2 H2O2 R 2 H2O1 O2 (Engdahl and Nelander 2002) This de-composition was thought to be due to the con-siderable distortion of the bonds in the dimerTherefore the configuration or distortion by thelattice of peroxide or any other precursor mole-cule in the ice may be critical (Sieger et al 1998)In addition more recent experiments show thatG values for the production of H2O2 by 30 keVincident ions are significant as are the steady-state ratios of H2O2H2O (Gomis et al 2004) Ra-tios of the order of 02ndash4 were found at 16K and

77K in the irradiated regions with incident H giv-ing the smallest fraction and incident O thelargest (Table 2) At these concentrations perox-ide molecules might segregate forming inclu-sions (Gurman et al 1967 Cooper et al 2003b)Since the amount of trapped peroxide also com-petes with the amount of trapped O studies areneeded to describe the chemical state of the solidproduced by radiolysis This will be critical in de-scribing the fate of the irradiation products at Eu-ropa In particular the chemical composition fa-vorable for producing O2 needs to be determinedas does the importance of trapped O H2O2 HO2or peroxide dimers as precursors

Other species

In the model below we will refer to atomic andmolecular species trapped in ice Such species areoften discussed in the literature on radiolysis andphotolysis (for summaries see Matich et al 1993Johnson and Quickenden 1997) Trapped OH OH2O2 and HO2 have been discussed andor mea-sured by a variety of techniques (absorption andluminescence spectroscopy electron spin reso-nance programmed thermal desorption etc)Care must be taken when using these data as re-sults for ice samples formed below 160K orthose exposed to penetrating ion beams likelyhave numerous defects voids and grain inter-faces so that a variety of trapping sites and trapdepths exist Therefore when a particular bandfor a trapped species (eg OH) disappears withincreasing temperature it could be due to achange in the nature of the trapping as well as toloss of the species by reaction

Although H is discussed below we do not dis-tinguish it in these simple models from H1

trapped as the hydronium ion H3O1 whichwould mean that there are corresponding nega-tive ions present Therefore below we refer to thevarious species as if they are neutral althoughthis might not be the case Although trapped elec-trons are detected only over a narrow range oftemperatures (100ndash120K) negative ions must beimportant over a broad range of temperature

We also note that SO2 and a sulfate and CO2and a carbonate appear to be trapped species inEuroparsquos icy surface Under radiolysis each ofthese can dissociate directly giving O2 Howeverwith the exception of the sulfate their densitiesare less than 1 so they are not relevant sourcesof O2 However their presence is indicative of the

OXIDANTS AT EUROPA 833

preference for forming oxygen-rich molecules inEuroparsquos ice and can affect the efficiency of pro-duction of O2 and H2O2

MODELS FOR FORMATION OF O2

Introduction

Based on the data described above the yieldsfor incident electrons and ions are found to benearly linear in fluence at the lowest fluences forwhich O2 is detected In addition the relevant al-terations of the ice samples by radiolysis appearto be stable and independent of the beam fluxRemarkably this is the case both for low-energyelectrons that make only a single excitation foreach impact and fast ions that make a density ofexcitations along their path through the solidThis suggests that O2 is produced and trapped inthe initial events and then caused to escape fromthe solid by the subsequent radiation or that astable chemical alteration of the ice is required be-fore the radiation can produce and eject O2

Because H produced by the dissociation of wa-ter molecules can attack trapped O2 and any pre-cursor oxidants the loss of H is important Thisoccurs by the production and loss of hydrogenmolecules as discussed leaving a surface that isslightly rich in O relative to H The O2 produc-tion mechanism could be direct excitation of aparticular precursor species or it might not bespecific That is a new OH or O produced by theincident radiation could react with the H-depleted solid producing O2 This would occurmore efficiently as the ratio of O to H in the nearsurface region increased

In their early work WL Brown and colleaguesassumed that processes similar to those used todescribe radiolysis of liquid and gaseous H2Ocould be used to describe the radiolysis in ice Thelow activation energy that they found was as-sumed to be associated with diffusion of somemobile radical species in ice such as H or O orpossibly diffusion of the O2 molecule itselfBased on their studies of ejection from isotopi-cally labeled D2O samples they concluded thatthe O2 is formed in the track of a single ion andthen ejected by subsequent ions (eg Benit andBrown 1990) More recently trapped specieshave been suggested as precursors These are ei-ther subsequently excited by the incident radia-tion or attacked by newly produced mobile

species We consider a simple model that incor-porates all of these processes and can be used toorganize and interpret the experimental data Wealso discuss its relationship to more completemodels In this discussion much of which is in theAppendix we consider the standard radiationproducts that have been often described How-ever in a solid the excited electron cloud pro-duced by absorption can overlap a number of lat-tice sites Therefore their relaxation pathwaysandor product species could be very differentfrom those that have been traditionally consideredin the literature on radiolysis and photolysis

Precursor model

The column density of a precursor species Npformed in ice by incident radiation can be calcu-lated using a simple model In this model cross sec-tions are used for the production sp and destruc-tion s of the precursor O2 is then formed byexcitation of the precursor or by reaction of a newlyformed radical with the precursor This is also de-scribed using a cross section sO2 For incident elec-trons having a small penetration depth the O2 pro-duced is assumed to rapidly diffuse to the surfaceand escape (Sieger et al 1998) More generally wedefine N as the column density of H2O from whichthe incident radiation can produce and eject O2 Apair of rate equations gives both Np and the pro-duction rate of oxygen dNO2dt That is

H2O 1 radiation R precursor

precursor 1 radiation R O2

giving

dNpdt 5 sp f N 2 s f Np (1)

dNO2dt 5 sO2 f Np (2)

In Eq 1 f is the incident flux and the total pre-cursor destruction cross section can be written ass 5 sp9 1 sO2 with sp9 accounting for precursordestruction processes other than the formation ofO2 Solving Eqs 1 and 2 and assuming N is nearlyconstant (10 level as discussed) the fractionof precursors cp in the affected column after timet is obtained (Appendix)

cp 5 NpN 5 [sps] [1 2 exp(2 s F)] (3)

Here F 5 [ft] is the fluence (flux 3 time) which

JOHNSON ET AL834

is the number of particles (photons electrons orions) incident on the surface per unit area Usingcp in Eq 2 the O2 yield Y at any fluence can bewritten in terms of the yield at steady state Y`

Y 5 [dNO2dt]f 5 sO2 cp N 5 Y`

[1 2 exp(2sF)] (4a)

Y` 5 [sp sO2 s] N (4b)

If we also account for the reduction in the den-sity of water molecules in the initial column Nthese forms change only slightly The size of thetwo parameters Y` and s can be obtained by fit-ting the yield versus fluence data such as that inFig 3

For the low-energy electron data Sieger et al(1998) showed that the ratio [YY`] for a givenincident electron energy is independent of tem-perature over a narrow range of temperatureTherefore the destruction cross section s in Eq4a is roughly independent of T and can be esti-mated from the [YY`] curve Sieger et al (1998)found a precursor destruction cross section s lt08ndash30 3 10216 cm2 for incident electron energiesfrom 30 to 100 eV This is of the order of the elec-tron impact ionization cross section for H2O

Based on the above model the lack of an ob-served fluence dependence for fluences down to1017 hncm2 in the Lyman-a irradiation of ice(Westley et al 1995) can be examined The low-est fluence used implies that the precursor de-struction cross section for Lyman-a is of the or-der of the Lyman-a absorption cross section in ice(09 3 10217 cm2)

Sieger et al (1998) also assumed that sp9 5 0This would mean that the precursor is only de-stroyed by production of O2 (ie s 5 sO2) Thatis unlikely since suggested precursors can be de-stroyed by reaction with H and other radicalsTheir assumption would also mean that the crosssection for production of O2 from a precursorsO2 is large This would be surprising and ap-pear to contradict some of the data discussedabove Instead we assume that s is not equal tosO2 and the total destruction cross section s isof the order of the ionization or absorption crosssection

Having obtained s the measured temperaturedependence for production of O2 is containedonly in Y` Over a narrow range of T one canwrite Y`(T) lt Y`9 exp(2EakT) and obtain the es-

timate of the activation energies in Table 2 Ifsp9 THORN 0 as discussed then sp and sO22 in Eq 4bcannot be obtained separately from the data forY` unless an independent measure of cp is ob-tained Such a measurement is critical for de-scribing the chemical state of the irradiated sur-faces

The observed temperature dependence of Y` iscontained in sp sO2 or N Here N is the ldquodepthrdquofrom which newly formed O2 can reach the sur-face and escape given as a column density of H2OThe physical depth is obtained by dividing thecolumn density N by the molecular number den-sity of the ice Assuming that in the low-energyelectron experiments N is independent of tem-perature and is about three monolayers (Sieger etal 1998) the measured Y` gives [sO2sps]05ndash20 3 10218 cm2 over the electron impactenergy range At 110ndash120K this ratio was foundto be proportional to (E 2 Eo) for E Eo 10 eVagain assuming fixed N (Orlando et al 1998 Or-lando and Sieger 2003)

In the following section we first consider a pre-cursor like peroxide that typically forms fromtwo dissociation events We use this to show un-der what circumstances such a model does notviolate the observed linearity of the O2 yield datawith fluence Following that we relate the sim-ple precursor model above to the case wheretrapped O is the precursor We then use themodel for trapped O to examine the effect of thetrap density on the production of O2 The densityof traps (defects) is determined by the sample for-mation temperature and the subsequent radiationdamage Finally we consider the escape depthN for penetrating radiation within the precursormodel These discussions are supported by moredetailed rate equations in the Appendix

Precursor trapped peroxide

The proposed precursor peroxide is typicallyformed from products of two dissociation eventsie 2 H2O R 2 H 1 2OH R H2 1 H2O2 This wasconfirmed in low temperature ice (Gerakines etal 1996) It is also possible that a dissociation pro-ducing a trapped O followed by an excitationevent can produce peroxide 2 (H2O 1 hn) R H2

1 H2O-O 1 hn R H2O2 1 H2 (eg Khriachtchevet al 2000) The O2 yield versus fluence for inci-dent electrons is reasonably well fit by the yieldexpression above for fluences greater than1014cm2 Therefore it has been suggested that

OXIDANTS AT EUROPA 835

any precursor species must be formed in a singleevent Using a model in the Appendix in whichthe precursor is formed by two events we showthis is not necessarily the case Precursor forma-tion if it is fast could involve two dissociationevents and give the measured fluence depen-dence Although we consider peroxide here theseconclusions apply to other possible precursors

In the example in the Appendix H2O is disso-ciated to H 1 OH according to the cross sectionsd The reactions of two OH units to form H2O2(rate constant k2OH) and 2 H to form H2 (k2H)compete with the recombination of H and OH toH2O (kHOH) If k2H is much greater than k2OH then H2 loss dominates at low fluences as ex-pected (Appendix) Steady state requires not sur-prisingly that the rate for forming H2O2 isroughly equal to the rate for forming H2 Finallyit is seen in the Appendix that the low H2O2 con-centrations suggested by experiment imply thatkHOH (k2H k2OH)05 Such sizes are reasonable

The set of rate equations in the Appendix thatuse number densities of species is integrated overdepth in order to obtain the forms used in Eqs 1and 2 which are based on column densities Inthis way the effective cross sections are related tothe rate constants for the interaction of freshlyproduced mobile species with previously pro-duced trapped species We show in the Appen-dix that the O2 yield versus fluence can be writ-ten in a form similar to that in Eq 4a but with asmall offset due to the two-step precursor for-mation process

Y lt Y`(T) [1 2 exp(2sF) 2 (ssr)]

for F sr21 (4c)

Here s is again the precursor destruction crosssection but sr 5 (2s2OH 1 sHOH) is the effectivecross section for loss of OH by formation of H2O2

and H2O The corresponding expression for Y`(T)is given in the Appendix

The form for the yield versus fluence in Eq 4cis consistent with the low-energy electron mea-surements if (ssr) 1 and if sr F21 lt10214 cm2 Such sizes are indeed possible Whatis required is that at the relevant temperatures(50ndash110K) the interaction length associated withsr is more than a few molecular layers (greaterthan 10 Aring) A cross section that is likely to belarge above 20K is sHOH which describes themean distance for diffusion of mobile H before it

reacts Since s discussed earlier is of the sameorder of magnitude as the H2O dissociation crosssection sd then the precursor concentration cp(in this case H2O2) is roughly proportional to[s2OHsHOH] 1 (Appendix)

Based on the above a two-step precursor for-mation process could be consistent with the elec-tron data contrary to the statements in Sieger etal (1998) Orlando et al (1998) and Orlando andSieger (2003) Experiments are needed to deter-mine the relevant cross sections Below we con-sider a precursor formed in a single excitationevent

Precursor trapped O

O2 can be formed for instance by diffusion ofa transiently mobile O to a previously formed andtrapped O or to the production of O near the siteof a previously produced and trapped O (Matichet al 1993 Johnson and Quickenden 1997) Werefer to this as the MatichndashJohnsonndashQuickenden(MJQ) model and note that O 1 OH R O2 1 Hcan also give molecular oxygen

Trapped O as H2O-O was in fact seen to be aprecursor to O2 when attacked by a mobile O ina Kr matrix (Khriachtchev et al 2000) ThereforeO trapped as H2O-O was also suggested as a pre-cursor in ice (Johnson 2001) Before proceedingwe note that the examination of the MJQ modelin Sieger et al (1998) was incorrect They consid-ered two mobile interacting species whose life-times were long compared with the bombard-ment rate rather than a new mobile O interactingwith a previously trapped O Although this hasnot been corrected recently they added Otrapped as H2O-O as a possible precursor (Or-lando and Sieger 2003) O trapped as H2O-O isseen in matrix isolation studies as discussedabove but it has not yet been identified in iceand like OH it may trap in a number of ways

The MJQ model is essentially the followingFormation of H2 1 O is followed by loss of H2and trapping of O (Ot) A subsequent dissocia-tion or excitation event can lead to the destruc-tion of Ot by reaction with for instance H or byconversion to H2O2 However if this dissociationevent also leads to H2 1 O a mobile O is pro-duced that can react with Ot with an interactionlength that depends on T This results in the for-mation of O2 [O 1 Ot R O2] giving a net pro-duction of 2 H2 and 1 O2 In the Appendix therelevant rate equations (Eqs A1cndashA3c) are con-

JOHNSON ET AL836

sidered Assuming the mobile O processes arefast these rate equations can also be integratedand simplified to give the form in Eqs 1 and 2The precursor column density Np now repre-sents the column density of Ot In this model sO2in Eq 2 is the ldquocross sectionrdquo for reaction of afreshly produced O with an Ot and sp9 indicatesdestruction of Ot Finally sp is the cross sectionfor production of H2 and Ot via processes dis-cussed above

Based on the sizes of the cross sections ex-tracted in the electron irradiation experimentsthe model with trapped O (Ot) as the precursoris quantitatively reasonable That is for a rangeof T both sp and sO2 might be described byevents of the type H2O R H2 1 O (whether di-rect or indirect) If these occur for 10 of theradiation-induced dissociation events as dis-cussed then the required product of cross sec-tions suggested by the sizes of s and Y` above isroughly obtained that is [sp sO2] 02ndash6 3 10234

cm2 in the low-energy electron experiments Thecross section for production of O might dependon the availability of dangling bonds at interfacesin pores or at defects as suggested by a numberof authors The importance of such bonds couldalso be consistent with the differences in the elec-tron-induced yield for initially crystalline and ini-tially amorphous ice (Orlando and Sieger 2003)

O2 production can increase in the presence ofoxygen-rich impurities such as CO2 or SO2 bothof which are known to be present at Europa Asin the MJQ model dissociation (eg CO2 R CO1 O) can provide a mobile O that can be trappedor react with a previously trapped O forming O2In addition such species are hydrogen scav-engers (Moore and Hudson 2000) This can re-duce the destruction of Ot affecting sp as seen inEq A5c

Trap density

In the simplified models above the density oftrapping sites nt was not explicitly accountedfor However in the discussion of both the ex-perimental data and models trapping sites havebeen assumed to be important It has beendemonstrated repeatedly that the density of de-fects voids and interfaces decreases as the tem-perature of formation of the ice increases How-ever the efficiency of production of O2 is theopposite it increases with increasing T There-fore the oxygen yield varies in a manner that is

the inverse of the trap density over a range oftemperatures For penetrating ions in which theenergy density deposited is high and O2 might beformed in a single track a simple explanation hasbeen given As the density of traps for O2 is re-duced and the temperature increases the O2 pro-duced can diffuse to the surface and escape moreefficiently limiting back reactions on further ir-radiation This is examined in the next sectionHowever such a process does not affect the low-energy electron yields for which the penetrationdepth is small and the formed O2 escapes read-ily at the measurement temperatures In Eqs A1cA1c9 A3c and A10 in the Appendix we againuse a simple set of rate equations for trapped andmobile O in order to determine the role of the trapdensity nt on the production of a precursorspecies The conclusions apply to any modelbased on a trapped precursor

In the MQJ model mobile O can be trappedforming Ot (described by kOt nt) and can reactwith previously trapped O to form O2 (kOOt) Inaddition it can react at other sites and Ot can bedestroyed by mobile reactive species such as H(kHOt nH) Mobile H can also react with othertrapped species (kHR nR) Incorporating these re-actions Eq A10 in the Appendix gives a steady-state yield Y` When the density of traps is largethe steady-state yield is given in Eq A5c

Y` R (sd9sd9sd) [kOOtkHR nr][kHOt kOt nt] N (5)

Here sd and sd9 are the dissociation cross sectionsthat determine the production of H (H 1 OH) andO (H2 1 O) by the incident radiation Comparingthe result in Eq 5 with Y` in the simple precur-sor model in Eq 4b the dissociation cross sec-tions in Eq 5 are now scaled by a ratio of the re-action rates

It is seen in Eq 5 that the steady-state yield Y`depends inversely on the density of traps nt Thatis there are competing destruction processes fortrapped O (Ot) Therefore in order to form O2 amobile O competes with mobile H to find atrapped O before it becomes trapped In this waythe increasing O2 yield correlates with the re-duction in the trapping sites with increasing tem-perature of formation The small activation ener-gies discussed earlier would be related to there-orientation of the lattice binding leading to thereduction in trap density At relatively high T thedensity of traps becomes small so that only the

OXIDANTS AT EUROPA 837

sample surface or grain interfaces might act astraps accounting for the observed drop in theyield at high T For this reason it is important todetermine if the yield for T greater than 150Kdepends on the incident flux f Since other quan-tities in the expression for Y` could also be tem-perature dependent experiments are neededThese would likely be pulsed radiolysis experi-ments in which the transients are studied ratherthan the steady-state yields However the avail-able data for lightly ionizing radiation do suggestthat the O2 yield depends inversely on trap den-sity as in the model above

Escape from depth

The quantity N in Eq 1 is the ldquodepthrdquo given asa molecular column density from which O2 formedbelow the surface can percolate to the surface andescape It can be converted to a physical depth bydividing by the molecular number density of theice In the case of the low-energy electrons thedepth below the surface at which ionizations andexcitations are produced is small However forfast ions that penetrate many monolayers this isnot the case Below we examine N in Eq 1

Although it has long been understood that twoexcitations events are required to produce O2(eg Brown et al 1982 Reimann et al 1984) theprecursor model suggests that at saturation(steady state) any new event can produce O2 fromthe steady-state density of trapped precursorsTherefore one might expect the yield at satura-tion to be nearly linear in the excitation cross sec-tion This is the case for the low-energy electronsand it was initially assumed to be the case for thefast ions (Brown et al 1980 Reimann et al 1984)However Baragiola et al (2003) showed that thetotal sputtering yield in the saturation region attemperatures greater than 100K which is dom-inated by decomposition (Brown et al 1982) isproportional to the square of the electronic en-ergy deposition per unit path length in the solid(dEdx)e Assuming that the production and de-struction cross sections in Y` in Eq 4b roughlyscale with (dEdx)e then their ratio [sp sO2s]also roughly varies as (dEdx)e If this is the casethen the observed quadratic dependence impliesthat the depth N from which O2 can be mobi-lized to escape must also depend on the excita-tion density (dEdx)e This is consistent with theobservation that the O2 yield depends on thick-ness for samples thinner than the ion penetration

depth (Reimann et al 1984 Benit and Brown1990) Therefore the percolation depth for escapeN is a critical factor in comparing yields for pen-etrating radiation with those for non-penetratingradiation like the low-energy electron and low-energy ion data It is also seen that percolationand trapping can be included within the simplemodel considered here

Model summary

We have examined the simple precursor modelintroduced by Sieger et al (1998) to explain theirelectron data We have shown how it can be gen-eralized and derived from an integration of theappropriate chemical rate equations with thecross sections and total column density contain-ing the reaction rates and the physics of diffusivetransport Therefore the parameters can havevery different meanings from those initially as-sumed The equations are useful when there arefast processes which go to completion in timesmuch shorter then the time between particle im-pacts and slow processes associated with speciesthat are assumed to be trapped between particleimpacts This separation allowed us to examinethose models proposed for precursor formationand destruction due to reacting mobile speciesformed from dissociation events Further fromthe size of the onset of the linearity of the yieldversus fluence we can place constraints on therole of nonlinear processes such as those thatmight be associated with the formation of a pre-cursor like peroxide

With these extensions this simple model is use-ful for approximating the full rate equations inthe temperature region for which the yields de-pend on the fluence (dose) and not on the doserate Based on the assumptions in the model thisimplies temperatures at which precursor speciesare stable over experimental times At the high-est temperatures for the yields shown in Fig 4athis might not be the case This should be testedexperimentally In this regard it is puzzling thatexperiments have shown that the absorptionbands associated with certain trapped speciessuch as OH and trapped electrons ldquodisappearrdquoat 100ndash120K (see eg Johnson and Quickenden1997) whereas the O2 yield data indicate that rel-evant precursor species must be stable to 140KTherefore the proposed O2 production processescan only be distinguished by directly determin-ing the chemical state of the irradiated ice

JOHNSON ET AL838

It may be the case that a particular precursoris not required as mentioned earlier Rather af-ter a sufficient level of oxidation a newly pro-duced O or OH can react in the oxygen-rich lat-tice producing O2 in a pseudo-first-order processThis can be tested experimentally Indeedtrapped O and peroxide may be changed into oneanother as in the matrix isolation experiments(Khriachtchev et al 2000) Therefore their rela-tive concentrations versus temperature and doseshould be studied

In the model considered here it is seen that theparameters in Y` must account for the strong tem-perature dependence Since the number of bulktrapping sites decreases with increasing T as theice sample is annealed one might expect that theprecursor concentration would decrease We sug-gest that the observed temperature dependenceis determined by the competition between theavailability of trapping sites and mean interactiondistances of a mobile species such as O and HThat is the O2 yield varies inversely with the den-sity of traps over a range of temperatures In thesimple model above this effect is included indi-rectly as shown using the more detailed equa-tions in the Appendix

EUROPA

Oxygen and peroxide production at Europa G values

The dependence of the yield on the thicknessof the ice sample for highly penetrating ions sug-gests that the O2 yield is roughly related to theamount of energy deposited in the sample There-fore the amount of O2 produced and lost fromice in steady state is often given as a G value thenumber of molecules produced per 100 eV de-posited in Table 2 In applying these G values tothe planetary environment care must be takensince the yields are temperature fluence and ex-citation density dependent They also are affectedby the presence of other proton scavengersknown to be present in Europarsquos ice

The G values for peroxide in Table 2 are thosefor production at low fluence but the G valuesfor O2 are for the production and escape of gas-phase oxygen at saturation Therefore they de-pend on the formation and destruction cross sec-tions in the model above Such G values can beused to describe the steady-state source of the am-

bient O2 atmosphere at Europa but are not help-ful in determining the ability to produce and trapO2 at depth into the icy surface Over the timescales of the experiments some O2 is seen to re-main trapped and is not destroyed by radicalsTypically there are a number of trapping sitesand recent estimates of the activation energy foran O2 molecule trapped in a substitutional site(Hori and Hondoh 2002) suggest that it can betrapped for times comparable to the time re-quired for the conversion of amorphous ice intoIh (Ayotte et al 2001) The stability of O2 inclu-sions has not been studied

To understand the relationship between the Gvalues for O2 in Table 2 we note that fast lightions and electrons lose their energy predomi-nantly by producing ionizations and excitationsThe ionization events produce secondary elec-trons that can result in further ionizations Thenumber of ionization events Ni produced by afast light ion stopping in a water ice is Ni 5EoW where Eo is the initial energy and W is theaverage energy expended per ionization event27 eV in water ice Therefore radiolysis by fastions is often described as the net effect of theshower of electrons set in motion by the incidention In this manner the G values for productionof O2 from ice by energetic light ions and elec-trons can be roughly estimated as the sum of theeffects of all the secondary electrons using theyield Ye for 10ndash100 eV electrons That is if weignore the percolation depth N we can write thenumber of O2 produced as the number of ioniza-tion events times Ye which already has in it thedestruction processes That is G NiYeEo YeW 0006 O2100 eV

This estimate of G is seen to be an order of mag-nitude larger than the keV proton value for O2 inTable 2 but is much smaller than the G value forenergetic Ne1 These differences from the simpleestimate are consistent with the observation thatthe yield in the saturation regime for penetratingions varies nonlinearly with excitation densityThe difference from the keV proton G value islikely due to the inability of O2 to escape fromdepth before destruction On the other hand thelarger G value seen using MeV Ne1 is likely dueto the high density of excitation This can resultin more efficient production of O2 increased pro-duction of damage sites for forming O2 andorincreased transport (percolation) along the tran-siently heated track That is the ability of theformed O2 to percolate to the surface during ion

OXIDANTS AT EUROPA 839

bombardment plays an essential role in deter-mining the measured yields as discussed aboveThis effect needs to be evaluated for a planetarysurface subject to particle fluxes that are many or-ders of magnitude smaller than those used in thelaboratory

In the absence of O2 yields for all of the rele-vant ions and energies an estimate of the totalsupply of O2 can be obtained using the G valuesin Table 2 For the energetic oxygen and sulfurions we use G values for neon and argon ionsThese are lower bounds as the data were obtainedat temperatures much lower than the surface tem-peratures at Europa (80ndash110K) For the fast pen-etrating electrons and protons we use the valuefor 200 keV protons in Table 2 For the heavy ionswe used the Ne1 and Ar1 values For compari-son we also give an estimate using the G valuesobtained from the low energy electron data andthe total ionization as discussed above

The G values are combined in Table 3 with theelectron and ion energy flux at Europa Althoughthere is some deflection of the flow due to in-duced fields the net energy flux to the surface isnot reduced much from these values (Paranicaset al 2002) In fact locally produced pick-up ionswhich are formed from the ejecta and acceleratedto the surface can contribute (Ip 1996) Furtheralthough G values are typically independent ofthe incident angle the loss of O2 is a surfaceprocess and is enhanced with angle That is thenumber of O2 produced within the escape depthincreases with increasing angle to the normal(steeper than the inverse cosine) This enhance-ment is 25ndash4 (Jurac et al 2001) Ignoring suchcorrections the total production rate calculated

(23ndash62 3 109 O2cm2s) is a lower bound to theactual production rate from the icy regions of Eu-roparsquos surface Recent modeling of the O2 atmo-sphere suggest that such a source rate is required(Shematovich and Johnson 2001)

Earlier estimates of the O2 source rate were in-correct because O2 production was assumed to bea fixed fraction of the H2O sputtering yield Atthe ambient temperature an ejected H2O sticks ef-ficiently (Smith and Kay 1997) to the surface ofneighboring grains in the porous regolith reduc-ing the sputtering yield for H2O (Johnson 1989)This reduction does not apply to the O2 fractionof the yield enhancing its relative importance toH2O Here we assume that all of the gas-phaseO2 produced can contribute to the ambient at-mosphere even if it is produced at a grain sur-face that is at some depth into the porous regolith

Finally Europarsquos surface is not pure ice every-where That is much of the trailing hemisphereis a hydrated sulfuric acid containing smallamounts of sulfur and SO2 and the leading hemi-sphere contains low levels of carbon CO2 andcarbonates Sodium and potassium atoms aresputtered from the icy regions (eg Leblanc et al2002) O2 production from surfaces containingthese species has not been measured C and S areboth proton scavengers and the other species canbe sources of O as discussed so that the O2 yieldfrom such regions could be enhanced Howeverthere can also be competing processes that mightreduce these rates In fact recent HST observa-tions suggest that the O2 might not be produceduniformly across Europarsquos surface (McGrath et al2000) The analysis of these observations is pre-liminary and is not simple since the observed

JOHNSON ET AL840

TABLE 3 APPLICATION TO EUROPA

OxygenImplantation Energy flux production

Species 109(cm2 s)a 109 keV(cm2 s)(1) 109(cm2 s)

H (10 keV) 0015(1) 12 0084 (072)bO (10 keV) 0015(1) 18 054cS (10 keV) 0009(1) 30 090cElectrons 62 043 (37)bPlasma (10 keV) (137 amu) 13(2) 10 03c

Total 79 23 (62)UV photons (64 eV) 46 01

aReferences are as follows 1Cooper et al (2001) [similar results in Paranicas et al (2002)] 2Mauk et al (1996)bUsing G for 200 keV protons (upper bounds in brackets are based on the low-energy electrons data as a model for

the secondary electrons produced)cUsing G for 15 MeV Ne1 at 7K a rough lower bound

emissions depend on the local plasma propertiesHowever the efficiency of O2 production in con-taminated ices needs to be studied

Porous regolith O2 interactions on grain surfaces

Although the O2 atmosphere is principally pro-duced in the near surface layers and is eventuallylost to space it also permeates Europarsquos porousregolith The porosity p (void space) of the re-golith is uncertain (Buratti 1995) It has been es-timated to be the most compact surface in theOuter Solar System p 025 on the trailing hemi-sphere and p 079 on the leading hemisphere(Buratti and Veverka 1983) and the most poroussurface p 096 globally (Domingue et al 1991Domingue and Hapke 1992) These large differ-ences are due to the uncertainties in the amountof backscattered light at small phase angles andon the photon absorption properties and grainsizes The more compact surface is likely correctGrain sizes rg of the order of 100 mm have beensuggested (Hansen and McCord 1999 2000Geissler et al 1998) Europarsquos regolith differs sig-nificantly from the lunar regolith which is asand-like surface of individual grains Analysis ofthe thermal inertia of Europarsquos surface (Grundyet al 2001) suggests that the ice grains are sin-tered forming a fairy castle-like structure Sin-tering can occur thermally or due to the incidentradiation In addition ices doped with acids andalkalis in the laboratory can be powdery solidswith the impurities primarily found at grain in-terfaces

The porous regolith affects the above estimatesof the oxygen production in three ways First forisotropic bombardment the mean distance d intothe surface at which an incident ion or photonstrikes a grain surface is d (89)rg(1 2 p) Us-ing the above value for rg then d 02 m if theporosity is very large p 96 Therefore de-sorbed molecules can come from a considerabledepth and will interact with grain surfaces Evenfor much lower porosities the effective sputteringyields can be significantly modified (Johnson1989) As discussed above the yields for ejectedH2O which sticks onto neighboring grains at theambient temperatures are reduced but those forejected O2 are not This enhances the relative im-portance of the O2 yields

Finally the surfaces and interfaces of the sin-tered grains in Europarsquos porous surface are placeswhere low temperature chemistry involving the

ambient oxygen can occur at depths that are wellbelow the nominal penetration depth of the inci-dent radiation The relative importance of suchchemistry is roughly indicated by the ratio of thetotal surface area in the regolith to the area of Eu-roparsquos visible surface Using the porosity andgrain size above and a regolith depth dr of theorder of a meter then this ratio is large 3(1 2p)drrg which is 103ndash2 3 104 for average porosi-ties of 96 and 25 respectively Therefore thegas-phase O2 molecules below the visible surfaceinteract frequently with the grain surfaces In thisway a very large volume of the subsurface couldbe altered chemically even if it is not directly ex-posed to the radiation

From the overlying structures seen in the high-resolution images of Europarsquos surface by theGalileo spacecraft (eg Prockter and Pappalardo2000) it is clear that the youngest surfaces are red-der The reddening in the visible has been attrib-uted to the presence of sulfur in the ice (McEwen1986 Johnson et al 1988 Calvin et al 1995 Carl-son et al 1999b) The amount of reddening in thespectra of the surface material decreases with ageand the surface materials appear to becomebrighter with age This might occur by the slowaccumulation of a frost over-layer or by radioly-sis (Carlson et al 2002 Johnson et al 2003) Ra-diolysis preferentially converts the sulfur in theice to sulfur dioxide or sulfuric acid both ofwhich have flat spectra in the visible In regionswhere the radiation does not directly penetratethe O2 permeating the regolith can oxidize thesulfur at the surfaces of the icy grains and at graininterfaces Similarly the carbon seen in the ice onEuroparsquos leading hemisphere (eg Carlson et al2004 Johnson et al 2003) can be oxidized by ra-diolysis in the near surface layers or by reactionson the grain surfaces at depth in the regolith bythe ambient O2 The meteoroid bombardmentthat produces the regolith also exposes fresh sur-faces for oxidation and can cause the collapse ofpores leading to the trapping of oxygen In thismanner both refractory and volatile oxidants canbe formed and trapped in the subsurface ice forpotential delivery to Europarsquos subsurface ocean

SUMMARY

The presence of a thin oxygen atmosphere andoxidants trapped in Europarsquos surface ice (egJohnson et al 2003) combined with the possible

OXIDANTS AT EUROPA 841

presence of a subsurface ocean (eg Carr et al1998) have made Europa an exciting target forfuture space exploration The molecular oxidantsproduced by radiolysis and photolysis could bespecies needed for carbon-based biochemistry inEuroparsquos putative subsurface ocean Howeverthere has been no model for the formation trap-ping and transport of relevant oxidants

In this paper we have reviewed the space ob-servations and laboratory studies related to thechemistry of formation of oxidants in ice by theincident radiation Because the laboratory dataare incomplete we examined the rate equationsfor formation of oxygen and related species byradiolysis and photolysis We showed how suchequations can be simplified to the analytic modelthat was used for fitting the fluence and temper-ature dependence of the O2 yield produced bylow-energy electrons (Sieger et al 1998) In de-riving the analytic model from the more completerate equations the competition between forma-tion and destruction processes can be studied ascan the role of trapping sites escape from depthand temperature history These equations also al-lowed us to put incident photon electron and iondata on a similar footing for the first time

The model can be used to suggest measure-ments that are needed to describe the chemicalstate of the irradiated ice and to interpret new lab-oratory data We showed that if reactions are fastthe yield data could be consistent with a precur-sor being formed by two events as is the case forperoxide However we favor trapped O as a pre-cursor as initially suggested by Matich et al(1993) for production of excited O2 Such aspecies if present could be detected spectro-scopically as can trapped O2 through bands ac-tivated by lattice perturbations (Cooper et al2003a)

The available oxygen yield data were used tocompare G values for different incident particlesThese were then used to estimate the oxygen pro-duction in the radiation environment at EuropaThe lower bounds were consistent with the pro-duction rates obtained from a recent model of Eu-roparsquos atmosphere The porosity of the regolithwas also considered It not only affects the rela-tive sputtering yields for O2 and H2O but alsoprovides an enormously increased surface areaon which the ambient gas-phase oxygen can re-act with species in the ice That this might be oc-curring is consistent with the change in re-

flectance associated with surface aging Such re-actions could also convert ambient molecularoxygen into other more refractory species that re-main trapped on subduction of Europarsquos crust

Although radiolysis produces a stable chemi-cally altered surface in the laboratory the parti-cle flux to Europarsquos surface is many orders ofmagnitude lower Therefore annealing can be ef-fective at temperatures that are much lower thanthose typically found experimentally Belowabout 120K the annealing times appear to be long(eg Baragiola 2003) so that laboratory data canoften be directly applied to Europarsquos icy surfaceHowever account has to be taken of the syner-gism between the UV and the energetic plasmaas discussed earlier (Johnson and Quickenden1997)

Transient surface melting might initiate lifeprocesses but subduction of oxidants to the pu-tative ocean is likely to be required if biochem-istry is to occur at Europa (Chyba and Hand2001) Although the temperature versus depthinto Europarsquos regolith is not well known belowthe regolith the temperature and pressure appar-ently increase with depth At some depth (ap-proximately kilometers) either very warm ice oran ocean exists (Ruiz and Tejero 2000) Thereforechemically refractory oxidants are likely to be re-quired in order to survive subduction to thewarmer subsurface regions Trapped species ifdelivered downward can become mobile andwill segregate forming inclusions (Johnson andJesser 1997) Indeed peroxide appears to segre-gate in ice (Gurman et al 1967) and these inclu-sions may remain relatively stable until the per-oxide is dissolved in the putative ocean Thevolatiles seen in reflectance such as O2 SO2 andCO2 will also segregate But at the higher tem-peratures they could also diffuse through thebulk react or percolate into the regolith atmo-sphere Therefore the observation of the trappedvolatiles in Europarsquos surface ice is of interest pri-marily because it suggests that oxidants are pre-sent The formation of oxygen-rich molecules fol-lows the preferential loss of hydrogen Howeverthe surface is in a local steady state with im-planted H O and S from the plasma and the lossto space Since the plasma flux to the surface canbe variable this balance can also vary Indeedtemporal changes in reflectance have been re-ported (Domingue and Hendrix 2004)

Based on the models for formation of O2 ex-

JOHNSON ET AL842

amined in this paper additional experiments areneeded to determine the chemical state of irra-diated ice at low fluences Of immediate inter-est is the identification of stable molecular pre-cursors required for the formation of O2 Basedon the available laboratory data individual per-oxide molecules trapped in the ice do not appearto be the principal precursors but trapped O(Matich et al 1993) remains a viable candidateAlthough the data suggest that the local hydro-gen-bonding network is important in the pro-duction of O2 the effect of the formation tem-perature and radiation damage on the structureneed to be measured It is also important to de-scribe quantitatively the relationship betweenH2 loss and the formation of O2 and H2O2 overa range of temperatures and fluences Finallyone would like to determine the steady-statedensities of biochemically interesting moleculesformed by radiolysis in an ice containing sulfurand carbon at Europarsquos surface temperaturesand then determine their stability with increas-ing temperature and pressure Such experimentsstill need to be carried out Here we have pre-sented a model that brings together the variouslaboratory data and now can be used to considerthe formation trapping and transport of relevantoxidants at Europa

APPENDIX

Introduction

In this Appendix we examine rate equationsthat describe various aspects of the radiolysis orphotolysis of ice These equations need to be un-derstood if one is to be able to calculate the for-mation trapping and transport of oxidants Theemphasis is on the formation of molecular oxy-gen We integrate these equations over the escapedepth of O2 to give the simplified rate equationsdiscussed in the text in which number densitiesare replaced by column densities (eg Eqs 1 and2) In this way we can examine the relationshipbetween the rate constants and the effective pre-cursor formation and destruction cross sectionsextracted from data We first consider a precur-sor such as H2O2 that is formed from two exci-tation events This bears on the observed linear-ity of the oxygen yield versus fluence We thenconsider the equations for trapped versus a tran-

siently mobile species We use a reaction betweena mobile and trapped O as an example althoughO 1 OH can also give oxygen Finally we ex-amine the possible dependence of the O2 forma-tion process on the density of traps (defectsvoids and grain boundaries) as a function of tem-perature The results are discussed in the text

For clarity all possible radiolytic species arenot included in the discussion below Thereforepotentially important products such as HO2 andthe charged species (eg H3O1 the trapped elec-tron etc) are missing In each section we attemptto make a point related to production of O2 andother oxidants by using a subset of the productsrather than attempt a complete description withmany unknown rate constants

Fast versus slow processes

To describe the O2 production rate from a pre-cursor such as H2O2 we first use a simple but in-structive set of rate equations to describe the for-mation of H2O2 We then relate these to thesimplified rate equation for the precursor in thetext Eq 1 and show how the postulated precur-sor formation cross section sp is related to theH2O dissociation cross section sd by the reactionrates

Using the reactions

H2O 1 radiation R H 1 OH (sd)

H 1 H R H2 (k2H)

H 1 OH R H2O (kHOH)

OH 1 OH R H2O2 (k2OH)

the reaction pathways are described by

H2

Hqk2H

1 OHH2O 1 radiation mdashRsd H

OHH2O2

OH Rk2OH

kHOH

HH2O

OXIDANTS AT EUROPA 843

The corresponding rate equations in an irradiatedvolume are

dnOHdt 5 sd f n 2 2 k2OH nOH nOH

2 kHOH nH nOH (A1a)

dnHdt 5 sd f n 2 2 k2H nH nH

2 kHOH nH nOH (A2a)

dnH2O2dt 5 k2OH nOH nOH

2 destruction processes (A3a)

Here sd is the dissociation cross section f is theradiation flux and n is the molecular numberdensity of ice with nOH and nH the density of dis-sociation products OH and H respectively Therate constants k2OH k2H and kHOH describe thereactions above In this model the H2 formed isassumed to escape for temperatures above about20K We do not distinguish between trappedspecies and mobile species although in fact thatdistinction is important as discussed in the nextsection That is any OH produced at low tem-perature by one ion would likely be trapped priorto interacting with an OH produced by a subse-quent ion At higher temperatures this can breakdown as discussed in the text We first relate theabove set of equations to Eq 1 in the text and thenconsider precursor destruction processes

Equation 1 in the text is obtained by solvingthese equations together and then by integratingover a depth Dx Here Dx is the smaller of the pen-etration depth of the particle and the O2 escape(percolation) depth Assuming the molecular den-sity of ice n is constant with depth Dx n 5 N inEq 1 Comparing Eq 1 with Eq A3a the precur-sor formation rate sp f N is equivalent to [k2OH

nOH nOH] Dx Assuming that the processes in EqsA1a and A2a are fast then the steady state is read-ily obtained In addition if k2H k2OH there ispreferential loss of H2 at short times as suggestedby the laboratory data If steady state is reachedthe production of H2 is equal to that of H2O2 us-ing only the equations above Further one obtains

[k2OH nOH nOH] Dx R [fp sd] f N (A4)

with

fp 5 05[1 1 05 kHOH(k2H k2OH)05] (A5a)

Therefore the precursor formation cross sectionsp in Eq 1 is a fraction fp of the total dissocia-

tion cross sd Since laboratory data suggest theH2O2 fraction in an irradiated ice is small thenkHOH (k2OH k2H)05 which is likely givingfp R [(k2OH k2H)05kHOH]

Trapped versus mobile species

To examine the fluence dependence discussedin the text we note that trapped OH and freshlyproduced transiently mobile OH can be treatedas different species (eg Matich et al 1993)Therefore assuming no trapping of H the OHrate equation above would be replaced by twoequations one for the trapped species nOHt andthe other for the freshly produced mobile speciesnOH Assuming the reaction of two mobile OHunits is unlikely and that peroxide is formed fromthe interaction of a trapped OH with a subse-quently produced OH an expanded set of equa-tions is obtained

OH 1 trap R OHt (kOHt)

OH 1 OHt R H2O2 (kOHOHt)

H 1 OHt R H2O (kHOHt)

then the reaction pathways are described by

qH2

H2O 1 radiation mdashRsdH

qk2H

1 OHH

trap OH

OH RkOHt

OHt kOHOHtH2O2

kHOHkHOHt

H H2O HH2O

The resulting set of equations is

dnOHdt 5 sd f n 2 kHOH nH nOH

2 kOHt nOH nt 2 kOHOHt nOH nOHt (A1b)

dnOHtdt 5 kOHt nOH nt 2 kOHOHt nOH nOHt

2 kHOHt nH nOHt (A1b9)

dnHdt 5 sd f n 2 2 k2H nH nH 2

kHOH nH nOH 2 kHOHt nH nOHt (A2b)

JOHNSON ET AL844

dnH2O2dt 5 kOHOHt nOH nOHt

2 destruction processes (A3b)

where nt is the trap density in the irradiated iceAt low T we can assume nt is large so thatdntdt 0 simplifying the set of equations If theprecursor of interest is trapped O as discussed inthe text then a similar set of equations can be con-structed for O and Ot as in the example to followOf course a full description over a broad rangeof temperatures would require the considerationof the both trapped and mobile H O and OHthe rate of change of trap density and the vari-ous destruction processes

The above equations should be solved simul-taneously but for the incident particle fluxes typ-ically used in these studies the diffusion of mo-bile species to traps is relatively fast This is trueeven at low temperatures if there is a significantdensity of trapping sites This is usually the casefor samples formed at low T Therefore in a lowtemperature solid at low dose rates there is not asignificant steady-state background of diffusingOH There are trapped or freshly produced OHunits which rapidly react Averaging over thebombardment rate this can be approximated bywriting dnOHdt 0 and then separating nOH outin Eq A1b Assuming also that H diffuses rapidlyand either reacts or forms H2 we also writednHdt 0 on the average Therefore betweenimpacts only species trapped at defects pores orgrain surfaces are assumed to be present as sug-gested by the lack of a dependence on the inci-dent beam flux In this limit the rate of change oftrapped OH roughly simplifies to

dnOHtdt 5 2 k2H nH nH 2 2 kOHOHt nOH nOHt

It is seen that the production of peroxide dependson the loss of H by formation of H2 as discussedand at steady state the production rate for per-oxide equals the rate of formation of H2 in theseequations

Treating the mobile and trapped species sepa-rately we can further simplify the rate equationsand obtain the precursor model used in the textThat is integrating over the relevant depth Dxthe densities are replaced by column densitiesThe rate constants ki are then related to effectivecross sections si In this way cross sections aregiven as an interaction length for each freshlyproduced mobile radical [ki nj] R [si f] The

trapped OH column density NOHt and peroxideproduction rate can be estimated from ldquolinearrdquorate equations

dNOHtdt lt [sd f] N 2 2 [s2OH f] NOHt

2 [sHOH f] NOHt (A6a)

dNH2O2dt lt [s2OH f] NOHt 2 s NH2O2 (A6b)

Here sd is again the dissociation cross sectionand the destruction processes in Eq A3b are con-tained in s The second and third terms on theright in Eq A6a are the loss of OH due to the pro-duction of H2O2 and H2O In the integration ofEq A1b9 [2 kOH OHt nOH] becomes [2 s2OH f] and[kHOHt nOH] becomes [sHOH f] Therefore thedensity and the interaction length of mobile H arecontained in sHOH If H2O2 is indeed the pre-cursor then in Eq 2 in the text [sp f N] is be-comes [s2OH f NOHt] from Eq A6b Accountingfor precursor destruction which can give backtwo OH and treating H2 can both be included inthe integrated model via the si values

The simplified set of equations above allows usto examine the relative time (fluence) scales forprecursor formation (in this case H2O2) SolvingEqs A6a and A6b the column density of OH andthe concentration of precursors versus fluence cpcan be obtained Using the column of H2O2 pro-duced as the precursor column Np in Eq 2 thenO2 yield can be obtained The results can be com-pared with the results from Eqs1 and 2 in thetext The average precursor (H2O2) concentrationover the column depth of water molecules N is

cp 5 fp (sds)[1 2 exp(2s F)] 2 (ssr)

[1 2 exp(2sr F)] (A7)

with

fp 5 2s2OH(sr 2 s) (A5b)

In these equations sr 5 (2s2OH 1 sHOH) andssr 1

It is seen that the precursor concentration withtime cp depends on two rates the reaction ratesthat determine sr and the destruction rates in sAs in Eq 4a the yield can be written as

Y 5 sO2 cp N (A8)

If sr and s are comparable then the yield at small

OXIDANTS AT EUROPA 845

fluence is quadratic in F However if sr sthen the linear dependence on fluence is obtainedwith a small shift in the fluence as shown in Eq4c in the text Further the steady-state yield isnow

Y` (T) 5 sO2 [fp sd] Ns (A9)

That is sp R [fp sd] as in Eq A4 and the tem-perature dependence for the incident electrondata must be contained in [sO2 fp] or in the ratio[sO2 s2OH]sHOH]

For comparison with their precursor modelSieger et al (1998) examined the possibility of ob-taining the O2 precursor by considering reactionsfor a density of diffusing dissociation products asin Eqs A1 and A2 rather than a trapped productinteracting with a freshly formed dissociationproduct Based on the above the lowest fluence atwhich O2 is first detected experimentally sets alimit on the time scale for mobilendashmobile reactions

Effect of trap density

Equations A1b A1b9 A2b and A3b can be ap-proximately solved assuming that the dnOHdtdnHdt and dntdt are very small on the averageas discussed above In order to examine the roleof the trap density nt which can depend on thetemperature and method of formation and on theradiation history of the ice we use the slightlysimpler precursor model In this model the O pro-duced directly by dissociation traps forming aprecursor O2 is then formed when a freshly pro-duced O or other species reacts with a trappedoxygen atom Ot

H2O 1 radiation R H2 1 O (sd)

O 1 trap R Ot (kOt)

O 1 Ot R O2 (kOOt)

The reaction pathways for the oxygen species are

H2O 1 radiation mdashRsd9

trap OO2

kOtOt kOOt

H2 1 OOHt

kOR H(kHO t)

ROR

We assume for simplicity both H2 and O2 escapefrom depth Dx after formation We also allow mo-bile H to react with Ot (kHOt) removing trappedO and both the mobile H and O to be removedby other reactants R (eg OH or contaminants)This leads to a simplified set of equations not di-rectly involving OH or peroxide

dnOdt 5 sd9 f n 2 kOt nO nt 2 kOR nO nR

2 kOOt nO nOt (A1c)

dnOtdt 5 kOt nt nO 2 kOOt nO nOt

2 kHOt nH nOt (A1c9)

dnHdt 5 sd f n 2 kHOt nH nOt

2 kHR nH nR (A2c)

dnO2dt 5 kOOt nO nOt (A3c)

These equations of course act in parallel with theequations above for OH and mobile OH may re-act with trapped O to form O2 Here we focusonly on mobile O

Since the diffusion of H and O are fast and thedensity of traps is large the nH and nt averagedover the bombardment rate can be taken veryroughly as time independent Further if we as-sume the number of trapped O nOt is small com-pared with the number of sites at which H canreact then nH [sd f nkHR nR] For a low ra-diation flux f nH is small on the average as as-sumed Since the O2 yield YO2 is equal to[(dnO2dt) Dxf] with Dx the escape depth thenthe steady-state yield Y` can be obtained ana-lytically

[Y`][sd9 N] 5 1 1 (q 2 1) 2 [(q 2 1)2

1 8 q d]054 d (A10)

q 5 [kOt (nt 1 nR)] (kHOt nH)(sd9 f n) kOOt

5(sdsd9) [kOt (nt 1 nR)] kHOt[kOOt kHR nR]

d 5 05 [1 1 nt(nt 1 nR)]

We note that for nt large q can be large Assum-ing that q 1 and nt nR then Eq A10 be-comes

Y` R [sd9 N]q 5 (sd9 N) (sd9sd)

[kOOt kHR nR][kHOt kOt nt] (A5c)

JOHNSON ET AL846

Comparing with Y` in the simple precursormodel in Eq 4b then s R [(kHOt nH)f] R [sd nkHOt][kHR nR] In addition the product [sO2sp] R sd9 sd9 [kOOt n][kOt nt] That is the dis-sociation cross sections are modified by reactionrates

Quite remarkably it is seen in Eq A5c that thesteady-state yield Y` can depend inversely onthe density of traps nt in the temperature andflux regimes for which the approximations arerelevant That is if there are competing destruc-tion process for trapped O then at high trap den-sity a mobile O must find a trapped O before itbecomes trapped or before mobile H destroys theOt In this way the increasing O2 yield with in-creasing T can be related to the change in the trap-ping density nt with the formation temperatureof the ice sample As the temperature increasesand the density of traps in the bulk decreases thesurface and grain interfaces become the only trap-ping sites

ACKNOWLEDGMENTS

REJ acknowledges support by a Gledden Fel-lowship in Chemistry School of Biomedical andChemical Sciences at The University of WesternAustralia and by grants from the NSF Astron-omy Program and NASArsquos Planetary Geologyand Geophysics Program PDC acknowledgessupport by an Australian Postgraduate Award

ABBREVIATIONS

HST Hubble Space Telescope MJQ MatichndashJohnsonndashQuickenden NIMS Near Infrared Map-ping Spectrometer UV ultraviolet

REFERENCES

Ayotte P Smith RS Stevenson KP Dohnarsquolek ZKimmel GA and Kay BD (2001) Effect of porosityon the adsorption desorption trapping and release ofvolatile gases by amorphous solid water J Geophys Res106(E12) 33387ndash33392

Bahr DA Famaacute M and Baragiola RA (2001) Radiol-ysis of water ice in the outer solar system sputteringand trapping of radiation products J Geophys Res 10633285ndash33290

Baragiola RA (2003) Microporous amorphous water icefilms and astronomical implications In Water in Con-

fining Geometries edited by JP Devlin and B Buch El-sevier New York in press

Baragiola RA and Bahr DA (1998) Laboratory studiesof the optical properties and stability of oxygen onGanymede J Geophys Res 103 25865ndash25872

Baragiola RA Atteberry CL and Bahr DA (1999) Ori-gin of solid oxygen on Ganymede Reply to commentby RE Johnson on ldquoLaboratory studies of the opticalproperties and stability of oxygen on Ganymederdquo JGeophys Res E104 14183ndash14187

Baragiola RA Vidal RA Svendsen W Schou J ShiM Bahr DA and Atteberrry CL (2003) Sputteringof water ice Nucl Instrum Methods Phys Res B 209294ndash303

Bar-Nun A Herman G Laufer D and Rappaport ML(1985) Trapping and release of gases by water ice andimplications for icy bodies Icarus 63 317ndash332

Baverstock KF and Burns WG (1976) Primary pro-duction of oxygen from irradiated water as an expla-nation for decreased radiobiological oxygen enhance-ment at high LET Nature 260 316ndash318

Benit J and Brown WL (1990) Electronic sputtering ofoxygen and water molecules from thin films of waterice bombarded by MeV Ne1 ions Nucl Instrum Meth-ods Phys Rev B 46 448ndash451

Benit J Bibring J-P della-Negra S LeBeyec YMendenhall M Rocard F and Standing K (1987)Erosion of ices by ion irradiation Nucl Instrum Meth-ods Phys Rev B 1920 838ndash842

Brown WL Augustyniak WM Lanzerotti LJ John-son RE and Evatt R (1980) Linear and nonlinearprocesses in the erosion of H2O ice by fast light ionsPhys Rev Lett 43 1632ndash1635

Brown WL Augustyniak W M Simmons E Mar-cantonio KJ Lanzerotti LJ Johnson RE BoringJW Reimann CT Foti G and Pirronello V (1982)Erosion and molecular formation in condensed gasfilms by electronic energy loss of fast ions Nucl In-strum Methods B 1 307ndash314

Brown WL Augustyniak WM Marcantonio KJ Sim-mons EN Boring JW Johnson RE and ReimannCT (1984) Electronic sputtering of low-temperaturemolecular solids Nucl Instrum Methods B 1 307

Buratti B (1995) Photometry and surface structure of theicy Galilean satellites J Geophys Res 100(E9) 19061ndash19066

Buratti B and Veverka J (1983) Voyager photometry ofEuropa Icarus 55 93ndash110

Calvin WM and Spencer JR (1997) Latitudinal distri-bution of O2 on Ganymede observations with the Hub-ble Space Telescope Icarus 130 505ndash516

Calvin WM Clark RN Brown RH and Spencer JA(1995) Spectra of the icy Galilean satellites from 02 to5 mm a compilation new observations and a recentsummary J Geophys Res 100 19041ndash19048

Calvin WM Johnson RE and Spencer JA (1996) O2

on Ganymede spectral characteristics and plasma for-mation mechanisms Geophys Res Lett 23 673ndash676

Calvin WM Aninich VG and Brown RH (2004) Oxy-gen temperature and phase effects Icarus (in press)

OXIDANTS AT EUROPA 847

The hydrogen readily escapes from the surfacematerial and in turn from Europarsquos atmospherebecause of the low gravitational field The optionsfor oxygen are the release into the tenuous at-mosphere where it is retained because of its highmolecular mass and retention in the ice lattice ashomogeneously dissolved oxygen as trappedbubbles of oxygen or as an oxidized trace speciesBecause of the preferential loss of hydrogen oxygen-rich species are formed preferentially inEuroparsquos surface ice (Johnson and Quickenden1997 Madey et al 2002) Therefore oxygen per-oxide and oxidized sulfur and carbon have beendetected as trapped species (eg Johnson et al2003)

The production and trapping of such species inthe surface layers are of course not sufficient forour purposes as indicated schematically in Fig1 Radiolysis can occur to depths of the order oftens of centimeters because of the penetration ofthe energetic component of the incident radiation(Cooper et al 2001 Paranicas et al 2001) Mixingof these radiolytic products to greater depths oc-curs because of meteoroid bombardment (Cooperet al 2001) This bombardment also produces aporous regolith (Buratti 1995) composed of sin-tered grains (Grundy et al 2001) which increasesthe effective radiation penetration depth In ad-dition the atmospheric O2 permeates pore spacein the regolith Therefore oxygen gas in theporous regolith can be trapped by the collapse ofpores on meteoroid impact or can react on grain

surfaces well below the typical radiation pene-tration depth Finally macroscopic mass trans-port of trapped species by crustal subduction(Prockter and Pappalardo 2000) must occur Be-cause of the temperature gradient relevant speciesmust be stable in the ice at temperatures wellabove Europarsquos surface temperature (80ndash110K)and at higher pressures The subduction processeffectively is a macroscopic mass transport pumpwhich is needed to carry oxidants to Europarsquosocean

As a first step we focus in this paper on thechemical processes leading to the formation ofoxygen and related oxidants in Europarsquos regolithThe goal is to characterize the chemical state andstability of the irradiated surface material Wefirst summarize the spacecraft and telescopic ob-servations of oxygen and related species at Eu-ropa Next we summarize the laboratory data onthe radiolysis and photolysis of ice Because dataare not available for all incident species and inci-dent energies we concentrate on the descriptionsof the chemistry of formation of molecular oxy-gen The focus on oxygen is not because this isthe species delivered to a subsurface ocean butbecause understanding the chemistry of its for-mation in ice is key to describing the chemicalstate and stability of Europarsquos irradiated surfacematerial We then use the data to correct previ-ous estimates of the production rate of oxygen atEuropa and suggest experiments in order to char-acterize the chemical state of the irradiated sur-face that leads to the production of O2

OBSERVATIONS

Molecular oxygen

Tenuous oxygen atmospheres have been re-ported at both Ganymede (Hall et al 1998) andEuropa (Hall et al 1995) based on Hubble SpaceTelescope (HST) observations The detection ofmolecular oxygen was indirect That is two lines(1356 and 1304 nm) due to excited O were ob-served in emission and their ratio (19) sug-gested that the excited atoms were produced byelectron impact dissociation of molecular oxygenIn addition two well-known transitions for in-teracting pairs of O2 molecules 1Dg 1 1Dg r3^g

2 1 3^g2 were observed in the reflectance

spectra of Ganymede (Calvin et al 1995 Spenceret al 1995) and Europa (Spencer and Klesman

JOHNSON ET AL824

FIG 1 Diagram of processes occurring on Europarsquossurface as discussed in the text

2001 Spencer and Calvin 2002) using ground-based telescopes and the HST These absorptionbands cannot be due to atmospheric O2 becausevery high pressures or very long line-of-sightcolumns of O2 would be required

The positions of the two signature absorptionsfor dimer O2 at 5772 nm and 6275 nm respec-tively (Landau et al 1962 Cooper et al 2003a)suggest that the O2 is locally quite dense (Calvinet al 1996 2003) Therefore these bands weresuggested to be due to inclusions of O2 trappedwithin Ganymedersquos icy surface (Calvin et al 1996Johnson and Jesser 1997) It is seen from Table 1that although the atmospheric and surface reser-voirs of O2 are coupled the oxygen content in thesurface ice is the largest

The observed O2-dimer band shapes and rela-tive band depths are distorted from solid or liq-uid O2 (Johnson 1999) At Ganymede where spa-tial resolution was obtained using the HST boththe position of the band minimum and its shapevary with latitude (Calvin and Spencer 1997)Therefore care must be taken when comparingthe peak positions of the bands seen in the dis-covery spectrum at Ganymede with those of solidor liquid oxygen This spectrum is a hemispher-ical average of spectral features that vary in theirpeak positions and band shapes It was obtainedby ratioing Ganymedersquos reflectance to that of Cal-listo (Spencer et al 1995) which we now knowexhibits a weak solid-like O2 feature It was re-cently shown that O2 is also trapped in Europarsquosicy surface (Spencer and Klesman 2001 Spencerand Calvin 2002) However the solid-like O2 fea-ture at Europa is much weaker than that atGanymede (5772 nm band depth 03 vs18) This could be due to the fact that Eu-roparsquos surface scatters light more efficiently in thevisible leading to shorter path lengths (Johnsonand Jesser 1997) It could also be due to compet-

ing chemical processes (Calvin et al 1996 John-son and Quickenden 1997)

Ozone and peroxide

Related to the oxygen observations the Hartleyband of O3 has been tentatively identified in the ul-traviolet (UV) reflectance spectra of Ganymede andin the icy satellites of Saturn (Noll et al 1996 1997)O3 trapped within the icy surface would be con-sistent with the presence of trapped oxygen ex-posed to radiolysis (Johnson and Quickenden1997) The presence of trapped O3 supports the in-terpretation that the dimer-O2 observations are dueto inclusions of oxygen trapped in the ice How-ever the shape of the O3-like feature also varies sig-nificantly with latitude at Ganymede (Hendrix etal 1999a) and is highly distorted Based on theband shape the presence of an additional absorb-ing species was inferred This was initially sug-gested to be trapped OH (028 mm) (Noll et al1996) but may more likely be due to organic mol-ecules (Johnson 2001) This O3-like feature is su-perimposed on a very broad UV absorption ex-tending from about 04 mm to shorter wavelengthsseen on all of the icy satellites The broad UV ab-sorption has been attributed to another radiolyti-cally produced oxidant H2O2 (Carlson et al 1999aHendrix et al 1999b) but may also include ab-sorption by trapped HO2 and OH (Taub and Eiben1968 Johnson and Quickenden 1997) Identifica-tion of specific radiolytic products in the UV is dif-ficult because the bands are very broad and over-lap The presence of H2O2 trapped in Europarsquos icewas confirmed by a band in the infrared observedusing the Near Infrared Mapping Spectrometer(NIMS) on the Galileo spacecraft (Carlson et al1999a) Remarkably the amount seen in the in-frared is roughly consistent with the amount sug-gested by the decrease in reflected light in the UV

OXIDANTS AT EUROPA 825

TABLE 1 GLOBAL OXIDANT INVENTORY

Reservoir Oxidant Column density (1016 molcm2)

Atmosphere1 O2 010Regolith (gas)2 O2 1025

Trapped in ice (1 mm at unit density)34 O2 50ndash500H2O2 400

1Based on Hall et al (1995 1998)2Estimated using average regolith depth of 1 m3Europa band depth approximately one-sixth that in Calvin et al (1996) who give 01ndash1 concentration on

Ganymede4For comparison sulfur species on the trailing hemisphere [SO2] 300 [sulfate] or 1000 3 1016 molcm2

Whereas H2O2 is observed in ice at Europa at013 level the O3-like UV feature seen atGanymede is not obvious at Europa (eg Fanaleet al 1999) This again may be due to the fact thatless trapped O2 is seen in reflectance at Europathan at Ganymede That is since O2 is a precur-sor of O3 the factor of 6 reduction in O2 meansthat the identification of this broad feature su-perimposed on other broad UV bands is prob-lematic However the absence of the O3-like fea-ture could also be due to the presence of sulfurspecies that strongly compete for excess oxygen(Johnson and Quickenden 1997)

Summary space observations

Although each of the individual identificationsof suggested products of water radiolysis or pho-tolysis (OH H2O2 O2 O3) has some uncertaintythe reflectance spectra taken as a whole give afairly clear self-consistent picture The irradiatedsurfaces of the icy satellites contain trapped mol-ecules that would be expected from radiolysis orphotolysis of an icy surface (Johnson and Quick-enden 1997) Indeed OH is a suggested precur-sor of H2O2 which is a suggested precursor ofO2 which in turn is a precursor of O3 Howeverthe separation into amounts of individual speciesbased on the absorption in the UV identificationsremains difficult Therefore the observed band oftrapped H2O2 in the infrared and the pair ofbands in the visible of a condensed form of O2are critical as they confirm the presence of radi-olytically produced oxidizing molecules in Eu-roparsquos icy surface

Consistent with the above oxygen-rich mole-cules derived from the radiation processing of anice containing sulfur (SO2 and a sulfate) and car-bon (CO2 and a carbonate) have been identified(see eg Johnson et al 2003) The sulfate hasbeen suggested to be a hydrated salt mineral (Mc-Cord et al 1999) or a hydrated sulfuric acid (Carl-son et al 1999b 2002) The latter is an efficientoxidant that is also refractory Together these ob-servations indicate that volatile and refractory ox-idizing molecules are readily formed by radioly-sis and photolysis in Europarsquos surface material Ifindeed that material can be delivered to Europarsquosputative subsurface ocean a source of free energymay be available for potential biological activityHere we consider the formation of molecular ox-idants by radiolysis with emphasis on O2

RADIOLYTIC PRODUCTIONLABORATORY DATA

Introduction

In this section we review the laboratory dataleading to the production of oxygen and relatedspecies by radiolysis and photolysis of ice Wefirst compare the data for production of oxygenby incident ions electrons and UV photons Thisis followed by a review of the data on the loss ofhydrogen which is the other principal product ofradiation-induced decomposition of ice We thenconsider the evidence that certain proposedspecies such as peroxide or trapped O might beprecursors to molecular oxygen formation in iceSome of these data have been summarized ear-lier but with a very different emphasis Here thegoal is to motivate the development of chemicalmodels for describing irradiated ice models thatcan be used to suggest new experiments and tointerpret data Individuals familiar with the datacan go directly to the modeling section in whichrate equations are developed and used to inter-pret aspects of the diverse database

Molecular oxygen

Gas-phase O2 has been produced by irradiat-ing low temperature ice using UV photons low-energy (eV) electrons keV ions and electrons andMeV ions (see eg reviews by Johnson andQuickenden 1997 Delitsky and Lane 1998 John-son 1990 1998 2001 Strazzulla 1998 Baragiola2003 Johnson et al 2003) In spite of the consid-erable differences in excitation density and pen-etration depth for these incident particles the re-sults exhibit important similarities Here wesummarize the principal findings We first notethat the efficiency of this process in pure ice islikely to be lower than estimates based on veryearly measurements of the production of O2 in ice(see eg Hart and Platzman 1961) because oflow levels of contaminants in those early experi-ments (Baragiola and Bahr 1998 Moore andHudson 2000) Therefore trace species can be im-portant in determining the O2 production ratesfrom Europarsquos icy surface and will be discussedshortly

In studying the ability of energetic plasma par-ticles to eject (sputter) molecules from ice aprocess occurring on small Outer Solar Systembodies Brown Lanzerotti and co-workers (Brown

JOHNSON ET AL826

et al 1982 1984) found that at temperatures rel-evant to the surface of Europa H2 and O2 wereimportant components of the ejecta They thenused those data to predict the presence of a radi-olytically generated O2 atmosphere on Europa(Johnson et al 1982) The production efficiencyfor gas-phase O2 was found to be strongly tem-perature dependent as seen in Fig 2a for MeVHe1 ions incident on ice and Fig 2b for keV Ar1

ions on ice In Fig 2 the production rate is givenas a yield which is the number of moleculesejected per incident ion (ie O2 production ratedivided by the flux of incident particles photonselectrons or ions) Yields are typically measuredat a low incident particle flux so that sample heat-ing does not occur The yield versus temperatureshown in Fig 2 appears to increase exponen-tially with T over a range of temperatures Al-though this is not an Arrhenius dependencerough activation energies can be extracted overnarrow ranges of T (eg Fig 2a inset) Theseare of the order of 001ndash007 eV (Table 2) Suchvalues are much smaller than those for diffusionof vacancies or trapped species in ice (Livingstonet al 2002) but are comparable to the reportedactivation energies for H1 or H diffusion in iceas noted earlier (Johnson and Quickenden1997) The lowest energies are also comparableto the enthalpy difference between amorphoussolid water and crystalline ice (Ayotte et al2001) Therefore the estimated energies are likely

related to H1 and H transport or to hydrogenbond re-orientation It has also been shown thatas the temperature is increased above 80ndash100Kmorphological changes occur that can affect thereaction kinetics and give incorrect activation en-ergies This is especially so when using measure-ments of average quantities such as yields ratherthan directly extracting rate constants by mea-suring transients (Selby 2003) In addition the ac-tivation energies can be representative of the com-peting (destruction) processes rather than theformation processes (eg Matich et al 1993) asdiscussed in the model section

The measured O2 production rates (yields)were also found to be fluence dependent wherefluence is the radiation flux 3 the time or the to-tal number of particles per unit area onto the iceMultiplying the fluence by the incident particleenergy and dividing by the average penetrationdepth gives the energy per unit volume oftencalled an average dose As seen in Fig 3 the O2yield versus fluence rises from zero implyingthat a single ion incident on a fresh surface doesnot efficiently produce and eject oxygen This isthe case even for an incident 15 MeV Ne1 forwhich the excitation density along the ion path ishigh However it is seen in Fig 3 that with in-creasing irradiation time individual impactingions produce and eject O2 with increasing effi-ciency as the material is chemically altered Atlarge fluences the yield appears to saturate indi-

OXIDANTS AT EUROPA 827

FIG 2 a Yield for D2 and O2 at saturation (steady state) from 15 MeV He1 on D2O ice versus the temperature ofthe ice (adapted from Brown et al 1982) Inset D2 yield minus the constant yield at low T versus 1kT giving an av-erage activation energy of 003 eV b Yield of O2 at saturation versus temperature for incident 30 keV Ar1 ions(adapted from Baragiola et al 2003)

a b

cating that destruction processes are in a roughsteady state with formation processes The yieldsshown in Fig 2 are purportedly those after satu-ration although full saturation is often ap-proached slowly

The slope of the O2 yield versus fluence shownin Fig 3 increases with increasing temperaturegiving the activation energies in Table 2 Thesevalues are similar to those obtained from Fig 2For energetic ions that fully penetrate the film asin Fig 2a and 3a the yield versus fluence can ex-hibit a second rise at higher fluences for whichthe radiation track cores begin to overlap This in-crease also depends on temperature and exhibitssimilar activation energies (Reimann et al 1984)The strong dependence on temperature is at firstsurprising since the energy density deposited bya 15 MeV Ne1 produces transient temperaturesin the track core that are much higher than theambient material temperature Therefore tracktemperature (local excitation density) alone doesnot determine the O2 yield

For all incident particle types and energies themeasured fluence dependence is found to be sta-ble over laboratory time scales That is for ex-periments carried out at high vacuum the yieldreturns to the value it had when the beam wasturned off This indicates that the irradiated ice isldquopermanentlyrdquo altered (Reimann et al 1984)Thus a relatively stable chemical state is createdthat is favorable for O2 production This could bethe formation of a particular trapped radiationproduct in ice that acts as a stable precursor or isdue to the altered chemical and physical state ofthe irradiated ice Trapped O H2O2 and HO2(Matich et al 1993 Johnson and Quickenden1997 Sieger et al 1998) have been proposed asstable precursor species

Using isotopically labeled H2O Benit andBrown (1990) showed that considerable transportof O occurs along the ion track when an energeticheavy ion penetrates an ice sample Because ofthis the O2 yield was found to depend on sam-ple thickness Therefore for thin ice samples the

JOHNSON ET AL828

TABLE 2 OXYGEN AND PEROXIDE PRODUCTION

Radiationa Product T (K) Ea (eV) Percent at saturation G (product100 eV)b

He1(1) O2 015g(23) O2 77 (300) 0003 (0)Ne1 (30 MeV)(3) O2 300 0031He21 (6 MeV)(3) O2 300 0008He21 (29 MeV)(3) O2 300 00016H1 (80 keV)(4) O2 70 00002H1 (200 keV)(5) O2 120 00007He21 (15 MeV)(6) O2 50 (100) 003 (07ndash4) 3 1024

Ne1 (15 MeV)(7) O2 7 10ndash15c 003(8) 70ndash140 005ndash007 01

Ar1 (01 MeV)(9) O2 30ndash140 001ndash004 0006ndash008160 001

e (30ndash100 eV)(10) O2 120 002ndash003 0005ndash0007hn (102 eV)(11) O2 50 (100) 003 00001 (00003)H1 (1 MeV)(12) H2O2 16 (77) 01 (001)

1 10 CO2 77 011 10 O2 77 04

H1 (30 keV)(13) H2O2 16 (77) 022 (038) 0098 (0086)C1 (30 keV) H2O2 16 (77) 17 (22) 023 (034)N1 (30 keV) H2O2 16 (77) 16 (14) 021 (011)O1 (30 keV) H2O2 16 (77) 14 (40) 026 (015)Ar1 (30 keV) H2O2 16 (77) 22 (20) 011 (006)

aReferences are as follows 1Lefort (1995) 2Ghormley and Stewart (1956) 3Baverstock and Burns (1976) 4Baragiolaet al (1999) 5Baragiola and Bahr (1998) 6Brown et al (1982) (ion does not stop in sample) 7Benit and Brown (1990)(ion does not stop in sample) 8Reimann et al (1984) (ion does not stop in sample) 9Baragiola (2003) 10Sieger et al(1998) (steady-state value) 11Westley et al (1995) (steady-state value as no fluence dependence seen) 12Moore andHudson (2000) 13Gomis et al (2003)

bG values are for escaping O2 For H2O2 and O2 formation both of which require multiple events yields after sat-uration are used In a few cases yields are obtained by melting the sample Values in parentheses are those at anothertemperature

cPercent of film eroded as O2 molecules for long-term irradiation

ldquosaturatedrdquo yields change slowly with fluenceMore importantly O from water molecules at aconsiderable depth into the sample contributedto the O2 ejected from the surface Since there wasno observed diffusion of the O2 in ice at the am-bient temperature (7K) over the time for their experiment they concluded that the oxygentransport and production are stimulated by theincident ions The transport was suggested as be-ing due to percolationdiffusion along the dam-aged and transiently heated tracks (Reimann etal 1984 Benit and Brown 1990) They also foundthat in fully eroding (sputtering) thin ice filmsthe total amount of O2 produced required about10ndash15 of the D2O molecules in the films at T 57K On warming irradiated samples some O2

formed at depth is found to remain trapped forexperimentally relevant times (see eg Hart andPlatzman 1961 Benit and Brown 1990 Baragiolaand Bahr 1998)

Matich et al (1993) detected luminescence in77K H2O and D2O ice from newly formed and ex-cited O2 They suggested that the O2 is formed

from the interaction of a freshly produced mo-bile O interacting with a previously producedand trapped O or H2O2 They favored the formerIn matrix isolation studies of peroxide irradiatedby UV photons luminescence from newly formedexcited O2 was seen However the oxygen wasformed from mobile O (greater than 30K in Krlattice) interacting with O trapped as H2O-O(Khriachtchev et al 2000) These studies also sug-gest that trapped O possibly in the form H2O-Omay be an important precursor

In Fig 4a the O2 yield for low-energy electronsplotted as a function of electron energy exhibits anapparent ldquothresholdrdquo at 10 eV (Orlando andSieger 2003) Although the uncertainties are largethe authors suggested there may be a lower thresh-old 6 eV corresponding to a very small crosssection for O2 production (10220 cm2) This couldbe associated with an excitation at the ice surfaceBulk ice exhibits very weak absorption at photonenergies below 6 eV but exhibits a strong ab-sorption that depending on the ice structure be-gins at about 73 eV with a peak at 86 eV This

OXIDANTS AT EUROPA 829

b

a

FIG 3 a Relative yields versus fluence (ion flux 3 time) for 15 MeV Ne1 incident on D2O ice at 7K (adapted fromReimann et al 1984) Top panel D2O yield Middle panel D2 yield Bottom panel O2 yield Yields at other tem-peratures exhibited similar dependences with the slope of O2 yield versus fluence varying with T Fits give activa-tion energies between 002 eV at the lowest fluences and lowest T to 007 eV at the highest T and highest fluencesb Oxygen yield versus fluence of low-energy electrons on D2O ice (from Sieger et al 1998)

is followed by a peak in absorption at 10 eVThe results in Fig 4a show that there is a dra-matic increase in the production of O2 at 10 eVwhich is the onset of secondary electron produc-tion in the bulk (Orlando and Sieger 2003)

For incident fast ions the yield at a given tem-perature was found after a small fluence to beindependent of whether the sample was crys-talline or amorphous ice (Brown et al 1982)However using low-energy electrons the yieldswere found to be smaller for an initially crys-talline ice (Orlando and Sieger 2003) Since verylow doses of ion radiation (a few eV per mole-cule) can convert crystalline ice to amorphous iceat low temperatures (Leto and Baratta 2003) thedamage produced by the penetrating ions usedby Brown et al (1982) affected the O2 yield Thestructural damage produced may create trappingsites cause changes in the electronic relaxationpathways or enhance percolation

The temperature dependence of the O2 yield isshown in Fig 5 for electrons incident on initiallyamorphous (Fig 5a) and initially crystalline (Fig5b) ice A small decrease in efficiency is seen inboth cases at 120ndash130K where Ic forms fromamorphous ice and pore collapse occurs (Row-land et al 1991) This is followed by an increas-ing yield with increasing T and then by a more

dramatic drop near 160K where Ih forms Dur-ing the transition to Ih trapped molecules havebeen shown to escape efficiently from ice samples (Ayotte et al 2001) A large drop in the yield isalso seen in Fig 2b for incident Ar1 ions

In modeling the dramatic decreases in the yieldin Figs 2b and 5 care must be taken That is thetime for crystallization of an amorphous sample(Baragiola 2003) even allowing for the large un-certainties is 30 min at 145K and becomes 1min at 160K Therefore pulsed radiolysis exper-iments should be carried out at these tempera-tures and the transients observed Since the timefor conversion from amorphous to crystalline icecan be shorter than the experimental times at145K the results are not comparable to thosefor which the chemical state of the ice at a givenfluence is stable over experimental times Belowwe focus on the data below 140K

We have emphasized processes initiated by theelectronic excitations and ionizations producedHowever momentum transfer collisions whichare the dominant energy loss process for low-en-ergy ions also produce dissociations defects andO2 in ice (Bar-Nun et al 1985) The relative effi-ciency of momentum transfer and electronicprocesses for producing O2 needs further studyAs the penetration depths are much smaller when

JOHNSON ET AL830

FIG 4 a Relative yield of O2 from D2O ice versus incident electron energy E at T 5 120K (from Orlando et al1998 Orlando and Sieger 2003) b Relative yield of D2 from D2O ice versus incident energy E at T 5 88K (fromKimmel et al 1994)

a

b

the former dominates the produced O2 likely es-capes readily

H2 formation and loss correlation with O2

The production of O2 from a low temperatureice by radiolysis correlates with the formationand loss of H2 (or D2 from D2O ice) (Reimann etal 1984) D2O is used to reduce the backgroundin mass spectrometry studies The yields of O2from D2O and H2O ices were shown to be com-parable However the total yields from D2O wereabout 10 less (Reimann et al 1984) consistentwith H (D) transport being important It is seenin Fig 3 that D2 is promptly produced and lostfrom the D2O ice At higher fluence the D2 yieldis seen to exhibit a dependence like that for O2For the 15 MeV Ne1 ions which fully penetratethe ice sample the D2 yield eventually decreasessince D2 is produced and escapes from the fullthickness of the film

Because of H2 (D2) loss the surface region of

the irradiated ice becomes depleted in H (D) rel-ative to O favoring the formation of oxygen-richspecies The change seen in the ratio of H (D) toO in the irradiated ice is small An irradiated film(100 nm) remains stoichiometric to better than2 (Brown et al 1980) and in the near surface re-gion Auger spectroscopy suggests only a slightoxygen enhancement due to irradiation by keVelectrons (Rye et al 1978) Sieger et al (1998) sug-gested a precursor concentration of the order ofa percent Such concentrations are an order ofmagnitude smaller than the steady-state concen-tration of dissociated H2O (Table 2)

Using infrared spectroscopy the damaged H2Oconcentrations at low temperatures in the satu-ration regime are about 10 over the irradiatedvolume (Benit et al 1987 Watanabe et al 2000Gomis et al 2003) This fraction is the ratio of thedissociation cross section to the effective crosssection for recombination to H2O as discussedlater At a 10 concentration most H2O will havea neighbor that is dissociated

OXIDANTS AT EUROPA 831

FIG 5 a Relative steady-state yields of O2 from a D2O ice sample versus T (upper panel) The ice samples wereformed and irradiated at these temperatures Film erosion as a yield (lower panel) (from Orlando and Sieger 2003)b Relative steady-state yields of O2 versus T as in (a) (upper panel) Here the D2O ice samples are formed at 160Kpresumably becoming crystalline and then irradiated at the T shown A new sample is used for each data point in(a) and (b) Film erosion as a yield (lower panel) (from Orlando and Sieger 2003)

At temperatures above 40K it is seen in Fig2 that the D2 and O2 yields exhibit the same tem-perature dependence suggesting a correlationbetween D2 and O2 loss H2 (D2) can be formedby reactions between diffusing hydrogen fromdissociated water or directly [ie H2O R H2 1O (eg Kimmel and Orlando 1995) H2O R (H1 OH)cage R H2 1 O (eg Matich et al 1993)H3O1 1 e R H2 1 OH (eg Williams et al1996)] Subtraction of the nearly constant D2 yieldin Fig 2a gives the Arrhenius plot in the insetSince molecular hydrogen readily escapes atthese temperature it is the H2 (D2) formationprocess that has an activation energy of 003eVmolecule The fact that slopes of the yields forH2 and O2 above 40K are similar suggests thatH2 (D2) formation and loss is the rate-limitingstep in determining the production and loss of O2at these temperatures

Although the data at low T in Fig 2a have con-siderable uncertainties below about 20ndash40K theD2 production is roughly independent of tem-perature but the O2 yield continues to decreasewith decreasing temperature Radiolytically pro-duced D2 is mobile and can escape above15ndash20K (Watanabe et al 2000) but O2 is im-mobile at such temperatures and the incidentHe1 ions are much less efficient at causing per-colation than the incident Ne1 discussed earlier

The production of H2 (D2) from ice has alsobeen studied using low-energy electrons and UVphotons at very low temperatures In the formerstudies absolute yields were not obtained butthere appeared to be a threshold at 63 eV (Fig4b) This is roughly the threshold for excitationof a surface state (Khusnatdinov and Petrenko1992) and the threshold for peroxide productionis close to this value The efficiency for H2 (D2)production increases around 11 eV with a sec-ond steep rise at 17 eV (Kimmel et al 1994) TheO2 and D2 yields versus electron energy in Fig 4exhibit some similarity in the threshold regionalthough the uncertainties are large and the sam-ple temperatures different Measuring the H2 (D2)and the O2 ejecta simultaneously over a range oftemperatures as in the Ne1 experiments wouldbe instructive

Using 98 eV photons at low T (12K) Watanabeet al (2000) measured the D2 ejected and trappedin ice They obtained a cross section for produc-tion of D2 (Table 2) at low fluences that is10ndash20 of the total absorption cross sectionwith significant uncertainties They also found

that the equilibrium damage fraction in the sam-ple was 10ndash20 of the D2O molecules but theequilibrium production of D2 was about an orderof magnitude smaller 1ndash2 at their tempera-tures They suggested that the D2 is formed atgrain interfaces and cracks and on the surfaces ofpores In a mixed ice D2O 1 CO exposed to UVphotolysis D2 and CO2 formation are correlated(Watanabe and Kouchi 2002) The D2O dissoci-ates to produce D2 plus O and this transientlymobile O reacts with CO to produce CO2

Precursor state and peroxide

In order to produce O2 from low temperatureice it appears that the ice sample must lose hy-drogen and form trapped oxygen-rich reactivespecies The efficiency of production also appearsto be affected by the level of defects and damagein the ice consistent with the suggestion that thehydrogen-bonding network is important in theproduction of O2 via availability of trapping sitesnew electronic relaxation pathways or the natureof the bonding of precursors Here we briefly ex-amine the possible role of the suggested precur-sor molecules H2O2 and HO2 We note that H2O2might not readily dissolve in ice possibly ac-counting for its low abundance in some experi-ments It appears to be heterogeneously distrib-uted in the form of microcrystals (Gurman et al1967) If H2O2 does act as a precursor the for-mation of aggregates could affect the dimer ab-sorption bands (Cooper et al 2003a) and affectthe production and stability of O2 (Cooper et al2003b)

The results of Gerakines et al (1996) in whichice at 10K is irradiated by UV photons confirmthat H2O2 forms in ice from two OH after a build-up of trapped OH They monitored the band as-sociated with H2O-HO (Langford et al 2000) inwhich the HO is bound to water with 024 eVTherefore they might only be monitoring a sub-set of the trapped OH seen in earlier experiments(eg Taub and Eiben 1968) The H2O2 concen-tration in ice reaches a steady-state level at flu-ences of about 1018 photonscm2 (Gerakines et al1996) As seen in the model section this gives anestimate of the H2O2 photodestruction cross sec-tion of the order of 10218 cm2 Similarly HO2 be-gins to be seen in absorption at about a few times1017cm2 shortly after they begin to detect H2O2The absolute yields were not measured

Westley et al (1995) detected H2 and O2 pro-

JOHNSON ET AL832

duced by Lyman-a (98 eV) radiation of ice butfound no fluence dependence for fluences greaterthan 1017cm2 They also saw small amounts ofH2O2 by programmed thermal desorption of thesample but made no quantitative evaluationComparing their lowest fluence for the produc-tion of O2 with the result of Gerakines et al (1996)for peroxide suggests that H2O2 and HO2 areprobably not important precursors of O2 In UVphotolysis of H2O2 using 64 eV photons the pro-duction of O2 was not observed (Vaghjiani et al1992) and the excitation of an irradiated icewhich presumably contained some peroxide by37 eV photons did not produce O2 (Baragiola etal 2003) Direct excitation of matrix-isolatedH2O2 at low temperature also did not yield O2but the production of a mobile O that reacted withtrapped O (H2O-O) did give O2 (Khriachtchev etal 2000)

Irradiating with 200 keV protons at tempera-tures of 40ndash129K Bahr et al (2001) found onlyabout 0001ndash002 H2O2 using programmed ther-mal desorption as compared with earlier experi-ments that suggested larger values (eg Table 2)Using infrared absorption Moore and Hudson(2000) found a G value of 01 for incident MeVprotons at low temperature but the signal in theobserved band became negligible at 80K sug-gesting that H2O2 was not being formed Theyalso found that adding O2 significantly increasedthe production efficiency of H2O2 That is thatthe amount of peroxide was affected by theamount of trapped oxygen Based on the matrixisolation studies (Khriachtchev et al 2000) dis-sociated mobile O might trap as H2O-O whichon excitation converts to H2O2 Therefore even ifH2O2 is not the precursor the peroxide andtrapped oxygen concentrations will be related

Although molecular oxygen was not a productof photolysis of matrix isolated H2O2 it was aproduct of photolysis at 266 nm (466 eV) of per-oxide dimers in an argon matrix 2 H2O2 R 2 H2O1 O2 (Engdahl and Nelander 2002) This de-composition was thought to be due to the con-siderable distortion of the bonds in the dimerTherefore the configuration or distortion by thelattice of peroxide or any other precursor mole-cule in the ice may be critical (Sieger et al 1998)In addition more recent experiments show thatG values for the production of H2O2 by 30 keVincident ions are significant as are the steady-state ratios of H2O2H2O (Gomis et al 2004) Ra-tios of the order of 02ndash4 were found at 16K and

77K in the irradiated regions with incident H giv-ing the smallest fraction and incident O thelargest (Table 2) At these concentrations perox-ide molecules might segregate forming inclu-sions (Gurman et al 1967 Cooper et al 2003b)Since the amount of trapped peroxide also com-petes with the amount of trapped O studies areneeded to describe the chemical state of the solidproduced by radiolysis This will be critical in de-scribing the fate of the irradiation products at Eu-ropa In particular the chemical composition fa-vorable for producing O2 needs to be determinedas does the importance of trapped O H2O2 HO2or peroxide dimers as precursors

Other species

In the model below we will refer to atomic andmolecular species trapped in ice Such species areoften discussed in the literature on radiolysis andphotolysis (for summaries see Matich et al 1993Johnson and Quickenden 1997) Trapped OH OH2O2 and HO2 have been discussed andor mea-sured by a variety of techniques (absorption andluminescence spectroscopy electron spin reso-nance programmed thermal desorption etc)Care must be taken when using these data as re-sults for ice samples formed below 160K orthose exposed to penetrating ion beams likelyhave numerous defects voids and grain inter-faces so that a variety of trapping sites and trapdepths exist Therefore when a particular bandfor a trapped species (eg OH) disappears withincreasing temperature it could be due to achange in the nature of the trapping as well as toloss of the species by reaction

Although H is discussed below we do not dis-tinguish it in these simple models from H1

trapped as the hydronium ion H3O1 whichwould mean that there are corresponding nega-tive ions present Therefore below we refer to thevarious species as if they are neutral althoughthis might not be the case Although trapped elec-trons are detected only over a narrow range oftemperatures (100ndash120K) negative ions must beimportant over a broad range of temperature

We also note that SO2 and a sulfate and CO2and a carbonate appear to be trapped species inEuroparsquos icy surface Under radiolysis each ofthese can dissociate directly giving O2 Howeverwith the exception of the sulfate their densitiesare less than 1 so they are not relevant sourcesof O2 However their presence is indicative of the

OXIDANTS AT EUROPA 833

preference for forming oxygen-rich molecules inEuroparsquos ice and can affect the efficiency of pro-duction of O2 and H2O2

MODELS FOR FORMATION OF O2

Introduction

Based on the data described above the yieldsfor incident electrons and ions are found to benearly linear in fluence at the lowest fluences forwhich O2 is detected In addition the relevant al-terations of the ice samples by radiolysis appearto be stable and independent of the beam fluxRemarkably this is the case both for low-energyelectrons that make only a single excitation foreach impact and fast ions that make a density ofexcitations along their path through the solidThis suggests that O2 is produced and trapped inthe initial events and then caused to escape fromthe solid by the subsequent radiation or that astable chemical alteration of the ice is required be-fore the radiation can produce and eject O2

Because H produced by the dissociation of wa-ter molecules can attack trapped O2 and any pre-cursor oxidants the loss of H is important Thisoccurs by the production and loss of hydrogenmolecules as discussed leaving a surface that isslightly rich in O relative to H The O2 produc-tion mechanism could be direct excitation of aparticular precursor species or it might not bespecific That is a new OH or O produced by theincident radiation could react with the H-depleted solid producing O2 This would occurmore efficiently as the ratio of O to H in the nearsurface region increased

In their early work WL Brown and colleaguesassumed that processes similar to those used todescribe radiolysis of liquid and gaseous H2Ocould be used to describe the radiolysis in ice Thelow activation energy that they found was as-sumed to be associated with diffusion of somemobile radical species in ice such as H or O orpossibly diffusion of the O2 molecule itselfBased on their studies of ejection from isotopi-cally labeled D2O samples they concluded thatthe O2 is formed in the track of a single ion andthen ejected by subsequent ions (eg Benit andBrown 1990) More recently trapped specieshave been suggested as precursors These are ei-ther subsequently excited by the incident radia-tion or attacked by newly produced mobile

species We consider a simple model that incor-porates all of these processes and can be used toorganize and interpret the experimental data Wealso discuss its relationship to more completemodels In this discussion much of which is in theAppendix we consider the standard radiationproducts that have been often described How-ever in a solid the excited electron cloud pro-duced by absorption can overlap a number of lat-tice sites Therefore their relaxation pathwaysandor product species could be very differentfrom those that have been traditionally consideredin the literature on radiolysis and photolysis

Precursor model

The column density of a precursor species Npformed in ice by incident radiation can be calcu-lated using a simple model In this model cross sec-tions are used for the production sp and destruc-tion s of the precursor O2 is then formed byexcitation of the precursor or by reaction of a newlyformed radical with the precursor This is also de-scribed using a cross section sO2 For incident elec-trons having a small penetration depth the O2 pro-duced is assumed to rapidly diffuse to the surfaceand escape (Sieger et al 1998) More generally wedefine N as the column density of H2O from whichthe incident radiation can produce and eject O2 Apair of rate equations gives both Np and the pro-duction rate of oxygen dNO2dt That is

H2O 1 radiation R precursor

precursor 1 radiation R O2

giving

dNpdt 5 sp f N 2 s f Np (1)

dNO2dt 5 sO2 f Np (2)

In Eq 1 f is the incident flux and the total pre-cursor destruction cross section can be written ass 5 sp9 1 sO2 with sp9 accounting for precursordestruction processes other than the formation ofO2 Solving Eqs 1 and 2 and assuming N is nearlyconstant (10 level as discussed) the fractionof precursors cp in the affected column after timet is obtained (Appendix)

cp 5 NpN 5 [sps] [1 2 exp(2 s F)] (3)

Here F 5 [ft] is the fluence (flux 3 time) which

JOHNSON ET AL834

is the number of particles (photons electrons orions) incident on the surface per unit area Usingcp in Eq 2 the O2 yield Y at any fluence can bewritten in terms of the yield at steady state Y`

Y 5 [dNO2dt]f 5 sO2 cp N 5 Y`

[1 2 exp(2sF)] (4a)

Y` 5 [sp sO2 s] N (4b)

If we also account for the reduction in the den-sity of water molecules in the initial column Nthese forms change only slightly The size of thetwo parameters Y` and s can be obtained by fit-ting the yield versus fluence data such as that inFig 3

For the low-energy electron data Sieger et al(1998) showed that the ratio [YY`] for a givenincident electron energy is independent of tem-perature over a narrow range of temperatureTherefore the destruction cross section s in Eq4a is roughly independent of T and can be esti-mated from the [YY`] curve Sieger et al (1998)found a precursor destruction cross section s lt08ndash30 3 10216 cm2 for incident electron energiesfrom 30 to 100 eV This is of the order of the elec-tron impact ionization cross section for H2O

Based on the above model the lack of an ob-served fluence dependence for fluences down to1017 hncm2 in the Lyman-a irradiation of ice(Westley et al 1995) can be examined The low-est fluence used implies that the precursor de-struction cross section for Lyman-a is of the or-der of the Lyman-a absorption cross section in ice(09 3 10217 cm2)

Sieger et al (1998) also assumed that sp9 5 0This would mean that the precursor is only de-stroyed by production of O2 (ie s 5 sO2) Thatis unlikely since suggested precursors can be de-stroyed by reaction with H and other radicalsTheir assumption would also mean that the crosssection for production of O2 from a precursorsO2 is large This would be surprising and ap-pear to contradict some of the data discussedabove Instead we assume that s is not equal tosO2 and the total destruction cross section s isof the order of the ionization or absorption crosssection

Having obtained s the measured temperaturedependence for production of O2 is containedonly in Y` Over a narrow range of T one canwrite Y`(T) lt Y`9 exp(2EakT) and obtain the es-

timate of the activation energies in Table 2 Ifsp9 THORN 0 as discussed then sp and sO22 in Eq 4bcannot be obtained separately from the data forY` unless an independent measure of cp is ob-tained Such a measurement is critical for de-scribing the chemical state of the irradiated sur-faces

The observed temperature dependence of Y` iscontained in sp sO2 or N Here N is the ldquodepthrdquofrom which newly formed O2 can reach the sur-face and escape given as a column density of H2OThe physical depth is obtained by dividing thecolumn density N by the molecular number den-sity of the ice Assuming that in the low-energyelectron experiments N is independent of tem-perature and is about three monolayers (Sieger etal 1998) the measured Y` gives [sO2sps]05ndash20 3 10218 cm2 over the electron impactenergy range At 110ndash120K this ratio was foundto be proportional to (E 2 Eo) for E Eo 10 eVagain assuming fixed N (Orlando et al 1998 Or-lando and Sieger 2003)

In the following section we first consider a pre-cursor like peroxide that typically forms fromtwo dissociation events We use this to show un-der what circumstances such a model does notviolate the observed linearity of the O2 yield datawith fluence Following that we relate the sim-ple precursor model above to the case wheretrapped O is the precursor We then use themodel for trapped O to examine the effect of thetrap density on the production of O2 The densityof traps (defects) is determined by the sample for-mation temperature and the subsequent radiationdamage Finally we consider the escape depthN for penetrating radiation within the precursormodel These discussions are supported by moredetailed rate equations in the Appendix

Precursor trapped peroxide

The proposed precursor peroxide is typicallyformed from products of two dissociation eventsie 2 H2O R 2 H 1 2OH R H2 1 H2O2 This wasconfirmed in low temperature ice (Gerakines etal 1996) It is also possible that a dissociation pro-ducing a trapped O followed by an excitationevent can produce peroxide 2 (H2O 1 hn) R H2

1 H2O-O 1 hn R H2O2 1 H2 (eg Khriachtchevet al 2000) The O2 yield versus fluence for inci-dent electrons is reasonably well fit by the yieldexpression above for fluences greater than1014cm2 Therefore it has been suggested that

OXIDANTS AT EUROPA 835

any precursor species must be formed in a singleevent Using a model in the Appendix in whichthe precursor is formed by two events we showthis is not necessarily the case Precursor forma-tion if it is fast could involve two dissociationevents and give the measured fluence depen-dence Although we consider peroxide here theseconclusions apply to other possible precursors

In the example in the Appendix H2O is disso-ciated to H 1 OH according to the cross sectionsd The reactions of two OH units to form H2O2(rate constant k2OH) and 2 H to form H2 (k2H)compete with the recombination of H and OH toH2O (kHOH) If k2H is much greater than k2OH then H2 loss dominates at low fluences as ex-pected (Appendix) Steady state requires not sur-prisingly that the rate for forming H2O2 isroughly equal to the rate for forming H2 Finallyit is seen in the Appendix that the low H2O2 con-centrations suggested by experiment imply thatkHOH (k2H k2OH)05 Such sizes are reasonable

The set of rate equations in the Appendix thatuse number densities of species is integrated overdepth in order to obtain the forms used in Eqs 1and 2 which are based on column densities Inthis way the effective cross sections are related tothe rate constants for the interaction of freshlyproduced mobile species with previously pro-duced trapped species We show in the Appen-dix that the O2 yield versus fluence can be writ-ten in a form similar to that in Eq 4a but with asmall offset due to the two-step precursor for-mation process

Y lt Y`(T) [1 2 exp(2sF) 2 (ssr)]

for F sr21 (4c)

Here s is again the precursor destruction crosssection but sr 5 (2s2OH 1 sHOH) is the effectivecross section for loss of OH by formation of H2O2

and H2O The corresponding expression for Y`(T)is given in the Appendix

The form for the yield versus fluence in Eq 4cis consistent with the low-energy electron mea-surements if (ssr) 1 and if sr F21 lt10214 cm2 Such sizes are indeed possible Whatis required is that at the relevant temperatures(50ndash110K) the interaction length associated withsr is more than a few molecular layers (greaterthan 10 Aring) A cross section that is likely to belarge above 20K is sHOH which describes themean distance for diffusion of mobile H before it

reacts Since s discussed earlier is of the sameorder of magnitude as the H2O dissociation crosssection sd then the precursor concentration cp(in this case H2O2) is roughly proportional to[s2OHsHOH] 1 (Appendix)

Based on the above a two-step precursor for-mation process could be consistent with the elec-tron data contrary to the statements in Sieger etal (1998) Orlando et al (1998) and Orlando andSieger (2003) Experiments are needed to deter-mine the relevant cross sections Below we con-sider a precursor formed in a single excitationevent

Precursor trapped O

O2 can be formed for instance by diffusion ofa transiently mobile O to a previously formed andtrapped O or to the production of O near the siteof a previously produced and trapped O (Matichet al 1993 Johnson and Quickenden 1997) Werefer to this as the MatichndashJohnsonndashQuickenden(MJQ) model and note that O 1 OH R O2 1 Hcan also give molecular oxygen

Trapped O as H2O-O was in fact seen to be aprecursor to O2 when attacked by a mobile O ina Kr matrix (Khriachtchev et al 2000) ThereforeO trapped as H2O-O was also suggested as a pre-cursor in ice (Johnson 2001) Before proceedingwe note that the examination of the MJQ modelin Sieger et al (1998) was incorrect They consid-ered two mobile interacting species whose life-times were long compared with the bombard-ment rate rather than a new mobile O interactingwith a previously trapped O Although this hasnot been corrected recently they added Otrapped as H2O-O as a possible precursor (Or-lando and Sieger 2003) O trapped as H2O-O isseen in matrix isolation studies as discussedabove but it has not yet been identified in iceand like OH it may trap in a number of ways

The MJQ model is essentially the followingFormation of H2 1 O is followed by loss of H2and trapping of O (Ot) A subsequent dissocia-tion or excitation event can lead to the destruc-tion of Ot by reaction with for instance H or byconversion to H2O2 However if this dissociationevent also leads to H2 1 O a mobile O is pro-duced that can react with Ot with an interactionlength that depends on T This results in the for-mation of O2 [O 1 Ot R O2] giving a net pro-duction of 2 H2 and 1 O2 In the Appendix therelevant rate equations (Eqs A1cndashA3c) are con-

JOHNSON ET AL836

sidered Assuming the mobile O processes arefast these rate equations can also be integratedand simplified to give the form in Eqs 1 and 2The precursor column density Np now repre-sents the column density of Ot In this model sO2in Eq 2 is the ldquocross sectionrdquo for reaction of afreshly produced O with an Ot and sp9 indicatesdestruction of Ot Finally sp is the cross sectionfor production of H2 and Ot via processes dis-cussed above

Based on the sizes of the cross sections ex-tracted in the electron irradiation experimentsthe model with trapped O (Ot) as the precursoris quantitatively reasonable That is for a rangeof T both sp and sO2 might be described byevents of the type H2O R H2 1 O (whether di-rect or indirect) If these occur for 10 of theradiation-induced dissociation events as dis-cussed then the required product of cross sec-tions suggested by the sizes of s and Y` above isroughly obtained that is [sp sO2] 02ndash6 3 10234

cm2 in the low-energy electron experiments Thecross section for production of O might dependon the availability of dangling bonds at interfacesin pores or at defects as suggested by a numberof authors The importance of such bonds couldalso be consistent with the differences in the elec-tron-induced yield for initially crystalline and ini-tially amorphous ice (Orlando and Sieger 2003)

O2 production can increase in the presence ofoxygen-rich impurities such as CO2 or SO2 bothof which are known to be present at Europa Asin the MJQ model dissociation (eg CO2 R CO1 O) can provide a mobile O that can be trappedor react with a previously trapped O forming O2In addition such species are hydrogen scav-engers (Moore and Hudson 2000) This can re-duce the destruction of Ot affecting sp as seen inEq A5c

Trap density

In the simplified models above the density oftrapping sites nt was not explicitly accountedfor However in the discussion of both the ex-perimental data and models trapping sites havebeen assumed to be important It has beendemonstrated repeatedly that the density of de-fects voids and interfaces decreases as the tem-perature of formation of the ice increases How-ever the efficiency of production of O2 is theopposite it increases with increasing T There-fore the oxygen yield varies in a manner that is

the inverse of the trap density over a range oftemperatures For penetrating ions in which theenergy density deposited is high and O2 might beformed in a single track a simple explanation hasbeen given As the density of traps for O2 is re-duced and the temperature increases the O2 pro-duced can diffuse to the surface and escape moreefficiently limiting back reactions on further ir-radiation This is examined in the next sectionHowever such a process does not affect the low-energy electron yields for which the penetrationdepth is small and the formed O2 escapes read-ily at the measurement temperatures In Eqs A1cA1c9 A3c and A10 in the Appendix we againuse a simple set of rate equations for trapped andmobile O in order to determine the role of the trapdensity nt on the production of a precursorspecies The conclusions apply to any modelbased on a trapped precursor

In the MQJ model mobile O can be trappedforming Ot (described by kOt nt) and can reactwith previously trapped O to form O2 (kOOt) Inaddition it can react at other sites and Ot can bedestroyed by mobile reactive species such as H(kHOt nH) Mobile H can also react with othertrapped species (kHR nR) Incorporating these re-actions Eq A10 in the Appendix gives a steady-state yield Y` When the density of traps is largethe steady-state yield is given in Eq A5c

Y` R (sd9sd9sd) [kOOtkHR nr][kHOt kOt nt] N (5)

Here sd and sd9 are the dissociation cross sectionsthat determine the production of H (H 1 OH) andO (H2 1 O) by the incident radiation Comparingthe result in Eq 5 with Y` in the simple precur-sor model in Eq 4b the dissociation cross sec-tions in Eq 5 are now scaled by a ratio of the re-action rates

It is seen in Eq 5 that the steady-state yield Y`depends inversely on the density of traps nt Thatis there are competing destruction processes fortrapped O (Ot) Therefore in order to form O2 amobile O competes with mobile H to find atrapped O before it becomes trapped In this waythe increasing O2 yield correlates with the re-duction in the trapping sites with increasing tem-perature of formation The small activation ener-gies discussed earlier would be related to there-orientation of the lattice binding leading to thereduction in trap density At relatively high T thedensity of traps becomes small so that only the

OXIDANTS AT EUROPA 837

sample surface or grain interfaces might act astraps accounting for the observed drop in theyield at high T For this reason it is important todetermine if the yield for T greater than 150Kdepends on the incident flux f Since other quan-tities in the expression for Y` could also be tem-perature dependent experiments are neededThese would likely be pulsed radiolysis experi-ments in which the transients are studied ratherthan the steady-state yields However the avail-able data for lightly ionizing radiation do suggestthat the O2 yield depends inversely on trap den-sity as in the model above

Escape from depth

The quantity N in Eq 1 is the ldquodepthrdquo given asa molecular column density from which O2 formedbelow the surface can percolate to the surface andescape It can be converted to a physical depth bydividing by the molecular number density of theice In the case of the low-energy electrons thedepth below the surface at which ionizations andexcitations are produced is small However forfast ions that penetrate many monolayers this isnot the case Below we examine N in Eq 1

Although it has long been understood that twoexcitations events are required to produce O2(eg Brown et al 1982 Reimann et al 1984) theprecursor model suggests that at saturation(steady state) any new event can produce O2 fromthe steady-state density of trapped precursorsTherefore one might expect the yield at satura-tion to be nearly linear in the excitation cross sec-tion This is the case for the low-energy electronsand it was initially assumed to be the case for thefast ions (Brown et al 1980 Reimann et al 1984)However Baragiola et al (2003) showed that thetotal sputtering yield in the saturation region attemperatures greater than 100K which is dom-inated by decomposition (Brown et al 1982) isproportional to the square of the electronic en-ergy deposition per unit path length in the solid(dEdx)e Assuming that the production and de-struction cross sections in Y` in Eq 4b roughlyscale with (dEdx)e then their ratio [sp sO2s]also roughly varies as (dEdx)e If this is the casethen the observed quadratic dependence impliesthat the depth N from which O2 can be mobi-lized to escape must also depend on the excita-tion density (dEdx)e This is consistent with theobservation that the O2 yield depends on thick-ness for samples thinner than the ion penetration

depth (Reimann et al 1984 Benit and Brown1990) Therefore the percolation depth for escapeN is a critical factor in comparing yields for pen-etrating radiation with those for non-penetratingradiation like the low-energy electron and low-energy ion data It is also seen that percolationand trapping can be included within the simplemodel considered here

Model summary

We have examined the simple precursor modelintroduced by Sieger et al (1998) to explain theirelectron data We have shown how it can be gen-eralized and derived from an integration of theappropriate chemical rate equations with thecross sections and total column density contain-ing the reaction rates and the physics of diffusivetransport Therefore the parameters can havevery different meanings from those initially as-sumed The equations are useful when there arefast processes which go to completion in timesmuch shorter then the time between particle im-pacts and slow processes associated with speciesthat are assumed to be trapped between particleimpacts This separation allowed us to examinethose models proposed for precursor formationand destruction due to reacting mobile speciesformed from dissociation events Further fromthe size of the onset of the linearity of the yieldversus fluence we can place constraints on therole of nonlinear processes such as those thatmight be associated with the formation of a pre-cursor like peroxide

With these extensions this simple model is use-ful for approximating the full rate equations inthe temperature region for which the yields de-pend on the fluence (dose) and not on the doserate Based on the assumptions in the model thisimplies temperatures at which precursor speciesare stable over experimental times At the high-est temperatures for the yields shown in Fig 4athis might not be the case This should be testedexperimentally In this regard it is puzzling thatexperiments have shown that the absorptionbands associated with certain trapped speciessuch as OH and trapped electrons ldquodisappearrdquoat 100ndash120K (see eg Johnson and Quickenden1997) whereas the O2 yield data indicate that rel-evant precursor species must be stable to 140KTherefore the proposed O2 production processescan only be distinguished by directly determin-ing the chemical state of the irradiated ice

JOHNSON ET AL838

It may be the case that a particular precursoris not required as mentioned earlier Rather af-ter a sufficient level of oxidation a newly pro-duced O or OH can react in the oxygen-rich lat-tice producing O2 in a pseudo-first-order processThis can be tested experimentally Indeedtrapped O and peroxide may be changed into oneanother as in the matrix isolation experiments(Khriachtchev et al 2000) Therefore their rela-tive concentrations versus temperature and doseshould be studied

In the model considered here it is seen that theparameters in Y` must account for the strong tem-perature dependence Since the number of bulktrapping sites decreases with increasing T as theice sample is annealed one might expect that theprecursor concentration would decrease We sug-gest that the observed temperature dependenceis determined by the competition between theavailability of trapping sites and mean interactiondistances of a mobile species such as O and HThat is the O2 yield varies inversely with the den-sity of traps over a range of temperatures In thesimple model above this effect is included indi-rectly as shown using the more detailed equa-tions in the Appendix

EUROPA

Oxygen and peroxide production at Europa G values

The dependence of the yield on the thicknessof the ice sample for highly penetrating ions sug-gests that the O2 yield is roughly related to theamount of energy deposited in the sample There-fore the amount of O2 produced and lost fromice in steady state is often given as a G value thenumber of molecules produced per 100 eV de-posited in Table 2 In applying these G values tothe planetary environment care must be takensince the yields are temperature fluence and ex-citation density dependent They also are affectedby the presence of other proton scavengersknown to be present in Europarsquos ice

The G values for peroxide in Table 2 are thosefor production at low fluence but the G valuesfor O2 are for the production and escape of gas-phase oxygen at saturation Therefore they de-pend on the formation and destruction cross sec-tions in the model above Such G values can beused to describe the steady-state source of the am-

bient O2 atmosphere at Europa but are not help-ful in determining the ability to produce and trapO2 at depth into the icy surface Over the timescales of the experiments some O2 is seen to re-main trapped and is not destroyed by radicalsTypically there are a number of trapping sitesand recent estimates of the activation energy foran O2 molecule trapped in a substitutional site(Hori and Hondoh 2002) suggest that it can betrapped for times comparable to the time re-quired for the conversion of amorphous ice intoIh (Ayotte et al 2001) The stability of O2 inclu-sions has not been studied

To understand the relationship between the Gvalues for O2 in Table 2 we note that fast lightions and electrons lose their energy predomi-nantly by producing ionizations and excitationsThe ionization events produce secondary elec-trons that can result in further ionizations Thenumber of ionization events Ni produced by afast light ion stopping in a water ice is Ni 5EoW where Eo is the initial energy and W is theaverage energy expended per ionization event27 eV in water ice Therefore radiolysis by fastions is often described as the net effect of theshower of electrons set in motion by the incidention In this manner the G values for productionof O2 from ice by energetic light ions and elec-trons can be roughly estimated as the sum of theeffects of all the secondary electrons using theyield Ye for 10ndash100 eV electrons That is if weignore the percolation depth N we can write thenumber of O2 produced as the number of ioniza-tion events times Ye which already has in it thedestruction processes That is G NiYeEo YeW 0006 O2100 eV

This estimate of G is seen to be an order of mag-nitude larger than the keV proton value for O2 inTable 2 but is much smaller than the G value forenergetic Ne1 These differences from the simpleestimate are consistent with the observation thatthe yield in the saturation regime for penetratingions varies nonlinearly with excitation densityThe difference from the keV proton G value islikely due to the inability of O2 to escape fromdepth before destruction On the other hand thelarger G value seen using MeV Ne1 is likely dueto the high density of excitation This can resultin more efficient production of O2 increased pro-duction of damage sites for forming O2 andorincreased transport (percolation) along the tran-siently heated track That is the ability of theformed O2 to percolate to the surface during ion

OXIDANTS AT EUROPA 839

bombardment plays an essential role in deter-mining the measured yields as discussed aboveThis effect needs to be evaluated for a planetarysurface subject to particle fluxes that are many or-ders of magnitude smaller than those used in thelaboratory

In the absence of O2 yields for all of the rele-vant ions and energies an estimate of the totalsupply of O2 can be obtained using the G valuesin Table 2 For the energetic oxygen and sulfurions we use G values for neon and argon ionsThese are lower bounds as the data were obtainedat temperatures much lower than the surface tem-peratures at Europa (80ndash110K) For the fast pen-etrating electrons and protons we use the valuefor 200 keV protons in Table 2 For the heavy ionswe used the Ne1 and Ar1 values For compari-son we also give an estimate using the G valuesobtained from the low energy electron data andthe total ionization as discussed above

The G values are combined in Table 3 with theelectron and ion energy flux at Europa Althoughthere is some deflection of the flow due to in-duced fields the net energy flux to the surface isnot reduced much from these values (Paranicaset al 2002) In fact locally produced pick-up ionswhich are formed from the ejecta and acceleratedto the surface can contribute (Ip 1996) Furtheralthough G values are typically independent ofthe incident angle the loss of O2 is a surfaceprocess and is enhanced with angle That is thenumber of O2 produced within the escape depthincreases with increasing angle to the normal(steeper than the inverse cosine) This enhance-ment is 25ndash4 (Jurac et al 2001) Ignoring suchcorrections the total production rate calculated

(23ndash62 3 109 O2cm2s) is a lower bound to theactual production rate from the icy regions of Eu-roparsquos surface Recent modeling of the O2 atmo-sphere suggest that such a source rate is required(Shematovich and Johnson 2001)

Earlier estimates of the O2 source rate were in-correct because O2 production was assumed to bea fixed fraction of the H2O sputtering yield Atthe ambient temperature an ejected H2O sticks ef-ficiently (Smith and Kay 1997) to the surface ofneighboring grains in the porous regolith reduc-ing the sputtering yield for H2O (Johnson 1989)This reduction does not apply to the O2 fractionof the yield enhancing its relative importance toH2O Here we assume that all of the gas-phaseO2 produced can contribute to the ambient at-mosphere even if it is produced at a grain sur-face that is at some depth into the porous regolith

Finally Europarsquos surface is not pure ice every-where That is much of the trailing hemisphereis a hydrated sulfuric acid containing smallamounts of sulfur and SO2 and the leading hemi-sphere contains low levels of carbon CO2 andcarbonates Sodium and potassium atoms aresputtered from the icy regions (eg Leblanc et al2002) O2 production from surfaces containingthese species has not been measured C and S areboth proton scavengers and the other species canbe sources of O as discussed so that the O2 yieldfrom such regions could be enhanced Howeverthere can also be competing processes that mightreduce these rates In fact recent HST observa-tions suggest that the O2 might not be produceduniformly across Europarsquos surface (McGrath et al2000) The analysis of these observations is pre-liminary and is not simple since the observed

JOHNSON ET AL840

TABLE 3 APPLICATION TO EUROPA

OxygenImplantation Energy flux production

Species 109(cm2 s)a 109 keV(cm2 s)(1) 109(cm2 s)

H (10 keV) 0015(1) 12 0084 (072)bO (10 keV) 0015(1) 18 054cS (10 keV) 0009(1) 30 090cElectrons 62 043 (37)bPlasma (10 keV) (137 amu) 13(2) 10 03c

Total 79 23 (62)UV photons (64 eV) 46 01

aReferences are as follows 1Cooper et al (2001) [similar results in Paranicas et al (2002)] 2Mauk et al (1996)bUsing G for 200 keV protons (upper bounds in brackets are based on the low-energy electrons data as a model for

the secondary electrons produced)cUsing G for 15 MeV Ne1 at 7K a rough lower bound

emissions depend on the local plasma propertiesHowever the efficiency of O2 production in con-taminated ices needs to be studied

Porous regolith O2 interactions on grain surfaces

Although the O2 atmosphere is principally pro-duced in the near surface layers and is eventuallylost to space it also permeates Europarsquos porousregolith The porosity p (void space) of the re-golith is uncertain (Buratti 1995) It has been es-timated to be the most compact surface in theOuter Solar System p 025 on the trailing hemi-sphere and p 079 on the leading hemisphere(Buratti and Veverka 1983) and the most poroussurface p 096 globally (Domingue et al 1991Domingue and Hapke 1992) These large differ-ences are due to the uncertainties in the amountof backscattered light at small phase angles andon the photon absorption properties and grainsizes The more compact surface is likely correctGrain sizes rg of the order of 100 mm have beensuggested (Hansen and McCord 1999 2000Geissler et al 1998) Europarsquos regolith differs sig-nificantly from the lunar regolith which is asand-like surface of individual grains Analysis ofthe thermal inertia of Europarsquos surface (Grundyet al 2001) suggests that the ice grains are sin-tered forming a fairy castle-like structure Sin-tering can occur thermally or due to the incidentradiation In addition ices doped with acids andalkalis in the laboratory can be powdery solidswith the impurities primarily found at grain in-terfaces

The porous regolith affects the above estimatesof the oxygen production in three ways First forisotropic bombardment the mean distance d intothe surface at which an incident ion or photonstrikes a grain surface is d (89)rg(1 2 p) Us-ing the above value for rg then d 02 m if theporosity is very large p 96 Therefore de-sorbed molecules can come from a considerabledepth and will interact with grain surfaces Evenfor much lower porosities the effective sputteringyields can be significantly modified (Johnson1989) As discussed above the yields for ejectedH2O which sticks onto neighboring grains at theambient temperatures are reduced but those forejected O2 are not This enhances the relative im-portance of the O2 yields

Finally the surfaces and interfaces of the sin-tered grains in Europarsquos porous surface are placeswhere low temperature chemistry involving the

ambient oxygen can occur at depths that are wellbelow the nominal penetration depth of the inci-dent radiation The relative importance of suchchemistry is roughly indicated by the ratio of thetotal surface area in the regolith to the area of Eu-roparsquos visible surface Using the porosity andgrain size above and a regolith depth dr of theorder of a meter then this ratio is large 3(1 2p)drrg which is 103ndash2 3 104 for average porosi-ties of 96 and 25 respectively Therefore thegas-phase O2 molecules below the visible surfaceinteract frequently with the grain surfaces In thisway a very large volume of the subsurface couldbe altered chemically even if it is not directly ex-posed to the radiation

From the overlying structures seen in the high-resolution images of Europarsquos surface by theGalileo spacecraft (eg Prockter and Pappalardo2000) it is clear that the youngest surfaces are red-der The reddening in the visible has been attrib-uted to the presence of sulfur in the ice (McEwen1986 Johnson et al 1988 Calvin et al 1995 Carl-son et al 1999b) The amount of reddening in thespectra of the surface material decreases with ageand the surface materials appear to becomebrighter with age This might occur by the slowaccumulation of a frost over-layer or by radioly-sis (Carlson et al 2002 Johnson et al 2003) Ra-diolysis preferentially converts the sulfur in theice to sulfur dioxide or sulfuric acid both ofwhich have flat spectra in the visible In regionswhere the radiation does not directly penetratethe O2 permeating the regolith can oxidize thesulfur at the surfaces of the icy grains and at graininterfaces Similarly the carbon seen in the ice onEuroparsquos leading hemisphere (eg Carlson et al2004 Johnson et al 2003) can be oxidized by ra-diolysis in the near surface layers or by reactionson the grain surfaces at depth in the regolith bythe ambient O2 The meteoroid bombardmentthat produces the regolith also exposes fresh sur-faces for oxidation and can cause the collapse ofpores leading to the trapping of oxygen In thismanner both refractory and volatile oxidants canbe formed and trapped in the subsurface ice forpotential delivery to Europarsquos subsurface ocean

SUMMARY

The presence of a thin oxygen atmosphere andoxidants trapped in Europarsquos surface ice (egJohnson et al 2003) combined with the possible

OXIDANTS AT EUROPA 841

presence of a subsurface ocean (eg Carr et al1998) have made Europa an exciting target forfuture space exploration The molecular oxidantsproduced by radiolysis and photolysis could bespecies needed for carbon-based biochemistry inEuroparsquos putative subsurface ocean Howeverthere has been no model for the formation trap-ping and transport of relevant oxidants

In this paper we have reviewed the space ob-servations and laboratory studies related to thechemistry of formation of oxidants in ice by theincident radiation Because the laboratory dataare incomplete we examined the rate equationsfor formation of oxygen and related species byradiolysis and photolysis We showed how suchequations can be simplified to the analytic modelthat was used for fitting the fluence and temper-ature dependence of the O2 yield produced bylow-energy electrons (Sieger et al 1998) In de-riving the analytic model from the more completerate equations the competition between forma-tion and destruction processes can be studied ascan the role of trapping sites escape from depthand temperature history These equations also al-lowed us to put incident photon electron and iondata on a similar footing for the first time

The model can be used to suggest measure-ments that are needed to describe the chemicalstate of the irradiated ice and to interpret new lab-oratory data We showed that if reactions are fastthe yield data could be consistent with a precur-sor being formed by two events as is the case forperoxide However we favor trapped O as a pre-cursor as initially suggested by Matich et al(1993) for production of excited O2 Such aspecies if present could be detected spectro-scopically as can trapped O2 through bands ac-tivated by lattice perturbations (Cooper et al2003a)

The available oxygen yield data were used tocompare G values for different incident particlesThese were then used to estimate the oxygen pro-duction in the radiation environment at EuropaThe lower bounds were consistent with the pro-duction rates obtained from a recent model of Eu-roparsquos atmosphere The porosity of the regolithwas also considered It not only affects the rela-tive sputtering yields for O2 and H2O but alsoprovides an enormously increased surface areaon which the ambient gas-phase oxygen can re-act with species in the ice That this might be oc-curring is consistent with the change in re-

flectance associated with surface aging Such re-actions could also convert ambient molecularoxygen into other more refractory species that re-main trapped on subduction of Europarsquos crust

Although radiolysis produces a stable chemi-cally altered surface in the laboratory the parti-cle flux to Europarsquos surface is many orders ofmagnitude lower Therefore annealing can be ef-fective at temperatures that are much lower thanthose typically found experimentally Belowabout 120K the annealing times appear to be long(eg Baragiola 2003) so that laboratory data canoften be directly applied to Europarsquos icy surfaceHowever account has to be taken of the syner-gism between the UV and the energetic plasmaas discussed earlier (Johnson and Quickenden1997)

Transient surface melting might initiate lifeprocesses but subduction of oxidants to the pu-tative ocean is likely to be required if biochem-istry is to occur at Europa (Chyba and Hand2001) Although the temperature versus depthinto Europarsquos regolith is not well known belowthe regolith the temperature and pressure appar-ently increase with depth At some depth (ap-proximately kilometers) either very warm ice oran ocean exists (Ruiz and Tejero 2000) Thereforechemically refractory oxidants are likely to be re-quired in order to survive subduction to thewarmer subsurface regions Trapped species ifdelivered downward can become mobile andwill segregate forming inclusions (Johnson andJesser 1997) Indeed peroxide appears to segre-gate in ice (Gurman et al 1967) and these inclu-sions may remain relatively stable until the per-oxide is dissolved in the putative ocean Thevolatiles seen in reflectance such as O2 SO2 andCO2 will also segregate But at the higher tem-peratures they could also diffuse through thebulk react or percolate into the regolith atmo-sphere Therefore the observation of the trappedvolatiles in Europarsquos surface ice is of interest pri-marily because it suggests that oxidants are pre-sent The formation of oxygen-rich molecules fol-lows the preferential loss of hydrogen Howeverthe surface is in a local steady state with im-planted H O and S from the plasma and the lossto space Since the plasma flux to the surface canbe variable this balance can also vary Indeedtemporal changes in reflectance have been re-ported (Domingue and Hendrix 2004)

Based on the models for formation of O2 ex-

JOHNSON ET AL842

amined in this paper additional experiments areneeded to determine the chemical state of irra-diated ice at low fluences Of immediate inter-est is the identification of stable molecular pre-cursors required for the formation of O2 Basedon the available laboratory data individual per-oxide molecules trapped in the ice do not appearto be the principal precursors but trapped O(Matich et al 1993) remains a viable candidateAlthough the data suggest that the local hydro-gen-bonding network is important in the pro-duction of O2 the effect of the formation tem-perature and radiation damage on the structureneed to be measured It is also important to de-scribe quantitatively the relationship betweenH2 loss and the formation of O2 and H2O2 overa range of temperatures and fluences Finallyone would like to determine the steady-statedensities of biochemically interesting moleculesformed by radiolysis in an ice containing sulfurand carbon at Europarsquos surface temperaturesand then determine their stability with increas-ing temperature and pressure Such experimentsstill need to be carried out Here we have pre-sented a model that brings together the variouslaboratory data and now can be used to considerthe formation trapping and transport of relevantoxidants at Europa

APPENDIX

Introduction

In this Appendix we examine rate equationsthat describe various aspects of the radiolysis orphotolysis of ice These equations need to be un-derstood if one is to be able to calculate the for-mation trapping and transport of oxidants Theemphasis is on the formation of molecular oxy-gen We integrate these equations over the escapedepth of O2 to give the simplified rate equationsdiscussed in the text in which number densitiesare replaced by column densities (eg Eqs 1 and2) In this way we can examine the relationshipbetween the rate constants and the effective pre-cursor formation and destruction cross sectionsextracted from data We first consider a precur-sor such as H2O2 that is formed from two exci-tation events This bears on the observed linear-ity of the oxygen yield versus fluence We thenconsider the equations for trapped versus a tran-

siently mobile species We use a reaction betweena mobile and trapped O as an example althoughO 1 OH can also give oxygen Finally we ex-amine the possible dependence of the O2 forma-tion process on the density of traps (defectsvoids and grain boundaries) as a function of tem-perature The results are discussed in the text

For clarity all possible radiolytic species arenot included in the discussion below Thereforepotentially important products such as HO2 andthe charged species (eg H3O1 the trapped elec-tron etc) are missing In each section we attemptto make a point related to production of O2 andother oxidants by using a subset of the productsrather than attempt a complete description withmany unknown rate constants

Fast versus slow processes

To describe the O2 production rate from a pre-cursor such as H2O2 we first use a simple but in-structive set of rate equations to describe the for-mation of H2O2 We then relate these to thesimplified rate equation for the precursor in thetext Eq 1 and show how the postulated precur-sor formation cross section sp is related to theH2O dissociation cross section sd by the reactionrates

Using the reactions

H2O 1 radiation R H 1 OH (sd)

H 1 H R H2 (k2H)

H 1 OH R H2O (kHOH)

OH 1 OH R H2O2 (k2OH)

the reaction pathways are described by

H2

Hqk2H

1 OHH2O 1 radiation mdashRsd H

OHH2O2

OH Rk2OH

kHOH

HH2O

OXIDANTS AT EUROPA 843

The corresponding rate equations in an irradiatedvolume are

dnOHdt 5 sd f n 2 2 k2OH nOH nOH

2 kHOH nH nOH (A1a)

dnHdt 5 sd f n 2 2 k2H nH nH

2 kHOH nH nOH (A2a)

dnH2O2dt 5 k2OH nOH nOH

2 destruction processes (A3a)

Here sd is the dissociation cross section f is theradiation flux and n is the molecular numberdensity of ice with nOH and nH the density of dis-sociation products OH and H respectively Therate constants k2OH k2H and kHOH describe thereactions above In this model the H2 formed isassumed to escape for temperatures above about20K We do not distinguish between trappedspecies and mobile species although in fact thatdistinction is important as discussed in the nextsection That is any OH produced at low tem-perature by one ion would likely be trapped priorto interacting with an OH produced by a subse-quent ion At higher temperatures this can breakdown as discussed in the text We first relate theabove set of equations to Eq 1 in the text and thenconsider precursor destruction processes

Equation 1 in the text is obtained by solvingthese equations together and then by integratingover a depth Dx Here Dx is the smaller of the pen-etration depth of the particle and the O2 escape(percolation) depth Assuming the molecular den-sity of ice n is constant with depth Dx n 5 N inEq 1 Comparing Eq 1 with Eq A3a the precur-sor formation rate sp f N is equivalent to [k2OH

nOH nOH] Dx Assuming that the processes in EqsA1a and A2a are fast then the steady state is read-ily obtained In addition if k2H k2OH there ispreferential loss of H2 at short times as suggestedby the laboratory data If steady state is reachedthe production of H2 is equal to that of H2O2 us-ing only the equations above Further one obtains

[k2OH nOH nOH] Dx R [fp sd] f N (A4)

with

fp 5 05[1 1 05 kHOH(k2H k2OH)05] (A5a)

Therefore the precursor formation cross sectionsp in Eq 1 is a fraction fp of the total dissocia-

tion cross sd Since laboratory data suggest theH2O2 fraction in an irradiated ice is small thenkHOH (k2OH k2H)05 which is likely givingfp R [(k2OH k2H)05kHOH]

Trapped versus mobile species

To examine the fluence dependence discussedin the text we note that trapped OH and freshlyproduced transiently mobile OH can be treatedas different species (eg Matich et al 1993)Therefore assuming no trapping of H the OHrate equation above would be replaced by twoequations one for the trapped species nOHt andthe other for the freshly produced mobile speciesnOH Assuming the reaction of two mobile OHunits is unlikely and that peroxide is formed fromthe interaction of a trapped OH with a subse-quently produced OH an expanded set of equa-tions is obtained

OH 1 trap R OHt (kOHt)

OH 1 OHt R H2O2 (kOHOHt)

H 1 OHt R H2O (kHOHt)

then the reaction pathways are described by

qH2

H2O 1 radiation mdashRsdH

qk2H

1 OHH

trap OH

OH RkOHt

OHt kOHOHtH2O2

kHOHkHOHt

H H2O HH2O

The resulting set of equations is

dnOHdt 5 sd f n 2 kHOH nH nOH

2 kOHt nOH nt 2 kOHOHt nOH nOHt (A1b)

dnOHtdt 5 kOHt nOH nt 2 kOHOHt nOH nOHt

2 kHOHt nH nOHt (A1b9)

dnHdt 5 sd f n 2 2 k2H nH nH 2

kHOH nH nOH 2 kHOHt nH nOHt (A2b)

JOHNSON ET AL844

dnH2O2dt 5 kOHOHt nOH nOHt

2 destruction processes (A3b)

where nt is the trap density in the irradiated iceAt low T we can assume nt is large so thatdntdt 0 simplifying the set of equations If theprecursor of interest is trapped O as discussed inthe text then a similar set of equations can be con-structed for O and Ot as in the example to followOf course a full description over a broad rangeof temperatures would require the considerationof the both trapped and mobile H O and OHthe rate of change of trap density and the vari-ous destruction processes

The above equations should be solved simul-taneously but for the incident particle fluxes typ-ically used in these studies the diffusion of mo-bile species to traps is relatively fast This is trueeven at low temperatures if there is a significantdensity of trapping sites This is usually the casefor samples formed at low T Therefore in a lowtemperature solid at low dose rates there is not asignificant steady-state background of diffusingOH There are trapped or freshly produced OHunits which rapidly react Averaging over thebombardment rate this can be approximated bywriting dnOHdt 0 and then separating nOH outin Eq A1b Assuming also that H diffuses rapidlyand either reacts or forms H2 we also writednHdt 0 on the average Therefore betweenimpacts only species trapped at defects pores orgrain surfaces are assumed to be present as sug-gested by the lack of a dependence on the inci-dent beam flux In this limit the rate of change oftrapped OH roughly simplifies to

dnOHtdt 5 2 k2H nH nH 2 2 kOHOHt nOH nOHt

It is seen that the production of peroxide dependson the loss of H by formation of H2 as discussedand at steady state the production rate for per-oxide equals the rate of formation of H2 in theseequations

Treating the mobile and trapped species sepa-rately we can further simplify the rate equationsand obtain the precursor model used in the textThat is integrating over the relevant depth Dxthe densities are replaced by column densitiesThe rate constants ki are then related to effectivecross sections si In this way cross sections aregiven as an interaction length for each freshlyproduced mobile radical [ki nj] R [si f] The

trapped OH column density NOHt and peroxideproduction rate can be estimated from ldquolinearrdquorate equations

dNOHtdt lt [sd f] N 2 2 [s2OH f] NOHt

2 [sHOH f] NOHt (A6a)

dNH2O2dt lt [s2OH f] NOHt 2 s NH2O2 (A6b)

Here sd is again the dissociation cross sectionand the destruction processes in Eq A3b are con-tained in s The second and third terms on theright in Eq A6a are the loss of OH due to the pro-duction of H2O2 and H2O In the integration ofEq A1b9 [2 kOH OHt nOH] becomes [2 s2OH f] and[kHOHt nOH] becomes [sHOH f] Therefore thedensity and the interaction length of mobile H arecontained in sHOH If H2O2 is indeed the pre-cursor then in Eq 2 in the text [sp f N] is be-comes [s2OH f NOHt] from Eq A6b Accountingfor precursor destruction which can give backtwo OH and treating H2 can both be included inthe integrated model via the si values

The simplified set of equations above allows usto examine the relative time (fluence) scales forprecursor formation (in this case H2O2) SolvingEqs A6a and A6b the column density of OH andthe concentration of precursors versus fluence cpcan be obtained Using the column of H2O2 pro-duced as the precursor column Np in Eq 2 thenO2 yield can be obtained The results can be com-pared with the results from Eqs1 and 2 in thetext The average precursor (H2O2) concentrationover the column depth of water molecules N is

cp 5 fp (sds)[1 2 exp(2s F)] 2 (ssr)

[1 2 exp(2sr F)] (A7)

with

fp 5 2s2OH(sr 2 s) (A5b)

In these equations sr 5 (2s2OH 1 sHOH) andssr 1

It is seen that the precursor concentration withtime cp depends on two rates the reaction ratesthat determine sr and the destruction rates in sAs in Eq 4a the yield can be written as

Y 5 sO2 cp N (A8)

If sr and s are comparable then the yield at small

OXIDANTS AT EUROPA 845

fluence is quadratic in F However if sr sthen the linear dependence on fluence is obtainedwith a small shift in the fluence as shown in Eq4c in the text Further the steady-state yield isnow

Y` (T) 5 sO2 [fp sd] Ns (A9)

That is sp R [fp sd] as in Eq A4 and the tem-perature dependence for the incident electrondata must be contained in [sO2 fp] or in the ratio[sO2 s2OH]sHOH]

For comparison with their precursor modelSieger et al (1998) examined the possibility of ob-taining the O2 precursor by considering reactionsfor a density of diffusing dissociation products asin Eqs A1 and A2 rather than a trapped productinteracting with a freshly formed dissociationproduct Based on the above the lowest fluence atwhich O2 is first detected experimentally sets alimit on the time scale for mobilendashmobile reactions

Effect of trap density

Equations A1b A1b9 A2b and A3b can be ap-proximately solved assuming that the dnOHdtdnHdt and dntdt are very small on the averageas discussed above In order to examine the roleof the trap density nt which can depend on thetemperature and method of formation and on theradiation history of the ice we use the slightlysimpler precursor model In this model the O pro-duced directly by dissociation traps forming aprecursor O2 is then formed when a freshly pro-duced O or other species reacts with a trappedoxygen atom Ot

H2O 1 radiation R H2 1 O (sd)

O 1 trap R Ot (kOt)

O 1 Ot R O2 (kOOt)

The reaction pathways for the oxygen species are

H2O 1 radiation mdashRsd9

trap OO2

kOtOt kOOt

H2 1 OOHt

kOR H(kHO t)

ROR

We assume for simplicity both H2 and O2 escapefrom depth Dx after formation We also allow mo-bile H to react with Ot (kHOt) removing trappedO and both the mobile H and O to be removedby other reactants R (eg OH or contaminants)This leads to a simplified set of equations not di-rectly involving OH or peroxide

dnOdt 5 sd9 f n 2 kOt nO nt 2 kOR nO nR

2 kOOt nO nOt (A1c)

dnOtdt 5 kOt nt nO 2 kOOt nO nOt

2 kHOt nH nOt (A1c9)

dnHdt 5 sd f n 2 kHOt nH nOt

2 kHR nH nR (A2c)

dnO2dt 5 kOOt nO nOt (A3c)

These equations of course act in parallel with theequations above for OH and mobile OH may re-act with trapped O to form O2 Here we focusonly on mobile O

Since the diffusion of H and O are fast and thedensity of traps is large the nH and nt averagedover the bombardment rate can be taken veryroughly as time independent Further if we as-sume the number of trapped O nOt is small com-pared with the number of sites at which H canreact then nH [sd f nkHR nR] For a low ra-diation flux f nH is small on the average as as-sumed Since the O2 yield YO2 is equal to[(dnO2dt) Dxf] with Dx the escape depth thenthe steady-state yield Y` can be obtained ana-lytically

[Y`][sd9 N] 5 1 1 (q 2 1) 2 [(q 2 1)2

1 8 q d]054 d (A10)

q 5 [kOt (nt 1 nR)] (kHOt nH)(sd9 f n) kOOt

5(sdsd9) [kOt (nt 1 nR)] kHOt[kOOt kHR nR]

d 5 05 [1 1 nt(nt 1 nR)]

We note that for nt large q can be large Assum-ing that q 1 and nt nR then Eq A10 be-comes

Y` R [sd9 N]q 5 (sd9 N) (sd9sd)

[kOOt kHR nR][kHOt kOt nt] (A5c)

JOHNSON ET AL846

Comparing with Y` in the simple precursormodel in Eq 4b then s R [(kHOt nH)f] R [sd nkHOt][kHR nR] In addition the product [sO2sp] R sd9 sd9 [kOOt n][kOt nt] That is the dis-sociation cross sections are modified by reactionrates

Quite remarkably it is seen in Eq A5c that thesteady-state yield Y` can depend inversely onthe density of traps nt in the temperature andflux regimes for which the approximations arerelevant That is if there are competing destruc-tion process for trapped O then at high trap den-sity a mobile O must find a trapped O before itbecomes trapped or before mobile H destroys theOt In this way the increasing O2 yield with in-creasing T can be related to the change in the trap-ping density nt with the formation temperatureof the ice sample As the temperature increasesand the density of traps in the bulk decreases thesurface and grain interfaces become the only trap-ping sites

ACKNOWLEDGMENTS

REJ acknowledges support by a Gledden Fel-lowship in Chemistry School of Biomedical andChemical Sciences at The University of WesternAustralia and by grants from the NSF Astron-omy Program and NASArsquos Planetary Geologyand Geophysics Program PDC acknowledgessupport by an Australian Postgraduate Award

ABBREVIATIONS

HST Hubble Space Telescope MJQ MatichndashJohnsonndashQuickenden NIMS Near Infrared Map-ping Spectrometer UV ultraviolet

REFERENCES

Ayotte P Smith RS Stevenson KP Dohnarsquolek ZKimmel GA and Kay BD (2001) Effect of porosityon the adsorption desorption trapping and release ofvolatile gases by amorphous solid water J Geophys Res106(E12) 33387ndash33392

Bahr DA Famaacute M and Baragiola RA (2001) Radiol-ysis of water ice in the outer solar system sputteringand trapping of radiation products J Geophys Res 10633285ndash33290

Baragiola RA (2003) Microporous amorphous water icefilms and astronomical implications In Water in Con-

fining Geometries edited by JP Devlin and B Buch El-sevier New York in press

Baragiola RA and Bahr DA (1998) Laboratory studiesof the optical properties and stability of oxygen onGanymede J Geophys Res 103 25865ndash25872

Baragiola RA Atteberry CL and Bahr DA (1999) Ori-gin of solid oxygen on Ganymede Reply to commentby RE Johnson on ldquoLaboratory studies of the opticalproperties and stability of oxygen on Ganymederdquo JGeophys Res E104 14183ndash14187

Baragiola RA Vidal RA Svendsen W Schou J ShiM Bahr DA and Atteberrry CL (2003) Sputteringof water ice Nucl Instrum Methods Phys Res B 209294ndash303

Bar-Nun A Herman G Laufer D and Rappaport ML(1985) Trapping and release of gases by water ice andimplications for icy bodies Icarus 63 317ndash332

Baverstock KF and Burns WG (1976) Primary pro-duction of oxygen from irradiated water as an expla-nation for decreased radiobiological oxygen enhance-ment at high LET Nature 260 316ndash318

Benit J and Brown WL (1990) Electronic sputtering ofoxygen and water molecules from thin films of waterice bombarded by MeV Ne1 ions Nucl Instrum Meth-ods Phys Rev B 46 448ndash451

Benit J Bibring J-P della-Negra S LeBeyec YMendenhall M Rocard F and Standing K (1987)Erosion of ices by ion irradiation Nucl Instrum Meth-ods Phys Rev B 1920 838ndash842

Brown WL Augustyniak WM Lanzerotti LJ John-son RE and Evatt R (1980) Linear and nonlinearprocesses in the erosion of H2O ice by fast light ionsPhys Rev Lett 43 1632ndash1635

Brown WL Augustyniak W M Simmons E Mar-cantonio KJ Lanzerotti LJ Johnson RE BoringJW Reimann CT Foti G and Pirronello V (1982)Erosion and molecular formation in condensed gasfilms by electronic energy loss of fast ions Nucl In-strum Methods B 1 307ndash314

Brown WL Augustyniak WM Marcantonio KJ Sim-mons EN Boring JW Johnson RE and ReimannCT (1984) Electronic sputtering of low-temperaturemolecular solids Nucl Instrum Methods B 1 307

Buratti B (1995) Photometry and surface structure of theicy Galilean satellites J Geophys Res 100(E9) 19061ndash19066

Buratti B and Veverka J (1983) Voyager photometry ofEuropa Icarus 55 93ndash110

Calvin WM and Spencer JR (1997) Latitudinal distri-bution of O2 on Ganymede observations with the Hub-ble Space Telescope Icarus 130 505ndash516

Calvin WM Clark RN Brown RH and Spencer JA(1995) Spectra of the icy Galilean satellites from 02 to5 mm a compilation new observations and a recentsummary J Geophys Res 100 19041ndash19048

Calvin WM Johnson RE and Spencer JA (1996) O2

on Ganymede spectral characteristics and plasma for-mation mechanisms Geophys Res Lett 23 673ndash676

Calvin WM Aninich VG and Brown RH (2004) Oxy-gen temperature and phase effects Icarus (in press)

OXIDANTS AT EUROPA 847

2001 Spencer and Calvin 2002) using ground-based telescopes and the HST These absorptionbands cannot be due to atmospheric O2 becausevery high pressures or very long line-of-sightcolumns of O2 would be required

The positions of the two signature absorptionsfor dimer O2 at 5772 nm and 6275 nm respec-tively (Landau et al 1962 Cooper et al 2003a)suggest that the O2 is locally quite dense (Calvinet al 1996 2003) Therefore these bands weresuggested to be due to inclusions of O2 trappedwithin Ganymedersquos icy surface (Calvin et al 1996Johnson and Jesser 1997) It is seen from Table 1that although the atmospheric and surface reser-voirs of O2 are coupled the oxygen content in thesurface ice is the largest

The observed O2-dimer band shapes and rela-tive band depths are distorted from solid or liq-uid O2 (Johnson 1999) At Ganymede where spa-tial resolution was obtained using the HST boththe position of the band minimum and its shapevary with latitude (Calvin and Spencer 1997)Therefore care must be taken when comparingthe peak positions of the bands seen in the dis-covery spectrum at Ganymede with those of solidor liquid oxygen This spectrum is a hemispher-ical average of spectral features that vary in theirpeak positions and band shapes It was obtainedby ratioing Ganymedersquos reflectance to that of Cal-listo (Spencer et al 1995) which we now knowexhibits a weak solid-like O2 feature It was re-cently shown that O2 is also trapped in Europarsquosicy surface (Spencer and Klesman 2001 Spencerand Calvin 2002) However the solid-like O2 fea-ture at Europa is much weaker than that atGanymede (5772 nm band depth 03 vs18) This could be due to the fact that Eu-roparsquos surface scatters light more efficiently in thevisible leading to shorter path lengths (Johnsonand Jesser 1997) It could also be due to compet-

ing chemical processes (Calvin et al 1996 John-son and Quickenden 1997)

Ozone and peroxide

Related to the oxygen observations the Hartleyband of O3 has been tentatively identified in the ul-traviolet (UV) reflectance spectra of Ganymede andin the icy satellites of Saturn (Noll et al 1996 1997)O3 trapped within the icy surface would be con-sistent with the presence of trapped oxygen ex-posed to radiolysis (Johnson and Quickenden1997) The presence of trapped O3 supports the in-terpretation that the dimer-O2 observations are dueto inclusions of oxygen trapped in the ice How-ever the shape of the O3-like feature also varies sig-nificantly with latitude at Ganymede (Hendrix etal 1999a) and is highly distorted Based on theband shape the presence of an additional absorb-ing species was inferred This was initially sug-gested to be trapped OH (028 mm) (Noll et al1996) but may more likely be due to organic mol-ecules (Johnson 2001) This O3-like feature is su-perimposed on a very broad UV absorption ex-tending from about 04 mm to shorter wavelengthsseen on all of the icy satellites The broad UV ab-sorption has been attributed to another radiolyti-cally produced oxidant H2O2 (Carlson et al 1999aHendrix et al 1999b) but may also include ab-sorption by trapped HO2 and OH (Taub and Eiben1968 Johnson and Quickenden 1997) Identifica-tion of specific radiolytic products in the UV is dif-ficult because the bands are very broad and over-lap The presence of H2O2 trapped in Europarsquos icewas confirmed by a band in the infrared observedusing the Near Infrared Mapping Spectrometer(NIMS) on the Galileo spacecraft (Carlson et al1999a) Remarkably the amount seen in the in-frared is roughly consistent with the amount sug-gested by the decrease in reflected light in the UV

OXIDANTS AT EUROPA 825

TABLE 1 GLOBAL OXIDANT INVENTORY

Reservoir Oxidant Column density (1016 molcm2)

Atmosphere1 O2 010Regolith (gas)2 O2 1025

Trapped in ice (1 mm at unit density)34 O2 50ndash500H2O2 400

1Based on Hall et al (1995 1998)2Estimated using average regolith depth of 1 m3Europa band depth approximately one-sixth that in Calvin et al (1996) who give 01ndash1 concentration on

Ganymede4For comparison sulfur species on the trailing hemisphere [SO2] 300 [sulfate] or 1000 3 1016 molcm2

Whereas H2O2 is observed in ice at Europa at013 level the O3-like UV feature seen atGanymede is not obvious at Europa (eg Fanaleet al 1999) This again may be due to the fact thatless trapped O2 is seen in reflectance at Europathan at Ganymede That is since O2 is a precur-sor of O3 the factor of 6 reduction in O2 meansthat the identification of this broad feature su-perimposed on other broad UV bands is prob-lematic However the absence of the O3-like fea-ture could also be due to the presence of sulfurspecies that strongly compete for excess oxygen(Johnson and Quickenden 1997)

Summary space observations

Although each of the individual identificationsof suggested products of water radiolysis or pho-tolysis (OH H2O2 O2 O3) has some uncertaintythe reflectance spectra taken as a whole give afairly clear self-consistent picture The irradiatedsurfaces of the icy satellites contain trapped mol-ecules that would be expected from radiolysis orphotolysis of an icy surface (Johnson and Quick-enden 1997) Indeed OH is a suggested precur-sor of H2O2 which is a suggested precursor ofO2 which in turn is a precursor of O3 Howeverthe separation into amounts of individual speciesbased on the absorption in the UV identificationsremains difficult Therefore the observed band oftrapped H2O2 in the infrared and the pair ofbands in the visible of a condensed form of O2are critical as they confirm the presence of radi-olytically produced oxidizing molecules in Eu-roparsquos icy surface

Consistent with the above oxygen-rich mole-cules derived from the radiation processing of anice containing sulfur (SO2 and a sulfate) and car-bon (CO2 and a carbonate) have been identified(see eg Johnson et al 2003) The sulfate hasbeen suggested to be a hydrated salt mineral (Mc-Cord et al 1999) or a hydrated sulfuric acid (Carl-son et al 1999b 2002) The latter is an efficientoxidant that is also refractory Together these ob-servations indicate that volatile and refractory ox-idizing molecules are readily formed by radioly-sis and photolysis in Europarsquos surface material Ifindeed that material can be delivered to Europarsquosputative subsurface ocean a source of free energymay be available for potential biological activityHere we consider the formation of molecular ox-idants by radiolysis with emphasis on O2

RADIOLYTIC PRODUCTIONLABORATORY DATA

Introduction

In this section we review the laboratory dataleading to the production of oxygen and relatedspecies by radiolysis and photolysis of ice Wefirst compare the data for production of oxygenby incident ions electrons and UV photons Thisis followed by a review of the data on the loss ofhydrogen which is the other principal product ofradiation-induced decomposition of ice We thenconsider the evidence that certain proposedspecies such as peroxide or trapped O might beprecursors to molecular oxygen formation in iceSome of these data have been summarized ear-lier but with a very different emphasis Here thegoal is to motivate the development of chemicalmodels for describing irradiated ice models thatcan be used to suggest new experiments and tointerpret data Individuals familiar with the datacan go directly to the modeling section in whichrate equations are developed and used to inter-pret aspects of the diverse database

Molecular oxygen

Gas-phase O2 has been produced by irradiat-ing low temperature ice using UV photons low-energy (eV) electrons keV ions and electrons andMeV ions (see eg reviews by Johnson andQuickenden 1997 Delitsky and Lane 1998 John-son 1990 1998 2001 Strazzulla 1998 Baragiola2003 Johnson et al 2003) In spite of the consid-erable differences in excitation density and pen-etration depth for these incident particles the re-sults exhibit important similarities Here wesummarize the principal findings We first notethat the efficiency of this process in pure ice islikely to be lower than estimates based on veryearly measurements of the production of O2 in ice(see eg Hart and Platzman 1961) because oflow levels of contaminants in those early experi-ments (Baragiola and Bahr 1998 Moore andHudson 2000) Therefore trace species can be im-portant in determining the O2 production ratesfrom Europarsquos icy surface and will be discussedshortly

In studying the ability of energetic plasma par-ticles to eject (sputter) molecules from ice aprocess occurring on small Outer Solar Systembodies Brown Lanzerotti and co-workers (Brown

JOHNSON ET AL826

et al 1982 1984) found that at temperatures rel-evant to the surface of Europa H2 and O2 wereimportant components of the ejecta They thenused those data to predict the presence of a radi-olytically generated O2 atmosphere on Europa(Johnson et al 1982) The production efficiencyfor gas-phase O2 was found to be strongly tem-perature dependent as seen in Fig 2a for MeVHe1 ions incident on ice and Fig 2b for keV Ar1

ions on ice In Fig 2 the production rate is givenas a yield which is the number of moleculesejected per incident ion (ie O2 production ratedivided by the flux of incident particles photonselectrons or ions) Yields are typically measuredat a low incident particle flux so that sample heat-ing does not occur The yield versus temperatureshown in Fig 2 appears to increase exponen-tially with T over a range of temperatures Al-though this is not an Arrhenius dependencerough activation energies can be extracted overnarrow ranges of T (eg Fig 2a inset) Theseare of the order of 001ndash007 eV (Table 2) Suchvalues are much smaller than those for diffusionof vacancies or trapped species in ice (Livingstonet al 2002) but are comparable to the reportedactivation energies for H1 or H diffusion in iceas noted earlier (Johnson and Quickenden1997) The lowest energies are also comparableto the enthalpy difference between amorphoussolid water and crystalline ice (Ayotte et al2001) Therefore the estimated energies are likely

related to H1 and H transport or to hydrogenbond re-orientation It has also been shown thatas the temperature is increased above 80ndash100Kmorphological changes occur that can affect thereaction kinetics and give incorrect activation en-ergies This is especially so when using measure-ments of average quantities such as yields ratherthan directly extracting rate constants by mea-suring transients (Selby 2003) In addition the ac-tivation energies can be representative of the com-peting (destruction) processes rather than theformation processes (eg Matich et al 1993) asdiscussed in the model section

The measured O2 production rates (yields)were also found to be fluence dependent wherefluence is the radiation flux 3 the time or the to-tal number of particles per unit area onto the iceMultiplying the fluence by the incident particleenergy and dividing by the average penetrationdepth gives the energy per unit volume oftencalled an average dose As seen in Fig 3 the O2yield versus fluence rises from zero implyingthat a single ion incident on a fresh surface doesnot efficiently produce and eject oxygen This isthe case even for an incident 15 MeV Ne1 forwhich the excitation density along the ion path ishigh However it is seen in Fig 3 that with in-creasing irradiation time individual impactingions produce and eject O2 with increasing effi-ciency as the material is chemically altered Atlarge fluences the yield appears to saturate indi-

OXIDANTS AT EUROPA 827

FIG 2 a Yield for D2 and O2 at saturation (steady state) from 15 MeV He1 on D2O ice versus the temperature ofthe ice (adapted from Brown et al 1982) Inset D2 yield minus the constant yield at low T versus 1kT giving an av-erage activation energy of 003 eV b Yield of O2 at saturation versus temperature for incident 30 keV Ar1 ions(adapted from Baragiola et al 2003)

a b

cating that destruction processes are in a roughsteady state with formation processes The yieldsshown in Fig 2 are purportedly those after satu-ration although full saturation is often ap-proached slowly

The slope of the O2 yield versus fluence shownin Fig 3 increases with increasing temperaturegiving the activation energies in Table 2 Thesevalues are similar to those obtained from Fig 2For energetic ions that fully penetrate the film asin Fig 2a and 3a the yield versus fluence can ex-hibit a second rise at higher fluences for whichthe radiation track cores begin to overlap This in-crease also depends on temperature and exhibitssimilar activation energies (Reimann et al 1984)The strong dependence on temperature is at firstsurprising since the energy density deposited bya 15 MeV Ne1 produces transient temperaturesin the track core that are much higher than theambient material temperature Therefore tracktemperature (local excitation density) alone doesnot determine the O2 yield

For all incident particle types and energies themeasured fluence dependence is found to be sta-ble over laboratory time scales That is for ex-periments carried out at high vacuum the yieldreturns to the value it had when the beam wasturned off This indicates that the irradiated ice isldquopermanentlyrdquo altered (Reimann et al 1984)Thus a relatively stable chemical state is createdthat is favorable for O2 production This could bethe formation of a particular trapped radiationproduct in ice that acts as a stable precursor or isdue to the altered chemical and physical state ofthe irradiated ice Trapped O H2O2 and HO2(Matich et al 1993 Johnson and Quickenden1997 Sieger et al 1998) have been proposed asstable precursor species

Using isotopically labeled H2O Benit andBrown (1990) showed that considerable transportof O occurs along the ion track when an energeticheavy ion penetrates an ice sample Because ofthis the O2 yield was found to depend on sam-ple thickness Therefore for thin ice samples the

JOHNSON ET AL828

TABLE 2 OXYGEN AND PEROXIDE PRODUCTION

Radiationa Product T (K) Ea (eV) Percent at saturation G (product100 eV)b

He1(1) O2 015g(23) O2 77 (300) 0003 (0)Ne1 (30 MeV)(3) O2 300 0031He21 (6 MeV)(3) O2 300 0008He21 (29 MeV)(3) O2 300 00016H1 (80 keV)(4) O2 70 00002H1 (200 keV)(5) O2 120 00007He21 (15 MeV)(6) O2 50 (100) 003 (07ndash4) 3 1024

Ne1 (15 MeV)(7) O2 7 10ndash15c 003(8) 70ndash140 005ndash007 01

Ar1 (01 MeV)(9) O2 30ndash140 001ndash004 0006ndash008160 001

e (30ndash100 eV)(10) O2 120 002ndash003 0005ndash0007hn (102 eV)(11) O2 50 (100) 003 00001 (00003)H1 (1 MeV)(12) H2O2 16 (77) 01 (001)

1 10 CO2 77 011 10 O2 77 04

H1 (30 keV)(13) H2O2 16 (77) 022 (038) 0098 (0086)C1 (30 keV) H2O2 16 (77) 17 (22) 023 (034)N1 (30 keV) H2O2 16 (77) 16 (14) 021 (011)O1 (30 keV) H2O2 16 (77) 14 (40) 026 (015)Ar1 (30 keV) H2O2 16 (77) 22 (20) 011 (006)

aReferences are as follows 1Lefort (1995) 2Ghormley and Stewart (1956) 3Baverstock and Burns (1976) 4Baragiolaet al (1999) 5Baragiola and Bahr (1998) 6Brown et al (1982) (ion does not stop in sample) 7Benit and Brown (1990)(ion does not stop in sample) 8Reimann et al (1984) (ion does not stop in sample) 9Baragiola (2003) 10Sieger et al(1998) (steady-state value) 11Westley et al (1995) (steady-state value as no fluence dependence seen) 12Moore andHudson (2000) 13Gomis et al (2003)

bG values are for escaping O2 For H2O2 and O2 formation both of which require multiple events yields after sat-uration are used In a few cases yields are obtained by melting the sample Values in parentheses are those at anothertemperature

cPercent of film eroded as O2 molecules for long-term irradiation

ldquosaturatedrdquo yields change slowly with fluenceMore importantly O from water molecules at aconsiderable depth into the sample contributedto the O2 ejected from the surface Since there wasno observed diffusion of the O2 in ice at the am-bient temperature (7K) over the time for their experiment they concluded that the oxygentransport and production are stimulated by theincident ions The transport was suggested as be-ing due to percolationdiffusion along the dam-aged and transiently heated tracks (Reimann etal 1984 Benit and Brown 1990) They also foundthat in fully eroding (sputtering) thin ice filmsthe total amount of O2 produced required about10ndash15 of the D2O molecules in the films at T 57K On warming irradiated samples some O2

formed at depth is found to remain trapped forexperimentally relevant times (see eg Hart andPlatzman 1961 Benit and Brown 1990 Baragiolaand Bahr 1998)

Matich et al (1993) detected luminescence in77K H2O and D2O ice from newly formed and ex-cited O2 They suggested that the O2 is formed

from the interaction of a freshly produced mo-bile O interacting with a previously producedand trapped O or H2O2 They favored the formerIn matrix isolation studies of peroxide irradiatedby UV photons luminescence from newly formedexcited O2 was seen However the oxygen wasformed from mobile O (greater than 30K in Krlattice) interacting with O trapped as H2O-O(Khriachtchev et al 2000) These studies also sug-gest that trapped O possibly in the form H2O-Omay be an important precursor

In Fig 4a the O2 yield for low-energy electronsplotted as a function of electron energy exhibits anapparent ldquothresholdrdquo at 10 eV (Orlando andSieger 2003) Although the uncertainties are largethe authors suggested there may be a lower thresh-old 6 eV corresponding to a very small crosssection for O2 production (10220 cm2) This couldbe associated with an excitation at the ice surfaceBulk ice exhibits very weak absorption at photonenergies below 6 eV but exhibits a strong ab-sorption that depending on the ice structure be-gins at about 73 eV with a peak at 86 eV This

OXIDANTS AT EUROPA 829

b

a

FIG 3 a Relative yields versus fluence (ion flux 3 time) for 15 MeV Ne1 incident on D2O ice at 7K (adapted fromReimann et al 1984) Top panel D2O yield Middle panel D2 yield Bottom panel O2 yield Yields at other tem-peratures exhibited similar dependences with the slope of O2 yield versus fluence varying with T Fits give activa-tion energies between 002 eV at the lowest fluences and lowest T to 007 eV at the highest T and highest fluencesb Oxygen yield versus fluence of low-energy electrons on D2O ice (from Sieger et al 1998)

is followed by a peak in absorption at 10 eVThe results in Fig 4a show that there is a dra-matic increase in the production of O2 at 10 eVwhich is the onset of secondary electron produc-tion in the bulk (Orlando and Sieger 2003)

For incident fast ions the yield at a given tem-perature was found after a small fluence to beindependent of whether the sample was crys-talline or amorphous ice (Brown et al 1982)However using low-energy electrons the yieldswere found to be smaller for an initially crys-talline ice (Orlando and Sieger 2003) Since verylow doses of ion radiation (a few eV per mole-cule) can convert crystalline ice to amorphous iceat low temperatures (Leto and Baratta 2003) thedamage produced by the penetrating ions usedby Brown et al (1982) affected the O2 yield Thestructural damage produced may create trappingsites cause changes in the electronic relaxationpathways or enhance percolation

The temperature dependence of the O2 yield isshown in Fig 5 for electrons incident on initiallyamorphous (Fig 5a) and initially crystalline (Fig5b) ice A small decrease in efficiency is seen inboth cases at 120ndash130K where Ic forms fromamorphous ice and pore collapse occurs (Row-land et al 1991) This is followed by an increas-ing yield with increasing T and then by a more

dramatic drop near 160K where Ih forms Dur-ing the transition to Ih trapped molecules havebeen shown to escape efficiently from ice samples (Ayotte et al 2001) A large drop in the yield isalso seen in Fig 2b for incident Ar1 ions

In modeling the dramatic decreases in the yieldin Figs 2b and 5 care must be taken That is thetime for crystallization of an amorphous sample(Baragiola 2003) even allowing for the large un-certainties is 30 min at 145K and becomes 1min at 160K Therefore pulsed radiolysis exper-iments should be carried out at these tempera-tures and the transients observed Since the timefor conversion from amorphous to crystalline icecan be shorter than the experimental times at145K the results are not comparable to thosefor which the chemical state of the ice at a givenfluence is stable over experimental times Belowwe focus on the data below 140K

We have emphasized processes initiated by theelectronic excitations and ionizations producedHowever momentum transfer collisions whichare the dominant energy loss process for low-en-ergy ions also produce dissociations defects andO2 in ice (Bar-Nun et al 1985) The relative effi-ciency of momentum transfer and electronicprocesses for producing O2 needs further studyAs the penetration depths are much smaller when

JOHNSON ET AL830

FIG 4 a Relative yield of O2 from D2O ice versus incident electron energy E at T 5 120K (from Orlando et al1998 Orlando and Sieger 2003) b Relative yield of D2 from D2O ice versus incident energy E at T 5 88K (fromKimmel et al 1994)

a

b

the former dominates the produced O2 likely es-capes readily

H2 formation and loss correlation with O2

The production of O2 from a low temperatureice by radiolysis correlates with the formationand loss of H2 (or D2 from D2O ice) (Reimann etal 1984) D2O is used to reduce the backgroundin mass spectrometry studies The yields of O2from D2O and H2O ices were shown to be com-parable However the total yields from D2O wereabout 10 less (Reimann et al 1984) consistentwith H (D) transport being important It is seenin Fig 3 that D2 is promptly produced and lostfrom the D2O ice At higher fluence the D2 yieldis seen to exhibit a dependence like that for O2For the 15 MeV Ne1 ions which fully penetratethe ice sample the D2 yield eventually decreasessince D2 is produced and escapes from the fullthickness of the film

Because of H2 (D2) loss the surface region of

the irradiated ice becomes depleted in H (D) rel-ative to O favoring the formation of oxygen-richspecies The change seen in the ratio of H (D) toO in the irradiated ice is small An irradiated film(100 nm) remains stoichiometric to better than2 (Brown et al 1980) and in the near surface re-gion Auger spectroscopy suggests only a slightoxygen enhancement due to irradiation by keVelectrons (Rye et al 1978) Sieger et al (1998) sug-gested a precursor concentration of the order ofa percent Such concentrations are an order ofmagnitude smaller than the steady-state concen-tration of dissociated H2O (Table 2)

Using infrared spectroscopy the damaged H2Oconcentrations at low temperatures in the satu-ration regime are about 10 over the irradiatedvolume (Benit et al 1987 Watanabe et al 2000Gomis et al 2003) This fraction is the ratio of thedissociation cross section to the effective crosssection for recombination to H2O as discussedlater At a 10 concentration most H2O will havea neighbor that is dissociated

OXIDANTS AT EUROPA 831

FIG 5 a Relative steady-state yields of O2 from a D2O ice sample versus T (upper panel) The ice samples wereformed and irradiated at these temperatures Film erosion as a yield (lower panel) (from Orlando and Sieger 2003)b Relative steady-state yields of O2 versus T as in (a) (upper panel) Here the D2O ice samples are formed at 160Kpresumably becoming crystalline and then irradiated at the T shown A new sample is used for each data point in(a) and (b) Film erosion as a yield (lower panel) (from Orlando and Sieger 2003)

At temperatures above 40K it is seen in Fig2 that the D2 and O2 yields exhibit the same tem-perature dependence suggesting a correlationbetween D2 and O2 loss H2 (D2) can be formedby reactions between diffusing hydrogen fromdissociated water or directly [ie H2O R H2 1O (eg Kimmel and Orlando 1995) H2O R (H1 OH)cage R H2 1 O (eg Matich et al 1993)H3O1 1 e R H2 1 OH (eg Williams et al1996)] Subtraction of the nearly constant D2 yieldin Fig 2a gives the Arrhenius plot in the insetSince molecular hydrogen readily escapes atthese temperature it is the H2 (D2) formationprocess that has an activation energy of 003eVmolecule The fact that slopes of the yields forH2 and O2 above 40K are similar suggests thatH2 (D2) formation and loss is the rate-limitingstep in determining the production and loss of O2at these temperatures

Although the data at low T in Fig 2a have con-siderable uncertainties below about 20ndash40K theD2 production is roughly independent of tem-perature but the O2 yield continues to decreasewith decreasing temperature Radiolytically pro-duced D2 is mobile and can escape above15ndash20K (Watanabe et al 2000) but O2 is im-mobile at such temperatures and the incidentHe1 ions are much less efficient at causing per-colation than the incident Ne1 discussed earlier

The production of H2 (D2) from ice has alsobeen studied using low-energy electrons and UVphotons at very low temperatures In the formerstudies absolute yields were not obtained butthere appeared to be a threshold at 63 eV (Fig4b) This is roughly the threshold for excitationof a surface state (Khusnatdinov and Petrenko1992) and the threshold for peroxide productionis close to this value The efficiency for H2 (D2)production increases around 11 eV with a sec-ond steep rise at 17 eV (Kimmel et al 1994) TheO2 and D2 yields versus electron energy in Fig 4exhibit some similarity in the threshold regionalthough the uncertainties are large and the sam-ple temperatures different Measuring the H2 (D2)and the O2 ejecta simultaneously over a range oftemperatures as in the Ne1 experiments wouldbe instructive

Using 98 eV photons at low T (12K) Watanabeet al (2000) measured the D2 ejected and trappedin ice They obtained a cross section for produc-tion of D2 (Table 2) at low fluences that is10ndash20 of the total absorption cross sectionwith significant uncertainties They also found

that the equilibrium damage fraction in the sam-ple was 10ndash20 of the D2O molecules but theequilibrium production of D2 was about an orderof magnitude smaller 1ndash2 at their tempera-tures They suggested that the D2 is formed atgrain interfaces and cracks and on the surfaces ofpores In a mixed ice D2O 1 CO exposed to UVphotolysis D2 and CO2 formation are correlated(Watanabe and Kouchi 2002) The D2O dissoci-ates to produce D2 plus O and this transientlymobile O reacts with CO to produce CO2

Precursor state and peroxide

In order to produce O2 from low temperatureice it appears that the ice sample must lose hy-drogen and form trapped oxygen-rich reactivespecies The efficiency of production also appearsto be affected by the level of defects and damagein the ice consistent with the suggestion that thehydrogen-bonding network is important in theproduction of O2 via availability of trapping sitesnew electronic relaxation pathways or the natureof the bonding of precursors Here we briefly ex-amine the possible role of the suggested precur-sor molecules H2O2 and HO2 We note that H2O2might not readily dissolve in ice possibly ac-counting for its low abundance in some experi-ments It appears to be heterogeneously distrib-uted in the form of microcrystals (Gurman et al1967) If H2O2 does act as a precursor the for-mation of aggregates could affect the dimer ab-sorption bands (Cooper et al 2003a) and affectthe production and stability of O2 (Cooper et al2003b)

The results of Gerakines et al (1996) in whichice at 10K is irradiated by UV photons confirmthat H2O2 forms in ice from two OH after a build-up of trapped OH They monitored the band as-sociated with H2O-HO (Langford et al 2000) inwhich the HO is bound to water with 024 eVTherefore they might only be monitoring a sub-set of the trapped OH seen in earlier experiments(eg Taub and Eiben 1968) The H2O2 concen-tration in ice reaches a steady-state level at flu-ences of about 1018 photonscm2 (Gerakines et al1996) As seen in the model section this gives anestimate of the H2O2 photodestruction cross sec-tion of the order of 10218 cm2 Similarly HO2 be-gins to be seen in absorption at about a few times1017cm2 shortly after they begin to detect H2O2The absolute yields were not measured

Westley et al (1995) detected H2 and O2 pro-

JOHNSON ET AL832

duced by Lyman-a (98 eV) radiation of ice butfound no fluence dependence for fluences greaterthan 1017cm2 They also saw small amounts ofH2O2 by programmed thermal desorption of thesample but made no quantitative evaluationComparing their lowest fluence for the produc-tion of O2 with the result of Gerakines et al (1996)for peroxide suggests that H2O2 and HO2 areprobably not important precursors of O2 In UVphotolysis of H2O2 using 64 eV photons the pro-duction of O2 was not observed (Vaghjiani et al1992) and the excitation of an irradiated icewhich presumably contained some peroxide by37 eV photons did not produce O2 (Baragiola etal 2003) Direct excitation of matrix-isolatedH2O2 at low temperature also did not yield O2but the production of a mobile O that reacted withtrapped O (H2O-O) did give O2 (Khriachtchev etal 2000)

Irradiating with 200 keV protons at tempera-tures of 40ndash129K Bahr et al (2001) found onlyabout 0001ndash002 H2O2 using programmed ther-mal desorption as compared with earlier experi-ments that suggested larger values (eg Table 2)Using infrared absorption Moore and Hudson(2000) found a G value of 01 for incident MeVprotons at low temperature but the signal in theobserved band became negligible at 80K sug-gesting that H2O2 was not being formed Theyalso found that adding O2 significantly increasedthe production efficiency of H2O2 That is thatthe amount of peroxide was affected by theamount of trapped oxygen Based on the matrixisolation studies (Khriachtchev et al 2000) dis-sociated mobile O might trap as H2O-O whichon excitation converts to H2O2 Therefore even ifH2O2 is not the precursor the peroxide andtrapped oxygen concentrations will be related

Although molecular oxygen was not a productof photolysis of matrix isolated H2O2 it was aproduct of photolysis at 266 nm (466 eV) of per-oxide dimers in an argon matrix 2 H2O2 R 2 H2O1 O2 (Engdahl and Nelander 2002) This de-composition was thought to be due to the con-siderable distortion of the bonds in the dimerTherefore the configuration or distortion by thelattice of peroxide or any other precursor mole-cule in the ice may be critical (Sieger et al 1998)In addition more recent experiments show thatG values for the production of H2O2 by 30 keVincident ions are significant as are the steady-state ratios of H2O2H2O (Gomis et al 2004) Ra-tios of the order of 02ndash4 were found at 16K and

77K in the irradiated regions with incident H giv-ing the smallest fraction and incident O thelargest (Table 2) At these concentrations perox-ide molecules might segregate forming inclu-sions (Gurman et al 1967 Cooper et al 2003b)Since the amount of trapped peroxide also com-petes with the amount of trapped O studies areneeded to describe the chemical state of the solidproduced by radiolysis This will be critical in de-scribing the fate of the irradiation products at Eu-ropa In particular the chemical composition fa-vorable for producing O2 needs to be determinedas does the importance of trapped O H2O2 HO2or peroxide dimers as precursors

Other species

In the model below we will refer to atomic andmolecular species trapped in ice Such species areoften discussed in the literature on radiolysis andphotolysis (for summaries see Matich et al 1993Johnson and Quickenden 1997) Trapped OH OH2O2 and HO2 have been discussed andor mea-sured by a variety of techniques (absorption andluminescence spectroscopy electron spin reso-nance programmed thermal desorption etc)Care must be taken when using these data as re-sults for ice samples formed below 160K orthose exposed to penetrating ion beams likelyhave numerous defects voids and grain inter-faces so that a variety of trapping sites and trapdepths exist Therefore when a particular bandfor a trapped species (eg OH) disappears withincreasing temperature it could be due to achange in the nature of the trapping as well as toloss of the species by reaction

Although H is discussed below we do not dis-tinguish it in these simple models from H1

trapped as the hydronium ion H3O1 whichwould mean that there are corresponding nega-tive ions present Therefore below we refer to thevarious species as if they are neutral althoughthis might not be the case Although trapped elec-trons are detected only over a narrow range oftemperatures (100ndash120K) negative ions must beimportant over a broad range of temperature

We also note that SO2 and a sulfate and CO2and a carbonate appear to be trapped species inEuroparsquos icy surface Under radiolysis each ofthese can dissociate directly giving O2 Howeverwith the exception of the sulfate their densitiesare less than 1 so they are not relevant sourcesof O2 However their presence is indicative of the

OXIDANTS AT EUROPA 833

preference for forming oxygen-rich molecules inEuroparsquos ice and can affect the efficiency of pro-duction of O2 and H2O2

MODELS FOR FORMATION OF O2

Introduction

Based on the data described above the yieldsfor incident electrons and ions are found to benearly linear in fluence at the lowest fluences forwhich O2 is detected In addition the relevant al-terations of the ice samples by radiolysis appearto be stable and independent of the beam fluxRemarkably this is the case both for low-energyelectrons that make only a single excitation foreach impact and fast ions that make a density ofexcitations along their path through the solidThis suggests that O2 is produced and trapped inthe initial events and then caused to escape fromthe solid by the subsequent radiation or that astable chemical alteration of the ice is required be-fore the radiation can produce and eject O2

Because H produced by the dissociation of wa-ter molecules can attack trapped O2 and any pre-cursor oxidants the loss of H is important Thisoccurs by the production and loss of hydrogenmolecules as discussed leaving a surface that isslightly rich in O relative to H The O2 produc-tion mechanism could be direct excitation of aparticular precursor species or it might not bespecific That is a new OH or O produced by theincident radiation could react with the H-depleted solid producing O2 This would occurmore efficiently as the ratio of O to H in the nearsurface region increased

In their early work WL Brown and colleaguesassumed that processes similar to those used todescribe radiolysis of liquid and gaseous H2Ocould be used to describe the radiolysis in ice Thelow activation energy that they found was as-sumed to be associated with diffusion of somemobile radical species in ice such as H or O orpossibly diffusion of the O2 molecule itselfBased on their studies of ejection from isotopi-cally labeled D2O samples they concluded thatthe O2 is formed in the track of a single ion andthen ejected by subsequent ions (eg Benit andBrown 1990) More recently trapped specieshave been suggested as precursors These are ei-ther subsequently excited by the incident radia-tion or attacked by newly produced mobile

species We consider a simple model that incor-porates all of these processes and can be used toorganize and interpret the experimental data Wealso discuss its relationship to more completemodels In this discussion much of which is in theAppendix we consider the standard radiationproducts that have been often described How-ever in a solid the excited electron cloud pro-duced by absorption can overlap a number of lat-tice sites Therefore their relaxation pathwaysandor product species could be very differentfrom those that have been traditionally consideredin the literature on radiolysis and photolysis

Precursor model

The column density of a precursor species Npformed in ice by incident radiation can be calcu-lated using a simple model In this model cross sec-tions are used for the production sp and destruc-tion s of the precursor O2 is then formed byexcitation of the precursor or by reaction of a newlyformed radical with the precursor This is also de-scribed using a cross section sO2 For incident elec-trons having a small penetration depth the O2 pro-duced is assumed to rapidly diffuse to the surfaceand escape (Sieger et al 1998) More generally wedefine N as the column density of H2O from whichthe incident radiation can produce and eject O2 Apair of rate equations gives both Np and the pro-duction rate of oxygen dNO2dt That is

H2O 1 radiation R precursor

precursor 1 radiation R O2

giving

dNpdt 5 sp f N 2 s f Np (1)

dNO2dt 5 sO2 f Np (2)

In Eq 1 f is the incident flux and the total pre-cursor destruction cross section can be written ass 5 sp9 1 sO2 with sp9 accounting for precursordestruction processes other than the formation ofO2 Solving Eqs 1 and 2 and assuming N is nearlyconstant (10 level as discussed) the fractionof precursors cp in the affected column after timet is obtained (Appendix)

cp 5 NpN 5 [sps] [1 2 exp(2 s F)] (3)

Here F 5 [ft] is the fluence (flux 3 time) which

JOHNSON ET AL834

is the number of particles (photons electrons orions) incident on the surface per unit area Usingcp in Eq 2 the O2 yield Y at any fluence can bewritten in terms of the yield at steady state Y`

Y 5 [dNO2dt]f 5 sO2 cp N 5 Y`

[1 2 exp(2sF)] (4a)

Y` 5 [sp sO2 s] N (4b)

If we also account for the reduction in the den-sity of water molecules in the initial column Nthese forms change only slightly The size of thetwo parameters Y` and s can be obtained by fit-ting the yield versus fluence data such as that inFig 3

For the low-energy electron data Sieger et al(1998) showed that the ratio [YY`] for a givenincident electron energy is independent of tem-perature over a narrow range of temperatureTherefore the destruction cross section s in Eq4a is roughly independent of T and can be esti-mated from the [YY`] curve Sieger et al (1998)found a precursor destruction cross section s lt08ndash30 3 10216 cm2 for incident electron energiesfrom 30 to 100 eV This is of the order of the elec-tron impact ionization cross section for H2O

Based on the above model the lack of an ob-served fluence dependence for fluences down to1017 hncm2 in the Lyman-a irradiation of ice(Westley et al 1995) can be examined The low-est fluence used implies that the precursor de-struction cross section for Lyman-a is of the or-der of the Lyman-a absorption cross section in ice(09 3 10217 cm2)

Sieger et al (1998) also assumed that sp9 5 0This would mean that the precursor is only de-stroyed by production of O2 (ie s 5 sO2) Thatis unlikely since suggested precursors can be de-stroyed by reaction with H and other radicalsTheir assumption would also mean that the crosssection for production of O2 from a precursorsO2 is large This would be surprising and ap-pear to contradict some of the data discussedabove Instead we assume that s is not equal tosO2 and the total destruction cross section s isof the order of the ionization or absorption crosssection

Having obtained s the measured temperaturedependence for production of O2 is containedonly in Y` Over a narrow range of T one canwrite Y`(T) lt Y`9 exp(2EakT) and obtain the es-

timate of the activation energies in Table 2 Ifsp9 THORN 0 as discussed then sp and sO22 in Eq 4bcannot be obtained separately from the data forY` unless an independent measure of cp is ob-tained Such a measurement is critical for de-scribing the chemical state of the irradiated sur-faces

The observed temperature dependence of Y` iscontained in sp sO2 or N Here N is the ldquodepthrdquofrom which newly formed O2 can reach the sur-face and escape given as a column density of H2OThe physical depth is obtained by dividing thecolumn density N by the molecular number den-sity of the ice Assuming that in the low-energyelectron experiments N is independent of tem-perature and is about three monolayers (Sieger etal 1998) the measured Y` gives [sO2sps]05ndash20 3 10218 cm2 over the electron impactenergy range At 110ndash120K this ratio was foundto be proportional to (E 2 Eo) for E Eo 10 eVagain assuming fixed N (Orlando et al 1998 Or-lando and Sieger 2003)

In the following section we first consider a pre-cursor like peroxide that typically forms fromtwo dissociation events We use this to show un-der what circumstances such a model does notviolate the observed linearity of the O2 yield datawith fluence Following that we relate the sim-ple precursor model above to the case wheretrapped O is the precursor We then use themodel for trapped O to examine the effect of thetrap density on the production of O2 The densityof traps (defects) is determined by the sample for-mation temperature and the subsequent radiationdamage Finally we consider the escape depthN for penetrating radiation within the precursormodel These discussions are supported by moredetailed rate equations in the Appendix

Precursor trapped peroxide

The proposed precursor peroxide is typicallyformed from products of two dissociation eventsie 2 H2O R 2 H 1 2OH R H2 1 H2O2 This wasconfirmed in low temperature ice (Gerakines etal 1996) It is also possible that a dissociation pro-ducing a trapped O followed by an excitationevent can produce peroxide 2 (H2O 1 hn) R H2

1 H2O-O 1 hn R H2O2 1 H2 (eg Khriachtchevet al 2000) The O2 yield versus fluence for inci-dent electrons is reasonably well fit by the yieldexpression above for fluences greater than1014cm2 Therefore it has been suggested that

OXIDANTS AT EUROPA 835

any precursor species must be formed in a singleevent Using a model in the Appendix in whichthe precursor is formed by two events we showthis is not necessarily the case Precursor forma-tion if it is fast could involve two dissociationevents and give the measured fluence depen-dence Although we consider peroxide here theseconclusions apply to other possible precursors

In the example in the Appendix H2O is disso-ciated to H 1 OH according to the cross sectionsd The reactions of two OH units to form H2O2(rate constant k2OH) and 2 H to form H2 (k2H)compete with the recombination of H and OH toH2O (kHOH) If k2H is much greater than k2OH then H2 loss dominates at low fluences as ex-pected (Appendix) Steady state requires not sur-prisingly that the rate for forming H2O2 isroughly equal to the rate for forming H2 Finallyit is seen in the Appendix that the low H2O2 con-centrations suggested by experiment imply thatkHOH (k2H k2OH)05 Such sizes are reasonable

The set of rate equations in the Appendix thatuse number densities of species is integrated overdepth in order to obtain the forms used in Eqs 1and 2 which are based on column densities Inthis way the effective cross sections are related tothe rate constants for the interaction of freshlyproduced mobile species with previously pro-duced trapped species We show in the Appen-dix that the O2 yield versus fluence can be writ-ten in a form similar to that in Eq 4a but with asmall offset due to the two-step precursor for-mation process

Y lt Y`(T) [1 2 exp(2sF) 2 (ssr)]

for F sr21 (4c)

Here s is again the precursor destruction crosssection but sr 5 (2s2OH 1 sHOH) is the effectivecross section for loss of OH by formation of H2O2

and H2O The corresponding expression for Y`(T)is given in the Appendix

The form for the yield versus fluence in Eq 4cis consistent with the low-energy electron mea-surements if (ssr) 1 and if sr F21 lt10214 cm2 Such sizes are indeed possible Whatis required is that at the relevant temperatures(50ndash110K) the interaction length associated withsr is more than a few molecular layers (greaterthan 10 Aring) A cross section that is likely to belarge above 20K is sHOH which describes themean distance for diffusion of mobile H before it

reacts Since s discussed earlier is of the sameorder of magnitude as the H2O dissociation crosssection sd then the precursor concentration cp(in this case H2O2) is roughly proportional to[s2OHsHOH] 1 (Appendix)

Based on the above a two-step precursor for-mation process could be consistent with the elec-tron data contrary to the statements in Sieger etal (1998) Orlando et al (1998) and Orlando andSieger (2003) Experiments are needed to deter-mine the relevant cross sections Below we con-sider a precursor formed in a single excitationevent

Precursor trapped O

O2 can be formed for instance by diffusion ofa transiently mobile O to a previously formed andtrapped O or to the production of O near the siteof a previously produced and trapped O (Matichet al 1993 Johnson and Quickenden 1997) Werefer to this as the MatichndashJohnsonndashQuickenden(MJQ) model and note that O 1 OH R O2 1 Hcan also give molecular oxygen

Trapped O as H2O-O was in fact seen to be aprecursor to O2 when attacked by a mobile O ina Kr matrix (Khriachtchev et al 2000) ThereforeO trapped as H2O-O was also suggested as a pre-cursor in ice (Johnson 2001) Before proceedingwe note that the examination of the MJQ modelin Sieger et al (1998) was incorrect They consid-ered two mobile interacting species whose life-times were long compared with the bombard-ment rate rather than a new mobile O interactingwith a previously trapped O Although this hasnot been corrected recently they added Otrapped as H2O-O as a possible precursor (Or-lando and Sieger 2003) O trapped as H2O-O isseen in matrix isolation studies as discussedabove but it has not yet been identified in iceand like OH it may trap in a number of ways

The MJQ model is essentially the followingFormation of H2 1 O is followed by loss of H2and trapping of O (Ot) A subsequent dissocia-tion or excitation event can lead to the destruc-tion of Ot by reaction with for instance H or byconversion to H2O2 However if this dissociationevent also leads to H2 1 O a mobile O is pro-duced that can react with Ot with an interactionlength that depends on T This results in the for-mation of O2 [O 1 Ot R O2] giving a net pro-duction of 2 H2 and 1 O2 In the Appendix therelevant rate equations (Eqs A1cndashA3c) are con-

JOHNSON ET AL836

sidered Assuming the mobile O processes arefast these rate equations can also be integratedand simplified to give the form in Eqs 1 and 2The precursor column density Np now repre-sents the column density of Ot In this model sO2in Eq 2 is the ldquocross sectionrdquo for reaction of afreshly produced O with an Ot and sp9 indicatesdestruction of Ot Finally sp is the cross sectionfor production of H2 and Ot via processes dis-cussed above

Based on the sizes of the cross sections ex-tracted in the electron irradiation experimentsthe model with trapped O (Ot) as the precursoris quantitatively reasonable That is for a rangeof T both sp and sO2 might be described byevents of the type H2O R H2 1 O (whether di-rect or indirect) If these occur for 10 of theradiation-induced dissociation events as dis-cussed then the required product of cross sec-tions suggested by the sizes of s and Y` above isroughly obtained that is [sp sO2] 02ndash6 3 10234

cm2 in the low-energy electron experiments Thecross section for production of O might dependon the availability of dangling bonds at interfacesin pores or at defects as suggested by a numberof authors The importance of such bonds couldalso be consistent with the differences in the elec-tron-induced yield for initially crystalline and ini-tially amorphous ice (Orlando and Sieger 2003)

O2 production can increase in the presence ofoxygen-rich impurities such as CO2 or SO2 bothof which are known to be present at Europa Asin the MJQ model dissociation (eg CO2 R CO1 O) can provide a mobile O that can be trappedor react with a previously trapped O forming O2In addition such species are hydrogen scav-engers (Moore and Hudson 2000) This can re-duce the destruction of Ot affecting sp as seen inEq A5c

Trap density

In the simplified models above the density oftrapping sites nt was not explicitly accountedfor However in the discussion of both the ex-perimental data and models trapping sites havebeen assumed to be important It has beendemonstrated repeatedly that the density of de-fects voids and interfaces decreases as the tem-perature of formation of the ice increases How-ever the efficiency of production of O2 is theopposite it increases with increasing T There-fore the oxygen yield varies in a manner that is

the inverse of the trap density over a range oftemperatures For penetrating ions in which theenergy density deposited is high and O2 might beformed in a single track a simple explanation hasbeen given As the density of traps for O2 is re-duced and the temperature increases the O2 pro-duced can diffuse to the surface and escape moreefficiently limiting back reactions on further ir-radiation This is examined in the next sectionHowever such a process does not affect the low-energy electron yields for which the penetrationdepth is small and the formed O2 escapes read-ily at the measurement temperatures In Eqs A1cA1c9 A3c and A10 in the Appendix we againuse a simple set of rate equations for trapped andmobile O in order to determine the role of the trapdensity nt on the production of a precursorspecies The conclusions apply to any modelbased on a trapped precursor

In the MQJ model mobile O can be trappedforming Ot (described by kOt nt) and can reactwith previously trapped O to form O2 (kOOt) Inaddition it can react at other sites and Ot can bedestroyed by mobile reactive species such as H(kHOt nH) Mobile H can also react with othertrapped species (kHR nR) Incorporating these re-actions Eq A10 in the Appendix gives a steady-state yield Y` When the density of traps is largethe steady-state yield is given in Eq A5c

Y` R (sd9sd9sd) [kOOtkHR nr][kHOt kOt nt] N (5)

Here sd and sd9 are the dissociation cross sectionsthat determine the production of H (H 1 OH) andO (H2 1 O) by the incident radiation Comparingthe result in Eq 5 with Y` in the simple precur-sor model in Eq 4b the dissociation cross sec-tions in Eq 5 are now scaled by a ratio of the re-action rates

It is seen in Eq 5 that the steady-state yield Y`depends inversely on the density of traps nt Thatis there are competing destruction processes fortrapped O (Ot) Therefore in order to form O2 amobile O competes with mobile H to find atrapped O before it becomes trapped In this waythe increasing O2 yield correlates with the re-duction in the trapping sites with increasing tem-perature of formation The small activation ener-gies discussed earlier would be related to there-orientation of the lattice binding leading to thereduction in trap density At relatively high T thedensity of traps becomes small so that only the

OXIDANTS AT EUROPA 837

sample surface or grain interfaces might act astraps accounting for the observed drop in theyield at high T For this reason it is important todetermine if the yield for T greater than 150Kdepends on the incident flux f Since other quan-tities in the expression for Y` could also be tem-perature dependent experiments are neededThese would likely be pulsed radiolysis experi-ments in which the transients are studied ratherthan the steady-state yields However the avail-able data for lightly ionizing radiation do suggestthat the O2 yield depends inversely on trap den-sity as in the model above

Escape from depth

The quantity N in Eq 1 is the ldquodepthrdquo given asa molecular column density from which O2 formedbelow the surface can percolate to the surface andescape It can be converted to a physical depth bydividing by the molecular number density of theice In the case of the low-energy electrons thedepth below the surface at which ionizations andexcitations are produced is small However forfast ions that penetrate many monolayers this isnot the case Below we examine N in Eq 1

Although it has long been understood that twoexcitations events are required to produce O2(eg Brown et al 1982 Reimann et al 1984) theprecursor model suggests that at saturation(steady state) any new event can produce O2 fromthe steady-state density of trapped precursorsTherefore one might expect the yield at satura-tion to be nearly linear in the excitation cross sec-tion This is the case for the low-energy electronsand it was initially assumed to be the case for thefast ions (Brown et al 1980 Reimann et al 1984)However Baragiola et al (2003) showed that thetotal sputtering yield in the saturation region attemperatures greater than 100K which is dom-inated by decomposition (Brown et al 1982) isproportional to the square of the electronic en-ergy deposition per unit path length in the solid(dEdx)e Assuming that the production and de-struction cross sections in Y` in Eq 4b roughlyscale with (dEdx)e then their ratio [sp sO2s]also roughly varies as (dEdx)e If this is the casethen the observed quadratic dependence impliesthat the depth N from which O2 can be mobi-lized to escape must also depend on the excita-tion density (dEdx)e This is consistent with theobservation that the O2 yield depends on thick-ness for samples thinner than the ion penetration

depth (Reimann et al 1984 Benit and Brown1990) Therefore the percolation depth for escapeN is a critical factor in comparing yields for pen-etrating radiation with those for non-penetratingradiation like the low-energy electron and low-energy ion data It is also seen that percolationand trapping can be included within the simplemodel considered here

Model summary

We have examined the simple precursor modelintroduced by Sieger et al (1998) to explain theirelectron data We have shown how it can be gen-eralized and derived from an integration of theappropriate chemical rate equations with thecross sections and total column density contain-ing the reaction rates and the physics of diffusivetransport Therefore the parameters can havevery different meanings from those initially as-sumed The equations are useful when there arefast processes which go to completion in timesmuch shorter then the time between particle im-pacts and slow processes associated with speciesthat are assumed to be trapped between particleimpacts This separation allowed us to examinethose models proposed for precursor formationand destruction due to reacting mobile speciesformed from dissociation events Further fromthe size of the onset of the linearity of the yieldversus fluence we can place constraints on therole of nonlinear processes such as those thatmight be associated with the formation of a pre-cursor like peroxide

With these extensions this simple model is use-ful for approximating the full rate equations inthe temperature region for which the yields de-pend on the fluence (dose) and not on the doserate Based on the assumptions in the model thisimplies temperatures at which precursor speciesare stable over experimental times At the high-est temperatures for the yields shown in Fig 4athis might not be the case This should be testedexperimentally In this regard it is puzzling thatexperiments have shown that the absorptionbands associated with certain trapped speciessuch as OH and trapped electrons ldquodisappearrdquoat 100ndash120K (see eg Johnson and Quickenden1997) whereas the O2 yield data indicate that rel-evant precursor species must be stable to 140KTherefore the proposed O2 production processescan only be distinguished by directly determin-ing the chemical state of the irradiated ice

JOHNSON ET AL838

It may be the case that a particular precursoris not required as mentioned earlier Rather af-ter a sufficient level of oxidation a newly pro-duced O or OH can react in the oxygen-rich lat-tice producing O2 in a pseudo-first-order processThis can be tested experimentally Indeedtrapped O and peroxide may be changed into oneanother as in the matrix isolation experiments(Khriachtchev et al 2000) Therefore their rela-tive concentrations versus temperature and doseshould be studied

In the model considered here it is seen that theparameters in Y` must account for the strong tem-perature dependence Since the number of bulktrapping sites decreases with increasing T as theice sample is annealed one might expect that theprecursor concentration would decrease We sug-gest that the observed temperature dependenceis determined by the competition between theavailability of trapping sites and mean interactiondistances of a mobile species such as O and HThat is the O2 yield varies inversely with the den-sity of traps over a range of temperatures In thesimple model above this effect is included indi-rectly as shown using the more detailed equa-tions in the Appendix

EUROPA

Oxygen and peroxide production at Europa G values

The dependence of the yield on the thicknessof the ice sample for highly penetrating ions sug-gests that the O2 yield is roughly related to theamount of energy deposited in the sample There-fore the amount of O2 produced and lost fromice in steady state is often given as a G value thenumber of molecules produced per 100 eV de-posited in Table 2 In applying these G values tothe planetary environment care must be takensince the yields are temperature fluence and ex-citation density dependent They also are affectedby the presence of other proton scavengersknown to be present in Europarsquos ice

The G values for peroxide in Table 2 are thosefor production at low fluence but the G valuesfor O2 are for the production and escape of gas-phase oxygen at saturation Therefore they de-pend on the formation and destruction cross sec-tions in the model above Such G values can beused to describe the steady-state source of the am-

bient O2 atmosphere at Europa but are not help-ful in determining the ability to produce and trapO2 at depth into the icy surface Over the timescales of the experiments some O2 is seen to re-main trapped and is not destroyed by radicalsTypically there are a number of trapping sitesand recent estimates of the activation energy foran O2 molecule trapped in a substitutional site(Hori and Hondoh 2002) suggest that it can betrapped for times comparable to the time re-quired for the conversion of amorphous ice intoIh (Ayotte et al 2001) The stability of O2 inclu-sions has not been studied

To understand the relationship between the Gvalues for O2 in Table 2 we note that fast lightions and electrons lose their energy predomi-nantly by producing ionizations and excitationsThe ionization events produce secondary elec-trons that can result in further ionizations Thenumber of ionization events Ni produced by afast light ion stopping in a water ice is Ni 5EoW where Eo is the initial energy and W is theaverage energy expended per ionization event27 eV in water ice Therefore radiolysis by fastions is often described as the net effect of theshower of electrons set in motion by the incidention In this manner the G values for productionof O2 from ice by energetic light ions and elec-trons can be roughly estimated as the sum of theeffects of all the secondary electrons using theyield Ye for 10ndash100 eV electrons That is if weignore the percolation depth N we can write thenumber of O2 produced as the number of ioniza-tion events times Ye which already has in it thedestruction processes That is G NiYeEo YeW 0006 O2100 eV

This estimate of G is seen to be an order of mag-nitude larger than the keV proton value for O2 inTable 2 but is much smaller than the G value forenergetic Ne1 These differences from the simpleestimate are consistent with the observation thatthe yield in the saturation regime for penetratingions varies nonlinearly with excitation densityThe difference from the keV proton G value islikely due to the inability of O2 to escape fromdepth before destruction On the other hand thelarger G value seen using MeV Ne1 is likely dueto the high density of excitation This can resultin more efficient production of O2 increased pro-duction of damage sites for forming O2 andorincreased transport (percolation) along the tran-siently heated track That is the ability of theformed O2 to percolate to the surface during ion

OXIDANTS AT EUROPA 839

bombardment plays an essential role in deter-mining the measured yields as discussed aboveThis effect needs to be evaluated for a planetarysurface subject to particle fluxes that are many or-ders of magnitude smaller than those used in thelaboratory

In the absence of O2 yields for all of the rele-vant ions and energies an estimate of the totalsupply of O2 can be obtained using the G valuesin Table 2 For the energetic oxygen and sulfurions we use G values for neon and argon ionsThese are lower bounds as the data were obtainedat temperatures much lower than the surface tem-peratures at Europa (80ndash110K) For the fast pen-etrating electrons and protons we use the valuefor 200 keV protons in Table 2 For the heavy ionswe used the Ne1 and Ar1 values For compari-son we also give an estimate using the G valuesobtained from the low energy electron data andthe total ionization as discussed above

The G values are combined in Table 3 with theelectron and ion energy flux at Europa Althoughthere is some deflection of the flow due to in-duced fields the net energy flux to the surface isnot reduced much from these values (Paranicaset al 2002) In fact locally produced pick-up ionswhich are formed from the ejecta and acceleratedto the surface can contribute (Ip 1996) Furtheralthough G values are typically independent ofthe incident angle the loss of O2 is a surfaceprocess and is enhanced with angle That is thenumber of O2 produced within the escape depthincreases with increasing angle to the normal(steeper than the inverse cosine) This enhance-ment is 25ndash4 (Jurac et al 2001) Ignoring suchcorrections the total production rate calculated

(23ndash62 3 109 O2cm2s) is a lower bound to theactual production rate from the icy regions of Eu-roparsquos surface Recent modeling of the O2 atmo-sphere suggest that such a source rate is required(Shematovich and Johnson 2001)

Earlier estimates of the O2 source rate were in-correct because O2 production was assumed to bea fixed fraction of the H2O sputtering yield Atthe ambient temperature an ejected H2O sticks ef-ficiently (Smith and Kay 1997) to the surface ofneighboring grains in the porous regolith reduc-ing the sputtering yield for H2O (Johnson 1989)This reduction does not apply to the O2 fractionof the yield enhancing its relative importance toH2O Here we assume that all of the gas-phaseO2 produced can contribute to the ambient at-mosphere even if it is produced at a grain sur-face that is at some depth into the porous regolith

Finally Europarsquos surface is not pure ice every-where That is much of the trailing hemisphereis a hydrated sulfuric acid containing smallamounts of sulfur and SO2 and the leading hemi-sphere contains low levels of carbon CO2 andcarbonates Sodium and potassium atoms aresputtered from the icy regions (eg Leblanc et al2002) O2 production from surfaces containingthese species has not been measured C and S areboth proton scavengers and the other species canbe sources of O as discussed so that the O2 yieldfrom such regions could be enhanced Howeverthere can also be competing processes that mightreduce these rates In fact recent HST observa-tions suggest that the O2 might not be produceduniformly across Europarsquos surface (McGrath et al2000) The analysis of these observations is pre-liminary and is not simple since the observed

JOHNSON ET AL840

TABLE 3 APPLICATION TO EUROPA

OxygenImplantation Energy flux production

Species 109(cm2 s)a 109 keV(cm2 s)(1) 109(cm2 s)

H (10 keV) 0015(1) 12 0084 (072)bO (10 keV) 0015(1) 18 054cS (10 keV) 0009(1) 30 090cElectrons 62 043 (37)bPlasma (10 keV) (137 amu) 13(2) 10 03c

Total 79 23 (62)UV photons (64 eV) 46 01

aReferences are as follows 1Cooper et al (2001) [similar results in Paranicas et al (2002)] 2Mauk et al (1996)bUsing G for 200 keV protons (upper bounds in brackets are based on the low-energy electrons data as a model for

the secondary electrons produced)cUsing G for 15 MeV Ne1 at 7K a rough lower bound

emissions depend on the local plasma propertiesHowever the efficiency of O2 production in con-taminated ices needs to be studied

Porous regolith O2 interactions on grain surfaces

Although the O2 atmosphere is principally pro-duced in the near surface layers and is eventuallylost to space it also permeates Europarsquos porousregolith The porosity p (void space) of the re-golith is uncertain (Buratti 1995) It has been es-timated to be the most compact surface in theOuter Solar System p 025 on the trailing hemi-sphere and p 079 on the leading hemisphere(Buratti and Veverka 1983) and the most poroussurface p 096 globally (Domingue et al 1991Domingue and Hapke 1992) These large differ-ences are due to the uncertainties in the amountof backscattered light at small phase angles andon the photon absorption properties and grainsizes The more compact surface is likely correctGrain sizes rg of the order of 100 mm have beensuggested (Hansen and McCord 1999 2000Geissler et al 1998) Europarsquos regolith differs sig-nificantly from the lunar regolith which is asand-like surface of individual grains Analysis ofthe thermal inertia of Europarsquos surface (Grundyet al 2001) suggests that the ice grains are sin-tered forming a fairy castle-like structure Sin-tering can occur thermally or due to the incidentradiation In addition ices doped with acids andalkalis in the laboratory can be powdery solidswith the impurities primarily found at grain in-terfaces

The porous regolith affects the above estimatesof the oxygen production in three ways First forisotropic bombardment the mean distance d intothe surface at which an incident ion or photonstrikes a grain surface is d (89)rg(1 2 p) Us-ing the above value for rg then d 02 m if theporosity is very large p 96 Therefore de-sorbed molecules can come from a considerabledepth and will interact with grain surfaces Evenfor much lower porosities the effective sputteringyields can be significantly modified (Johnson1989) As discussed above the yields for ejectedH2O which sticks onto neighboring grains at theambient temperatures are reduced but those forejected O2 are not This enhances the relative im-portance of the O2 yields

Finally the surfaces and interfaces of the sin-tered grains in Europarsquos porous surface are placeswhere low temperature chemistry involving the

ambient oxygen can occur at depths that are wellbelow the nominal penetration depth of the inci-dent radiation The relative importance of suchchemistry is roughly indicated by the ratio of thetotal surface area in the regolith to the area of Eu-roparsquos visible surface Using the porosity andgrain size above and a regolith depth dr of theorder of a meter then this ratio is large 3(1 2p)drrg which is 103ndash2 3 104 for average porosi-ties of 96 and 25 respectively Therefore thegas-phase O2 molecules below the visible surfaceinteract frequently with the grain surfaces In thisway a very large volume of the subsurface couldbe altered chemically even if it is not directly ex-posed to the radiation

From the overlying structures seen in the high-resolution images of Europarsquos surface by theGalileo spacecraft (eg Prockter and Pappalardo2000) it is clear that the youngest surfaces are red-der The reddening in the visible has been attrib-uted to the presence of sulfur in the ice (McEwen1986 Johnson et al 1988 Calvin et al 1995 Carl-son et al 1999b) The amount of reddening in thespectra of the surface material decreases with ageand the surface materials appear to becomebrighter with age This might occur by the slowaccumulation of a frost over-layer or by radioly-sis (Carlson et al 2002 Johnson et al 2003) Ra-diolysis preferentially converts the sulfur in theice to sulfur dioxide or sulfuric acid both ofwhich have flat spectra in the visible In regionswhere the radiation does not directly penetratethe O2 permeating the regolith can oxidize thesulfur at the surfaces of the icy grains and at graininterfaces Similarly the carbon seen in the ice onEuroparsquos leading hemisphere (eg Carlson et al2004 Johnson et al 2003) can be oxidized by ra-diolysis in the near surface layers or by reactionson the grain surfaces at depth in the regolith bythe ambient O2 The meteoroid bombardmentthat produces the regolith also exposes fresh sur-faces for oxidation and can cause the collapse ofpores leading to the trapping of oxygen In thismanner both refractory and volatile oxidants canbe formed and trapped in the subsurface ice forpotential delivery to Europarsquos subsurface ocean

SUMMARY

The presence of a thin oxygen atmosphere andoxidants trapped in Europarsquos surface ice (egJohnson et al 2003) combined with the possible

OXIDANTS AT EUROPA 841

presence of a subsurface ocean (eg Carr et al1998) have made Europa an exciting target forfuture space exploration The molecular oxidantsproduced by radiolysis and photolysis could bespecies needed for carbon-based biochemistry inEuroparsquos putative subsurface ocean Howeverthere has been no model for the formation trap-ping and transport of relevant oxidants

In this paper we have reviewed the space ob-servations and laboratory studies related to thechemistry of formation of oxidants in ice by theincident radiation Because the laboratory dataare incomplete we examined the rate equationsfor formation of oxygen and related species byradiolysis and photolysis We showed how suchequations can be simplified to the analytic modelthat was used for fitting the fluence and temper-ature dependence of the O2 yield produced bylow-energy electrons (Sieger et al 1998) In de-riving the analytic model from the more completerate equations the competition between forma-tion and destruction processes can be studied ascan the role of trapping sites escape from depthand temperature history These equations also al-lowed us to put incident photon electron and iondata on a similar footing for the first time

The model can be used to suggest measure-ments that are needed to describe the chemicalstate of the irradiated ice and to interpret new lab-oratory data We showed that if reactions are fastthe yield data could be consistent with a precur-sor being formed by two events as is the case forperoxide However we favor trapped O as a pre-cursor as initially suggested by Matich et al(1993) for production of excited O2 Such aspecies if present could be detected spectro-scopically as can trapped O2 through bands ac-tivated by lattice perturbations (Cooper et al2003a)

The available oxygen yield data were used tocompare G values for different incident particlesThese were then used to estimate the oxygen pro-duction in the radiation environment at EuropaThe lower bounds were consistent with the pro-duction rates obtained from a recent model of Eu-roparsquos atmosphere The porosity of the regolithwas also considered It not only affects the rela-tive sputtering yields for O2 and H2O but alsoprovides an enormously increased surface areaon which the ambient gas-phase oxygen can re-act with species in the ice That this might be oc-curring is consistent with the change in re-

flectance associated with surface aging Such re-actions could also convert ambient molecularoxygen into other more refractory species that re-main trapped on subduction of Europarsquos crust

Although radiolysis produces a stable chemi-cally altered surface in the laboratory the parti-cle flux to Europarsquos surface is many orders ofmagnitude lower Therefore annealing can be ef-fective at temperatures that are much lower thanthose typically found experimentally Belowabout 120K the annealing times appear to be long(eg Baragiola 2003) so that laboratory data canoften be directly applied to Europarsquos icy surfaceHowever account has to be taken of the syner-gism between the UV and the energetic plasmaas discussed earlier (Johnson and Quickenden1997)

Transient surface melting might initiate lifeprocesses but subduction of oxidants to the pu-tative ocean is likely to be required if biochem-istry is to occur at Europa (Chyba and Hand2001) Although the temperature versus depthinto Europarsquos regolith is not well known belowthe regolith the temperature and pressure appar-ently increase with depth At some depth (ap-proximately kilometers) either very warm ice oran ocean exists (Ruiz and Tejero 2000) Thereforechemically refractory oxidants are likely to be re-quired in order to survive subduction to thewarmer subsurface regions Trapped species ifdelivered downward can become mobile andwill segregate forming inclusions (Johnson andJesser 1997) Indeed peroxide appears to segre-gate in ice (Gurman et al 1967) and these inclu-sions may remain relatively stable until the per-oxide is dissolved in the putative ocean Thevolatiles seen in reflectance such as O2 SO2 andCO2 will also segregate But at the higher tem-peratures they could also diffuse through thebulk react or percolate into the regolith atmo-sphere Therefore the observation of the trappedvolatiles in Europarsquos surface ice is of interest pri-marily because it suggests that oxidants are pre-sent The formation of oxygen-rich molecules fol-lows the preferential loss of hydrogen Howeverthe surface is in a local steady state with im-planted H O and S from the plasma and the lossto space Since the plasma flux to the surface canbe variable this balance can also vary Indeedtemporal changes in reflectance have been re-ported (Domingue and Hendrix 2004)

Based on the models for formation of O2 ex-

JOHNSON ET AL842

amined in this paper additional experiments areneeded to determine the chemical state of irra-diated ice at low fluences Of immediate inter-est is the identification of stable molecular pre-cursors required for the formation of O2 Basedon the available laboratory data individual per-oxide molecules trapped in the ice do not appearto be the principal precursors but trapped O(Matich et al 1993) remains a viable candidateAlthough the data suggest that the local hydro-gen-bonding network is important in the pro-duction of O2 the effect of the formation tem-perature and radiation damage on the structureneed to be measured It is also important to de-scribe quantitatively the relationship betweenH2 loss and the formation of O2 and H2O2 overa range of temperatures and fluences Finallyone would like to determine the steady-statedensities of biochemically interesting moleculesformed by radiolysis in an ice containing sulfurand carbon at Europarsquos surface temperaturesand then determine their stability with increas-ing temperature and pressure Such experimentsstill need to be carried out Here we have pre-sented a model that brings together the variouslaboratory data and now can be used to considerthe formation trapping and transport of relevantoxidants at Europa

APPENDIX

Introduction

In this Appendix we examine rate equationsthat describe various aspects of the radiolysis orphotolysis of ice These equations need to be un-derstood if one is to be able to calculate the for-mation trapping and transport of oxidants Theemphasis is on the formation of molecular oxy-gen We integrate these equations over the escapedepth of O2 to give the simplified rate equationsdiscussed in the text in which number densitiesare replaced by column densities (eg Eqs 1 and2) In this way we can examine the relationshipbetween the rate constants and the effective pre-cursor formation and destruction cross sectionsextracted from data We first consider a precur-sor such as H2O2 that is formed from two exci-tation events This bears on the observed linear-ity of the oxygen yield versus fluence We thenconsider the equations for trapped versus a tran-

siently mobile species We use a reaction betweena mobile and trapped O as an example althoughO 1 OH can also give oxygen Finally we ex-amine the possible dependence of the O2 forma-tion process on the density of traps (defectsvoids and grain boundaries) as a function of tem-perature The results are discussed in the text

For clarity all possible radiolytic species arenot included in the discussion below Thereforepotentially important products such as HO2 andthe charged species (eg H3O1 the trapped elec-tron etc) are missing In each section we attemptto make a point related to production of O2 andother oxidants by using a subset of the productsrather than attempt a complete description withmany unknown rate constants

Fast versus slow processes

To describe the O2 production rate from a pre-cursor such as H2O2 we first use a simple but in-structive set of rate equations to describe the for-mation of H2O2 We then relate these to thesimplified rate equation for the precursor in thetext Eq 1 and show how the postulated precur-sor formation cross section sp is related to theH2O dissociation cross section sd by the reactionrates

Using the reactions

H2O 1 radiation R H 1 OH (sd)

H 1 H R H2 (k2H)

H 1 OH R H2O (kHOH)

OH 1 OH R H2O2 (k2OH)

the reaction pathways are described by

H2

Hqk2H

1 OHH2O 1 radiation mdashRsd H

OHH2O2

OH Rk2OH

kHOH

HH2O

OXIDANTS AT EUROPA 843

The corresponding rate equations in an irradiatedvolume are

dnOHdt 5 sd f n 2 2 k2OH nOH nOH

2 kHOH nH nOH (A1a)

dnHdt 5 sd f n 2 2 k2H nH nH

2 kHOH nH nOH (A2a)

dnH2O2dt 5 k2OH nOH nOH

2 destruction processes (A3a)

Here sd is the dissociation cross section f is theradiation flux and n is the molecular numberdensity of ice with nOH and nH the density of dis-sociation products OH and H respectively Therate constants k2OH k2H and kHOH describe thereactions above In this model the H2 formed isassumed to escape for temperatures above about20K We do not distinguish between trappedspecies and mobile species although in fact thatdistinction is important as discussed in the nextsection That is any OH produced at low tem-perature by one ion would likely be trapped priorto interacting with an OH produced by a subse-quent ion At higher temperatures this can breakdown as discussed in the text We first relate theabove set of equations to Eq 1 in the text and thenconsider precursor destruction processes

Equation 1 in the text is obtained by solvingthese equations together and then by integratingover a depth Dx Here Dx is the smaller of the pen-etration depth of the particle and the O2 escape(percolation) depth Assuming the molecular den-sity of ice n is constant with depth Dx n 5 N inEq 1 Comparing Eq 1 with Eq A3a the precur-sor formation rate sp f N is equivalent to [k2OH

nOH nOH] Dx Assuming that the processes in EqsA1a and A2a are fast then the steady state is read-ily obtained In addition if k2H k2OH there ispreferential loss of H2 at short times as suggestedby the laboratory data If steady state is reachedthe production of H2 is equal to that of H2O2 us-ing only the equations above Further one obtains

[k2OH nOH nOH] Dx R [fp sd] f N (A4)

with

fp 5 05[1 1 05 kHOH(k2H k2OH)05] (A5a)

Therefore the precursor formation cross sectionsp in Eq 1 is a fraction fp of the total dissocia-

tion cross sd Since laboratory data suggest theH2O2 fraction in an irradiated ice is small thenkHOH (k2OH k2H)05 which is likely givingfp R [(k2OH k2H)05kHOH]

Trapped versus mobile species

To examine the fluence dependence discussedin the text we note that trapped OH and freshlyproduced transiently mobile OH can be treatedas different species (eg Matich et al 1993)Therefore assuming no trapping of H the OHrate equation above would be replaced by twoequations one for the trapped species nOHt andthe other for the freshly produced mobile speciesnOH Assuming the reaction of two mobile OHunits is unlikely and that peroxide is formed fromthe interaction of a trapped OH with a subse-quently produced OH an expanded set of equa-tions is obtained

OH 1 trap R OHt (kOHt)

OH 1 OHt R H2O2 (kOHOHt)

H 1 OHt R H2O (kHOHt)

then the reaction pathways are described by

qH2

H2O 1 radiation mdashRsdH

qk2H

1 OHH

trap OH

OH RkOHt

OHt kOHOHtH2O2

kHOHkHOHt

H H2O HH2O

The resulting set of equations is

dnOHdt 5 sd f n 2 kHOH nH nOH

2 kOHt nOH nt 2 kOHOHt nOH nOHt (A1b)

dnOHtdt 5 kOHt nOH nt 2 kOHOHt nOH nOHt

2 kHOHt nH nOHt (A1b9)

dnHdt 5 sd f n 2 2 k2H nH nH 2

kHOH nH nOH 2 kHOHt nH nOHt (A2b)

JOHNSON ET AL844

dnH2O2dt 5 kOHOHt nOH nOHt

2 destruction processes (A3b)

where nt is the trap density in the irradiated iceAt low T we can assume nt is large so thatdntdt 0 simplifying the set of equations If theprecursor of interest is trapped O as discussed inthe text then a similar set of equations can be con-structed for O and Ot as in the example to followOf course a full description over a broad rangeof temperatures would require the considerationof the both trapped and mobile H O and OHthe rate of change of trap density and the vari-ous destruction processes

The above equations should be solved simul-taneously but for the incident particle fluxes typ-ically used in these studies the diffusion of mo-bile species to traps is relatively fast This is trueeven at low temperatures if there is a significantdensity of trapping sites This is usually the casefor samples formed at low T Therefore in a lowtemperature solid at low dose rates there is not asignificant steady-state background of diffusingOH There are trapped or freshly produced OHunits which rapidly react Averaging over thebombardment rate this can be approximated bywriting dnOHdt 0 and then separating nOH outin Eq A1b Assuming also that H diffuses rapidlyand either reacts or forms H2 we also writednHdt 0 on the average Therefore betweenimpacts only species trapped at defects pores orgrain surfaces are assumed to be present as sug-gested by the lack of a dependence on the inci-dent beam flux In this limit the rate of change oftrapped OH roughly simplifies to

dnOHtdt 5 2 k2H nH nH 2 2 kOHOHt nOH nOHt

It is seen that the production of peroxide dependson the loss of H by formation of H2 as discussedand at steady state the production rate for per-oxide equals the rate of formation of H2 in theseequations

Treating the mobile and trapped species sepa-rately we can further simplify the rate equationsand obtain the precursor model used in the textThat is integrating over the relevant depth Dxthe densities are replaced by column densitiesThe rate constants ki are then related to effectivecross sections si In this way cross sections aregiven as an interaction length for each freshlyproduced mobile radical [ki nj] R [si f] The

trapped OH column density NOHt and peroxideproduction rate can be estimated from ldquolinearrdquorate equations

dNOHtdt lt [sd f] N 2 2 [s2OH f] NOHt

2 [sHOH f] NOHt (A6a)

dNH2O2dt lt [s2OH f] NOHt 2 s NH2O2 (A6b)

Here sd is again the dissociation cross sectionand the destruction processes in Eq A3b are con-tained in s The second and third terms on theright in Eq A6a are the loss of OH due to the pro-duction of H2O2 and H2O In the integration ofEq A1b9 [2 kOH OHt nOH] becomes [2 s2OH f] and[kHOHt nOH] becomes [sHOH f] Therefore thedensity and the interaction length of mobile H arecontained in sHOH If H2O2 is indeed the pre-cursor then in Eq 2 in the text [sp f N] is be-comes [s2OH f NOHt] from Eq A6b Accountingfor precursor destruction which can give backtwo OH and treating H2 can both be included inthe integrated model via the si values

The simplified set of equations above allows usto examine the relative time (fluence) scales forprecursor formation (in this case H2O2) SolvingEqs A6a and A6b the column density of OH andthe concentration of precursors versus fluence cpcan be obtained Using the column of H2O2 pro-duced as the precursor column Np in Eq 2 thenO2 yield can be obtained The results can be com-pared with the results from Eqs1 and 2 in thetext The average precursor (H2O2) concentrationover the column depth of water molecules N is

cp 5 fp (sds)[1 2 exp(2s F)] 2 (ssr)

[1 2 exp(2sr F)] (A7)

with

fp 5 2s2OH(sr 2 s) (A5b)

In these equations sr 5 (2s2OH 1 sHOH) andssr 1

It is seen that the precursor concentration withtime cp depends on two rates the reaction ratesthat determine sr and the destruction rates in sAs in Eq 4a the yield can be written as

Y 5 sO2 cp N (A8)

If sr and s are comparable then the yield at small

OXIDANTS AT EUROPA 845

fluence is quadratic in F However if sr sthen the linear dependence on fluence is obtainedwith a small shift in the fluence as shown in Eq4c in the text Further the steady-state yield isnow

Y` (T) 5 sO2 [fp sd] Ns (A9)

That is sp R [fp sd] as in Eq A4 and the tem-perature dependence for the incident electrondata must be contained in [sO2 fp] or in the ratio[sO2 s2OH]sHOH]

For comparison with their precursor modelSieger et al (1998) examined the possibility of ob-taining the O2 precursor by considering reactionsfor a density of diffusing dissociation products asin Eqs A1 and A2 rather than a trapped productinteracting with a freshly formed dissociationproduct Based on the above the lowest fluence atwhich O2 is first detected experimentally sets alimit on the time scale for mobilendashmobile reactions

Effect of trap density

Equations A1b A1b9 A2b and A3b can be ap-proximately solved assuming that the dnOHdtdnHdt and dntdt are very small on the averageas discussed above In order to examine the roleof the trap density nt which can depend on thetemperature and method of formation and on theradiation history of the ice we use the slightlysimpler precursor model In this model the O pro-duced directly by dissociation traps forming aprecursor O2 is then formed when a freshly pro-duced O or other species reacts with a trappedoxygen atom Ot

H2O 1 radiation R H2 1 O (sd)

O 1 trap R Ot (kOt)

O 1 Ot R O2 (kOOt)

The reaction pathways for the oxygen species are

H2O 1 radiation mdashRsd9

trap OO2

kOtOt kOOt

H2 1 OOHt

kOR H(kHO t)

ROR

We assume for simplicity both H2 and O2 escapefrom depth Dx after formation We also allow mo-bile H to react with Ot (kHOt) removing trappedO and both the mobile H and O to be removedby other reactants R (eg OH or contaminants)This leads to a simplified set of equations not di-rectly involving OH or peroxide

dnOdt 5 sd9 f n 2 kOt nO nt 2 kOR nO nR

2 kOOt nO nOt (A1c)

dnOtdt 5 kOt nt nO 2 kOOt nO nOt

2 kHOt nH nOt (A1c9)

dnHdt 5 sd f n 2 kHOt nH nOt

2 kHR nH nR (A2c)

dnO2dt 5 kOOt nO nOt (A3c)

These equations of course act in parallel with theequations above for OH and mobile OH may re-act with trapped O to form O2 Here we focusonly on mobile O

Since the diffusion of H and O are fast and thedensity of traps is large the nH and nt averagedover the bombardment rate can be taken veryroughly as time independent Further if we as-sume the number of trapped O nOt is small com-pared with the number of sites at which H canreact then nH [sd f nkHR nR] For a low ra-diation flux f nH is small on the average as as-sumed Since the O2 yield YO2 is equal to[(dnO2dt) Dxf] with Dx the escape depth thenthe steady-state yield Y` can be obtained ana-lytically

[Y`][sd9 N] 5 1 1 (q 2 1) 2 [(q 2 1)2

1 8 q d]054 d (A10)

q 5 [kOt (nt 1 nR)] (kHOt nH)(sd9 f n) kOOt

5(sdsd9) [kOt (nt 1 nR)] kHOt[kOOt kHR nR]

d 5 05 [1 1 nt(nt 1 nR)]

We note that for nt large q can be large Assum-ing that q 1 and nt nR then Eq A10 be-comes

Y` R [sd9 N]q 5 (sd9 N) (sd9sd)

[kOOt kHR nR][kHOt kOt nt] (A5c)

JOHNSON ET AL846

Comparing with Y` in the simple precursormodel in Eq 4b then s R [(kHOt nH)f] R [sd nkHOt][kHR nR] In addition the product [sO2sp] R sd9 sd9 [kOOt n][kOt nt] That is the dis-sociation cross sections are modified by reactionrates

Quite remarkably it is seen in Eq A5c that thesteady-state yield Y` can depend inversely onthe density of traps nt in the temperature andflux regimes for which the approximations arerelevant That is if there are competing destruc-tion process for trapped O then at high trap den-sity a mobile O must find a trapped O before itbecomes trapped or before mobile H destroys theOt In this way the increasing O2 yield with in-creasing T can be related to the change in the trap-ping density nt with the formation temperatureof the ice sample As the temperature increasesand the density of traps in the bulk decreases thesurface and grain interfaces become the only trap-ping sites

ACKNOWLEDGMENTS

REJ acknowledges support by a Gledden Fel-lowship in Chemistry School of Biomedical andChemical Sciences at The University of WesternAustralia and by grants from the NSF Astron-omy Program and NASArsquos Planetary Geologyand Geophysics Program PDC acknowledgessupport by an Australian Postgraduate Award

ABBREVIATIONS

HST Hubble Space Telescope MJQ MatichndashJohnsonndashQuickenden NIMS Near Infrared Map-ping Spectrometer UV ultraviolet

REFERENCES

Ayotte P Smith RS Stevenson KP Dohnarsquolek ZKimmel GA and Kay BD (2001) Effect of porosityon the adsorption desorption trapping and release ofvolatile gases by amorphous solid water J Geophys Res106(E12) 33387ndash33392

Bahr DA Famaacute M and Baragiola RA (2001) Radiol-ysis of water ice in the outer solar system sputteringand trapping of radiation products J Geophys Res 10633285ndash33290

Baragiola RA (2003) Microporous amorphous water icefilms and astronomical implications In Water in Con-

fining Geometries edited by JP Devlin and B Buch El-sevier New York in press

Baragiola RA and Bahr DA (1998) Laboratory studiesof the optical properties and stability of oxygen onGanymede J Geophys Res 103 25865ndash25872

Baragiola RA Atteberry CL and Bahr DA (1999) Ori-gin of solid oxygen on Ganymede Reply to commentby RE Johnson on ldquoLaboratory studies of the opticalproperties and stability of oxygen on Ganymederdquo JGeophys Res E104 14183ndash14187

Baragiola RA Vidal RA Svendsen W Schou J ShiM Bahr DA and Atteberrry CL (2003) Sputteringof water ice Nucl Instrum Methods Phys Res B 209294ndash303

Bar-Nun A Herman G Laufer D and Rappaport ML(1985) Trapping and release of gases by water ice andimplications for icy bodies Icarus 63 317ndash332

Baverstock KF and Burns WG (1976) Primary pro-duction of oxygen from irradiated water as an expla-nation for decreased radiobiological oxygen enhance-ment at high LET Nature 260 316ndash318

Benit J and Brown WL (1990) Electronic sputtering ofoxygen and water molecules from thin films of waterice bombarded by MeV Ne1 ions Nucl Instrum Meth-ods Phys Rev B 46 448ndash451

Benit J Bibring J-P della-Negra S LeBeyec YMendenhall M Rocard F and Standing K (1987)Erosion of ices by ion irradiation Nucl Instrum Meth-ods Phys Rev B 1920 838ndash842

Brown WL Augustyniak WM Lanzerotti LJ John-son RE and Evatt R (1980) Linear and nonlinearprocesses in the erosion of H2O ice by fast light ionsPhys Rev Lett 43 1632ndash1635

Brown WL Augustyniak W M Simmons E Mar-cantonio KJ Lanzerotti LJ Johnson RE BoringJW Reimann CT Foti G and Pirronello V (1982)Erosion and molecular formation in condensed gasfilms by electronic energy loss of fast ions Nucl In-strum Methods B 1 307ndash314

Brown WL Augustyniak WM Marcantonio KJ Sim-mons EN Boring JW Johnson RE and ReimannCT (1984) Electronic sputtering of low-temperaturemolecular solids Nucl Instrum Methods B 1 307

Buratti B (1995) Photometry and surface structure of theicy Galilean satellites J Geophys Res 100(E9) 19061ndash19066

Buratti B and Veverka J (1983) Voyager photometry ofEuropa Icarus 55 93ndash110

Calvin WM and Spencer JR (1997) Latitudinal distri-bution of O2 on Ganymede observations with the Hub-ble Space Telescope Icarus 130 505ndash516

Calvin WM Clark RN Brown RH and Spencer JA(1995) Spectra of the icy Galilean satellites from 02 to5 mm a compilation new observations and a recentsummary J Geophys Res 100 19041ndash19048

Calvin WM Johnson RE and Spencer JA (1996) O2

on Ganymede spectral characteristics and plasma for-mation mechanisms Geophys Res Lett 23 673ndash676

Calvin WM Aninich VG and Brown RH (2004) Oxy-gen temperature and phase effects Icarus (in press)

OXIDANTS AT EUROPA 847

Whereas H2O2 is observed in ice at Europa at013 level the O3-like UV feature seen atGanymede is not obvious at Europa (eg Fanaleet al 1999) This again may be due to the fact thatless trapped O2 is seen in reflectance at Europathan at Ganymede That is since O2 is a precur-sor of O3 the factor of 6 reduction in O2 meansthat the identification of this broad feature su-perimposed on other broad UV bands is prob-lematic However the absence of the O3-like fea-ture could also be due to the presence of sulfurspecies that strongly compete for excess oxygen(Johnson and Quickenden 1997)

Summary space observations

Although each of the individual identificationsof suggested products of water radiolysis or pho-tolysis (OH H2O2 O2 O3) has some uncertaintythe reflectance spectra taken as a whole give afairly clear self-consistent picture The irradiatedsurfaces of the icy satellites contain trapped mol-ecules that would be expected from radiolysis orphotolysis of an icy surface (Johnson and Quick-enden 1997) Indeed OH is a suggested precur-sor of H2O2 which is a suggested precursor ofO2 which in turn is a precursor of O3 Howeverthe separation into amounts of individual speciesbased on the absorption in the UV identificationsremains difficult Therefore the observed band oftrapped H2O2 in the infrared and the pair ofbands in the visible of a condensed form of O2are critical as they confirm the presence of radi-olytically produced oxidizing molecules in Eu-roparsquos icy surface

Consistent with the above oxygen-rich mole-cules derived from the radiation processing of anice containing sulfur (SO2 and a sulfate) and car-bon (CO2 and a carbonate) have been identified(see eg Johnson et al 2003) The sulfate hasbeen suggested to be a hydrated salt mineral (Mc-Cord et al 1999) or a hydrated sulfuric acid (Carl-son et al 1999b 2002) The latter is an efficientoxidant that is also refractory Together these ob-servations indicate that volatile and refractory ox-idizing molecules are readily formed by radioly-sis and photolysis in Europarsquos surface material Ifindeed that material can be delivered to Europarsquosputative subsurface ocean a source of free energymay be available for potential biological activityHere we consider the formation of molecular ox-idants by radiolysis with emphasis on O2

RADIOLYTIC PRODUCTIONLABORATORY DATA

Introduction

In this section we review the laboratory dataleading to the production of oxygen and relatedspecies by radiolysis and photolysis of ice Wefirst compare the data for production of oxygenby incident ions electrons and UV photons Thisis followed by a review of the data on the loss ofhydrogen which is the other principal product ofradiation-induced decomposition of ice We thenconsider the evidence that certain proposedspecies such as peroxide or trapped O might beprecursors to molecular oxygen formation in iceSome of these data have been summarized ear-lier but with a very different emphasis Here thegoal is to motivate the development of chemicalmodels for describing irradiated ice models thatcan be used to suggest new experiments and tointerpret data Individuals familiar with the datacan go directly to the modeling section in whichrate equations are developed and used to inter-pret aspects of the diverse database

Molecular oxygen

Gas-phase O2 has been produced by irradiat-ing low temperature ice using UV photons low-energy (eV) electrons keV ions and electrons andMeV ions (see eg reviews by Johnson andQuickenden 1997 Delitsky and Lane 1998 John-son 1990 1998 2001 Strazzulla 1998 Baragiola2003 Johnson et al 2003) In spite of the consid-erable differences in excitation density and pen-etration depth for these incident particles the re-sults exhibit important similarities Here wesummarize the principal findings We first notethat the efficiency of this process in pure ice islikely to be lower than estimates based on veryearly measurements of the production of O2 in ice(see eg Hart and Platzman 1961) because oflow levels of contaminants in those early experi-ments (Baragiola and Bahr 1998 Moore andHudson 2000) Therefore trace species can be im-portant in determining the O2 production ratesfrom Europarsquos icy surface and will be discussedshortly

In studying the ability of energetic plasma par-ticles to eject (sputter) molecules from ice aprocess occurring on small Outer Solar Systembodies Brown Lanzerotti and co-workers (Brown

JOHNSON ET AL826

et al 1982 1984) found that at temperatures rel-evant to the surface of Europa H2 and O2 wereimportant components of the ejecta They thenused those data to predict the presence of a radi-olytically generated O2 atmosphere on Europa(Johnson et al 1982) The production efficiencyfor gas-phase O2 was found to be strongly tem-perature dependent as seen in Fig 2a for MeVHe1 ions incident on ice and Fig 2b for keV Ar1

ions on ice In Fig 2 the production rate is givenas a yield which is the number of moleculesejected per incident ion (ie O2 production ratedivided by the flux of incident particles photonselectrons or ions) Yields are typically measuredat a low incident particle flux so that sample heat-ing does not occur The yield versus temperatureshown in Fig 2 appears to increase exponen-tially with T over a range of temperatures Al-though this is not an Arrhenius dependencerough activation energies can be extracted overnarrow ranges of T (eg Fig 2a inset) Theseare of the order of 001ndash007 eV (Table 2) Suchvalues are much smaller than those for diffusionof vacancies or trapped species in ice (Livingstonet al 2002) but are comparable to the reportedactivation energies for H1 or H diffusion in iceas noted earlier (Johnson and Quickenden1997) The lowest energies are also comparableto the enthalpy difference between amorphoussolid water and crystalline ice (Ayotte et al2001) Therefore the estimated energies are likely

related to H1 and H transport or to hydrogenbond re-orientation It has also been shown thatas the temperature is increased above 80ndash100Kmorphological changes occur that can affect thereaction kinetics and give incorrect activation en-ergies This is especially so when using measure-ments of average quantities such as yields ratherthan directly extracting rate constants by mea-suring transients (Selby 2003) In addition the ac-tivation energies can be representative of the com-peting (destruction) processes rather than theformation processes (eg Matich et al 1993) asdiscussed in the model section

The measured O2 production rates (yields)were also found to be fluence dependent wherefluence is the radiation flux 3 the time or the to-tal number of particles per unit area onto the iceMultiplying the fluence by the incident particleenergy and dividing by the average penetrationdepth gives the energy per unit volume oftencalled an average dose As seen in Fig 3 the O2yield versus fluence rises from zero implyingthat a single ion incident on a fresh surface doesnot efficiently produce and eject oxygen This isthe case even for an incident 15 MeV Ne1 forwhich the excitation density along the ion path ishigh However it is seen in Fig 3 that with in-creasing irradiation time individual impactingions produce and eject O2 with increasing effi-ciency as the material is chemically altered Atlarge fluences the yield appears to saturate indi-

OXIDANTS AT EUROPA 827

FIG 2 a Yield for D2 and O2 at saturation (steady state) from 15 MeV He1 on D2O ice versus the temperature ofthe ice (adapted from Brown et al 1982) Inset D2 yield minus the constant yield at low T versus 1kT giving an av-erage activation energy of 003 eV b Yield of O2 at saturation versus temperature for incident 30 keV Ar1 ions(adapted from Baragiola et al 2003)

a b

cating that destruction processes are in a roughsteady state with formation processes The yieldsshown in Fig 2 are purportedly those after satu-ration although full saturation is often ap-proached slowly

The slope of the O2 yield versus fluence shownin Fig 3 increases with increasing temperaturegiving the activation energies in Table 2 Thesevalues are similar to those obtained from Fig 2For energetic ions that fully penetrate the film asin Fig 2a and 3a the yield versus fluence can ex-hibit a second rise at higher fluences for whichthe radiation track cores begin to overlap This in-crease also depends on temperature and exhibitssimilar activation energies (Reimann et al 1984)The strong dependence on temperature is at firstsurprising since the energy density deposited bya 15 MeV Ne1 produces transient temperaturesin the track core that are much higher than theambient material temperature Therefore tracktemperature (local excitation density) alone doesnot determine the O2 yield

For all incident particle types and energies themeasured fluence dependence is found to be sta-ble over laboratory time scales That is for ex-periments carried out at high vacuum the yieldreturns to the value it had when the beam wasturned off This indicates that the irradiated ice isldquopermanentlyrdquo altered (Reimann et al 1984)Thus a relatively stable chemical state is createdthat is favorable for O2 production This could bethe formation of a particular trapped radiationproduct in ice that acts as a stable precursor or isdue to the altered chemical and physical state ofthe irradiated ice Trapped O H2O2 and HO2(Matich et al 1993 Johnson and Quickenden1997 Sieger et al 1998) have been proposed asstable precursor species

Using isotopically labeled H2O Benit andBrown (1990) showed that considerable transportof O occurs along the ion track when an energeticheavy ion penetrates an ice sample Because ofthis the O2 yield was found to depend on sam-ple thickness Therefore for thin ice samples the

JOHNSON ET AL828

TABLE 2 OXYGEN AND PEROXIDE PRODUCTION

Radiationa Product T (K) Ea (eV) Percent at saturation G (product100 eV)b

He1(1) O2 015g(23) O2 77 (300) 0003 (0)Ne1 (30 MeV)(3) O2 300 0031He21 (6 MeV)(3) O2 300 0008He21 (29 MeV)(3) O2 300 00016H1 (80 keV)(4) O2 70 00002H1 (200 keV)(5) O2 120 00007He21 (15 MeV)(6) O2 50 (100) 003 (07ndash4) 3 1024

Ne1 (15 MeV)(7) O2 7 10ndash15c 003(8) 70ndash140 005ndash007 01

Ar1 (01 MeV)(9) O2 30ndash140 001ndash004 0006ndash008160 001

e (30ndash100 eV)(10) O2 120 002ndash003 0005ndash0007hn (102 eV)(11) O2 50 (100) 003 00001 (00003)H1 (1 MeV)(12) H2O2 16 (77) 01 (001)

1 10 CO2 77 011 10 O2 77 04

H1 (30 keV)(13) H2O2 16 (77) 022 (038) 0098 (0086)C1 (30 keV) H2O2 16 (77) 17 (22) 023 (034)N1 (30 keV) H2O2 16 (77) 16 (14) 021 (011)O1 (30 keV) H2O2 16 (77) 14 (40) 026 (015)Ar1 (30 keV) H2O2 16 (77) 22 (20) 011 (006)

aReferences are as follows 1Lefort (1995) 2Ghormley and Stewart (1956) 3Baverstock and Burns (1976) 4Baragiolaet al (1999) 5Baragiola and Bahr (1998) 6Brown et al (1982) (ion does not stop in sample) 7Benit and Brown (1990)(ion does not stop in sample) 8Reimann et al (1984) (ion does not stop in sample) 9Baragiola (2003) 10Sieger et al(1998) (steady-state value) 11Westley et al (1995) (steady-state value as no fluence dependence seen) 12Moore andHudson (2000) 13Gomis et al (2003)

bG values are for escaping O2 For H2O2 and O2 formation both of which require multiple events yields after sat-uration are used In a few cases yields are obtained by melting the sample Values in parentheses are those at anothertemperature

cPercent of film eroded as O2 molecules for long-term irradiation

ldquosaturatedrdquo yields change slowly with fluenceMore importantly O from water molecules at aconsiderable depth into the sample contributedto the O2 ejected from the surface Since there wasno observed diffusion of the O2 in ice at the am-bient temperature (7K) over the time for their experiment they concluded that the oxygentransport and production are stimulated by theincident ions The transport was suggested as be-ing due to percolationdiffusion along the dam-aged and transiently heated tracks (Reimann etal 1984 Benit and Brown 1990) They also foundthat in fully eroding (sputtering) thin ice filmsthe total amount of O2 produced required about10ndash15 of the D2O molecules in the films at T 57K On warming irradiated samples some O2

formed at depth is found to remain trapped forexperimentally relevant times (see eg Hart andPlatzman 1961 Benit and Brown 1990 Baragiolaand Bahr 1998)

Matich et al (1993) detected luminescence in77K H2O and D2O ice from newly formed and ex-cited O2 They suggested that the O2 is formed

from the interaction of a freshly produced mo-bile O interacting with a previously producedand trapped O or H2O2 They favored the formerIn matrix isolation studies of peroxide irradiatedby UV photons luminescence from newly formedexcited O2 was seen However the oxygen wasformed from mobile O (greater than 30K in Krlattice) interacting with O trapped as H2O-O(Khriachtchev et al 2000) These studies also sug-gest that trapped O possibly in the form H2O-Omay be an important precursor

In Fig 4a the O2 yield for low-energy electronsplotted as a function of electron energy exhibits anapparent ldquothresholdrdquo at 10 eV (Orlando andSieger 2003) Although the uncertainties are largethe authors suggested there may be a lower thresh-old 6 eV corresponding to a very small crosssection for O2 production (10220 cm2) This couldbe associated with an excitation at the ice surfaceBulk ice exhibits very weak absorption at photonenergies below 6 eV but exhibits a strong ab-sorption that depending on the ice structure be-gins at about 73 eV with a peak at 86 eV This

OXIDANTS AT EUROPA 829

b

a

FIG 3 a Relative yields versus fluence (ion flux 3 time) for 15 MeV Ne1 incident on D2O ice at 7K (adapted fromReimann et al 1984) Top panel D2O yield Middle panel D2 yield Bottom panel O2 yield Yields at other tem-peratures exhibited similar dependences with the slope of O2 yield versus fluence varying with T Fits give activa-tion energies between 002 eV at the lowest fluences and lowest T to 007 eV at the highest T and highest fluencesb Oxygen yield versus fluence of low-energy electrons on D2O ice (from Sieger et al 1998)

is followed by a peak in absorption at 10 eVThe results in Fig 4a show that there is a dra-matic increase in the production of O2 at 10 eVwhich is the onset of secondary electron produc-tion in the bulk (Orlando and Sieger 2003)

For incident fast ions the yield at a given tem-perature was found after a small fluence to beindependent of whether the sample was crys-talline or amorphous ice (Brown et al 1982)However using low-energy electrons the yieldswere found to be smaller for an initially crys-talline ice (Orlando and Sieger 2003) Since verylow doses of ion radiation (a few eV per mole-cule) can convert crystalline ice to amorphous iceat low temperatures (Leto and Baratta 2003) thedamage produced by the penetrating ions usedby Brown et al (1982) affected the O2 yield Thestructural damage produced may create trappingsites cause changes in the electronic relaxationpathways or enhance percolation

The temperature dependence of the O2 yield isshown in Fig 5 for electrons incident on initiallyamorphous (Fig 5a) and initially crystalline (Fig5b) ice A small decrease in efficiency is seen inboth cases at 120ndash130K where Ic forms fromamorphous ice and pore collapse occurs (Row-land et al 1991) This is followed by an increas-ing yield with increasing T and then by a more

dramatic drop near 160K where Ih forms Dur-ing the transition to Ih trapped molecules havebeen shown to escape efficiently from ice samples (Ayotte et al 2001) A large drop in the yield isalso seen in Fig 2b for incident Ar1 ions

In modeling the dramatic decreases in the yieldin Figs 2b and 5 care must be taken That is thetime for crystallization of an amorphous sample(Baragiola 2003) even allowing for the large un-certainties is 30 min at 145K and becomes 1min at 160K Therefore pulsed radiolysis exper-iments should be carried out at these tempera-tures and the transients observed Since the timefor conversion from amorphous to crystalline icecan be shorter than the experimental times at145K the results are not comparable to thosefor which the chemical state of the ice at a givenfluence is stable over experimental times Belowwe focus on the data below 140K

We have emphasized processes initiated by theelectronic excitations and ionizations producedHowever momentum transfer collisions whichare the dominant energy loss process for low-en-ergy ions also produce dissociations defects andO2 in ice (Bar-Nun et al 1985) The relative effi-ciency of momentum transfer and electronicprocesses for producing O2 needs further studyAs the penetration depths are much smaller when

JOHNSON ET AL830

FIG 4 a Relative yield of O2 from D2O ice versus incident electron energy E at T 5 120K (from Orlando et al1998 Orlando and Sieger 2003) b Relative yield of D2 from D2O ice versus incident energy E at T 5 88K (fromKimmel et al 1994)

a

b

the former dominates the produced O2 likely es-capes readily

H2 formation and loss correlation with O2

The production of O2 from a low temperatureice by radiolysis correlates with the formationand loss of H2 (or D2 from D2O ice) (Reimann etal 1984) D2O is used to reduce the backgroundin mass spectrometry studies The yields of O2from D2O and H2O ices were shown to be com-parable However the total yields from D2O wereabout 10 less (Reimann et al 1984) consistentwith H (D) transport being important It is seenin Fig 3 that D2 is promptly produced and lostfrom the D2O ice At higher fluence the D2 yieldis seen to exhibit a dependence like that for O2For the 15 MeV Ne1 ions which fully penetratethe ice sample the D2 yield eventually decreasessince D2 is produced and escapes from the fullthickness of the film

Because of H2 (D2) loss the surface region of

the irradiated ice becomes depleted in H (D) rel-ative to O favoring the formation of oxygen-richspecies The change seen in the ratio of H (D) toO in the irradiated ice is small An irradiated film(100 nm) remains stoichiometric to better than2 (Brown et al 1980) and in the near surface re-gion Auger spectroscopy suggests only a slightoxygen enhancement due to irradiation by keVelectrons (Rye et al 1978) Sieger et al (1998) sug-gested a precursor concentration of the order ofa percent Such concentrations are an order ofmagnitude smaller than the steady-state concen-tration of dissociated H2O (Table 2)

Using infrared spectroscopy the damaged H2Oconcentrations at low temperatures in the satu-ration regime are about 10 over the irradiatedvolume (Benit et al 1987 Watanabe et al 2000Gomis et al 2003) This fraction is the ratio of thedissociation cross section to the effective crosssection for recombination to H2O as discussedlater At a 10 concentration most H2O will havea neighbor that is dissociated

OXIDANTS AT EUROPA 831

FIG 5 a Relative steady-state yields of O2 from a D2O ice sample versus T (upper panel) The ice samples wereformed and irradiated at these temperatures Film erosion as a yield (lower panel) (from Orlando and Sieger 2003)b Relative steady-state yields of O2 versus T as in (a) (upper panel) Here the D2O ice samples are formed at 160Kpresumably becoming crystalline and then irradiated at the T shown A new sample is used for each data point in(a) and (b) Film erosion as a yield (lower panel) (from Orlando and Sieger 2003)

At temperatures above 40K it is seen in Fig2 that the D2 and O2 yields exhibit the same tem-perature dependence suggesting a correlationbetween D2 and O2 loss H2 (D2) can be formedby reactions between diffusing hydrogen fromdissociated water or directly [ie H2O R H2 1O (eg Kimmel and Orlando 1995) H2O R (H1 OH)cage R H2 1 O (eg Matich et al 1993)H3O1 1 e R H2 1 OH (eg Williams et al1996)] Subtraction of the nearly constant D2 yieldin Fig 2a gives the Arrhenius plot in the insetSince molecular hydrogen readily escapes atthese temperature it is the H2 (D2) formationprocess that has an activation energy of 003eVmolecule The fact that slopes of the yields forH2 and O2 above 40K are similar suggests thatH2 (D2) formation and loss is the rate-limitingstep in determining the production and loss of O2at these temperatures

Although the data at low T in Fig 2a have con-siderable uncertainties below about 20ndash40K theD2 production is roughly independent of tem-perature but the O2 yield continues to decreasewith decreasing temperature Radiolytically pro-duced D2 is mobile and can escape above15ndash20K (Watanabe et al 2000) but O2 is im-mobile at such temperatures and the incidentHe1 ions are much less efficient at causing per-colation than the incident Ne1 discussed earlier

The production of H2 (D2) from ice has alsobeen studied using low-energy electrons and UVphotons at very low temperatures In the formerstudies absolute yields were not obtained butthere appeared to be a threshold at 63 eV (Fig4b) This is roughly the threshold for excitationof a surface state (Khusnatdinov and Petrenko1992) and the threshold for peroxide productionis close to this value The efficiency for H2 (D2)production increases around 11 eV with a sec-ond steep rise at 17 eV (Kimmel et al 1994) TheO2 and D2 yields versus electron energy in Fig 4exhibit some similarity in the threshold regionalthough the uncertainties are large and the sam-ple temperatures different Measuring the H2 (D2)and the O2 ejecta simultaneously over a range oftemperatures as in the Ne1 experiments wouldbe instructive

Using 98 eV photons at low T (12K) Watanabeet al (2000) measured the D2 ejected and trappedin ice They obtained a cross section for produc-tion of D2 (Table 2) at low fluences that is10ndash20 of the total absorption cross sectionwith significant uncertainties They also found

that the equilibrium damage fraction in the sam-ple was 10ndash20 of the D2O molecules but theequilibrium production of D2 was about an orderof magnitude smaller 1ndash2 at their tempera-tures They suggested that the D2 is formed atgrain interfaces and cracks and on the surfaces ofpores In a mixed ice D2O 1 CO exposed to UVphotolysis D2 and CO2 formation are correlated(Watanabe and Kouchi 2002) The D2O dissoci-ates to produce D2 plus O and this transientlymobile O reacts with CO to produce CO2

Precursor state and peroxide

In order to produce O2 from low temperatureice it appears that the ice sample must lose hy-drogen and form trapped oxygen-rich reactivespecies The efficiency of production also appearsto be affected by the level of defects and damagein the ice consistent with the suggestion that thehydrogen-bonding network is important in theproduction of O2 via availability of trapping sitesnew electronic relaxation pathways or the natureof the bonding of precursors Here we briefly ex-amine the possible role of the suggested precur-sor molecules H2O2 and HO2 We note that H2O2might not readily dissolve in ice possibly ac-counting for its low abundance in some experi-ments It appears to be heterogeneously distrib-uted in the form of microcrystals (Gurman et al1967) If H2O2 does act as a precursor the for-mation of aggregates could affect the dimer ab-sorption bands (Cooper et al 2003a) and affectthe production and stability of O2 (Cooper et al2003b)

The results of Gerakines et al (1996) in whichice at 10K is irradiated by UV photons confirmthat H2O2 forms in ice from two OH after a build-up of trapped OH They monitored the band as-sociated with H2O-HO (Langford et al 2000) inwhich the HO is bound to water with 024 eVTherefore they might only be monitoring a sub-set of the trapped OH seen in earlier experiments(eg Taub and Eiben 1968) The H2O2 concen-tration in ice reaches a steady-state level at flu-ences of about 1018 photonscm2 (Gerakines et al1996) As seen in the model section this gives anestimate of the H2O2 photodestruction cross sec-tion of the order of 10218 cm2 Similarly HO2 be-gins to be seen in absorption at about a few times1017cm2 shortly after they begin to detect H2O2The absolute yields were not measured

Westley et al (1995) detected H2 and O2 pro-

JOHNSON ET AL832

duced by Lyman-a (98 eV) radiation of ice butfound no fluence dependence for fluences greaterthan 1017cm2 They also saw small amounts ofH2O2 by programmed thermal desorption of thesample but made no quantitative evaluationComparing their lowest fluence for the produc-tion of O2 with the result of Gerakines et al (1996)for peroxide suggests that H2O2 and HO2 areprobably not important precursors of O2 In UVphotolysis of H2O2 using 64 eV photons the pro-duction of O2 was not observed (Vaghjiani et al1992) and the excitation of an irradiated icewhich presumably contained some peroxide by37 eV photons did not produce O2 (Baragiola etal 2003) Direct excitation of matrix-isolatedH2O2 at low temperature also did not yield O2but the production of a mobile O that reacted withtrapped O (H2O-O) did give O2 (Khriachtchev etal 2000)

Irradiating with 200 keV protons at tempera-tures of 40ndash129K Bahr et al (2001) found onlyabout 0001ndash002 H2O2 using programmed ther-mal desorption as compared with earlier experi-ments that suggested larger values (eg Table 2)Using infrared absorption Moore and Hudson(2000) found a G value of 01 for incident MeVprotons at low temperature but the signal in theobserved band became negligible at 80K sug-gesting that H2O2 was not being formed Theyalso found that adding O2 significantly increasedthe production efficiency of H2O2 That is thatthe amount of peroxide was affected by theamount of trapped oxygen Based on the matrixisolation studies (Khriachtchev et al 2000) dis-sociated mobile O might trap as H2O-O whichon excitation converts to H2O2 Therefore even ifH2O2 is not the precursor the peroxide andtrapped oxygen concentrations will be related

Although molecular oxygen was not a productof photolysis of matrix isolated H2O2 it was aproduct of photolysis at 266 nm (466 eV) of per-oxide dimers in an argon matrix 2 H2O2 R 2 H2O1 O2 (Engdahl and Nelander 2002) This de-composition was thought to be due to the con-siderable distortion of the bonds in the dimerTherefore the configuration or distortion by thelattice of peroxide or any other precursor mole-cule in the ice may be critical (Sieger et al 1998)In addition more recent experiments show thatG values for the production of H2O2 by 30 keVincident ions are significant as are the steady-state ratios of H2O2H2O (Gomis et al 2004) Ra-tios of the order of 02ndash4 were found at 16K and

77K in the irradiated regions with incident H giv-ing the smallest fraction and incident O thelargest (Table 2) At these concentrations perox-ide molecules might segregate forming inclu-sions (Gurman et al 1967 Cooper et al 2003b)Since the amount of trapped peroxide also com-petes with the amount of trapped O studies areneeded to describe the chemical state of the solidproduced by radiolysis This will be critical in de-scribing the fate of the irradiation products at Eu-ropa In particular the chemical composition fa-vorable for producing O2 needs to be determinedas does the importance of trapped O H2O2 HO2or peroxide dimers as precursors

Other species

In the model below we will refer to atomic andmolecular species trapped in ice Such species areoften discussed in the literature on radiolysis andphotolysis (for summaries see Matich et al 1993Johnson and Quickenden 1997) Trapped OH OH2O2 and HO2 have been discussed andor mea-sured by a variety of techniques (absorption andluminescence spectroscopy electron spin reso-nance programmed thermal desorption etc)Care must be taken when using these data as re-sults for ice samples formed below 160K orthose exposed to penetrating ion beams likelyhave numerous defects voids and grain inter-faces so that a variety of trapping sites and trapdepths exist Therefore when a particular bandfor a trapped species (eg OH) disappears withincreasing temperature it could be due to achange in the nature of the trapping as well as toloss of the species by reaction

Although H is discussed below we do not dis-tinguish it in these simple models from H1

trapped as the hydronium ion H3O1 whichwould mean that there are corresponding nega-tive ions present Therefore below we refer to thevarious species as if they are neutral althoughthis might not be the case Although trapped elec-trons are detected only over a narrow range oftemperatures (100ndash120K) negative ions must beimportant over a broad range of temperature

We also note that SO2 and a sulfate and CO2and a carbonate appear to be trapped species inEuroparsquos icy surface Under radiolysis each ofthese can dissociate directly giving O2 Howeverwith the exception of the sulfate their densitiesare less than 1 so they are not relevant sourcesof O2 However their presence is indicative of the

OXIDANTS AT EUROPA 833

preference for forming oxygen-rich molecules inEuroparsquos ice and can affect the efficiency of pro-duction of O2 and H2O2

MODELS FOR FORMATION OF O2

Introduction

Based on the data described above the yieldsfor incident electrons and ions are found to benearly linear in fluence at the lowest fluences forwhich O2 is detected In addition the relevant al-terations of the ice samples by radiolysis appearto be stable and independent of the beam fluxRemarkably this is the case both for low-energyelectrons that make only a single excitation foreach impact and fast ions that make a density ofexcitations along their path through the solidThis suggests that O2 is produced and trapped inthe initial events and then caused to escape fromthe solid by the subsequent radiation or that astable chemical alteration of the ice is required be-fore the radiation can produce and eject O2

Because H produced by the dissociation of wa-ter molecules can attack trapped O2 and any pre-cursor oxidants the loss of H is important Thisoccurs by the production and loss of hydrogenmolecules as discussed leaving a surface that isslightly rich in O relative to H The O2 produc-tion mechanism could be direct excitation of aparticular precursor species or it might not bespecific That is a new OH or O produced by theincident radiation could react with the H-depleted solid producing O2 This would occurmore efficiently as the ratio of O to H in the nearsurface region increased

In their early work WL Brown and colleaguesassumed that processes similar to those used todescribe radiolysis of liquid and gaseous H2Ocould be used to describe the radiolysis in ice Thelow activation energy that they found was as-sumed to be associated with diffusion of somemobile radical species in ice such as H or O orpossibly diffusion of the O2 molecule itselfBased on their studies of ejection from isotopi-cally labeled D2O samples they concluded thatthe O2 is formed in the track of a single ion andthen ejected by subsequent ions (eg Benit andBrown 1990) More recently trapped specieshave been suggested as precursors These are ei-ther subsequently excited by the incident radia-tion or attacked by newly produced mobile

species We consider a simple model that incor-porates all of these processes and can be used toorganize and interpret the experimental data Wealso discuss its relationship to more completemodels In this discussion much of which is in theAppendix we consider the standard radiationproducts that have been often described How-ever in a solid the excited electron cloud pro-duced by absorption can overlap a number of lat-tice sites Therefore their relaxation pathwaysandor product species could be very differentfrom those that have been traditionally consideredin the literature on radiolysis and photolysis

Precursor model

The column density of a precursor species Npformed in ice by incident radiation can be calcu-lated using a simple model In this model cross sec-tions are used for the production sp and destruc-tion s of the precursor O2 is then formed byexcitation of the precursor or by reaction of a newlyformed radical with the precursor This is also de-scribed using a cross section sO2 For incident elec-trons having a small penetration depth the O2 pro-duced is assumed to rapidly diffuse to the surfaceand escape (Sieger et al 1998) More generally wedefine N as the column density of H2O from whichthe incident radiation can produce and eject O2 Apair of rate equations gives both Np and the pro-duction rate of oxygen dNO2dt That is

H2O 1 radiation R precursor

precursor 1 radiation R O2

giving

dNpdt 5 sp f N 2 s f Np (1)

dNO2dt 5 sO2 f Np (2)

In Eq 1 f is the incident flux and the total pre-cursor destruction cross section can be written ass 5 sp9 1 sO2 with sp9 accounting for precursordestruction processes other than the formation ofO2 Solving Eqs 1 and 2 and assuming N is nearlyconstant (10 level as discussed) the fractionof precursors cp in the affected column after timet is obtained (Appendix)

cp 5 NpN 5 [sps] [1 2 exp(2 s F)] (3)

Here F 5 [ft] is the fluence (flux 3 time) which

JOHNSON ET AL834

is the number of particles (photons electrons orions) incident on the surface per unit area Usingcp in Eq 2 the O2 yield Y at any fluence can bewritten in terms of the yield at steady state Y`

Y 5 [dNO2dt]f 5 sO2 cp N 5 Y`

[1 2 exp(2sF)] (4a)

Y` 5 [sp sO2 s] N (4b)

If we also account for the reduction in the den-sity of water molecules in the initial column Nthese forms change only slightly The size of thetwo parameters Y` and s can be obtained by fit-ting the yield versus fluence data such as that inFig 3

For the low-energy electron data Sieger et al(1998) showed that the ratio [YY`] for a givenincident electron energy is independent of tem-perature over a narrow range of temperatureTherefore the destruction cross section s in Eq4a is roughly independent of T and can be esti-mated from the [YY`] curve Sieger et al (1998)found a precursor destruction cross section s lt08ndash30 3 10216 cm2 for incident electron energiesfrom 30 to 100 eV This is of the order of the elec-tron impact ionization cross section for H2O

Based on the above model the lack of an ob-served fluence dependence for fluences down to1017 hncm2 in the Lyman-a irradiation of ice(Westley et al 1995) can be examined The low-est fluence used implies that the precursor de-struction cross section for Lyman-a is of the or-der of the Lyman-a absorption cross section in ice(09 3 10217 cm2)

Sieger et al (1998) also assumed that sp9 5 0This would mean that the precursor is only de-stroyed by production of O2 (ie s 5 sO2) Thatis unlikely since suggested precursors can be de-stroyed by reaction with H and other radicalsTheir assumption would also mean that the crosssection for production of O2 from a precursorsO2 is large This would be surprising and ap-pear to contradict some of the data discussedabove Instead we assume that s is not equal tosO2 and the total destruction cross section s isof the order of the ionization or absorption crosssection

Having obtained s the measured temperaturedependence for production of O2 is containedonly in Y` Over a narrow range of T one canwrite Y`(T) lt Y`9 exp(2EakT) and obtain the es-

timate of the activation energies in Table 2 Ifsp9 THORN 0 as discussed then sp and sO22 in Eq 4bcannot be obtained separately from the data forY` unless an independent measure of cp is ob-tained Such a measurement is critical for de-scribing the chemical state of the irradiated sur-faces

The observed temperature dependence of Y` iscontained in sp sO2 or N Here N is the ldquodepthrdquofrom which newly formed O2 can reach the sur-face and escape given as a column density of H2OThe physical depth is obtained by dividing thecolumn density N by the molecular number den-sity of the ice Assuming that in the low-energyelectron experiments N is independent of tem-perature and is about three monolayers (Sieger etal 1998) the measured Y` gives [sO2sps]05ndash20 3 10218 cm2 over the electron impactenergy range At 110ndash120K this ratio was foundto be proportional to (E 2 Eo) for E Eo 10 eVagain assuming fixed N (Orlando et al 1998 Or-lando and Sieger 2003)

In the following section we first consider a pre-cursor like peroxide that typically forms fromtwo dissociation events We use this to show un-der what circumstances such a model does notviolate the observed linearity of the O2 yield datawith fluence Following that we relate the sim-ple precursor model above to the case wheretrapped O is the precursor We then use themodel for trapped O to examine the effect of thetrap density on the production of O2 The densityof traps (defects) is determined by the sample for-mation temperature and the subsequent radiationdamage Finally we consider the escape depthN for penetrating radiation within the precursormodel These discussions are supported by moredetailed rate equations in the Appendix

Precursor trapped peroxide

The proposed precursor peroxide is typicallyformed from products of two dissociation eventsie 2 H2O R 2 H 1 2OH R H2 1 H2O2 This wasconfirmed in low temperature ice (Gerakines etal 1996) It is also possible that a dissociation pro-ducing a trapped O followed by an excitationevent can produce peroxide 2 (H2O 1 hn) R H2

1 H2O-O 1 hn R H2O2 1 H2 (eg Khriachtchevet al 2000) The O2 yield versus fluence for inci-dent electrons is reasonably well fit by the yieldexpression above for fluences greater than1014cm2 Therefore it has been suggested that

OXIDANTS AT EUROPA 835

any precursor species must be formed in a singleevent Using a model in the Appendix in whichthe precursor is formed by two events we showthis is not necessarily the case Precursor forma-tion if it is fast could involve two dissociationevents and give the measured fluence depen-dence Although we consider peroxide here theseconclusions apply to other possible precursors

In the example in the Appendix H2O is disso-ciated to H 1 OH according to the cross sectionsd The reactions of two OH units to form H2O2(rate constant k2OH) and 2 H to form H2 (k2H)compete with the recombination of H and OH toH2O (kHOH) If k2H is much greater than k2OH then H2 loss dominates at low fluences as ex-pected (Appendix) Steady state requires not sur-prisingly that the rate for forming H2O2 isroughly equal to the rate for forming H2 Finallyit is seen in the Appendix that the low H2O2 con-centrations suggested by experiment imply thatkHOH (k2H k2OH)05 Such sizes are reasonable

The set of rate equations in the Appendix thatuse number densities of species is integrated overdepth in order to obtain the forms used in Eqs 1and 2 which are based on column densities Inthis way the effective cross sections are related tothe rate constants for the interaction of freshlyproduced mobile species with previously pro-duced trapped species We show in the Appen-dix that the O2 yield versus fluence can be writ-ten in a form similar to that in Eq 4a but with asmall offset due to the two-step precursor for-mation process

Y lt Y`(T) [1 2 exp(2sF) 2 (ssr)]

for F sr21 (4c)

Here s is again the precursor destruction crosssection but sr 5 (2s2OH 1 sHOH) is the effectivecross section for loss of OH by formation of H2O2

and H2O The corresponding expression for Y`(T)is given in the Appendix

The form for the yield versus fluence in Eq 4cis consistent with the low-energy electron mea-surements if (ssr) 1 and if sr F21 lt10214 cm2 Such sizes are indeed possible Whatis required is that at the relevant temperatures(50ndash110K) the interaction length associated withsr is more than a few molecular layers (greaterthan 10 Aring) A cross section that is likely to belarge above 20K is sHOH which describes themean distance for diffusion of mobile H before it

reacts Since s discussed earlier is of the sameorder of magnitude as the H2O dissociation crosssection sd then the precursor concentration cp(in this case H2O2) is roughly proportional to[s2OHsHOH] 1 (Appendix)

Based on the above a two-step precursor for-mation process could be consistent with the elec-tron data contrary to the statements in Sieger etal (1998) Orlando et al (1998) and Orlando andSieger (2003) Experiments are needed to deter-mine the relevant cross sections Below we con-sider a precursor formed in a single excitationevent

Precursor trapped O

O2 can be formed for instance by diffusion ofa transiently mobile O to a previously formed andtrapped O or to the production of O near the siteof a previously produced and trapped O (Matichet al 1993 Johnson and Quickenden 1997) Werefer to this as the MatichndashJohnsonndashQuickenden(MJQ) model and note that O 1 OH R O2 1 Hcan also give molecular oxygen

Trapped O as H2O-O was in fact seen to be aprecursor to O2 when attacked by a mobile O ina Kr matrix (Khriachtchev et al 2000) ThereforeO trapped as H2O-O was also suggested as a pre-cursor in ice (Johnson 2001) Before proceedingwe note that the examination of the MJQ modelin Sieger et al (1998) was incorrect They consid-ered two mobile interacting species whose life-times were long compared with the bombard-ment rate rather than a new mobile O interactingwith a previously trapped O Although this hasnot been corrected recently they added Otrapped as H2O-O as a possible precursor (Or-lando and Sieger 2003) O trapped as H2O-O isseen in matrix isolation studies as discussedabove but it has not yet been identified in iceand like OH it may trap in a number of ways

The MJQ model is essentially the followingFormation of H2 1 O is followed by loss of H2and trapping of O (Ot) A subsequent dissocia-tion or excitation event can lead to the destruc-tion of Ot by reaction with for instance H or byconversion to H2O2 However if this dissociationevent also leads to H2 1 O a mobile O is pro-duced that can react with Ot with an interactionlength that depends on T This results in the for-mation of O2 [O 1 Ot R O2] giving a net pro-duction of 2 H2 and 1 O2 In the Appendix therelevant rate equations (Eqs A1cndashA3c) are con-

JOHNSON ET AL836

sidered Assuming the mobile O processes arefast these rate equations can also be integratedand simplified to give the form in Eqs 1 and 2The precursor column density Np now repre-sents the column density of Ot In this model sO2in Eq 2 is the ldquocross sectionrdquo for reaction of afreshly produced O with an Ot and sp9 indicatesdestruction of Ot Finally sp is the cross sectionfor production of H2 and Ot via processes dis-cussed above

Based on the sizes of the cross sections ex-tracted in the electron irradiation experimentsthe model with trapped O (Ot) as the precursoris quantitatively reasonable That is for a rangeof T both sp and sO2 might be described byevents of the type H2O R H2 1 O (whether di-rect or indirect) If these occur for 10 of theradiation-induced dissociation events as dis-cussed then the required product of cross sec-tions suggested by the sizes of s and Y` above isroughly obtained that is [sp sO2] 02ndash6 3 10234

cm2 in the low-energy electron experiments Thecross section for production of O might dependon the availability of dangling bonds at interfacesin pores or at defects as suggested by a numberof authors The importance of such bonds couldalso be consistent with the differences in the elec-tron-induced yield for initially crystalline and ini-tially amorphous ice (Orlando and Sieger 2003)

O2 production can increase in the presence ofoxygen-rich impurities such as CO2 or SO2 bothof which are known to be present at Europa Asin the MJQ model dissociation (eg CO2 R CO1 O) can provide a mobile O that can be trappedor react with a previously trapped O forming O2In addition such species are hydrogen scav-engers (Moore and Hudson 2000) This can re-duce the destruction of Ot affecting sp as seen inEq A5c

Trap density

In the simplified models above the density oftrapping sites nt was not explicitly accountedfor However in the discussion of both the ex-perimental data and models trapping sites havebeen assumed to be important It has beendemonstrated repeatedly that the density of de-fects voids and interfaces decreases as the tem-perature of formation of the ice increases How-ever the efficiency of production of O2 is theopposite it increases with increasing T There-fore the oxygen yield varies in a manner that is

the inverse of the trap density over a range oftemperatures For penetrating ions in which theenergy density deposited is high and O2 might beformed in a single track a simple explanation hasbeen given As the density of traps for O2 is re-duced and the temperature increases the O2 pro-duced can diffuse to the surface and escape moreefficiently limiting back reactions on further ir-radiation This is examined in the next sectionHowever such a process does not affect the low-energy electron yields for which the penetrationdepth is small and the formed O2 escapes read-ily at the measurement temperatures In Eqs A1cA1c9 A3c and A10 in the Appendix we againuse a simple set of rate equations for trapped andmobile O in order to determine the role of the trapdensity nt on the production of a precursorspecies The conclusions apply to any modelbased on a trapped precursor

In the MQJ model mobile O can be trappedforming Ot (described by kOt nt) and can reactwith previously trapped O to form O2 (kOOt) Inaddition it can react at other sites and Ot can bedestroyed by mobile reactive species such as H(kHOt nH) Mobile H can also react with othertrapped species (kHR nR) Incorporating these re-actions Eq A10 in the Appendix gives a steady-state yield Y` When the density of traps is largethe steady-state yield is given in Eq A5c

Y` R (sd9sd9sd) [kOOtkHR nr][kHOt kOt nt] N (5)

Here sd and sd9 are the dissociation cross sectionsthat determine the production of H (H 1 OH) andO (H2 1 O) by the incident radiation Comparingthe result in Eq 5 with Y` in the simple precur-sor model in Eq 4b the dissociation cross sec-tions in Eq 5 are now scaled by a ratio of the re-action rates

It is seen in Eq 5 that the steady-state yield Y`depends inversely on the density of traps nt Thatis there are competing destruction processes fortrapped O (Ot) Therefore in order to form O2 amobile O competes with mobile H to find atrapped O before it becomes trapped In this waythe increasing O2 yield correlates with the re-duction in the trapping sites with increasing tem-perature of formation The small activation ener-gies discussed earlier would be related to there-orientation of the lattice binding leading to thereduction in trap density At relatively high T thedensity of traps becomes small so that only the

OXIDANTS AT EUROPA 837

sample surface or grain interfaces might act astraps accounting for the observed drop in theyield at high T For this reason it is important todetermine if the yield for T greater than 150Kdepends on the incident flux f Since other quan-tities in the expression for Y` could also be tem-perature dependent experiments are neededThese would likely be pulsed radiolysis experi-ments in which the transients are studied ratherthan the steady-state yields However the avail-able data for lightly ionizing radiation do suggestthat the O2 yield depends inversely on trap den-sity as in the model above

Escape from depth

The quantity N in Eq 1 is the ldquodepthrdquo given asa molecular column density from which O2 formedbelow the surface can percolate to the surface andescape It can be converted to a physical depth bydividing by the molecular number density of theice In the case of the low-energy electrons thedepth below the surface at which ionizations andexcitations are produced is small However forfast ions that penetrate many monolayers this isnot the case Below we examine N in Eq 1

Although it has long been understood that twoexcitations events are required to produce O2(eg Brown et al 1982 Reimann et al 1984) theprecursor model suggests that at saturation(steady state) any new event can produce O2 fromthe steady-state density of trapped precursorsTherefore one might expect the yield at satura-tion to be nearly linear in the excitation cross sec-tion This is the case for the low-energy electronsand it was initially assumed to be the case for thefast ions (Brown et al 1980 Reimann et al 1984)However Baragiola et al (2003) showed that thetotal sputtering yield in the saturation region attemperatures greater than 100K which is dom-inated by decomposition (Brown et al 1982) isproportional to the square of the electronic en-ergy deposition per unit path length in the solid(dEdx)e Assuming that the production and de-struction cross sections in Y` in Eq 4b roughlyscale with (dEdx)e then their ratio [sp sO2s]also roughly varies as (dEdx)e If this is the casethen the observed quadratic dependence impliesthat the depth N from which O2 can be mobi-lized to escape must also depend on the excita-tion density (dEdx)e This is consistent with theobservation that the O2 yield depends on thick-ness for samples thinner than the ion penetration

depth (Reimann et al 1984 Benit and Brown1990) Therefore the percolation depth for escapeN is a critical factor in comparing yields for pen-etrating radiation with those for non-penetratingradiation like the low-energy electron and low-energy ion data It is also seen that percolationand trapping can be included within the simplemodel considered here

Model summary

We have examined the simple precursor modelintroduced by Sieger et al (1998) to explain theirelectron data We have shown how it can be gen-eralized and derived from an integration of theappropriate chemical rate equations with thecross sections and total column density contain-ing the reaction rates and the physics of diffusivetransport Therefore the parameters can havevery different meanings from those initially as-sumed The equations are useful when there arefast processes which go to completion in timesmuch shorter then the time between particle im-pacts and slow processes associated with speciesthat are assumed to be trapped between particleimpacts This separation allowed us to examinethose models proposed for precursor formationand destruction due to reacting mobile speciesformed from dissociation events Further fromthe size of the onset of the linearity of the yieldversus fluence we can place constraints on therole of nonlinear processes such as those thatmight be associated with the formation of a pre-cursor like peroxide

With these extensions this simple model is use-ful for approximating the full rate equations inthe temperature region for which the yields de-pend on the fluence (dose) and not on the doserate Based on the assumptions in the model thisimplies temperatures at which precursor speciesare stable over experimental times At the high-est temperatures for the yields shown in Fig 4athis might not be the case This should be testedexperimentally In this regard it is puzzling thatexperiments have shown that the absorptionbands associated with certain trapped speciessuch as OH and trapped electrons ldquodisappearrdquoat 100ndash120K (see eg Johnson and Quickenden1997) whereas the O2 yield data indicate that rel-evant precursor species must be stable to 140KTherefore the proposed O2 production processescan only be distinguished by directly determin-ing the chemical state of the irradiated ice

JOHNSON ET AL838

It may be the case that a particular precursoris not required as mentioned earlier Rather af-ter a sufficient level of oxidation a newly pro-duced O or OH can react in the oxygen-rich lat-tice producing O2 in a pseudo-first-order processThis can be tested experimentally Indeedtrapped O and peroxide may be changed into oneanother as in the matrix isolation experiments(Khriachtchev et al 2000) Therefore their rela-tive concentrations versus temperature and doseshould be studied

In the model considered here it is seen that theparameters in Y` must account for the strong tem-perature dependence Since the number of bulktrapping sites decreases with increasing T as theice sample is annealed one might expect that theprecursor concentration would decrease We sug-gest that the observed temperature dependenceis determined by the competition between theavailability of trapping sites and mean interactiondistances of a mobile species such as O and HThat is the O2 yield varies inversely with the den-sity of traps over a range of temperatures In thesimple model above this effect is included indi-rectly as shown using the more detailed equa-tions in the Appendix

EUROPA

Oxygen and peroxide production at Europa G values

The dependence of the yield on the thicknessof the ice sample for highly penetrating ions sug-gests that the O2 yield is roughly related to theamount of energy deposited in the sample There-fore the amount of O2 produced and lost fromice in steady state is often given as a G value thenumber of molecules produced per 100 eV de-posited in Table 2 In applying these G values tothe planetary environment care must be takensince the yields are temperature fluence and ex-citation density dependent They also are affectedby the presence of other proton scavengersknown to be present in Europarsquos ice

The G values for peroxide in Table 2 are thosefor production at low fluence but the G valuesfor O2 are for the production and escape of gas-phase oxygen at saturation Therefore they de-pend on the formation and destruction cross sec-tions in the model above Such G values can beused to describe the steady-state source of the am-

bient O2 atmosphere at Europa but are not help-ful in determining the ability to produce and trapO2 at depth into the icy surface Over the timescales of the experiments some O2 is seen to re-main trapped and is not destroyed by radicalsTypically there are a number of trapping sitesand recent estimates of the activation energy foran O2 molecule trapped in a substitutional site(Hori and Hondoh 2002) suggest that it can betrapped for times comparable to the time re-quired for the conversion of amorphous ice intoIh (Ayotte et al 2001) The stability of O2 inclu-sions has not been studied

To understand the relationship between the Gvalues for O2 in Table 2 we note that fast lightions and electrons lose their energy predomi-nantly by producing ionizations and excitationsThe ionization events produce secondary elec-trons that can result in further ionizations Thenumber of ionization events Ni produced by afast light ion stopping in a water ice is Ni 5EoW where Eo is the initial energy and W is theaverage energy expended per ionization event27 eV in water ice Therefore radiolysis by fastions is often described as the net effect of theshower of electrons set in motion by the incidention In this manner the G values for productionof O2 from ice by energetic light ions and elec-trons can be roughly estimated as the sum of theeffects of all the secondary electrons using theyield Ye for 10ndash100 eV electrons That is if weignore the percolation depth N we can write thenumber of O2 produced as the number of ioniza-tion events times Ye which already has in it thedestruction processes That is G NiYeEo YeW 0006 O2100 eV

This estimate of G is seen to be an order of mag-nitude larger than the keV proton value for O2 inTable 2 but is much smaller than the G value forenergetic Ne1 These differences from the simpleestimate are consistent with the observation thatthe yield in the saturation regime for penetratingions varies nonlinearly with excitation densityThe difference from the keV proton G value islikely due to the inability of O2 to escape fromdepth before destruction On the other hand thelarger G value seen using MeV Ne1 is likely dueto the high density of excitation This can resultin more efficient production of O2 increased pro-duction of damage sites for forming O2 andorincreased transport (percolation) along the tran-siently heated track That is the ability of theformed O2 to percolate to the surface during ion

OXIDANTS AT EUROPA 839

bombardment plays an essential role in deter-mining the measured yields as discussed aboveThis effect needs to be evaluated for a planetarysurface subject to particle fluxes that are many or-ders of magnitude smaller than those used in thelaboratory

In the absence of O2 yields for all of the rele-vant ions and energies an estimate of the totalsupply of O2 can be obtained using the G valuesin Table 2 For the energetic oxygen and sulfurions we use G values for neon and argon ionsThese are lower bounds as the data were obtainedat temperatures much lower than the surface tem-peratures at Europa (80ndash110K) For the fast pen-etrating electrons and protons we use the valuefor 200 keV protons in Table 2 For the heavy ionswe used the Ne1 and Ar1 values For compari-son we also give an estimate using the G valuesobtained from the low energy electron data andthe total ionization as discussed above

The G values are combined in Table 3 with theelectron and ion energy flux at Europa Althoughthere is some deflection of the flow due to in-duced fields the net energy flux to the surface isnot reduced much from these values (Paranicaset al 2002) In fact locally produced pick-up ionswhich are formed from the ejecta and acceleratedto the surface can contribute (Ip 1996) Furtheralthough G values are typically independent ofthe incident angle the loss of O2 is a surfaceprocess and is enhanced with angle That is thenumber of O2 produced within the escape depthincreases with increasing angle to the normal(steeper than the inverse cosine) This enhance-ment is 25ndash4 (Jurac et al 2001) Ignoring suchcorrections the total production rate calculated

(23ndash62 3 109 O2cm2s) is a lower bound to theactual production rate from the icy regions of Eu-roparsquos surface Recent modeling of the O2 atmo-sphere suggest that such a source rate is required(Shematovich and Johnson 2001)

Earlier estimates of the O2 source rate were in-correct because O2 production was assumed to bea fixed fraction of the H2O sputtering yield Atthe ambient temperature an ejected H2O sticks ef-ficiently (Smith and Kay 1997) to the surface ofneighboring grains in the porous regolith reduc-ing the sputtering yield for H2O (Johnson 1989)This reduction does not apply to the O2 fractionof the yield enhancing its relative importance toH2O Here we assume that all of the gas-phaseO2 produced can contribute to the ambient at-mosphere even if it is produced at a grain sur-face that is at some depth into the porous regolith

Finally Europarsquos surface is not pure ice every-where That is much of the trailing hemisphereis a hydrated sulfuric acid containing smallamounts of sulfur and SO2 and the leading hemi-sphere contains low levels of carbon CO2 andcarbonates Sodium and potassium atoms aresputtered from the icy regions (eg Leblanc et al2002) O2 production from surfaces containingthese species has not been measured C and S areboth proton scavengers and the other species canbe sources of O as discussed so that the O2 yieldfrom such regions could be enhanced Howeverthere can also be competing processes that mightreduce these rates In fact recent HST observa-tions suggest that the O2 might not be produceduniformly across Europarsquos surface (McGrath et al2000) The analysis of these observations is pre-liminary and is not simple since the observed

JOHNSON ET AL840

TABLE 3 APPLICATION TO EUROPA

OxygenImplantation Energy flux production

Species 109(cm2 s)a 109 keV(cm2 s)(1) 109(cm2 s)

H (10 keV) 0015(1) 12 0084 (072)bO (10 keV) 0015(1) 18 054cS (10 keV) 0009(1) 30 090cElectrons 62 043 (37)bPlasma (10 keV) (137 amu) 13(2) 10 03c

Total 79 23 (62)UV photons (64 eV) 46 01

aReferences are as follows 1Cooper et al (2001) [similar results in Paranicas et al (2002)] 2Mauk et al (1996)bUsing G for 200 keV protons (upper bounds in brackets are based on the low-energy electrons data as a model for

the secondary electrons produced)cUsing G for 15 MeV Ne1 at 7K a rough lower bound

emissions depend on the local plasma propertiesHowever the efficiency of O2 production in con-taminated ices needs to be studied

Porous regolith O2 interactions on grain surfaces

Although the O2 atmosphere is principally pro-duced in the near surface layers and is eventuallylost to space it also permeates Europarsquos porousregolith The porosity p (void space) of the re-golith is uncertain (Buratti 1995) It has been es-timated to be the most compact surface in theOuter Solar System p 025 on the trailing hemi-sphere and p 079 on the leading hemisphere(Buratti and Veverka 1983) and the most poroussurface p 096 globally (Domingue et al 1991Domingue and Hapke 1992) These large differ-ences are due to the uncertainties in the amountof backscattered light at small phase angles andon the photon absorption properties and grainsizes The more compact surface is likely correctGrain sizes rg of the order of 100 mm have beensuggested (Hansen and McCord 1999 2000Geissler et al 1998) Europarsquos regolith differs sig-nificantly from the lunar regolith which is asand-like surface of individual grains Analysis ofthe thermal inertia of Europarsquos surface (Grundyet al 2001) suggests that the ice grains are sin-tered forming a fairy castle-like structure Sin-tering can occur thermally or due to the incidentradiation In addition ices doped with acids andalkalis in the laboratory can be powdery solidswith the impurities primarily found at grain in-terfaces

The porous regolith affects the above estimatesof the oxygen production in three ways First forisotropic bombardment the mean distance d intothe surface at which an incident ion or photonstrikes a grain surface is d (89)rg(1 2 p) Us-ing the above value for rg then d 02 m if theporosity is very large p 96 Therefore de-sorbed molecules can come from a considerabledepth and will interact with grain surfaces Evenfor much lower porosities the effective sputteringyields can be significantly modified (Johnson1989) As discussed above the yields for ejectedH2O which sticks onto neighboring grains at theambient temperatures are reduced but those forejected O2 are not This enhances the relative im-portance of the O2 yields

Finally the surfaces and interfaces of the sin-tered grains in Europarsquos porous surface are placeswhere low temperature chemistry involving the

ambient oxygen can occur at depths that are wellbelow the nominal penetration depth of the inci-dent radiation The relative importance of suchchemistry is roughly indicated by the ratio of thetotal surface area in the regolith to the area of Eu-roparsquos visible surface Using the porosity andgrain size above and a regolith depth dr of theorder of a meter then this ratio is large 3(1 2p)drrg which is 103ndash2 3 104 for average porosi-ties of 96 and 25 respectively Therefore thegas-phase O2 molecules below the visible surfaceinteract frequently with the grain surfaces In thisway a very large volume of the subsurface couldbe altered chemically even if it is not directly ex-posed to the radiation

From the overlying structures seen in the high-resolution images of Europarsquos surface by theGalileo spacecraft (eg Prockter and Pappalardo2000) it is clear that the youngest surfaces are red-der The reddening in the visible has been attrib-uted to the presence of sulfur in the ice (McEwen1986 Johnson et al 1988 Calvin et al 1995 Carl-son et al 1999b) The amount of reddening in thespectra of the surface material decreases with ageand the surface materials appear to becomebrighter with age This might occur by the slowaccumulation of a frost over-layer or by radioly-sis (Carlson et al 2002 Johnson et al 2003) Ra-diolysis preferentially converts the sulfur in theice to sulfur dioxide or sulfuric acid both ofwhich have flat spectra in the visible In regionswhere the radiation does not directly penetratethe O2 permeating the regolith can oxidize thesulfur at the surfaces of the icy grains and at graininterfaces Similarly the carbon seen in the ice onEuroparsquos leading hemisphere (eg Carlson et al2004 Johnson et al 2003) can be oxidized by ra-diolysis in the near surface layers or by reactionson the grain surfaces at depth in the regolith bythe ambient O2 The meteoroid bombardmentthat produces the regolith also exposes fresh sur-faces for oxidation and can cause the collapse ofpores leading to the trapping of oxygen In thismanner both refractory and volatile oxidants canbe formed and trapped in the subsurface ice forpotential delivery to Europarsquos subsurface ocean

SUMMARY

The presence of a thin oxygen atmosphere andoxidants trapped in Europarsquos surface ice (egJohnson et al 2003) combined with the possible

OXIDANTS AT EUROPA 841

presence of a subsurface ocean (eg Carr et al1998) have made Europa an exciting target forfuture space exploration The molecular oxidantsproduced by radiolysis and photolysis could bespecies needed for carbon-based biochemistry inEuroparsquos putative subsurface ocean Howeverthere has been no model for the formation trap-ping and transport of relevant oxidants

In this paper we have reviewed the space ob-servations and laboratory studies related to thechemistry of formation of oxidants in ice by theincident radiation Because the laboratory dataare incomplete we examined the rate equationsfor formation of oxygen and related species byradiolysis and photolysis We showed how suchequations can be simplified to the analytic modelthat was used for fitting the fluence and temper-ature dependence of the O2 yield produced bylow-energy electrons (Sieger et al 1998) In de-riving the analytic model from the more completerate equations the competition between forma-tion and destruction processes can be studied ascan the role of trapping sites escape from depthand temperature history These equations also al-lowed us to put incident photon electron and iondata on a similar footing for the first time

The model can be used to suggest measure-ments that are needed to describe the chemicalstate of the irradiated ice and to interpret new lab-oratory data We showed that if reactions are fastthe yield data could be consistent with a precur-sor being formed by two events as is the case forperoxide However we favor trapped O as a pre-cursor as initially suggested by Matich et al(1993) for production of excited O2 Such aspecies if present could be detected spectro-scopically as can trapped O2 through bands ac-tivated by lattice perturbations (Cooper et al2003a)

The available oxygen yield data were used tocompare G values for different incident particlesThese were then used to estimate the oxygen pro-duction in the radiation environment at EuropaThe lower bounds were consistent with the pro-duction rates obtained from a recent model of Eu-roparsquos atmosphere The porosity of the regolithwas also considered It not only affects the rela-tive sputtering yields for O2 and H2O but alsoprovides an enormously increased surface areaon which the ambient gas-phase oxygen can re-act with species in the ice That this might be oc-curring is consistent with the change in re-

flectance associated with surface aging Such re-actions could also convert ambient molecularoxygen into other more refractory species that re-main trapped on subduction of Europarsquos crust

Although radiolysis produces a stable chemi-cally altered surface in the laboratory the parti-cle flux to Europarsquos surface is many orders ofmagnitude lower Therefore annealing can be ef-fective at temperatures that are much lower thanthose typically found experimentally Belowabout 120K the annealing times appear to be long(eg Baragiola 2003) so that laboratory data canoften be directly applied to Europarsquos icy surfaceHowever account has to be taken of the syner-gism between the UV and the energetic plasmaas discussed earlier (Johnson and Quickenden1997)

Transient surface melting might initiate lifeprocesses but subduction of oxidants to the pu-tative ocean is likely to be required if biochem-istry is to occur at Europa (Chyba and Hand2001) Although the temperature versus depthinto Europarsquos regolith is not well known belowthe regolith the temperature and pressure appar-ently increase with depth At some depth (ap-proximately kilometers) either very warm ice oran ocean exists (Ruiz and Tejero 2000) Thereforechemically refractory oxidants are likely to be re-quired in order to survive subduction to thewarmer subsurface regions Trapped species ifdelivered downward can become mobile andwill segregate forming inclusions (Johnson andJesser 1997) Indeed peroxide appears to segre-gate in ice (Gurman et al 1967) and these inclu-sions may remain relatively stable until the per-oxide is dissolved in the putative ocean Thevolatiles seen in reflectance such as O2 SO2 andCO2 will also segregate But at the higher tem-peratures they could also diffuse through thebulk react or percolate into the regolith atmo-sphere Therefore the observation of the trappedvolatiles in Europarsquos surface ice is of interest pri-marily because it suggests that oxidants are pre-sent The formation of oxygen-rich molecules fol-lows the preferential loss of hydrogen Howeverthe surface is in a local steady state with im-planted H O and S from the plasma and the lossto space Since the plasma flux to the surface canbe variable this balance can also vary Indeedtemporal changes in reflectance have been re-ported (Domingue and Hendrix 2004)

Based on the models for formation of O2 ex-

JOHNSON ET AL842

amined in this paper additional experiments areneeded to determine the chemical state of irra-diated ice at low fluences Of immediate inter-est is the identification of stable molecular pre-cursors required for the formation of O2 Basedon the available laboratory data individual per-oxide molecules trapped in the ice do not appearto be the principal precursors but trapped O(Matich et al 1993) remains a viable candidateAlthough the data suggest that the local hydro-gen-bonding network is important in the pro-duction of O2 the effect of the formation tem-perature and radiation damage on the structureneed to be measured It is also important to de-scribe quantitatively the relationship betweenH2 loss and the formation of O2 and H2O2 overa range of temperatures and fluences Finallyone would like to determine the steady-statedensities of biochemically interesting moleculesformed by radiolysis in an ice containing sulfurand carbon at Europarsquos surface temperaturesand then determine their stability with increas-ing temperature and pressure Such experimentsstill need to be carried out Here we have pre-sented a model that brings together the variouslaboratory data and now can be used to considerthe formation trapping and transport of relevantoxidants at Europa

APPENDIX

Introduction

In this Appendix we examine rate equationsthat describe various aspects of the radiolysis orphotolysis of ice These equations need to be un-derstood if one is to be able to calculate the for-mation trapping and transport of oxidants Theemphasis is on the formation of molecular oxy-gen We integrate these equations over the escapedepth of O2 to give the simplified rate equationsdiscussed in the text in which number densitiesare replaced by column densities (eg Eqs 1 and2) In this way we can examine the relationshipbetween the rate constants and the effective pre-cursor formation and destruction cross sectionsextracted from data We first consider a precur-sor such as H2O2 that is formed from two exci-tation events This bears on the observed linear-ity of the oxygen yield versus fluence We thenconsider the equations for trapped versus a tran-

siently mobile species We use a reaction betweena mobile and trapped O as an example althoughO 1 OH can also give oxygen Finally we ex-amine the possible dependence of the O2 forma-tion process on the density of traps (defectsvoids and grain boundaries) as a function of tem-perature The results are discussed in the text

For clarity all possible radiolytic species arenot included in the discussion below Thereforepotentially important products such as HO2 andthe charged species (eg H3O1 the trapped elec-tron etc) are missing In each section we attemptto make a point related to production of O2 andother oxidants by using a subset of the productsrather than attempt a complete description withmany unknown rate constants

Fast versus slow processes

To describe the O2 production rate from a pre-cursor such as H2O2 we first use a simple but in-structive set of rate equations to describe the for-mation of H2O2 We then relate these to thesimplified rate equation for the precursor in thetext Eq 1 and show how the postulated precur-sor formation cross section sp is related to theH2O dissociation cross section sd by the reactionrates

Using the reactions

H2O 1 radiation R H 1 OH (sd)

H 1 H R H2 (k2H)

H 1 OH R H2O (kHOH)

OH 1 OH R H2O2 (k2OH)

the reaction pathways are described by

H2

Hqk2H

1 OHH2O 1 radiation mdashRsd H

OHH2O2

OH Rk2OH

kHOH

HH2O

OXIDANTS AT EUROPA 843

The corresponding rate equations in an irradiatedvolume are

dnOHdt 5 sd f n 2 2 k2OH nOH nOH

2 kHOH nH nOH (A1a)

dnHdt 5 sd f n 2 2 k2H nH nH

2 kHOH nH nOH (A2a)

dnH2O2dt 5 k2OH nOH nOH

2 destruction processes (A3a)

Here sd is the dissociation cross section f is theradiation flux and n is the molecular numberdensity of ice with nOH and nH the density of dis-sociation products OH and H respectively Therate constants k2OH k2H and kHOH describe thereactions above In this model the H2 formed isassumed to escape for temperatures above about20K We do not distinguish between trappedspecies and mobile species although in fact thatdistinction is important as discussed in the nextsection That is any OH produced at low tem-perature by one ion would likely be trapped priorto interacting with an OH produced by a subse-quent ion At higher temperatures this can breakdown as discussed in the text We first relate theabove set of equations to Eq 1 in the text and thenconsider precursor destruction processes

Equation 1 in the text is obtained by solvingthese equations together and then by integratingover a depth Dx Here Dx is the smaller of the pen-etration depth of the particle and the O2 escape(percolation) depth Assuming the molecular den-sity of ice n is constant with depth Dx n 5 N inEq 1 Comparing Eq 1 with Eq A3a the precur-sor formation rate sp f N is equivalent to [k2OH

nOH nOH] Dx Assuming that the processes in EqsA1a and A2a are fast then the steady state is read-ily obtained In addition if k2H k2OH there ispreferential loss of H2 at short times as suggestedby the laboratory data If steady state is reachedthe production of H2 is equal to that of H2O2 us-ing only the equations above Further one obtains

[k2OH nOH nOH] Dx R [fp sd] f N (A4)

with

fp 5 05[1 1 05 kHOH(k2H k2OH)05] (A5a)

Therefore the precursor formation cross sectionsp in Eq 1 is a fraction fp of the total dissocia-

tion cross sd Since laboratory data suggest theH2O2 fraction in an irradiated ice is small thenkHOH (k2OH k2H)05 which is likely givingfp R [(k2OH k2H)05kHOH]

Trapped versus mobile species

To examine the fluence dependence discussedin the text we note that trapped OH and freshlyproduced transiently mobile OH can be treatedas different species (eg Matich et al 1993)Therefore assuming no trapping of H the OHrate equation above would be replaced by twoequations one for the trapped species nOHt andthe other for the freshly produced mobile speciesnOH Assuming the reaction of two mobile OHunits is unlikely and that peroxide is formed fromthe interaction of a trapped OH with a subse-quently produced OH an expanded set of equa-tions is obtained

OH 1 trap R OHt (kOHt)

OH 1 OHt R H2O2 (kOHOHt)

H 1 OHt R H2O (kHOHt)

then the reaction pathways are described by

qH2

H2O 1 radiation mdashRsdH

qk2H

1 OHH

trap OH

OH RkOHt

OHt kOHOHtH2O2

kHOHkHOHt

H H2O HH2O

The resulting set of equations is

dnOHdt 5 sd f n 2 kHOH nH nOH

2 kOHt nOH nt 2 kOHOHt nOH nOHt (A1b)

dnOHtdt 5 kOHt nOH nt 2 kOHOHt nOH nOHt

2 kHOHt nH nOHt (A1b9)

dnHdt 5 sd f n 2 2 k2H nH nH 2

kHOH nH nOH 2 kHOHt nH nOHt (A2b)

JOHNSON ET AL844

dnH2O2dt 5 kOHOHt nOH nOHt

2 destruction processes (A3b)

where nt is the trap density in the irradiated iceAt low T we can assume nt is large so thatdntdt 0 simplifying the set of equations If theprecursor of interest is trapped O as discussed inthe text then a similar set of equations can be con-structed for O and Ot as in the example to followOf course a full description over a broad rangeof temperatures would require the considerationof the both trapped and mobile H O and OHthe rate of change of trap density and the vari-ous destruction processes

The above equations should be solved simul-taneously but for the incident particle fluxes typ-ically used in these studies the diffusion of mo-bile species to traps is relatively fast This is trueeven at low temperatures if there is a significantdensity of trapping sites This is usually the casefor samples formed at low T Therefore in a lowtemperature solid at low dose rates there is not asignificant steady-state background of diffusingOH There are trapped or freshly produced OHunits which rapidly react Averaging over thebombardment rate this can be approximated bywriting dnOHdt 0 and then separating nOH outin Eq A1b Assuming also that H diffuses rapidlyand either reacts or forms H2 we also writednHdt 0 on the average Therefore betweenimpacts only species trapped at defects pores orgrain surfaces are assumed to be present as sug-gested by the lack of a dependence on the inci-dent beam flux In this limit the rate of change oftrapped OH roughly simplifies to

dnOHtdt 5 2 k2H nH nH 2 2 kOHOHt nOH nOHt

It is seen that the production of peroxide dependson the loss of H by formation of H2 as discussedand at steady state the production rate for per-oxide equals the rate of formation of H2 in theseequations

Treating the mobile and trapped species sepa-rately we can further simplify the rate equationsand obtain the precursor model used in the textThat is integrating over the relevant depth Dxthe densities are replaced by column densitiesThe rate constants ki are then related to effectivecross sections si In this way cross sections aregiven as an interaction length for each freshlyproduced mobile radical [ki nj] R [si f] The

trapped OH column density NOHt and peroxideproduction rate can be estimated from ldquolinearrdquorate equations

dNOHtdt lt [sd f] N 2 2 [s2OH f] NOHt

2 [sHOH f] NOHt (A6a)

dNH2O2dt lt [s2OH f] NOHt 2 s NH2O2 (A6b)

Here sd is again the dissociation cross sectionand the destruction processes in Eq A3b are con-tained in s The second and third terms on theright in Eq A6a are the loss of OH due to the pro-duction of H2O2 and H2O In the integration ofEq A1b9 [2 kOH OHt nOH] becomes [2 s2OH f] and[kHOHt nOH] becomes [sHOH f] Therefore thedensity and the interaction length of mobile H arecontained in sHOH If H2O2 is indeed the pre-cursor then in Eq 2 in the text [sp f N] is be-comes [s2OH f NOHt] from Eq A6b Accountingfor precursor destruction which can give backtwo OH and treating H2 can both be included inthe integrated model via the si values

The simplified set of equations above allows usto examine the relative time (fluence) scales forprecursor formation (in this case H2O2) SolvingEqs A6a and A6b the column density of OH andthe concentration of precursors versus fluence cpcan be obtained Using the column of H2O2 pro-duced as the precursor column Np in Eq 2 thenO2 yield can be obtained The results can be com-pared with the results from Eqs1 and 2 in thetext The average precursor (H2O2) concentrationover the column depth of water molecules N is

cp 5 fp (sds)[1 2 exp(2s F)] 2 (ssr)

[1 2 exp(2sr F)] (A7)

with

fp 5 2s2OH(sr 2 s) (A5b)

In these equations sr 5 (2s2OH 1 sHOH) andssr 1

It is seen that the precursor concentration withtime cp depends on two rates the reaction ratesthat determine sr and the destruction rates in sAs in Eq 4a the yield can be written as

Y 5 sO2 cp N (A8)

If sr and s are comparable then the yield at small

OXIDANTS AT EUROPA 845

fluence is quadratic in F However if sr sthen the linear dependence on fluence is obtainedwith a small shift in the fluence as shown in Eq4c in the text Further the steady-state yield isnow

Y` (T) 5 sO2 [fp sd] Ns (A9)

That is sp R [fp sd] as in Eq A4 and the tem-perature dependence for the incident electrondata must be contained in [sO2 fp] or in the ratio[sO2 s2OH]sHOH]

For comparison with their precursor modelSieger et al (1998) examined the possibility of ob-taining the O2 precursor by considering reactionsfor a density of diffusing dissociation products asin Eqs A1 and A2 rather than a trapped productinteracting with a freshly formed dissociationproduct Based on the above the lowest fluence atwhich O2 is first detected experimentally sets alimit on the time scale for mobilendashmobile reactions

Effect of trap density

Equations A1b A1b9 A2b and A3b can be ap-proximately solved assuming that the dnOHdtdnHdt and dntdt are very small on the averageas discussed above In order to examine the roleof the trap density nt which can depend on thetemperature and method of formation and on theradiation history of the ice we use the slightlysimpler precursor model In this model the O pro-duced directly by dissociation traps forming aprecursor O2 is then formed when a freshly pro-duced O or other species reacts with a trappedoxygen atom Ot

H2O 1 radiation R H2 1 O (sd)

O 1 trap R Ot (kOt)

O 1 Ot R O2 (kOOt)

The reaction pathways for the oxygen species are

H2O 1 radiation mdashRsd9

trap OO2

kOtOt kOOt

H2 1 OOHt

kOR H(kHO t)

ROR

We assume for simplicity both H2 and O2 escapefrom depth Dx after formation We also allow mo-bile H to react with Ot (kHOt) removing trappedO and both the mobile H and O to be removedby other reactants R (eg OH or contaminants)This leads to a simplified set of equations not di-rectly involving OH or peroxide

dnOdt 5 sd9 f n 2 kOt nO nt 2 kOR nO nR

2 kOOt nO nOt (A1c)

dnOtdt 5 kOt nt nO 2 kOOt nO nOt

2 kHOt nH nOt (A1c9)

dnHdt 5 sd f n 2 kHOt nH nOt

2 kHR nH nR (A2c)

dnO2dt 5 kOOt nO nOt (A3c)

These equations of course act in parallel with theequations above for OH and mobile OH may re-act with trapped O to form O2 Here we focusonly on mobile O

Since the diffusion of H and O are fast and thedensity of traps is large the nH and nt averagedover the bombardment rate can be taken veryroughly as time independent Further if we as-sume the number of trapped O nOt is small com-pared with the number of sites at which H canreact then nH [sd f nkHR nR] For a low ra-diation flux f nH is small on the average as as-sumed Since the O2 yield YO2 is equal to[(dnO2dt) Dxf] with Dx the escape depth thenthe steady-state yield Y` can be obtained ana-lytically

[Y`][sd9 N] 5 1 1 (q 2 1) 2 [(q 2 1)2

1 8 q d]054 d (A10)

q 5 [kOt (nt 1 nR)] (kHOt nH)(sd9 f n) kOOt

5(sdsd9) [kOt (nt 1 nR)] kHOt[kOOt kHR nR]

d 5 05 [1 1 nt(nt 1 nR)]

We note that for nt large q can be large Assum-ing that q 1 and nt nR then Eq A10 be-comes

Y` R [sd9 N]q 5 (sd9 N) (sd9sd)

[kOOt kHR nR][kHOt kOt nt] (A5c)

JOHNSON ET AL846

Comparing with Y` in the simple precursormodel in Eq 4b then s R [(kHOt nH)f] R [sd nkHOt][kHR nR] In addition the product [sO2sp] R sd9 sd9 [kOOt n][kOt nt] That is the dis-sociation cross sections are modified by reactionrates

Quite remarkably it is seen in Eq A5c that thesteady-state yield Y` can depend inversely onthe density of traps nt in the temperature andflux regimes for which the approximations arerelevant That is if there are competing destruc-tion process for trapped O then at high trap den-sity a mobile O must find a trapped O before itbecomes trapped or before mobile H destroys theOt In this way the increasing O2 yield with in-creasing T can be related to the change in the trap-ping density nt with the formation temperatureof the ice sample As the temperature increasesand the density of traps in the bulk decreases thesurface and grain interfaces become the only trap-ping sites

ACKNOWLEDGMENTS

REJ acknowledges support by a Gledden Fel-lowship in Chemistry School of Biomedical andChemical Sciences at The University of WesternAustralia and by grants from the NSF Astron-omy Program and NASArsquos Planetary Geologyand Geophysics Program PDC acknowledgessupport by an Australian Postgraduate Award

ABBREVIATIONS

HST Hubble Space Telescope MJQ MatichndashJohnsonndashQuickenden NIMS Near Infrared Map-ping Spectrometer UV ultraviolet

REFERENCES

Ayotte P Smith RS Stevenson KP Dohnarsquolek ZKimmel GA and Kay BD (2001) Effect of porosityon the adsorption desorption trapping and release ofvolatile gases by amorphous solid water J Geophys Res106(E12) 33387ndash33392

Bahr DA Famaacute M and Baragiola RA (2001) Radiol-ysis of water ice in the outer solar system sputteringand trapping of radiation products J Geophys Res 10633285ndash33290

Baragiola RA (2003) Microporous amorphous water icefilms and astronomical implications In Water in Con-

fining Geometries edited by JP Devlin and B Buch El-sevier New York in press

Baragiola RA and Bahr DA (1998) Laboratory studiesof the optical properties and stability of oxygen onGanymede J Geophys Res 103 25865ndash25872

Baragiola RA Atteberry CL and Bahr DA (1999) Ori-gin of solid oxygen on Ganymede Reply to commentby RE Johnson on ldquoLaboratory studies of the opticalproperties and stability of oxygen on Ganymederdquo JGeophys Res E104 14183ndash14187

Baragiola RA Vidal RA Svendsen W Schou J ShiM Bahr DA and Atteberrry CL (2003) Sputteringof water ice Nucl Instrum Methods Phys Res B 209294ndash303

Bar-Nun A Herman G Laufer D and Rappaport ML(1985) Trapping and release of gases by water ice andimplications for icy bodies Icarus 63 317ndash332

Baverstock KF and Burns WG (1976) Primary pro-duction of oxygen from irradiated water as an expla-nation for decreased radiobiological oxygen enhance-ment at high LET Nature 260 316ndash318

Benit J and Brown WL (1990) Electronic sputtering ofoxygen and water molecules from thin films of waterice bombarded by MeV Ne1 ions Nucl Instrum Meth-ods Phys Rev B 46 448ndash451

Benit J Bibring J-P della-Negra S LeBeyec YMendenhall M Rocard F and Standing K (1987)Erosion of ices by ion irradiation Nucl Instrum Meth-ods Phys Rev B 1920 838ndash842

Brown WL Augustyniak WM Lanzerotti LJ John-son RE and Evatt R (1980) Linear and nonlinearprocesses in the erosion of H2O ice by fast light ionsPhys Rev Lett 43 1632ndash1635

Brown WL Augustyniak W M Simmons E Mar-cantonio KJ Lanzerotti LJ Johnson RE BoringJW Reimann CT Foti G and Pirronello V (1982)Erosion and molecular formation in condensed gasfilms by electronic energy loss of fast ions Nucl In-strum Methods B 1 307ndash314

Brown WL Augustyniak WM Marcantonio KJ Sim-mons EN Boring JW Johnson RE and ReimannCT (1984) Electronic sputtering of low-temperaturemolecular solids Nucl Instrum Methods B 1 307

Buratti B (1995) Photometry and surface structure of theicy Galilean satellites J Geophys Res 100(E9) 19061ndash19066

Buratti B and Veverka J (1983) Voyager photometry ofEuropa Icarus 55 93ndash110

Calvin WM and Spencer JR (1997) Latitudinal distri-bution of O2 on Ganymede observations with the Hub-ble Space Telescope Icarus 130 505ndash516

Calvin WM Clark RN Brown RH and Spencer JA(1995) Spectra of the icy Galilean satellites from 02 to5 mm a compilation new observations and a recentsummary J Geophys Res 100 19041ndash19048

Calvin WM Johnson RE and Spencer JA (1996) O2

on Ganymede spectral characteristics and plasma for-mation mechanisms Geophys Res Lett 23 673ndash676

Calvin WM Aninich VG and Brown RH (2004) Oxy-gen temperature and phase effects Icarus (in press)

OXIDANTS AT EUROPA 847

et al 1982 1984) found that at temperatures rel-evant to the surface of Europa H2 and O2 wereimportant components of the ejecta They thenused those data to predict the presence of a radi-olytically generated O2 atmosphere on Europa(Johnson et al 1982) The production efficiencyfor gas-phase O2 was found to be strongly tem-perature dependent as seen in Fig 2a for MeVHe1 ions incident on ice and Fig 2b for keV Ar1

ions on ice In Fig 2 the production rate is givenas a yield which is the number of moleculesejected per incident ion (ie O2 production ratedivided by the flux of incident particles photonselectrons or ions) Yields are typically measuredat a low incident particle flux so that sample heat-ing does not occur The yield versus temperatureshown in Fig 2 appears to increase exponen-tially with T over a range of temperatures Al-though this is not an Arrhenius dependencerough activation energies can be extracted overnarrow ranges of T (eg Fig 2a inset) Theseare of the order of 001ndash007 eV (Table 2) Suchvalues are much smaller than those for diffusionof vacancies or trapped species in ice (Livingstonet al 2002) but are comparable to the reportedactivation energies for H1 or H diffusion in iceas noted earlier (Johnson and Quickenden1997) The lowest energies are also comparableto the enthalpy difference between amorphoussolid water and crystalline ice (Ayotte et al2001) Therefore the estimated energies are likely

related to H1 and H transport or to hydrogenbond re-orientation It has also been shown thatas the temperature is increased above 80ndash100Kmorphological changes occur that can affect thereaction kinetics and give incorrect activation en-ergies This is especially so when using measure-ments of average quantities such as yields ratherthan directly extracting rate constants by mea-suring transients (Selby 2003) In addition the ac-tivation energies can be representative of the com-peting (destruction) processes rather than theformation processes (eg Matich et al 1993) asdiscussed in the model section

The measured O2 production rates (yields)were also found to be fluence dependent wherefluence is the radiation flux 3 the time or the to-tal number of particles per unit area onto the iceMultiplying the fluence by the incident particleenergy and dividing by the average penetrationdepth gives the energy per unit volume oftencalled an average dose As seen in Fig 3 the O2yield versus fluence rises from zero implyingthat a single ion incident on a fresh surface doesnot efficiently produce and eject oxygen This isthe case even for an incident 15 MeV Ne1 forwhich the excitation density along the ion path ishigh However it is seen in Fig 3 that with in-creasing irradiation time individual impactingions produce and eject O2 with increasing effi-ciency as the material is chemically altered Atlarge fluences the yield appears to saturate indi-

OXIDANTS AT EUROPA 827

FIG 2 a Yield for D2 and O2 at saturation (steady state) from 15 MeV He1 on D2O ice versus the temperature ofthe ice (adapted from Brown et al 1982) Inset D2 yield minus the constant yield at low T versus 1kT giving an av-erage activation energy of 003 eV b Yield of O2 at saturation versus temperature for incident 30 keV Ar1 ions(adapted from Baragiola et al 2003)

a b

cating that destruction processes are in a roughsteady state with formation processes The yieldsshown in Fig 2 are purportedly those after satu-ration although full saturation is often ap-proached slowly

The slope of the O2 yield versus fluence shownin Fig 3 increases with increasing temperaturegiving the activation energies in Table 2 Thesevalues are similar to those obtained from Fig 2For energetic ions that fully penetrate the film asin Fig 2a and 3a the yield versus fluence can ex-hibit a second rise at higher fluences for whichthe radiation track cores begin to overlap This in-crease also depends on temperature and exhibitssimilar activation energies (Reimann et al 1984)The strong dependence on temperature is at firstsurprising since the energy density deposited bya 15 MeV Ne1 produces transient temperaturesin the track core that are much higher than theambient material temperature Therefore tracktemperature (local excitation density) alone doesnot determine the O2 yield

For all incident particle types and energies themeasured fluence dependence is found to be sta-ble over laboratory time scales That is for ex-periments carried out at high vacuum the yieldreturns to the value it had when the beam wasturned off This indicates that the irradiated ice isldquopermanentlyrdquo altered (Reimann et al 1984)Thus a relatively stable chemical state is createdthat is favorable for O2 production This could bethe formation of a particular trapped radiationproduct in ice that acts as a stable precursor or isdue to the altered chemical and physical state ofthe irradiated ice Trapped O H2O2 and HO2(Matich et al 1993 Johnson and Quickenden1997 Sieger et al 1998) have been proposed asstable precursor species

Using isotopically labeled H2O Benit andBrown (1990) showed that considerable transportof O occurs along the ion track when an energeticheavy ion penetrates an ice sample Because ofthis the O2 yield was found to depend on sam-ple thickness Therefore for thin ice samples the

JOHNSON ET AL828

TABLE 2 OXYGEN AND PEROXIDE PRODUCTION

Radiationa Product T (K) Ea (eV) Percent at saturation G (product100 eV)b

He1(1) O2 015g(23) O2 77 (300) 0003 (0)Ne1 (30 MeV)(3) O2 300 0031He21 (6 MeV)(3) O2 300 0008He21 (29 MeV)(3) O2 300 00016H1 (80 keV)(4) O2 70 00002H1 (200 keV)(5) O2 120 00007He21 (15 MeV)(6) O2 50 (100) 003 (07ndash4) 3 1024

Ne1 (15 MeV)(7) O2 7 10ndash15c 003(8) 70ndash140 005ndash007 01

Ar1 (01 MeV)(9) O2 30ndash140 001ndash004 0006ndash008160 001

e (30ndash100 eV)(10) O2 120 002ndash003 0005ndash0007hn (102 eV)(11) O2 50 (100) 003 00001 (00003)H1 (1 MeV)(12) H2O2 16 (77) 01 (001)

1 10 CO2 77 011 10 O2 77 04

H1 (30 keV)(13) H2O2 16 (77) 022 (038) 0098 (0086)C1 (30 keV) H2O2 16 (77) 17 (22) 023 (034)N1 (30 keV) H2O2 16 (77) 16 (14) 021 (011)O1 (30 keV) H2O2 16 (77) 14 (40) 026 (015)Ar1 (30 keV) H2O2 16 (77) 22 (20) 011 (006)

aReferences are as follows 1Lefort (1995) 2Ghormley and Stewart (1956) 3Baverstock and Burns (1976) 4Baragiolaet al (1999) 5Baragiola and Bahr (1998) 6Brown et al (1982) (ion does not stop in sample) 7Benit and Brown (1990)(ion does not stop in sample) 8Reimann et al (1984) (ion does not stop in sample) 9Baragiola (2003) 10Sieger et al(1998) (steady-state value) 11Westley et al (1995) (steady-state value as no fluence dependence seen) 12Moore andHudson (2000) 13Gomis et al (2003)

bG values are for escaping O2 For H2O2 and O2 formation both of which require multiple events yields after sat-uration are used In a few cases yields are obtained by melting the sample Values in parentheses are those at anothertemperature

cPercent of film eroded as O2 molecules for long-term irradiation

ldquosaturatedrdquo yields change slowly with fluenceMore importantly O from water molecules at aconsiderable depth into the sample contributedto the O2 ejected from the surface Since there wasno observed diffusion of the O2 in ice at the am-bient temperature (7K) over the time for their experiment they concluded that the oxygentransport and production are stimulated by theincident ions The transport was suggested as be-ing due to percolationdiffusion along the dam-aged and transiently heated tracks (Reimann etal 1984 Benit and Brown 1990) They also foundthat in fully eroding (sputtering) thin ice filmsthe total amount of O2 produced required about10ndash15 of the D2O molecules in the films at T 57K On warming irradiated samples some O2

formed at depth is found to remain trapped forexperimentally relevant times (see eg Hart andPlatzman 1961 Benit and Brown 1990 Baragiolaand Bahr 1998)

Matich et al (1993) detected luminescence in77K H2O and D2O ice from newly formed and ex-cited O2 They suggested that the O2 is formed

from the interaction of a freshly produced mo-bile O interacting with a previously producedand trapped O or H2O2 They favored the formerIn matrix isolation studies of peroxide irradiatedby UV photons luminescence from newly formedexcited O2 was seen However the oxygen wasformed from mobile O (greater than 30K in Krlattice) interacting with O trapped as H2O-O(Khriachtchev et al 2000) These studies also sug-gest that trapped O possibly in the form H2O-Omay be an important precursor

In Fig 4a the O2 yield for low-energy electronsplotted as a function of electron energy exhibits anapparent ldquothresholdrdquo at 10 eV (Orlando andSieger 2003) Although the uncertainties are largethe authors suggested there may be a lower thresh-old 6 eV corresponding to a very small crosssection for O2 production (10220 cm2) This couldbe associated with an excitation at the ice surfaceBulk ice exhibits very weak absorption at photonenergies below 6 eV but exhibits a strong ab-sorption that depending on the ice structure be-gins at about 73 eV with a peak at 86 eV This

OXIDANTS AT EUROPA 829

b

a

FIG 3 a Relative yields versus fluence (ion flux 3 time) for 15 MeV Ne1 incident on D2O ice at 7K (adapted fromReimann et al 1984) Top panel D2O yield Middle panel D2 yield Bottom panel O2 yield Yields at other tem-peratures exhibited similar dependences with the slope of O2 yield versus fluence varying with T Fits give activa-tion energies between 002 eV at the lowest fluences and lowest T to 007 eV at the highest T and highest fluencesb Oxygen yield versus fluence of low-energy electrons on D2O ice (from Sieger et al 1998)

is followed by a peak in absorption at 10 eVThe results in Fig 4a show that there is a dra-matic increase in the production of O2 at 10 eVwhich is the onset of secondary electron produc-tion in the bulk (Orlando and Sieger 2003)

For incident fast ions the yield at a given tem-perature was found after a small fluence to beindependent of whether the sample was crys-talline or amorphous ice (Brown et al 1982)However using low-energy electrons the yieldswere found to be smaller for an initially crys-talline ice (Orlando and Sieger 2003) Since verylow doses of ion radiation (a few eV per mole-cule) can convert crystalline ice to amorphous iceat low temperatures (Leto and Baratta 2003) thedamage produced by the penetrating ions usedby Brown et al (1982) affected the O2 yield Thestructural damage produced may create trappingsites cause changes in the electronic relaxationpathways or enhance percolation

The temperature dependence of the O2 yield isshown in Fig 5 for electrons incident on initiallyamorphous (Fig 5a) and initially crystalline (Fig5b) ice A small decrease in efficiency is seen inboth cases at 120ndash130K where Ic forms fromamorphous ice and pore collapse occurs (Row-land et al 1991) This is followed by an increas-ing yield with increasing T and then by a more

dramatic drop near 160K where Ih forms Dur-ing the transition to Ih trapped molecules havebeen shown to escape efficiently from ice samples (Ayotte et al 2001) A large drop in the yield isalso seen in Fig 2b for incident Ar1 ions

In modeling the dramatic decreases in the yieldin Figs 2b and 5 care must be taken That is thetime for crystallization of an amorphous sample(Baragiola 2003) even allowing for the large un-certainties is 30 min at 145K and becomes 1min at 160K Therefore pulsed radiolysis exper-iments should be carried out at these tempera-tures and the transients observed Since the timefor conversion from amorphous to crystalline icecan be shorter than the experimental times at145K the results are not comparable to thosefor which the chemical state of the ice at a givenfluence is stable over experimental times Belowwe focus on the data below 140K

We have emphasized processes initiated by theelectronic excitations and ionizations producedHowever momentum transfer collisions whichare the dominant energy loss process for low-en-ergy ions also produce dissociations defects andO2 in ice (Bar-Nun et al 1985) The relative effi-ciency of momentum transfer and electronicprocesses for producing O2 needs further studyAs the penetration depths are much smaller when

JOHNSON ET AL830

FIG 4 a Relative yield of O2 from D2O ice versus incident electron energy E at T 5 120K (from Orlando et al1998 Orlando and Sieger 2003) b Relative yield of D2 from D2O ice versus incident energy E at T 5 88K (fromKimmel et al 1994)

a

b

the former dominates the produced O2 likely es-capes readily

H2 formation and loss correlation with O2

The production of O2 from a low temperatureice by radiolysis correlates with the formationand loss of H2 (or D2 from D2O ice) (Reimann etal 1984) D2O is used to reduce the backgroundin mass spectrometry studies The yields of O2from D2O and H2O ices were shown to be com-parable However the total yields from D2O wereabout 10 less (Reimann et al 1984) consistentwith H (D) transport being important It is seenin Fig 3 that D2 is promptly produced and lostfrom the D2O ice At higher fluence the D2 yieldis seen to exhibit a dependence like that for O2For the 15 MeV Ne1 ions which fully penetratethe ice sample the D2 yield eventually decreasessince D2 is produced and escapes from the fullthickness of the film

Because of H2 (D2) loss the surface region of

the irradiated ice becomes depleted in H (D) rel-ative to O favoring the formation of oxygen-richspecies The change seen in the ratio of H (D) toO in the irradiated ice is small An irradiated film(100 nm) remains stoichiometric to better than2 (Brown et al 1980) and in the near surface re-gion Auger spectroscopy suggests only a slightoxygen enhancement due to irradiation by keVelectrons (Rye et al 1978) Sieger et al (1998) sug-gested a precursor concentration of the order ofa percent Such concentrations are an order ofmagnitude smaller than the steady-state concen-tration of dissociated H2O (Table 2)

Using infrared spectroscopy the damaged H2Oconcentrations at low temperatures in the satu-ration regime are about 10 over the irradiatedvolume (Benit et al 1987 Watanabe et al 2000Gomis et al 2003) This fraction is the ratio of thedissociation cross section to the effective crosssection for recombination to H2O as discussedlater At a 10 concentration most H2O will havea neighbor that is dissociated

OXIDANTS AT EUROPA 831

FIG 5 a Relative steady-state yields of O2 from a D2O ice sample versus T (upper panel) The ice samples wereformed and irradiated at these temperatures Film erosion as a yield (lower panel) (from Orlando and Sieger 2003)b Relative steady-state yields of O2 versus T as in (a) (upper panel) Here the D2O ice samples are formed at 160Kpresumably becoming crystalline and then irradiated at the T shown A new sample is used for each data point in(a) and (b) Film erosion as a yield (lower panel) (from Orlando and Sieger 2003)

At temperatures above 40K it is seen in Fig2 that the D2 and O2 yields exhibit the same tem-perature dependence suggesting a correlationbetween D2 and O2 loss H2 (D2) can be formedby reactions between diffusing hydrogen fromdissociated water or directly [ie H2O R H2 1O (eg Kimmel and Orlando 1995) H2O R (H1 OH)cage R H2 1 O (eg Matich et al 1993)H3O1 1 e R H2 1 OH (eg Williams et al1996)] Subtraction of the nearly constant D2 yieldin Fig 2a gives the Arrhenius plot in the insetSince molecular hydrogen readily escapes atthese temperature it is the H2 (D2) formationprocess that has an activation energy of 003eVmolecule The fact that slopes of the yields forH2 and O2 above 40K are similar suggests thatH2 (D2) formation and loss is the rate-limitingstep in determining the production and loss of O2at these temperatures

Although the data at low T in Fig 2a have con-siderable uncertainties below about 20ndash40K theD2 production is roughly independent of tem-perature but the O2 yield continues to decreasewith decreasing temperature Radiolytically pro-duced D2 is mobile and can escape above15ndash20K (Watanabe et al 2000) but O2 is im-mobile at such temperatures and the incidentHe1 ions are much less efficient at causing per-colation than the incident Ne1 discussed earlier

The production of H2 (D2) from ice has alsobeen studied using low-energy electrons and UVphotons at very low temperatures In the formerstudies absolute yields were not obtained butthere appeared to be a threshold at 63 eV (Fig4b) This is roughly the threshold for excitationof a surface state (Khusnatdinov and Petrenko1992) and the threshold for peroxide productionis close to this value The efficiency for H2 (D2)production increases around 11 eV with a sec-ond steep rise at 17 eV (Kimmel et al 1994) TheO2 and D2 yields versus electron energy in Fig 4exhibit some similarity in the threshold regionalthough the uncertainties are large and the sam-ple temperatures different Measuring the H2 (D2)and the O2 ejecta simultaneously over a range oftemperatures as in the Ne1 experiments wouldbe instructive

Using 98 eV photons at low T (12K) Watanabeet al (2000) measured the D2 ejected and trappedin ice They obtained a cross section for produc-tion of D2 (Table 2) at low fluences that is10ndash20 of the total absorption cross sectionwith significant uncertainties They also found

that the equilibrium damage fraction in the sam-ple was 10ndash20 of the D2O molecules but theequilibrium production of D2 was about an orderof magnitude smaller 1ndash2 at their tempera-tures They suggested that the D2 is formed atgrain interfaces and cracks and on the surfaces ofpores In a mixed ice D2O 1 CO exposed to UVphotolysis D2 and CO2 formation are correlated(Watanabe and Kouchi 2002) The D2O dissoci-ates to produce D2 plus O and this transientlymobile O reacts with CO to produce CO2

Precursor state and peroxide

In order to produce O2 from low temperatureice it appears that the ice sample must lose hy-drogen and form trapped oxygen-rich reactivespecies The efficiency of production also appearsto be affected by the level of defects and damagein the ice consistent with the suggestion that thehydrogen-bonding network is important in theproduction of O2 via availability of trapping sitesnew electronic relaxation pathways or the natureof the bonding of precursors Here we briefly ex-amine the possible role of the suggested precur-sor molecules H2O2 and HO2 We note that H2O2might not readily dissolve in ice possibly ac-counting for its low abundance in some experi-ments It appears to be heterogeneously distrib-uted in the form of microcrystals (Gurman et al1967) If H2O2 does act as a precursor the for-mation of aggregates could affect the dimer ab-sorption bands (Cooper et al 2003a) and affectthe production and stability of O2 (Cooper et al2003b)

The results of Gerakines et al (1996) in whichice at 10K is irradiated by UV photons confirmthat H2O2 forms in ice from two OH after a build-up of trapped OH They monitored the band as-sociated with H2O-HO (Langford et al 2000) inwhich the HO is bound to water with 024 eVTherefore they might only be monitoring a sub-set of the trapped OH seen in earlier experiments(eg Taub and Eiben 1968) The H2O2 concen-tration in ice reaches a steady-state level at flu-ences of about 1018 photonscm2 (Gerakines et al1996) As seen in the model section this gives anestimate of the H2O2 photodestruction cross sec-tion of the order of 10218 cm2 Similarly HO2 be-gins to be seen in absorption at about a few times1017cm2 shortly after they begin to detect H2O2The absolute yields were not measured

Westley et al (1995) detected H2 and O2 pro-

JOHNSON ET AL832

duced by Lyman-a (98 eV) radiation of ice butfound no fluence dependence for fluences greaterthan 1017cm2 They also saw small amounts ofH2O2 by programmed thermal desorption of thesample but made no quantitative evaluationComparing their lowest fluence for the produc-tion of O2 with the result of Gerakines et al (1996)for peroxide suggests that H2O2 and HO2 areprobably not important precursors of O2 In UVphotolysis of H2O2 using 64 eV photons the pro-duction of O2 was not observed (Vaghjiani et al1992) and the excitation of an irradiated icewhich presumably contained some peroxide by37 eV photons did not produce O2 (Baragiola etal 2003) Direct excitation of matrix-isolatedH2O2 at low temperature also did not yield O2but the production of a mobile O that reacted withtrapped O (H2O-O) did give O2 (Khriachtchev etal 2000)

Irradiating with 200 keV protons at tempera-tures of 40ndash129K Bahr et al (2001) found onlyabout 0001ndash002 H2O2 using programmed ther-mal desorption as compared with earlier experi-ments that suggested larger values (eg Table 2)Using infrared absorption Moore and Hudson(2000) found a G value of 01 for incident MeVprotons at low temperature but the signal in theobserved band became negligible at 80K sug-gesting that H2O2 was not being formed Theyalso found that adding O2 significantly increasedthe production efficiency of H2O2 That is thatthe amount of peroxide was affected by theamount of trapped oxygen Based on the matrixisolation studies (Khriachtchev et al 2000) dis-sociated mobile O might trap as H2O-O whichon excitation converts to H2O2 Therefore even ifH2O2 is not the precursor the peroxide andtrapped oxygen concentrations will be related

Although molecular oxygen was not a productof photolysis of matrix isolated H2O2 it was aproduct of photolysis at 266 nm (466 eV) of per-oxide dimers in an argon matrix 2 H2O2 R 2 H2O1 O2 (Engdahl and Nelander 2002) This de-composition was thought to be due to the con-siderable distortion of the bonds in the dimerTherefore the configuration or distortion by thelattice of peroxide or any other precursor mole-cule in the ice may be critical (Sieger et al 1998)In addition more recent experiments show thatG values for the production of H2O2 by 30 keVincident ions are significant as are the steady-state ratios of H2O2H2O (Gomis et al 2004) Ra-tios of the order of 02ndash4 were found at 16K and

77K in the irradiated regions with incident H giv-ing the smallest fraction and incident O thelargest (Table 2) At these concentrations perox-ide molecules might segregate forming inclu-sions (Gurman et al 1967 Cooper et al 2003b)Since the amount of trapped peroxide also com-petes with the amount of trapped O studies areneeded to describe the chemical state of the solidproduced by radiolysis This will be critical in de-scribing the fate of the irradiation products at Eu-ropa In particular the chemical composition fa-vorable for producing O2 needs to be determinedas does the importance of trapped O H2O2 HO2or peroxide dimers as precursors

Other species

In the model below we will refer to atomic andmolecular species trapped in ice Such species areoften discussed in the literature on radiolysis andphotolysis (for summaries see Matich et al 1993Johnson and Quickenden 1997) Trapped OH OH2O2 and HO2 have been discussed andor mea-sured by a variety of techniques (absorption andluminescence spectroscopy electron spin reso-nance programmed thermal desorption etc)Care must be taken when using these data as re-sults for ice samples formed below 160K orthose exposed to penetrating ion beams likelyhave numerous defects voids and grain inter-faces so that a variety of trapping sites and trapdepths exist Therefore when a particular bandfor a trapped species (eg OH) disappears withincreasing temperature it could be due to achange in the nature of the trapping as well as toloss of the species by reaction

Although H is discussed below we do not dis-tinguish it in these simple models from H1

trapped as the hydronium ion H3O1 whichwould mean that there are corresponding nega-tive ions present Therefore below we refer to thevarious species as if they are neutral althoughthis might not be the case Although trapped elec-trons are detected only over a narrow range oftemperatures (100ndash120K) negative ions must beimportant over a broad range of temperature

We also note that SO2 and a sulfate and CO2and a carbonate appear to be trapped species inEuroparsquos icy surface Under radiolysis each ofthese can dissociate directly giving O2 Howeverwith the exception of the sulfate their densitiesare less than 1 so they are not relevant sourcesof O2 However their presence is indicative of the

OXIDANTS AT EUROPA 833

preference for forming oxygen-rich molecules inEuroparsquos ice and can affect the efficiency of pro-duction of O2 and H2O2

MODELS FOR FORMATION OF O2

Introduction

Based on the data described above the yieldsfor incident electrons and ions are found to benearly linear in fluence at the lowest fluences forwhich O2 is detected In addition the relevant al-terations of the ice samples by radiolysis appearto be stable and independent of the beam fluxRemarkably this is the case both for low-energyelectrons that make only a single excitation foreach impact and fast ions that make a density ofexcitations along their path through the solidThis suggests that O2 is produced and trapped inthe initial events and then caused to escape fromthe solid by the subsequent radiation or that astable chemical alteration of the ice is required be-fore the radiation can produce and eject O2

Because H produced by the dissociation of wa-ter molecules can attack trapped O2 and any pre-cursor oxidants the loss of H is important Thisoccurs by the production and loss of hydrogenmolecules as discussed leaving a surface that isslightly rich in O relative to H The O2 produc-tion mechanism could be direct excitation of aparticular precursor species or it might not bespecific That is a new OH or O produced by theincident radiation could react with the H-depleted solid producing O2 This would occurmore efficiently as the ratio of O to H in the nearsurface region increased

In their early work WL Brown and colleaguesassumed that processes similar to those used todescribe radiolysis of liquid and gaseous H2Ocould be used to describe the radiolysis in ice Thelow activation energy that they found was as-sumed to be associated with diffusion of somemobile radical species in ice such as H or O orpossibly diffusion of the O2 molecule itselfBased on their studies of ejection from isotopi-cally labeled D2O samples they concluded thatthe O2 is formed in the track of a single ion andthen ejected by subsequent ions (eg Benit andBrown 1990) More recently trapped specieshave been suggested as precursors These are ei-ther subsequently excited by the incident radia-tion or attacked by newly produced mobile

species We consider a simple model that incor-porates all of these processes and can be used toorganize and interpret the experimental data Wealso discuss its relationship to more completemodels In this discussion much of which is in theAppendix we consider the standard radiationproducts that have been often described How-ever in a solid the excited electron cloud pro-duced by absorption can overlap a number of lat-tice sites Therefore their relaxation pathwaysandor product species could be very differentfrom those that have been traditionally consideredin the literature on radiolysis and photolysis

Precursor model

The column density of a precursor species Npformed in ice by incident radiation can be calcu-lated using a simple model In this model cross sec-tions are used for the production sp and destruc-tion s of the precursor O2 is then formed byexcitation of the precursor or by reaction of a newlyformed radical with the precursor This is also de-scribed using a cross section sO2 For incident elec-trons having a small penetration depth the O2 pro-duced is assumed to rapidly diffuse to the surfaceand escape (Sieger et al 1998) More generally wedefine N as the column density of H2O from whichthe incident radiation can produce and eject O2 Apair of rate equations gives both Np and the pro-duction rate of oxygen dNO2dt That is

H2O 1 radiation R precursor

precursor 1 radiation R O2

giving

dNpdt 5 sp f N 2 s f Np (1)

dNO2dt 5 sO2 f Np (2)

In Eq 1 f is the incident flux and the total pre-cursor destruction cross section can be written ass 5 sp9 1 sO2 with sp9 accounting for precursordestruction processes other than the formation ofO2 Solving Eqs 1 and 2 and assuming N is nearlyconstant (10 level as discussed) the fractionof precursors cp in the affected column after timet is obtained (Appendix)

cp 5 NpN 5 [sps] [1 2 exp(2 s F)] (3)

Here F 5 [ft] is the fluence (flux 3 time) which

JOHNSON ET AL834

is the number of particles (photons electrons orions) incident on the surface per unit area Usingcp in Eq 2 the O2 yield Y at any fluence can bewritten in terms of the yield at steady state Y`

Y 5 [dNO2dt]f 5 sO2 cp N 5 Y`

[1 2 exp(2sF)] (4a)

Y` 5 [sp sO2 s] N (4b)

If we also account for the reduction in the den-sity of water molecules in the initial column Nthese forms change only slightly The size of thetwo parameters Y` and s can be obtained by fit-ting the yield versus fluence data such as that inFig 3

For the low-energy electron data Sieger et al(1998) showed that the ratio [YY`] for a givenincident electron energy is independent of tem-perature over a narrow range of temperatureTherefore the destruction cross section s in Eq4a is roughly independent of T and can be esti-mated from the [YY`] curve Sieger et al (1998)found a precursor destruction cross section s lt08ndash30 3 10216 cm2 for incident electron energiesfrom 30 to 100 eV This is of the order of the elec-tron impact ionization cross section for H2O

Based on the above model the lack of an ob-served fluence dependence for fluences down to1017 hncm2 in the Lyman-a irradiation of ice(Westley et al 1995) can be examined The low-est fluence used implies that the precursor de-struction cross section for Lyman-a is of the or-der of the Lyman-a absorption cross section in ice(09 3 10217 cm2)

Sieger et al (1998) also assumed that sp9 5 0This would mean that the precursor is only de-stroyed by production of O2 (ie s 5 sO2) Thatis unlikely since suggested precursors can be de-stroyed by reaction with H and other radicalsTheir assumption would also mean that the crosssection for production of O2 from a precursorsO2 is large This would be surprising and ap-pear to contradict some of the data discussedabove Instead we assume that s is not equal tosO2 and the total destruction cross section s isof the order of the ionization or absorption crosssection

Having obtained s the measured temperaturedependence for production of O2 is containedonly in Y` Over a narrow range of T one canwrite Y`(T) lt Y`9 exp(2EakT) and obtain the es-

timate of the activation energies in Table 2 Ifsp9 THORN 0 as discussed then sp and sO22 in Eq 4bcannot be obtained separately from the data forY` unless an independent measure of cp is ob-tained Such a measurement is critical for de-scribing the chemical state of the irradiated sur-faces

The observed temperature dependence of Y` iscontained in sp sO2 or N Here N is the ldquodepthrdquofrom which newly formed O2 can reach the sur-face and escape given as a column density of H2OThe physical depth is obtained by dividing thecolumn density N by the molecular number den-sity of the ice Assuming that in the low-energyelectron experiments N is independent of tem-perature and is about three monolayers (Sieger etal 1998) the measured Y` gives [sO2sps]05ndash20 3 10218 cm2 over the electron impactenergy range At 110ndash120K this ratio was foundto be proportional to (E 2 Eo) for E Eo 10 eVagain assuming fixed N (Orlando et al 1998 Or-lando and Sieger 2003)

In the following section we first consider a pre-cursor like peroxide that typically forms fromtwo dissociation events We use this to show un-der what circumstances such a model does notviolate the observed linearity of the O2 yield datawith fluence Following that we relate the sim-ple precursor model above to the case wheretrapped O is the precursor We then use themodel for trapped O to examine the effect of thetrap density on the production of O2 The densityof traps (defects) is determined by the sample for-mation temperature and the subsequent radiationdamage Finally we consider the escape depthN for penetrating radiation within the precursormodel These discussions are supported by moredetailed rate equations in the Appendix

Precursor trapped peroxide

The proposed precursor peroxide is typicallyformed from products of two dissociation eventsie 2 H2O R 2 H 1 2OH R H2 1 H2O2 This wasconfirmed in low temperature ice (Gerakines etal 1996) It is also possible that a dissociation pro-ducing a trapped O followed by an excitationevent can produce peroxide 2 (H2O 1 hn) R H2

1 H2O-O 1 hn R H2O2 1 H2 (eg Khriachtchevet al 2000) The O2 yield versus fluence for inci-dent electrons is reasonably well fit by the yieldexpression above for fluences greater than1014cm2 Therefore it has been suggested that

OXIDANTS AT EUROPA 835

any precursor species must be formed in a singleevent Using a model in the Appendix in whichthe precursor is formed by two events we showthis is not necessarily the case Precursor forma-tion if it is fast could involve two dissociationevents and give the measured fluence depen-dence Although we consider peroxide here theseconclusions apply to other possible precursors

In the example in the Appendix H2O is disso-ciated to H 1 OH according to the cross sectionsd The reactions of two OH units to form H2O2(rate constant k2OH) and 2 H to form H2 (k2H)compete with the recombination of H and OH toH2O (kHOH) If k2H is much greater than k2OH then H2 loss dominates at low fluences as ex-pected (Appendix) Steady state requires not sur-prisingly that the rate for forming H2O2 isroughly equal to the rate for forming H2 Finallyit is seen in the Appendix that the low H2O2 con-centrations suggested by experiment imply thatkHOH (k2H k2OH)05 Such sizes are reasonable

The set of rate equations in the Appendix thatuse number densities of species is integrated overdepth in order to obtain the forms used in Eqs 1and 2 which are based on column densities Inthis way the effective cross sections are related tothe rate constants for the interaction of freshlyproduced mobile species with previously pro-duced trapped species We show in the Appen-dix that the O2 yield versus fluence can be writ-ten in a form similar to that in Eq 4a but with asmall offset due to the two-step precursor for-mation process

Y lt Y`(T) [1 2 exp(2sF) 2 (ssr)]

for F sr21 (4c)

Here s is again the precursor destruction crosssection but sr 5 (2s2OH 1 sHOH) is the effectivecross section for loss of OH by formation of H2O2

and H2O The corresponding expression for Y`(T)is given in the Appendix

The form for the yield versus fluence in Eq 4cis consistent with the low-energy electron mea-surements if (ssr) 1 and if sr F21 lt10214 cm2 Such sizes are indeed possible Whatis required is that at the relevant temperatures(50ndash110K) the interaction length associated withsr is more than a few molecular layers (greaterthan 10 Aring) A cross section that is likely to belarge above 20K is sHOH which describes themean distance for diffusion of mobile H before it

reacts Since s discussed earlier is of the sameorder of magnitude as the H2O dissociation crosssection sd then the precursor concentration cp(in this case H2O2) is roughly proportional to[s2OHsHOH] 1 (Appendix)

Based on the above a two-step precursor for-mation process could be consistent with the elec-tron data contrary to the statements in Sieger etal (1998) Orlando et al (1998) and Orlando andSieger (2003) Experiments are needed to deter-mine the relevant cross sections Below we con-sider a precursor formed in a single excitationevent

Precursor trapped O

O2 can be formed for instance by diffusion ofa transiently mobile O to a previously formed andtrapped O or to the production of O near the siteof a previously produced and trapped O (Matichet al 1993 Johnson and Quickenden 1997) Werefer to this as the MatichndashJohnsonndashQuickenden(MJQ) model and note that O 1 OH R O2 1 Hcan also give molecular oxygen

Trapped O as H2O-O was in fact seen to be aprecursor to O2 when attacked by a mobile O ina Kr matrix (Khriachtchev et al 2000) ThereforeO trapped as H2O-O was also suggested as a pre-cursor in ice (Johnson 2001) Before proceedingwe note that the examination of the MJQ modelin Sieger et al (1998) was incorrect They consid-ered two mobile interacting species whose life-times were long compared with the bombard-ment rate rather than a new mobile O interactingwith a previously trapped O Although this hasnot been corrected recently they added Otrapped as H2O-O as a possible precursor (Or-lando and Sieger 2003) O trapped as H2O-O isseen in matrix isolation studies as discussedabove but it has not yet been identified in iceand like OH it may trap in a number of ways

The MJQ model is essentially the followingFormation of H2 1 O is followed by loss of H2and trapping of O (Ot) A subsequent dissocia-tion or excitation event can lead to the destruc-tion of Ot by reaction with for instance H or byconversion to H2O2 However if this dissociationevent also leads to H2 1 O a mobile O is pro-duced that can react with Ot with an interactionlength that depends on T This results in the for-mation of O2 [O 1 Ot R O2] giving a net pro-duction of 2 H2 and 1 O2 In the Appendix therelevant rate equations (Eqs A1cndashA3c) are con-

JOHNSON ET AL836

sidered Assuming the mobile O processes arefast these rate equations can also be integratedand simplified to give the form in Eqs 1 and 2The precursor column density Np now repre-sents the column density of Ot In this model sO2in Eq 2 is the ldquocross sectionrdquo for reaction of afreshly produced O with an Ot and sp9 indicatesdestruction of Ot Finally sp is the cross sectionfor production of H2 and Ot via processes dis-cussed above

Based on the sizes of the cross sections ex-tracted in the electron irradiation experimentsthe model with trapped O (Ot) as the precursoris quantitatively reasonable That is for a rangeof T both sp and sO2 might be described byevents of the type H2O R H2 1 O (whether di-rect or indirect) If these occur for 10 of theradiation-induced dissociation events as dis-cussed then the required product of cross sec-tions suggested by the sizes of s and Y` above isroughly obtained that is [sp sO2] 02ndash6 3 10234

cm2 in the low-energy electron experiments Thecross section for production of O might dependon the availability of dangling bonds at interfacesin pores or at defects as suggested by a numberof authors The importance of such bonds couldalso be consistent with the differences in the elec-tron-induced yield for initially crystalline and ini-tially amorphous ice (Orlando and Sieger 2003)

O2 production can increase in the presence ofoxygen-rich impurities such as CO2 or SO2 bothof which are known to be present at Europa Asin the MJQ model dissociation (eg CO2 R CO1 O) can provide a mobile O that can be trappedor react with a previously trapped O forming O2In addition such species are hydrogen scav-engers (Moore and Hudson 2000) This can re-duce the destruction of Ot affecting sp as seen inEq A5c

Trap density

In the simplified models above the density oftrapping sites nt was not explicitly accountedfor However in the discussion of both the ex-perimental data and models trapping sites havebeen assumed to be important It has beendemonstrated repeatedly that the density of de-fects voids and interfaces decreases as the tem-perature of formation of the ice increases How-ever the efficiency of production of O2 is theopposite it increases with increasing T There-fore the oxygen yield varies in a manner that is

the inverse of the trap density over a range oftemperatures For penetrating ions in which theenergy density deposited is high and O2 might beformed in a single track a simple explanation hasbeen given As the density of traps for O2 is re-duced and the temperature increases the O2 pro-duced can diffuse to the surface and escape moreefficiently limiting back reactions on further ir-radiation This is examined in the next sectionHowever such a process does not affect the low-energy electron yields for which the penetrationdepth is small and the formed O2 escapes read-ily at the measurement temperatures In Eqs A1cA1c9 A3c and A10 in the Appendix we againuse a simple set of rate equations for trapped andmobile O in order to determine the role of the trapdensity nt on the production of a precursorspecies The conclusions apply to any modelbased on a trapped precursor

In the MQJ model mobile O can be trappedforming Ot (described by kOt nt) and can reactwith previously trapped O to form O2 (kOOt) Inaddition it can react at other sites and Ot can bedestroyed by mobile reactive species such as H(kHOt nH) Mobile H can also react with othertrapped species (kHR nR) Incorporating these re-actions Eq A10 in the Appendix gives a steady-state yield Y` When the density of traps is largethe steady-state yield is given in Eq A5c

Y` R (sd9sd9sd) [kOOtkHR nr][kHOt kOt nt] N (5)

Here sd and sd9 are the dissociation cross sectionsthat determine the production of H (H 1 OH) andO (H2 1 O) by the incident radiation Comparingthe result in Eq 5 with Y` in the simple precur-sor model in Eq 4b the dissociation cross sec-tions in Eq 5 are now scaled by a ratio of the re-action rates

It is seen in Eq 5 that the steady-state yield Y`depends inversely on the density of traps nt Thatis there are competing destruction processes fortrapped O (Ot) Therefore in order to form O2 amobile O competes with mobile H to find atrapped O before it becomes trapped In this waythe increasing O2 yield correlates with the re-duction in the trapping sites with increasing tem-perature of formation The small activation ener-gies discussed earlier would be related to there-orientation of the lattice binding leading to thereduction in trap density At relatively high T thedensity of traps becomes small so that only the

OXIDANTS AT EUROPA 837

sample surface or grain interfaces might act astraps accounting for the observed drop in theyield at high T For this reason it is important todetermine if the yield for T greater than 150Kdepends on the incident flux f Since other quan-tities in the expression for Y` could also be tem-perature dependent experiments are neededThese would likely be pulsed radiolysis experi-ments in which the transients are studied ratherthan the steady-state yields However the avail-able data for lightly ionizing radiation do suggestthat the O2 yield depends inversely on trap den-sity as in the model above

Escape from depth

The quantity N in Eq 1 is the ldquodepthrdquo given asa molecular column density from which O2 formedbelow the surface can percolate to the surface andescape It can be converted to a physical depth bydividing by the molecular number density of theice In the case of the low-energy electrons thedepth below the surface at which ionizations andexcitations are produced is small However forfast ions that penetrate many monolayers this isnot the case Below we examine N in Eq 1

Although it has long been understood that twoexcitations events are required to produce O2(eg Brown et al 1982 Reimann et al 1984) theprecursor model suggests that at saturation(steady state) any new event can produce O2 fromthe steady-state density of trapped precursorsTherefore one might expect the yield at satura-tion to be nearly linear in the excitation cross sec-tion This is the case for the low-energy electronsand it was initially assumed to be the case for thefast ions (Brown et al 1980 Reimann et al 1984)However Baragiola et al (2003) showed that thetotal sputtering yield in the saturation region attemperatures greater than 100K which is dom-inated by decomposition (Brown et al 1982) isproportional to the square of the electronic en-ergy deposition per unit path length in the solid(dEdx)e Assuming that the production and de-struction cross sections in Y` in Eq 4b roughlyscale with (dEdx)e then their ratio [sp sO2s]also roughly varies as (dEdx)e If this is the casethen the observed quadratic dependence impliesthat the depth N from which O2 can be mobi-lized to escape must also depend on the excita-tion density (dEdx)e This is consistent with theobservation that the O2 yield depends on thick-ness for samples thinner than the ion penetration

depth (Reimann et al 1984 Benit and Brown1990) Therefore the percolation depth for escapeN is a critical factor in comparing yields for pen-etrating radiation with those for non-penetratingradiation like the low-energy electron and low-energy ion data It is also seen that percolationand trapping can be included within the simplemodel considered here

Model summary

We have examined the simple precursor modelintroduced by Sieger et al (1998) to explain theirelectron data We have shown how it can be gen-eralized and derived from an integration of theappropriate chemical rate equations with thecross sections and total column density contain-ing the reaction rates and the physics of diffusivetransport Therefore the parameters can havevery different meanings from those initially as-sumed The equations are useful when there arefast processes which go to completion in timesmuch shorter then the time between particle im-pacts and slow processes associated with speciesthat are assumed to be trapped between particleimpacts This separation allowed us to examinethose models proposed for precursor formationand destruction due to reacting mobile speciesformed from dissociation events Further fromthe size of the onset of the linearity of the yieldversus fluence we can place constraints on therole of nonlinear processes such as those thatmight be associated with the formation of a pre-cursor like peroxide

With these extensions this simple model is use-ful for approximating the full rate equations inthe temperature region for which the yields de-pend on the fluence (dose) and not on the doserate Based on the assumptions in the model thisimplies temperatures at which precursor speciesare stable over experimental times At the high-est temperatures for the yields shown in Fig 4athis might not be the case This should be testedexperimentally In this regard it is puzzling thatexperiments have shown that the absorptionbands associated with certain trapped speciessuch as OH and trapped electrons ldquodisappearrdquoat 100ndash120K (see eg Johnson and Quickenden1997) whereas the O2 yield data indicate that rel-evant precursor species must be stable to 140KTherefore the proposed O2 production processescan only be distinguished by directly determin-ing the chemical state of the irradiated ice

JOHNSON ET AL838

It may be the case that a particular precursoris not required as mentioned earlier Rather af-ter a sufficient level of oxidation a newly pro-duced O or OH can react in the oxygen-rich lat-tice producing O2 in a pseudo-first-order processThis can be tested experimentally Indeedtrapped O and peroxide may be changed into oneanother as in the matrix isolation experiments(Khriachtchev et al 2000) Therefore their rela-tive concentrations versus temperature and doseshould be studied

In the model considered here it is seen that theparameters in Y` must account for the strong tem-perature dependence Since the number of bulktrapping sites decreases with increasing T as theice sample is annealed one might expect that theprecursor concentration would decrease We sug-gest that the observed temperature dependenceis determined by the competition between theavailability of trapping sites and mean interactiondistances of a mobile species such as O and HThat is the O2 yield varies inversely with the den-sity of traps over a range of temperatures In thesimple model above this effect is included indi-rectly as shown using the more detailed equa-tions in the Appendix

EUROPA

Oxygen and peroxide production at Europa G values

The dependence of the yield on the thicknessof the ice sample for highly penetrating ions sug-gests that the O2 yield is roughly related to theamount of energy deposited in the sample There-fore the amount of O2 produced and lost fromice in steady state is often given as a G value thenumber of molecules produced per 100 eV de-posited in Table 2 In applying these G values tothe planetary environment care must be takensince the yields are temperature fluence and ex-citation density dependent They also are affectedby the presence of other proton scavengersknown to be present in Europarsquos ice

The G values for peroxide in Table 2 are thosefor production at low fluence but the G valuesfor O2 are for the production and escape of gas-phase oxygen at saturation Therefore they de-pend on the formation and destruction cross sec-tions in the model above Such G values can beused to describe the steady-state source of the am-

bient O2 atmosphere at Europa but are not help-ful in determining the ability to produce and trapO2 at depth into the icy surface Over the timescales of the experiments some O2 is seen to re-main trapped and is not destroyed by radicalsTypically there are a number of trapping sitesand recent estimates of the activation energy foran O2 molecule trapped in a substitutional site(Hori and Hondoh 2002) suggest that it can betrapped for times comparable to the time re-quired for the conversion of amorphous ice intoIh (Ayotte et al 2001) The stability of O2 inclu-sions has not been studied

To understand the relationship between the Gvalues for O2 in Table 2 we note that fast lightions and electrons lose their energy predomi-nantly by producing ionizations and excitationsThe ionization events produce secondary elec-trons that can result in further ionizations Thenumber of ionization events Ni produced by afast light ion stopping in a water ice is Ni 5EoW where Eo is the initial energy and W is theaverage energy expended per ionization event27 eV in water ice Therefore radiolysis by fastions is often described as the net effect of theshower of electrons set in motion by the incidention In this manner the G values for productionof O2 from ice by energetic light ions and elec-trons can be roughly estimated as the sum of theeffects of all the secondary electrons using theyield Ye for 10ndash100 eV electrons That is if weignore the percolation depth N we can write thenumber of O2 produced as the number of ioniza-tion events times Ye which already has in it thedestruction processes That is G NiYeEo YeW 0006 O2100 eV

This estimate of G is seen to be an order of mag-nitude larger than the keV proton value for O2 inTable 2 but is much smaller than the G value forenergetic Ne1 These differences from the simpleestimate are consistent with the observation thatthe yield in the saturation regime for penetratingions varies nonlinearly with excitation densityThe difference from the keV proton G value islikely due to the inability of O2 to escape fromdepth before destruction On the other hand thelarger G value seen using MeV Ne1 is likely dueto the high density of excitation This can resultin more efficient production of O2 increased pro-duction of damage sites for forming O2 andorincreased transport (percolation) along the tran-siently heated track That is the ability of theformed O2 to percolate to the surface during ion

OXIDANTS AT EUROPA 839

bombardment plays an essential role in deter-mining the measured yields as discussed aboveThis effect needs to be evaluated for a planetarysurface subject to particle fluxes that are many or-ders of magnitude smaller than those used in thelaboratory

In the absence of O2 yields for all of the rele-vant ions and energies an estimate of the totalsupply of O2 can be obtained using the G valuesin Table 2 For the energetic oxygen and sulfurions we use G values for neon and argon ionsThese are lower bounds as the data were obtainedat temperatures much lower than the surface tem-peratures at Europa (80ndash110K) For the fast pen-etrating electrons and protons we use the valuefor 200 keV protons in Table 2 For the heavy ionswe used the Ne1 and Ar1 values For compari-son we also give an estimate using the G valuesobtained from the low energy electron data andthe total ionization as discussed above

The G values are combined in Table 3 with theelectron and ion energy flux at Europa Althoughthere is some deflection of the flow due to in-duced fields the net energy flux to the surface isnot reduced much from these values (Paranicaset al 2002) In fact locally produced pick-up ionswhich are formed from the ejecta and acceleratedto the surface can contribute (Ip 1996) Furtheralthough G values are typically independent ofthe incident angle the loss of O2 is a surfaceprocess and is enhanced with angle That is thenumber of O2 produced within the escape depthincreases with increasing angle to the normal(steeper than the inverse cosine) This enhance-ment is 25ndash4 (Jurac et al 2001) Ignoring suchcorrections the total production rate calculated

(23ndash62 3 109 O2cm2s) is a lower bound to theactual production rate from the icy regions of Eu-roparsquos surface Recent modeling of the O2 atmo-sphere suggest that such a source rate is required(Shematovich and Johnson 2001)

Earlier estimates of the O2 source rate were in-correct because O2 production was assumed to bea fixed fraction of the H2O sputtering yield Atthe ambient temperature an ejected H2O sticks ef-ficiently (Smith and Kay 1997) to the surface ofneighboring grains in the porous regolith reduc-ing the sputtering yield for H2O (Johnson 1989)This reduction does not apply to the O2 fractionof the yield enhancing its relative importance toH2O Here we assume that all of the gas-phaseO2 produced can contribute to the ambient at-mosphere even if it is produced at a grain sur-face that is at some depth into the porous regolith

Finally Europarsquos surface is not pure ice every-where That is much of the trailing hemisphereis a hydrated sulfuric acid containing smallamounts of sulfur and SO2 and the leading hemi-sphere contains low levels of carbon CO2 andcarbonates Sodium and potassium atoms aresputtered from the icy regions (eg Leblanc et al2002) O2 production from surfaces containingthese species has not been measured C and S areboth proton scavengers and the other species canbe sources of O as discussed so that the O2 yieldfrom such regions could be enhanced Howeverthere can also be competing processes that mightreduce these rates In fact recent HST observa-tions suggest that the O2 might not be produceduniformly across Europarsquos surface (McGrath et al2000) The analysis of these observations is pre-liminary and is not simple since the observed

JOHNSON ET AL840

TABLE 3 APPLICATION TO EUROPA

OxygenImplantation Energy flux production

Species 109(cm2 s)a 109 keV(cm2 s)(1) 109(cm2 s)

H (10 keV) 0015(1) 12 0084 (072)bO (10 keV) 0015(1) 18 054cS (10 keV) 0009(1) 30 090cElectrons 62 043 (37)bPlasma (10 keV) (137 amu) 13(2) 10 03c

Total 79 23 (62)UV photons (64 eV) 46 01

aReferences are as follows 1Cooper et al (2001) [similar results in Paranicas et al (2002)] 2Mauk et al (1996)bUsing G for 200 keV protons (upper bounds in brackets are based on the low-energy electrons data as a model for

the secondary electrons produced)cUsing G for 15 MeV Ne1 at 7K a rough lower bound

emissions depend on the local plasma propertiesHowever the efficiency of O2 production in con-taminated ices needs to be studied

Porous regolith O2 interactions on grain surfaces

Although the O2 atmosphere is principally pro-duced in the near surface layers and is eventuallylost to space it also permeates Europarsquos porousregolith The porosity p (void space) of the re-golith is uncertain (Buratti 1995) It has been es-timated to be the most compact surface in theOuter Solar System p 025 on the trailing hemi-sphere and p 079 on the leading hemisphere(Buratti and Veverka 1983) and the most poroussurface p 096 globally (Domingue et al 1991Domingue and Hapke 1992) These large differ-ences are due to the uncertainties in the amountof backscattered light at small phase angles andon the photon absorption properties and grainsizes The more compact surface is likely correctGrain sizes rg of the order of 100 mm have beensuggested (Hansen and McCord 1999 2000Geissler et al 1998) Europarsquos regolith differs sig-nificantly from the lunar regolith which is asand-like surface of individual grains Analysis ofthe thermal inertia of Europarsquos surface (Grundyet al 2001) suggests that the ice grains are sin-tered forming a fairy castle-like structure Sin-tering can occur thermally or due to the incidentradiation In addition ices doped with acids andalkalis in the laboratory can be powdery solidswith the impurities primarily found at grain in-terfaces

The porous regolith affects the above estimatesof the oxygen production in three ways First forisotropic bombardment the mean distance d intothe surface at which an incident ion or photonstrikes a grain surface is d (89)rg(1 2 p) Us-ing the above value for rg then d 02 m if theporosity is very large p 96 Therefore de-sorbed molecules can come from a considerabledepth and will interact with grain surfaces Evenfor much lower porosities the effective sputteringyields can be significantly modified (Johnson1989) As discussed above the yields for ejectedH2O which sticks onto neighboring grains at theambient temperatures are reduced but those forejected O2 are not This enhances the relative im-portance of the O2 yields

Finally the surfaces and interfaces of the sin-tered grains in Europarsquos porous surface are placeswhere low temperature chemistry involving the

ambient oxygen can occur at depths that are wellbelow the nominal penetration depth of the inci-dent radiation The relative importance of suchchemistry is roughly indicated by the ratio of thetotal surface area in the regolith to the area of Eu-roparsquos visible surface Using the porosity andgrain size above and a regolith depth dr of theorder of a meter then this ratio is large 3(1 2p)drrg which is 103ndash2 3 104 for average porosi-ties of 96 and 25 respectively Therefore thegas-phase O2 molecules below the visible surfaceinteract frequently with the grain surfaces In thisway a very large volume of the subsurface couldbe altered chemically even if it is not directly ex-posed to the radiation

From the overlying structures seen in the high-resolution images of Europarsquos surface by theGalileo spacecraft (eg Prockter and Pappalardo2000) it is clear that the youngest surfaces are red-der The reddening in the visible has been attrib-uted to the presence of sulfur in the ice (McEwen1986 Johnson et al 1988 Calvin et al 1995 Carl-son et al 1999b) The amount of reddening in thespectra of the surface material decreases with ageand the surface materials appear to becomebrighter with age This might occur by the slowaccumulation of a frost over-layer or by radioly-sis (Carlson et al 2002 Johnson et al 2003) Ra-diolysis preferentially converts the sulfur in theice to sulfur dioxide or sulfuric acid both ofwhich have flat spectra in the visible In regionswhere the radiation does not directly penetratethe O2 permeating the regolith can oxidize thesulfur at the surfaces of the icy grains and at graininterfaces Similarly the carbon seen in the ice onEuroparsquos leading hemisphere (eg Carlson et al2004 Johnson et al 2003) can be oxidized by ra-diolysis in the near surface layers or by reactionson the grain surfaces at depth in the regolith bythe ambient O2 The meteoroid bombardmentthat produces the regolith also exposes fresh sur-faces for oxidation and can cause the collapse ofpores leading to the trapping of oxygen In thismanner both refractory and volatile oxidants canbe formed and trapped in the subsurface ice forpotential delivery to Europarsquos subsurface ocean

SUMMARY

The presence of a thin oxygen atmosphere andoxidants trapped in Europarsquos surface ice (egJohnson et al 2003) combined with the possible

OXIDANTS AT EUROPA 841

presence of a subsurface ocean (eg Carr et al1998) have made Europa an exciting target forfuture space exploration The molecular oxidantsproduced by radiolysis and photolysis could bespecies needed for carbon-based biochemistry inEuroparsquos putative subsurface ocean Howeverthere has been no model for the formation trap-ping and transport of relevant oxidants

In this paper we have reviewed the space ob-servations and laboratory studies related to thechemistry of formation of oxidants in ice by theincident radiation Because the laboratory dataare incomplete we examined the rate equationsfor formation of oxygen and related species byradiolysis and photolysis We showed how suchequations can be simplified to the analytic modelthat was used for fitting the fluence and temper-ature dependence of the O2 yield produced bylow-energy electrons (Sieger et al 1998) In de-riving the analytic model from the more completerate equations the competition between forma-tion and destruction processes can be studied ascan the role of trapping sites escape from depthand temperature history These equations also al-lowed us to put incident photon electron and iondata on a similar footing for the first time

The model can be used to suggest measure-ments that are needed to describe the chemicalstate of the irradiated ice and to interpret new lab-oratory data We showed that if reactions are fastthe yield data could be consistent with a precur-sor being formed by two events as is the case forperoxide However we favor trapped O as a pre-cursor as initially suggested by Matich et al(1993) for production of excited O2 Such aspecies if present could be detected spectro-scopically as can trapped O2 through bands ac-tivated by lattice perturbations (Cooper et al2003a)

The available oxygen yield data were used tocompare G values for different incident particlesThese were then used to estimate the oxygen pro-duction in the radiation environment at EuropaThe lower bounds were consistent with the pro-duction rates obtained from a recent model of Eu-roparsquos atmosphere The porosity of the regolithwas also considered It not only affects the rela-tive sputtering yields for O2 and H2O but alsoprovides an enormously increased surface areaon which the ambient gas-phase oxygen can re-act with species in the ice That this might be oc-curring is consistent with the change in re-

flectance associated with surface aging Such re-actions could also convert ambient molecularoxygen into other more refractory species that re-main trapped on subduction of Europarsquos crust

Although radiolysis produces a stable chemi-cally altered surface in the laboratory the parti-cle flux to Europarsquos surface is many orders ofmagnitude lower Therefore annealing can be ef-fective at temperatures that are much lower thanthose typically found experimentally Belowabout 120K the annealing times appear to be long(eg Baragiola 2003) so that laboratory data canoften be directly applied to Europarsquos icy surfaceHowever account has to be taken of the syner-gism between the UV and the energetic plasmaas discussed earlier (Johnson and Quickenden1997)

Transient surface melting might initiate lifeprocesses but subduction of oxidants to the pu-tative ocean is likely to be required if biochem-istry is to occur at Europa (Chyba and Hand2001) Although the temperature versus depthinto Europarsquos regolith is not well known belowthe regolith the temperature and pressure appar-ently increase with depth At some depth (ap-proximately kilometers) either very warm ice oran ocean exists (Ruiz and Tejero 2000) Thereforechemically refractory oxidants are likely to be re-quired in order to survive subduction to thewarmer subsurface regions Trapped species ifdelivered downward can become mobile andwill segregate forming inclusions (Johnson andJesser 1997) Indeed peroxide appears to segre-gate in ice (Gurman et al 1967) and these inclu-sions may remain relatively stable until the per-oxide is dissolved in the putative ocean Thevolatiles seen in reflectance such as O2 SO2 andCO2 will also segregate But at the higher tem-peratures they could also diffuse through thebulk react or percolate into the regolith atmo-sphere Therefore the observation of the trappedvolatiles in Europarsquos surface ice is of interest pri-marily because it suggests that oxidants are pre-sent The formation of oxygen-rich molecules fol-lows the preferential loss of hydrogen Howeverthe surface is in a local steady state with im-planted H O and S from the plasma and the lossto space Since the plasma flux to the surface canbe variable this balance can also vary Indeedtemporal changes in reflectance have been re-ported (Domingue and Hendrix 2004)

Based on the models for formation of O2 ex-

JOHNSON ET AL842

amined in this paper additional experiments areneeded to determine the chemical state of irra-diated ice at low fluences Of immediate inter-est is the identification of stable molecular pre-cursors required for the formation of O2 Basedon the available laboratory data individual per-oxide molecules trapped in the ice do not appearto be the principal precursors but trapped O(Matich et al 1993) remains a viable candidateAlthough the data suggest that the local hydro-gen-bonding network is important in the pro-duction of O2 the effect of the formation tem-perature and radiation damage on the structureneed to be measured It is also important to de-scribe quantitatively the relationship betweenH2 loss and the formation of O2 and H2O2 overa range of temperatures and fluences Finallyone would like to determine the steady-statedensities of biochemically interesting moleculesformed by radiolysis in an ice containing sulfurand carbon at Europarsquos surface temperaturesand then determine their stability with increas-ing temperature and pressure Such experimentsstill need to be carried out Here we have pre-sented a model that brings together the variouslaboratory data and now can be used to considerthe formation trapping and transport of relevantoxidants at Europa

APPENDIX

Introduction

In this Appendix we examine rate equationsthat describe various aspects of the radiolysis orphotolysis of ice These equations need to be un-derstood if one is to be able to calculate the for-mation trapping and transport of oxidants Theemphasis is on the formation of molecular oxy-gen We integrate these equations over the escapedepth of O2 to give the simplified rate equationsdiscussed in the text in which number densitiesare replaced by column densities (eg Eqs 1 and2) In this way we can examine the relationshipbetween the rate constants and the effective pre-cursor formation and destruction cross sectionsextracted from data We first consider a precur-sor such as H2O2 that is formed from two exci-tation events This bears on the observed linear-ity of the oxygen yield versus fluence We thenconsider the equations for trapped versus a tran-

siently mobile species We use a reaction betweena mobile and trapped O as an example althoughO 1 OH can also give oxygen Finally we ex-amine the possible dependence of the O2 forma-tion process on the density of traps (defectsvoids and grain boundaries) as a function of tem-perature The results are discussed in the text

For clarity all possible radiolytic species arenot included in the discussion below Thereforepotentially important products such as HO2 andthe charged species (eg H3O1 the trapped elec-tron etc) are missing In each section we attemptto make a point related to production of O2 andother oxidants by using a subset of the productsrather than attempt a complete description withmany unknown rate constants

Fast versus slow processes

To describe the O2 production rate from a pre-cursor such as H2O2 we first use a simple but in-structive set of rate equations to describe the for-mation of H2O2 We then relate these to thesimplified rate equation for the precursor in thetext Eq 1 and show how the postulated precur-sor formation cross section sp is related to theH2O dissociation cross section sd by the reactionrates

Using the reactions

H2O 1 radiation R H 1 OH (sd)

H 1 H R H2 (k2H)

H 1 OH R H2O (kHOH)

OH 1 OH R H2O2 (k2OH)

the reaction pathways are described by

H2

Hqk2H

1 OHH2O 1 radiation mdashRsd H

OHH2O2

OH Rk2OH

kHOH

HH2O

OXIDANTS AT EUROPA 843

The corresponding rate equations in an irradiatedvolume are

dnOHdt 5 sd f n 2 2 k2OH nOH nOH

2 kHOH nH nOH (A1a)

dnHdt 5 sd f n 2 2 k2H nH nH

2 kHOH nH nOH (A2a)

dnH2O2dt 5 k2OH nOH nOH

2 destruction processes (A3a)

Here sd is the dissociation cross section f is theradiation flux and n is the molecular numberdensity of ice with nOH and nH the density of dis-sociation products OH and H respectively Therate constants k2OH k2H and kHOH describe thereactions above In this model the H2 formed isassumed to escape for temperatures above about20K We do not distinguish between trappedspecies and mobile species although in fact thatdistinction is important as discussed in the nextsection That is any OH produced at low tem-perature by one ion would likely be trapped priorto interacting with an OH produced by a subse-quent ion At higher temperatures this can breakdown as discussed in the text We first relate theabove set of equations to Eq 1 in the text and thenconsider precursor destruction processes

Equation 1 in the text is obtained by solvingthese equations together and then by integratingover a depth Dx Here Dx is the smaller of the pen-etration depth of the particle and the O2 escape(percolation) depth Assuming the molecular den-sity of ice n is constant with depth Dx n 5 N inEq 1 Comparing Eq 1 with Eq A3a the precur-sor formation rate sp f N is equivalent to [k2OH

nOH nOH] Dx Assuming that the processes in EqsA1a and A2a are fast then the steady state is read-ily obtained In addition if k2H k2OH there ispreferential loss of H2 at short times as suggestedby the laboratory data If steady state is reachedthe production of H2 is equal to that of H2O2 us-ing only the equations above Further one obtains

[k2OH nOH nOH] Dx R [fp sd] f N (A4)

with

fp 5 05[1 1 05 kHOH(k2H k2OH)05] (A5a)

Therefore the precursor formation cross sectionsp in Eq 1 is a fraction fp of the total dissocia-

tion cross sd Since laboratory data suggest theH2O2 fraction in an irradiated ice is small thenkHOH (k2OH k2H)05 which is likely givingfp R [(k2OH k2H)05kHOH]

Trapped versus mobile species

To examine the fluence dependence discussedin the text we note that trapped OH and freshlyproduced transiently mobile OH can be treatedas different species (eg Matich et al 1993)Therefore assuming no trapping of H the OHrate equation above would be replaced by twoequations one for the trapped species nOHt andthe other for the freshly produced mobile speciesnOH Assuming the reaction of two mobile OHunits is unlikely and that peroxide is formed fromthe interaction of a trapped OH with a subse-quently produced OH an expanded set of equa-tions is obtained

OH 1 trap R OHt (kOHt)

OH 1 OHt R H2O2 (kOHOHt)

H 1 OHt R H2O (kHOHt)

then the reaction pathways are described by

qH2

H2O 1 radiation mdashRsdH

qk2H

1 OHH

trap OH

OH RkOHt

OHt kOHOHtH2O2

kHOHkHOHt

H H2O HH2O

The resulting set of equations is

dnOHdt 5 sd f n 2 kHOH nH nOH

2 kOHt nOH nt 2 kOHOHt nOH nOHt (A1b)

dnOHtdt 5 kOHt nOH nt 2 kOHOHt nOH nOHt

2 kHOHt nH nOHt (A1b9)

dnHdt 5 sd f n 2 2 k2H nH nH 2

kHOH nH nOH 2 kHOHt nH nOHt (A2b)

JOHNSON ET AL844

dnH2O2dt 5 kOHOHt nOH nOHt

2 destruction processes (A3b)

where nt is the trap density in the irradiated iceAt low T we can assume nt is large so thatdntdt 0 simplifying the set of equations If theprecursor of interest is trapped O as discussed inthe text then a similar set of equations can be con-structed for O and Ot as in the example to followOf course a full description over a broad rangeof temperatures would require the considerationof the both trapped and mobile H O and OHthe rate of change of trap density and the vari-ous destruction processes

The above equations should be solved simul-taneously but for the incident particle fluxes typ-ically used in these studies the diffusion of mo-bile species to traps is relatively fast This is trueeven at low temperatures if there is a significantdensity of trapping sites This is usually the casefor samples formed at low T Therefore in a lowtemperature solid at low dose rates there is not asignificant steady-state background of diffusingOH There are trapped or freshly produced OHunits which rapidly react Averaging over thebombardment rate this can be approximated bywriting dnOHdt 0 and then separating nOH outin Eq A1b Assuming also that H diffuses rapidlyand either reacts or forms H2 we also writednHdt 0 on the average Therefore betweenimpacts only species trapped at defects pores orgrain surfaces are assumed to be present as sug-gested by the lack of a dependence on the inci-dent beam flux In this limit the rate of change oftrapped OH roughly simplifies to

dnOHtdt 5 2 k2H nH nH 2 2 kOHOHt nOH nOHt

It is seen that the production of peroxide dependson the loss of H by formation of H2 as discussedand at steady state the production rate for per-oxide equals the rate of formation of H2 in theseequations

Treating the mobile and trapped species sepa-rately we can further simplify the rate equationsand obtain the precursor model used in the textThat is integrating over the relevant depth Dxthe densities are replaced by column densitiesThe rate constants ki are then related to effectivecross sections si In this way cross sections aregiven as an interaction length for each freshlyproduced mobile radical [ki nj] R [si f] The

trapped OH column density NOHt and peroxideproduction rate can be estimated from ldquolinearrdquorate equations

dNOHtdt lt [sd f] N 2 2 [s2OH f] NOHt

2 [sHOH f] NOHt (A6a)

dNH2O2dt lt [s2OH f] NOHt 2 s NH2O2 (A6b)

Here sd is again the dissociation cross sectionand the destruction processes in Eq A3b are con-tained in s The second and third terms on theright in Eq A6a are the loss of OH due to the pro-duction of H2O2 and H2O In the integration ofEq A1b9 [2 kOH OHt nOH] becomes [2 s2OH f] and[kHOHt nOH] becomes [sHOH f] Therefore thedensity and the interaction length of mobile H arecontained in sHOH If H2O2 is indeed the pre-cursor then in Eq 2 in the text [sp f N] is be-comes [s2OH f NOHt] from Eq A6b Accountingfor precursor destruction which can give backtwo OH and treating H2 can both be included inthe integrated model via the si values

The simplified set of equations above allows usto examine the relative time (fluence) scales forprecursor formation (in this case H2O2) SolvingEqs A6a and A6b the column density of OH andthe concentration of precursors versus fluence cpcan be obtained Using the column of H2O2 pro-duced as the precursor column Np in Eq 2 thenO2 yield can be obtained The results can be com-pared with the results from Eqs1 and 2 in thetext The average precursor (H2O2) concentrationover the column depth of water molecules N is

cp 5 fp (sds)[1 2 exp(2s F)] 2 (ssr)

[1 2 exp(2sr F)] (A7)

with

fp 5 2s2OH(sr 2 s) (A5b)

In these equations sr 5 (2s2OH 1 sHOH) andssr 1

It is seen that the precursor concentration withtime cp depends on two rates the reaction ratesthat determine sr and the destruction rates in sAs in Eq 4a the yield can be written as

Y 5 sO2 cp N (A8)

If sr and s are comparable then the yield at small

OXIDANTS AT EUROPA 845

fluence is quadratic in F However if sr sthen the linear dependence on fluence is obtainedwith a small shift in the fluence as shown in Eq4c in the text Further the steady-state yield isnow

Y` (T) 5 sO2 [fp sd] Ns (A9)

That is sp R [fp sd] as in Eq A4 and the tem-perature dependence for the incident electrondata must be contained in [sO2 fp] or in the ratio[sO2 s2OH]sHOH]

For comparison with their precursor modelSieger et al (1998) examined the possibility of ob-taining the O2 precursor by considering reactionsfor a density of diffusing dissociation products asin Eqs A1 and A2 rather than a trapped productinteracting with a freshly formed dissociationproduct Based on the above the lowest fluence atwhich O2 is first detected experimentally sets alimit on the time scale for mobilendashmobile reactions

Effect of trap density

Equations A1b A1b9 A2b and A3b can be ap-proximately solved assuming that the dnOHdtdnHdt and dntdt are very small on the averageas discussed above In order to examine the roleof the trap density nt which can depend on thetemperature and method of formation and on theradiation history of the ice we use the slightlysimpler precursor model In this model the O pro-duced directly by dissociation traps forming aprecursor O2 is then formed when a freshly pro-duced O or other species reacts with a trappedoxygen atom Ot

H2O 1 radiation R H2 1 O (sd)

O 1 trap R Ot (kOt)

O 1 Ot R O2 (kOOt)

The reaction pathways for the oxygen species are

H2O 1 radiation mdashRsd9

trap OO2

kOtOt kOOt

H2 1 OOHt

kOR H(kHO t)

ROR

We assume for simplicity both H2 and O2 escapefrom depth Dx after formation We also allow mo-bile H to react with Ot (kHOt) removing trappedO and both the mobile H and O to be removedby other reactants R (eg OH or contaminants)This leads to a simplified set of equations not di-rectly involving OH or peroxide

dnOdt 5 sd9 f n 2 kOt nO nt 2 kOR nO nR

2 kOOt nO nOt (A1c)

dnOtdt 5 kOt nt nO 2 kOOt nO nOt

2 kHOt nH nOt (A1c9)

dnHdt 5 sd f n 2 kHOt nH nOt

2 kHR nH nR (A2c)

dnO2dt 5 kOOt nO nOt (A3c)

These equations of course act in parallel with theequations above for OH and mobile OH may re-act with trapped O to form O2 Here we focusonly on mobile O

Since the diffusion of H and O are fast and thedensity of traps is large the nH and nt averagedover the bombardment rate can be taken veryroughly as time independent Further if we as-sume the number of trapped O nOt is small com-pared with the number of sites at which H canreact then nH [sd f nkHR nR] For a low ra-diation flux f nH is small on the average as as-sumed Since the O2 yield YO2 is equal to[(dnO2dt) Dxf] with Dx the escape depth thenthe steady-state yield Y` can be obtained ana-lytically

[Y`][sd9 N] 5 1 1 (q 2 1) 2 [(q 2 1)2

1 8 q d]054 d (A10)

q 5 [kOt (nt 1 nR)] (kHOt nH)(sd9 f n) kOOt

5(sdsd9) [kOt (nt 1 nR)] kHOt[kOOt kHR nR]

d 5 05 [1 1 nt(nt 1 nR)]

We note that for nt large q can be large Assum-ing that q 1 and nt nR then Eq A10 be-comes

Y` R [sd9 N]q 5 (sd9 N) (sd9sd)

[kOOt kHR nR][kHOt kOt nt] (A5c)

JOHNSON ET AL846

Comparing with Y` in the simple precursormodel in Eq 4b then s R [(kHOt nH)f] R [sd nkHOt][kHR nR] In addition the product [sO2sp] R sd9 sd9 [kOOt n][kOt nt] That is the dis-sociation cross sections are modified by reactionrates

Quite remarkably it is seen in Eq A5c that thesteady-state yield Y` can depend inversely onthe density of traps nt in the temperature andflux regimes for which the approximations arerelevant That is if there are competing destruc-tion process for trapped O then at high trap den-sity a mobile O must find a trapped O before itbecomes trapped or before mobile H destroys theOt In this way the increasing O2 yield with in-creasing T can be related to the change in the trap-ping density nt with the formation temperatureof the ice sample As the temperature increasesand the density of traps in the bulk decreases thesurface and grain interfaces become the only trap-ping sites

ACKNOWLEDGMENTS

REJ acknowledges support by a Gledden Fel-lowship in Chemistry School of Biomedical andChemical Sciences at The University of WesternAustralia and by grants from the NSF Astron-omy Program and NASArsquos Planetary Geologyand Geophysics Program PDC acknowledgessupport by an Australian Postgraduate Award

ABBREVIATIONS

HST Hubble Space Telescope MJQ MatichndashJohnsonndashQuickenden NIMS Near Infrared Map-ping Spectrometer UV ultraviolet

REFERENCES

Ayotte P Smith RS Stevenson KP Dohnarsquolek ZKimmel GA and Kay BD (2001) Effect of porosityon the adsorption desorption trapping and release ofvolatile gases by amorphous solid water J Geophys Res106(E12) 33387ndash33392

Bahr DA Famaacute M and Baragiola RA (2001) Radiol-ysis of water ice in the outer solar system sputteringand trapping of radiation products J Geophys Res 10633285ndash33290

Baragiola RA (2003) Microporous amorphous water icefilms and astronomical implications In Water in Con-

fining Geometries edited by JP Devlin and B Buch El-sevier New York in press

Baragiola RA and Bahr DA (1998) Laboratory studiesof the optical properties and stability of oxygen onGanymede J Geophys Res 103 25865ndash25872

Baragiola RA Atteberry CL and Bahr DA (1999) Ori-gin of solid oxygen on Ganymede Reply to commentby RE Johnson on ldquoLaboratory studies of the opticalproperties and stability of oxygen on Ganymederdquo JGeophys Res E104 14183ndash14187

Baragiola RA Vidal RA Svendsen W Schou J ShiM Bahr DA and Atteberrry CL (2003) Sputteringof water ice Nucl Instrum Methods Phys Res B 209294ndash303

Bar-Nun A Herman G Laufer D and Rappaport ML(1985) Trapping and release of gases by water ice andimplications for icy bodies Icarus 63 317ndash332

Baverstock KF and Burns WG (1976) Primary pro-duction of oxygen from irradiated water as an expla-nation for decreased radiobiological oxygen enhance-ment at high LET Nature 260 316ndash318

Benit J and Brown WL (1990) Electronic sputtering ofoxygen and water molecules from thin films of waterice bombarded by MeV Ne1 ions Nucl Instrum Meth-ods Phys Rev B 46 448ndash451

Benit J Bibring J-P della-Negra S LeBeyec YMendenhall M Rocard F and Standing K (1987)Erosion of ices by ion irradiation Nucl Instrum Meth-ods Phys Rev B 1920 838ndash842

Brown WL Augustyniak WM Lanzerotti LJ John-son RE and Evatt R (1980) Linear and nonlinearprocesses in the erosion of H2O ice by fast light ionsPhys Rev Lett 43 1632ndash1635

Brown WL Augustyniak W M Simmons E Mar-cantonio KJ Lanzerotti LJ Johnson RE BoringJW Reimann CT Foti G and Pirronello V (1982)Erosion and molecular formation in condensed gasfilms by electronic energy loss of fast ions Nucl In-strum Methods B 1 307ndash314

Brown WL Augustyniak WM Marcantonio KJ Sim-mons EN Boring JW Johnson RE and ReimannCT (1984) Electronic sputtering of low-temperaturemolecular solids Nucl Instrum Methods B 1 307

Buratti B (1995) Photometry and surface structure of theicy Galilean satellites J Geophys Res 100(E9) 19061ndash19066

Buratti B and Veverka J (1983) Voyager photometry ofEuropa Icarus 55 93ndash110

Calvin WM and Spencer JR (1997) Latitudinal distri-bution of O2 on Ganymede observations with the Hub-ble Space Telescope Icarus 130 505ndash516

Calvin WM Clark RN Brown RH and Spencer JA(1995) Spectra of the icy Galilean satellites from 02 to5 mm a compilation new observations and a recentsummary J Geophys Res 100 19041ndash19048

Calvin WM Johnson RE and Spencer JA (1996) O2

on Ganymede spectral characteristics and plasma for-mation mechanisms Geophys Res Lett 23 673ndash676

Calvin WM Aninich VG and Brown RH (2004) Oxy-gen temperature and phase effects Icarus (in press)

OXIDANTS AT EUROPA 847

cating that destruction processes are in a roughsteady state with formation processes The yieldsshown in Fig 2 are purportedly those after satu-ration although full saturation is often ap-proached slowly

The slope of the O2 yield versus fluence shownin Fig 3 increases with increasing temperaturegiving the activation energies in Table 2 Thesevalues are similar to those obtained from Fig 2For energetic ions that fully penetrate the film asin Fig 2a and 3a the yield versus fluence can ex-hibit a second rise at higher fluences for whichthe radiation track cores begin to overlap This in-crease also depends on temperature and exhibitssimilar activation energies (Reimann et al 1984)The strong dependence on temperature is at firstsurprising since the energy density deposited bya 15 MeV Ne1 produces transient temperaturesin the track core that are much higher than theambient material temperature Therefore tracktemperature (local excitation density) alone doesnot determine the O2 yield

For all incident particle types and energies themeasured fluence dependence is found to be sta-ble over laboratory time scales That is for ex-periments carried out at high vacuum the yieldreturns to the value it had when the beam wasturned off This indicates that the irradiated ice isldquopermanentlyrdquo altered (Reimann et al 1984)Thus a relatively stable chemical state is createdthat is favorable for O2 production This could bethe formation of a particular trapped radiationproduct in ice that acts as a stable precursor or isdue to the altered chemical and physical state ofthe irradiated ice Trapped O H2O2 and HO2(Matich et al 1993 Johnson and Quickenden1997 Sieger et al 1998) have been proposed asstable precursor species

Using isotopically labeled H2O Benit andBrown (1990) showed that considerable transportof O occurs along the ion track when an energeticheavy ion penetrates an ice sample Because ofthis the O2 yield was found to depend on sam-ple thickness Therefore for thin ice samples the

JOHNSON ET AL828

TABLE 2 OXYGEN AND PEROXIDE PRODUCTION

Radiationa Product T (K) Ea (eV) Percent at saturation G (product100 eV)b

He1(1) O2 015g(23) O2 77 (300) 0003 (0)Ne1 (30 MeV)(3) O2 300 0031He21 (6 MeV)(3) O2 300 0008He21 (29 MeV)(3) O2 300 00016H1 (80 keV)(4) O2 70 00002H1 (200 keV)(5) O2 120 00007He21 (15 MeV)(6) O2 50 (100) 003 (07ndash4) 3 1024

Ne1 (15 MeV)(7) O2 7 10ndash15c 003(8) 70ndash140 005ndash007 01

Ar1 (01 MeV)(9) O2 30ndash140 001ndash004 0006ndash008160 001

e (30ndash100 eV)(10) O2 120 002ndash003 0005ndash0007hn (102 eV)(11) O2 50 (100) 003 00001 (00003)H1 (1 MeV)(12) H2O2 16 (77) 01 (001)

1 10 CO2 77 011 10 O2 77 04

H1 (30 keV)(13) H2O2 16 (77) 022 (038) 0098 (0086)C1 (30 keV) H2O2 16 (77) 17 (22) 023 (034)N1 (30 keV) H2O2 16 (77) 16 (14) 021 (011)O1 (30 keV) H2O2 16 (77) 14 (40) 026 (015)Ar1 (30 keV) H2O2 16 (77) 22 (20) 011 (006)

aReferences are as follows 1Lefort (1995) 2Ghormley and Stewart (1956) 3Baverstock and Burns (1976) 4Baragiolaet al (1999) 5Baragiola and Bahr (1998) 6Brown et al (1982) (ion does not stop in sample) 7Benit and Brown (1990)(ion does not stop in sample) 8Reimann et al (1984) (ion does not stop in sample) 9Baragiola (2003) 10Sieger et al(1998) (steady-state value) 11Westley et al (1995) (steady-state value as no fluence dependence seen) 12Moore andHudson (2000) 13Gomis et al (2003)

bG values are for escaping O2 For H2O2 and O2 formation both of which require multiple events yields after sat-uration are used In a few cases yields are obtained by melting the sample Values in parentheses are those at anothertemperature

cPercent of film eroded as O2 molecules for long-term irradiation

ldquosaturatedrdquo yields change slowly with fluenceMore importantly O from water molecules at aconsiderable depth into the sample contributedto the O2 ejected from the surface Since there wasno observed diffusion of the O2 in ice at the am-bient temperature (7K) over the time for their experiment they concluded that the oxygentransport and production are stimulated by theincident ions The transport was suggested as be-ing due to percolationdiffusion along the dam-aged and transiently heated tracks (Reimann etal 1984 Benit and Brown 1990) They also foundthat in fully eroding (sputtering) thin ice filmsthe total amount of O2 produced required about10ndash15 of the D2O molecules in the films at T 57K On warming irradiated samples some O2

formed at depth is found to remain trapped forexperimentally relevant times (see eg Hart andPlatzman 1961 Benit and Brown 1990 Baragiolaand Bahr 1998)

Matich et al (1993) detected luminescence in77K H2O and D2O ice from newly formed and ex-cited O2 They suggested that the O2 is formed

from the interaction of a freshly produced mo-bile O interacting with a previously producedand trapped O or H2O2 They favored the formerIn matrix isolation studies of peroxide irradiatedby UV photons luminescence from newly formedexcited O2 was seen However the oxygen wasformed from mobile O (greater than 30K in Krlattice) interacting with O trapped as H2O-O(Khriachtchev et al 2000) These studies also sug-gest that trapped O possibly in the form H2O-Omay be an important precursor

In Fig 4a the O2 yield for low-energy electronsplotted as a function of electron energy exhibits anapparent ldquothresholdrdquo at 10 eV (Orlando andSieger 2003) Although the uncertainties are largethe authors suggested there may be a lower thresh-old 6 eV corresponding to a very small crosssection for O2 production (10220 cm2) This couldbe associated with an excitation at the ice surfaceBulk ice exhibits very weak absorption at photonenergies below 6 eV but exhibits a strong ab-sorption that depending on the ice structure be-gins at about 73 eV with a peak at 86 eV This

OXIDANTS AT EUROPA 829

b

a

FIG 3 a Relative yields versus fluence (ion flux 3 time) for 15 MeV Ne1 incident on D2O ice at 7K (adapted fromReimann et al 1984) Top panel D2O yield Middle panel D2 yield Bottom panel O2 yield Yields at other tem-peratures exhibited similar dependences with the slope of O2 yield versus fluence varying with T Fits give activa-tion energies between 002 eV at the lowest fluences and lowest T to 007 eV at the highest T and highest fluencesb Oxygen yield versus fluence of low-energy electrons on D2O ice (from Sieger et al 1998)

is followed by a peak in absorption at 10 eVThe results in Fig 4a show that there is a dra-matic increase in the production of O2 at 10 eVwhich is the onset of secondary electron produc-tion in the bulk (Orlando and Sieger 2003)

For incident fast ions the yield at a given tem-perature was found after a small fluence to beindependent of whether the sample was crys-talline or amorphous ice (Brown et al 1982)However using low-energy electrons the yieldswere found to be smaller for an initially crys-talline ice (Orlando and Sieger 2003) Since verylow doses of ion radiation (a few eV per mole-cule) can convert crystalline ice to amorphous iceat low temperatures (Leto and Baratta 2003) thedamage produced by the penetrating ions usedby Brown et al (1982) affected the O2 yield Thestructural damage produced may create trappingsites cause changes in the electronic relaxationpathways or enhance percolation

The temperature dependence of the O2 yield isshown in Fig 5 for electrons incident on initiallyamorphous (Fig 5a) and initially crystalline (Fig5b) ice A small decrease in efficiency is seen inboth cases at 120ndash130K where Ic forms fromamorphous ice and pore collapse occurs (Row-land et al 1991) This is followed by an increas-ing yield with increasing T and then by a more

dramatic drop near 160K where Ih forms Dur-ing the transition to Ih trapped molecules havebeen shown to escape efficiently from ice samples (Ayotte et al 2001) A large drop in the yield isalso seen in Fig 2b for incident Ar1 ions

In modeling the dramatic decreases in the yieldin Figs 2b and 5 care must be taken That is thetime for crystallization of an amorphous sample(Baragiola 2003) even allowing for the large un-certainties is 30 min at 145K and becomes 1min at 160K Therefore pulsed radiolysis exper-iments should be carried out at these tempera-tures and the transients observed Since the timefor conversion from amorphous to crystalline icecan be shorter than the experimental times at145K the results are not comparable to thosefor which the chemical state of the ice at a givenfluence is stable over experimental times Belowwe focus on the data below 140K

We have emphasized processes initiated by theelectronic excitations and ionizations producedHowever momentum transfer collisions whichare the dominant energy loss process for low-en-ergy ions also produce dissociations defects andO2 in ice (Bar-Nun et al 1985) The relative effi-ciency of momentum transfer and electronicprocesses for producing O2 needs further studyAs the penetration depths are much smaller when

JOHNSON ET AL830

FIG 4 a Relative yield of O2 from D2O ice versus incident electron energy E at T 5 120K (from Orlando et al1998 Orlando and Sieger 2003) b Relative yield of D2 from D2O ice versus incident energy E at T 5 88K (fromKimmel et al 1994)

a

b

the former dominates the produced O2 likely es-capes readily

H2 formation and loss correlation with O2

The production of O2 from a low temperatureice by radiolysis correlates with the formationand loss of H2 (or D2 from D2O ice) (Reimann etal 1984) D2O is used to reduce the backgroundin mass spectrometry studies The yields of O2from D2O and H2O ices were shown to be com-parable However the total yields from D2O wereabout 10 less (Reimann et al 1984) consistentwith H (D) transport being important It is seenin Fig 3 that D2 is promptly produced and lostfrom the D2O ice At higher fluence the D2 yieldis seen to exhibit a dependence like that for O2For the 15 MeV Ne1 ions which fully penetratethe ice sample the D2 yield eventually decreasessince D2 is produced and escapes from the fullthickness of the film

Because of H2 (D2) loss the surface region of

the irradiated ice becomes depleted in H (D) rel-ative to O favoring the formation of oxygen-richspecies The change seen in the ratio of H (D) toO in the irradiated ice is small An irradiated film(100 nm) remains stoichiometric to better than2 (Brown et al 1980) and in the near surface re-gion Auger spectroscopy suggests only a slightoxygen enhancement due to irradiation by keVelectrons (Rye et al 1978) Sieger et al (1998) sug-gested a precursor concentration of the order ofa percent Such concentrations are an order ofmagnitude smaller than the steady-state concen-tration of dissociated H2O (Table 2)

Using infrared spectroscopy the damaged H2Oconcentrations at low temperatures in the satu-ration regime are about 10 over the irradiatedvolume (Benit et al 1987 Watanabe et al 2000Gomis et al 2003) This fraction is the ratio of thedissociation cross section to the effective crosssection for recombination to H2O as discussedlater At a 10 concentration most H2O will havea neighbor that is dissociated

OXIDANTS AT EUROPA 831

FIG 5 a Relative steady-state yields of O2 from a D2O ice sample versus T (upper panel) The ice samples wereformed and irradiated at these temperatures Film erosion as a yield (lower panel) (from Orlando and Sieger 2003)b Relative steady-state yields of O2 versus T as in (a) (upper panel) Here the D2O ice samples are formed at 160Kpresumably becoming crystalline and then irradiated at the T shown A new sample is used for each data point in(a) and (b) Film erosion as a yield (lower panel) (from Orlando and Sieger 2003)

At temperatures above 40K it is seen in Fig2 that the D2 and O2 yields exhibit the same tem-perature dependence suggesting a correlationbetween D2 and O2 loss H2 (D2) can be formedby reactions between diffusing hydrogen fromdissociated water or directly [ie H2O R H2 1O (eg Kimmel and Orlando 1995) H2O R (H1 OH)cage R H2 1 O (eg Matich et al 1993)H3O1 1 e R H2 1 OH (eg Williams et al1996)] Subtraction of the nearly constant D2 yieldin Fig 2a gives the Arrhenius plot in the insetSince molecular hydrogen readily escapes atthese temperature it is the H2 (D2) formationprocess that has an activation energy of 003eVmolecule The fact that slopes of the yields forH2 and O2 above 40K are similar suggests thatH2 (D2) formation and loss is the rate-limitingstep in determining the production and loss of O2at these temperatures

Although the data at low T in Fig 2a have con-siderable uncertainties below about 20ndash40K theD2 production is roughly independent of tem-perature but the O2 yield continues to decreasewith decreasing temperature Radiolytically pro-duced D2 is mobile and can escape above15ndash20K (Watanabe et al 2000) but O2 is im-mobile at such temperatures and the incidentHe1 ions are much less efficient at causing per-colation than the incident Ne1 discussed earlier

The production of H2 (D2) from ice has alsobeen studied using low-energy electrons and UVphotons at very low temperatures In the formerstudies absolute yields were not obtained butthere appeared to be a threshold at 63 eV (Fig4b) This is roughly the threshold for excitationof a surface state (Khusnatdinov and Petrenko1992) and the threshold for peroxide productionis close to this value The efficiency for H2 (D2)production increases around 11 eV with a sec-ond steep rise at 17 eV (Kimmel et al 1994) TheO2 and D2 yields versus electron energy in Fig 4exhibit some similarity in the threshold regionalthough the uncertainties are large and the sam-ple temperatures different Measuring the H2 (D2)and the O2 ejecta simultaneously over a range oftemperatures as in the Ne1 experiments wouldbe instructive

Using 98 eV photons at low T (12K) Watanabeet al (2000) measured the D2 ejected and trappedin ice They obtained a cross section for produc-tion of D2 (Table 2) at low fluences that is10ndash20 of the total absorption cross sectionwith significant uncertainties They also found

that the equilibrium damage fraction in the sam-ple was 10ndash20 of the D2O molecules but theequilibrium production of D2 was about an orderof magnitude smaller 1ndash2 at their tempera-tures They suggested that the D2 is formed atgrain interfaces and cracks and on the surfaces ofpores In a mixed ice D2O 1 CO exposed to UVphotolysis D2 and CO2 formation are correlated(Watanabe and Kouchi 2002) The D2O dissoci-ates to produce D2 plus O and this transientlymobile O reacts with CO to produce CO2

Precursor state and peroxide

In order to produce O2 from low temperatureice it appears that the ice sample must lose hy-drogen and form trapped oxygen-rich reactivespecies The efficiency of production also appearsto be affected by the level of defects and damagein the ice consistent with the suggestion that thehydrogen-bonding network is important in theproduction of O2 via availability of trapping sitesnew electronic relaxation pathways or the natureof the bonding of precursors Here we briefly ex-amine the possible role of the suggested precur-sor molecules H2O2 and HO2 We note that H2O2might not readily dissolve in ice possibly ac-counting for its low abundance in some experi-ments It appears to be heterogeneously distrib-uted in the form of microcrystals (Gurman et al1967) If H2O2 does act as a precursor the for-mation of aggregates could affect the dimer ab-sorption bands (Cooper et al 2003a) and affectthe production and stability of O2 (Cooper et al2003b)

The results of Gerakines et al (1996) in whichice at 10K is irradiated by UV photons confirmthat H2O2 forms in ice from two OH after a build-up of trapped OH They monitored the band as-sociated with H2O-HO (Langford et al 2000) inwhich the HO is bound to water with 024 eVTherefore they might only be monitoring a sub-set of the trapped OH seen in earlier experiments(eg Taub and Eiben 1968) The H2O2 concen-tration in ice reaches a steady-state level at flu-ences of about 1018 photonscm2 (Gerakines et al1996) As seen in the model section this gives anestimate of the H2O2 photodestruction cross sec-tion of the order of 10218 cm2 Similarly HO2 be-gins to be seen in absorption at about a few times1017cm2 shortly after they begin to detect H2O2The absolute yields were not measured

Westley et al (1995) detected H2 and O2 pro-

JOHNSON ET AL832

duced by Lyman-a (98 eV) radiation of ice butfound no fluence dependence for fluences greaterthan 1017cm2 They also saw small amounts ofH2O2 by programmed thermal desorption of thesample but made no quantitative evaluationComparing their lowest fluence for the produc-tion of O2 with the result of Gerakines et al (1996)for peroxide suggests that H2O2 and HO2 areprobably not important precursors of O2 In UVphotolysis of H2O2 using 64 eV photons the pro-duction of O2 was not observed (Vaghjiani et al1992) and the excitation of an irradiated icewhich presumably contained some peroxide by37 eV photons did not produce O2 (Baragiola etal 2003) Direct excitation of matrix-isolatedH2O2 at low temperature also did not yield O2but the production of a mobile O that reacted withtrapped O (H2O-O) did give O2 (Khriachtchev etal 2000)

Irradiating with 200 keV protons at tempera-tures of 40ndash129K Bahr et al (2001) found onlyabout 0001ndash002 H2O2 using programmed ther-mal desorption as compared with earlier experi-ments that suggested larger values (eg Table 2)Using infrared absorption Moore and Hudson(2000) found a G value of 01 for incident MeVprotons at low temperature but the signal in theobserved band became negligible at 80K sug-gesting that H2O2 was not being formed Theyalso found that adding O2 significantly increasedthe production efficiency of H2O2 That is thatthe amount of peroxide was affected by theamount of trapped oxygen Based on the matrixisolation studies (Khriachtchev et al 2000) dis-sociated mobile O might trap as H2O-O whichon excitation converts to H2O2 Therefore even ifH2O2 is not the precursor the peroxide andtrapped oxygen concentrations will be related

Although molecular oxygen was not a productof photolysis of matrix isolated H2O2 it was aproduct of photolysis at 266 nm (466 eV) of per-oxide dimers in an argon matrix 2 H2O2 R 2 H2O1 O2 (Engdahl and Nelander 2002) This de-composition was thought to be due to the con-siderable distortion of the bonds in the dimerTherefore the configuration or distortion by thelattice of peroxide or any other precursor mole-cule in the ice may be critical (Sieger et al 1998)In addition more recent experiments show thatG values for the production of H2O2 by 30 keVincident ions are significant as are the steady-state ratios of H2O2H2O (Gomis et al 2004) Ra-tios of the order of 02ndash4 were found at 16K and

77K in the irradiated regions with incident H giv-ing the smallest fraction and incident O thelargest (Table 2) At these concentrations perox-ide molecules might segregate forming inclu-sions (Gurman et al 1967 Cooper et al 2003b)Since the amount of trapped peroxide also com-petes with the amount of trapped O studies areneeded to describe the chemical state of the solidproduced by radiolysis This will be critical in de-scribing the fate of the irradiation products at Eu-ropa In particular the chemical composition fa-vorable for producing O2 needs to be determinedas does the importance of trapped O H2O2 HO2or peroxide dimers as precursors

Other species

In the model below we will refer to atomic andmolecular species trapped in ice Such species areoften discussed in the literature on radiolysis andphotolysis (for summaries see Matich et al 1993Johnson and Quickenden 1997) Trapped OH OH2O2 and HO2 have been discussed andor mea-sured by a variety of techniques (absorption andluminescence spectroscopy electron spin reso-nance programmed thermal desorption etc)Care must be taken when using these data as re-sults for ice samples formed below 160K orthose exposed to penetrating ion beams likelyhave numerous defects voids and grain inter-faces so that a variety of trapping sites and trapdepths exist Therefore when a particular bandfor a trapped species (eg OH) disappears withincreasing temperature it could be due to achange in the nature of the trapping as well as toloss of the species by reaction

Although H is discussed below we do not dis-tinguish it in these simple models from H1

trapped as the hydronium ion H3O1 whichwould mean that there are corresponding nega-tive ions present Therefore below we refer to thevarious species as if they are neutral althoughthis might not be the case Although trapped elec-trons are detected only over a narrow range oftemperatures (100ndash120K) negative ions must beimportant over a broad range of temperature

We also note that SO2 and a sulfate and CO2and a carbonate appear to be trapped species inEuroparsquos icy surface Under radiolysis each ofthese can dissociate directly giving O2 Howeverwith the exception of the sulfate their densitiesare less than 1 so they are not relevant sourcesof O2 However their presence is indicative of the

OXIDANTS AT EUROPA 833

preference for forming oxygen-rich molecules inEuroparsquos ice and can affect the efficiency of pro-duction of O2 and H2O2

MODELS FOR FORMATION OF O2

Introduction

Based on the data described above the yieldsfor incident electrons and ions are found to benearly linear in fluence at the lowest fluences forwhich O2 is detected In addition the relevant al-terations of the ice samples by radiolysis appearto be stable and independent of the beam fluxRemarkably this is the case both for low-energyelectrons that make only a single excitation foreach impact and fast ions that make a density ofexcitations along their path through the solidThis suggests that O2 is produced and trapped inthe initial events and then caused to escape fromthe solid by the subsequent radiation or that astable chemical alteration of the ice is required be-fore the radiation can produce and eject O2

Because H produced by the dissociation of wa-ter molecules can attack trapped O2 and any pre-cursor oxidants the loss of H is important Thisoccurs by the production and loss of hydrogenmolecules as discussed leaving a surface that isslightly rich in O relative to H The O2 produc-tion mechanism could be direct excitation of aparticular precursor species or it might not bespecific That is a new OH or O produced by theincident radiation could react with the H-depleted solid producing O2 This would occurmore efficiently as the ratio of O to H in the nearsurface region increased

In their early work WL Brown and colleaguesassumed that processes similar to those used todescribe radiolysis of liquid and gaseous H2Ocould be used to describe the radiolysis in ice Thelow activation energy that they found was as-sumed to be associated with diffusion of somemobile radical species in ice such as H or O orpossibly diffusion of the O2 molecule itselfBased on their studies of ejection from isotopi-cally labeled D2O samples they concluded thatthe O2 is formed in the track of a single ion andthen ejected by subsequent ions (eg Benit andBrown 1990) More recently trapped specieshave been suggested as precursors These are ei-ther subsequently excited by the incident radia-tion or attacked by newly produced mobile

species We consider a simple model that incor-porates all of these processes and can be used toorganize and interpret the experimental data Wealso discuss its relationship to more completemodels In this discussion much of which is in theAppendix we consider the standard radiationproducts that have been often described How-ever in a solid the excited electron cloud pro-duced by absorption can overlap a number of lat-tice sites Therefore their relaxation pathwaysandor product species could be very differentfrom those that have been traditionally consideredin the literature on radiolysis and photolysis

Precursor model

The column density of a precursor species Npformed in ice by incident radiation can be calcu-lated using a simple model In this model cross sec-tions are used for the production sp and destruc-tion s of the precursor O2 is then formed byexcitation of the precursor or by reaction of a newlyformed radical with the precursor This is also de-scribed using a cross section sO2 For incident elec-trons having a small penetration depth the O2 pro-duced is assumed to rapidly diffuse to the surfaceand escape (Sieger et al 1998) More generally wedefine N as the column density of H2O from whichthe incident radiation can produce and eject O2 Apair of rate equations gives both Np and the pro-duction rate of oxygen dNO2dt That is

H2O 1 radiation R precursor

precursor 1 radiation R O2

giving

dNpdt 5 sp f N 2 s f Np (1)

dNO2dt 5 sO2 f Np (2)

In Eq 1 f is the incident flux and the total pre-cursor destruction cross section can be written ass 5 sp9 1 sO2 with sp9 accounting for precursordestruction processes other than the formation ofO2 Solving Eqs 1 and 2 and assuming N is nearlyconstant (10 level as discussed) the fractionof precursors cp in the affected column after timet is obtained (Appendix)

cp 5 NpN 5 [sps] [1 2 exp(2 s F)] (3)

Here F 5 [ft] is the fluence (flux 3 time) which

JOHNSON ET AL834

is the number of particles (photons electrons orions) incident on the surface per unit area Usingcp in Eq 2 the O2 yield Y at any fluence can bewritten in terms of the yield at steady state Y`

Y 5 [dNO2dt]f 5 sO2 cp N 5 Y`

[1 2 exp(2sF)] (4a)

Y` 5 [sp sO2 s] N (4b)

If we also account for the reduction in the den-sity of water molecules in the initial column Nthese forms change only slightly The size of thetwo parameters Y` and s can be obtained by fit-ting the yield versus fluence data such as that inFig 3

For the low-energy electron data Sieger et al(1998) showed that the ratio [YY`] for a givenincident electron energy is independent of tem-perature over a narrow range of temperatureTherefore the destruction cross section s in Eq4a is roughly independent of T and can be esti-mated from the [YY`] curve Sieger et al (1998)found a precursor destruction cross section s lt08ndash30 3 10216 cm2 for incident electron energiesfrom 30 to 100 eV This is of the order of the elec-tron impact ionization cross section for H2O

Based on the above model the lack of an ob-served fluence dependence for fluences down to1017 hncm2 in the Lyman-a irradiation of ice(Westley et al 1995) can be examined The low-est fluence used implies that the precursor de-struction cross section for Lyman-a is of the or-der of the Lyman-a absorption cross section in ice(09 3 10217 cm2)

Sieger et al (1998) also assumed that sp9 5 0This would mean that the precursor is only de-stroyed by production of O2 (ie s 5 sO2) Thatis unlikely since suggested precursors can be de-stroyed by reaction with H and other radicalsTheir assumption would also mean that the crosssection for production of O2 from a precursorsO2 is large This would be surprising and ap-pear to contradict some of the data discussedabove Instead we assume that s is not equal tosO2 and the total destruction cross section s isof the order of the ionization or absorption crosssection

Having obtained s the measured temperaturedependence for production of O2 is containedonly in Y` Over a narrow range of T one canwrite Y`(T) lt Y`9 exp(2EakT) and obtain the es-

timate of the activation energies in Table 2 Ifsp9 THORN 0 as discussed then sp and sO22 in Eq 4bcannot be obtained separately from the data forY` unless an independent measure of cp is ob-tained Such a measurement is critical for de-scribing the chemical state of the irradiated sur-faces

The observed temperature dependence of Y` iscontained in sp sO2 or N Here N is the ldquodepthrdquofrom which newly formed O2 can reach the sur-face and escape given as a column density of H2OThe physical depth is obtained by dividing thecolumn density N by the molecular number den-sity of the ice Assuming that in the low-energyelectron experiments N is independent of tem-perature and is about three monolayers (Sieger etal 1998) the measured Y` gives [sO2sps]05ndash20 3 10218 cm2 over the electron impactenergy range At 110ndash120K this ratio was foundto be proportional to (E 2 Eo) for E Eo 10 eVagain assuming fixed N (Orlando et al 1998 Or-lando and Sieger 2003)

In the following section we first consider a pre-cursor like peroxide that typically forms fromtwo dissociation events We use this to show un-der what circumstances such a model does notviolate the observed linearity of the O2 yield datawith fluence Following that we relate the sim-ple precursor model above to the case wheretrapped O is the precursor We then use themodel for trapped O to examine the effect of thetrap density on the production of O2 The densityof traps (defects) is determined by the sample for-mation temperature and the subsequent radiationdamage Finally we consider the escape depthN for penetrating radiation within the precursormodel These discussions are supported by moredetailed rate equations in the Appendix

Precursor trapped peroxide

The proposed precursor peroxide is typicallyformed from products of two dissociation eventsie 2 H2O R 2 H 1 2OH R H2 1 H2O2 This wasconfirmed in low temperature ice (Gerakines etal 1996) It is also possible that a dissociation pro-ducing a trapped O followed by an excitationevent can produce peroxide 2 (H2O 1 hn) R H2

1 H2O-O 1 hn R H2O2 1 H2 (eg Khriachtchevet al 2000) The O2 yield versus fluence for inci-dent electrons is reasonably well fit by the yieldexpression above for fluences greater than1014cm2 Therefore it has been suggested that

OXIDANTS AT EUROPA 835

any precursor species must be formed in a singleevent Using a model in the Appendix in whichthe precursor is formed by two events we showthis is not necessarily the case Precursor forma-tion if it is fast could involve two dissociationevents and give the measured fluence depen-dence Although we consider peroxide here theseconclusions apply to other possible precursors

In the example in the Appendix H2O is disso-ciated to H 1 OH according to the cross sectionsd The reactions of two OH units to form H2O2(rate constant k2OH) and 2 H to form H2 (k2H)compete with the recombination of H and OH toH2O (kHOH) If k2H is much greater than k2OH then H2 loss dominates at low fluences as ex-pected (Appendix) Steady state requires not sur-prisingly that the rate for forming H2O2 isroughly equal to the rate for forming H2 Finallyit is seen in the Appendix that the low H2O2 con-centrations suggested by experiment imply thatkHOH (k2H k2OH)05 Such sizes are reasonable

The set of rate equations in the Appendix thatuse number densities of species is integrated overdepth in order to obtain the forms used in Eqs 1and 2 which are based on column densities Inthis way the effective cross sections are related tothe rate constants for the interaction of freshlyproduced mobile species with previously pro-duced trapped species We show in the Appen-dix that the O2 yield versus fluence can be writ-ten in a form similar to that in Eq 4a but with asmall offset due to the two-step precursor for-mation process

Y lt Y`(T) [1 2 exp(2sF) 2 (ssr)]

for F sr21 (4c)

Here s is again the precursor destruction crosssection but sr 5 (2s2OH 1 sHOH) is the effectivecross section for loss of OH by formation of H2O2

and H2O The corresponding expression for Y`(T)is given in the Appendix

The form for the yield versus fluence in Eq 4cis consistent with the low-energy electron mea-surements if (ssr) 1 and if sr F21 lt10214 cm2 Such sizes are indeed possible Whatis required is that at the relevant temperatures(50ndash110K) the interaction length associated withsr is more than a few molecular layers (greaterthan 10 Aring) A cross section that is likely to belarge above 20K is sHOH which describes themean distance for diffusion of mobile H before it

reacts Since s discussed earlier is of the sameorder of magnitude as the H2O dissociation crosssection sd then the precursor concentration cp(in this case H2O2) is roughly proportional to[s2OHsHOH] 1 (Appendix)

Based on the above a two-step precursor for-mation process could be consistent with the elec-tron data contrary to the statements in Sieger etal (1998) Orlando et al (1998) and Orlando andSieger (2003) Experiments are needed to deter-mine the relevant cross sections Below we con-sider a precursor formed in a single excitationevent

Precursor trapped O

O2 can be formed for instance by diffusion ofa transiently mobile O to a previously formed andtrapped O or to the production of O near the siteof a previously produced and trapped O (Matichet al 1993 Johnson and Quickenden 1997) Werefer to this as the MatichndashJohnsonndashQuickenden(MJQ) model and note that O 1 OH R O2 1 Hcan also give molecular oxygen

Trapped O as H2O-O was in fact seen to be aprecursor to O2 when attacked by a mobile O ina Kr matrix (Khriachtchev et al 2000) ThereforeO trapped as H2O-O was also suggested as a pre-cursor in ice (Johnson 2001) Before proceedingwe note that the examination of the MJQ modelin Sieger et al (1998) was incorrect They consid-ered two mobile interacting species whose life-times were long compared with the bombard-ment rate rather than a new mobile O interactingwith a previously trapped O Although this hasnot been corrected recently they added Otrapped as H2O-O as a possible precursor (Or-lando and Sieger 2003) O trapped as H2O-O isseen in matrix isolation studies as discussedabove but it has not yet been identified in iceand like OH it may trap in a number of ways

The MJQ model is essentially the followingFormation of H2 1 O is followed by loss of H2and trapping of O (Ot) A subsequent dissocia-tion or excitation event can lead to the destruc-tion of Ot by reaction with for instance H or byconversion to H2O2 However if this dissociationevent also leads to H2 1 O a mobile O is pro-duced that can react with Ot with an interactionlength that depends on T This results in the for-mation of O2 [O 1 Ot R O2] giving a net pro-duction of 2 H2 and 1 O2 In the Appendix therelevant rate equations (Eqs A1cndashA3c) are con-

JOHNSON ET AL836

sidered Assuming the mobile O processes arefast these rate equations can also be integratedand simplified to give the form in Eqs 1 and 2The precursor column density Np now repre-sents the column density of Ot In this model sO2in Eq 2 is the ldquocross sectionrdquo for reaction of afreshly produced O with an Ot and sp9 indicatesdestruction of Ot Finally sp is the cross sectionfor production of H2 and Ot via processes dis-cussed above

Based on the sizes of the cross sections ex-tracted in the electron irradiation experimentsthe model with trapped O (Ot) as the precursoris quantitatively reasonable That is for a rangeof T both sp and sO2 might be described byevents of the type H2O R H2 1 O (whether di-rect or indirect) If these occur for 10 of theradiation-induced dissociation events as dis-cussed then the required product of cross sec-tions suggested by the sizes of s and Y` above isroughly obtained that is [sp sO2] 02ndash6 3 10234

cm2 in the low-energy electron experiments Thecross section for production of O might dependon the availability of dangling bonds at interfacesin pores or at defects as suggested by a numberof authors The importance of such bonds couldalso be consistent with the differences in the elec-tron-induced yield for initially crystalline and ini-tially amorphous ice (Orlando and Sieger 2003)

O2 production can increase in the presence ofoxygen-rich impurities such as CO2 or SO2 bothof which are known to be present at Europa Asin the MJQ model dissociation (eg CO2 R CO1 O) can provide a mobile O that can be trappedor react with a previously trapped O forming O2In addition such species are hydrogen scav-engers (Moore and Hudson 2000) This can re-duce the destruction of Ot affecting sp as seen inEq A5c

Trap density

In the simplified models above the density oftrapping sites nt was not explicitly accountedfor However in the discussion of both the ex-perimental data and models trapping sites havebeen assumed to be important It has beendemonstrated repeatedly that the density of de-fects voids and interfaces decreases as the tem-perature of formation of the ice increases How-ever the efficiency of production of O2 is theopposite it increases with increasing T There-fore the oxygen yield varies in a manner that is

the inverse of the trap density over a range oftemperatures For penetrating ions in which theenergy density deposited is high and O2 might beformed in a single track a simple explanation hasbeen given As the density of traps for O2 is re-duced and the temperature increases the O2 pro-duced can diffuse to the surface and escape moreefficiently limiting back reactions on further ir-radiation This is examined in the next sectionHowever such a process does not affect the low-energy electron yields for which the penetrationdepth is small and the formed O2 escapes read-ily at the measurement temperatures In Eqs A1cA1c9 A3c and A10 in the Appendix we againuse a simple set of rate equations for trapped andmobile O in order to determine the role of the trapdensity nt on the production of a precursorspecies The conclusions apply to any modelbased on a trapped precursor

In the MQJ model mobile O can be trappedforming Ot (described by kOt nt) and can reactwith previously trapped O to form O2 (kOOt) Inaddition it can react at other sites and Ot can bedestroyed by mobile reactive species such as H(kHOt nH) Mobile H can also react with othertrapped species (kHR nR) Incorporating these re-actions Eq A10 in the Appendix gives a steady-state yield Y` When the density of traps is largethe steady-state yield is given in Eq A5c

Y` R (sd9sd9sd) [kOOtkHR nr][kHOt kOt nt] N (5)

Here sd and sd9 are the dissociation cross sectionsthat determine the production of H (H 1 OH) andO (H2 1 O) by the incident radiation Comparingthe result in Eq 5 with Y` in the simple precur-sor model in Eq 4b the dissociation cross sec-tions in Eq 5 are now scaled by a ratio of the re-action rates

It is seen in Eq 5 that the steady-state yield Y`depends inversely on the density of traps nt Thatis there are competing destruction processes fortrapped O (Ot) Therefore in order to form O2 amobile O competes with mobile H to find atrapped O before it becomes trapped In this waythe increasing O2 yield correlates with the re-duction in the trapping sites with increasing tem-perature of formation The small activation ener-gies discussed earlier would be related to there-orientation of the lattice binding leading to thereduction in trap density At relatively high T thedensity of traps becomes small so that only the

OXIDANTS AT EUROPA 837

sample surface or grain interfaces might act astraps accounting for the observed drop in theyield at high T For this reason it is important todetermine if the yield for T greater than 150Kdepends on the incident flux f Since other quan-tities in the expression for Y` could also be tem-perature dependent experiments are neededThese would likely be pulsed radiolysis experi-ments in which the transients are studied ratherthan the steady-state yields However the avail-able data for lightly ionizing radiation do suggestthat the O2 yield depends inversely on trap den-sity as in the model above

Escape from depth

The quantity N in Eq 1 is the ldquodepthrdquo given asa molecular column density from which O2 formedbelow the surface can percolate to the surface andescape It can be converted to a physical depth bydividing by the molecular number density of theice In the case of the low-energy electrons thedepth below the surface at which ionizations andexcitations are produced is small However forfast ions that penetrate many monolayers this isnot the case Below we examine N in Eq 1

Although it has long been understood that twoexcitations events are required to produce O2(eg Brown et al 1982 Reimann et al 1984) theprecursor model suggests that at saturation(steady state) any new event can produce O2 fromthe steady-state density of trapped precursorsTherefore one might expect the yield at satura-tion to be nearly linear in the excitation cross sec-tion This is the case for the low-energy electronsand it was initially assumed to be the case for thefast ions (Brown et al 1980 Reimann et al 1984)However Baragiola et al (2003) showed that thetotal sputtering yield in the saturation region attemperatures greater than 100K which is dom-inated by decomposition (Brown et al 1982) isproportional to the square of the electronic en-ergy deposition per unit path length in the solid(dEdx)e Assuming that the production and de-struction cross sections in Y` in Eq 4b roughlyscale with (dEdx)e then their ratio [sp sO2s]also roughly varies as (dEdx)e If this is the casethen the observed quadratic dependence impliesthat the depth N from which O2 can be mobi-lized to escape must also depend on the excita-tion density (dEdx)e This is consistent with theobservation that the O2 yield depends on thick-ness for samples thinner than the ion penetration

depth (Reimann et al 1984 Benit and Brown1990) Therefore the percolation depth for escapeN is a critical factor in comparing yields for pen-etrating radiation with those for non-penetratingradiation like the low-energy electron and low-energy ion data It is also seen that percolationand trapping can be included within the simplemodel considered here

Model summary

We have examined the simple precursor modelintroduced by Sieger et al (1998) to explain theirelectron data We have shown how it can be gen-eralized and derived from an integration of theappropriate chemical rate equations with thecross sections and total column density contain-ing the reaction rates and the physics of diffusivetransport Therefore the parameters can havevery different meanings from those initially as-sumed The equations are useful when there arefast processes which go to completion in timesmuch shorter then the time between particle im-pacts and slow processes associated with speciesthat are assumed to be trapped between particleimpacts This separation allowed us to examinethose models proposed for precursor formationand destruction due to reacting mobile speciesformed from dissociation events Further fromthe size of the onset of the linearity of the yieldversus fluence we can place constraints on therole of nonlinear processes such as those thatmight be associated with the formation of a pre-cursor like peroxide

With these extensions this simple model is use-ful for approximating the full rate equations inthe temperature region for which the yields de-pend on the fluence (dose) and not on the doserate Based on the assumptions in the model thisimplies temperatures at which precursor speciesare stable over experimental times At the high-est temperatures for the yields shown in Fig 4athis might not be the case This should be testedexperimentally In this regard it is puzzling thatexperiments have shown that the absorptionbands associated with certain trapped speciessuch as OH and trapped electrons ldquodisappearrdquoat 100ndash120K (see eg Johnson and Quickenden1997) whereas the O2 yield data indicate that rel-evant precursor species must be stable to 140KTherefore the proposed O2 production processescan only be distinguished by directly determin-ing the chemical state of the irradiated ice

JOHNSON ET AL838

It may be the case that a particular precursoris not required as mentioned earlier Rather af-ter a sufficient level of oxidation a newly pro-duced O or OH can react in the oxygen-rich lat-tice producing O2 in a pseudo-first-order processThis can be tested experimentally Indeedtrapped O and peroxide may be changed into oneanother as in the matrix isolation experiments(Khriachtchev et al 2000) Therefore their rela-tive concentrations versus temperature and doseshould be studied

In the model considered here it is seen that theparameters in Y` must account for the strong tem-perature dependence Since the number of bulktrapping sites decreases with increasing T as theice sample is annealed one might expect that theprecursor concentration would decrease We sug-gest that the observed temperature dependenceis determined by the competition between theavailability of trapping sites and mean interactiondistances of a mobile species such as O and HThat is the O2 yield varies inversely with the den-sity of traps over a range of temperatures In thesimple model above this effect is included indi-rectly as shown using the more detailed equa-tions in the Appendix

EUROPA

Oxygen and peroxide production at Europa G values

The dependence of the yield on the thicknessof the ice sample for highly penetrating ions sug-gests that the O2 yield is roughly related to theamount of energy deposited in the sample There-fore the amount of O2 produced and lost fromice in steady state is often given as a G value thenumber of molecules produced per 100 eV de-posited in Table 2 In applying these G values tothe planetary environment care must be takensince the yields are temperature fluence and ex-citation density dependent They also are affectedby the presence of other proton scavengersknown to be present in Europarsquos ice

The G values for peroxide in Table 2 are thosefor production at low fluence but the G valuesfor O2 are for the production and escape of gas-phase oxygen at saturation Therefore they de-pend on the formation and destruction cross sec-tions in the model above Such G values can beused to describe the steady-state source of the am-

bient O2 atmosphere at Europa but are not help-ful in determining the ability to produce and trapO2 at depth into the icy surface Over the timescales of the experiments some O2 is seen to re-main trapped and is not destroyed by radicalsTypically there are a number of trapping sitesand recent estimates of the activation energy foran O2 molecule trapped in a substitutional site(Hori and Hondoh 2002) suggest that it can betrapped for times comparable to the time re-quired for the conversion of amorphous ice intoIh (Ayotte et al 2001) The stability of O2 inclu-sions has not been studied

To understand the relationship between the Gvalues for O2 in Table 2 we note that fast lightions and electrons lose their energy predomi-nantly by producing ionizations and excitationsThe ionization events produce secondary elec-trons that can result in further ionizations Thenumber of ionization events Ni produced by afast light ion stopping in a water ice is Ni 5EoW where Eo is the initial energy and W is theaverage energy expended per ionization event27 eV in water ice Therefore radiolysis by fastions is often described as the net effect of theshower of electrons set in motion by the incidention In this manner the G values for productionof O2 from ice by energetic light ions and elec-trons can be roughly estimated as the sum of theeffects of all the secondary electrons using theyield Ye for 10ndash100 eV electrons That is if weignore the percolation depth N we can write thenumber of O2 produced as the number of ioniza-tion events times Ye which already has in it thedestruction processes That is G NiYeEo YeW 0006 O2100 eV

This estimate of G is seen to be an order of mag-nitude larger than the keV proton value for O2 inTable 2 but is much smaller than the G value forenergetic Ne1 These differences from the simpleestimate are consistent with the observation thatthe yield in the saturation regime for penetratingions varies nonlinearly with excitation densityThe difference from the keV proton G value islikely due to the inability of O2 to escape fromdepth before destruction On the other hand thelarger G value seen using MeV Ne1 is likely dueto the high density of excitation This can resultin more efficient production of O2 increased pro-duction of damage sites for forming O2 andorincreased transport (percolation) along the tran-siently heated track That is the ability of theformed O2 to percolate to the surface during ion

OXIDANTS AT EUROPA 839

bombardment plays an essential role in deter-mining the measured yields as discussed aboveThis effect needs to be evaluated for a planetarysurface subject to particle fluxes that are many or-ders of magnitude smaller than those used in thelaboratory

In the absence of O2 yields for all of the rele-vant ions and energies an estimate of the totalsupply of O2 can be obtained using the G valuesin Table 2 For the energetic oxygen and sulfurions we use G values for neon and argon ionsThese are lower bounds as the data were obtainedat temperatures much lower than the surface tem-peratures at Europa (80ndash110K) For the fast pen-etrating electrons and protons we use the valuefor 200 keV protons in Table 2 For the heavy ionswe used the Ne1 and Ar1 values For compari-son we also give an estimate using the G valuesobtained from the low energy electron data andthe total ionization as discussed above

The G values are combined in Table 3 with theelectron and ion energy flux at Europa Althoughthere is some deflection of the flow due to in-duced fields the net energy flux to the surface isnot reduced much from these values (Paranicaset al 2002) In fact locally produced pick-up ionswhich are formed from the ejecta and acceleratedto the surface can contribute (Ip 1996) Furtheralthough G values are typically independent ofthe incident angle the loss of O2 is a surfaceprocess and is enhanced with angle That is thenumber of O2 produced within the escape depthincreases with increasing angle to the normal(steeper than the inverse cosine) This enhance-ment is 25ndash4 (Jurac et al 2001) Ignoring suchcorrections the total production rate calculated

(23ndash62 3 109 O2cm2s) is a lower bound to theactual production rate from the icy regions of Eu-roparsquos surface Recent modeling of the O2 atmo-sphere suggest that such a source rate is required(Shematovich and Johnson 2001)

Earlier estimates of the O2 source rate were in-correct because O2 production was assumed to bea fixed fraction of the H2O sputtering yield Atthe ambient temperature an ejected H2O sticks ef-ficiently (Smith and Kay 1997) to the surface ofneighboring grains in the porous regolith reduc-ing the sputtering yield for H2O (Johnson 1989)This reduction does not apply to the O2 fractionof the yield enhancing its relative importance toH2O Here we assume that all of the gas-phaseO2 produced can contribute to the ambient at-mosphere even if it is produced at a grain sur-face that is at some depth into the porous regolith

Finally Europarsquos surface is not pure ice every-where That is much of the trailing hemisphereis a hydrated sulfuric acid containing smallamounts of sulfur and SO2 and the leading hemi-sphere contains low levels of carbon CO2 andcarbonates Sodium and potassium atoms aresputtered from the icy regions (eg Leblanc et al2002) O2 production from surfaces containingthese species has not been measured C and S areboth proton scavengers and the other species canbe sources of O as discussed so that the O2 yieldfrom such regions could be enhanced Howeverthere can also be competing processes that mightreduce these rates In fact recent HST observa-tions suggest that the O2 might not be produceduniformly across Europarsquos surface (McGrath et al2000) The analysis of these observations is pre-liminary and is not simple since the observed

JOHNSON ET AL840

TABLE 3 APPLICATION TO EUROPA

OxygenImplantation Energy flux production

Species 109(cm2 s)a 109 keV(cm2 s)(1) 109(cm2 s)

H (10 keV) 0015(1) 12 0084 (072)bO (10 keV) 0015(1) 18 054cS (10 keV) 0009(1) 30 090cElectrons 62 043 (37)bPlasma (10 keV) (137 amu) 13(2) 10 03c

Total 79 23 (62)UV photons (64 eV) 46 01

aReferences are as follows 1Cooper et al (2001) [similar results in Paranicas et al (2002)] 2Mauk et al (1996)bUsing G for 200 keV protons (upper bounds in brackets are based on the low-energy electrons data as a model for

the secondary electrons produced)cUsing G for 15 MeV Ne1 at 7K a rough lower bound

emissions depend on the local plasma propertiesHowever the efficiency of O2 production in con-taminated ices needs to be studied

Porous regolith O2 interactions on grain surfaces

Although the O2 atmosphere is principally pro-duced in the near surface layers and is eventuallylost to space it also permeates Europarsquos porousregolith The porosity p (void space) of the re-golith is uncertain (Buratti 1995) It has been es-timated to be the most compact surface in theOuter Solar System p 025 on the trailing hemi-sphere and p 079 on the leading hemisphere(Buratti and Veverka 1983) and the most poroussurface p 096 globally (Domingue et al 1991Domingue and Hapke 1992) These large differ-ences are due to the uncertainties in the amountof backscattered light at small phase angles andon the photon absorption properties and grainsizes The more compact surface is likely correctGrain sizes rg of the order of 100 mm have beensuggested (Hansen and McCord 1999 2000Geissler et al 1998) Europarsquos regolith differs sig-nificantly from the lunar regolith which is asand-like surface of individual grains Analysis ofthe thermal inertia of Europarsquos surface (Grundyet al 2001) suggests that the ice grains are sin-tered forming a fairy castle-like structure Sin-tering can occur thermally or due to the incidentradiation In addition ices doped with acids andalkalis in the laboratory can be powdery solidswith the impurities primarily found at grain in-terfaces

The porous regolith affects the above estimatesof the oxygen production in three ways First forisotropic bombardment the mean distance d intothe surface at which an incident ion or photonstrikes a grain surface is d (89)rg(1 2 p) Us-ing the above value for rg then d 02 m if theporosity is very large p 96 Therefore de-sorbed molecules can come from a considerabledepth and will interact with grain surfaces Evenfor much lower porosities the effective sputteringyields can be significantly modified (Johnson1989) As discussed above the yields for ejectedH2O which sticks onto neighboring grains at theambient temperatures are reduced but those forejected O2 are not This enhances the relative im-portance of the O2 yields

Finally the surfaces and interfaces of the sin-tered grains in Europarsquos porous surface are placeswhere low temperature chemistry involving the

ambient oxygen can occur at depths that are wellbelow the nominal penetration depth of the inci-dent radiation The relative importance of suchchemistry is roughly indicated by the ratio of thetotal surface area in the regolith to the area of Eu-roparsquos visible surface Using the porosity andgrain size above and a regolith depth dr of theorder of a meter then this ratio is large 3(1 2p)drrg which is 103ndash2 3 104 for average porosi-ties of 96 and 25 respectively Therefore thegas-phase O2 molecules below the visible surfaceinteract frequently with the grain surfaces In thisway a very large volume of the subsurface couldbe altered chemically even if it is not directly ex-posed to the radiation

From the overlying structures seen in the high-resolution images of Europarsquos surface by theGalileo spacecraft (eg Prockter and Pappalardo2000) it is clear that the youngest surfaces are red-der The reddening in the visible has been attrib-uted to the presence of sulfur in the ice (McEwen1986 Johnson et al 1988 Calvin et al 1995 Carl-son et al 1999b) The amount of reddening in thespectra of the surface material decreases with ageand the surface materials appear to becomebrighter with age This might occur by the slowaccumulation of a frost over-layer or by radioly-sis (Carlson et al 2002 Johnson et al 2003) Ra-diolysis preferentially converts the sulfur in theice to sulfur dioxide or sulfuric acid both ofwhich have flat spectra in the visible In regionswhere the radiation does not directly penetratethe O2 permeating the regolith can oxidize thesulfur at the surfaces of the icy grains and at graininterfaces Similarly the carbon seen in the ice onEuroparsquos leading hemisphere (eg Carlson et al2004 Johnson et al 2003) can be oxidized by ra-diolysis in the near surface layers or by reactionson the grain surfaces at depth in the regolith bythe ambient O2 The meteoroid bombardmentthat produces the regolith also exposes fresh sur-faces for oxidation and can cause the collapse ofpores leading to the trapping of oxygen In thismanner both refractory and volatile oxidants canbe formed and trapped in the subsurface ice forpotential delivery to Europarsquos subsurface ocean

SUMMARY

The presence of a thin oxygen atmosphere andoxidants trapped in Europarsquos surface ice (egJohnson et al 2003) combined with the possible

OXIDANTS AT EUROPA 841

presence of a subsurface ocean (eg Carr et al1998) have made Europa an exciting target forfuture space exploration The molecular oxidantsproduced by radiolysis and photolysis could bespecies needed for carbon-based biochemistry inEuroparsquos putative subsurface ocean Howeverthere has been no model for the formation trap-ping and transport of relevant oxidants

In this paper we have reviewed the space ob-servations and laboratory studies related to thechemistry of formation of oxidants in ice by theincident radiation Because the laboratory dataare incomplete we examined the rate equationsfor formation of oxygen and related species byradiolysis and photolysis We showed how suchequations can be simplified to the analytic modelthat was used for fitting the fluence and temper-ature dependence of the O2 yield produced bylow-energy electrons (Sieger et al 1998) In de-riving the analytic model from the more completerate equations the competition between forma-tion and destruction processes can be studied ascan the role of trapping sites escape from depthand temperature history These equations also al-lowed us to put incident photon electron and iondata on a similar footing for the first time

The model can be used to suggest measure-ments that are needed to describe the chemicalstate of the irradiated ice and to interpret new lab-oratory data We showed that if reactions are fastthe yield data could be consistent with a precur-sor being formed by two events as is the case forperoxide However we favor trapped O as a pre-cursor as initially suggested by Matich et al(1993) for production of excited O2 Such aspecies if present could be detected spectro-scopically as can trapped O2 through bands ac-tivated by lattice perturbations (Cooper et al2003a)

The available oxygen yield data were used tocompare G values for different incident particlesThese were then used to estimate the oxygen pro-duction in the radiation environment at EuropaThe lower bounds were consistent with the pro-duction rates obtained from a recent model of Eu-roparsquos atmosphere The porosity of the regolithwas also considered It not only affects the rela-tive sputtering yields for O2 and H2O but alsoprovides an enormously increased surface areaon which the ambient gas-phase oxygen can re-act with species in the ice That this might be oc-curring is consistent with the change in re-

flectance associated with surface aging Such re-actions could also convert ambient molecularoxygen into other more refractory species that re-main trapped on subduction of Europarsquos crust

Although radiolysis produces a stable chemi-cally altered surface in the laboratory the parti-cle flux to Europarsquos surface is many orders ofmagnitude lower Therefore annealing can be ef-fective at temperatures that are much lower thanthose typically found experimentally Belowabout 120K the annealing times appear to be long(eg Baragiola 2003) so that laboratory data canoften be directly applied to Europarsquos icy surfaceHowever account has to be taken of the syner-gism between the UV and the energetic plasmaas discussed earlier (Johnson and Quickenden1997)

Transient surface melting might initiate lifeprocesses but subduction of oxidants to the pu-tative ocean is likely to be required if biochem-istry is to occur at Europa (Chyba and Hand2001) Although the temperature versus depthinto Europarsquos regolith is not well known belowthe regolith the temperature and pressure appar-ently increase with depth At some depth (ap-proximately kilometers) either very warm ice oran ocean exists (Ruiz and Tejero 2000) Thereforechemically refractory oxidants are likely to be re-quired in order to survive subduction to thewarmer subsurface regions Trapped species ifdelivered downward can become mobile andwill segregate forming inclusions (Johnson andJesser 1997) Indeed peroxide appears to segre-gate in ice (Gurman et al 1967) and these inclu-sions may remain relatively stable until the per-oxide is dissolved in the putative ocean Thevolatiles seen in reflectance such as O2 SO2 andCO2 will also segregate But at the higher tem-peratures they could also diffuse through thebulk react or percolate into the regolith atmo-sphere Therefore the observation of the trappedvolatiles in Europarsquos surface ice is of interest pri-marily because it suggests that oxidants are pre-sent The formation of oxygen-rich molecules fol-lows the preferential loss of hydrogen Howeverthe surface is in a local steady state with im-planted H O and S from the plasma and the lossto space Since the plasma flux to the surface canbe variable this balance can also vary Indeedtemporal changes in reflectance have been re-ported (Domingue and Hendrix 2004)

Based on the models for formation of O2 ex-

JOHNSON ET AL842

amined in this paper additional experiments areneeded to determine the chemical state of irra-diated ice at low fluences Of immediate inter-est is the identification of stable molecular pre-cursors required for the formation of O2 Basedon the available laboratory data individual per-oxide molecules trapped in the ice do not appearto be the principal precursors but trapped O(Matich et al 1993) remains a viable candidateAlthough the data suggest that the local hydro-gen-bonding network is important in the pro-duction of O2 the effect of the formation tem-perature and radiation damage on the structureneed to be measured It is also important to de-scribe quantitatively the relationship betweenH2 loss and the formation of O2 and H2O2 overa range of temperatures and fluences Finallyone would like to determine the steady-statedensities of biochemically interesting moleculesformed by radiolysis in an ice containing sulfurand carbon at Europarsquos surface temperaturesand then determine their stability with increas-ing temperature and pressure Such experimentsstill need to be carried out Here we have pre-sented a model that brings together the variouslaboratory data and now can be used to considerthe formation trapping and transport of relevantoxidants at Europa

APPENDIX

Introduction

In this Appendix we examine rate equationsthat describe various aspects of the radiolysis orphotolysis of ice These equations need to be un-derstood if one is to be able to calculate the for-mation trapping and transport of oxidants Theemphasis is on the formation of molecular oxy-gen We integrate these equations over the escapedepth of O2 to give the simplified rate equationsdiscussed in the text in which number densitiesare replaced by column densities (eg Eqs 1 and2) In this way we can examine the relationshipbetween the rate constants and the effective pre-cursor formation and destruction cross sectionsextracted from data We first consider a precur-sor such as H2O2 that is formed from two exci-tation events This bears on the observed linear-ity of the oxygen yield versus fluence We thenconsider the equations for trapped versus a tran-

siently mobile species We use a reaction betweena mobile and trapped O as an example althoughO 1 OH can also give oxygen Finally we ex-amine the possible dependence of the O2 forma-tion process on the density of traps (defectsvoids and grain boundaries) as a function of tem-perature The results are discussed in the text

For clarity all possible radiolytic species arenot included in the discussion below Thereforepotentially important products such as HO2 andthe charged species (eg H3O1 the trapped elec-tron etc) are missing In each section we attemptto make a point related to production of O2 andother oxidants by using a subset of the productsrather than attempt a complete description withmany unknown rate constants

Fast versus slow processes

To describe the O2 production rate from a pre-cursor such as H2O2 we first use a simple but in-structive set of rate equations to describe the for-mation of H2O2 We then relate these to thesimplified rate equation for the precursor in thetext Eq 1 and show how the postulated precur-sor formation cross section sp is related to theH2O dissociation cross section sd by the reactionrates

Using the reactions

H2O 1 radiation R H 1 OH (sd)

H 1 H R H2 (k2H)

H 1 OH R H2O (kHOH)

OH 1 OH R H2O2 (k2OH)

the reaction pathways are described by

H2

Hqk2H

1 OHH2O 1 radiation mdashRsd H

OHH2O2

OH Rk2OH

kHOH

HH2O

OXIDANTS AT EUROPA 843

The corresponding rate equations in an irradiatedvolume are

dnOHdt 5 sd f n 2 2 k2OH nOH nOH

2 kHOH nH nOH (A1a)

dnHdt 5 sd f n 2 2 k2H nH nH

2 kHOH nH nOH (A2a)

dnH2O2dt 5 k2OH nOH nOH

2 destruction processes (A3a)

Here sd is the dissociation cross section f is theradiation flux and n is the molecular numberdensity of ice with nOH and nH the density of dis-sociation products OH and H respectively Therate constants k2OH k2H and kHOH describe thereactions above In this model the H2 formed isassumed to escape for temperatures above about20K We do not distinguish between trappedspecies and mobile species although in fact thatdistinction is important as discussed in the nextsection That is any OH produced at low tem-perature by one ion would likely be trapped priorto interacting with an OH produced by a subse-quent ion At higher temperatures this can breakdown as discussed in the text We first relate theabove set of equations to Eq 1 in the text and thenconsider precursor destruction processes

Equation 1 in the text is obtained by solvingthese equations together and then by integratingover a depth Dx Here Dx is the smaller of the pen-etration depth of the particle and the O2 escape(percolation) depth Assuming the molecular den-sity of ice n is constant with depth Dx n 5 N inEq 1 Comparing Eq 1 with Eq A3a the precur-sor formation rate sp f N is equivalent to [k2OH

nOH nOH] Dx Assuming that the processes in EqsA1a and A2a are fast then the steady state is read-ily obtained In addition if k2H k2OH there ispreferential loss of H2 at short times as suggestedby the laboratory data If steady state is reachedthe production of H2 is equal to that of H2O2 us-ing only the equations above Further one obtains

[k2OH nOH nOH] Dx R [fp sd] f N (A4)

with

fp 5 05[1 1 05 kHOH(k2H k2OH)05] (A5a)

Therefore the precursor formation cross sectionsp in Eq 1 is a fraction fp of the total dissocia-

tion cross sd Since laboratory data suggest theH2O2 fraction in an irradiated ice is small thenkHOH (k2OH k2H)05 which is likely givingfp R [(k2OH k2H)05kHOH]

Trapped versus mobile species

To examine the fluence dependence discussedin the text we note that trapped OH and freshlyproduced transiently mobile OH can be treatedas different species (eg Matich et al 1993)Therefore assuming no trapping of H the OHrate equation above would be replaced by twoequations one for the trapped species nOHt andthe other for the freshly produced mobile speciesnOH Assuming the reaction of two mobile OHunits is unlikely and that peroxide is formed fromthe interaction of a trapped OH with a subse-quently produced OH an expanded set of equa-tions is obtained

OH 1 trap R OHt (kOHt)

OH 1 OHt R H2O2 (kOHOHt)

H 1 OHt R H2O (kHOHt)

then the reaction pathways are described by

qH2

H2O 1 radiation mdashRsdH

qk2H

1 OHH

trap OH

OH RkOHt

OHt kOHOHtH2O2

kHOHkHOHt

H H2O HH2O

The resulting set of equations is

dnOHdt 5 sd f n 2 kHOH nH nOH

2 kOHt nOH nt 2 kOHOHt nOH nOHt (A1b)

dnOHtdt 5 kOHt nOH nt 2 kOHOHt nOH nOHt

2 kHOHt nH nOHt (A1b9)

dnHdt 5 sd f n 2 2 k2H nH nH 2

kHOH nH nOH 2 kHOHt nH nOHt (A2b)

JOHNSON ET AL844

dnH2O2dt 5 kOHOHt nOH nOHt

2 destruction processes (A3b)

where nt is the trap density in the irradiated iceAt low T we can assume nt is large so thatdntdt 0 simplifying the set of equations If theprecursor of interest is trapped O as discussed inthe text then a similar set of equations can be con-structed for O and Ot as in the example to followOf course a full description over a broad rangeof temperatures would require the considerationof the both trapped and mobile H O and OHthe rate of change of trap density and the vari-ous destruction processes

The above equations should be solved simul-taneously but for the incident particle fluxes typ-ically used in these studies the diffusion of mo-bile species to traps is relatively fast This is trueeven at low temperatures if there is a significantdensity of trapping sites This is usually the casefor samples formed at low T Therefore in a lowtemperature solid at low dose rates there is not asignificant steady-state background of diffusingOH There are trapped or freshly produced OHunits which rapidly react Averaging over thebombardment rate this can be approximated bywriting dnOHdt 0 and then separating nOH outin Eq A1b Assuming also that H diffuses rapidlyand either reacts or forms H2 we also writednHdt 0 on the average Therefore betweenimpacts only species trapped at defects pores orgrain surfaces are assumed to be present as sug-gested by the lack of a dependence on the inci-dent beam flux In this limit the rate of change oftrapped OH roughly simplifies to

dnOHtdt 5 2 k2H nH nH 2 2 kOHOHt nOH nOHt

It is seen that the production of peroxide dependson the loss of H by formation of H2 as discussedand at steady state the production rate for per-oxide equals the rate of formation of H2 in theseequations

Treating the mobile and trapped species sepa-rately we can further simplify the rate equationsand obtain the precursor model used in the textThat is integrating over the relevant depth Dxthe densities are replaced by column densitiesThe rate constants ki are then related to effectivecross sections si In this way cross sections aregiven as an interaction length for each freshlyproduced mobile radical [ki nj] R [si f] The

trapped OH column density NOHt and peroxideproduction rate can be estimated from ldquolinearrdquorate equations

dNOHtdt lt [sd f] N 2 2 [s2OH f] NOHt

2 [sHOH f] NOHt (A6a)

dNH2O2dt lt [s2OH f] NOHt 2 s NH2O2 (A6b)

Here sd is again the dissociation cross sectionand the destruction processes in Eq A3b are con-tained in s The second and third terms on theright in Eq A6a are the loss of OH due to the pro-duction of H2O2 and H2O In the integration ofEq A1b9 [2 kOH OHt nOH] becomes [2 s2OH f] and[kHOHt nOH] becomes [sHOH f] Therefore thedensity and the interaction length of mobile H arecontained in sHOH If H2O2 is indeed the pre-cursor then in Eq 2 in the text [sp f N] is be-comes [s2OH f NOHt] from Eq A6b Accountingfor precursor destruction which can give backtwo OH and treating H2 can both be included inthe integrated model via the si values

The simplified set of equations above allows usto examine the relative time (fluence) scales forprecursor formation (in this case H2O2) SolvingEqs A6a and A6b the column density of OH andthe concentration of precursors versus fluence cpcan be obtained Using the column of H2O2 pro-duced as the precursor column Np in Eq 2 thenO2 yield can be obtained The results can be com-pared with the results from Eqs1 and 2 in thetext The average precursor (H2O2) concentrationover the column depth of water molecules N is

cp 5 fp (sds)[1 2 exp(2s F)] 2 (ssr)

[1 2 exp(2sr F)] (A7)

with

fp 5 2s2OH(sr 2 s) (A5b)

In these equations sr 5 (2s2OH 1 sHOH) andssr 1

It is seen that the precursor concentration withtime cp depends on two rates the reaction ratesthat determine sr and the destruction rates in sAs in Eq 4a the yield can be written as

Y 5 sO2 cp N (A8)

If sr and s are comparable then the yield at small

OXIDANTS AT EUROPA 845

fluence is quadratic in F However if sr sthen the linear dependence on fluence is obtainedwith a small shift in the fluence as shown in Eq4c in the text Further the steady-state yield isnow

Y` (T) 5 sO2 [fp sd] Ns (A9)

That is sp R [fp sd] as in Eq A4 and the tem-perature dependence for the incident electrondata must be contained in [sO2 fp] or in the ratio[sO2 s2OH]sHOH]

For comparison with their precursor modelSieger et al (1998) examined the possibility of ob-taining the O2 precursor by considering reactionsfor a density of diffusing dissociation products asin Eqs A1 and A2 rather than a trapped productinteracting with a freshly formed dissociationproduct Based on the above the lowest fluence atwhich O2 is first detected experimentally sets alimit on the time scale for mobilendashmobile reactions

Effect of trap density

Equations A1b A1b9 A2b and A3b can be ap-proximately solved assuming that the dnOHdtdnHdt and dntdt are very small on the averageas discussed above In order to examine the roleof the trap density nt which can depend on thetemperature and method of formation and on theradiation history of the ice we use the slightlysimpler precursor model In this model the O pro-duced directly by dissociation traps forming aprecursor O2 is then formed when a freshly pro-duced O or other species reacts with a trappedoxygen atom Ot

H2O 1 radiation R H2 1 O (sd)

O 1 trap R Ot (kOt)

O 1 Ot R O2 (kOOt)

The reaction pathways for the oxygen species are

H2O 1 radiation mdashRsd9

trap OO2

kOtOt kOOt

H2 1 OOHt

kOR H(kHO t)

ROR

We assume for simplicity both H2 and O2 escapefrom depth Dx after formation We also allow mo-bile H to react with Ot (kHOt) removing trappedO and both the mobile H and O to be removedby other reactants R (eg OH or contaminants)This leads to a simplified set of equations not di-rectly involving OH or peroxide

dnOdt 5 sd9 f n 2 kOt nO nt 2 kOR nO nR

2 kOOt nO nOt (A1c)

dnOtdt 5 kOt nt nO 2 kOOt nO nOt

2 kHOt nH nOt (A1c9)

dnHdt 5 sd f n 2 kHOt nH nOt

2 kHR nH nR (A2c)

dnO2dt 5 kOOt nO nOt (A3c)

These equations of course act in parallel with theequations above for OH and mobile OH may re-act with trapped O to form O2 Here we focusonly on mobile O

Since the diffusion of H and O are fast and thedensity of traps is large the nH and nt averagedover the bombardment rate can be taken veryroughly as time independent Further if we as-sume the number of trapped O nOt is small com-pared with the number of sites at which H canreact then nH [sd f nkHR nR] For a low ra-diation flux f nH is small on the average as as-sumed Since the O2 yield YO2 is equal to[(dnO2dt) Dxf] with Dx the escape depth thenthe steady-state yield Y` can be obtained ana-lytically

[Y`][sd9 N] 5 1 1 (q 2 1) 2 [(q 2 1)2

1 8 q d]054 d (A10)

q 5 [kOt (nt 1 nR)] (kHOt nH)(sd9 f n) kOOt

5(sdsd9) [kOt (nt 1 nR)] kHOt[kOOt kHR nR]

d 5 05 [1 1 nt(nt 1 nR)]

We note that for nt large q can be large Assum-ing that q 1 and nt nR then Eq A10 be-comes

Y` R [sd9 N]q 5 (sd9 N) (sd9sd)

[kOOt kHR nR][kHOt kOt nt] (A5c)

JOHNSON ET AL846

Comparing with Y` in the simple precursormodel in Eq 4b then s R [(kHOt nH)f] R [sd nkHOt][kHR nR] In addition the product [sO2sp] R sd9 sd9 [kOOt n][kOt nt] That is the dis-sociation cross sections are modified by reactionrates

Quite remarkably it is seen in Eq A5c that thesteady-state yield Y` can depend inversely onthe density of traps nt in the temperature andflux regimes for which the approximations arerelevant That is if there are competing destruc-tion process for trapped O then at high trap den-sity a mobile O must find a trapped O before itbecomes trapped or before mobile H destroys theOt In this way the increasing O2 yield with in-creasing T can be related to the change in the trap-ping density nt with the formation temperatureof the ice sample As the temperature increasesand the density of traps in the bulk decreases thesurface and grain interfaces become the only trap-ping sites

ACKNOWLEDGMENTS

REJ acknowledges support by a Gledden Fel-lowship in Chemistry School of Biomedical andChemical Sciences at The University of WesternAustralia and by grants from the NSF Astron-omy Program and NASArsquos Planetary Geologyand Geophysics Program PDC acknowledgessupport by an Australian Postgraduate Award

ABBREVIATIONS

HST Hubble Space Telescope MJQ MatichndashJohnsonndashQuickenden NIMS Near Infrared Map-ping Spectrometer UV ultraviolet

REFERENCES

Ayotte P Smith RS Stevenson KP Dohnarsquolek ZKimmel GA and Kay BD (2001) Effect of porosityon the adsorption desorption trapping and release ofvolatile gases by amorphous solid water J Geophys Res106(E12) 33387ndash33392

Bahr DA Famaacute M and Baragiola RA (2001) Radiol-ysis of water ice in the outer solar system sputteringand trapping of radiation products J Geophys Res 10633285ndash33290

Baragiola RA (2003) Microporous amorphous water icefilms and astronomical implications In Water in Con-

fining Geometries edited by JP Devlin and B Buch El-sevier New York in press

Baragiola RA and Bahr DA (1998) Laboratory studiesof the optical properties and stability of oxygen onGanymede J Geophys Res 103 25865ndash25872

Baragiola RA Atteberry CL and Bahr DA (1999) Ori-gin of solid oxygen on Ganymede Reply to commentby RE Johnson on ldquoLaboratory studies of the opticalproperties and stability of oxygen on Ganymederdquo JGeophys Res E104 14183ndash14187

Baragiola RA Vidal RA Svendsen W Schou J ShiM Bahr DA and Atteberrry CL (2003) Sputteringof water ice Nucl Instrum Methods Phys Res B 209294ndash303

Bar-Nun A Herman G Laufer D and Rappaport ML(1985) Trapping and release of gases by water ice andimplications for icy bodies Icarus 63 317ndash332

Baverstock KF and Burns WG (1976) Primary pro-duction of oxygen from irradiated water as an expla-nation for decreased radiobiological oxygen enhance-ment at high LET Nature 260 316ndash318

Benit J and Brown WL (1990) Electronic sputtering ofoxygen and water molecules from thin films of waterice bombarded by MeV Ne1 ions Nucl Instrum Meth-ods Phys Rev B 46 448ndash451

Benit J Bibring J-P della-Negra S LeBeyec YMendenhall M Rocard F and Standing K (1987)Erosion of ices by ion irradiation Nucl Instrum Meth-ods Phys Rev B 1920 838ndash842

Brown WL Augustyniak WM Lanzerotti LJ John-son RE and Evatt R (1980) Linear and nonlinearprocesses in the erosion of H2O ice by fast light ionsPhys Rev Lett 43 1632ndash1635

Brown WL Augustyniak W M Simmons E Mar-cantonio KJ Lanzerotti LJ Johnson RE BoringJW Reimann CT Foti G and Pirronello V (1982)Erosion and molecular formation in condensed gasfilms by electronic energy loss of fast ions Nucl In-strum Methods B 1 307ndash314

Brown WL Augustyniak WM Marcantonio KJ Sim-mons EN Boring JW Johnson RE and ReimannCT (1984) Electronic sputtering of low-temperaturemolecular solids Nucl Instrum Methods B 1 307

Buratti B (1995) Photometry and surface structure of theicy Galilean satellites J Geophys Res 100(E9) 19061ndash19066

Buratti B and Veverka J (1983) Voyager photometry ofEuropa Icarus 55 93ndash110

Calvin WM and Spencer JR (1997) Latitudinal distri-bution of O2 on Ganymede observations with the Hub-ble Space Telescope Icarus 130 505ndash516

Calvin WM Clark RN Brown RH and Spencer JA(1995) Spectra of the icy Galilean satellites from 02 to5 mm a compilation new observations and a recentsummary J Geophys Res 100 19041ndash19048

Calvin WM Johnson RE and Spencer JA (1996) O2

on Ganymede spectral characteristics and plasma for-mation mechanisms Geophys Res Lett 23 673ndash676

Calvin WM Aninich VG and Brown RH (2004) Oxy-gen temperature and phase effects Icarus (in press)

OXIDANTS AT EUROPA 847

ldquosaturatedrdquo yields change slowly with fluenceMore importantly O from water molecules at aconsiderable depth into the sample contributedto the O2 ejected from the surface Since there wasno observed diffusion of the O2 in ice at the am-bient temperature (7K) over the time for their experiment they concluded that the oxygentransport and production are stimulated by theincident ions The transport was suggested as be-ing due to percolationdiffusion along the dam-aged and transiently heated tracks (Reimann etal 1984 Benit and Brown 1990) They also foundthat in fully eroding (sputtering) thin ice filmsthe total amount of O2 produced required about10ndash15 of the D2O molecules in the films at T 57K On warming irradiated samples some O2

formed at depth is found to remain trapped forexperimentally relevant times (see eg Hart andPlatzman 1961 Benit and Brown 1990 Baragiolaand Bahr 1998)

Matich et al (1993) detected luminescence in77K H2O and D2O ice from newly formed and ex-cited O2 They suggested that the O2 is formed

from the interaction of a freshly produced mo-bile O interacting with a previously producedand trapped O or H2O2 They favored the formerIn matrix isolation studies of peroxide irradiatedby UV photons luminescence from newly formedexcited O2 was seen However the oxygen wasformed from mobile O (greater than 30K in Krlattice) interacting with O trapped as H2O-O(Khriachtchev et al 2000) These studies also sug-gest that trapped O possibly in the form H2O-Omay be an important precursor

In Fig 4a the O2 yield for low-energy electronsplotted as a function of electron energy exhibits anapparent ldquothresholdrdquo at 10 eV (Orlando andSieger 2003) Although the uncertainties are largethe authors suggested there may be a lower thresh-old 6 eV corresponding to a very small crosssection for O2 production (10220 cm2) This couldbe associated with an excitation at the ice surfaceBulk ice exhibits very weak absorption at photonenergies below 6 eV but exhibits a strong ab-sorption that depending on the ice structure be-gins at about 73 eV with a peak at 86 eV This

OXIDANTS AT EUROPA 829

b

a

FIG 3 a Relative yields versus fluence (ion flux 3 time) for 15 MeV Ne1 incident on D2O ice at 7K (adapted fromReimann et al 1984) Top panel D2O yield Middle panel D2 yield Bottom panel O2 yield Yields at other tem-peratures exhibited similar dependences with the slope of O2 yield versus fluence varying with T Fits give activa-tion energies between 002 eV at the lowest fluences and lowest T to 007 eV at the highest T and highest fluencesb Oxygen yield versus fluence of low-energy electrons on D2O ice (from Sieger et al 1998)

is followed by a peak in absorption at 10 eVThe results in Fig 4a show that there is a dra-matic increase in the production of O2 at 10 eVwhich is the onset of secondary electron produc-tion in the bulk (Orlando and Sieger 2003)

For incident fast ions the yield at a given tem-perature was found after a small fluence to beindependent of whether the sample was crys-talline or amorphous ice (Brown et al 1982)However using low-energy electrons the yieldswere found to be smaller for an initially crys-talline ice (Orlando and Sieger 2003) Since verylow doses of ion radiation (a few eV per mole-cule) can convert crystalline ice to amorphous iceat low temperatures (Leto and Baratta 2003) thedamage produced by the penetrating ions usedby Brown et al (1982) affected the O2 yield Thestructural damage produced may create trappingsites cause changes in the electronic relaxationpathways or enhance percolation

The temperature dependence of the O2 yield isshown in Fig 5 for electrons incident on initiallyamorphous (Fig 5a) and initially crystalline (Fig5b) ice A small decrease in efficiency is seen inboth cases at 120ndash130K where Ic forms fromamorphous ice and pore collapse occurs (Row-land et al 1991) This is followed by an increas-ing yield with increasing T and then by a more

dramatic drop near 160K where Ih forms Dur-ing the transition to Ih trapped molecules havebeen shown to escape efficiently from ice samples (Ayotte et al 2001) A large drop in the yield isalso seen in Fig 2b for incident Ar1 ions

In modeling the dramatic decreases in the yieldin Figs 2b and 5 care must be taken That is thetime for crystallization of an amorphous sample(Baragiola 2003) even allowing for the large un-certainties is 30 min at 145K and becomes 1min at 160K Therefore pulsed radiolysis exper-iments should be carried out at these tempera-tures and the transients observed Since the timefor conversion from amorphous to crystalline icecan be shorter than the experimental times at145K the results are not comparable to thosefor which the chemical state of the ice at a givenfluence is stable over experimental times Belowwe focus on the data below 140K

We have emphasized processes initiated by theelectronic excitations and ionizations producedHowever momentum transfer collisions whichare the dominant energy loss process for low-en-ergy ions also produce dissociations defects andO2 in ice (Bar-Nun et al 1985) The relative effi-ciency of momentum transfer and electronicprocesses for producing O2 needs further studyAs the penetration depths are much smaller when

JOHNSON ET AL830

FIG 4 a Relative yield of O2 from D2O ice versus incident electron energy E at T 5 120K (from Orlando et al1998 Orlando and Sieger 2003) b Relative yield of D2 from D2O ice versus incident energy E at T 5 88K (fromKimmel et al 1994)

a

b

the former dominates the produced O2 likely es-capes readily

H2 formation and loss correlation with O2

The production of O2 from a low temperatureice by radiolysis correlates with the formationand loss of H2 (or D2 from D2O ice) (Reimann etal 1984) D2O is used to reduce the backgroundin mass spectrometry studies The yields of O2from D2O and H2O ices were shown to be com-parable However the total yields from D2O wereabout 10 less (Reimann et al 1984) consistentwith H (D) transport being important It is seenin Fig 3 that D2 is promptly produced and lostfrom the D2O ice At higher fluence the D2 yieldis seen to exhibit a dependence like that for O2For the 15 MeV Ne1 ions which fully penetratethe ice sample the D2 yield eventually decreasessince D2 is produced and escapes from the fullthickness of the film

Because of H2 (D2) loss the surface region of

the irradiated ice becomes depleted in H (D) rel-ative to O favoring the formation of oxygen-richspecies The change seen in the ratio of H (D) toO in the irradiated ice is small An irradiated film(100 nm) remains stoichiometric to better than2 (Brown et al 1980) and in the near surface re-gion Auger spectroscopy suggests only a slightoxygen enhancement due to irradiation by keVelectrons (Rye et al 1978) Sieger et al (1998) sug-gested a precursor concentration of the order ofa percent Such concentrations are an order ofmagnitude smaller than the steady-state concen-tration of dissociated H2O (Table 2)

Using infrared spectroscopy the damaged H2Oconcentrations at low temperatures in the satu-ration regime are about 10 over the irradiatedvolume (Benit et al 1987 Watanabe et al 2000Gomis et al 2003) This fraction is the ratio of thedissociation cross section to the effective crosssection for recombination to H2O as discussedlater At a 10 concentration most H2O will havea neighbor that is dissociated

OXIDANTS AT EUROPA 831

FIG 5 a Relative steady-state yields of O2 from a D2O ice sample versus T (upper panel) The ice samples wereformed and irradiated at these temperatures Film erosion as a yield (lower panel) (from Orlando and Sieger 2003)b Relative steady-state yields of O2 versus T as in (a) (upper panel) Here the D2O ice samples are formed at 160Kpresumably becoming crystalline and then irradiated at the T shown A new sample is used for each data point in(a) and (b) Film erosion as a yield (lower panel) (from Orlando and Sieger 2003)

At temperatures above 40K it is seen in Fig2 that the D2 and O2 yields exhibit the same tem-perature dependence suggesting a correlationbetween D2 and O2 loss H2 (D2) can be formedby reactions between diffusing hydrogen fromdissociated water or directly [ie H2O R H2 1O (eg Kimmel and Orlando 1995) H2O R (H1 OH)cage R H2 1 O (eg Matich et al 1993)H3O1 1 e R H2 1 OH (eg Williams et al1996)] Subtraction of the nearly constant D2 yieldin Fig 2a gives the Arrhenius plot in the insetSince molecular hydrogen readily escapes atthese temperature it is the H2 (D2) formationprocess that has an activation energy of 003eVmolecule The fact that slopes of the yields forH2 and O2 above 40K are similar suggests thatH2 (D2) formation and loss is the rate-limitingstep in determining the production and loss of O2at these temperatures

Although the data at low T in Fig 2a have con-siderable uncertainties below about 20ndash40K theD2 production is roughly independent of tem-perature but the O2 yield continues to decreasewith decreasing temperature Radiolytically pro-duced D2 is mobile and can escape above15ndash20K (Watanabe et al 2000) but O2 is im-mobile at such temperatures and the incidentHe1 ions are much less efficient at causing per-colation than the incident Ne1 discussed earlier

The production of H2 (D2) from ice has alsobeen studied using low-energy electrons and UVphotons at very low temperatures In the formerstudies absolute yields were not obtained butthere appeared to be a threshold at 63 eV (Fig4b) This is roughly the threshold for excitationof a surface state (Khusnatdinov and Petrenko1992) and the threshold for peroxide productionis close to this value The efficiency for H2 (D2)production increases around 11 eV with a sec-ond steep rise at 17 eV (Kimmel et al 1994) TheO2 and D2 yields versus electron energy in Fig 4exhibit some similarity in the threshold regionalthough the uncertainties are large and the sam-ple temperatures different Measuring the H2 (D2)and the O2 ejecta simultaneously over a range oftemperatures as in the Ne1 experiments wouldbe instructive

Using 98 eV photons at low T (12K) Watanabeet al (2000) measured the D2 ejected and trappedin ice They obtained a cross section for produc-tion of D2 (Table 2) at low fluences that is10ndash20 of the total absorption cross sectionwith significant uncertainties They also found

that the equilibrium damage fraction in the sam-ple was 10ndash20 of the D2O molecules but theequilibrium production of D2 was about an orderof magnitude smaller 1ndash2 at their tempera-tures They suggested that the D2 is formed atgrain interfaces and cracks and on the surfaces ofpores In a mixed ice D2O 1 CO exposed to UVphotolysis D2 and CO2 formation are correlated(Watanabe and Kouchi 2002) The D2O dissoci-ates to produce D2 plus O and this transientlymobile O reacts with CO to produce CO2

Precursor state and peroxide

In order to produce O2 from low temperatureice it appears that the ice sample must lose hy-drogen and form trapped oxygen-rich reactivespecies The efficiency of production also appearsto be affected by the level of defects and damagein the ice consistent with the suggestion that thehydrogen-bonding network is important in theproduction of O2 via availability of trapping sitesnew electronic relaxation pathways or the natureof the bonding of precursors Here we briefly ex-amine the possible role of the suggested precur-sor molecules H2O2 and HO2 We note that H2O2might not readily dissolve in ice possibly ac-counting for its low abundance in some experi-ments It appears to be heterogeneously distrib-uted in the form of microcrystals (Gurman et al1967) If H2O2 does act as a precursor the for-mation of aggregates could affect the dimer ab-sorption bands (Cooper et al 2003a) and affectthe production and stability of O2 (Cooper et al2003b)

The results of Gerakines et al (1996) in whichice at 10K is irradiated by UV photons confirmthat H2O2 forms in ice from two OH after a build-up of trapped OH They monitored the band as-sociated with H2O-HO (Langford et al 2000) inwhich the HO is bound to water with 024 eVTherefore they might only be monitoring a sub-set of the trapped OH seen in earlier experiments(eg Taub and Eiben 1968) The H2O2 concen-tration in ice reaches a steady-state level at flu-ences of about 1018 photonscm2 (Gerakines et al1996) As seen in the model section this gives anestimate of the H2O2 photodestruction cross sec-tion of the order of 10218 cm2 Similarly HO2 be-gins to be seen in absorption at about a few times1017cm2 shortly after they begin to detect H2O2The absolute yields were not measured

Westley et al (1995) detected H2 and O2 pro-

JOHNSON ET AL832

duced by Lyman-a (98 eV) radiation of ice butfound no fluence dependence for fluences greaterthan 1017cm2 They also saw small amounts ofH2O2 by programmed thermal desorption of thesample but made no quantitative evaluationComparing their lowest fluence for the produc-tion of O2 with the result of Gerakines et al (1996)for peroxide suggests that H2O2 and HO2 areprobably not important precursors of O2 In UVphotolysis of H2O2 using 64 eV photons the pro-duction of O2 was not observed (Vaghjiani et al1992) and the excitation of an irradiated icewhich presumably contained some peroxide by37 eV photons did not produce O2 (Baragiola etal 2003) Direct excitation of matrix-isolatedH2O2 at low temperature also did not yield O2but the production of a mobile O that reacted withtrapped O (H2O-O) did give O2 (Khriachtchev etal 2000)

Irradiating with 200 keV protons at tempera-tures of 40ndash129K Bahr et al (2001) found onlyabout 0001ndash002 H2O2 using programmed ther-mal desorption as compared with earlier experi-ments that suggested larger values (eg Table 2)Using infrared absorption Moore and Hudson(2000) found a G value of 01 for incident MeVprotons at low temperature but the signal in theobserved band became negligible at 80K sug-gesting that H2O2 was not being formed Theyalso found that adding O2 significantly increasedthe production efficiency of H2O2 That is thatthe amount of peroxide was affected by theamount of trapped oxygen Based on the matrixisolation studies (Khriachtchev et al 2000) dis-sociated mobile O might trap as H2O-O whichon excitation converts to H2O2 Therefore even ifH2O2 is not the precursor the peroxide andtrapped oxygen concentrations will be related

Although molecular oxygen was not a productof photolysis of matrix isolated H2O2 it was aproduct of photolysis at 266 nm (466 eV) of per-oxide dimers in an argon matrix 2 H2O2 R 2 H2O1 O2 (Engdahl and Nelander 2002) This de-composition was thought to be due to the con-siderable distortion of the bonds in the dimerTherefore the configuration or distortion by thelattice of peroxide or any other precursor mole-cule in the ice may be critical (Sieger et al 1998)In addition more recent experiments show thatG values for the production of H2O2 by 30 keVincident ions are significant as are the steady-state ratios of H2O2H2O (Gomis et al 2004) Ra-tios of the order of 02ndash4 were found at 16K and

77K in the irradiated regions with incident H giv-ing the smallest fraction and incident O thelargest (Table 2) At these concentrations perox-ide molecules might segregate forming inclu-sions (Gurman et al 1967 Cooper et al 2003b)Since the amount of trapped peroxide also com-petes with the amount of trapped O studies areneeded to describe the chemical state of the solidproduced by radiolysis This will be critical in de-scribing the fate of the irradiation products at Eu-ropa In particular the chemical composition fa-vorable for producing O2 needs to be determinedas does the importance of trapped O H2O2 HO2or peroxide dimers as precursors

Other species

In the model below we will refer to atomic andmolecular species trapped in ice Such species areoften discussed in the literature on radiolysis andphotolysis (for summaries see Matich et al 1993Johnson and Quickenden 1997) Trapped OH OH2O2 and HO2 have been discussed andor mea-sured by a variety of techniques (absorption andluminescence spectroscopy electron spin reso-nance programmed thermal desorption etc)Care must be taken when using these data as re-sults for ice samples formed below 160K orthose exposed to penetrating ion beams likelyhave numerous defects voids and grain inter-faces so that a variety of trapping sites and trapdepths exist Therefore when a particular bandfor a trapped species (eg OH) disappears withincreasing temperature it could be due to achange in the nature of the trapping as well as toloss of the species by reaction

Although H is discussed below we do not dis-tinguish it in these simple models from H1

trapped as the hydronium ion H3O1 whichwould mean that there are corresponding nega-tive ions present Therefore below we refer to thevarious species as if they are neutral althoughthis might not be the case Although trapped elec-trons are detected only over a narrow range oftemperatures (100ndash120K) negative ions must beimportant over a broad range of temperature

We also note that SO2 and a sulfate and CO2and a carbonate appear to be trapped species inEuroparsquos icy surface Under radiolysis each ofthese can dissociate directly giving O2 Howeverwith the exception of the sulfate their densitiesare less than 1 so they are not relevant sourcesof O2 However their presence is indicative of the

OXIDANTS AT EUROPA 833

preference for forming oxygen-rich molecules inEuroparsquos ice and can affect the efficiency of pro-duction of O2 and H2O2

MODELS FOR FORMATION OF O2

Introduction

Based on the data described above the yieldsfor incident electrons and ions are found to benearly linear in fluence at the lowest fluences forwhich O2 is detected In addition the relevant al-terations of the ice samples by radiolysis appearto be stable and independent of the beam fluxRemarkably this is the case both for low-energyelectrons that make only a single excitation foreach impact and fast ions that make a density ofexcitations along their path through the solidThis suggests that O2 is produced and trapped inthe initial events and then caused to escape fromthe solid by the subsequent radiation or that astable chemical alteration of the ice is required be-fore the radiation can produce and eject O2

Because H produced by the dissociation of wa-ter molecules can attack trapped O2 and any pre-cursor oxidants the loss of H is important Thisoccurs by the production and loss of hydrogenmolecules as discussed leaving a surface that isslightly rich in O relative to H The O2 produc-tion mechanism could be direct excitation of aparticular precursor species or it might not bespecific That is a new OH or O produced by theincident radiation could react with the H-depleted solid producing O2 This would occurmore efficiently as the ratio of O to H in the nearsurface region increased

In their early work WL Brown and colleaguesassumed that processes similar to those used todescribe radiolysis of liquid and gaseous H2Ocould be used to describe the radiolysis in ice Thelow activation energy that they found was as-sumed to be associated with diffusion of somemobile radical species in ice such as H or O orpossibly diffusion of the O2 molecule itselfBased on their studies of ejection from isotopi-cally labeled D2O samples they concluded thatthe O2 is formed in the track of a single ion andthen ejected by subsequent ions (eg Benit andBrown 1990) More recently trapped specieshave been suggested as precursors These are ei-ther subsequently excited by the incident radia-tion or attacked by newly produced mobile

species We consider a simple model that incor-porates all of these processes and can be used toorganize and interpret the experimental data Wealso discuss its relationship to more completemodels In this discussion much of which is in theAppendix we consider the standard radiationproducts that have been often described How-ever in a solid the excited electron cloud pro-duced by absorption can overlap a number of lat-tice sites Therefore their relaxation pathwaysandor product species could be very differentfrom those that have been traditionally consideredin the literature on radiolysis and photolysis

Precursor model

The column density of a precursor species Npformed in ice by incident radiation can be calcu-lated using a simple model In this model cross sec-tions are used for the production sp and destruc-tion s of the precursor O2 is then formed byexcitation of the precursor or by reaction of a newlyformed radical with the precursor This is also de-scribed using a cross section sO2 For incident elec-trons having a small penetration depth the O2 pro-duced is assumed to rapidly diffuse to the surfaceand escape (Sieger et al 1998) More generally wedefine N as the column density of H2O from whichthe incident radiation can produce and eject O2 Apair of rate equations gives both Np and the pro-duction rate of oxygen dNO2dt That is

H2O 1 radiation R precursor

precursor 1 radiation R O2

giving

dNpdt 5 sp f N 2 s f Np (1)

dNO2dt 5 sO2 f Np (2)

In Eq 1 f is the incident flux and the total pre-cursor destruction cross section can be written ass 5 sp9 1 sO2 with sp9 accounting for precursordestruction processes other than the formation ofO2 Solving Eqs 1 and 2 and assuming N is nearlyconstant (10 level as discussed) the fractionof precursors cp in the affected column after timet is obtained (Appendix)

cp 5 NpN 5 [sps] [1 2 exp(2 s F)] (3)

Here F 5 [ft] is the fluence (flux 3 time) which

JOHNSON ET AL834

is the number of particles (photons electrons orions) incident on the surface per unit area Usingcp in Eq 2 the O2 yield Y at any fluence can bewritten in terms of the yield at steady state Y`

Y 5 [dNO2dt]f 5 sO2 cp N 5 Y`

[1 2 exp(2sF)] (4a)

Y` 5 [sp sO2 s] N (4b)

If we also account for the reduction in the den-sity of water molecules in the initial column Nthese forms change only slightly The size of thetwo parameters Y` and s can be obtained by fit-ting the yield versus fluence data such as that inFig 3

For the low-energy electron data Sieger et al(1998) showed that the ratio [YY`] for a givenincident electron energy is independent of tem-perature over a narrow range of temperatureTherefore the destruction cross section s in Eq4a is roughly independent of T and can be esti-mated from the [YY`] curve Sieger et al (1998)found a precursor destruction cross section s lt08ndash30 3 10216 cm2 for incident electron energiesfrom 30 to 100 eV This is of the order of the elec-tron impact ionization cross section for H2O

Based on the above model the lack of an ob-served fluence dependence for fluences down to1017 hncm2 in the Lyman-a irradiation of ice(Westley et al 1995) can be examined The low-est fluence used implies that the precursor de-struction cross section for Lyman-a is of the or-der of the Lyman-a absorption cross section in ice(09 3 10217 cm2)

Sieger et al (1998) also assumed that sp9 5 0This would mean that the precursor is only de-stroyed by production of O2 (ie s 5 sO2) Thatis unlikely since suggested precursors can be de-stroyed by reaction with H and other radicalsTheir assumption would also mean that the crosssection for production of O2 from a precursorsO2 is large This would be surprising and ap-pear to contradict some of the data discussedabove Instead we assume that s is not equal tosO2 and the total destruction cross section s isof the order of the ionization or absorption crosssection

Having obtained s the measured temperaturedependence for production of O2 is containedonly in Y` Over a narrow range of T one canwrite Y`(T) lt Y`9 exp(2EakT) and obtain the es-

timate of the activation energies in Table 2 Ifsp9 THORN 0 as discussed then sp and sO22 in Eq 4bcannot be obtained separately from the data forY` unless an independent measure of cp is ob-tained Such a measurement is critical for de-scribing the chemical state of the irradiated sur-faces

The observed temperature dependence of Y` iscontained in sp sO2 or N Here N is the ldquodepthrdquofrom which newly formed O2 can reach the sur-face and escape given as a column density of H2OThe physical depth is obtained by dividing thecolumn density N by the molecular number den-sity of the ice Assuming that in the low-energyelectron experiments N is independent of tem-perature and is about three monolayers (Sieger etal 1998) the measured Y` gives [sO2sps]05ndash20 3 10218 cm2 over the electron impactenergy range At 110ndash120K this ratio was foundto be proportional to (E 2 Eo) for E Eo 10 eVagain assuming fixed N (Orlando et al 1998 Or-lando and Sieger 2003)

In the following section we first consider a pre-cursor like peroxide that typically forms fromtwo dissociation events We use this to show un-der what circumstances such a model does notviolate the observed linearity of the O2 yield datawith fluence Following that we relate the sim-ple precursor model above to the case wheretrapped O is the precursor We then use themodel for trapped O to examine the effect of thetrap density on the production of O2 The densityof traps (defects) is determined by the sample for-mation temperature and the subsequent radiationdamage Finally we consider the escape depthN for penetrating radiation within the precursormodel These discussions are supported by moredetailed rate equations in the Appendix

Precursor trapped peroxide

The proposed precursor peroxide is typicallyformed from products of two dissociation eventsie 2 H2O R 2 H 1 2OH R H2 1 H2O2 This wasconfirmed in low temperature ice (Gerakines etal 1996) It is also possible that a dissociation pro-ducing a trapped O followed by an excitationevent can produce peroxide 2 (H2O 1 hn) R H2

1 H2O-O 1 hn R H2O2 1 H2 (eg Khriachtchevet al 2000) The O2 yield versus fluence for inci-dent electrons is reasonably well fit by the yieldexpression above for fluences greater than1014cm2 Therefore it has been suggested that

OXIDANTS AT EUROPA 835

any precursor species must be formed in a singleevent Using a model in the Appendix in whichthe precursor is formed by two events we showthis is not necessarily the case Precursor forma-tion if it is fast could involve two dissociationevents and give the measured fluence depen-dence Although we consider peroxide here theseconclusions apply to other possible precursors

In the example in the Appendix H2O is disso-ciated to H 1 OH according to the cross sectionsd The reactions of two OH units to form H2O2(rate constant k2OH) and 2 H to form H2 (k2H)compete with the recombination of H and OH toH2O (kHOH) If k2H is much greater than k2OH then H2 loss dominates at low fluences as ex-pected (Appendix) Steady state requires not sur-prisingly that the rate for forming H2O2 isroughly equal to the rate for forming H2 Finallyit is seen in the Appendix that the low H2O2 con-centrations suggested by experiment imply thatkHOH (k2H k2OH)05 Such sizes are reasonable

The set of rate equations in the Appendix thatuse number densities of species is integrated overdepth in order to obtain the forms used in Eqs 1and 2 which are based on column densities Inthis way the effective cross sections are related tothe rate constants for the interaction of freshlyproduced mobile species with previously pro-duced trapped species We show in the Appen-dix that the O2 yield versus fluence can be writ-ten in a form similar to that in Eq 4a but with asmall offset due to the two-step precursor for-mation process

Y lt Y`(T) [1 2 exp(2sF) 2 (ssr)]

for F sr21 (4c)

Here s is again the precursor destruction crosssection but sr 5 (2s2OH 1 sHOH) is the effectivecross section for loss of OH by formation of H2O2

and H2O The corresponding expression for Y`(T)is given in the Appendix

The form for the yield versus fluence in Eq 4cis consistent with the low-energy electron mea-surements if (ssr) 1 and if sr F21 lt10214 cm2 Such sizes are indeed possible Whatis required is that at the relevant temperatures(50ndash110K) the interaction length associated withsr is more than a few molecular layers (greaterthan 10 Aring) A cross section that is likely to belarge above 20K is sHOH which describes themean distance for diffusion of mobile H before it

reacts Since s discussed earlier is of the sameorder of magnitude as the H2O dissociation crosssection sd then the precursor concentration cp(in this case H2O2) is roughly proportional to[s2OHsHOH] 1 (Appendix)

Based on the above a two-step precursor for-mation process could be consistent with the elec-tron data contrary to the statements in Sieger etal (1998) Orlando et al (1998) and Orlando andSieger (2003) Experiments are needed to deter-mine the relevant cross sections Below we con-sider a precursor formed in a single excitationevent

Precursor trapped O

O2 can be formed for instance by diffusion ofa transiently mobile O to a previously formed andtrapped O or to the production of O near the siteof a previously produced and trapped O (Matichet al 1993 Johnson and Quickenden 1997) Werefer to this as the MatichndashJohnsonndashQuickenden(MJQ) model and note that O 1 OH R O2 1 Hcan also give molecular oxygen

Trapped O as H2O-O was in fact seen to be aprecursor to O2 when attacked by a mobile O ina Kr matrix (Khriachtchev et al 2000) ThereforeO trapped as H2O-O was also suggested as a pre-cursor in ice (Johnson 2001) Before proceedingwe note that the examination of the MJQ modelin Sieger et al (1998) was incorrect They consid-ered two mobile interacting species whose life-times were long compared with the bombard-ment rate rather than a new mobile O interactingwith a previously trapped O Although this hasnot been corrected recently they added Otrapped as H2O-O as a possible precursor (Or-lando and Sieger 2003) O trapped as H2O-O isseen in matrix isolation studies as discussedabove but it has not yet been identified in iceand like OH it may trap in a number of ways

The MJQ model is essentially the followingFormation of H2 1 O is followed by loss of H2and trapping of O (Ot) A subsequent dissocia-tion or excitation event can lead to the destruc-tion of Ot by reaction with for instance H or byconversion to H2O2 However if this dissociationevent also leads to H2 1 O a mobile O is pro-duced that can react with Ot with an interactionlength that depends on T This results in the for-mation of O2 [O 1 Ot R O2] giving a net pro-duction of 2 H2 and 1 O2 In the Appendix therelevant rate equations (Eqs A1cndashA3c) are con-

JOHNSON ET AL836

sidered Assuming the mobile O processes arefast these rate equations can also be integratedand simplified to give the form in Eqs 1 and 2The precursor column density Np now repre-sents the column density of Ot In this model sO2in Eq 2 is the ldquocross sectionrdquo for reaction of afreshly produced O with an Ot and sp9 indicatesdestruction of Ot Finally sp is the cross sectionfor production of H2 and Ot via processes dis-cussed above

Based on the sizes of the cross sections ex-tracted in the electron irradiation experimentsthe model with trapped O (Ot) as the precursoris quantitatively reasonable That is for a rangeof T both sp and sO2 might be described byevents of the type H2O R H2 1 O (whether di-rect or indirect) If these occur for 10 of theradiation-induced dissociation events as dis-cussed then the required product of cross sec-tions suggested by the sizes of s and Y` above isroughly obtained that is [sp sO2] 02ndash6 3 10234

cm2 in the low-energy electron experiments Thecross section for production of O might dependon the availability of dangling bonds at interfacesin pores or at defects as suggested by a numberof authors The importance of such bonds couldalso be consistent with the differences in the elec-tron-induced yield for initially crystalline and ini-tially amorphous ice (Orlando and Sieger 2003)

O2 production can increase in the presence ofoxygen-rich impurities such as CO2 or SO2 bothof which are known to be present at Europa Asin the MJQ model dissociation (eg CO2 R CO1 O) can provide a mobile O that can be trappedor react with a previously trapped O forming O2In addition such species are hydrogen scav-engers (Moore and Hudson 2000) This can re-duce the destruction of Ot affecting sp as seen inEq A5c

Trap density

In the simplified models above the density oftrapping sites nt was not explicitly accountedfor However in the discussion of both the ex-perimental data and models trapping sites havebeen assumed to be important It has beendemonstrated repeatedly that the density of de-fects voids and interfaces decreases as the tem-perature of formation of the ice increases How-ever the efficiency of production of O2 is theopposite it increases with increasing T There-fore the oxygen yield varies in a manner that is

the inverse of the trap density over a range oftemperatures For penetrating ions in which theenergy density deposited is high and O2 might beformed in a single track a simple explanation hasbeen given As the density of traps for O2 is re-duced and the temperature increases the O2 pro-duced can diffuse to the surface and escape moreefficiently limiting back reactions on further ir-radiation This is examined in the next sectionHowever such a process does not affect the low-energy electron yields for which the penetrationdepth is small and the formed O2 escapes read-ily at the measurement temperatures In Eqs A1cA1c9 A3c and A10 in the Appendix we againuse a simple set of rate equations for trapped andmobile O in order to determine the role of the trapdensity nt on the production of a precursorspecies The conclusions apply to any modelbased on a trapped precursor

In the MQJ model mobile O can be trappedforming Ot (described by kOt nt) and can reactwith previously trapped O to form O2 (kOOt) Inaddition it can react at other sites and Ot can bedestroyed by mobile reactive species such as H(kHOt nH) Mobile H can also react with othertrapped species (kHR nR) Incorporating these re-actions Eq A10 in the Appendix gives a steady-state yield Y` When the density of traps is largethe steady-state yield is given in Eq A5c

Y` R (sd9sd9sd) [kOOtkHR nr][kHOt kOt nt] N (5)

Here sd and sd9 are the dissociation cross sectionsthat determine the production of H (H 1 OH) andO (H2 1 O) by the incident radiation Comparingthe result in Eq 5 with Y` in the simple precur-sor model in Eq 4b the dissociation cross sec-tions in Eq 5 are now scaled by a ratio of the re-action rates

It is seen in Eq 5 that the steady-state yield Y`depends inversely on the density of traps nt Thatis there are competing destruction processes fortrapped O (Ot) Therefore in order to form O2 amobile O competes with mobile H to find atrapped O before it becomes trapped In this waythe increasing O2 yield correlates with the re-duction in the trapping sites with increasing tem-perature of formation The small activation ener-gies discussed earlier would be related to there-orientation of the lattice binding leading to thereduction in trap density At relatively high T thedensity of traps becomes small so that only the

OXIDANTS AT EUROPA 837

sample surface or grain interfaces might act astraps accounting for the observed drop in theyield at high T For this reason it is important todetermine if the yield for T greater than 150Kdepends on the incident flux f Since other quan-tities in the expression for Y` could also be tem-perature dependent experiments are neededThese would likely be pulsed radiolysis experi-ments in which the transients are studied ratherthan the steady-state yields However the avail-able data for lightly ionizing radiation do suggestthat the O2 yield depends inversely on trap den-sity as in the model above

Escape from depth

The quantity N in Eq 1 is the ldquodepthrdquo given asa molecular column density from which O2 formedbelow the surface can percolate to the surface andescape It can be converted to a physical depth bydividing by the molecular number density of theice In the case of the low-energy electrons thedepth below the surface at which ionizations andexcitations are produced is small However forfast ions that penetrate many monolayers this isnot the case Below we examine N in Eq 1

Although it has long been understood that twoexcitations events are required to produce O2(eg Brown et al 1982 Reimann et al 1984) theprecursor model suggests that at saturation(steady state) any new event can produce O2 fromthe steady-state density of trapped precursorsTherefore one might expect the yield at satura-tion to be nearly linear in the excitation cross sec-tion This is the case for the low-energy electronsand it was initially assumed to be the case for thefast ions (Brown et al 1980 Reimann et al 1984)However Baragiola et al (2003) showed that thetotal sputtering yield in the saturation region attemperatures greater than 100K which is dom-inated by decomposition (Brown et al 1982) isproportional to the square of the electronic en-ergy deposition per unit path length in the solid(dEdx)e Assuming that the production and de-struction cross sections in Y` in Eq 4b roughlyscale with (dEdx)e then their ratio [sp sO2s]also roughly varies as (dEdx)e If this is the casethen the observed quadratic dependence impliesthat the depth N from which O2 can be mobi-lized to escape must also depend on the excita-tion density (dEdx)e This is consistent with theobservation that the O2 yield depends on thick-ness for samples thinner than the ion penetration

depth (Reimann et al 1984 Benit and Brown1990) Therefore the percolation depth for escapeN is a critical factor in comparing yields for pen-etrating radiation with those for non-penetratingradiation like the low-energy electron and low-energy ion data It is also seen that percolationand trapping can be included within the simplemodel considered here

Model summary

We have examined the simple precursor modelintroduced by Sieger et al (1998) to explain theirelectron data We have shown how it can be gen-eralized and derived from an integration of theappropriate chemical rate equations with thecross sections and total column density contain-ing the reaction rates and the physics of diffusivetransport Therefore the parameters can havevery different meanings from those initially as-sumed The equations are useful when there arefast processes which go to completion in timesmuch shorter then the time between particle im-pacts and slow processes associated with speciesthat are assumed to be trapped between particleimpacts This separation allowed us to examinethose models proposed for precursor formationand destruction due to reacting mobile speciesformed from dissociation events Further fromthe size of the onset of the linearity of the yieldversus fluence we can place constraints on therole of nonlinear processes such as those thatmight be associated with the formation of a pre-cursor like peroxide

With these extensions this simple model is use-ful for approximating the full rate equations inthe temperature region for which the yields de-pend on the fluence (dose) and not on the doserate Based on the assumptions in the model thisimplies temperatures at which precursor speciesare stable over experimental times At the high-est temperatures for the yields shown in Fig 4athis might not be the case This should be testedexperimentally In this regard it is puzzling thatexperiments have shown that the absorptionbands associated with certain trapped speciessuch as OH and trapped electrons ldquodisappearrdquoat 100ndash120K (see eg Johnson and Quickenden1997) whereas the O2 yield data indicate that rel-evant precursor species must be stable to 140KTherefore the proposed O2 production processescan only be distinguished by directly determin-ing the chemical state of the irradiated ice

JOHNSON ET AL838

It may be the case that a particular precursoris not required as mentioned earlier Rather af-ter a sufficient level of oxidation a newly pro-duced O or OH can react in the oxygen-rich lat-tice producing O2 in a pseudo-first-order processThis can be tested experimentally Indeedtrapped O and peroxide may be changed into oneanother as in the matrix isolation experiments(Khriachtchev et al 2000) Therefore their rela-tive concentrations versus temperature and doseshould be studied

In the model considered here it is seen that theparameters in Y` must account for the strong tem-perature dependence Since the number of bulktrapping sites decreases with increasing T as theice sample is annealed one might expect that theprecursor concentration would decrease We sug-gest that the observed temperature dependenceis determined by the competition between theavailability of trapping sites and mean interactiondistances of a mobile species such as O and HThat is the O2 yield varies inversely with the den-sity of traps over a range of temperatures In thesimple model above this effect is included indi-rectly as shown using the more detailed equa-tions in the Appendix

EUROPA

Oxygen and peroxide production at Europa G values

The dependence of the yield on the thicknessof the ice sample for highly penetrating ions sug-gests that the O2 yield is roughly related to theamount of energy deposited in the sample There-fore the amount of O2 produced and lost fromice in steady state is often given as a G value thenumber of molecules produced per 100 eV de-posited in Table 2 In applying these G values tothe planetary environment care must be takensince the yields are temperature fluence and ex-citation density dependent They also are affectedby the presence of other proton scavengersknown to be present in Europarsquos ice

The G values for peroxide in Table 2 are thosefor production at low fluence but the G valuesfor O2 are for the production and escape of gas-phase oxygen at saturation Therefore they de-pend on the formation and destruction cross sec-tions in the model above Such G values can beused to describe the steady-state source of the am-

bient O2 atmosphere at Europa but are not help-ful in determining the ability to produce and trapO2 at depth into the icy surface Over the timescales of the experiments some O2 is seen to re-main trapped and is not destroyed by radicalsTypically there are a number of trapping sitesand recent estimates of the activation energy foran O2 molecule trapped in a substitutional site(Hori and Hondoh 2002) suggest that it can betrapped for times comparable to the time re-quired for the conversion of amorphous ice intoIh (Ayotte et al 2001) The stability of O2 inclu-sions has not been studied

To understand the relationship between the Gvalues for O2 in Table 2 we note that fast lightions and electrons lose their energy predomi-nantly by producing ionizations and excitationsThe ionization events produce secondary elec-trons that can result in further ionizations Thenumber of ionization events Ni produced by afast light ion stopping in a water ice is Ni 5EoW where Eo is the initial energy and W is theaverage energy expended per ionization event27 eV in water ice Therefore radiolysis by fastions is often described as the net effect of theshower of electrons set in motion by the incidention In this manner the G values for productionof O2 from ice by energetic light ions and elec-trons can be roughly estimated as the sum of theeffects of all the secondary electrons using theyield Ye for 10ndash100 eV electrons That is if weignore the percolation depth N we can write thenumber of O2 produced as the number of ioniza-tion events times Ye which already has in it thedestruction processes That is G NiYeEo YeW 0006 O2100 eV

This estimate of G is seen to be an order of mag-nitude larger than the keV proton value for O2 inTable 2 but is much smaller than the G value forenergetic Ne1 These differences from the simpleestimate are consistent with the observation thatthe yield in the saturation regime for penetratingions varies nonlinearly with excitation densityThe difference from the keV proton G value islikely due to the inability of O2 to escape fromdepth before destruction On the other hand thelarger G value seen using MeV Ne1 is likely dueto the high density of excitation This can resultin more efficient production of O2 increased pro-duction of damage sites for forming O2 andorincreased transport (percolation) along the tran-siently heated track That is the ability of theformed O2 to percolate to the surface during ion

OXIDANTS AT EUROPA 839

bombardment plays an essential role in deter-mining the measured yields as discussed aboveThis effect needs to be evaluated for a planetarysurface subject to particle fluxes that are many or-ders of magnitude smaller than those used in thelaboratory

In the absence of O2 yields for all of the rele-vant ions and energies an estimate of the totalsupply of O2 can be obtained using the G valuesin Table 2 For the energetic oxygen and sulfurions we use G values for neon and argon ionsThese are lower bounds as the data were obtainedat temperatures much lower than the surface tem-peratures at Europa (80ndash110K) For the fast pen-etrating electrons and protons we use the valuefor 200 keV protons in Table 2 For the heavy ionswe used the Ne1 and Ar1 values For compari-son we also give an estimate using the G valuesobtained from the low energy electron data andthe total ionization as discussed above

The G values are combined in Table 3 with theelectron and ion energy flux at Europa Althoughthere is some deflection of the flow due to in-duced fields the net energy flux to the surface isnot reduced much from these values (Paranicaset al 2002) In fact locally produced pick-up ionswhich are formed from the ejecta and acceleratedto the surface can contribute (Ip 1996) Furtheralthough G values are typically independent ofthe incident angle the loss of O2 is a surfaceprocess and is enhanced with angle That is thenumber of O2 produced within the escape depthincreases with increasing angle to the normal(steeper than the inverse cosine) This enhance-ment is 25ndash4 (Jurac et al 2001) Ignoring suchcorrections the total production rate calculated

(23ndash62 3 109 O2cm2s) is a lower bound to theactual production rate from the icy regions of Eu-roparsquos surface Recent modeling of the O2 atmo-sphere suggest that such a source rate is required(Shematovich and Johnson 2001)

Earlier estimates of the O2 source rate were in-correct because O2 production was assumed to bea fixed fraction of the H2O sputtering yield Atthe ambient temperature an ejected H2O sticks ef-ficiently (Smith and Kay 1997) to the surface ofneighboring grains in the porous regolith reduc-ing the sputtering yield for H2O (Johnson 1989)This reduction does not apply to the O2 fractionof the yield enhancing its relative importance toH2O Here we assume that all of the gas-phaseO2 produced can contribute to the ambient at-mosphere even if it is produced at a grain sur-face that is at some depth into the porous regolith

Finally Europarsquos surface is not pure ice every-where That is much of the trailing hemisphereis a hydrated sulfuric acid containing smallamounts of sulfur and SO2 and the leading hemi-sphere contains low levels of carbon CO2 andcarbonates Sodium and potassium atoms aresputtered from the icy regions (eg Leblanc et al2002) O2 production from surfaces containingthese species has not been measured C and S areboth proton scavengers and the other species canbe sources of O as discussed so that the O2 yieldfrom such regions could be enhanced Howeverthere can also be competing processes that mightreduce these rates In fact recent HST observa-tions suggest that the O2 might not be produceduniformly across Europarsquos surface (McGrath et al2000) The analysis of these observations is pre-liminary and is not simple since the observed

JOHNSON ET AL840

TABLE 3 APPLICATION TO EUROPA

OxygenImplantation Energy flux production

Species 109(cm2 s)a 109 keV(cm2 s)(1) 109(cm2 s)

H (10 keV) 0015(1) 12 0084 (072)bO (10 keV) 0015(1) 18 054cS (10 keV) 0009(1) 30 090cElectrons 62 043 (37)bPlasma (10 keV) (137 amu) 13(2) 10 03c

Total 79 23 (62)UV photons (64 eV) 46 01

aReferences are as follows 1Cooper et al (2001) [similar results in Paranicas et al (2002)] 2Mauk et al (1996)bUsing G for 200 keV protons (upper bounds in brackets are based on the low-energy electrons data as a model for

the secondary electrons produced)cUsing G for 15 MeV Ne1 at 7K a rough lower bound

emissions depend on the local plasma propertiesHowever the efficiency of O2 production in con-taminated ices needs to be studied

Porous regolith O2 interactions on grain surfaces

Although the O2 atmosphere is principally pro-duced in the near surface layers and is eventuallylost to space it also permeates Europarsquos porousregolith The porosity p (void space) of the re-golith is uncertain (Buratti 1995) It has been es-timated to be the most compact surface in theOuter Solar System p 025 on the trailing hemi-sphere and p 079 on the leading hemisphere(Buratti and Veverka 1983) and the most poroussurface p 096 globally (Domingue et al 1991Domingue and Hapke 1992) These large differ-ences are due to the uncertainties in the amountof backscattered light at small phase angles andon the photon absorption properties and grainsizes The more compact surface is likely correctGrain sizes rg of the order of 100 mm have beensuggested (Hansen and McCord 1999 2000Geissler et al 1998) Europarsquos regolith differs sig-nificantly from the lunar regolith which is asand-like surface of individual grains Analysis ofthe thermal inertia of Europarsquos surface (Grundyet al 2001) suggests that the ice grains are sin-tered forming a fairy castle-like structure Sin-tering can occur thermally or due to the incidentradiation In addition ices doped with acids andalkalis in the laboratory can be powdery solidswith the impurities primarily found at grain in-terfaces

The porous regolith affects the above estimatesof the oxygen production in three ways First forisotropic bombardment the mean distance d intothe surface at which an incident ion or photonstrikes a grain surface is d (89)rg(1 2 p) Us-ing the above value for rg then d 02 m if theporosity is very large p 96 Therefore de-sorbed molecules can come from a considerabledepth and will interact with grain surfaces Evenfor much lower porosities the effective sputteringyields can be significantly modified (Johnson1989) As discussed above the yields for ejectedH2O which sticks onto neighboring grains at theambient temperatures are reduced but those forejected O2 are not This enhances the relative im-portance of the O2 yields

Finally the surfaces and interfaces of the sin-tered grains in Europarsquos porous surface are placeswhere low temperature chemistry involving the

ambient oxygen can occur at depths that are wellbelow the nominal penetration depth of the inci-dent radiation The relative importance of suchchemistry is roughly indicated by the ratio of thetotal surface area in the regolith to the area of Eu-roparsquos visible surface Using the porosity andgrain size above and a regolith depth dr of theorder of a meter then this ratio is large 3(1 2p)drrg which is 103ndash2 3 104 for average porosi-ties of 96 and 25 respectively Therefore thegas-phase O2 molecules below the visible surfaceinteract frequently with the grain surfaces In thisway a very large volume of the subsurface couldbe altered chemically even if it is not directly ex-posed to the radiation

From the overlying structures seen in the high-resolution images of Europarsquos surface by theGalileo spacecraft (eg Prockter and Pappalardo2000) it is clear that the youngest surfaces are red-der The reddening in the visible has been attrib-uted to the presence of sulfur in the ice (McEwen1986 Johnson et al 1988 Calvin et al 1995 Carl-son et al 1999b) The amount of reddening in thespectra of the surface material decreases with ageand the surface materials appear to becomebrighter with age This might occur by the slowaccumulation of a frost over-layer or by radioly-sis (Carlson et al 2002 Johnson et al 2003) Ra-diolysis preferentially converts the sulfur in theice to sulfur dioxide or sulfuric acid both ofwhich have flat spectra in the visible In regionswhere the radiation does not directly penetratethe O2 permeating the regolith can oxidize thesulfur at the surfaces of the icy grains and at graininterfaces Similarly the carbon seen in the ice onEuroparsquos leading hemisphere (eg Carlson et al2004 Johnson et al 2003) can be oxidized by ra-diolysis in the near surface layers or by reactionson the grain surfaces at depth in the regolith bythe ambient O2 The meteoroid bombardmentthat produces the regolith also exposes fresh sur-faces for oxidation and can cause the collapse ofpores leading to the trapping of oxygen In thismanner both refractory and volatile oxidants canbe formed and trapped in the subsurface ice forpotential delivery to Europarsquos subsurface ocean

SUMMARY

The presence of a thin oxygen atmosphere andoxidants trapped in Europarsquos surface ice (egJohnson et al 2003) combined with the possible

OXIDANTS AT EUROPA 841

presence of a subsurface ocean (eg Carr et al1998) have made Europa an exciting target forfuture space exploration The molecular oxidantsproduced by radiolysis and photolysis could bespecies needed for carbon-based biochemistry inEuroparsquos putative subsurface ocean Howeverthere has been no model for the formation trap-ping and transport of relevant oxidants

In this paper we have reviewed the space ob-servations and laboratory studies related to thechemistry of formation of oxidants in ice by theincident radiation Because the laboratory dataare incomplete we examined the rate equationsfor formation of oxygen and related species byradiolysis and photolysis We showed how suchequations can be simplified to the analytic modelthat was used for fitting the fluence and temper-ature dependence of the O2 yield produced bylow-energy electrons (Sieger et al 1998) In de-riving the analytic model from the more completerate equations the competition between forma-tion and destruction processes can be studied ascan the role of trapping sites escape from depthand temperature history These equations also al-lowed us to put incident photon electron and iondata on a similar footing for the first time

The model can be used to suggest measure-ments that are needed to describe the chemicalstate of the irradiated ice and to interpret new lab-oratory data We showed that if reactions are fastthe yield data could be consistent with a precur-sor being formed by two events as is the case forperoxide However we favor trapped O as a pre-cursor as initially suggested by Matich et al(1993) for production of excited O2 Such aspecies if present could be detected spectro-scopically as can trapped O2 through bands ac-tivated by lattice perturbations (Cooper et al2003a)

The available oxygen yield data were used tocompare G values for different incident particlesThese were then used to estimate the oxygen pro-duction in the radiation environment at EuropaThe lower bounds were consistent with the pro-duction rates obtained from a recent model of Eu-roparsquos atmosphere The porosity of the regolithwas also considered It not only affects the rela-tive sputtering yields for O2 and H2O but alsoprovides an enormously increased surface areaon which the ambient gas-phase oxygen can re-act with species in the ice That this might be oc-curring is consistent with the change in re-

flectance associated with surface aging Such re-actions could also convert ambient molecularoxygen into other more refractory species that re-main trapped on subduction of Europarsquos crust

Although radiolysis produces a stable chemi-cally altered surface in the laboratory the parti-cle flux to Europarsquos surface is many orders ofmagnitude lower Therefore annealing can be ef-fective at temperatures that are much lower thanthose typically found experimentally Belowabout 120K the annealing times appear to be long(eg Baragiola 2003) so that laboratory data canoften be directly applied to Europarsquos icy surfaceHowever account has to be taken of the syner-gism between the UV and the energetic plasmaas discussed earlier (Johnson and Quickenden1997)

Transient surface melting might initiate lifeprocesses but subduction of oxidants to the pu-tative ocean is likely to be required if biochem-istry is to occur at Europa (Chyba and Hand2001) Although the temperature versus depthinto Europarsquos regolith is not well known belowthe regolith the temperature and pressure appar-ently increase with depth At some depth (ap-proximately kilometers) either very warm ice oran ocean exists (Ruiz and Tejero 2000) Thereforechemically refractory oxidants are likely to be re-quired in order to survive subduction to thewarmer subsurface regions Trapped species ifdelivered downward can become mobile andwill segregate forming inclusions (Johnson andJesser 1997) Indeed peroxide appears to segre-gate in ice (Gurman et al 1967) and these inclu-sions may remain relatively stable until the per-oxide is dissolved in the putative ocean Thevolatiles seen in reflectance such as O2 SO2 andCO2 will also segregate But at the higher tem-peratures they could also diffuse through thebulk react or percolate into the regolith atmo-sphere Therefore the observation of the trappedvolatiles in Europarsquos surface ice is of interest pri-marily because it suggests that oxidants are pre-sent The formation of oxygen-rich molecules fol-lows the preferential loss of hydrogen Howeverthe surface is in a local steady state with im-planted H O and S from the plasma and the lossto space Since the plasma flux to the surface canbe variable this balance can also vary Indeedtemporal changes in reflectance have been re-ported (Domingue and Hendrix 2004)

Based on the models for formation of O2 ex-

JOHNSON ET AL842

amined in this paper additional experiments areneeded to determine the chemical state of irra-diated ice at low fluences Of immediate inter-est is the identification of stable molecular pre-cursors required for the formation of O2 Basedon the available laboratory data individual per-oxide molecules trapped in the ice do not appearto be the principal precursors but trapped O(Matich et al 1993) remains a viable candidateAlthough the data suggest that the local hydro-gen-bonding network is important in the pro-duction of O2 the effect of the formation tem-perature and radiation damage on the structureneed to be measured It is also important to de-scribe quantitatively the relationship betweenH2 loss and the formation of O2 and H2O2 overa range of temperatures and fluences Finallyone would like to determine the steady-statedensities of biochemically interesting moleculesformed by radiolysis in an ice containing sulfurand carbon at Europarsquos surface temperaturesand then determine their stability with increas-ing temperature and pressure Such experimentsstill need to be carried out Here we have pre-sented a model that brings together the variouslaboratory data and now can be used to considerthe formation trapping and transport of relevantoxidants at Europa

APPENDIX

Introduction

In this Appendix we examine rate equationsthat describe various aspects of the radiolysis orphotolysis of ice These equations need to be un-derstood if one is to be able to calculate the for-mation trapping and transport of oxidants Theemphasis is on the formation of molecular oxy-gen We integrate these equations over the escapedepth of O2 to give the simplified rate equationsdiscussed in the text in which number densitiesare replaced by column densities (eg Eqs 1 and2) In this way we can examine the relationshipbetween the rate constants and the effective pre-cursor formation and destruction cross sectionsextracted from data We first consider a precur-sor such as H2O2 that is formed from two exci-tation events This bears on the observed linear-ity of the oxygen yield versus fluence We thenconsider the equations for trapped versus a tran-

siently mobile species We use a reaction betweena mobile and trapped O as an example althoughO 1 OH can also give oxygen Finally we ex-amine the possible dependence of the O2 forma-tion process on the density of traps (defectsvoids and grain boundaries) as a function of tem-perature The results are discussed in the text

For clarity all possible radiolytic species arenot included in the discussion below Thereforepotentially important products such as HO2 andthe charged species (eg H3O1 the trapped elec-tron etc) are missing In each section we attemptto make a point related to production of O2 andother oxidants by using a subset of the productsrather than attempt a complete description withmany unknown rate constants

Fast versus slow processes

To describe the O2 production rate from a pre-cursor such as H2O2 we first use a simple but in-structive set of rate equations to describe the for-mation of H2O2 We then relate these to thesimplified rate equation for the precursor in thetext Eq 1 and show how the postulated precur-sor formation cross section sp is related to theH2O dissociation cross section sd by the reactionrates

Using the reactions

H2O 1 radiation R H 1 OH (sd)

H 1 H R H2 (k2H)

H 1 OH R H2O (kHOH)

OH 1 OH R H2O2 (k2OH)

the reaction pathways are described by

H2

Hqk2H

1 OHH2O 1 radiation mdashRsd H

OHH2O2

OH Rk2OH

kHOH

HH2O

OXIDANTS AT EUROPA 843

The corresponding rate equations in an irradiatedvolume are

dnOHdt 5 sd f n 2 2 k2OH nOH nOH

2 kHOH nH nOH (A1a)

dnHdt 5 sd f n 2 2 k2H nH nH

2 kHOH nH nOH (A2a)

dnH2O2dt 5 k2OH nOH nOH

2 destruction processes (A3a)

Here sd is the dissociation cross section f is theradiation flux and n is the molecular numberdensity of ice with nOH and nH the density of dis-sociation products OH and H respectively Therate constants k2OH k2H and kHOH describe thereactions above In this model the H2 formed isassumed to escape for temperatures above about20K We do not distinguish between trappedspecies and mobile species although in fact thatdistinction is important as discussed in the nextsection That is any OH produced at low tem-perature by one ion would likely be trapped priorto interacting with an OH produced by a subse-quent ion At higher temperatures this can breakdown as discussed in the text We first relate theabove set of equations to Eq 1 in the text and thenconsider precursor destruction processes

Equation 1 in the text is obtained by solvingthese equations together and then by integratingover a depth Dx Here Dx is the smaller of the pen-etration depth of the particle and the O2 escape(percolation) depth Assuming the molecular den-sity of ice n is constant with depth Dx n 5 N inEq 1 Comparing Eq 1 with Eq A3a the precur-sor formation rate sp f N is equivalent to [k2OH

nOH nOH] Dx Assuming that the processes in EqsA1a and A2a are fast then the steady state is read-ily obtained In addition if k2H k2OH there ispreferential loss of H2 at short times as suggestedby the laboratory data If steady state is reachedthe production of H2 is equal to that of H2O2 us-ing only the equations above Further one obtains

[k2OH nOH nOH] Dx R [fp sd] f N (A4)

with

fp 5 05[1 1 05 kHOH(k2H k2OH)05] (A5a)

Therefore the precursor formation cross sectionsp in Eq 1 is a fraction fp of the total dissocia-

tion cross sd Since laboratory data suggest theH2O2 fraction in an irradiated ice is small thenkHOH (k2OH k2H)05 which is likely givingfp R [(k2OH k2H)05kHOH]

Trapped versus mobile species

To examine the fluence dependence discussedin the text we note that trapped OH and freshlyproduced transiently mobile OH can be treatedas different species (eg Matich et al 1993)Therefore assuming no trapping of H the OHrate equation above would be replaced by twoequations one for the trapped species nOHt andthe other for the freshly produced mobile speciesnOH Assuming the reaction of two mobile OHunits is unlikely and that peroxide is formed fromthe interaction of a trapped OH with a subse-quently produced OH an expanded set of equa-tions is obtained

OH 1 trap R OHt (kOHt)

OH 1 OHt R H2O2 (kOHOHt)

H 1 OHt R H2O (kHOHt)

then the reaction pathways are described by

qH2

H2O 1 radiation mdashRsdH

qk2H

1 OHH

trap OH

OH RkOHt

OHt kOHOHtH2O2

kHOHkHOHt

H H2O HH2O

The resulting set of equations is

dnOHdt 5 sd f n 2 kHOH nH nOH

2 kOHt nOH nt 2 kOHOHt nOH nOHt (A1b)

dnOHtdt 5 kOHt nOH nt 2 kOHOHt nOH nOHt

2 kHOHt nH nOHt (A1b9)

dnHdt 5 sd f n 2 2 k2H nH nH 2

kHOH nH nOH 2 kHOHt nH nOHt (A2b)

JOHNSON ET AL844

dnH2O2dt 5 kOHOHt nOH nOHt

2 destruction processes (A3b)

where nt is the trap density in the irradiated iceAt low T we can assume nt is large so thatdntdt 0 simplifying the set of equations If theprecursor of interest is trapped O as discussed inthe text then a similar set of equations can be con-structed for O and Ot as in the example to followOf course a full description over a broad rangeof temperatures would require the considerationof the both trapped and mobile H O and OHthe rate of change of trap density and the vari-ous destruction processes

The above equations should be solved simul-taneously but for the incident particle fluxes typ-ically used in these studies the diffusion of mo-bile species to traps is relatively fast This is trueeven at low temperatures if there is a significantdensity of trapping sites This is usually the casefor samples formed at low T Therefore in a lowtemperature solid at low dose rates there is not asignificant steady-state background of diffusingOH There are trapped or freshly produced OHunits which rapidly react Averaging over thebombardment rate this can be approximated bywriting dnOHdt 0 and then separating nOH outin Eq A1b Assuming also that H diffuses rapidlyand either reacts or forms H2 we also writednHdt 0 on the average Therefore betweenimpacts only species trapped at defects pores orgrain surfaces are assumed to be present as sug-gested by the lack of a dependence on the inci-dent beam flux In this limit the rate of change oftrapped OH roughly simplifies to

dnOHtdt 5 2 k2H nH nH 2 2 kOHOHt nOH nOHt

It is seen that the production of peroxide dependson the loss of H by formation of H2 as discussedand at steady state the production rate for per-oxide equals the rate of formation of H2 in theseequations

Treating the mobile and trapped species sepa-rately we can further simplify the rate equationsand obtain the precursor model used in the textThat is integrating over the relevant depth Dxthe densities are replaced by column densitiesThe rate constants ki are then related to effectivecross sections si In this way cross sections aregiven as an interaction length for each freshlyproduced mobile radical [ki nj] R [si f] The

trapped OH column density NOHt and peroxideproduction rate can be estimated from ldquolinearrdquorate equations

dNOHtdt lt [sd f] N 2 2 [s2OH f] NOHt

2 [sHOH f] NOHt (A6a)

dNH2O2dt lt [s2OH f] NOHt 2 s NH2O2 (A6b)

Here sd is again the dissociation cross sectionand the destruction processes in Eq A3b are con-tained in s The second and third terms on theright in Eq A6a are the loss of OH due to the pro-duction of H2O2 and H2O In the integration ofEq A1b9 [2 kOH OHt nOH] becomes [2 s2OH f] and[kHOHt nOH] becomes [sHOH f] Therefore thedensity and the interaction length of mobile H arecontained in sHOH If H2O2 is indeed the pre-cursor then in Eq 2 in the text [sp f N] is be-comes [s2OH f NOHt] from Eq A6b Accountingfor precursor destruction which can give backtwo OH and treating H2 can both be included inthe integrated model via the si values

The simplified set of equations above allows usto examine the relative time (fluence) scales forprecursor formation (in this case H2O2) SolvingEqs A6a and A6b the column density of OH andthe concentration of precursors versus fluence cpcan be obtained Using the column of H2O2 pro-duced as the precursor column Np in Eq 2 thenO2 yield can be obtained The results can be com-pared with the results from Eqs1 and 2 in thetext The average precursor (H2O2) concentrationover the column depth of water molecules N is

cp 5 fp (sds)[1 2 exp(2s F)] 2 (ssr)

[1 2 exp(2sr F)] (A7)

with

fp 5 2s2OH(sr 2 s) (A5b)

In these equations sr 5 (2s2OH 1 sHOH) andssr 1

It is seen that the precursor concentration withtime cp depends on two rates the reaction ratesthat determine sr and the destruction rates in sAs in Eq 4a the yield can be written as

Y 5 sO2 cp N (A8)

If sr and s are comparable then the yield at small

OXIDANTS AT EUROPA 845

fluence is quadratic in F However if sr sthen the linear dependence on fluence is obtainedwith a small shift in the fluence as shown in Eq4c in the text Further the steady-state yield isnow

Y` (T) 5 sO2 [fp sd] Ns (A9)

That is sp R [fp sd] as in Eq A4 and the tem-perature dependence for the incident electrondata must be contained in [sO2 fp] or in the ratio[sO2 s2OH]sHOH]

For comparison with their precursor modelSieger et al (1998) examined the possibility of ob-taining the O2 precursor by considering reactionsfor a density of diffusing dissociation products asin Eqs A1 and A2 rather than a trapped productinteracting with a freshly formed dissociationproduct Based on the above the lowest fluence atwhich O2 is first detected experimentally sets alimit on the time scale for mobilendashmobile reactions

Effect of trap density

Equations A1b A1b9 A2b and A3b can be ap-proximately solved assuming that the dnOHdtdnHdt and dntdt are very small on the averageas discussed above In order to examine the roleof the trap density nt which can depend on thetemperature and method of formation and on theradiation history of the ice we use the slightlysimpler precursor model In this model the O pro-duced directly by dissociation traps forming aprecursor O2 is then formed when a freshly pro-duced O or other species reacts with a trappedoxygen atom Ot

H2O 1 radiation R H2 1 O (sd)

O 1 trap R Ot (kOt)

O 1 Ot R O2 (kOOt)

The reaction pathways for the oxygen species are

H2O 1 radiation mdashRsd9

trap OO2

kOtOt kOOt

H2 1 OOHt

kOR H(kHO t)

ROR

We assume for simplicity both H2 and O2 escapefrom depth Dx after formation We also allow mo-bile H to react with Ot (kHOt) removing trappedO and both the mobile H and O to be removedby other reactants R (eg OH or contaminants)This leads to a simplified set of equations not di-rectly involving OH or peroxide

dnOdt 5 sd9 f n 2 kOt nO nt 2 kOR nO nR

2 kOOt nO nOt (A1c)

dnOtdt 5 kOt nt nO 2 kOOt nO nOt

2 kHOt nH nOt (A1c9)

dnHdt 5 sd f n 2 kHOt nH nOt

2 kHR nH nR (A2c)

dnO2dt 5 kOOt nO nOt (A3c)

These equations of course act in parallel with theequations above for OH and mobile OH may re-act with trapped O to form O2 Here we focusonly on mobile O

Since the diffusion of H and O are fast and thedensity of traps is large the nH and nt averagedover the bombardment rate can be taken veryroughly as time independent Further if we as-sume the number of trapped O nOt is small com-pared with the number of sites at which H canreact then nH [sd f nkHR nR] For a low ra-diation flux f nH is small on the average as as-sumed Since the O2 yield YO2 is equal to[(dnO2dt) Dxf] with Dx the escape depth thenthe steady-state yield Y` can be obtained ana-lytically

[Y`][sd9 N] 5 1 1 (q 2 1) 2 [(q 2 1)2

1 8 q d]054 d (A10)

q 5 [kOt (nt 1 nR)] (kHOt nH)(sd9 f n) kOOt

5(sdsd9) [kOt (nt 1 nR)] kHOt[kOOt kHR nR]

d 5 05 [1 1 nt(nt 1 nR)]

We note that for nt large q can be large Assum-ing that q 1 and nt nR then Eq A10 be-comes

Y` R [sd9 N]q 5 (sd9 N) (sd9sd)

[kOOt kHR nR][kHOt kOt nt] (A5c)

JOHNSON ET AL846

Comparing with Y` in the simple precursormodel in Eq 4b then s R [(kHOt nH)f] R [sd nkHOt][kHR nR] In addition the product [sO2sp] R sd9 sd9 [kOOt n][kOt nt] That is the dis-sociation cross sections are modified by reactionrates

Quite remarkably it is seen in Eq A5c that thesteady-state yield Y` can depend inversely onthe density of traps nt in the temperature andflux regimes for which the approximations arerelevant That is if there are competing destruc-tion process for trapped O then at high trap den-sity a mobile O must find a trapped O before itbecomes trapped or before mobile H destroys theOt In this way the increasing O2 yield with in-creasing T can be related to the change in the trap-ping density nt with the formation temperatureof the ice sample As the temperature increasesand the density of traps in the bulk decreases thesurface and grain interfaces become the only trap-ping sites

ACKNOWLEDGMENTS

REJ acknowledges support by a Gledden Fel-lowship in Chemistry School of Biomedical andChemical Sciences at The University of WesternAustralia and by grants from the NSF Astron-omy Program and NASArsquos Planetary Geologyand Geophysics Program PDC acknowledgessupport by an Australian Postgraduate Award

ABBREVIATIONS

HST Hubble Space Telescope MJQ MatichndashJohnsonndashQuickenden NIMS Near Infrared Map-ping Spectrometer UV ultraviolet

REFERENCES

Ayotte P Smith RS Stevenson KP Dohnarsquolek ZKimmel GA and Kay BD (2001) Effect of porosityon the adsorption desorption trapping and release ofvolatile gases by amorphous solid water J Geophys Res106(E12) 33387ndash33392

Bahr DA Famaacute M and Baragiola RA (2001) Radiol-ysis of water ice in the outer solar system sputteringand trapping of radiation products J Geophys Res 10633285ndash33290

Baragiola RA (2003) Microporous amorphous water icefilms and astronomical implications In Water in Con-

fining Geometries edited by JP Devlin and B Buch El-sevier New York in press

Baragiola RA and Bahr DA (1998) Laboratory studiesof the optical properties and stability of oxygen onGanymede J Geophys Res 103 25865ndash25872

Baragiola RA Atteberry CL and Bahr DA (1999) Ori-gin of solid oxygen on Ganymede Reply to commentby RE Johnson on ldquoLaboratory studies of the opticalproperties and stability of oxygen on Ganymederdquo JGeophys Res E104 14183ndash14187

Baragiola RA Vidal RA Svendsen W Schou J ShiM Bahr DA and Atteberrry CL (2003) Sputteringof water ice Nucl Instrum Methods Phys Res B 209294ndash303

Bar-Nun A Herman G Laufer D and Rappaport ML(1985) Trapping and release of gases by water ice andimplications for icy bodies Icarus 63 317ndash332

Baverstock KF and Burns WG (1976) Primary pro-duction of oxygen from irradiated water as an expla-nation for decreased radiobiological oxygen enhance-ment at high LET Nature 260 316ndash318

Benit J and Brown WL (1990) Electronic sputtering ofoxygen and water molecules from thin films of waterice bombarded by MeV Ne1 ions Nucl Instrum Meth-ods Phys Rev B 46 448ndash451

Benit J Bibring J-P della-Negra S LeBeyec YMendenhall M Rocard F and Standing K (1987)Erosion of ices by ion irradiation Nucl Instrum Meth-ods Phys Rev B 1920 838ndash842

Brown WL Augustyniak WM Lanzerotti LJ John-son RE and Evatt R (1980) Linear and nonlinearprocesses in the erosion of H2O ice by fast light ionsPhys Rev Lett 43 1632ndash1635

Brown WL Augustyniak W M Simmons E Mar-cantonio KJ Lanzerotti LJ Johnson RE BoringJW Reimann CT Foti G and Pirronello V (1982)Erosion and molecular formation in condensed gasfilms by electronic energy loss of fast ions Nucl In-strum Methods B 1 307ndash314

Brown WL Augustyniak WM Marcantonio KJ Sim-mons EN Boring JW Johnson RE and ReimannCT (1984) Electronic sputtering of low-temperaturemolecular solids Nucl Instrum Methods B 1 307

Buratti B (1995) Photometry and surface structure of theicy Galilean satellites J Geophys Res 100(E9) 19061ndash19066

Buratti B and Veverka J (1983) Voyager photometry ofEuropa Icarus 55 93ndash110

Calvin WM and Spencer JR (1997) Latitudinal distri-bution of O2 on Ganymede observations with the Hub-ble Space Telescope Icarus 130 505ndash516

Calvin WM Clark RN Brown RH and Spencer JA(1995) Spectra of the icy Galilean satellites from 02 to5 mm a compilation new observations and a recentsummary J Geophys Res 100 19041ndash19048

Calvin WM Johnson RE and Spencer JA (1996) O2

on Ganymede spectral characteristics and plasma for-mation mechanisms Geophys Res Lett 23 673ndash676

Calvin WM Aninich VG and Brown RH (2004) Oxy-gen temperature and phase effects Icarus (in press)

OXIDANTS AT EUROPA 847

is followed by a peak in absorption at 10 eVThe results in Fig 4a show that there is a dra-matic increase in the production of O2 at 10 eVwhich is the onset of secondary electron produc-tion in the bulk (Orlando and Sieger 2003)

For incident fast ions the yield at a given tem-perature was found after a small fluence to beindependent of whether the sample was crys-talline or amorphous ice (Brown et al 1982)However using low-energy electrons the yieldswere found to be smaller for an initially crys-talline ice (Orlando and Sieger 2003) Since verylow doses of ion radiation (a few eV per mole-cule) can convert crystalline ice to amorphous iceat low temperatures (Leto and Baratta 2003) thedamage produced by the penetrating ions usedby Brown et al (1982) affected the O2 yield Thestructural damage produced may create trappingsites cause changes in the electronic relaxationpathways or enhance percolation

The temperature dependence of the O2 yield isshown in Fig 5 for electrons incident on initiallyamorphous (Fig 5a) and initially crystalline (Fig5b) ice A small decrease in efficiency is seen inboth cases at 120ndash130K where Ic forms fromamorphous ice and pore collapse occurs (Row-land et al 1991) This is followed by an increas-ing yield with increasing T and then by a more

dramatic drop near 160K where Ih forms Dur-ing the transition to Ih trapped molecules havebeen shown to escape efficiently from ice samples (Ayotte et al 2001) A large drop in the yield isalso seen in Fig 2b for incident Ar1 ions

In modeling the dramatic decreases in the yieldin Figs 2b and 5 care must be taken That is thetime for crystallization of an amorphous sample(Baragiola 2003) even allowing for the large un-certainties is 30 min at 145K and becomes 1min at 160K Therefore pulsed radiolysis exper-iments should be carried out at these tempera-tures and the transients observed Since the timefor conversion from amorphous to crystalline icecan be shorter than the experimental times at145K the results are not comparable to thosefor which the chemical state of the ice at a givenfluence is stable over experimental times Belowwe focus on the data below 140K

We have emphasized processes initiated by theelectronic excitations and ionizations producedHowever momentum transfer collisions whichare the dominant energy loss process for low-en-ergy ions also produce dissociations defects andO2 in ice (Bar-Nun et al 1985) The relative effi-ciency of momentum transfer and electronicprocesses for producing O2 needs further studyAs the penetration depths are much smaller when

JOHNSON ET AL830

FIG 4 a Relative yield of O2 from D2O ice versus incident electron energy E at T 5 120K (from Orlando et al1998 Orlando and Sieger 2003) b Relative yield of D2 from D2O ice versus incident energy E at T 5 88K (fromKimmel et al 1994)

a

b

the former dominates the produced O2 likely es-capes readily

H2 formation and loss correlation with O2

The production of O2 from a low temperatureice by radiolysis correlates with the formationand loss of H2 (or D2 from D2O ice) (Reimann etal 1984) D2O is used to reduce the backgroundin mass spectrometry studies The yields of O2from D2O and H2O ices were shown to be com-parable However the total yields from D2O wereabout 10 less (Reimann et al 1984) consistentwith H (D) transport being important It is seenin Fig 3 that D2 is promptly produced and lostfrom the D2O ice At higher fluence the D2 yieldis seen to exhibit a dependence like that for O2For the 15 MeV Ne1 ions which fully penetratethe ice sample the D2 yield eventually decreasessince D2 is produced and escapes from the fullthickness of the film

Because of H2 (D2) loss the surface region of

the irradiated ice becomes depleted in H (D) rel-ative to O favoring the formation of oxygen-richspecies The change seen in the ratio of H (D) toO in the irradiated ice is small An irradiated film(100 nm) remains stoichiometric to better than2 (Brown et al 1980) and in the near surface re-gion Auger spectroscopy suggests only a slightoxygen enhancement due to irradiation by keVelectrons (Rye et al 1978) Sieger et al (1998) sug-gested a precursor concentration of the order ofa percent Such concentrations are an order ofmagnitude smaller than the steady-state concen-tration of dissociated H2O (Table 2)

Using infrared spectroscopy the damaged H2Oconcentrations at low temperatures in the satu-ration regime are about 10 over the irradiatedvolume (Benit et al 1987 Watanabe et al 2000Gomis et al 2003) This fraction is the ratio of thedissociation cross section to the effective crosssection for recombination to H2O as discussedlater At a 10 concentration most H2O will havea neighbor that is dissociated

OXIDANTS AT EUROPA 831

FIG 5 a Relative steady-state yields of O2 from a D2O ice sample versus T (upper panel) The ice samples wereformed and irradiated at these temperatures Film erosion as a yield (lower panel) (from Orlando and Sieger 2003)b Relative steady-state yields of O2 versus T as in (a) (upper panel) Here the D2O ice samples are formed at 160Kpresumably becoming crystalline and then irradiated at the T shown A new sample is used for each data point in(a) and (b) Film erosion as a yield (lower panel) (from Orlando and Sieger 2003)

At temperatures above 40K it is seen in Fig2 that the D2 and O2 yields exhibit the same tem-perature dependence suggesting a correlationbetween D2 and O2 loss H2 (D2) can be formedby reactions between diffusing hydrogen fromdissociated water or directly [ie H2O R H2 1O (eg Kimmel and Orlando 1995) H2O R (H1 OH)cage R H2 1 O (eg Matich et al 1993)H3O1 1 e R H2 1 OH (eg Williams et al1996)] Subtraction of the nearly constant D2 yieldin Fig 2a gives the Arrhenius plot in the insetSince molecular hydrogen readily escapes atthese temperature it is the H2 (D2) formationprocess that has an activation energy of 003eVmolecule The fact that slopes of the yields forH2 and O2 above 40K are similar suggests thatH2 (D2) formation and loss is the rate-limitingstep in determining the production and loss of O2at these temperatures

Although the data at low T in Fig 2a have con-siderable uncertainties below about 20ndash40K theD2 production is roughly independent of tem-perature but the O2 yield continues to decreasewith decreasing temperature Radiolytically pro-duced D2 is mobile and can escape above15ndash20K (Watanabe et al 2000) but O2 is im-mobile at such temperatures and the incidentHe1 ions are much less efficient at causing per-colation than the incident Ne1 discussed earlier

The production of H2 (D2) from ice has alsobeen studied using low-energy electrons and UVphotons at very low temperatures In the formerstudies absolute yields were not obtained butthere appeared to be a threshold at 63 eV (Fig4b) This is roughly the threshold for excitationof a surface state (Khusnatdinov and Petrenko1992) and the threshold for peroxide productionis close to this value The efficiency for H2 (D2)production increases around 11 eV with a sec-ond steep rise at 17 eV (Kimmel et al 1994) TheO2 and D2 yields versus electron energy in Fig 4exhibit some similarity in the threshold regionalthough the uncertainties are large and the sam-ple temperatures different Measuring the H2 (D2)and the O2 ejecta simultaneously over a range oftemperatures as in the Ne1 experiments wouldbe instructive

Using 98 eV photons at low T (12K) Watanabeet al (2000) measured the D2 ejected and trappedin ice They obtained a cross section for produc-tion of D2 (Table 2) at low fluences that is10ndash20 of the total absorption cross sectionwith significant uncertainties They also found

that the equilibrium damage fraction in the sam-ple was 10ndash20 of the D2O molecules but theequilibrium production of D2 was about an orderof magnitude smaller 1ndash2 at their tempera-tures They suggested that the D2 is formed atgrain interfaces and cracks and on the surfaces ofpores In a mixed ice D2O 1 CO exposed to UVphotolysis D2 and CO2 formation are correlated(Watanabe and Kouchi 2002) The D2O dissoci-ates to produce D2 plus O and this transientlymobile O reacts with CO to produce CO2

Precursor state and peroxide

In order to produce O2 from low temperatureice it appears that the ice sample must lose hy-drogen and form trapped oxygen-rich reactivespecies The efficiency of production also appearsto be affected by the level of defects and damagein the ice consistent with the suggestion that thehydrogen-bonding network is important in theproduction of O2 via availability of trapping sitesnew electronic relaxation pathways or the natureof the bonding of precursors Here we briefly ex-amine the possible role of the suggested precur-sor molecules H2O2 and HO2 We note that H2O2might not readily dissolve in ice possibly ac-counting for its low abundance in some experi-ments It appears to be heterogeneously distrib-uted in the form of microcrystals (Gurman et al1967) If H2O2 does act as a precursor the for-mation of aggregates could affect the dimer ab-sorption bands (Cooper et al 2003a) and affectthe production and stability of O2 (Cooper et al2003b)

The results of Gerakines et al (1996) in whichice at 10K is irradiated by UV photons confirmthat H2O2 forms in ice from two OH after a build-up of trapped OH They monitored the band as-sociated with H2O-HO (Langford et al 2000) inwhich the HO is bound to water with 024 eVTherefore they might only be monitoring a sub-set of the trapped OH seen in earlier experiments(eg Taub and Eiben 1968) The H2O2 concen-tration in ice reaches a steady-state level at flu-ences of about 1018 photonscm2 (Gerakines et al1996) As seen in the model section this gives anestimate of the H2O2 photodestruction cross sec-tion of the order of 10218 cm2 Similarly HO2 be-gins to be seen in absorption at about a few times1017cm2 shortly after they begin to detect H2O2The absolute yields were not measured

Westley et al (1995) detected H2 and O2 pro-

JOHNSON ET AL832

duced by Lyman-a (98 eV) radiation of ice butfound no fluence dependence for fluences greaterthan 1017cm2 They also saw small amounts ofH2O2 by programmed thermal desorption of thesample but made no quantitative evaluationComparing their lowest fluence for the produc-tion of O2 with the result of Gerakines et al (1996)for peroxide suggests that H2O2 and HO2 areprobably not important precursors of O2 In UVphotolysis of H2O2 using 64 eV photons the pro-duction of O2 was not observed (Vaghjiani et al1992) and the excitation of an irradiated icewhich presumably contained some peroxide by37 eV photons did not produce O2 (Baragiola etal 2003) Direct excitation of matrix-isolatedH2O2 at low temperature also did not yield O2but the production of a mobile O that reacted withtrapped O (H2O-O) did give O2 (Khriachtchev etal 2000)

Irradiating with 200 keV protons at tempera-tures of 40ndash129K Bahr et al (2001) found onlyabout 0001ndash002 H2O2 using programmed ther-mal desorption as compared with earlier experi-ments that suggested larger values (eg Table 2)Using infrared absorption Moore and Hudson(2000) found a G value of 01 for incident MeVprotons at low temperature but the signal in theobserved band became negligible at 80K sug-gesting that H2O2 was not being formed Theyalso found that adding O2 significantly increasedthe production efficiency of H2O2 That is thatthe amount of peroxide was affected by theamount of trapped oxygen Based on the matrixisolation studies (Khriachtchev et al 2000) dis-sociated mobile O might trap as H2O-O whichon excitation converts to H2O2 Therefore even ifH2O2 is not the precursor the peroxide andtrapped oxygen concentrations will be related

Although molecular oxygen was not a productof photolysis of matrix isolated H2O2 it was aproduct of photolysis at 266 nm (466 eV) of per-oxide dimers in an argon matrix 2 H2O2 R 2 H2O1 O2 (Engdahl and Nelander 2002) This de-composition was thought to be due to the con-siderable distortion of the bonds in the dimerTherefore the configuration or distortion by thelattice of peroxide or any other precursor mole-cule in the ice may be critical (Sieger et al 1998)In addition more recent experiments show thatG values for the production of H2O2 by 30 keVincident ions are significant as are the steady-state ratios of H2O2H2O (Gomis et al 2004) Ra-tios of the order of 02ndash4 were found at 16K and

77K in the irradiated regions with incident H giv-ing the smallest fraction and incident O thelargest (Table 2) At these concentrations perox-ide molecules might segregate forming inclu-sions (Gurman et al 1967 Cooper et al 2003b)Since the amount of trapped peroxide also com-petes with the amount of trapped O studies areneeded to describe the chemical state of the solidproduced by radiolysis This will be critical in de-scribing the fate of the irradiation products at Eu-ropa In particular the chemical composition fa-vorable for producing O2 needs to be determinedas does the importance of trapped O H2O2 HO2or peroxide dimers as precursors

Other species

In the model below we will refer to atomic andmolecular species trapped in ice Such species areoften discussed in the literature on radiolysis andphotolysis (for summaries see Matich et al 1993Johnson and Quickenden 1997) Trapped OH OH2O2 and HO2 have been discussed andor mea-sured by a variety of techniques (absorption andluminescence spectroscopy electron spin reso-nance programmed thermal desorption etc)Care must be taken when using these data as re-sults for ice samples formed below 160K orthose exposed to penetrating ion beams likelyhave numerous defects voids and grain inter-faces so that a variety of trapping sites and trapdepths exist Therefore when a particular bandfor a trapped species (eg OH) disappears withincreasing temperature it could be due to achange in the nature of the trapping as well as toloss of the species by reaction

Although H is discussed below we do not dis-tinguish it in these simple models from H1

trapped as the hydronium ion H3O1 whichwould mean that there are corresponding nega-tive ions present Therefore below we refer to thevarious species as if they are neutral althoughthis might not be the case Although trapped elec-trons are detected only over a narrow range oftemperatures (100ndash120K) negative ions must beimportant over a broad range of temperature

We also note that SO2 and a sulfate and CO2and a carbonate appear to be trapped species inEuroparsquos icy surface Under radiolysis each ofthese can dissociate directly giving O2 Howeverwith the exception of the sulfate their densitiesare less than 1 so they are not relevant sourcesof O2 However their presence is indicative of the

OXIDANTS AT EUROPA 833

preference for forming oxygen-rich molecules inEuroparsquos ice and can affect the efficiency of pro-duction of O2 and H2O2

MODELS FOR FORMATION OF O2

Introduction

Based on the data described above the yieldsfor incident electrons and ions are found to benearly linear in fluence at the lowest fluences forwhich O2 is detected In addition the relevant al-terations of the ice samples by radiolysis appearto be stable and independent of the beam fluxRemarkably this is the case both for low-energyelectrons that make only a single excitation foreach impact and fast ions that make a density ofexcitations along their path through the solidThis suggests that O2 is produced and trapped inthe initial events and then caused to escape fromthe solid by the subsequent radiation or that astable chemical alteration of the ice is required be-fore the radiation can produce and eject O2

Because H produced by the dissociation of wa-ter molecules can attack trapped O2 and any pre-cursor oxidants the loss of H is important Thisoccurs by the production and loss of hydrogenmolecules as discussed leaving a surface that isslightly rich in O relative to H The O2 produc-tion mechanism could be direct excitation of aparticular precursor species or it might not bespecific That is a new OH or O produced by theincident radiation could react with the H-depleted solid producing O2 This would occurmore efficiently as the ratio of O to H in the nearsurface region increased

In their early work WL Brown and colleaguesassumed that processes similar to those used todescribe radiolysis of liquid and gaseous H2Ocould be used to describe the radiolysis in ice Thelow activation energy that they found was as-sumed to be associated with diffusion of somemobile radical species in ice such as H or O orpossibly diffusion of the O2 molecule itselfBased on their studies of ejection from isotopi-cally labeled D2O samples they concluded thatthe O2 is formed in the track of a single ion andthen ejected by subsequent ions (eg Benit andBrown 1990) More recently trapped specieshave been suggested as precursors These are ei-ther subsequently excited by the incident radia-tion or attacked by newly produced mobile

species We consider a simple model that incor-porates all of these processes and can be used toorganize and interpret the experimental data Wealso discuss its relationship to more completemodels In this discussion much of which is in theAppendix we consider the standard radiationproducts that have been often described How-ever in a solid the excited electron cloud pro-duced by absorption can overlap a number of lat-tice sites Therefore their relaxation pathwaysandor product species could be very differentfrom those that have been traditionally consideredin the literature on radiolysis and photolysis

Precursor model

The column density of a precursor species Npformed in ice by incident radiation can be calcu-lated using a simple model In this model cross sec-tions are used for the production sp and destruc-tion s of the precursor O2 is then formed byexcitation of the precursor or by reaction of a newlyformed radical with the precursor This is also de-scribed using a cross section sO2 For incident elec-trons having a small penetration depth the O2 pro-duced is assumed to rapidly diffuse to the surfaceand escape (Sieger et al 1998) More generally wedefine N as the column density of H2O from whichthe incident radiation can produce and eject O2 Apair of rate equations gives both Np and the pro-duction rate of oxygen dNO2dt That is

H2O 1 radiation R precursor

precursor 1 radiation R O2

giving

dNpdt 5 sp f N 2 s f Np (1)

dNO2dt 5 sO2 f Np (2)

In Eq 1 f is the incident flux and the total pre-cursor destruction cross section can be written ass 5 sp9 1 sO2 with sp9 accounting for precursordestruction processes other than the formation ofO2 Solving Eqs 1 and 2 and assuming N is nearlyconstant (10 level as discussed) the fractionof precursors cp in the affected column after timet is obtained (Appendix)

cp 5 NpN 5 [sps] [1 2 exp(2 s F)] (3)

Here F 5 [ft] is the fluence (flux 3 time) which

JOHNSON ET AL834

is the number of particles (photons electrons orions) incident on the surface per unit area Usingcp in Eq 2 the O2 yield Y at any fluence can bewritten in terms of the yield at steady state Y`

Y 5 [dNO2dt]f 5 sO2 cp N 5 Y`

[1 2 exp(2sF)] (4a)

Y` 5 [sp sO2 s] N (4b)

If we also account for the reduction in the den-sity of water molecules in the initial column Nthese forms change only slightly The size of thetwo parameters Y` and s can be obtained by fit-ting the yield versus fluence data such as that inFig 3

For the low-energy electron data Sieger et al(1998) showed that the ratio [YY`] for a givenincident electron energy is independent of tem-perature over a narrow range of temperatureTherefore the destruction cross section s in Eq4a is roughly independent of T and can be esti-mated from the [YY`] curve Sieger et al (1998)found a precursor destruction cross section s lt08ndash30 3 10216 cm2 for incident electron energiesfrom 30 to 100 eV This is of the order of the elec-tron impact ionization cross section for H2O

Based on the above model the lack of an ob-served fluence dependence for fluences down to1017 hncm2 in the Lyman-a irradiation of ice(Westley et al 1995) can be examined The low-est fluence used implies that the precursor de-struction cross section for Lyman-a is of the or-der of the Lyman-a absorption cross section in ice(09 3 10217 cm2)

Sieger et al (1998) also assumed that sp9 5 0This would mean that the precursor is only de-stroyed by production of O2 (ie s 5 sO2) Thatis unlikely since suggested precursors can be de-stroyed by reaction with H and other radicalsTheir assumption would also mean that the crosssection for production of O2 from a precursorsO2 is large This would be surprising and ap-pear to contradict some of the data discussedabove Instead we assume that s is not equal tosO2 and the total destruction cross section s isof the order of the ionization or absorption crosssection

Having obtained s the measured temperaturedependence for production of O2 is containedonly in Y` Over a narrow range of T one canwrite Y`(T) lt Y`9 exp(2EakT) and obtain the es-

timate of the activation energies in Table 2 Ifsp9 THORN 0 as discussed then sp and sO22 in Eq 4bcannot be obtained separately from the data forY` unless an independent measure of cp is ob-tained Such a measurement is critical for de-scribing the chemical state of the irradiated sur-faces

The observed temperature dependence of Y` iscontained in sp sO2 or N Here N is the ldquodepthrdquofrom which newly formed O2 can reach the sur-face and escape given as a column density of H2OThe physical depth is obtained by dividing thecolumn density N by the molecular number den-sity of the ice Assuming that in the low-energyelectron experiments N is independent of tem-perature and is about three monolayers (Sieger etal 1998) the measured Y` gives [sO2sps]05ndash20 3 10218 cm2 over the electron impactenergy range At 110ndash120K this ratio was foundto be proportional to (E 2 Eo) for E Eo 10 eVagain assuming fixed N (Orlando et al 1998 Or-lando and Sieger 2003)

In the following section we first consider a pre-cursor like peroxide that typically forms fromtwo dissociation events We use this to show un-der what circumstances such a model does notviolate the observed linearity of the O2 yield datawith fluence Following that we relate the sim-ple precursor model above to the case wheretrapped O is the precursor We then use themodel for trapped O to examine the effect of thetrap density on the production of O2 The densityof traps (defects) is determined by the sample for-mation temperature and the subsequent radiationdamage Finally we consider the escape depthN for penetrating radiation within the precursormodel These discussions are supported by moredetailed rate equations in the Appendix

Precursor trapped peroxide

The proposed precursor peroxide is typicallyformed from products of two dissociation eventsie 2 H2O R 2 H 1 2OH R H2 1 H2O2 This wasconfirmed in low temperature ice (Gerakines etal 1996) It is also possible that a dissociation pro-ducing a trapped O followed by an excitationevent can produce peroxide 2 (H2O 1 hn) R H2

1 H2O-O 1 hn R H2O2 1 H2 (eg Khriachtchevet al 2000) The O2 yield versus fluence for inci-dent electrons is reasonably well fit by the yieldexpression above for fluences greater than1014cm2 Therefore it has been suggested that

OXIDANTS AT EUROPA 835

any precursor species must be formed in a singleevent Using a model in the Appendix in whichthe precursor is formed by two events we showthis is not necessarily the case Precursor forma-tion if it is fast could involve two dissociationevents and give the measured fluence depen-dence Although we consider peroxide here theseconclusions apply to other possible precursors

In the example in the Appendix H2O is disso-ciated to H 1 OH according to the cross sectionsd The reactions of two OH units to form H2O2(rate constant k2OH) and 2 H to form H2 (k2H)compete with the recombination of H and OH toH2O (kHOH) If k2H is much greater than k2OH then H2 loss dominates at low fluences as ex-pected (Appendix) Steady state requires not sur-prisingly that the rate for forming H2O2 isroughly equal to the rate for forming H2 Finallyit is seen in the Appendix that the low H2O2 con-centrations suggested by experiment imply thatkHOH (k2H k2OH)05 Such sizes are reasonable

The set of rate equations in the Appendix thatuse number densities of species is integrated overdepth in order to obtain the forms used in Eqs 1and 2 which are based on column densities Inthis way the effective cross sections are related tothe rate constants for the interaction of freshlyproduced mobile species with previously pro-duced trapped species We show in the Appen-dix that the O2 yield versus fluence can be writ-ten in a form similar to that in Eq 4a but with asmall offset due to the two-step precursor for-mation process

Y lt Y`(T) [1 2 exp(2sF) 2 (ssr)]

for F sr21 (4c)

Here s is again the precursor destruction crosssection but sr 5 (2s2OH 1 sHOH) is the effectivecross section for loss of OH by formation of H2O2

and H2O The corresponding expression for Y`(T)is given in the Appendix

The form for the yield versus fluence in Eq 4cis consistent with the low-energy electron mea-surements if (ssr) 1 and if sr F21 lt10214 cm2 Such sizes are indeed possible Whatis required is that at the relevant temperatures(50ndash110K) the interaction length associated withsr is more than a few molecular layers (greaterthan 10 Aring) A cross section that is likely to belarge above 20K is sHOH which describes themean distance for diffusion of mobile H before it

reacts Since s discussed earlier is of the sameorder of magnitude as the H2O dissociation crosssection sd then the precursor concentration cp(in this case H2O2) is roughly proportional to[s2OHsHOH] 1 (Appendix)

Based on the above a two-step precursor for-mation process could be consistent with the elec-tron data contrary to the statements in Sieger etal (1998) Orlando et al (1998) and Orlando andSieger (2003) Experiments are needed to deter-mine the relevant cross sections Below we con-sider a precursor formed in a single excitationevent

Precursor trapped O

O2 can be formed for instance by diffusion ofa transiently mobile O to a previously formed andtrapped O or to the production of O near the siteof a previously produced and trapped O (Matichet al 1993 Johnson and Quickenden 1997) Werefer to this as the MatichndashJohnsonndashQuickenden(MJQ) model and note that O 1 OH R O2 1 Hcan also give molecular oxygen

Trapped O as H2O-O was in fact seen to be aprecursor to O2 when attacked by a mobile O ina Kr matrix (Khriachtchev et al 2000) ThereforeO trapped as H2O-O was also suggested as a pre-cursor in ice (Johnson 2001) Before proceedingwe note that the examination of the MJQ modelin Sieger et al (1998) was incorrect They consid-ered two mobile interacting species whose life-times were long compared with the bombard-ment rate rather than a new mobile O interactingwith a previously trapped O Although this hasnot been corrected recently they added Otrapped as H2O-O as a possible precursor (Or-lando and Sieger 2003) O trapped as H2O-O isseen in matrix isolation studies as discussedabove but it has not yet been identified in iceand like OH it may trap in a number of ways

The MJQ model is essentially the followingFormation of H2 1 O is followed by loss of H2and trapping of O (Ot) A subsequent dissocia-tion or excitation event can lead to the destruc-tion of Ot by reaction with for instance H or byconversion to H2O2 However if this dissociationevent also leads to H2 1 O a mobile O is pro-duced that can react with Ot with an interactionlength that depends on T This results in the for-mation of O2 [O 1 Ot R O2] giving a net pro-duction of 2 H2 and 1 O2 In the Appendix therelevant rate equations (Eqs A1cndashA3c) are con-

JOHNSON ET AL836

sidered Assuming the mobile O processes arefast these rate equations can also be integratedand simplified to give the form in Eqs 1 and 2The precursor column density Np now repre-sents the column density of Ot In this model sO2in Eq 2 is the ldquocross sectionrdquo for reaction of afreshly produced O with an Ot and sp9 indicatesdestruction of Ot Finally sp is the cross sectionfor production of H2 and Ot via processes dis-cussed above

Based on the sizes of the cross sections ex-tracted in the electron irradiation experimentsthe model with trapped O (Ot) as the precursoris quantitatively reasonable That is for a rangeof T both sp and sO2 might be described byevents of the type H2O R H2 1 O (whether di-rect or indirect) If these occur for 10 of theradiation-induced dissociation events as dis-cussed then the required product of cross sec-tions suggested by the sizes of s and Y` above isroughly obtained that is [sp sO2] 02ndash6 3 10234

cm2 in the low-energy electron experiments Thecross section for production of O might dependon the availability of dangling bonds at interfacesin pores or at defects as suggested by a numberof authors The importance of such bonds couldalso be consistent with the differences in the elec-tron-induced yield for initially crystalline and ini-tially amorphous ice (Orlando and Sieger 2003)

O2 production can increase in the presence ofoxygen-rich impurities such as CO2 or SO2 bothof which are known to be present at Europa Asin the MJQ model dissociation (eg CO2 R CO1 O) can provide a mobile O that can be trappedor react with a previously trapped O forming O2In addition such species are hydrogen scav-engers (Moore and Hudson 2000) This can re-duce the destruction of Ot affecting sp as seen inEq A5c

Trap density

In the simplified models above the density oftrapping sites nt was not explicitly accountedfor However in the discussion of both the ex-perimental data and models trapping sites havebeen assumed to be important It has beendemonstrated repeatedly that the density of de-fects voids and interfaces decreases as the tem-perature of formation of the ice increases How-ever the efficiency of production of O2 is theopposite it increases with increasing T There-fore the oxygen yield varies in a manner that is

the inverse of the trap density over a range oftemperatures For penetrating ions in which theenergy density deposited is high and O2 might beformed in a single track a simple explanation hasbeen given As the density of traps for O2 is re-duced and the temperature increases the O2 pro-duced can diffuse to the surface and escape moreefficiently limiting back reactions on further ir-radiation This is examined in the next sectionHowever such a process does not affect the low-energy electron yields for which the penetrationdepth is small and the formed O2 escapes read-ily at the measurement temperatures In Eqs A1cA1c9 A3c and A10 in the Appendix we againuse a simple set of rate equations for trapped andmobile O in order to determine the role of the trapdensity nt on the production of a precursorspecies The conclusions apply to any modelbased on a trapped precursor

In the MQJ model mobile O can be trappedforming Ot (described by kOt nt) and can reactwith previously trapped O to form O2 (kOOt) Inaddition it can react at other sites and Ot can bedestroyed by mobile reactive species such as H(kHOt nH) Mobile H can also react with othertrapped species (kHR nR) Incorporating these re-actions Eq A10 in the Appendix gives a steady-state yield Y` When the density of traps is largethe steady-state yield is given in Eq A5c

Y` R (sd9sd9sd) [kOOtkHR nr][kHOt kOt nt] N (5)

Here sd and sd9 are the dissociation cross sectionsthat determine the production of H (H 1 OH) andO (H2 1 O) by the incident radiation Comparingthe result in Eq 5 with Y` in the simple precur-sor model in Eq 4b the dissociation cross sec-tions in Eq 5 are now scaled by a ratio of the re-action rates

It is seen in Eq 5 that the steady-state yield Y`depends inversely on the density of traps nt Thatis there are competing destruction processes fortrapped O (Ot) Therefore in order to form O2 amobile O competes with mobile H to find atrapped O before it becomes trapped In this waythe increasing O2 yield correlates with the re-duction in the trapping sites with increasing tem-perature of formation The small activation ener-gies discussed earlier would be related to there-orientation of the lattice binding leading to thereduction in trap density At relatively high T thedensity of traps becomes small so that only the

OXIDANTS AT EUROPA 837

sample surface or grain interfaces might act astraps accounting for the observed drop in theyield at high T For this reason it is important todetermine if the yield for T greater than 150Kdepends on the incident flux f Since other quan-tities in the expression for Y` could also be tem-perature dependent experiments are neededThese would likely be pulsed radiolysis experi-ments in which the transients are studied ratherthan the steady-state yields However the avail-able data for lightly ionizing radiation do suggestthat the O2 yield depends inversely on trap den-sity as in the model above

Escape from depth

The quantity N in Eq 1 is the ldquodepthrdquo given asa molecular column density from which O2 formedbelow the surface can percolate to the surface andescape It can be converted to a physical depth bydividing by the molecular number density of theice In the case of the low-energy electrons thedepth below the surface at which ionizations andexcitations are produced is small However forfast ions that penetrate many monolayers this isnot the case Below we examine N in Eq 1

Although it has long been understood that twoexcitations events are required to produce O2(eg Brown et al 1982 Reimann et al 1984) theprecursor model suggests that at saturation(steady state) any new event can produce O2 fromthe steady-state density of trapped precursorsTherefore one might expect the yield at satura-tion to be nearly linear in the excitation cross sec-tion This is the case for the low-energy electronsand it was initially assumed to be the case for thefast ions (Brown et al 1980 Reimann et al 1984)However Baragiola et al (2003) showed that thetotal sputtering yield in the saturation region attemperatures greater than 100K which is dom-inated by decomposition (Brown et al 1982) isproportional to the square of the electronic en-ergy deposition per unit path length in the solid(dEdx)e Assuming that the production and de-struction cross sections in Y` in Eq 4b roughlyscale with (dEdx)e then their ratio [sp sO2s]also roughly varies as (dEdx)e If this is the casethen the observed quadratic dependence impliesthat the depth N from which O2 can be mobi-lized to escape must also depend on the excita-tion density (dEdx)e This is consistent with theobservation that the O2 yield depends on thick-ness for samples thinner than the ion penetration

depth (Reimann et al 1984 Benit and Brown1990) Therefore the percolation depth for escapeN is a critical factor in comparing yields for pen-etrating radiation with those for non-penetratingradiation like the low-energy electron and low-energy ion data It is also seen that percolationand trapping can be included within the simplemodel considered here

Model summary

We have examined the simple precursor modelintroduced by Sieger et al (1998) to explain theirelectron data We have shown how it can be gen-eralized and derived from an integration of theappropriate chemical rate equations with thecross sections and total column density contain-ing the reaction rates and the physics of diffusivetransport Therefore the parameters can havevery different meanings from those initially as-sumed The equations are useful when there arefast processes which go to completion in timesmuch shorter then the time between particle im-pacts and slow processes associated with speciesthat are assumed to be trapped between particleimpacts This separation allowed us to examinethose models proposed for precursor formationand destruction due to reacting mobile speciesformed from dissociation events Further fromthe size of the onset of the linearity of the yieldversus fluence we can place constraints on therole of nonlinear processes such as those thatmight be associated with the formation of a pre-cursor like peroxide

With these extensions this simple model is use-ful for approximating the full rate equations inthe temperature region for which the yields de-pend on the fluence (dose) and not on the doserate Based on the assumptions in the model thisimplies temperatures at which precursor speciesare stable over experimental times At the high-est temperatures for the yields shown in Fig 4athis might not be the case This should be testedexperimentally In this regard it is puzzling thatexperiments have shown that the absorptionbands associated with certain trapped speciessuch as OH and trapped electrons ldquodisappearrdquoat 100ndash120K (see eg Johnson and Quickenden1997) whereas the O2 yield data indicate that rel-evant precursor species must be stable to 140KTherefore the proposed O2 production processescan only be distinguished by directly determin-ing the chemical state of the irradiated ice

JOHNSON ET AL838

It may be the case that a particular precursoris not required as mentioned earlier Rather af-ter a sufficient level of oxidation a newly pro-duced O or OH can react in the oxygen-rich lat-tice producing O2 in a pseudo-first-order processThis can be tested experimentally Indeedtrapped O and peroxide may be changed into oneanother as in the matrix isolation experiments(Khriachtchev et al 2000) Therefore their rela-tive concentrations versus temperature and doseshould be studied

In the model considered here it is seen that theparameters in Y` must account for the strong tem-perature dependence Since the number of bulktrapping sites decreases with increasing T as theice sample is annealed one might expect that theprecursor concentration would decrease We sug-gest that the observed temperature dependenceis determined by the competition between theavailability of trapping sites and mean interactiondistances of a mobile species such as O and HThat is the O2 yield varies inversely with the den-sity of traps over a range of temperatures In thesimple model above this effect is included indi-rectly as shown using the more detailed equa-tions in the Appendix

EUROPA

Oxygen and peroxide production at Europa G values

The dependence of the yield on the thicknessof the ice sample for highly penetrating ions sug-gests that the O2 yield is roughly related to theamount of energy deposited in the sample There-fore the amount of O2 produced and lost fromice in steady state is often given as a G value thenumber of molecules produced per 100 eV de-posited in Table 2 In applying these G values tothe planetary environment care must be takensince the yields are temperature fluence and ex-citation density dependent They also are affectedby the presence of other proton scavengersknown to be present in Europarsquos ice

The G values for peroxide in Table 2 are thosefor production at low fluence but the G valuesfor O2 are for the production and escape of gas-phase oxygen at saturation Therefore they de-pend on the formation and destruction cross sec-tions in the model above Such G values can beused to describe the steady-state source of the am-

bient O2 atmosphere at Europa but are not help-ful in determining the ability to produce and trapO2 at depth into the icy surface Over the timescales of the experiments some O2 is seen to re-main trapped and is not destroyed by radicalsTypically there are a number of trapping sitesand recent estimates of the activation energy foran O2 molecule trapped in a substitutional site(Hori and Hondoh 2002) suggest that it can betrapped for times comparable to the time re-quired for the conversion of amorphous ice intoIh (Ayotte et al 2001) The stability of O2 inclu-sions has not been studied

To understand the relationship between the Gvalues for O2 in Table 2 we note that fast lightions and electrons lose their energy predomi-nantly by producing ionizations and excitationsThe ionization events produce secondary elec-trons that can result in further ionizations Thenumber of ionization events Ni produced by afast light ion stopping in a water ice is Ni 5EoW where Eo is the initial energy and W is theaverage energy expended per ionization event27 eV in water ice Therefore radiolysis by fastions is often described as the net effect of theshower of electrons set in motion by the incidention In this manner the G values for productionof O2 from ice by energetic light ions and elec-trons can be roughly estimated as the sum of theeffects of all the secondary electrons using theyield Ye for 10ndash100 eV electrons That is if weignore the percolation depth N we can write thenumber of O2 produced as the number of ioniza-tion events times Ye which already has in it thedestruction processes That is G NiYeEo YeW 0006 O2100 eV

This estimate of G is seen to be an order of mag-nitude larger than the keV proton value for O2 inTable 2 but is much smaller than the G value forenergetic Ne1 These differences from the simpleestimate are consistent with the observation thatthe yield in the saturation regime for penetratingions varies nonlinearly with excitation densityThe difference from the keV proton G value islikely due to the inability of O2 to escape fromdepth before destruction On the other hand thelarger G value seen using MeV Ne1 is likely dueto the high density of excitation This can resultin more efficient production of O2 increased pro-duction of damage sites for forming O2 andorincreased transport (percolation) along the tran-siently heated track That is the ability of theformed O2 to percolate to the surface during ion

OXIDANTS AT EUROPA 839

bombardment plays an essential role in deter-mining the measured yields as discussed aboveThis effect needs to be evaluated for a planetarysurface subject to particle fluxes that are many or-ders of magnitude smaller than those used in thelaboratory

In the absence of O2 yields for all of the rele-vant ions and energies an estimate of the totalsupply of O2 can be obtained using the G valuesin Table 2 For the energetic oxygen and sulfurions we use G values for neon and argon ionsThese are lower bounds as the data were obtainedat temperatures much lower than the surface tem-peratures at Europa (80ndash110K) For the fast pen-etrating electrons and protons we use the valuefor 200 keV protons in Table 2 For the heavy ionswe used the Ne1 and Ar1 values For compari-son we also give an estimate using the G valuesobtained from the low energy electron data andthe total ionization as discussed above

The G values are combined in Table 3 with theelectron and ion energy flux at Europa Althoughthere is some deflection of the flow due to in-duced fields the net energy flux to the surface isnot reduced much from these values (Paranicaset al 2002) In fact locally produced pick-up ionswhich are formed from the ejecta and acceleratedto the surface can contribute (Ip 1996) Furtheralthough G values are typically independent ofthe incident angle the loss of O2 is a surfaceprocess and is enhanced with angle That is thenumber of O2 produced within the escape depthincreases with increasing angle to the normal(steeper than the inverse cosine) This enhance-ment is 25ndash4 (Jurac et al 2001) Ignoring suchcorrections the total production rate calculated

(23ndash62 3 109 O2cm2s) is a lower bound to theactual production rate from the icy regions of Eu-roparsquos surface Recent modeling of the O2 atmo-sphere suggest that such a source rate is required(Shematovich and Johnson 2001)

Earlier estimates of the O2 source rate were in-correct because O2 production was assumed to bea fixed fraction of the H2O sputtering yield Atthe ambient temperature an ejected H2O sticks ef-ficiently (Smith and Kay 1997) to the surface ofneighboring grains in the porous regolith reduc-ing the sputtering yield for H2O (Johnson 1989)This reduction does not apply to the O2 fractionof the yield enhancing its relative importance toH2O Here we assume that all of the gas-phaseO2 produced can contribute to the ambient at-mosphere even if it is produced at a grain sur-face that is at some depth into the porous regolith

Finally Europarsquos surface is not pure ice every-where That is much of the trailing hemisphereis a hydrated sulfuric acid containing smallamounts of sulfur and SO2 and the leading hemi-sphere contains low levels of carbon CO2 andcarbonates Sodium and potassium atoms aresputtered from the icy regions (eg Leblanc et al2002) O2 production from surfaces containingthese species has not been measured C and S areboth proton scavengers and the other species canbe sources of O as discussed so that the O2 yieldfrom such regions could be enhanced Howeverthere can also be competing processes that mightreduce these rates In fact recent HST observa-tions suggest that the O2 might not be produceduniformly across Europarsquos surface (McGrath et al2000) The analysis of these observations is pre-liminary and is not simple since the observed

JOHNSON ET AL840

TABLE 3 APPLICATION TO EUROPA

OxygenImplantation Energy flux production

Species 109(cm2 s)a 109 keV(cm2 s)(1) 109(cm2 s)

H (10 keV) 0015(1) 12 0084 (072)bO (10 keV) 0015(1) 18 054cS (10 keV) 0009(1) 30 090cElectrons 62 043 (37)bPlasma (10 keV) (137 amu) 13(2) 10 03c

Total 79 23 (62)UV photons (64 eV) 46 01

aReferences are as follows 1Cooper et al (2001) [similar results in Paranicas et al (2002)] 2Mauk et al (1996)bUsing G for 200 keV protons (upper bounds in brackets are based on the low-energy electrons data as a model for

the secondary electrons produced)cUsing G for 15 MeV Ne1 at 7K a rough lower bound

emissions depend on the local plasma propertiesHowever the efficiency of O2 production in con-taminated ices needs to be studied

Porous regolith O2 interactions on grain surfaces

Although the O2 atmosphere is principally pro-duced in the near surface layers and is eventuallylost to space it also permeates Europarsquos porousregolith The porosity p (void space) of the re-golith is uncertain (Buratti 1995) It has been es-timated to be the most compact surface in theOuter Solar System p 025 on the trailing hemi-sphere and p 079 on the leading hemisphere(Buratti and Veverka 1983) and the most poroussurface p 096 globally (Domingue et al 1991Domingue and Hapke 1992) These large differ-ences are due to the uncertainties in the amountof backscattered light at small phase angles andon the photon absorption properties and grainsizes The more compact surface is likely correctGrain sizes rg of the order of 100 mm have beensuggested (Hansen and McCord 1999 2000Geissler et al 1998) Europarsquos regolith differs sig-nificantly from the lunar regolith which is asand-like surface of individual grains Analysis ofthe thermal inertia of Europarsquos surface (Grundyet al 2001) suggests that the ice grains are sin-tered forming a fairy castle-like structure Sin-tering can occur thermally or due to the incidentradiation In addition ices doped with acids andalkalis in the laboratory can be powdery solidswith the impurities primarily found at grain in-terfaces

The porous regolith affects the above estimatesof the oxygen production in three ways First forisotropic bombardment the mean distance d intothe surface at which an incident ion or photonstrikes a grain surface is d (89)rg(1 2 p) Us-ing the above value for rg then d 02 m if theporosity is very large p 96 Therefore de-sorbed molecules can come from a considerabledepth and will interact with grain surfaces Evenfor much lower porosities the effective sputteringyields can be significantly modified (Johnson1989) As discussed above the yields for ejectedH2O which sticks onto neighboring grains at theambient temperatures are reduced but those forejected O2 are not This enhances the relative im-portance of the O2 yields

Finally the surfaces and interfaces of the sin-tered grains in Europarsquos porous surface are placeswhere low temperature chemistry involving the

ambient oxygen can occur at depths that are wellbelow the nominal penetration depth of the inci-dent radiation The relative importance of suchchemistry is roughly indicated by the ratio of thetotal surface area in the regolith to the area of Eu-roparsquos visible surface Using the porosity andgrain size above and a regolith depth dr of theorder of a meter then this ratio is large 3(1 2p)drrg which is 103ndash2 3 104 for average porosi-ties of 96 and 25 respectively Therefore thegas-phase O2 molecules below the visible surfaceinteract frequently with the grain surfaces In thisway a very large volume of the subsurface couldbe altered chemically even if it is not directly ex-posed to the radiation

From the overlying structures seen in the high-resolution images of Europarsquos surface by theGalileo spacecraft (eg Prockter and Pappalardo2000) it is clear that the youngest surfaces are red-der The reddening in the visible has been attrib-uted to the presence of sulfur in the ice (McEwen1986 Johnson et al 1988 Calvin et al 1995 Carl-son et al 1999b) The amount of reddening in thespectra of the surface material decreases with ageand the surface materials appear to becomebrighter with age This might occur by the slowaccumulation of a frost over-layer or by radioly-sis (Carlson et al 2002 Johnson et al 2003) Ra-diolysis preferentially converts the sulfur in theice to sulfur dioxide or sulfuric acid both ofwhich have flat spectra in the visible In regionswhere the radiation does not directly penetratethe O2 permeating the regolith can oxidize thesulfur at the surfaces of the icy grains and at graininterfaces Similarly the carbon seen in the ice onEuroparsquos leading hemisphere (eg Carlson et al2004 Johnson et al 2003) can be oxidized by ra-diolysis in the near surface layers or by reactionson the grain surfaces at depth in the regolith bythe ambient O2 The meteoroid bombardmentthat produces the regolith also exposes fresh sur-faces for oxidation and can cause the collapse ofpores leading to the trapping of oxygen In thismanner both refractory and volatile oxidants canbe formed and trapped in the subsurface ice forpotential delivery to Europarsquos subsurface ocean

SUMMARY

The presence of a thin oxygen atmosphere andoxidants trapped in Europarsquos surface ice (egJohnson et al 2003) combined with the possible

OXIDANTS AT EUROPA 841

presence of a subsurface ocean (eg Carr et al1998) have made Europa an exciting target forfuture space exploration The molecular oxidantsproduced by radiolysis and photolysis could bespecies needed for carbon-based biochemistry inEuroparsquos putative subsurface ocean Howeverthere has been no model for the formation trap-ping and transport of relevant oxidants

In this paper we have reviewed the space ob-servations and laboratory studies related to thechemistry of formation of oxidants in ice by theincident radiation Because the laboratory dataare incomplete we examined the rate equationsfor formation of oxygen and related species byradiolysis and photolysis We showed how suchequations can be simplified to the analytic modelthat was used for fitting the fluence and temper-ature dependence of the O2 yield produced bylow-energy electrons (Sieger et al 1998) In de-riving the analytic model from the more completerate equations the competition between forma-tion and destruction processes can be studied ascan the role of trapping sites escape from depthand temperature history These equations also al-lowed us to put incident photon electron and iondata on a similar footing for the first time

The model can be used to suggest measure-ments that are needed to describe the chemicalstate of the irradiated ice and to interpret new lab-oratory data We showed that if reactions are fastthe yield data could be consistent with a precur-sor being formed by two events as is the case forperoxide However we favor trapped O as a pre-cursor as initially suggested by Matich et al(1993) for production of excited O2 Such aspecies if present could be detected spectro-scopically as can trapped O2 through bands ac-tivated by lattice perturbations (Cooper et al2003a)

The available oxygen yield data were used tocompare G values for different incident particlesThese were then used to estimate the oxygen pro-duction in the radiation environment at EuropaThe lower bounds were consistent with the pro-duction rates obtained from a recent model of Eu-roparsquos atmosphere The porosity of the regolithwas also considered It not only affects the rela-tive sputtering yields for O2 and H2O but alsoprovides an enormously increased surface areaon which the ambient gas-phase oxygen can re-act with species in the ice That this might be oc-curring is consistent with the change in re-

flectance associated with surface aging Such re-actions could also convert ambient molecularoxygen into other more refractory species that re-main trapped on subduction of Europarsquos crust

Although radiolysis produces a stable chemi-cally altered surface in the laboratory the parti-cle flux to Europarsquos surface is many orders ofmagnitude lower Therefore annealing can be ef-fective at temperatures that are much lower thanthose typically found experimentally Belowabout 120K the annealing times appear to be long(eg Baragiola 2003) so that laboratory data canoften be directly applied to Europarsquos icy surfaceHowever account has to be taken of the syner-gism between the UV and the energetic plasmaas discussed earlier (Johnson and Quickenden1997)

Transient surface melting might initiate lifeprocesses but subduction of oxidants to the pu-tative ocean is likely to be required if biochem-istry is to occur at Europa (Chyba and Hand2001) Although the temperature versus depthinto Europarsquos regolith is not well known belowthe regolith the temperature and pressure appar-ently increase with depth At some depth (ap-proximately kilometers) either very warm ice oran ocean exists (Ruiz and Tejero 2000) Thereforechemically refractory oxidants are likely to be re-quired in order to survive subduction to thewarmer subsurface regions Trapped species ifdelivered downward can become mobile andwill segregate forming inclusions (Johnson andJesser 1997) Indeed peroxide appears to segre-gate in ice (Gurman et al 1967) and these inclu-sions may remain relatively stable until the per-oxide is dissolved in the putative ocean Thevolatiles seen in reflectance such as O2 SO2 andCO2 will also segregate But at the higher tem-peratures they could also diffuse through thebulk react or percolate into the regolith atmo-sphere Therefore the observation of the trappedvolatiles in Europarsquos surface ice is of interest pri-marily because it suggests that oxidants are pre-sent The formation of oxygen-rich molecules fol-lows the preferential loss of hydrogen Howeverthe surface is in a local steady state with im-planted H O and S from the plasma and the lossto space Since the plasma flux to the surface canbe variable this balance can also vary Indeedtemporal changes in reflectance have been re-ported (Domingue and Hendrix 2004)

Based on the models for formation of O2 ex-

JOHNSON ET AL842

amined in this paper additional experiments areneeded to determine the chemical state of irra-diated ice at low fluences Of immediate inter-est is the identification of stable molecular pre-cursors required for the formation of O2 Basedon the available laboratory data individual per-oxide molecules trapped in the ice do not appearto be the principal precursors but trapped O(Matich et al 1993) remains a viable candidateAlthough the data suggest that the local hydro-gen-bonding network is important in the pro-duction of O2 the effect of the formation tem-perature and radiation damage on the structureneed to be measured It is also important to de-scribe quantitatively the relationship betweenH2 loss and the formation of O2 and H2O2 overa range of temperatures and fluences Finallyone would like to determine the steady-statedensities of biochemically interesting moleculesformed by radiolysis in an ice containing sulfurand carbon at Europarsquos surface temperaturesand then determine their stability with increas-ing temperature and pressure Such experimentsstill need to be carried out Here we have pre-sented a model that brings together the variouslaboratory data and now can be used to considerthe formation trapping and transport of relevantoxidants at Europa

APPENDIX

Introduction

In this Appendix we examine rate equationsthat describe various aspects of the radiolysis orphotolysis of ice These equations need to be un-derstood if one is to be able to calculate the for-mation trapping and transport of oxidants Theemphasis is on the formation of molecular oxy-gen We integrate these equations over the escapedepth of O2 to give the simplified rate equationsdiscussed in the text in which number densitiesare replaced by column densities (eg Eqs 1 and2) In this way we can examine the relationshipbetween the rate constants and the effective pre-cursor formation and destruction cross sectionsextracted from data We first consider a precur-sor such as H2O2 that is formed from two exci-tation events This bears on the observed linear-ity of the oxygen yield versus fluence We thenconsider the equations for trapped versus a tran-

siently mobile species We use a reaction betweena mobile and trapped O as an example althoughO 1 OH can also give oxygen Finally we ex-amine the possible dependence of the O2 forma-tion process on the density of traps (defectsvoids and grain boundaries) as a function of tem-perature The results are discussed in the text

For clarity all possible radiolytic species arenot included in the discussion below Thereforepotentially important products such as HO2 andthe charged species (eg H3O1 the trapped elec-tron etc) are missing In each section we attemptto make a point related to production of O2 andother oxidants by using a subset of the productsrather than attempt a complete description withmany unknown rate constants

Fast versus slow processes

To describe the O2 production rate from a pre-cursor such as H2O2 we first use a simple but in-structive set of rate equations to describe the for-mation of H2O2 We then relate these to thesimplified rate equation for the precursor in thetext Eq 1 and show how the postulated precur-sor formation cross section sp is related to theH2O dissociation cross section sd by the reactionrates

Using the reactions

H2O 1 radiation R H 1 OH (sd)

H 1 H R H2 (k2H)

H 1 OH R H2O (kHOH)

OH 1 OH R H2O2 (k2OH)

the reaction pathways are described by

H2

Hqk2H

1 OHH2O 1 radiation mdashRsd H

OHH2O2

OH Rk2OH

kHOH

HH2O

OXIDANTS AT EUROPA 843

The corresponding rate equations in an irradiatedvolume are

dnOHdt 5 sd f n 2 2 k2OH nOH nOH

2 kHOH nH nOH (A1a)

dnHdt 5 sd f n 2 2 k2H nH nH

2 kHOH nH nOH (A2a)

dnH2O2dt 5 k2OH nOH nOH

2 destruction processes (A3a)

Here sd is the dissociation cross section f is theradiation flux and n is the molecular numberdensity of ice with nOH and nH the density of dis-sociation products OH and H respectively Therate constants k2OH k2H and kHOH describe thereactions above In this model the H2 formed isassumed to escape for temperatures above about20K We do not distinguish between trappedspecies and mobile species although in fact thatdistinction is important as discussed in the nextsection That is any OH produced at low tem-perature by one ion would likely be trapped priorto interacting with an OH produced by a subse-quent ion At higher temperatures this can breakdown as discussed in the text We first relate theabove set of equations to Eq 1 in the text and thenconsider precursor destruction processes

Equation 1 in the text is obtained by solvingthese equations together and then by integratingover a depth Dx Here Dx is the smaller of the pen-etration depth of the particle and the O2 escape(percolation) depth Assuming the molecular den-sity of ice n is constant with depth Dx n 5 N inEq 1 Comparing Eq 1 with Eq A3a the precur-sor formation rate sp f N is equivalent to [k2OH

nOH nOH] Dx Assuming that the processes in EqsA1a and A2a are fast then the steady state is read-ily obtained In addition if k2H k2OH there ispreferential loss of H2 at short times as suggestedby the laboratory data If steady state is reachedthe production of H2 is equal to that of H2O2 us-ing only the equations above Further one obtains

[k2OH nOH nOH] Dx R [fp sd] f N (A4)

with

fp 5 05[1 1 05 kHOH(k2H k2OH)05] (A5a)

Therefore the precursor formation cross sectionsp in Eq 1 is a fraction fp of the total dissocia-

tion cross sd Since laboratory data suggest theH2O2 fraction in an irradiated ice is small thenkHOH (k2OH k2H)05 which is likely givingfp R [(k2OH k2H)05kHOH]

Trapped versus mobile species

To examine the fluence dependence discussedin the text we note that trapped OH and freshlyproduced transiently mobile OH can be treatedas different species (eg Matich et al 1993)Therefore assuming no trapping of H the OHrate equation above would be replaced by twoequations one for the trapped species nOHt andthe other for the freshly produced mobile speciesnOH Assuming the reaction of two mobile OHunits is unlikely and that peroxide is formed fromthe interaction of a trapped OH with a subse-quently produced OH an expanded set of equa-tions is obtained

OH 1 trap R OHt (kOHt)

OH 1 OHt R H2O2 (kOHOHt)

H 1 OHt R H2O (kHOHt)

then the reaction pathways are described by

qH2

H2O 1 radiation mdashRsdH

qk2H

1 OHH

trap OH

OH RkOHt

OHt kOHOHtH2O2

kHOHkHOHt

H H2O HH2O

The resulting set of equations is

dnOHdt 5 sd f n 2 kHOH nH nOH

2 kOHt nOH nt 2 kOHOHt nOH nOHt (A1b)

dnOHtdt 5 kOHt nOH nt 2 kOHOHt nOH nOHt

2 kHOHt nH nOHt (A1b9)

dnHdt 5 sd f n 2 2 k2H nH nH 2

kHOH nH nOH 2 kHOHt nH nOHt (A2b)

JOHNSON ET AL844

dnH2O2dt 5 kOHOHt nOH nOHt

2 destruction processes (A3b)

where nt is the trap density in the irradiated iceAt low T we can assume nt is large so thatdntdt 0 simplifying the set of equations If theprecursor of interest is trapped O as discussed inthe text then a similar set of equations can be con-structed for O and Ot as in the example to followOf course a full description over a broad rangeof temperatures would require the considerationof the both trapped and mobile H O and OHthe rate of change of trap density and the vari-ous destruction processes

The above equations should be solved simul-taneously but for the incident particle fluxes typ-ically used in these studies the diffusion of mo-bile species to traps is relatively fast This is trueeven at low temperatures if there is a significantdensity of trapping sites This is usually the casefor samples formed at low T Therefore in a lowtemperature solid at low dose rates there is not asignificant steady-state background of diffusingOH There are trapped or freshly produced OHunits which rapidly react Averaging over thebombardment rate this can be approximated bywriting dnOHdt 0 and then separating nOH outin Eq A1b Assuming also that H diffuses rapidlyand either reacts or forms H2 we also writednHdt 0 on the average Therefore betweenimpacts only species trapped at defects pores orgrain surfaces are assumed to be present as sug-gested by the lack of a dependence on the inci-dent beam flux In this limit the rate of change oftrapped OH roughly simplifies to

dnOHtdt 5 2 k2H nH nH 2 2 kOHOHt nOH nOHt

It is seen that the production of peroxide dependson the loss of H by formation of H2 as discussedand at steady state the production rate for per-oxide equals the rate of formation of H2 in theseequations

Treating the mobile and trapped species sepa-rately we can further simplify the rate equationsand obtain the precursor model used in the textThat is integrating over the relevant depth Dxthe densities are replaced by column densitiesThe rate constants ki are then related to effectivecross sections si In this way cross sections aregiven as an interaction length for each freshlyproduced mobile radical [ki nj] R [si f] The

trapped OH column density NOHt and peroxideproduction rate can be estimated from ldquolinearrdquorate equations

dNOHtdt lt [sd f] N 2 2 [s2OH f] NOHt

2 [sHOH f] NOHt (A6a)

dNH2O2dt lt [s2OH f] NOHt 2 s NH2O2 (A6b)

Here sd is again the dissociation cross sectionand the destruction processes in Eq A3b are con-tained in s The second and third terms on theright in Eq A6a are the loss of OH due to the pro-duction of H2O2 and H2O In the integration ofEq A1b9 [2 kOH OHt nOH] becomes [2 s2OH f] and[kHOHt nOH] becomes [sHOH f] Therefore thedensity and the interaction length of mobile H arecontained in sHOH If H2O2 is indeed the pre-cursor then in Eq 2 in the text [sp f N] is be-comes [s2OH f NOHt] from Eq A6b Accountingfor precursor destruction which can give backtwo OH and treating H2 can both be included inthe integrated model via the si values

The simplified set of equations above allows usto examine the relative time (fluence) scales forprecursor formation (in this case H2O2) SolvingEqs A6a and A6b the column density of OH andthe concentration of precursors versus fluence cpcan be obtained Using the column of H2O2 pro-duced as the precursor column Np in Eq 2 thenO2 yield can be obtained The results can be com-pared with the results from Eqs1 and 2 in thetext The average precursor (H2O2) concentrationover the column depth of water molecules N is

cp 5 fp (sds)[1 2 exp(2s F)] 2 (ssr)

[1 2 exp(2sr F)] (A7)

with

fp 5 2s2OH(sr 2 s) (A5b)

In these equations sr 5 (2s2OH 1 sHOH) andssr 1

It is seen that the precursor concentration withtime cp depends on two rates the reaction ratesthat determine sr and the destruction rates in sAs in Eq 4a the yield can be written as

Y 5 sO2 cp N (A8)

If sr and s are comparable then the yield at small

OXIDANTS AT EUROPA 845

fluence is quadratic in F However if sr sthen the linear dependence on fluence is obtainedwith a small shift in the fluence as shown in Eq4c in the text Further the steady-state yield isnow

Y` (T) 5 sO2 [fp sd] Ns (A9)

That is sp R [fp sd] as in Eq A4 and the tem-perature dependence for the incident electrondata must be contained in [sO2 fp] or in the ratio[sO2 s2OH]sHOH]

For comparison with their precursor modelSieger et al (1998) examined the possibility of ob-taining the O2 precursor by considering reactionsfor a density of diffusing dissociation products asin Eqs A1 and A2 rather than a trapped productinteracting with a freshly formed dissociationproduct Based on the above the lowest fluence atwhich O2 is first detected experimentally sets alimit on the time scale for mobilendashmobile reactions

Effect of trap density

Equations A1b A1b9 A2b and A3b can be ap-proximately solved assuming that the dnOHdtdnHdt and dntdt are very small on the averageas discussed above In order to examine the roleof the trap density nt which can depend on thetemperature and method of formation and on theradiation history of the ice we use the slightlysimpler precursor model In this model the O pro-duced directly by dissociation traps forming aprecursor O2 is then formed when a freshly pro-duced O or other species reacts with a trappedoxygen atom Ot

H2O 1 radiation R H2 1 O (sd)

O 1 trap R Ot (kOt)

O 1 Ot R O2 (kOOt)

The reaction pathways for the oxygen species are

H2O 1 radiation mdashRsd9

trap OO2

kOtOt kOOt

H2 1 OOHt

kOR H(kHO t)

ROR

We assume for simplicity both H2 and O2 escapefrom depth Dx after formation We also allow mo-bile H to react with Ot (kHOt) removing trappedO and both the mobile H and O to be removedby other reactants R (eg OH or contaminants)This leads to a simplified set of equations not di-rectly involving OH or peroxide

dnOdt 5 sd9 f n 2 kOt nO nt 2 kOR nO nR

2 kOOt nO nOt (A1c)

dnOtdt 5 kOt nt nO 2 kOOt nO nOt

2 kHOt nH nOt (A1c9)

dnHdt 5 sd f n 2 kHOt nH nOt

2 kHR nH nR (A2c)

dnO2dt 5 kOOt nO nOt (A3c)

These equations of course act in parallel with theequations above for OH and mobile OH may re-act with trapped O to form O2 Here we focusonly on mobile O

Since the diffusion of H and O are fast and thedensity of traps is large the nH and nt averagedover the bombardment rate can be taken veryroughly as time independent Further if we as-sume the number of trapped O nOt is small com-pared with the number of sites at which H canreact then nH [sd f nkHR nR] For a low ra-diation flux f nH is small on the average as as-sumed Since the O2 yield YO2 is equal to[(dnO2dt) Dxf] with Dx the escape depth thenthe steady-state yield Y` can be obtained ana-lytically

[Y`][sd9 N] 5 1 1 (q 2 1) 2 [(q 2 1)2

1 8 q d]054 d (A10)

q 5 [kOt (nt 1 nR)] (kHOt nH)(sd9 f n) kOOt

5(sdsd9) [kOt (nt 1 nR)] kHOt[kOOt kHR nR]

d 5 05 [1 1 nt(nt 1 nR)]

We note that for nt large q can be large Assum-ing that q 1 and nt nR then Eq A10 be-comes

Y` R [sd9 N]q 5 (sd9 N) (sd9sd)

[kOOt kHR nR][kHOt kOt nt] (A5c)

JOHNSON ET AL846

Comparing with Y` in the simple precursormodel in Eq 4b then s R [(kHOt nH)f] R [sd nkHOt][kHR nR] In addition the product [sO2sp] R sd9 sd9 [kOOt n][kOt nt] That is the dis-sociation cross sections are modified by reactionrates

Quite remarkably it is seen in Eq A5c that thesteady-state yield Y` can depend inversely onthe density of traps nt in the temperature andflux regimes for which the approximations arerelevant That is if there are competing destruc-tion process for trapped O then at high trap den-sity a mobile O must find a trapped O before itbecomes trapped or before mobile H destroys theOt In this way the increasing O2 yield with in-creasing T can be related to the change in the trap-ping density nt with the formation temperatureof the ice sample As the temperature increasesand the density of traps in the bulk decreases thesurface and grain interfaces become the only trap-ping sites

ACKNOWLEDGMENTS

REJ acknowledges support by a Gledden Fel-lowship in Chemistry School of Biomedical andChemical Sciences at The University of WesternAustralia and by grants from the NSF Astron-omy Program and NASArsquos Planetary Geologyand Geophysics Program PDC acknowledgessupport by an Australian Postgraduate Award

ABBREVIATIONS

HST Hubble Space Telescope MJQ MatichndashJohnsonndashQuickenden NIMS Near Infrared Map-ping Spectrometer UV ultraviolet

REFERENCES

Ayotte P Smith RS Stevenson KP Dohnarsquolek ZKimmel GA and Kay BD (2001) Effect of porosityon the adsorption desorption trapping and release ofvolatile gases by amorphous solid water J Geophys Res106(E12) 33387ndash33392

Bahr DA Famaacute M and Baragiola RA (2001) Radiol-ysis of water ice in the outer solar system sputteringand trapping of radiation products J Geophys Res 10633285ndash33290

Baragiola RA (2003) Microporous amorphous water icefilms and astronomical implications In Water in Con-

fining Geometries edited by JP Devlin and B Buch El-sevier New York in press

Baragiola RA and Bahr DA (1998) Laboratory studiesof the optical properties and stability of oxygen onGanymede J Geophys Res 103 25865ndash25872

Baragiola RA Atteberry CL and Bahr DA (1999) Ori-gin of solid oxygen on Ganymede Reply to commentby RE Johnson on ldquoLaboratory studies of the opticalproperties and stability of oxygen on Ganymederdquo JGeophys Res E104 14183ndash14187

Baragiola RA Vidal RA Svendsen W Schou J ShiM Bahr DA and Atteberrry CL (2003) Sputteringof water ice Nucl Instrum Methods Phys Res B 209294ndash303

Bar-Nun A Herman G Laufer D and Rappaport ML(1985) Trapping and release of gases by water ice andimplications for icy bodies Icarus 63 317ndash332

Baverstock KF and Burns WG (1976) Primary pro-duction of oxygen from irradiated water as an expla-nation for decreased radiobiological oxygen enhance-ment at high LET Nature 260 316ndash318

Benit J and Brown WL (1990) Electronic sputtering ofoxygen and water molecules from thin films of waterice bombarded by MeV Ne1 ions Nucl Instrum Meth-ods Phys Rev B 46 448ndash451

Benit J Bibring J-P della-Negra S LeBeyec YMendenhall M Rocard F and Standing K (1987)Erosion of ices by ion irradiation Nucl Instrum Meth-ods Phys Rev B 1920 838ndash842

Brown WL Augustyniak WM Lanzerotti LJ John-son RE and Evatt R (1980) Linear and nonlinearprocesses in the erosion of H2O ice by fast light ionsPhys Rev Lett 43 1632ndash1635

Brown WL Augustyniak W M Simmons E Mar-cantonio KJ Lanzerotti LJ Johnson RE BoringJW Reimann CT Foti G and Pirronello V (1982)Erosion and molecular formation in condensed gasfilms by electronic energy loss of fast ions Nucl In-strum Methods B 1 307ndash314

Brown WL Augustyniak WM Marcantonio KJ Sim-mons EN Boring JW Johnson RE and ReimannCT (1984) Electronic sputtering of low-temperaturemolecular solids Nucl Instrum Methods B 1 307

Buratti B (1995) Photometry and surface structure of theicy Galilean satellites J Geophys Res 100(E9) 19061ndash19066

Buratti B and Veverka J (1983) Voyager photometry ofEuropa Icarus 55 93ndash110

Calvin WM and Spencer JR (1997) Latitudinal distri-bution of O2 on Ganymede observations with the Hub-ble Space Telescope Icarus 130 505ndash516

Calvin WM Clark RN Brown RH and Spencer JA(1995) Spectra of the icy Galilean satellites from 02 to5 mm a compilation new observations and a recentsummary J Geophys Res 100 19041ndash19048

Calvin WM Johnson RE and Spencer JA (1996) O2

on Ganymede spectral characteristics and plasma for-mation mechanisms Geophys Res Lett 23 673ndash676

Calvin WM Aninich VG and Brown RH (2004) Oxy-gen temperature and phase effects Icarus (in press)

OXIDANTS AT EUROPA 847

the former dominates the produced O2 likely es-capes readily

H2 formation and loss correlation with O2

The production of O2 from a low temperatureice by radiolysis correlates with the formationand loss of H2 (or D2 from D2O ice) (Reimann etal 1984) D2O is used to reduce the backgroundin mass spectrometry studies The yields of O2from D2O and H2O ices were shown to be com-parable However the total yields from D2O wereabout 10 less (Reimann et al 1984) consistentwith H (D) transport being important It is seenin Fig 3 that D2 is promptly produced and lostfrom the D2O ice At higher fluence the D2 yieldis seen to exhibit a dependence like that for O2For the 15 MeV Ne1 ions which fully penetratethe ice sample the D2 yield eventually decreasessince D2 is produced and escapes from the fullthickness of the film

Because of H2 (D2) loss the surface region of

the irradiated ice becomes depleted in H (D) rel-ative to O favoring the formation of oxygen-richspecies The change seen in the ratio of H (D) toO in the irradiated ice is small An irradiated film(100 nm) remains stoichiometric to better than2 (Brown et al 1980) and in the near surface re-gion Auger spectroscopy suggests only a slightoxygen enhancement due to irradiation by keVelectrons (Rye et al 1978) Sieger et al (1998) sug-gested a precursor concentration of the order ofa percent Such concentrations are an order ofmagnitude smaller than the steady-state concen-tration of dissociated H2O (Table 2)

Using infrared spectroscopy the damaged H2Oconcentrations at low temperatures in the satu-ration regime are about 10 over the irradiatedvolume (Benit et al 1987 Watanabe et al 2000Gomis et al 2003) This fraction is the ratio of thedissociation cross section to the effective crosssection for recombination to H2O as discussedlater At a 10 concentration most H2O will havea neighbor that is dissociated

OXIDANTS AT EUROPA 831

FIG 5 a Relative steady-state yields of O2 from a D2O ice sample versus T (upper panel) The ice samples wereformed and irradiated at these temperatures Film erosion as a yield (lower panel) (from Orlando and Sieger 2003)b Relative steady-state yields of O2 versus T as in (a) (upper panel) Here the D2O ice samples are formed at 160Kpresumably becoming crystalline and then irradiated at the T shown A new sample is used for each data point in(a) and (b) Film erosion as a yield (lower panel) (from Orlando and Sieger 2003)

At temperatures above 40K it is seen in Fig2 that the D2 and O2 yields exhibit the same tem-perature dependence suggesting a correlationbetween D2 and O2 loss H2 (D2) can be formedby reactions between diffusing hydrogen fromdissociated water or directly [ie H2O R H2 1O (eg Kimmel and Orlando 1995) H2O R (H1 OH)cage R H2 1 O (eg Matich et al 1993)H3O1 1 e R H2 1 OH (eg Williams et al1996)] Subtraction of the nearly constant D2 yieldin Fig 2a gives the Arrhenius plot in the insetSince molecular hydrogen readily escapes atthese temperature it is the H2 (D2) formationprocess that has an activation energy of 003eVmolecule The fact that slopes of the yields forH2 and O2 above 40K are similar suggests thatH2 (D2) formation and loss is the rate-limitingstep in determining the production and loss of O2at these temperatures

Although the data at low T in Fig 2a have con-siderable uncertainties below about 20ndash40K theD2 production is roughly independent of tem-perature but the O2 yield continues to decreasewith decreasing temperature Radiolytically pro-duced D2 is mobile and can escape above15ndash20K (Watanabe et al 2000) but O2 is im-mobile at such temperatures and the incidentHe1 ions are much less efficient at causing per-colation than the incident Ne1 discussed earlier

The production of H2 (D2) from ice has alsobeen studied using low-energy electrons and UVphotons at very low temperatures In the formerstudies absolute yields were not obtained butthere appeared to be a threshold at 63 eV (Fig4b) This is roughly the threshold for excitationof a surface state (Khusnatdinov and Petrenko1992) and the threshold for peroxide productionis close to this value The efficiency for H2 (D2)production increases around 11 eV with a sec-ond steep rise at 17 eV (Kimmel et al 1994) TheO2 and D2 yields versus electron energy in Fig 4exhibit some similarity in the threshold regionalthough the uncertainties are large and the sam-ple temperatures different Measuring the H2 (D2)and the O2 ejecta simultaneously over a range oftemperatures as in the Ne1 experiments wouldbe instructive

Using 98 eV photons at low T (12K) Watanabeet al (2000) measured the D2 ejected and trappedin ice They obtained a cross section for produc-tion of D2 (Table 2) at low fluences that is10ndash20 of the total absorption cross sectionwith significant uncertainties They also found

that the equilibrium damage fraction in the sam-ple was 10ndash20 of the D2O molecules but theequilibrium production of D2 was about an orderof magnitude smaller 1ndash2 at their tempera-tures They suggested that the D2 is formed atgrain interfaces and cracks and on the surfaces ofpores In a mixed ice D2O 1 CO exposed to UVphotolysis D2 and CO2 formation are correlated(Watanabe and Kouchi 2002) The D2O dissoci-ates to produce D2 plus O and this transientlymobile O reacts with CO to produce CO2

Precursor state and peroxide

In order to produce O2 from low temperatureice it appears that the ice sample must lose hy-drogen and form trapped oxygen-rich reactivespecies The efficiency of production also appearsto be affected by the level of defects and damagein the ice consistent with the suggestion that thehydrogen-bonding network is important in theproduction of O2 via availability of trapping sitesnew electronic relaxation pathways or the natureof the bonding of precursors Here we briefly ex-amine the possible role of the suggested precur-sor molecules H2O2 and HO2 We note that H2O2might not readily dissolve in ice possibly ac-counting for its low abundance in some experi-ments It appears to be heterogeneously distrib-uted in the form of microcrystals (Gurman et al1967) If H2O2 does act as a precursor the for-mation of aggregates could affect the dimer ab-sorption bands (Cooper et al 2003a) and affectthe production and stability of O2 (Cooper et al2003b)

The results of Gerakines et al (1996) in whichice at 10K is irradiated by UV photons confirmthat H2O2 forms in ice from two OH after a build-up of trapped OH They monitored the band as-sociated with H2O-HO (Langford et al 2000) inwhich the HO is bound to water with 024 eVTherefore they might only be monitoring a sub-set of the trapped OH seen in earlier experiments(eg Taub and Eiben 1968) The H2O2 concen-tration in ice reaches a steady-state level at flu-ences of about 1018 photonscm2 (Gerakines et al1996) As seen in the model section this gives anestimate of the H2O2 photodestruction cross sec-tion of the order of 10218 cm2 Similarly HO2 be-gins to be seen in absorption at about a few times1017cm2 shortly after they begin to detect H2O2The absolute yields were not measured

Westley et al (1995) detected H2 and O2 pro-

JOHNSON ET AL832

duced by Lyman-a (98 eV) radiation of ice butfound no fluence dependence for fluences greaterthan 1017cm2 They also saw small amounts ofH2O2 by programmed thermal desorption of thesample but made no quantitative evaluationComparing their lowest fluence for the produc-tion of O2 with the result of Gerakines et al (1996)for peroxide suggests that H2O2 and HO2 areprobably not important precursors of O2 In UVphotolysis of H2O2 using 64 eV photons the pro-duction of O2 was not observed (Vaghjiani et al1992) and the excitation of an irradiated icewhich presumably contained some peroxide by37 eV photons did not produce O2 (Baragiola etal 2003) Direct excitation of matrix-isolatedH2O2 at low temperature also did not yield O2but the production of a mobile O that reacted withtrapped O (H2O-O) did give O2 (Khriachtchev etal 2000)

Irradiating with 200 keV protons at tempera-tures of 40ndash129K Bahr et al (2001) found onlyabout 0001ndash002 H2O2 using programmed ther-mal desorption as compared with earlier experi-ments that suggested larger values (eg Table 2)Using infrared absorption Moore and Hudson(2000) found a G value of 01 for incident MeVprotons at low temperature but the signal in theobserved band became negligible at 80K sug-gesting that H2O2 was not being formed Theyalso found that adding O2 significantly increasedthe production efficiency of H2O2 That is thatthe amount of peroxide was affected by theamount of trapped oxygen Based on the matrixisolation studies (Khriachtchev et al 2000) dis-sociated mobile O might trap as H2O-O whichon excitation converts to H2O2 Therefore even ifH2O2 is not the precursor the peroxide andtrapped oxygen concentrations will be related

Although molecular oxygen was not a productof photolysis of matrix isolated H2O2 it was aproduct of photolysis at 266 nm (466 eV) of per-oxide dimers in an argon matrix 2 H2O2 R 2 H2O1 O2 (Engdahl and Nelander 2002) This de-composition was thought to be due to the con-siderable distortion of the bonds in the dimerTherefore the configuration or distortion by thelattice of peroxide or any other precursor mole-cule in the ice may be critical (Sieger et al 1998)In addition more recent experiments show thatG values for the production of H2O2 by 30 keVincident ions are significant as are the steady-state ratios of H2O2H2O (Gomis et al 2004) Ra-tios of the order of 02ndash4 were found at 16K and

77K in the irradiated regions with incident H giv-ing the smallest fraction and incident O thelargest (Table 2) At these concentrations perox-ide molecules might segregate forming inclu-sions (Gurman et al 1967 Cooper et al 2003b)Since the amount of trapped peroxide also com-petes with the amount of trapped O studies areneeded to describe the chemical state of the solidproduced by radiolysis This will be critical in de-scribing the fate of the irradiation products at Eu-ropa In particular the chemical composition fa-vorable for producing O2 needs to be determinedas does the importance of trapped O H2O2 HO2or peroxide dimers as precursors

Other species

In the model below we will refer to atomic andmolecular species trapped in ice Such species areoften discussed in the literature on radiolysis andphotolysis (for summaries see Matich et al 1993Johnson and Quickenden 1997) Trapped OH OH2O2 and HO2 have been discussed andor mea-sured by a variety of techniques (absorption andluminescence spectroscopy electron spin reso-nance programmed thermal desorption etc)Care must be taken when using these data as re-sults for ice samples formed below 160K orthose exposed to penetrating ion beams likelyhave numerous defects voids and grain inter-faces so that a variety of trapping sites and trapdepths exist Therefore when a particular bandfor a trapped species (eg OH) disappears withincreasing temperature it could be due to achange in the nature of the trapping as well as toloss of the species by reaction

Although H is discussed below we do not dis-tinguish it in these simple models from H1

trapped as the hydronium ion H3O1 whichwould mean that there are corresponding nega-tive ions present Therefore below we refer to thevarious species as if they are neutral althoughthis might not be the case Although trapped elec-trons are detected only over a narrow range oftemperatures (100ndash120K) negative ions must beimportant over a broad range of temperature

We also note that SO2 and a sulfate and CO2and a carbonate appear to be trapped species inEuroparsquos icy surface Under radiolysis each ofthese can dissociate directly giving O2 Howeverwith the exception of the sulfate their densitiesare less than 1 so they are not relevant sourcesof O2 However their presence is indicative of the

OXIDANTS AT EUROPA 833

preference for forming oxygen-rich molecules inEuroparsquos ice and can affect the efficiency of pro-duction of O2 and H2O2

MODELS FOR FORMATION OF O2

Introduction

Based on the data described above the yieldsfor incident electrons and ions are found to benearly linear in fluence at the lowest fluences forwhich O2 is detected In addition the relevant al-terations of the ice samples by radiolysis appearto be stable and independent of the beam fluxRemarkably this is the case both for low-energyelectrons that make only a single excitation foreach impact and fast ions that make a density ofexcitations along their path through the solidThis suggests that O2 is produced and trapped inthe initial events and then caused to escape fromthe solid by the subsequent radiation or that astable chemical alteration of the ice is required be-fore the radiation can produce and eject O2

Because H produced by the dissociation of wa-ter molecules can attack trapped O2 and any pre-cursor oxidants the loss of H is important Thisoccurs by the production and loss of hydrogenmolecules as discussed leaving a surface that isslightly rich in O relative to H The O2 produc-tion mechanism could be direct excitation of aparticular precursor species or it might not bespecific That is a new OH or O produced by theincident radiation could react with the H-depleted solid producing O2 This would occurmore efficiently as the ratio of O to H in the nearsurface region increased

In their early work WL Brown and colleaguesassumed that processes similar to those used todescribe radiolysis of liquid and gaseous H2Ocould be used to describe the radiolysis in ice Thelow activation energy that they found was as-sumed to be associated with diffusion of somemobile radical species in ice such as H or O orpossibly diffusion of the O2 molecule itselfBased on their studies of ejection from isotopi-cally labeled D2O samples they concluded thatthe O2 is formed in the track of a single ion andthen ejected by subsequent ions (eg Benit andBrown 1990) More recently trapped specieshave been suggested as precursors These are ei-ther subsequently excited by the incident radia-tion or attacked by newly produced mobile

species We consider a simple model that incor-porates all of these processes and can be used toorganize and interpret the experimental data Wealso discuss its relationship to more completemodels In this discussion much of which is in theAppendix we consider the standard radiationproducts that have been often described How-ever in a solid the excited electron cloud pro-duced by absorption can overlap a number of lat-tice sites Therefore their relaxation pathwaysandor product species could be very differentfrom those that have been traditionally consideredin the literature on radiolysis and photolysis

Precursor model

The column density of a precursor species Npformed in ice by incident radiation can be calcu-lated using a simple model In this model cross sec-tions are used for the production sp and destruc-tion s of the precursor O2 is then formed byexcitation of the precursor or by reaction of a newlyformed radical with the precursor This is also de-scribed using a cross section sO2 For incident elec-trons having a small penetration depth the O2 pro-duced is assumed to rapidly diffuse to the surfaceand escape (Sieger et al 1998) More generally wedefine N as the column density of H2O from whichthe incident radiation can produce and eject O2 Apair of rate equations gives both Np and the pro-duction rate of oxygen dNO2dt That is

H2O 1 radiation R precursor

precursor 1 radiation R O2

giving

dNpdt 5 sp f N 2 s f Np (1)

dNO2dt 5 sO2 f Np (2)

In Eq 1 f is the incident flux and the total pre-cursor destruction cross section can be written ass 5 sp9 1 sO2 with sp9 accounting for precursordestruction processes other than the formation ofO2 Solving Eqs 1 and 2 and assuming N is nearlyconstant (10 level as discussed) the fractionof precursors cp in the affected column after timet is obtained (Appendix)

cp 5 NpN 5 [sps] [1 2 exp(2 s F)] (3)

Here F 5 [ft] is the fluence (flux 3 time) which

JOHNSON ET AL834

is the number of particles (photons electrons orions) incident on the surface per unit area Usingcp in Eq 2 the O2 yield Y at any fluence can bewritten in terms of the yield at steady state Y`

Y 5 [dNO2dt]f 5 sO2 cp N 5 Y`

[1 2 exp(2sF)] (4a)

Y` 5 [sp sO2 s] N (4b)

If we also account for the reduction in the den-sity of water molecules in the initial column Nthese forms change only slightly The size of thetwo parameters Y` and s can be obtained by fit-ting the yield versus fluence data such as that inFig 3

For the low-energy electron data Sieger et al(1998) showed that the ratio [YY`] for a givenincident electron energy is independent of tem-perature over a narrow range of temperatureTherefore the destruction cross section s in Eq4a is roughly independent of T and can be esti-mated from the [YY`] curve Sieger et al (1998)found a precursor destruction cross section s lt08ndash30 3 10216 cm2 for incident electron energiesfrom 30 to 100 eV This is of the order of the elec-tron impact ionization cross section for H2O

Based on the above model the lack of an ob-served fluence dependence for fluences down to1017 hncm2 in the Lyman-a irradiation of ice(Westley et al 1995) can be examined The low-est fluence used implies that the precursor de-struction cross section for Lyman-a is of the or-der of the Lyman-a absorption cross section in ice(09 3 10217 cm2)

Sieger et al (1998) also assumed that sp9 5 0This would mean that the precursor is only de-stroyed by production of O2 (ie s 5 sO2) Thatis unlikely since suggested precursors can be de-stroyed by reaction with H and other radicalsTheir assumption would also mean that the crosssection for production of O2 from a precursorsO2 is large This would be surprising and ap-pear to contradict some of the data discussedabove Instead we assume that s is not equal tosO2 and the total destruction cross section s isof the order of the ionization or absorption crosssection

Having obtained s the measured temperaturedependence for production of O2 is containedonly in Y` Over a narrow range of T one canwrite Y`(T) lt Y`9 exp(2EakT) and obtain the es-

timate of the activation energies in Table 2 Ifsp9 THORN 0 as discussed then sp and sO22 in Eq 4bcannot be obtained separately from the data forY` unless an independent measure of cp is ob-tained Such a measurement is critical for de-scribing the chemical state of the irradiated sur-faces

The observed temperature dependence of Y` iscontained in sp sO2 or N Here N is the ldquodepthrdquofrom which newly formed O2 can reach the sur-face and escape given as a column density of H2OThe physical depth is obtained by dividing thecolumn density N by the molecular number den-sity of the ice Assuming that in the low-energyelectron experiments N is independent of tem-perature and is about three monolayers (Sieger etal 1998) the measured Y` gives [sO2sps]05ndash20 3 10218 cm2 over the electron impactenergy range At 110ndash120K this ratio was foundto be proportional to (E 2 Eo) for E Eo 10 eVagain assuming fixed N (Orlando et al 1998 Or-lando and Sieger 2003)

In the following section we first consider a pre-cursor like peroxide that typically forms fromtwo dissociation events We use this to show un-der what circumstances such a model does notviolate the observed linearity of the O2 yield datawith fluence Following that we relate the sim-ple precursor model above to the case wheretrapped O is the precursor We then use themodel for trapped O to examine the effect of thetrap density on the production of O2 The densityof traps (defects) is determined by the sample for-mation temperature and the subsequent radiationdamage Finally we consider the escape depthN for penetrating radiation within the precursormodel These discussions are supported by moredetailed rate equations in the Appendix

Precursor trapped peroxide

The proposed precursor peroxide is typicallyformed from products of two dissociation eventsie 2 H2O R 2 H 1 2OH R H2 1 H2O2 This wasconfirmed in low temperature ice (Gerakines etal 1996) It is also possible that a dissociation pro-ducing a trapped O followed by an excitationevent can produce peroxide 2 (H2O 1 hn) R H2

1 H2O-O 1 hn R H2O2 1 H2 (eg Khriachtchevet al 2000) The O2 yield versus fluence for inci-dent electrons is reasonably well fit by the yieldexpression above for fluences greater than1014cm2 Therefore it has been suggested that

OXIDANTS AT EUROPA 835

any precursor species must be formed in a singleevent Using a model in the Appendix in whichthe precursor is formed by two events we showthis is not necessarily the case Precursor forma-tion if it is fast could involve two dissociationevents and give the measured fluence depen-dence Although we consider peroxide here theseconclusions apply to other possible precursors

In the example in the Appendix H2O is disso-ciated to H 1 OH according to the cross sectionsd The reactions of two OH units to form H2O2(rate constant k2OH) and 2 H to form H2 (k2H)compete with the recombination of H and OH toH2O (kHOH) If k2H is much greater than k2OH then H2 loss dominates at low fluences as ex-pected (Appendix) Steady state requires not sur-prisingly that the rate for forming H2O2 isroughly equal to the rate for forming H2 Finallyit is seen in the Appendix that the low H2O2 con-centrations suggested by experiment imply thatkHOH (k2H k2OH)05 Such sizes are reasonable

The set of rate equations in the Appendix thatuse number densities of species is integrated overdepth in order to obtain the forms used in Eqs 1and 2 which are based on column densities Inthis way the effective cross sections are related tothe rate constants for the interaction of freshlyproduced mobile species with previously pro-duced trapped species We show in the Appen-dix that the O2 yield versus fluence can be writ-ten in a form similar to that in Eq 4a but with asmall offset due to the two-step precursor for-mation process

Y lt Y`(T) [1 2 exp(2sF) 2 (ssr)]

for F sr21 (4c)

Here s is again the precursor destruction crosssection but sr 5 (2s2OH 1 sHOH) is the effectivecross section for loss of OH by formation of H2O2

and H2O The corresponding expression for Y`(T)is given in the Appendix

The form for the yield versus fluence in Eq 4cis consistent with the low-energy electron mea-surements if (ssr) 1 and if sr F21 lt10214 cm2 Such sizes are indeed possible Whatis required is that at the relevant temperatures(50ndash110K) the interaction length associated withsr is more than a few molecular layers (greaterthan 10 Aring) A cross section that is likely to belarge above 20K is sHOH which describes themean distance for diffusion of mobile H before it

reacts Since s discussed earlier is of the sameorder of magnitude as the H2O dissociation crosssection sd then the precursor concentration cp(in this case H2O2) is roughly proportional to[s2OHsHOH] 1 (Appendix)

Based on the above a two-step precursor for-mation process could be consistent with the elec-tron data contrary to the statements in Sieger etal (1998) Orlando et al (1998) and Orlando andSieger (2003) Experiments are needed to deter-mine the relevant cross sections Below we con-sider a precursor formed in a single excitationevent

Precursor trapped O

O2 can be formed for instance by diffusion ofa transiently mobile O to a previously formed andtrapped O or to the production of O near the siteof a previously produced and trapped O (Matichet al 1993 Johnson and Quickenden 1997) Werefer to this as the MatichndashJohnsonndashQuickenden(MJQ) model and note that O 1 OH R O2 1 Hcan also give molecular oxygen

Trapped O as H2O-O was in fact seen to be aprecursor to O2 when attacked by a mobile O ina Kr matrix (Khriachtchev et al 2000) ThereforeO trapped as H2O-O was also suggested as a pre-cursor in ice (Johnson 2001) Before proceedingwe note that the examination of the MJQ modelin Sieger et al (1998) was incorrect They consid-ered two mobile interacting species whose life-times were long compared with the bombard-ment rate rather than a new mobile O interactingwith a previously trapped O Although this hasnot been corrected recently they added Otrapped as H2O-O as a possible precursor (Or-lando and Sieger 2003) O trapped as H2O-O isseen in matrix isolation studies as discussedabove but it has not yet been identified in iceand like OH it may trap in a number of ways

The MJQ model is essentially the followingFormation of H2 1 O is followed by loss of H2and trapping of O (Ot) A subsequent dissocia-tion or excitation event can lead to the destruc-tion of Ot by reaction with for instance H or byconversion to H2O2 However if this dissociationevent also leads to H2 1 O a mobile O is pro-duced that can react with Ot with an interactionlength that depends on T This results in the for-mation of O2 [O 1 Ot R O2] giving a net pro-duction of 2 H2 and 1 O2 In the Appendix therelevant rate equations (Eqs A1cndashA3c) are con-

JOHNSON ET AL836

sidered Assuming the mobile O processes arefast these rate equations can also be integratedand simplified to give the form in Eqs 1 and 2The precursor column density Np now repre-sents the column density of Ot In this model sO2in Eq 2 is the ldquocross sectionrdquo for reaction of afreshly produced O with an Ot and sp9 indicatesdestruction of Ot Finally sp is the cross sectionfor production of H2 and Ot via processes dis-cussed above

Based on the sizes of the cross sections ex-tracted in the electron irradiation experimentsthe model with trapped O (Ot) as the precursoris quantitatively reasonable That is for a rangeof T both sp and sO2 might be described byevents of the type H2O R H2 1 O (whether di-rect or indirect) If these occur for 10 of theradiation-induced dissociation events as dis-cussed then the required product of cross sec-tions suggested by the sizes of s and Y` above isroughly obtained that is [sp sO2] 02ndash6 3 10234

cm2 in the low-energy electron experiments Thecross section for production of O might dependon the availability of dangling bonds at interfacesin pores or at defects as suggested by a numberof authors The importance of such bonds couldalso be consistent with the differences in the elec-tron-induced yield for initially crystalline and ini-tially amorphous ice (Orlando and Sieger 2003)

O2 production can increase in the presence ofoxygen-rich impurities such as CO2 or SO2 bothof which are known to be present at Europa Asin the MJQ model dissociation (eg CO2 R CO1 O) can provide a mobile O that can be trappedor react with a previously trapped O forming O2In addition such species are hydrogen scav-engers (Moore and Hudson 2000) This can re-duce the destruction of Ot affecting sp as seen inEq A5c

Trap density

In the simplified models above the density oftrapping sites nt was not explicitly accountedfor However in the discussion of both the ex-perimental data and models trapping sites havebeen assumed to be important It has beendemonstrated repeatedly that the density of de-fects voids and interfaces decreases as the tem-perature of formation of the ice increases How-ever the efficiency of production of O2 is theopposite it increases with increasing T There-fore the oxygen yield varies in a manner that is

the inverse of the trap density over a range oftemperatures For penetrating ions in which theenergy density deposited is high and O2 might beformed in a single track a simple explanation hasbeen given As the density of traps for O2 is re-duced and the temperature increases the O2 pro-duced can diffuse to the surface and escape moreefficiently limiting back reactions on further ir-radiation This is examined in the next sectionHowever such a process does not affect the low-energy electron yields for which the penetrationdepth is small and the formed O2 escapes read-ily at the measurement temperatures In Eqs A1cA1c9 A3c and A10 in the Appendix we againuse a simple set of rate equations for trapped andmobile O in order to determine the role of the trapdensity nt on the production of a precursorspecies The conclusions apply to any modelbased on a trapped precursor

In the MQJ model mobile O can be trappedforming Ot (described by kOt nt) and can reactwith previously trapped O to form O2 (kOOt) Inaddition it can react at other sites and Ot can bedestroyed by mobile reactive species such as H(kHOt nH) Mobile H can also react with othertrapped species (kHR nR) Incorporating these re-actions Eq A10 in the Appendix gives a steady-state yield Y` When the density of traps is largethe steady-state yield is given in Eq A5c

Y` R (sd9sd9sd) [kOOtkHR nr][kHOt kOt nt] N (5)

Here sd and sd9 are the dissociation cross sectionsthat determine the production of H (H 1 OH) andO (H2 1 O) by the incident radiation Comparingthe result in Eq 5 with Y` in the simple precur-sor model in Eq 4b the dissociation cross sec-tions in Eq 5 are now scaled by a ratio of the re-action rates

It is seen in Eq 5 that the steady-state yield Y`depends inversely on the density of traps nt Thatis there are competing destruction processes fortrapped O (Ot) Therefore in order to form O2 amobile O competes with mobile H to find atrapped O before it becomes trapped In this waythe increasing O2 yield correlates with the re-duction in the trapping sites with increasing tem-perature of formation The small activation ener-gies discussed earlier would be related to there-orientation of the lattice binding leading to thereduction in trap density At relatively high T thedensity of traps becomes small so that only the

OXIDANTS AT EUROPA 837

sample surface or grain interfaces might act astraps accounting for the observed drop in theyield at high T For this reason it is important todetermine if the yield for T greater than 150Kdepends on the incident flux f Since other quan-tities in the expression for Y` could also be tem-perature dependent experiments are neededThese would likely be pulsed radiolysis experi-ments in which the transients are studied ratherthan the steady-state yields However the avail-able data for lightly ionizing radiation do suggestthat the O2 yield depends inversely on trap den-sity as in the model above

Escape from depth

The quantity N in Eq 1 is the ldquodepthrdquo given asa molecular column density from which O2 formedbelow the surface can percolate to the surface andescape It can be converted to a physical depth bydividing by the molecular number density of theice In the case of the low-energy electrons thedepth below the surface at which ionizations andexcitations are produced is small However forfast ions that penetrate many monolayers this isnot the case Below we examine N in Eq 1

Although it has long been understood that twoexcitations events are required to produce O2(eg Brown et al 1982 Reimann et al 1984) theprecursor model suggests that at saturation(steady state) any new event can produce O2 fromthe steady-state density of trapped precursorsTherefore one might expect the yield at satura-tion to be nearly linear in the excitation cross sec-tion This is the case for the low-energy electronsand it was initially assumed to be the case for thefast ions (Brown et al 1980 Reimann et al 1984)However Baragiola et al (2003) showed that thetotal sputtering yield in the saturation region attemperatures greater than 100K which is dom-inated by decomposition (Brown et al 1982) isproportional to the square of the electronic en-ergy deposition per unit path length in the solid(dEdx)e Assuming that the production and de-struction cross sections in Y` in Eq 4b roughlyscale with (dEdx)e then their ratio [sp sO2s]also roughly varies as (dEdx)e If this is the casethen the observed quadratic dependence impliesthat the depth N from which O2 can be mobi-lized to escape must also depend on the excita-tion density (dEdx)e This is consistent with theobservation that the O2 yield depends on thick-ness for samples thinner than the ion penetration

depth (Reimann et al 1984 Benit and Brown1990) Therefore the percolation depth for escapeN is a critical factor in comparing yields for pen-etrating radiation with those for non-penetratingradiation like the low-energy electron and low-energy ion data It is also seen that percolationand trapping can be included within the simplemodel considered here

Model summary

We have examined the simple precursor modelintroduced by Sieger et al (1998) to explain theirelectron data We have shown how it can be gen-eralized and derived from an integration of theappropriate chemical rate equations with thecross sections and total column density contain-ing the reaction rates and the physics of diffusivetransport Therefore the parameters can havevery different meanings from those initially as-sumed The equations are useful when there arefast processes which go to completion in timesmuch shorter then the time between particle im-pacts and slow processes associated with speciesthat are assumed to be trapped between particleimpacts This separation allowed us to examinethose models proposed for precursor formationand destruction due to reacting mobile speciesformed from dissociation events Further fromthe size of the onset of the linearity of the yieldversus fluence we can place constraints on therole of nonlinear processes such as those thatmight be associated with the formation of a pre-cursor like peroxide

With these extensions this simple model is use-ful for approximating the full rate equations inthe temperature region for which the yields de-pend on the fluence (dose) and not on the doserate Based on the assumptions in the model thisimplies temperatures at which precursor speciesare stable over experimental times At the high-est temperatures for the yields shown in Fig 4athis might not be the case This should be testedexperimentally In this regard it is puzzling thatexperiments have shown that the absorptionbands associated with certain trapped speciessuch as OH and trapped electrons ldquodisappearrdquoat 100ndash120K (see eg Johnson and Quickenden1997) whereas the O2 yield data indicate that rel-evant precursor species must be stable to 140KTherefore the proposed O2 production processescan only be distinguished by directly determin-ing the chemical state of the irradiated ice

JOHNSON ET AL838

It may be the case that a particular precursoris not required as mentioned earlier Rather af-ter a sufficient level of oxidation a newly pro-duced O or OH can react in the oxygen-rich lat-tice producing O2 in a pseudo-first-order processThis can be tested experimentally Indeedtrapped O and peroxide may be changed into oneanother as in the matrix isolation experiments(Khriachtchev et al 2000) Therefore their rela-tive concentrations versus temperature and doseshould be studied

In the model considered here it is seen that theparameters in Y` must account for the strong tem-perature dependence Since the number of bulktrapping sites decreases with increasing T as theice sample is annealed one might expect that theprecursor concentration would decrease We sug-gest that the observed temperature dependenceis determined by the competition between theavailability of trapping sites and mean interactiondistances of a mobile species such as O and HThat is the O2 yield varies inversely with the den-sity of traps over a range of temperatures In thesimple model above this effect is included indi-rectly as shown using the more detailed equa-tions in the Appendix

EUROPA

Oxygen and peroxide production at Europa G values

The dependence of the yield on the thicknessof the ice sample for highly penetrating ions sug-gests that the O2 yield is roughly related to theamount of energy deposited in the sample There-fore the amount of O2 produced and lost fromice in steady state is often given as a G value thenumber of molecules produced per 100 eV de-posited in Table 2 In applying these G values tothe planetary environment care must be takensince the yields are temperature fluence and ex-citation density dependent They also are affectedby the presence of other proton scavengersknown to be present in Europarsquos ice

The G values for peroxide in Table 2 are thosefor production at low fluence but the G valuesfor O2 are for the production and escape of gas-phase oxygen at saturation Therefore they de-pend on the formation and destruction cross sec-tions in the model above Such G values can beused to describe the steady-state source of the am-

bient O2 atmosphere at Europa but are not help-ful in determining the ability to produce and trapO2 at depth into the icy surface Over the timescales of the experiments some O2 is seen to re-main trapped and is not destroyed by radicalsTypically there are a number of trapping sitesand recent estimates of the activation energy foran O2 molecule trapped in a substitutional site(Hori and Hondoh 2002) suggest that it can betrapped for times comparable to the time re-quired for the conversion of amorphous ice intoIh (Ayotte et al 2001) The stability of O2 inclu-sions has not been studied

To understand the relationship between the Gvalues for O2 in Table 2 we note that fast lightions and electrons lose their energy predomi-nantly by producing ionizations and excitationsThe ionization events produce secondary elec-trons that can result in further ionizations Thenumber of ionization events Ni produced by afast light ion stopping in a water ice is Ni 5EoW where Eo is the initial energy and W is theaverage energy expended per ionization event27 eV in water ice Therefore radiolysis by fastions is often described as the net effect of theshower of electrons set in motion by the incidention In this manner the G values for productionof O2 from ice by energetic light ions and elec-trons can be roughly estimated as the sum of theeffects of all the secondary electrons using theyield Ye for 10ndash100 eV electrons That is if weignore the percolation depth N we can write thenumber of O2 produced as the number of ioniza-tion events times Ye which already has in it thedestruction processes That is G NiYeEo YeW 0006 O2100 eV

This estimate of G is seen to be an order of mag-nitude larger than the keV proton value for O2 inTable 2 but is much smaller than the G value forenergetic Ne1 These differences from the simpleestimate are consistent with the observation thatthe yield in the saturation regime for penetratingions varies nonlinearly with excitation densityThe difference from the keV proton G value islikely due to the inability of O2 to escape fromdepth before destruction On the other hand thelarger G value seen using MeV Ne1 is likely dueto the high density of excitation This can resultin more efficient production of O2 increased pro-duction of damage sites for forming O2 andorincreased transport (percolation) along the tran-siently heated track That is the ability of theformed O2 to percolate to the surface during ion

OXIDANTS AT EUROPA 839

bombardment plays an essential role in deter-mining the measured yields as discussed aboveThis effect needs to be evaluated for a planetarysurface subject to particle fluxes that are many or-ders of magnitude smaller than those used in thelaboratory

In the absence of O2 yields for all of the rele-vant ions and energies an estimate of the totalsupply of O2 can be obtained using the G valuesin Table 2 For the energetic oxygen and sulfurions we use G values for neon and argon ionsThese are lower bounds as the data were obtainedat temperatures much lower than the surface tem-peratures at Europa (80ndash110K) For the fast pen-etrating electrons and protons we use the valuefor 200 keV protons in Table 2 For the heavy ionswe used the Ne1 and Ar1 values For compari-son we also give an estimate using the G valuesobtained from the low energy electron data andthe total ionization as discussed above

The G values are combined in Table 3 with theelectron and ion energy flux at Europa Althoughthere is some deflection of the flow due to in-duced fields the net energy flux to the surface isnot reduced much from these values (Paranicaset al 2002) In fact locally produced pick-up ionswhich are formed from the ejecta and acceleratedto the surface can contribute (Ip 1996) Furtheralthough G values are typically independent ofthe incident angle the loss of O2 is a surfaceprocess and is enhanced with angle That is thenumber of O2 produced within the escape depthincreases with increasing angle to the normal(steeper than the inverse cosine) This enhance-ment is 25ndash4 (Jurac et al 2001) Ignoring suchcorrections the total production rate calculated

(23ndash62 3 109 O2cm2s) is a lower bound to theactual production rate from the icy regions of Eu-roparsquos surface Recent modeling of the O2 atmo-sphere suggest that such a source rate is required(Shematovich and Johnson 2001)

Earlier estimates of the O2 source rate were in-correct because O2 production was assumed to bea fixed fraction of the H2O sputtering yield Atthe ambient temperature an ejected H2O sticks ef-ficiently (Smith and Kay 1997) to the surface ofneighboring grains in the porous regolith reduc-ing the sputtering yield for H2O (Johnson 1989)This reduction does not apply to the O2 fractionof the yield enhancing its relative importance toH2O Here we assume that all of the gas-phaseO2 produced can contribute to the ambient at-mosphere even if it is produced at a grain sur-face that is at some depth into the porous regolith

Finally Europarsquos surface is not pure ice every-where That is much of the trailing hemisphereis a hydrated sulfuric acid containing smallamounts of sulfur and SO2 and the leading hemi-sphere contains low levels of carbon CO2 andcarbonates Sodium and potassium atoms aresputtered from the icy regions (eg Leblanc et al2002) O2 production from surfaces containingthese species has not been measured C and S areboth proton scavengers and the other species canbe sources of O as discussed so that the O2 yieldfrom such regions could be enhanced Howeverthere can also be competing processes that mightreduce these rates In fact recent HST observa-tions suggest that the O2 might not be produceduniformly across Europarsquos surface (McGrath et al2000) The analysis of these observations is pre-liminary and is not simple since the observed

JOHNSON ET AL840

TABLE 3 APPLICATION TO EUROPA

OxygenImplantation Energy flux production

Species 109(cm2 s)a 109 keV(cm2 s)(1) 109(cm2 s)

H (10 keV) 0015(1) 12 0084 (072)bO (10 keV) 0015(1) 18 054cS (10 keV) 0009(1) 30 090cElectrons 62 043 (37)bPlasma (10 keV) (137 amu) 13(2) 10 03c

Total 79 23 (62)UV photons (64 eV) 46 01

aReferences are as follows 1Cooper et al (2001) [similar results in Paranicas et al (2002)] 2Mauk et al (1996)bUsing G for 200 keV protons (upper bounds in brackets are based on the low-energy electrons data as a model for

the secondary electrons produced)cUsing G for 15 MeV Ne1 at 7K a rough lower bound

emissions depend on the local plasma propertiesHowever the efficiency of O2 production in con-taminated ices needs to be studied

Porous regolith O2 interactions on grain surfaces

Although the O2 atmosphere is principally pro-duced in the near surface layers and is eventuallylost to space it also permeates Europarsquos porousregolith The porosity p (void space) of the re-golith is uncertain (Buratti 1995) It has been es-timated to be the most compact surface in theOuter Solar System p 025 on the trailing hemi-sphere and p 079 on the leading hemisphere(Buratti and Veverka 1983) and the most poroussurface p 096 globally (Domingue et al 1991Domingue and Hapke 1992) These large differ-ences are due to the uncertainties in the amountof backscattered light at small phase angles andon the photon absorption properties and grainsizes The more compact surface is likely correctGrain sizes rg of the order of 100 mm have beensuggested (Hansen and McCord 1999 2000Geissler et al 1998) Europarsquos regolith differs sig-nificantly from the lunar regolith which is asand-like surface of individual grains Analysis ofthe thermal inertia of Europarsquos surface (Grundyet al 2001) suggests that the ice grains are sin-tered forming a fairy castle-like structure Sin-tering can occur thermally or due to the incidentradiation In addition ices doped with acids andalkalis in the laboratory can be powdery solidswith the impurities primarily found at grain in-terfaces

The porous regolith affects the above estimatesof the oxygen production in three ways First forisotropic bombardment the mean distance d intothe surface at which an incident ion or photonstrikes a grain surface is d (89)rg(1 2 p) Us-ing the above value for rg then d 02 m if theporosity is very large p 96 Therefore de-sorbed molecules can come from a considerabledepth and will interact with grain surfaces Evenfor much lower porosities the effective sputteringyields can be significantly modified (Johnson1989) As discussed above the yields for ejectedH2O which sticks onto neighboring grains at theambient temperatures are reduced but those forejected O2 are not This enhances the relative im-portance of the O2 yields

Finally the surfaces and interfaces of the sin-tered grains in Europarsquos porous surface are placeswhere low temperature chemistry involving the

ambient oxygen can occur at depths that are wellbelow the nominal penetration depth of the inci-dent radiation The relative importance of suchchemistry is roughly indicated by the ratio of thetotal surface area in the regolith to the area of Eu-roparsquos visible surface Using the porosity andgrain size above and a regolith depth dr of theorder of a meter then this ratio is large 3(1 2p)drrg which is 103ndash2 3 104 for average porosi-ties of 96 and 25 respectively Therefore thegas-phase O2 molecules below the visible surfaceinteract frequently with the grain surfaces In thisway a very large volume of the subsurface couldbe altered chemically even if it is not directly ex-posed to the radiation

From the overlying structures seen in the high-resolution images of Europarsquos surface by theGalileo spacecraft (eg Prockter and Pappalardo2000) it is clear that the youngest surfaces are red-der The reddening in the visible has been attrib-uted to the presence of sulfur in the ice (McEwen1986 Johnson et al 1988 Calvin et al 1995 Carl-son et al 1999b) The amount of reddening in thespectra of the surface material decreases with ageand the surface materials appear to becomebrighter with age This might occur by the slowaccumulation of a frost over-layer or by radioly-sis (Carlson et al 2002 Johnson et al 2003) Ra-diolysis preferentially converts the sulfur in theice to sulfur dioxide or sulfuric acid both ofwhich have flat spectra in the visible In regionswhere the radiation does not directly penetratethe O2 permeating the regolith can oxidize thesulfur at the surfaces of the icy grains and at graininterfaces Similarly the carbon seen in the ice onEuroparsquos leading hemisphere (eg Carlson et al2004 Johnson et al 2003) can be oxidized by ra-diolysis in the near surface layers or by reactionson the grain surfaces at depth in the regolith bythe ambient O2 The meteoroid bombardmentthat produces the regolith also exposes fresh sur-faces for oxidation and can cause the collapse ofpores leading to the trapping of oxygen In thismanner both refractory and volatile oxidants canbe formed and trapped in the subsurface ice forpotential delivery to Europarsquos subsurface ocean

SUMMARY

The presence of a thin oxygen atmosphere andoxidants trapped in Europarsquos surface ice (egJohnson et al 2003) combined with the possible

OXIDANTS AT EUROPA 841

presence of a subsurface ocean (eg Carr et al1998) have made Europa an exciting target forfuture space exploration The molecular oxidantsproduced by radiolysis and photolysis could bespecies needed for carbon-based biochemistry inEuroparsquos putative subsurface ocean Howeverthere has been no model for the formation trap-ping and transport of relevant oxidants

In this paper we have reviewed the space ob-servations and laboratory studies related to thechemistry of formation of oxidants in ice by theincident radiation Because the laboratory dataare incomplete we examined the rate equationsfor formation of oxygen and related species byradiolysis and photolysis We showed how suchequations can be simplified to the analytic modelthat was used for fitting the fluence and temper-ature dependence of the O2 yield produced bylow-energy electrons (Sieger et al 1998) In de-riving the analytic model from the more completerate equations the competition between forma-tion and destruction processes can be studied ascan the role of trapping sites escape from depthand temperature history These equations also al-lowed us to put incident photon electron and iondata on a similar footing for the first time

The model can be used to suggest measure-ments that are needed to describe the chemicalstate of the irradiated ice and to interpret new lab-oratory data We showed that if reactions are fastthe yield data could be consistent with a precur-sor being formed by two events as is the case forperoxide However we favor trapped O as a pre-cursor as initially suggested by Matich et al(1993) for production of excited O2 Such aspecies if present could be detected spectro-scopically as can trapped O2 through bands ac-tivated by lattice perturbations (Cooper et al2003a)

The available oxygen yield data were used tocompare G values for different incident particlesThese were then used to estimate the oxygen pro-duction in the radiation environment at EuropaThe lower bounds were consistent with the pro-duction rates obtained from a recent model of Eu-roparsquos atmosphere The porosity of the regolithwas also considered It not only affects the rela-tive sputtering yields for O2 and H2O but alsoprovides an enormously increased surface areaon which the ambient gas-phase oxygen can re-act with species in the ice That this might be oc-curring is consistent with the change in re-

flectance associated with surface aging Such re-actions could also convert ambient molecularoxygen into other more refractory species that re-main trapped on subduction of Europarsquos crust

Although radiolysis produces a stable chemi-cally altered surface in the laboratory the parti-cle flux to Europarsquos surface is many orders ofmagnitude lower Therefore annealing can be ef-fective at temperatures that are much lower thanthose typically found experimentally Belowabout 120K the annealing times appear to be long(eg Baragiola 2003) so that laboratory data canoften be directly applied to Europarsquos icy surfaceHowever account has to be taken of the syner-gism between the UV and the energetic plasmaas discussed earlier (Johnson and Quickenden1997)

Transient surface melting might initiate lifeprocesses but subduction of oxidants to the pu-tative ocean is likely to be required if biochem-istry is to occur at Europa (Chyba and Hand2001) Although the temperature versus depthinto Europarsquos regolith is not well known belowthe regolith the temperature and pressure appar-ently increase with depth At some depth (ap-proximately kilometers) either very warm ice oran ocean exists (Ruiz and Tejero 2000) Thereforechemically refractory oxidants are likely to be re-quired in order to survive subduction to thewarmer subsurface regions Trapped species ifdelivered downward can become mobile andwill segregate forming inclusions (Johnson andJesser 1997) Indeed peroxide appears to segre-gate in ice (Gurman et al 1967) and these inclu-sions may remain relatively stable until the per-oxide is dissolved in the putative ocean Thevolatiles seen in reflectance such as O2 SO2 andCO2 will also segregate But at the higher tem-peratures they could also diffuse through thebulk react or percolate into the regolith atmo-sphere Therefore the observation of the trappedvolatiles in Europarsquos surface ice is of interest pri-marily because it suggests that oxidants are pre-sent The formation of oxygen-rich molecules fol-lows the preferential loss of hydrogen Howeverthe surface is in a local steady state with im-planted H O and S from the plasma and the lossto space Since the plasma flux to the surface canbe variable this balance can also vary Indeedtemporal changes in reflectance have been re-ported (Domingue and Hendrix 2004)

Based on the models for formation of O2 ex-

JOHNSON ET AL842

amined in this paper additional experiments areneeded to determine the chemical state of irra-diated ice at low fluences Of immediate inter-est is the identification of stable molecular pre-cursors required for the formation of O2 Basedon the available laboratory data individual per-oxide molecules trapped in the ice do not appearto be the principal precursors but trapped O(Matich et al 1993) remains a viable candidateAlthough the data suggest that the local hydro-gen-bonding network is important in the pro-duction of O2 the effect of the formation tem-perature and radiation damage on the structureneed to be measured It is also important to de-scribe quantitatively the relationship betweenH2 loss and the formation of O2 and H2O2 overa range of temperatures and fluences Finallyone would like to determine the steady-statedensities of biochemically interesting moleculesformed by radiolysis in an ice containing sulfurand carbon at Europarsquos surface temperaturesand then determine their stability with increas-ing temperature and pressure Such experimentsstill need to be carried out Here we have pre-sented a model that brings together the variouslaboratory data and now can be used to considerthe formation trapping and transport of relevantoxidants at Europa

APPENDIX

Introduction

In this Appendix we examine rate equationsthat describe various aspects of the radiolysis orphotolysis of ice These equations need to be un-derstood if one is to be able to calculate the for-mation trapping and transport of oxidants Theemphasis is on the formation of molecular oxy-gen We integrate these equations over the escapedepth of O2 to give the simplified rate equationsdiscussed in the text in which number densitiesare replaced by column densities (eg Eqs 1 and2) In this way we can examine the relationshipbetween the rate constants and the effective pre-cursor formation and destruction cross sectionsextracted from data We first consider a precur-sor such as H2O2 that is formed from two exci-tation events This bears on the observed linear-ity of the oxygen yield versus fluence We thenconsider the equations for trapped versus a tran-

siently mobile species We use a reaction betweena mobile and trapped O as an example althoughO 1 OH can also give oxygen Finally we ex-amine the possible dependence of the O2 forma-tion process on the density of traps (defectsvoids and grain boundaries) as a function of tem-perature The results are discussed in the text

For clarity all possible radiolytic species arenot included in the discussion below Thereforepotentially important products such as HO2 andthe charged species (eg H3O1 the trapped elec-tron etc) are missing In each section we attemptto make a point related to production of O2 andother oxidants by using a subset of the productsrather than attempt a complete description withmany unknown rate constants

Fast versus slow processes

To describe the O2 production rate from a pre-cursor such as H2O2 we first use a simple but in-structive set of rate equations to describe the for-mation of H2O2 We then relate these to thesimplified rate equation for the precursor in thetext Eq 1 and show how the postulated precur-sor formation cross section sp is related to theH2O dissociation cross section sd by the reactionrates

Using the reactions

H2O 1 radiation R H 1 OH (sd)

H 1 H R H2 (k2H)

H 1 OH R H2O (kHOH)

OH 1 OH R H2O2 (k2OH)

the reaction pathways are described by

H2

Hqk2H

1 OHH2O 1 radiation mdashRsd H

OHH2O2

OH Rk2OH

kHOH

HH2O

OXIDANTS AT EUROPA 843

The corresponding rate equations in an irradiatedvolume are

dnOHdt 5 sd f n 2 2 k2OH nOH nOH

2 kHOH nH nOH (A1a)

dnHdt 5 sd f n 2 2 k2H nH nH

2 kHOH nH nOH (A2a)

dnH2O2dt 5 k2OH nOH nOH

2 destruction processes (A3a)

Here sd is the dissociation cross section f is theradiation flux and n is the molecular numberdensity of ice with nOH and nH the density of dis-sociation products OH and H respectively Therate constants k2OH k2H and kHOH describe thereactions above In this model the H2 formed isassumed to escape for temperatures above about20K We do not distinguish between trappedspecies and mobile species although in fact thatdistinction is important as discussed in the nextsection That is any OH produced at low tem-perature by one ion would likely be trapped priorto interacting with an OH produced by a subse-quent ion At higher temperatures this can breakdown as discussed in the text We first relate theabove set of equations to Eq 1 in the text and thenconsider precursor destruction processes

Equation 1 in the text is obtained by solvingthese equations together and then by integratingover a depth Dx Here Dx is the smaller of the pen-etration depth of the particle and the O2 escape(percolation) depth Assuming the molecular den-sity of ice n is constant with depth Dx n 5 N inEq 1 Comparing Eq 1 with Eq A3a the precur-sor formation rate sp f N is equivalent to [k2OH

nOH nOH] Dx Assuming that the processes in EqsA1a and A2a are fast then the steady state is read-ily obtained In addition if k2H k2OH there ispreferential loss of H2 at short times as suggestedby the laboratory data If steady state is reachedthe production of H2 is equal to that of H2O2 us-ing only the equations above Further one obtains

[k2OH nOH nOH] Dx R [fp sd] f N (A4)

with

fp 5 05[1 1 05 kHOH(k2H k2OH)05] (A5a)

Therefore the precursor formation cross sectionsp in Eq 1 is a fraction fp of the total dissocia-

tion cross sd Since laboratory data suggest theH2O2 fraction in an irradiated ice is small thenkHOH (k2OH k2H)05 which is likely givingfp R [(k2OH k2H)05kHOH]

Trapped versus mobile species

To examine the fluence dependence discussedin the text we note that trapped OH and freshlyproduced transiently mobile OH can be treatedas different species (eg Matich et al 1993)Therefore assuming no trapping of H the OHrate equation above would be replaced by twoequations one for the trapped species nOHt andthe other for the freshly produced mobile speciesnOH Assuming the reaction of two mobile OHunits is unlikely and that peroxide is formed fromthe interaction of a trapped OH with a subse-quently produced OH an expanded set of equa-tions is obtained

OH 1 trap R OHt (kOHt)

OH 1 OHt R H2O2 (kOHOHt)

H 1 OHt R H2O (kHOHt)

then the reaction pathways are described by

qH2

H2O 1 radiation mdashRsdH

qk2H

1 OHH

trap OH

OH RkOHt

OHt kOHOHtH2O2

kHOHkHOHt

H H2O HH2O

The resulting set of equations is

dnOHdt 5 sd f n 2 kHOH nH nOH

2 kOHt nOH nt 2 kOHOHt nOH nOHt (A1b)

dnOHtdt 5 kOHt nOH nt 2 kOHOHt nOH nOHt

2 kHOHt nH nOHt (A1b9)

dnHdt 5 sd f n 2 2 k2H nH nH 2

kHOH nH nOH 2 kHOHt nH nOHt (A2b)

JOHNSON ET AL844

dnH2O2dt 5 kOHOHt nOH nOHt

2 destruction processes (A3b)

where nt is the trap density in the irradiated iceAt low T we can assume nt is large so thatdntdt 0 simplifying the set of equations If theprecursor of interest is trapped O as discussed inthe text then a similar set of equations can be con-structed for O and Ot as in the example to followOf course a full description over a broad rangeof temperatures would require the considerationof the both trapped and mobile H O and OHthe rate of change of trap density and the vari-ous destruction processes

The above equations should be solved simul-taneously but for the incident particle fluxes typ-ically used in these studies the diffusion of mo-bile species to traps is relatively fast This is trueeven at low temperatures if there is a significantdensity of trapping sites This is usually the casefor samples formed at low T Therefore in a lowtemperature solid at low dose rates there is not asignificant steady-state background of diffusingOH There are trapped or freshly produced OHunits which rapidly react Averaging over thebombardment rate this can be approximated bywriting dnOHdt 0 and then separating nOH outin Eq A1b Assuming also that H diffuses rapidlyand either reacts or forms H2 we also writednHdt 0 on the average Therefore betweenimpacts only species trapped at defects pores orgrain surfaces are assumed to be present as sug-gested by the lack of a dependence on the inci-dent beam flux In this limit the rate of change oftrapped OH roughly simplifies to

dnOHtdt 5 2 k2H nH nH 2 2 kOHOHt nOH nOHt

It is seen that the production of peroxide dependson the loss of H by formation of H2 as discussedand at steady state the production rate for per-oxide equals the rate of formation of H2 in theseequations

Treating the mobile and trapped species sepa-rately we can further simplify the rate equationsand obtain the precursor model used in the textThat is integrating over the relevant depth Dxthe densities are replaced by column densitiesThe rate constants ki are then related to effectivecross sections si In this way cross sections aregiven as an interaction length for each freshlyproduced mobile radical [ki nj] R [si f] The

trapped OH column density NOHt and peroxideproduction rate can be estimated from ldquolinearrdquorate equations

dNOHtdt lt [sd f] N 2 2 [s2OH f] NOHt

2 [sHOH f] NOHt (A6a)

dNH2O2dt lt [s2OH f] NOHt 2 s NH2O2 (A6b)

Here sd is again the dissociation cross sectionand the destruction processes in Eq A3b are con-tained in s The second and third terms on theright in Eq A6a are the loss of OH due to the pro-duction of H2O2 and H2O In the integration ofEq A1b9 [2 kOH OHt nOH] becomes [2 s2OH f] and[kHOHt nOH] becomes [sHOH f] Therefore thedensity and the interaction length of mobile H arecontained in sHOH If H2O2 is indeed the pre-cursor then in Eq 2 in the text [sp f N] is be-comes [s2OH f NOHt] from Eq A6b Accountingfor precursor destruction which can give backtwo OH and treating H2 can both be included inthe integrated model via the si values

The simplified set of equations above allows usto examine the relative time (fluence) scales forprecursor formation (in this case H2O2) SolvingEqs A6a and A6b the column density of OH andthe concentration of precursors versus fluence cpcan be obtained Using the column of H2O2 pro-duced as the precursor column Np in Eq 2 thenO2 yield can be obtained The results can be com-pared with the results from Eqs1 and 2 in thetext The average precursor (H2O2) concentrationover the column depth of water molecules N is

cp 5 fp (sds)[1 2 exp(2s F)] 2 (ssr)

[1 2 exp(2sr F)] (A7)

with

fp 5 2s2OH(sr 2 s) (A5b)

In these equations sr 5 (2s2OH 1 sHOH) andssr 1

It is seen that the precursor concentration withtime cp depends on two rates the reaction ratesthat determine sr and the destruction rates in sAs in Eq 4a the yield can be written as

Y 5 sO2 cp N (A8)

If sr and s are comparable then the yield at small

OXIDANTS AT EUROPA 845

fluence is quadratic in F However if sr sthen the linear dependence on fluence is obtainedwith a small shift in the fluence as shown in Eq4c in the text Further the steady-state yield isnow

Y` (T) 5 sO2 [fp sd] Ns (A9)

That is sp R [fp sd] as in Eq A4 and the tem-perature dependence for the incident electrondata must be contained in [sO2 fp] or in the ratio[sO2 s2OH]sHOH]

For comparison with their precursor modelSieger et al (1998) examined the possibility of ob-taining the O2 precursor by considering reactionsfor a density of diffusing dissociation products asin Eqs A1 and A2 rather than a trapped productinteracting with a freshly formed dissociationproduct Based on the above the lowest fluence atwhich O2 is first detected experimentally sets alimit on the time scale for mobilendashmobile reactions

Effect of trap density

Equations A1b A1b9 A2b and A3b can be ap-proximately solved assuming that the dnOHdtdnHdt and dntdt are very small on the averageas discussed above In order to examine the roleof the trap density nt which can depend on thetemperature and method of formation and on theradiation history of the ice we use the slightlysimpler precursor model In this model the O pro-duced directly by dissociation traps forming aprecursor O2 is then formed when a freshly pro-duced O or other species reacts with a trappedoxygen atom Ot

H2O 1 radiation R H2 1 O (sd)

O 1 trap R Ot (kOt)

O 1 Ot R O2 (kOOt)

The reaction pathways for the oxygen species are

H2O 1 radiation mdashRsd9

trap OO2

kOtOt kOOt

H2 1 OOHt

kOR H(kHO t)

ROR

We assume for simplicity both H2 and O2 escapefrom depth Dx after formation We also allow mo-bile H to react with Ot (kHOt) removing trappedO and both the mobile H and O to be removedby other reactants R (eg OH or contaminants)This leads to a simplified set of equations not di-rectly involving OH or peroxide

dnOdt 5 sd9 f n 2 kOt nO nt 2 kOR nO nR

2 kOOt nO nOt (A1c)

dnOtdt 5 kOt nt nO 2 kOOt nO nOt

2 kHOt nH nOt (A1c9)

dnHdt 5 sd f n 2 kHOt nH nOt

2 kHR nH nR (A2c)

dnO2dt 5 kOOt nO nOt (A3c)

These equations of course act in parallel with theequations above for OH and mobile OH may re-act with trapped O to form O2 Here we focusonly on mobile O

Since the diffusion of H and O are fast and thedensity of traps is large the nH and nt averagedover the bombardment rate can be taken veryroughly as time independent Further if we as-sume the number of trapped O nOt is small com-pared with the number of sites at which H canreact then nH [sd f nkHR nR] For a low ra-diation flux f nH is small on the average as as-sumed Since the O2 yield YO2 is equal to[(dnO2dt) Dxf] with Dx the escape depth thenthe steady-state yield Y` can be obtained ana-lytically

[Y`][sd9 N] 5 1 1 (q 2 1) 2 [(q 2 1)2

1 8 q d]054 d (A10)

q 5 [kOt (nt 1 nR)] (kHOt nH)(sd9 f n) kOOt

5(sdsd9) [kOt (nt 1 nR)] kHOt[kOOt kHR nR]

d 5 05 [1 1 nt(nt 1 nR)]

We note that for nt large q can be large Assum-ing that q 1 and nt nR then Eq A10 be-comes

Y` R [sd9 N]q 5 (sd9 N) (sd9sd)

[kOOt kHR nR][kHOt kOt nt] (A5c)

JOHNSON ET AL846

Comparing with Y` in the simple precursormodel in Eq 4b then s R [(kHOt nH)f] R [sd nkHOt][kHR nR] In addition the product [sO2sp] R sd9 sd9 [kOOt n][kOt nt] That is the dis-sociation cross sections are modified by reactionrates

Quite remarkably it is seen in Eq A5c that thesteady-state yield Y` can depend inversely onthe density of traps nt in the temperature andflux regimes for which the approximations arerelevant That is if there are competing destruc-tion process for trapped O then at high trap den-sity a mobile O must find a trapped O before itbecomes trapped or before mobile H destroys theOt In this way the increasing O2 yield with in-creasing T can be related to the change in the trap-ping density nt with the formation temperatureof the ice sample As the temperature increasesand the density of traps in the bulk decreases thesurface and grain interfaces become the only trap-ping sites

ACKNOWLEDGMENTS

REJ acknowledges support by a Gledden Fel-lowship in Chemistry School of Biomedical andChemical Sciences at The University of WesternAustralia and by grants from the NSF Astron-omy Program and NASArsquos Planetary Geologyand Geophysics Program PDC acknowledgessupport by an Australian Postgraduate Award

ABBREVIATIONS

HST Hubble Space Telescope MJQ MatichndashJohnsonndashQuickenden NIMS Near Infrared Map-ping Spectrometer UV ultraviolet

REFERENCES

Ayotte P Smith RS Stevenson KP Dohnarsquolek ZKimmel GA and Kay BD (2001) Effect of porosityon the adsorption desorption trapping and release ofvolatile gases by amorphous solid water J Geophys Res106(E12) 33387ndash33392

Bahr DA Famaacute M and Baragiola RA (2001) Radiol-ysis of water ice in the outer solar system sputteringand trapping of radiation products J Geophys Res 10633285ndash33290

Baragiola RA (2003) Microporous amorphous water icefilms and astronomical implications In Water in Con-

fining Geometries edited by JP Devlin and B Buch El-sevier New York in press

Baragiola RA and Bahr DA (1998) Laboratory studiesof the optical properties and stability of oxygen onGanymede J Geophys Res 103 25865ndash25872

Baragiola RA Atteberry CL and Bahr DA (1999) Ori-gin of solid oxygen on Ganymede Reply to commentby RE Johnson on ldquoLaboratory studies of the opticalproperties and stability of oxygen on Ganymederdquo JGeophys Res E104 14183ndash14187

Baragiola RA Vidal RA Svendsen W Schou J ShiM Bahr DA and Atteberrry CL (2003) Sputteringof water ice Nucl Instrum Methods Phys Res B 209294ndash303

Bar-Nun A Herman G Laufer D and Rappaport ML(1985) Trapping and release of gases by water ice andimplications for icy bodies Icarus 63 317ndash332

Baverstock KF and Burns WG (1976) Primary pro-duction of oxygen from irradiated water as an expla-nation for decreased radiobiological oxygen enhance-ment at high LET Nature 260 316ndash318

Benit J and Brown WL (1990) Electronic sputtering ofoxygen and water molecules from thin films of waterice bombarded by MeV Ne1 ions Nucl Instrum Meth-ods Phys Rev B 46 448ndash451

Benit J Bibring J-P della-Negra S LeBeyec YMendenhall M Rocard F and Standing K (1987)Erosion of ices by ion irradiation Nucl Instrum Meth-ods Phys Rev B 1920 838ndash842

Brown WL Augustyniak WM Lanzerotti LJ John-son RE and Evatt R (1980) Linear and nonlinearprocesses in the erosion of H2O ice by fast light ionsPhys Rev Lett 43 1632ndash1635

Brown WL Augustyniak W M Simmons E Mar-cantonio KJ Lanzerotti LJ Johnson RE BoringJW Reimann CT Foti G and Pirronello V (1982)Erosion and molecular formation in condensed gasfilms by electronic energy loss of fast ions Nucl In-strum Methods B 1 307ndash314

Brown WL Augustyniak WM Marcantonio KJ Sim-mons EN Boring JW Johnson RE and ReimannCT (1984) Electronic sputtering of low-temperaturemolecular solids Nucl Instrum Methods B 1 307

Buratti B (1995) Photometry and surface structure of theicy Galilean satellites J Geophys Res 100(E9) 19061ndash19066

Buratti B and Veverka J (1983) Voyager photometry ofEuropa Icarus 55 93ndash110

Calvin WM and Spencer JR (1997) Latitudinal distri-bution of O2 on Ganymede observations with the Hub-ble Space Telescope Icarus 130 505ndash516

Calvin WM Clark RN Brown RH and Spencer JA(1995) Spectra of the icy Galilean satellites from 02 to5 mm a compilation new observations and a recentsummary J Geophys Res 100 19041ndash19048

Calvin WM Johnson RE and Spencer JA (1996) O2

on Ganymede spectral characteristics and plasma for-mation mechanisms Geophys Res Lett 23 673ndash676

Calvin WM Aninich VG and Brown RH (2004) Oxy-gen temperature and phase effects Icarus (in press)

OXIDANTS AT EUROPA 847

At temperatures above 40K it is seen in Fig2 that the D2 and O2 yields exhibit the same tem-perature dependence suggesting a correlationbetween D2 and O2 loss H2 (D2) can be formedby reactions between diffusing hydrogen fromdissociated water or directly [ie H2O R H2 1O (eg Kimmel and Orlando 1995) H2O R (H1 OH)cage R H2 1 O (eg Matich et al 1993)H3O1 1 e R H2 1 OH (eg Williams et al1996)] Subtraction of the nearly constant D2 yieldin Fig 2a gives the Arrhenius plot in the insetSince molecular hydrogen readily escapes atthese temperature it is the H2 (D2) formationprocess that has an activation energy of 003eVmolecule The fact that slopes of the yields forH2 and O2 above 40K are similar suggests thatH2 (D2) formation and loss is the rate-limitingstep in determining the production and loss of O2at these temperatures

Although the data at low T in Fig 2a have con-siderable uncertainties below about 20ndash40K theD2 production is roughly independent of tem-perature but the O2 yield continues to decreasewith decreasing temperature Radiolytically pro-duced D2 is mobile and can escape above15ndash20K (Watanabe et al 2000) but O2 is im-mobile at such temperatures and the incidentHe1 ions are much less efficient at causing per-colation than the incident Ne1 discussed earlier

The production of H2 (D2) from ice has alsobeen studied using low-energy electrons and UVphotons at very low temperatures In the formerstudies absolute yields were not obtained butthere appeared to be a threshold at 63 eV (Fig4b) This is roughly the threshold for excitationof a surface state (Khusnatdinov and Petrenko1992) and the threshold for peroxide productionis close to this value The efficiency for H2 (D2)production increases around 11 eV with a sec-ond steep rise at 17 eV (Kimmel et al 1994) TheO2 and D2 yields versus electron energy in Fig 4exhibit some similarity in the threshold regionalthough the uncertainties are large and the sam-ple temperatures different Measuring the H2 (D2)and the O2 ejecta simultaneously over a range oftemperatures as in the Ne1 experiments wouldbe instructive

Using 98 eV photons at low T (12K) Watanabeet al (2000) measured the D2 ejected and trappedin ice They obtained a cross section for produc-tion of D2 (Table 2) at low fluences that is10ndash20 of the total absorption cross sectionwith significant uncertainties They also found

that the equilibrium damage fraction in the sam-ple was 10ndash20 of the D2O molecules but theequilibrium production of D2 was about an orderof magnitude smaller 1ndash2 at their tempera-tures They suggested that the D2 is formed atgrain interfaces and cracks and on the surfaces ofpores In a mixed ice D2O 1 CO exposed to UVphotolysis D2 and CO2 formation are correlated(Watanabe and Kouchi 2002) The D2O dissoci-ates to produce D2 plus O and this transientlymobile O reacts with CO to produce CO2

Precursor state and peroxide

In order to produce O2 from low temperatureice it appears that the ice sample must lose hy-drogen and form trapped oxygen-rich reactivespecies The efficiency of production also appearsto be affected by the level of defects and damagein the ice consistent with the suggestion that thehydrogen-bonding network is important in theproduction of O2 via availability of trapping sitesnew electronic relaxation pathways or the natureof the bonding of precursors Here we briefly ex-amine the possible role of the suggested precur-sor molecules H2O2 and HO2 We note that H2O2might not readily dissolve in ice possibly ac-counting for its low abundance in some experi-ments It appears to be heterogeneously distrib-uted in the form of microcrystals (Gurman et al1967) If H2O2 does act as a precursor the for-mation of aggregates could affect the dimer ab-sorption bands (Cooper et al 2003a) and affectthe production and stability of O2 (Cooper et al2003b)

The results of Gerakines et al (1996) in whichice at 10K is irradiated by UV photons confirmthat H2O2 forms in ice from two OH after a build-up of trapped OH They monitored the band as-sociated with H2O-HO (Langford et al 2000) inwhich the HO is bound to water with 024 eVTherefore they might only be monitoring a sub-set of the trapped OH seen in earlier experiments(eg Taub and Eiben 1968) The H2O2 concen-tration in ice reaches a steady-state level at flu-ences of about 1018 photonscm2 (Gerakines et al1996) As seen in the model section this gives anestimate of the H2O2 photodestruction cross sec-tion of the order of 10218 cm2 Similarly HO2 be-gins to be seen in absorption at about a few times1017cm2 shortly after they begin to detect H2O2The absolute yields were not measured

Westley et al (1995) detected H2 and O2 pro-

JOHNSON ET AL832

duced by Lyman-a (98 eV) radiation of ice butfound no fluence dependence for fluences greaterthan 1017cm2 They also saw small amounts ofH2O2 by programmed thermal desorption of thesample but made no quantitative evaluationComparing their lowest fluence for the produc-tion of O2 with the result of Gerakines et al (1996)for peroxide suggests that H2O2 and HO2 areprobably not important precursors of O2 In UVphotolysis of H2O2 using 64 eV photons the pro-duction of O2 was not observed (Vaghjiani et al1992) and the excitation of an irradiated icewhich presumably contained some peroxide by37 eV photons did not produce O2 (Baragiola etal 2003) Direct excitation of matrix-isolatedH2O2 at low temperature also did not yield O2but the production of a mobile O that reacted withtrapped O (H2O-O) did give O2 (Khriachtchev etal 2000)

Irradiating with 200 keV protons at tempera-tures of 40ndash129K Bahr et al (2001) found onlyabout 0001ndash002 H2O2 using programmed ther-mal desorption as compared with earlier experi-ments that suggested larger values (eg Table 2)Using infrared absorption Moore and Hudson(2000) found a G value of 01 for incident MeVprotons at low temperature but the signal in theobserved band became negligible at 80K sug-gesting that H2O2 was not being formed Theyalso found that adding O2 significantly increasedthe production efficiency of H2O2 That is thatthe amount of peroxide was affected by theamount of trapped oxygen Based on the matrixisolation studies (Khriachtchev et al 2000) dis-sociated mobile O might trap as H2O-O whichon excitation converts to H2O2 Therefore even ifH2O2 is not the precursor the peroxide andtrapped oxygen concentrations will be related

Although molecular oxygen was not a productof photolysis of matrix isolated H2O2 it was aproduct of photolysis at 266 nm (466 eV) of per-oxide dimers in an argon matrix 2 H2O2 R 2 H2O1 O2 (Engdahl and Nelander 2002) This de-composition was thought to be due to the con-siderable distortion of the bonds in the dimerTherefore the configuration or distortion by thelattice of peroxide or any other precursor mole-cule in the ice may be critical (Sieger et al 1998)In addition more recent experiments show thatG values for the production of H2O2 by 30 keVincident ions are significant as are the steady-state ratios of H2O2H2O (Gomis et al 2004) Ra-tios of the order of 02ndash4 were found at 16K and

77K in the irradiated regions with incident H giv-ing the smallest fraction and incident O thelargest (Table 2) At these concentrations perox-ide molecules might segregate forming inclu-sions (Gurman et al 1967 Cooper et al 2003b)Since the amount of trapped peroxide also com-petes with the amount of trapped O studies areneeded to describe the chemical state of the solidproduced by radiolysis This will be critical in de-scribing the fate of the irradiation products at Eu-ropa In particular the chemical composition fa-vorable for producing O2 needs to be determinedas does the importance of trapped O H2O2 HO2or peroxide dimers as precursors

Other species

In the model below we will refer to atomic andmolecular species trapped in ice Such species areoften discussed in the literature on radiolysis andphotolysis (for summaries see Matich et al 1993Johnson and Quickenden 1997) Trapped OH OH2O2 and HO2 have been discussed andor mea-sured by a variety of techniques (absorption andluminescence spectroscopy electron spin reso-nance programmed thermal desorption etc)Care must be taken when using these data as re-sults for ice samples formed below 160K orthose exposed to penetrating ion beams likelyhave numerous defects voids and grain inter-faces so that a variety of trapping sites and trapdepths exist Therefore when a particular bandfor a trapped species (eg OH) disappears withincreasing temperature it could be due to achange in the nature of the trapping as well as toloss of the species by reaction

Although H is discussed below we do not dis-tinguish it in these simple models from H1

trapped as the hydronium ion H3O1 whichwould mean that there are corresponding nega-tive ions present Therefore below we refer to thevarious species as if they are neutral althoughthis might not be the case Although trapped elec-trons are detected only over a narrow range oftemperatures (100ndash120K) negative ions must beimportant over a broad range of temperature

We also note that SO2 and a sulfate and CO2and a carbonate appear to be trapped species inEuroparsquos icy surface Under radiolysis each ofthese can dissociate directly giving O2 Howeverwith the exception of the sulfate their densitiesare less than 1 so they are not relevant sourcesof O2 However their presence is indicative of the

OXIDANTS AT EUROPA 833

preference for forming oxygen-rich molecules inEuroparsquos ice and can affect the efficiency of pro-duction of O2 and H2O2

MODELS FOR FORMATION OF O2

Introduction

Based on the data described above the yieldsfor incident electrons and ions are found to benearly linear in fluence at the lowest fluences forwhich O2 is detected In addition the relevant al-terations of the ice samples by radiolysis appearto be stable and independent of the beam fluxRemarkably this is the case both for low-energyelectrons that make only a single excitation foreach impact and fast ions that make a density ofexcitations along their path through the solidThis suggests that O2 is produced and trapped inthe initial events and then caused to escape fromthe solid by the subsequent radiation or that astable chemical alteration of the ice is required be-fore the radiation can produce and eject O2

Because H produced by the dissociation of wa-ter molecules can attack trapped O2 and any pre-cursor oxidants the loss of H is important Thisoccurs by the production and loss of hydrogenmolecules as discussed leaving a surface that isslightly rich in O relative to H The O2 produc-tion mechanism could be direct excitation of aparticular precursor species or it might not bespecific That is a new OH or O produced by theincident radiation could react with the H-depleted solid producing O2 This would occurmore efficiently as the ratio of O to H in the nearsurface region increased

In their early work WL Brown and colleaguesassumed that processes similar to those used todescribe radiolysis of liquid and gaseous H2Ocould be used to describe the radiolysis in ice Thelow activation energy that they found was as-sumed to be associated with diffusion of somemobile radical species in ice such as H or O orpossibly diffusion of the O2 molecule itselfBased on their studies of ejection from isotopi-cally labeled D2O samples they concluded thatthe O2 is formed in the track of a single ion andthen ejected by subsequent ions (eg Benit andBrown 1990) More recently trapped specieshave been suggested as precursors These are ei-ther subsequently excited by the incident radia-tion or attacked by newly produced mobile

species We consider a simple model that incor-porates all of these processes and can be used toorganize and interpret the experimental data Wealso discuss its relationship to more completemodels In this discussion much of which is in theAppendix we consider the standard radiationproducts that have been often described How-ever in a solid the excited electron cloud pro-duced by absorption can overlap a number of lat-tice sites Therefore their relaxation pathwaysandor product species could be very differentfrom those that have been traditionally consideredin the literature on radiolysis and photolysis

Precursor model

The column density of a precursor species Npformed in ice by incident radiation can be calcu-lated using a simple model In this model cross sec-tions are used for the production sp and destruc-tion s of the precursor O2 is then formed byexcitation of the precursor or by reaction of a newlyformed radical with the precursor This is also de-scribed using a cross section sO2 For incident elec-trons having a small penetration depth the O2 pro-duced is assumed to rapidly diffuse to the surfaceand escape (Sieger et al 1998) More generally wedefine N as the column density of H2O from whichthe incident radiation can produce and eject O2 Apair of rate equations gives both Np and the pro-duction rate of oxygen dNO2dt That is

H2O 1 radiation R precursor

precursor 1 radiation R O2

giving

dNpdt 5 sp f N 2 s f Np (1)

dNO2dt 5 sO2 f Np (2)

In Eq 1 f is the incident flux and the total pre-cursor destruction cross section can be written ass 5 sp9 1 sO2 with sp9 accounting for precursordestruction processes other than the formation ofO2 Solving Eqs 1 and 2 and assuming N is nearlyconstant (10 level as discussed) the fractionof precursors cp in the affected column after timet is obtained (Appendix)

cp 5 NpN 5 [sps] [1 2 exp(2 s F)] (3)

Here F 5 [ft] is the fluence (flux 3 time) which

JOHNSON ET AL834

is the number of particles (photons electrons orions) incident on the surface per unit area Usingcp in Eq 2 the O2 yield Y at any fluence can bewritten in terms of the yield at steady state Y`

Y 5 [dNO2dt]f 5 sO2 cp N 5 Y`

[1 2 exp(2sF)] (4a)

Y` 5 [sp sO2 s] N (4b)

If we also account for the reduction in the den-sity of water molecules in the initial column Nthese forms change only slightly The size of thetwo parameters Y` and s can be obtained by fit-ting the yield versus fluence data such as that inFig 3

For the low-energy electron data Sieger et al(1998) showed that the ratio [YY`] for a givenincident electron energy is independent of tem-perature over a narrow range of temperatureTherefore the destruction cross section s in Eq4a is roughly independent of T and can be esti-mated from the [YY`] curve Sieger et al (1998)found a precursor destruction cross section s lt08ndash30 3 10216 cm2 for incident electron energiesfrom 30 to 100 eV This is of the order of the elec-tron impact ionization cross section for H2O

Based on the above model the lack of an ob-served fluence dependence for fluences down to1017 hncm2 in the Lyman-a irradiation of ice(Westley et al 1995) can be examined The low-est fluence used implies that the precursor de-struction cross section for Lyman-a is of the or-der of the Lyman-a absorption cross section in ice(09 3 10217 cm2)

Sieger et al (1998) also assumed that sp9 5 0This would mean that the precursor is only de-stroyed by production of O2 (ie s 5 sO2) Thatis unlikely since suggested precursors can be de-stroyed by reaction with H and other radicalsTheir assumption would also mean that the crosssection for production of O2 from a precursorsO2 is large This would be surprising and ap-pear to contradict some of the data discussedabove Instead we assume that s is not equal tosO2 and the total destruction cross section s isof the order of the ionization or absorption crosssection

Having obtained s the measured temperaturedependence for production of O2 is containedonly in Y` Over a narrow range of T one canwrite Y`(T) lt Y`9 exp(2EakT) and obtain the es-

timate of the activation energies in Table 2 Ifsp9 THORN 0 as discussed then sp and sO22 in Eq 4bcannot be obtained separately from the data forY` unless an independent measure of cp is ob-tained Such a measurement is critical for de-scribing the chemical state of the irradiated sur-faces

The observed temperature dependence of Y` iscontained in sp sO2 or N Here N is the ldquodepthrdquofrom which newly formed O2 can reach the sur-face and escape given as a column density of H2OThe physical depth is obtained by dividing thecolumn density N by the molecular number den-sity of the ice Assuming that in the low-energyelectron experiments N is independent of tem-perature and is about three monolayers (Sieger etal 1998) the measured Y` gives [sO2sps]05ndash20 3 10218 cm2 over the electron impactenergy range At 110ndash120K this ratio was foundto be proportional to (E 2 Eo) for E Eo 10 eVagain assuming fixed N (Orlando et al 1998 Or-lando and Sieger 2003)

In the following section we first consider a pre-cursor like peroxide that typically forms fromtwo dissociation events We use this to show un-der what circumstances such a model does notviolate the observed linearity of the O2 yield datawith fluence Following that we relate the sim-ple precursor model above to the case wheretrapped O is the precursor We then use themodel for trapped O to examine the effect of thetrap density on the production of O2 The densityof traps (defects) is determined by the sample for-mation temperature and the subsequent radiationdamage Finally we consider the escape depthN for penetrating radiation within the precursormodel These discussions are supported by moredetailed rate equations in the Appendix

Precursor trapped peroxide

The proposed precursor peroxide is typicallyformed from products of two dissociation eventsie 2 H2O R 2 H 1 2OH R H2 1 H2O2 This wasconfirmed in low temperature ice (Gerakines etal 1996) It is also possible that a dissociation pro-ducing a trapped O followed by an excitationevent can produce peroxide 2 (H2O 1 hn) R H2

1 H2O-O 1 hn R H2O2 1 H2 (eg Khriachtchevet al 2000) The O2 yield versus fluence for inci-dent electrons is reasonably well fit by the yieldexpression above for fluences greater than1014cm2 Therefore it has been suggested that

OXIDANTS AT EUROPA 835

any precursor species must be formed in a singleevent Using a model in the Appendix in whichthe precursor is formed by two events we showthis is not necessarily the case Precursor forma-tion if it is fast could involve two dissociationevents and give the measured fluence depen-dence Although we consider peroxide here theseconclusions apply to other possible precursors

In the example in the Appendix H2O is disso-ciated to H 1 OH according to the cross sectionsd The reactions of two OH units to form H2O2(rate constant k2OH) and 2 H to form H2 (k2H)compete with the recombination of H and OH toH2O (kHOH) If k2H is much greater than k2OH then H2 loss dominates at low fluences as ex-pected (Appendix) Steady state requires not sur-prisingly that the rate for forming H2O2 isroughly equal to the rate for forming H2 Finallyit is seen in the Appendix that the low H2O2 con-centrations suggested by experiment imply thatkHOH (k2H k2OH)05 Such sizes are reasonable

The set of rate equations in the Appendix thatuse number densities of species is integrated overdepth in order to obtain the forms used in Eqs 1and 2 which are based on column densities Inthis way the effective cross sections are related tothe rate constants for the interaction of freshlyproduced mobile species with previously pro-duced trapped species We show in the Appen-dix that the O2 yield versus fluence can be writ-ten in a form similar to that in Eq 4a but with asmall offset due to the two-step precursor for-mation process

Y lt Y`(T) [1 2 exp(2sF) 2 (ssr)]

for F sr21 (4c)

Here s is again the precursor destruction crosssection but sr 5 (2s2OH 1 sHOH) is the effectivecross section for loss of OH by formation of H2O2

and H2O The corresponding expression for Y`(T)is given in the Appendix

The form for the yield versus fluence in Eq 4cis consistent with the low-energy electron mea-surements if (ssr) 1 and if sr F21 lt10214 cm2 Such sizes are indeed possible Whatis required is that at the relevant temperatures(50ndash110K) the interaction length associated withsr is more than a few molecular layers (greaterthan 10 Aring) A cross section that is likely to belarge above 20K is sHOH which describes themean distance for diffusion of mobile H before it

reacts Since s discussed earlier is of the sameorder of magnitude as the H2O dissociation crosssection sd then the precursor concentration cp(in this case H2O2) is roughly proportional to[s2OHsHOH] 1 (Appendix)

Based on the above a two-step precursor for-mation process could be consistent with the elec-tron data contrary to the statements in Sieger etal (1998) Orlando et al (1998) and Orlando andSieger (2003) Experiments are needed to deter-mine the relevant cross sections Below we con-sider a precursor formed in a single excitationevent

Precursor trapped O

O2 can be formed for instance by diffusion ofa transiently mobile O to a previously formed andtrapped O or to the production of O near the siteof a previously produced and trapped O (Matichet al 1993 Johnson and Quickenden 1997) Werefer to this as the MatichndashJohnsonndashQuickenden(MJQ) model and note that O 1 OH R O2 1 Hcan also give molecular oxygen

Trapped O as H2O-O was in fact seen to be aprecursor to O2 when attacked by a mobile O ina Kr matrix (Khriachtchev et al 2000) ThereforeO trapped as H2O-O was also suggested as a pre-cursor in ice (Johnson 2001) Before proceedingwe note that the examination of the MJQ modelin Sieger et al (1998) was incorrect They consid-ered two mobile interacting species whose life-times were long compared with the bombard-ment rate rather than a new mobile O interactingwith a previously trapped O Although this hasnot been corrected recently they added Otrapped as H2O-O as a possible precursor (Or-lando and Sieger 2003) O trapped as H2O-O isseen in matrix isolation studies as discussedabove but it has not yet been identified in iceand like OH it may trap in a number of ways

The MJQ model is essentially the followingFormation of H2 1 O is followed by loss of H2and trapping of O (Ot) A subsequent dissocia-tion or excitation event can lead to the destruc-tion of Ot by reaction with for instance H or byconversion to H2O2 However if this dissociationevent also leads to H2 1 O a mobile O is pro-duced that can react with Ot with an interactionlength that depends on T This results in the for-mation of O2 [O 1 Ot R O2] giving a net pro-duction of 2 H2 and 1 O2 In the Appendix therelevant rate equations (Eqs A1cndashA3c) are con-

JOHNSON ET AL836

sidered Assuming the mobile O processes arefast these rate equations can also be integratedand simplified to give the form in Eqs 1 and 2The precursor column density Np now repre-sents the column density of Ot In this model sO2in Eq 2 is the ldquocross sectionrdquo for reaction of afreshly produced O with an Ot and sp9 indicatesdestruction of Ot Finally sp is the cross sectionfor production of H2 and Ot via processes dis-cussed above

Based on the sizes of the cross sections ex-tracted in the electron irradiation experimentsthe model with trapped O (Ot) as the precursoris quantitatively reasonable That is for a rangeof T both sp and sO2 might be described byevents of the type H2O R H2 1 O (whether di-rect or indirect) If these occur for 10 of theradiation-induced dissociation events as dis-cussed then the required product of cross sec-tions suggested by the sizes of s and Y` above isroughly obtained that is [sp sO2] 02ndash6 3 10234

cm2 in the low-energy electron experiments Thecross section for production of O might dependon the availability of dangling bonds at interfacesin pores or at defects as suggested by a numberof authors The importance of such bonds couldalso be consistent with the differences in the elec-tron-induced yield for initially crystalline and ini-tially amorphous ice (Orlando and Sieger 2003)

O2 production can increase in the presence ofoxygen-rich impurities such as CO2 or SO2 bothof which are known to be present at Europa Asin the MJQ model dissociation (eg CO2 R CO1 O) can provide a mobile O that can be trappedor react with a previously trapped O forming O2In addition such species are hydrogen scav-engers (Moore and Hudson 2000) This can re-duce the destruction of Ot affecting sp as seen inEq A5c

Trap density

In the simplified models above the density oftrapping sites nt was not explicitly accountedfor However in the discussion of both the ex-perimental data and models trapping sites havebeen assumed to be important It has beendemonstrated repeatedly that the density of de-fects voids and interfaces decreases as the tem-perature of formation of the ice increases How-ever the efficiency of production of O2 is theopposite it increases with increasing T There-fore the oxygen yield varies in a manner that is

the inverse of the trap density over a range oftemperatures For penetrating ions in which theenergy density deposited is high and O2 might beformed in a single track a simple explanation hasbeen given As the density of traps for O2 is re-duced and the temperature increases the O2 pro-duced can diffuse to the surface and escape moreefficiently limiting back reactions on further ir-radiation This is examined in the next sectionHowever such a process does not affect the low-energy electron yields for which the penetrationdepth is small and the formed O2 escapes read-ily at the measurement temperatures In Eqs A1cA1c9 A3c and A10 in the Appendix we againuse a simple set of rate equations for trapped andmobile O in order to determine the role of the trapdensity nt on the production of a precursorspecies The conclusions apply to any modelbased on a trapped precursor

In the MQJ model mobile O can be trappedforming Ot (described by kOt nt) and can reactwith previously trapped O to form O2 (kOOt) Inaddition it can react at other sites and Ot can bedestroyed by mobile reactive species such as H(kHOt nH) Mobile H can also react with othertrapped species (kHR nR) Incorporating these re-actions Eq A10 in the Appendix gives a steady-state yield Y` When the density of traps is largethe steady-state yield is given in Eq A5c

Y` R (sd9sd9sd) [kOOtkHR nr][kHOt kOt nt] N (5)

Here sd and sd9 are the dissociation cross sectionsthat determine the production of H (H 1 OH) andO (H2 1 O) by the incident radiation Comparingthe result in Eq 5 with Y` in the simple precur-sor model in Eq 4b the dissociation cross sec-tions in Eq 5 are now scaled by a ratio of the re-action rates

It is seen in Eq 5 that the steady-state yield Y`depends inversely on the density of traps nt Thatis there are competing destruction processes fortrapped O (Ot) Therefore in order to form O2 amobile O competes with mobile H to find atrapped O before it becomes trapped In this waythe increasing O2 yield correlates with the re-duction in the trapping sites with increasing tem-perature of formation The small activation ener-gies discussed earlier would be related to there-orientation of the lattice binding leading to thereduction in trap density At relatively high T thedensity of traps becomes small so that only the

OXIDANTS AT EUROPA 837

sample surface or grain interfaces might act astraps accounting for the observed drop in theyield at high T For this reason it is important todetermine if the yield for T greater than 150Kdepends on the incident flux f Since other quan-tities in the expression for Y` could also be tem-perature dependent experiments are neededThese would likely be pulsed radiolysis experi-ments in which the transients are studied ratherthan the steady-state yields However the avail-able data for lightly ionizing radiation do suggestthat the O2 yield depends inversely on trap den-sity as in the model above

Escape from depth

The quantity N in Eq 1 is the ldquodepthrdquo given asa molecular column density from which O2 formedbelow the surface can percolate to the surface andescape It can be converted to a physical depth bydividing by the molecular number density of theice In the case of the low-energy electrons thedepth below the surface at which ionizations andexcitations are produced is small However forfast ions that penetrate many monolayers this isnot the case Below we examine N in Eq 1

Although it has long been understood that twoexcitations events are required to produce O2(eg Brown et al 1982 Reimann et al 1984) theprecursor model suggests that at saturation(steady state) any new event can produce O2 fromthe steady-state density of trapped precursorsTherefore one might expect the yield at satura-tion to be nearly linear in the excitation cross sec-tion This is the case for the low-energy electronsand it was initially assumed to be the case for thefast ions (Brown et al 1980 Reimann et al 1984)However Baragiola et al (2003) showed that thetotal sputtering yield in the saturation region attemperatures greater than 100K which is dom-inated by decomposition (Brown et al 1982) isproportional to the square of the electronic en-ergy deposition per unit path length in the solid(dEdx)e Assuming that the production and de-struction cross sections in Y` in Eq 4b roughlyscale with (dEdx)e then their ratio [sp sO2s]also roughly varies as (dEdx)e If this is the casethen the observed quadratic dependence impliesthat the depth N from which O2 can be mobi-lized to escape must also depend on the excita-tion density (dEdx)e This is consistent with theobservation that the O2 yield depends on thick-ness for samples thinner than the ion penetration

depth (Reimann et al 1984 Benit and Brown1990) Therefore the percolation depth for escapeN is a critical factor in comparing yields for pen-etrating radiation with those for non-penetratingradiation like the low-energy electron and low-energy ion data It is also seen that percolationand trapping can be included within the simplemodel considered here

Model summary

We have examined the simple precursor modelintroduced by Sieger et al (1998) to explain theirelectron data We have shown how it can be gen-eralized and derived from an integration of theappropriate chemical rate equations with thecross sections and total column density contain-ing the reaction rates and the physics of diffusivetransport Therefore the parameters can havevery different meanings from those initially as-sumed The equations are useful when there arefast processes which go to completion in timesmuch shorter then the time between particle im-pacts and slow processes associated with speciesthat are assumed to be trapped between particleimpacts This separation allowed us to examinethose models proposed for precursor formationand destruction due to reacting mobile speciesformed from dissociation events Further fromthe size of the onset of the linearity of the yieldversus fluence we can place constraints on therole of nonlinear processes such as those thatmight be associated with the formation of a pre-cursor like peroxide

With these extensions this simple model is use-ful for approximating the full rate equations inthe temperature region for which the yields de-pend on the fluence (dose) and not on the doserate Based on the assumptions in the model thisimplies temperatures at which precursor speciesare stable over experimental times At the high-est temperatures for the yields shown in Fig 4athis might not be the case This should be testedexperimentally In this regard it is puzzling thatexperiments have shown that the absorptionbands associated with certain trapped speciessuch as OH and trapped electrons ldquodisappearrdquoat 100ndash120K (see eg Johnson and Quickenden1997) whereas the O2 yield data indicate that rel-evant precursor species must be stable to 140KTherefore the proposed O2 production processescan only be distinguished by directly determin-ing the chemical state of the irradiated ice

JOHNSON ET AL838

It may be the case that a particular precursoris not required as mentioned earlier Rather af-ter a sufficient level of oxidation a newly pro-duced O or OH can react in the oxygen-rich lat-tice producing O2 in a pseudo-first-order processThis can be tested experimentally Indeedtrapped O and peroxide may be changed into oneanother as in the matrix isolation experiments(Khriachtchev et al 2000) Therefore their rela-tive concentrations versus temperature and doseshould be studied

In the model considered here it is seen that theparameters in Y` must account for the strong tem-perature dependence Since the number of bulktrapping sites decreases with increasing T as theice sample is annealed one might expect that theprecursor concentration would decrease We sug-gest that the observed temperature dependenceis determined by the competition between theavailability of trapping sites and mean interactiondistances of a mobile species such as O and HThat is the O2 yield varies inversely with the den-sity of traps over a range of temperatures In thesimple model above this effect is included indi-rectly as shown using the more detailed equa-tions in the Appendix

EUROPA

Oxygen and peroxide production at Europa G values

The dependence of the yield on the thicknessof the ice sample for highly penetrating ions sug-gests that the O2 yield is roughly related to theamount of energy deposited in the sample There-fore the amount of O2 produced and lost fromice in steady state is often given as a G value thenumber of molecules produced per 100 eV de-posited in Table 2 In applying these G values tothe planetary environment care must be takensince the yields are temperature fluence and ex-citation density dependent They also are affectedby the presence of other proton scavengersknown to be present in Europarsquos ice

The G values for peroxide in Table 2 are thosefor production at low fluence but the G valuesfor O2 are for the production and escape of gas-phase oxygen at saturation Therefore they de-pend on the formation and destruction cross sec-tions in the model above Such G values can beused to describe the steady-state source of the am-

bient O2 atmosphere at Europa but are not help-ful in determining the ability to produce and trapO2 at depth into the icy surface Over the timescales of the experiments some O2 is seen to re-main trapped and is not destroyed by radicalsTypically there are a number of trapping sitesand recent estimates of the activation energy foran O2 molecule trapped in a substitutional site(Hori and Hondoh 2002) suggest that it can betrapped for times comparable to the time re-quired for the conversion of amorphous ice intoIh (Ayotte et al 2001) The stability of O2 inclu-sions has not been studied

To understand the relationship between the Gvalues for O2 in Table 2 we note that fast lightions and electrons lose their energy predomi-nantly by producing ionizations and excitationsThe ionization events produce secondary elec-trons that can result in further ionizations Thenumber of ionization events Ni produced by afast light ion stopping in a water ice is Ni 5EoW where Eo is the initial energy and W is theaverage energy expended per ionization event27 eV in water ice Therefore radiolysis by fastions is often described as the net effect of theshower of electrons set in motion by the incidention In this manner the G values for productionof O2 from ice by energetic light ions and elec-trons can be roughly estimated as the sum of theeffects of all the secondary electrons using theyield Ye for 10ndash100 eV electrons That is if weignore the percolation depth N we can write thenumber of O2 produced as the number of ioniza-tion events times Ye which already has in it thedestruction processes That is G NiYeEo YeW 0006 O2100 eV

This estimate of G is seen to be an order of mag-nitude larger than the keV proton value for O2 inTable 2 but is much smaller than the G value forenergetic Ne1 These differences from the simpleestimate are consistent with the observation thatthe yield in the saturation regime for penetratingions varies nonlinearly with excitation densityThe difference from the keV proton G value islikely due to the inability of O2 to escape fromdepth before destruction On the other hand thelarger G value seen using MeV Ne1 is likely dueto the high density of excitation This can resultin more efficient production of O2 increased pro-duction of damage sites for forming O2 andorincreased transport (percolation) along the tran-siently heated track That is the ability of theformed O2 to percolate to the surface during ion

OXIDANTS AT EUROPA 839

bombardment plays an essential role in deter-mining the measured yields as discussed aboveThis effect needs to be evaluated for a planetarysurface subject to particle fluxes that are many or-ders of magnitude smaller than those used in thelaboratory

In the absence of O2 yields for all of the rele-vant ions and energies an estimate of the totalsupply of O2 can be obtained using the G valuesin Table 2 For the energetic oxygen and sulfurions we use G values for neon and argon ionsThese are lower bounds as the data were obtainedat temperatures much lower than the surface tem-peratures at Europa (80ndash110K) For the fast pen-etrating electrons and protons we use the valuefor 200 keV protons in Table 2 For the heavy ionswe used the Ne1 and Ar1 values For compari-son we also give an estimate using the G valuesobtained from the low energy electron data andthe total ionization as discussed above

The G values are combined in Table 3 with theelectron and ion energy flux at Europa Althoughthere is some deflection of the flow due to in-duced fields the net energy flux to the surface isnot reduced much from these values (Paranicaset al 2002) In fact locally produced pick-up ionswhich are formed from the ejecta and acceleratedto the surface can contribute (Ip 1996) Furtheralthough G values are typically independent ofthe incident angle the loss of O2 is a surfaceprocess and is enhanced with angle That is thenumber of O2 produced within the escape depthincreases with increasing angle to the normal(steeper than the inverse cosine) This enhance-ment is 25ndash4 (Jurac et al 2001) Ignoring suchcorrections the total production rate calculated

(23ndash62 3 109 O2cm2s) is a lower bound to theactual production rate from the icy regions of Eu-roparsquos surface Recent modeling of the O2 atmo-sphere suggest that such a source rate is required(Shematovich and Johnson 2001)

Earlier estimates of the O2 source rate were in-correct because O2 production was assumed to bea fixed fraction of the H2O sputtering yield Atthe ambient temperature an ejected H2O sticks ef-ficiently (Smith and Kay 1997) to the surface ofneighboring grains in the porous regolith reduc-ing the sputtering yield for H2O (Johnson 1989)This reduction does not apply to the O2 fractionof the yield enhancing its relative importance toH2O Here we assume that all of the gas-phaseO2 produced can contribute to the ambient at-mosphere even if it is produced at a grain sur-face that is at some depth into the porous regolith

Finally Europarsquos surface is not pure ice every-where That is much of the trailing hemisphereis a hydrated sulfuric acid containing smallamounts of sulfur and SO2 and the leading hemi-sphere contains low levels of carbon CO2 andcarbonates Sodium and potassium atoms aresputtered from the icy regions (eg Leblanc et al2002) O2 production from surfaces containingthese species has not been measured C and S areboth proton scavengers and the other species canbe sources of O as discussed so that the O2 yieldfrom such regions could be enhanced Howeverthere can also be competing processes that mightreduce these rates In fact recent HST observa-tions suggest that the O2 might not be produceduniformly across Europarsquos surface (McGrath et al2000) The analysis of these observations is pre-liminary and is not simple since the observed

JOHNSON ET AL840

TABLE 3 APPLICATION TO EUROPA

OxygenImplantation Energy flux production

Species 109(cm2 s)a 109 keV(cm2 s)(1) 109(cm2 s)

H (10 keV) 0015(1) 12 0084 (072)bO (10 keV) 0015(1) 18 054cS (10 keV) 0009(1) 30 090cElectrons 62 043 (37)bPlasma (10 keV) (137 amu) 13(2) 10 03c

Total 79 23 (62)UV photons (64 eV) 46 01

aReferences are as follows 1Cooper et al (2001) [similar results in Paranicas et al (2002)] 2Mauk et al (1996)bUsing G for 200 keV protons (upper bounds in brackets are based on the low-energy electrons data as a model for

the secondary electrons produced)cUsing G for 15 MeV Ne1 at 7K a rough lower bound

emissions depend on the local plasma propertiesHowever the efficiency of O2 production in con-taminated ices needs to be studied

Porous regolith O2 interactions on grain surfaces

Although the O2 atmosphere is principally pro-duced in the near surface layers and is eventuallylost to space it also permeates Europarsquos porousregolith The porosity p (void space) of the re-golith is uncertain (Buratti 1995) It has been es-timated to be the most compact surface in theOuter Solar System p 025 on the trailing hemi-sphere and p 079 on the leading hemisphere(Buratti and Veverka 1983) and the most poroussurface p 096 globally (Domingue et al 1991Domingue and Hapke 1992) These large differ-ences are due to the uncertainties in the amountof backscattered light at small phase angles andon the photon absorption properties and grainsizes The more compact surface is likely correctGrain sizes rg of the order of 100 mm have beensuggested (Hansen and McCord 1999 2000Geissler et al 1998) Europarsquos regolith differs sig-nificantly from the lunar regolith which is asand-like surface of individual grains Analysis ofthe thermal inertia of Europarsquos surface (Grundyet al 2001) suggests that the ice grains are sin-tered forming a fairy castle-like structure Sin-tering can occur thermally or due to the incidentradiation In addition ices doped with acids andalkalis in the laboratory can be powdery solidswith the impurities primarily found at grain in-terfaces

The porous regolith affects the above estimatesof the oxygen production in three ways First forisotropic bombardment the mean distance d intothe surface at which an incident ion or photonstrikes a grain surface is d (89)rg(1 2 p) Us-ing the above value for rg then d 02 m if theporosity is very large p 96 Therefore de-sorbed molecules can come from a considerabledepth and will interact with grain surfaces Evenfor much lower porosities the effective sputteringyields can be significantly modified (Johnson1989) As discussed above the yields for ejectedH2O which sticks onto neighboring grains at theambient temperatures are reduced but those forejected O2 are not This enhances the relative im-portance of the O2 yields

Finally the surfaces and interfaces of the sin-tered grains in Europarsquos porous surface are placeswhere low temperature chemistry involving the

ambient oxygen can occur at depths that are wellbelow the nominal penetration depth of the inci-dent radiation The relative importance of suchchemistry is roughly indicated by the ratio of thetotal surface area in the regolith to the area of Eu-roparsquos visible surface Using the porosity andgrain size above and a regolith depth dr of theorder of a meter then this ratio is large 3(1 2p)drrg which is 103ndash2 3 104 for average porosi-ties of 96 and 25 respectively Therefore thegas-phase O2 molecules below the visible surfaceinteract frequently with the grain surfaces In thisway a very large volume of the subsurface couldbe altered chemically even if it is not directly ex-posed to the radiation

From the overlying structures seen in the high-resolution images of Europarsquos surface by theGalileo spacecraft (eg Prockter and Pappalardo2000) it is clear that the youngest surfaces are red-der The reddening in the visible has been attrib-uted to the presence of sulfur in the ice (McEwen1986 Johnson et al 1988 Calvin et al 1995 Carl-son et al 1999b) The amount of reddening in thespectra of the surface material decreases with ageand the surface materials appear to becomebrighter with age This might occur by the slowaccumulation of a frost over-layer or by radioly-sis (Carlson et al 2002 Johnson et al 2003) Ra-diolysis preferentially converts the sulfur in theice to sulfur dioxide or sulfuric acid both ofwhich have flat spectra in the visible In regionswhere the radiation does not directly penetratethe O2 permeating the regolith can oxidize thesulfur at the surfaces of the icy grains and at graininterfaces Similarly the carbon seen in the ice onEuroparsquos leading hemisphere (eg Carlson et al2004 Johnson et al 2003) can be oxidized by ra-diolysis in the near surface layers or by reactionson the grain surfaces at depth in the regolith bythe ambient O2 The meteoroid bombardmentthat produces the regolith also exposes fresh sur-faces for oxidation and can cause the collapse ofpores leading to the trapping of oxygen In thismanner both refractory and volatile oxidants canbe formed and trapped in the subsurface ice forpotential delivery to Europarsquos subsurface ocean

SUMMARY

The presence of a thin oxygen atmosphere andoxidants trapped in Europarsquos surface ice (egJohnson et al 2003) combined with the possible

OXIDANTS AT EUROPA 841

presence of a subsurface ocean (eg Carr et al1998) have made Europa an exciting target forfuture space exploration The molecular oxidantsproduced by radiolysis and photolysis could bespecies needed for carbon-based biochemistry inEuroparsquos putative subsurface ocean Howeverthere has been no model for the formation trap-ping and transport of relevant oxidants

In this paper we have reviewed the space ob-servations and laboratory studies related to thechemistry of formation of oxidants in ice by theincident radiation Because the laboratory dataare incomplete we examined the rate equationsfor formation of oxygen and related species byradiolysis and photolysis We showed how suchequations can be simplified to the analytic modelthat was used for fitting the fluence and temper-ature dependence of the O2 yield produced bylow-energy electrons (Sieger et al 1998) In de-riving the analytic model from the more completerate equations the competition between forma-tion and destruction processes can be studied ascan the role of trapping sites escape from depthand temperature history These equations also al-lowed us to put incident photon electron and iondata on a similar footing for the first time

The model can be used to suggest measure-ments that are needed to describe the chemicalstate of the irradiated ice and to interpret new lab-oratory data We showed that if reactions are fastthe yield data could be consistent with a precur-sor being formed by two events as is the case forperoxide However we favor trapped O as a pre-cursor as initially suggested by Matich et al(1993) for production of excited O2 Such aspecies if present could be detected spectro-scopically as can trapped O2 through bands ac-tivated by lattice perturbations (Cooper et al2003a)

The available oxygen yield data were used tocompare G values for different incident particlesThese were then used to estimate the oxygen pro-duction in the radiation environment at EuropaThe lower bounds were consistent with the pro-duction rates obtained from a recent model of Eu-roparsquos atmosphere The porosity of the regolithwas also considered It not only affects the rela-tive sputtering yields for O2 and H2O but alsoprovides an enormously increased surface areaon which the ambient gas-phase oxygen can re-act with species in the ice That this might be oc-curring is consistent with the change in re-

flectance associated with surface aging Such re-actions could also convert ambient molecularoxygen into other more refractory species that re-main trapped on subduction of Europarsquos crust

Although radiolysis produces a stable chemi-cally altered surface in the laboratory the parti-cle flux to Europarsquos surface is many orders ofmagnitude lower Therefore annealing can be ef-fective at temperatures that are much lower thanthose typically found experimentally Belowabout 120K the annealing times appear to be long(eg Baragiola 2003) so that laboratory data canoften be directly applied to Europarsquos icy surfaceHowever account has to be taken of the syner-gism between the UV and the energetic plasmaas discussed earlier (Johnson and Quickenden1997)

Transient surface melting might initiate lifeprocesses but subduction of oxidants to the pu-tative ocean is likely to be required if biochem-istry is to occur at Europa (Chyba and Hand2001) Although the temperature versus depthinto Europarsquos regolith is not well known belowthe regolith the temperature and pressure appar-ently increase with depth At some depth (ap-proximately kilometers) either very warm ice oran ocean exists (Ruiz and Tejero 2000) Thereforechemically refractory oxidants are likely to be re-quired in order to survive subduction to thewarmer subsurface regions Trapped species ifdelivered downward can become mobile andwill segregate forming inclusions (Johnson andJesser 1997) Indeed peroxide appears to segre-gate in ice (Gurman et al 1967) and these inclu-sions may remain relatively stable until the per-oxide is dissolved in the putative ocean Thevolatiles seen in reflectance such as O2 SO2 andCO2 will also segregate But at the higher tem-peratures they could also diffuse through thebulk react or percolate into the regolith atmo-sphere Therefore the observation of the trappedvolatiles in Europarsquos surface ice is of interest pri-marily because it suggests that oxidants are pre-sent The formation of oxygen-rich molecules fol-lows the preferential loss of hydrogen Howeverthe surface is in a local steady state with im-planted H O and S from the plasma and the lossto space Since the plasma flux to the surface canbe variable this balance can also vary Indeedtemporal changes in reflectance have been re-ported (Domingue and Hendrix 2004)

Based on the models for formation of O2 ex-

JOHNSON ET AL842

amined in this paper additional experiments areneeded to determine the chemical state of irra-diated ice at low fluences Of immediate inter-est is the identification of stable molecular pre-cursors required for the formation of O2 Basedon the available laboratory data individual per-oxide molecules trapped in the ice do not appearto be the principal precursors but trapped O(Matich et al 1993) remains a viable candidateAlthough the data suggest that the local hydro-gen-bonding network is important in the pro-duction of O2 the effect of the formation tem-perature and radiation damage on the structureneed to be measured It is also important to de-scribe quantitatively the relationship betweenH2 loss and the formation of O2 and H2O2 overa range of temperatures and fluences Finallyone would like to determine the steady-statedensities of biochemically interesting moleculesformed by radiolysis in an ice containing sulfurand carbon at Europarsquos surface temperaturesand then determine their stability with increas-ing temperature and pressure Such experimentsstill need to be carried out Here we have pre-sented a model that brings together the variouslaboratory data and now can be used to considerthe formation trapping and transport of relevantoxidants at Europa

APPENDIX

Introduction

In this Appendix we examine rate equationsthat describe various aspects of the radiolysis orphotolysis of ice These equations need to be un-derstood if one is to be able to calculate the for-mation trapping and transport of oxidants Theemphasis is on the formation of molecular oxy-gen We integrate these equations over the escapedepth of O2 to give the simplified rate equationsdiscussed in the text in which number densitiesare replaced by column densities (eg Eqs 1 and2) In this way we can examine the relationshipbetween the rate constants and the effective pre-cursor formation and destruction cross sectionsextracted from data We first consider a precur-sor such as H2O2 that is formed from two exci-tation events This bears on the observed linear-ity of the oxygen yield versus fluence We thenconsider the equations for trapped versus a tran-

siently mobile species We use a reaction betweena mobile and trapped O as an example althoughO 1 OH can also give oxygen Finally we ex-amine the possible dependence of the O2 forma-tion process on the density of traps (defectsvoids and grain boundaries) as a function of tem-perature The results are discussed in the text

For clarity all possible radiolytic species arenot included in the discussion below Thereforepotentially important products such as HO2 andthe charged species (eg H3O1 the trapped elec-tron etc) are missing In each section we attemptto make a point related to production of O2 andother oxidants by using a subset of the productsrather than attempt a complete description withmany unknown rate constants

Fast versus slow processes

To describe the O2 production rate from a pre-cursor such as H2O2 we first use a simple but in-structive set of rate equations to describe the for-mation of H2O2 We then relate these to thesimplified rate equation for the precursor in thetext Eq 1 and show how the postulated precur-sor formation cross section sp is related to theH2O dissociation cross section sd by the reactionrates

Using the reactions

H2O 1 radiation R H 1 OH (sd)

H 1 H R H2 (k2H)

H 1 OH R H2O (kHOH)

OH 1 OH R H2O2 (k2OH)

the reaction pathways are described by

H2

Hqk2H

1 OHH2O 1 radiation mdashRsd H

OHH2O2

OH Rk2OH

kHOH

HH2O

OXIDANTS AT EUROPA 843

The corresponding rate equations in an irradiatedvolume are

dnOHdt 5 sd f n 2 2 k2OH nOH nOH

2 kHOH nH nOH (A1a)

dnHdt 5 sd f n 2 2 k2H nH nH

2 kHOH nH nOH (A2a)

dnH2O2dt 5 k2OH nOH nOH

2 destruction processes (A3a)

Here sd is the dissociation cross section f is theradiation flux and n is the molecular numberdensity of ice with nOH and nH the density of dis-sociation products OH and H respectively Therate constants k2OH k2H and kHOH describe thereactions above In this model the H2 formed isassumed to escape for temperatures above about20K We do not distinguish between trappedspecies and mobile species although in fact thatdistinction is important as discussed in the nextsection That is any OH produced at low tem-perature by one ion would likely be trapped priorto interacting with an OH produced by a subse-quent ion At higher temperatures this can breakdown as discussed in the text We first relate theabove set of equations to Eq 1 in the text and thenconsider precursor destruction processes

Equation 1 in the text is obtained by solvingthese equations together and then by integratingover a depth Dx Here Dx is the smaller of the pen-etration depth of the particle and the O2 escape(percolation) depth Assuming the molecular den-sity of ice n is constant with depth Dx n 5 N inEq 1 Comparing Eq 1 with Eq A3a the precur-sor formation rate sp f N is equivalent to [k2OH

nOH nOH] Dx Assuming that the processes in EqsA1a and A2a are fast then the steady state is read-ily obtained In addition if k2H k2OH there ispreferential loss of H2 at short times as suggestedby the laboratory data If steady state is reachedthe production of H2 is equal to that of H2O2 us-ing only the equations above Further one obtains

[k2OH nOH nOH] Dx R [fp sd] f N (A4)

with

fp 5 05[1 1 05 kHOH(k2H k2OH)05] (A5a)

Therefore the precursor formation cross sectionsp in Eq 1 is a fraction fp of the total dissocia-

tion cross sd Since laboratory data suggest theH2O2 fraction in an irradiated ice is small thenkHOH (k2OH k2H)05 which is likely givingfp R [(k2OH k2H)05kHOH]

Trapped versus mobile species

To examine the fluence dependence discussedin the text we note that trapped OH and freshlyproduced transiently mobile OH can be treatedas different species (eg Matich et al 1993)Therefore assuming no trapping of H the OHrate equation above would be replaced by twoequations one for the trapped species nOHt andthe other for the freshly produced mobile speciesnOH Assuming the reaction of two mobile OHunits is unlikely and that peroxide is formed fromthe interaction of a trapped OH with a subse-quently produced OH an expanded set of equa-tions is obtained

OH 1 trap R OHt (kOHt)

OH 1 OHt R H2O2 (kOHOHt)

H 1 OHt R H2O (kHOHt)

then the reaction pathways are described by

qH2

H2O 1 radiation mdashRsdH

qk2H

1 OHH

trap OH

OH RkOHt

OHt kOHOHtH2O2

kHOHkHOHt

H H2O HH2O

The resulting set of equations is

dnOHdt 5 sd f n 2 kHOH nH nOH

2 kOHt nOH nt 2 kOHOHt nOH nOHt (A1b)

dnOHtdt 5 kOHt nOH nt 2 kOHOHt nOH nOHt

2 kHOHt nH nOHt (A1b9)

dnHdt 5 sd f n 2 2 k2H nH nH 2

kHOH nH nOH 2 kHOHt nH nOHt (A2b)

JOHNSON ET AL844

dnH2O2dt 5 kOHOHt nOH nOHt

2 destruction processes (A3b)

where nt is the trap density in the irradiated iceAt low T we can assume nt is large so thatdntdt 0 simplifying the set of equations If theprecursor of interest is trapped O as discussed inthe text then a similar set of equations can be con-structed for O and Ot as in the example to followOf course a full description over a broad rangeof temperatures would require the considerationof the both trapped and mobile H O and OHthe rate of change of trap density and the vari-ous destruction processes

The above equations should be solved simul-taneously but for the incident particle fluxes typ-ically used in these studies the diffusion of mo-bile species to traps is relatively fast This is trueeven at low temperatures if there is a significantdensity of trapping sites This is usually the casefor samples formed at low T Therefore in a lowtemperature solid at low dose rates there is not asignificant steady-state background of diffusingOH There are trapped or freshly produced OHunits which rapidly react Averaging over thebombardment rate this can be approximated bywriting dnOHdt 0 and then separating nOH outin Eq A1b Assuming also that H diffuses rapidlyand either reacts or forms H2 we also writednHdt 0 on the average Therefore betweenimpacts only species trapped at defects pores orgrain surfaces are assumed to be present as sug-gested by the lack of a dependence on the inci-dent beam flux In this limit the rate of change oftrapped OH roughly simplifies to

dnOHtdt 5 2 k2H nH nH 2 2 kOHOHt nOH nOHt

It is seen that the production of peroxide dependson the loss of H by formation of H2 as discussedand at steady state the production rate for per-oxide equals the rate of formation of H2 in theseequations

Treating the mobile and trapped species sepa-rately we can further simplify the rate equationsand obtain the precursor model used in the textThat is integrating over the relevant depth Dxthe densities are replaced by column densitiesThe rate constants ki are then related to effectivecross sections si In this way cross sections aregiven as an interaction length for each freshlyproduced mobile radical [ki nj] R [si f] The

trapped OH column density NOHt and peroxideproduction rate can be estimated from ldquolinearrdquorate equations

dNOHtdt lt [sd f] N 2 2 [s2OH f] NOHt

2 [sHOH f] NOHt (A6a)

dNH2O2dt lt [s2OH f] NOHt 2 s NH2O2 (A6b)

Here sd is again the dissociation cross sectionand the destruction processes in Eq A3b are con-tained in s The second and third terms on theright in Eq A6a are the loss of OH due to the pro-duction of H2O2 and H2O In the integration ofEq A1b9 [2 kOH OHt nOH] becomes [2 s2OH f] and[kHOHt nOH] becomes [sHOH f] Therefore thedensity and the interaction length of mobile H arecontained in sHOH If H2O2 is indeed the pre-cursor then in Eq 2 in the text [sp f N] is be-comes [s2OH f NOHt] from Eq A6b Accountingfor precursor destruction which can give backtwo OH and treating H2 can both be included inthe integrated model via the si values

The simplified set of equations above allows usto examine the relative time (fluence) scales forprecursor formation (in this case H2O2) SolvingEqs A6a and A6b the column density of OH andthe concentration of precursors versus fluence cpcan be obtained Using the column of H2O2 pro-duced as the precursor column Np in Eq 2 thenO2 yield can be obtained The results can be com-pared with the results from Eqs1 and 2 in thetext The average precursor (H2O2) concentrationover the column depth of water molecules N is

cp 5 fp (sds)[1 2 exp(2s F)] 2 (ssr)

[1 2 exp(2sr F)] (A7)

with

fp 5 2s2OH(sr 2 s) (A5b)

In these equations sr 5 (2s2OH 1 sHOH) andssr 1

It is seen that the precursor concentration withtime cp depends on two rates the reaction ratesthat determine sr and the destruction rates in sAs in Eq 4a the yield can be written as

Y 5 sO2 cp N (A8)

If sr and s are comparable then the yield at small

OXIDANTS AT EUROPA 845

fluence is quadratic in F However if sr sthen the linear dependence on fluence is obtainedwith a small shift in the fluence as shown in Eq4c in the text Further the steady-state yield isnow

Y` (T) 5 sO2 [fp sd] Ns (A9)

That is sp R [fp sd] as in Eq A4 and the tem-perature dependence for the incident electrondata must be contained in [sO2 fp] or in the ratio[sO2 s2OH]sHOH]

For comparison with their precursor modelSieger et al (1998) examined the possibility of ob-taining the O2 precursor by considering reactionsfor a density of diffusing dissociation products asin Eqs A1 and A2 rather than a trapped productinteracting with a freshly formed dissociationproduct Based on the above the lowest fluence atwhich O2 is first detected experimentally sets alimit on the time scale for mobilendashmobile reactions

Effect of trap density

Equations A1b A1b9 A2b and A3b can be ap-proximately solved assuming that the dnOHdtdnHdt and dntdt are very small on the averageas discussed above In order to examine the roleof the trap density nt which can depend on thetemperature and method of formation and on theradiation history of the ice we use the slightlysimpler precursor model In this model the O pro-duced directly by dissociation traps forming aprecursor O2 is then formed when a freshly pro-duced O or other species reacts with a trappedoxygen atom Ot

H2O 1 radiation R H2 1 O (sd)

O 1 trap R Ot (kOt)

O 1 Ot R O2 (kOOt)

The reaction pathways for the oxygen species are

H2O 1 radiation mdashRsd9

trap OO2

kOtOt kOOt

H2 1 OOHt

kOR H(kHO t)

ROR

We assume for simplicity both H2 and O2 escapefrom depth Dx after formation We also allow mo-bile H to react with Ot (kHOt) removing trappedO and both the mobile H and O to be removedby other reactants R (eg OH or contaminants)This leads to a simplified set of equations not di-rectly involving OH or peroxide

dnOdt 5 sd9 f n 2 kOt nO nt 2 kOR nO nR

2 kOOt nO nOt (A1c)

dnOtdt 5 kOt nt nO 2 kOOt nO nOt

2 kHOt nH nOt (A1c9)

dnHdt 5 sd f n 2 kHOt nH nOt

2 kHR nH nR (A2c)

dnO2dt 5 kOOt nO nOt (A3c)

These equations of course act in parallel with theequations above for OH and mobile OH may re-act with trapped O to form O2 Here we focusonly on mobile O

Since the diffusion of H and O are fast and thedensity of traps is large the nH and nt averagedover the bombardment rate can be taken veryroughly as time independent Further if we as-sume the number of trapped O nOt is small com-pared with the number of sites at which H canreact then nH [sd f nkHR nR] For a low ra-diation flux f nH is small on the average as as-sumed Since the O2 yield YO2 is equal to[(dnO2dt) Dxf] with Dx the escape depth thenthe steady-state yield Y` can be obtained ana-lytically

[Y`][sd9 N] 5 1 1 (q 2 1) 2 [(q 2 1)2

1 8 q d]054 d (A10)

q 5 [kOt (nt 1 nR)] (kHOt nH)(sd9 f n) kOOt

5(sdsd9) [kOt (nt 1 nR)] kHOt[kOOt kHR nR]

d 5 05 [1 1 nt(nt 1 nR)]

We note that for nt large q can be large Assum-ing that q 1 and nt nR then Eq A10 be-comes

Y` R [sd9 N]q 5 (sd9 N) (sd9sd)

[kOOt kHR nR][kHOt kOt nt] (A5c)

JOHNSON ET AL846

Comparing with Y` in the simple precursormodel in Eq 4b then s R [(kHOt nH)f] R [sd nkHOt][kHR nR] In addition the product [sO2sp] R sd9 sd9 [kOOt n][kOt nt] That is the dis-sociation cross sections are modified by reactionrates

Quite remarkably it is seen in Eq A5c that thesteady-state yield Y` can depend inversely onthe density of traps nt in the temperature andflux regimes for which the approximations arerelevant That is if there are competing destruc-tion process for trapped O then at high trap den-sity a mobile O must find a trapped O before itbecomes trapped or before mobile H destroys theOt In this way the increasing O2 yield with in-creasing T can be related to the change in the trap-ping density nt with the formation temperatureof the ice sample As the temperature increasesand the density of traps in the bulk decreases thesurface and grain interfaces become the only trap-ping sites

ACKNOWLEDGMENTS

REJ acknowledges support by a Gledden Fel-lowship in Chemistry School of Biomedical andChemical Sciences at The University of WesternAustralia and by grants from the NSF Astron-omy Program and NASArsquos Planetary Geologyand Geophysics Program PDC acknowledgessupport by an Australian Postgraduate Award

ABBREVIATIONS

HST Hubble Space Telescope MJQ MatichndashJohnsonndashQuickenden NIMS Near Infrared Map-ping Spectrometer UV ultraviolet

REFERENCES

Ayotte P Smith RS Stevenson KP Dohnarsquolek ZKimmel GA and Kay BD (2001) Effect of porosityon the adsorption desorption trapping and release ofvolatile gases by amorphous solid water J Geophys Res106(E12) 33387ndash33392

Bahr DA Famaacute M and Baragiola RA (2001) Radiol-ysis of water ice in the outer solar system sputteringand trapping of radiation products J Geophys Res 10633285ndash33290

Baragiola RA (2003) Microporous amorphous water icefilms and astronomical implications In Water in Con-

fining Geometries edited by JP Devlin and B Buch El-sevier New York in press

Baragiola RA and Bahr DA (1998) Laboratory studiesof the optical properties and stability of oxygen onGanymede J Geophys Res 103 25865ndash25872

Baragiola RA Atteberry CL and Bahr DA (1999) Ori-gin of solid oxygen on Ganymede Reply to commentby RE Johnson on ldquoLaboratory studies of the opticalproperties and stability of oxygen on Ganymederdquo JGeophys Res E104 14183ndash14187

Baragiola RA Vidal RA Svendsen W Schou J ShiM Bahr DA and Atteberrry CL (2003) Sputteringof water ice Nucl Instrum Methods Phys Res B 209294ndash303

Bar-Nun A Herman G Laufer D and Rappaport ML(1985) Trapping and release of gases by water ice andimplications for icy bodies Icarus 63 317ndash332

Baverstock KF and Burns WG (1976) Primary pro-duction of oxygen from irradiated water as an expla-nation for decreased radiobiological oxygen enhance-ment at high LET Nature 260 316ndash318

Benit J and Brown WL (1990) Electronic sputtering ofoxygen and water molecules from thin films of waterice bombarded by MeV Ne1 ions Nucl Instrum Meth-ods Phys Rev B 46 448ndash451

Benit J Bibring J-P della-Negra S LeBeyec YMendenhall M Rocard F and Standing K (1987)Erosion of ices by ion irradiation Nucl Instrum Meth-ods Phys Rev B 1920 838ndash842

Brown WL Augustyniak WM Lanzerotti LJ John-son RE and Evatt R (1980) Linear and nonlinearprocesses in the erosion of H2O ice by fast light ionsPhys Rev Lett 43 1632ndash1635

Brown WL Augustyniak W M Simmons E Mar-cantonio KJ Lanzerotti LJ Johnson RE BoringJW Reimann CT Foti G and Pirronello V (1982)Erosion and molecular formation in condensed gasfilms by electronic energy loss of fast ions Nucl In-strum Methods B 1 307ndash314

Brown WL Augustyniak WM Marcantonio KJ Sim-mons EN Boring JW Johnson RE and ReimannCT (1984) Electronic sputtering of low-temperaturemolecular solids Nucl Instrum Methods B 1 307

Buratti B (1995) Photometry and surface structure of theicy Galilean satellites J Geophys Res 100(E9) 19061ndash19066

Buratti B and Veverka J (1983) Voyager photometry ofEuropa Icarus 55 93ndash110

Calvin WM and Spencer JR (1997) Latitudinal distri-bution of O2 on Ganymede observations with the Hub-ble Space Telescope Icarus 130 505ndash516

Calvin WM Clark RN Brown RH and Spencer JA(1995) Spectra of the icy Galilean satellites from 02 to5 mm a compilation new observations and a recentsummary J Geophys Res 100 19041ndash19048

Calvin WM Johnson RE and Spencer JA (1996) O2

on Ganymede spectral characteristics and plasma for-mation mechanisms Geophys Res Lett 23 673ndash676

Calvin WM Aninich VG and Brown RH (2004) Oxy-gen temperature and phase effects Icarus (in press)

OXIDANTS AT EUROPA 847

duced by Lyman-a (98 eV) radiation of ice butfound no fluence dependence for fluences greaterthan 1017cm2 They also saw small amounts ofH2O2 by programmed thermal desorption of thesample but made no quantitative evaluationComparing their lowest fluence for the produc-tion of O2 with the result of Gerakines et al (1996)for peroxide suggests that H2O2 and HO2 areprobably not important precursors of O2 In UVphotolysis of H2O2 using 64 eV photons the pro-duction of O2 was not observed (Vaghjiani et al1992) and the excitation of an irradiated icewhich presumably contained some peroxide by37 eV photons did not produce O2 (Baragiola etal 2003) Direct excitation of matrix-isolatedH2O2 at low temperature also did not yield O2but the production of a mobile O that reacted withtrapped O (H2O-O) did give O2 (Khriachtchev etal 2000)

Irradiating with 200 keV protons at tempera-tures of 40ndash129K Bahr et al (2001) found onlyabout 0001ndash002 H2O2 using programmed ther-mal desorption as compared with earlier experi-ments that suggested larger values (eg Table 2)Using infrared absorption Moore and Hudson(2000) found a G value of 01 for incident MeVprotons at low temperature but the signal in theobserved band became negligible at 80K sug-gesting that H2O2 was not being formed Theyalso found that adding O2 significantly increasedthe production efficiency of H2O2 That is thatthe amount of peroxide was affected by theamount of trapped oxygen Based on the matrixisolation studies (Khriachtchev et al 2000) dis-sociated mobile O might trap as H2O-O whichon excitation converts to H2O2 Therefore even ifH2O2 is not the precursor the peroxide andtrapped oxygen concentrations will be related

Although molecular oxygen was not a productof photolysis of matrix isolated H2O2 it was aproduct of photolysis at 266 nm (466 eV) of per-oxide dimers in an argon matrix 2 H2O2 R 2 H2O1 O2 (Engdahl and Nelander 2002) This de-composition was thought to be due to the con-siderable distortion of the bonds in the dimerTherefore the configuration or distortion by thelattice of peroxide or any other precursor mole-cule in the ice may be critical (Sieger et al 1998)In addition more recent experiments show thatG values for the production of H2O2 by 30 keVincident ions are significant as are the steady-state ratios of H2O2H2O (Gomis et al 2004) Ra-tios of the order of 02ndash4 were found at 16K and

77K in the irradiated regions with incident H giv-ing the smallest fraction and incident O thelargest (Table 2) At these concentrations perox-ide molecules might segregate forming inclu-sions (Gurman et al 1967 Cooper et al 2003b)Since the amount of trapped peroxide also com-petes with the amount of trapped O studies areneeded to describe the chemical state of the solidproduced by radiolysis This will be critical in de-scribing the fate of the irradiation products at Eu-ropa In particular the chemical composition fa-vorable for producing O2 needs to be determinedas does the importance of trapped O H2O2 HO2or peroxide dimers as precursors

Other species

In the model below we will refer to atomic andmolecular species trapped in ice Such species areoften discussed in the literature on radiolysis andphotolysis (for summaries see Matich et al 1993Johnson and Quickenden 1997) Trapped OH OH2O2 and HO2 have been discussed andor mea-sured by a variety of techniques (absorption andluminescence spectroscopy electron spin reso-nance programmed thermal desorption etc)Care must be taken when using these data as re-sults for ice samples formed below 160K orthose exposed to penetrating ion beams likelyhave numerous defects voids and grain inter-faces so that a variety of trapping sites and trapdepths exist Therefore when a particular bandfor a trapped species (eg OH) disappears withincreasing temperature it could be due to achange in the nature of the trapping as well as toloss of the species by reaction

Although H is discussed below we do not dis-tinguish it in these simple models from H1

trapped as the hydronium ion H3O1 whichwould mean that there are corresponding nega-tive ions present Therefore below we refer to thevarious species as if they are neutral althoughthis might not be the case Although trapped elec-trons are detected only over a narrow range oftemperatures (100ndash120K) negative ions must beimportant over a broad range of temperature

We also note that SO2 and a sulfate and CO2and a carbonate appear to be trapped species inEuroparsquos icy surface Under radiolysis each ofthese can dissociate directly giving O2 Howeverwith the exception of the sulfate their densitiesare less than 1 so they are not relevant sourcesof O2 However their presence is indicative of the

OXIDANTS AT EUROPA 833

preference for forming oxygen-rich molecules inEuroparsquos ice and can affect the efficiency of pro-duction of O2 and H2O2

MODELS FOR FORMATION OF O2

Introduction

Based on the data described above the yieldsfor incident electrons and ions are found to benearly linear in fluence at the lowest fluences forwhich O2 is detected In addition the relevant al-terations of the ice samples by radiolysis appearto be stable and independent of the beam fluxRemarkably this is the case both for low-energyelectrons that make only a single excitation foreach impact and fast ions that make a density ofexcitations along their path through the solidThis suggests that O2 is produced and trapped inthe initial events and then caused to escape fromthe solid by the subsequent radiation or that astable chemical alteration of the ice is required be-fore the radiation can produce and eject O2

Because H produced by the dissociation of wa-ter molecules can attack trapped O2 and any pre-cursor oxidants the loss of H is important Thisoccurs by the production and loss of hydrogenmolecules as discussed leaving a surface that isslightly rich in O relative to H The O2 produc-tion mechanism could be direct excitation of aparticular precursor species or it might not bespecific That is a new OH or O produced by theincident radiation could react with the H-depleted solid producing O2 This would occurmore efficiently as the ratio of O to H in the nearsurface region increased

In their early work WL Brown and colleaguesassumed that processes similar to those used todescribe radiolysis of liquid and gaseous H2Ocould be used to describe the radiolysis in ice Thelow activation energy that they found was as-sumed to be associated with diffusion of somemobile radical species in ice such as H or O orpossibly diffusion of the O2 molecule itselfBased on their studies of ejection from isotopi-cally labeled D2O samples they concluded thatthe O2 is formed in the track of a single ion andthen ejected by subsequent ions (eg Benit andBrown 1990) More recently trapped specieshave been suggested as precursors These are ei-ther subsequently excited by the incident radia-tion or attacked by newly produced mobile

species We consider a simple model that incor-porates all of these processes and can be used toorganize and interpret the experimental data Wealso discuss its relationship to more completemodels In this discussion much of which is in theAppendix we consider the standard radiationproducts that have been often described How-ever in a solid the excited electron cloud pro-duced by absorption can overlap a number of lat-tice sites Therefore their relaxation pathwaysandor product species could be very differentfrom those that have been traditionally consideredin the literature on radiolysis and photolysis

Precursor model

The column density of a precursor species Npformed in ice by incident radiation can be calcu-lated using a simple model In this model cross sec-tions are used for the production sp and destruc-tion s of the precursor O2 is then formed byexcitation of the precursor or by reaction of a newlyformed radical with the precursor This is also de-scribed using a cross section sO2 For incident elec-trons having a small penetration depth the O2 pro-duced is assumed to rapidly diffuse to the surfaceand escape (Sieger et al 1998) More generally wedefine N as the column density of H2O from whichthe incident radiation can produce and eject O2 Apair of rate equations gives both Np and the pro-duction rate of oxygen dNO2dt That is

H2O 1 radiation R precursor

precursor 1 radiation R O2

giving

dNpdt 5 sp f N 2 s f Np (1)

dNO2dt 5 sO2 f Np (2)

In Eq 1 f is the incident flux and the total pre-cursor destruction cross section can be written ass 5 sp9 1 sO2 with sp9 accounting for precursordestruction processes other than the formation ofO2 Solving Eqs 1 and 2 and assuming N is nearlyconstant (10 level as discussed) the fractionof precursors cp in the affected column after timet is obtained (Appendix)

cp 5 NpN 5 [sps] [1 2 exp(2 s F)] (3)

Here F 5 [ft] is the fluence (flux 3 time) which

JOHNSON ET AL834

is the number of particles (photons electrons orions) incident on the surface per unit area Usingcp in Eq 2 the O2 yield Y at any fluence can bewritten in terms of the yield at steady state Y`

Y 5 [dNO2dt]f 5 sO2 cp N 5 Y`

[1 2 exp(2sF)] (4a)

Y` 5 [sp sO2 s] N (4b)

If we also account for the reduction in the den-sity of water molecules in the initial column Nthese forms change only slightly The size of thetwo parameters Y` and s can be obtained by fit-ting the yield versus fluence data such as that inFig 3

For the low-energy electron data Sieger et al(1998) showed that the ratio [YY`] for a givenincident electron energy is independent of tem-perature over a narrow range of temperatureTherefore the destruction cross section s in Eq4a is roughly independent of T and can be esti-mated from the [YY`] curve Sieger et al (1998)found a precursor destruction cross section s lt08ndash30 3 10216 cm2 for incident electron energiesfrom 30 to 100 eV This is of the order of the elec-tron impact ionization cross section for H2O

Based on the above model the lack of an ob-served fluence dependence for fluences down to1017 hncm2 in the Lyman-a irradiation of ice(Westley et al 1995) can be examined The low-est fluence used implies that the precursor de-struction cross section for Lyman-a is of the or-der of the Lyman-a absorption cross section in ice(09 3 10217 cm2)

Sieger et al (1998) also assumed that sp9 5 0This would mean that the precursor is only de-stroyed by production of O2 (ie s 5 sO2) Thatis unlikely since suggested precursors can be de-stroyed by reaction with H and other radicalsTheir assumption would also mean that the crosssection for production of O2 from a precursorsO2 is large This would be surprising and ap-pear to contradict some of the data discussedabove Instead we assume that s is not equal tosO2 and the total destruction cross section s isof the order of the ionization or absorption crosssection

Having obtained s the measured temperaturedependence for production of O2 is containedonly in Y` Over a narrow range of T one canwrite Y`(T) lt Y`9 exp(2EakT) and obtain the es-

timate of the activation energies in Table 2 Ifsp9 THORN 0 as discussed then sp and sO22 in Eq 4bcannot be obtained separately from the data forY` unless an independent measure of cp is ob-tained Such a measurement is critical for de-scribing the chemical state of the irradiated sur-faces

The observed temperature dependence of Y` iscontained in sp sO2 or N Here N is the ldquodepthrdquofrom which newly formed O2 can reach the sur-face and escape given as a column density of H2OThe physical depth is obtained by dividing thecolumn density N by the molecular number den-sity of the ice Assuming that in the low-energyelectron experiments N is independent of tem-perature and is about three monolayers (Sieger etal 1998) the measured Y` gives [sO2sps]05ndash20 3 10218 cm2 over the electron impactenergy range At 110ndash120K this ratio was foundto be proportional to (E 2 Eo) for E Eo 10 eVagain assuming fixed N (Orlando et al 1998 Or-lando and Sieger 2003)

In the following section we first consider a pre-cursor like peroxide that typically forms fromtwo dissociation events We use this to show un-der what circumstances such a model does notviolate the observed linearity of the O2 yield datawith fluence Following that we relate the sim-ple precursor model above to the case wheretrapped O is the precursor We then use themodel for trapped O to examine the effect of thetrap density on the production of O2 The densityof traps (defects) is determined by the sample for-mation temperature and the subsequent radiationdamage Finally we consider the escape depthN for penetrating radiation within the precursormodel These discussions are supported by moredetailed rate equations in the Appendix

Precursor trapped peroxide

The proposed precursor peroxide is typicallyformed from products of two dissociation eventsie 2 H2O R 2 H 1 2OH R H2 1 H2O2 This wasconfirmed in low temperature ice (Gerakines etal 1996) It is also possible that a dissociation pro-ducing a trapped O followed by an excitationevent can produce peroxide 2 (H2O 1 hn) R H2

1 H2O-O 1 hn R H2O2 1 H2 (eg Khriachtchevet al 2000) The O2 yield versus fluence for inci-dent electrons is reasonably well fit by the yieldexpression above for fluences greater than1014cm2 Therefore it has been suggested that

OXIDANTS AT EUROPA 835

any precursor species must be formed in a singleevent Using a model in the Appendix in whichthe precursor is formed by two events we showthis is not necessarily the case Precursor forma-tion if it is fast could involve two dissociationevents and give the measured fluence depen-dence Although we consider peroxide here theseconclusions apply to other possible precursors

In the example in the Appendix H2O is disso-ciated to H 1 OH according to the cross sectionsd The reactions of two OH units to form H2O2(rate constant k2OH) and 2 H to form H2 (k2H)compete with the recombination of H and OH toH2O (kHOH) If k2H is much greater than k2OH then H2 loss dominates at low fluences as ex-pected (Appendix) Steady state requires not sur-prisingly that the rate for forming H2O2 isroughly equal to the rate for forming H2 Finallyit is seen in the Appendix that the low H2O2 con-centrations suggested by experiment imply thatkHOH (k2H k2OH)05 Such sizes are reasonable

The set of rate equations in the Appendix thatuse number densities of species is integrated overdepth in order to obtain the forms used in Eqs 1and 2 which are based on column densities Inthis way the effective cross sections are related tothe rate constants for the interaction of freshlyproduced mobile species with previously pro-duced trapped species We show in the Appen-dix that the O2 yield versus fluence can be writ-ten in a form similar to that in Eq 4a but with asmall offset due to the two-step precursor for-mation process

Y lt Y`(T) [1 2 exp(2sF) 2 (ssr)]

for F sr21 (4c)

Here s is again the precursor destruction crosssection but sr 5 (2s2OH 1 sHOH) is the effectivecross section for loss of OH by formation of H2O2

and H2O The corresponding expression for Y`(T)is given in the Appendix

The form for the yield versus fluence in Eq 4cis consistent with the low-energy electron mea-surements if (ssr) 1 and if sr F21 lt10214 cm2 Such sizes are indeed possible Whatis required is that at the relevant temperatures(50ndash110K) the interaction length associated withsr is more than a few molecular layers (greaterthan 10 Aring) A cross section that is likely to belarge above 20K is sHOH which describes themean distance for diffusion of mobile H before it

reacts Since s discussed earlier is of the sameorder of magnitude as the H2O dissociation crosssection sd then the precursor concentration cp(in this case H2O2) is roughly proportional to[s2OHsHOH] 1 (Appendix)

Based on the above a two-step precursor for-mation process could be consistent with the elec-tron data contrary to the statements in Sieger etal (1998) Orlando et al (1998) and Orlando andSieger (2003) Experiments are needed to deter-mine the relevant cross sections Below we con-sider a precursor formed in a single excitationevent

Precursor trapped O

O2 can be formed for instance by diffusion ofa transiently mobile O to a previously formed andtrapped O or to the production of O near the siteof a previously produced and trapped O (Matichet al 1993 Johnson and Quickenden 1997) Werefer to this as the MatichndashJohnsonndashQuickenden(MJQ) model and note that O 1 OH R O2 1 Hcan also give molecular oxygen

Trapped O as H2O-O was in fact seen to be aprecursor to O2 when attacked by a mobile O ina Kr matrix (Khriachtchev et al 2000) ThereforeO trapped as H2O-O was also suggested as a pre-cursor in ice (Johnson 2001) Before proceedingwe note that the examination of the MJQ modelin Sieger et al (1998) was incorrect They consid-ered two mobile interacting species whose life-times were long compared with the bombard-ment rate rather than a new mobile O interactingwith a previously trapped O Although this hasnot been corrected recently they added Otrapped as H2O-O as a possible precursor (Or-lando and Sieger 2003) O trapped as H2O-O isseen in matrix isolation studies as discussedabove but it has not yet been identified in iceand like OH it may trap in a number of ways

The MJQ model is essentially the followingFormation of H2 1 O is followed by loss of H2and trapping of O (Ot) A subsequent dissocia-tion or excitation event can lead to the destruc-tion of Ot by reaction with for instance H or byconversion to H2O2 However if this dissociationevent also leads to H2 1 O a mobile O is pro-duced that can react with Ot with an interactionlength that depends on T This results in the for-mation of O2 [O 1 Ot R O2] giving a net pro-duction of 2 H2 and 1 O2 In the Appendix therelevant rate equations (Eqs A1cndashA3c) are con-

JOHNSON ET AL836

sidered Assuming the mobile O processes arefast these rate equations can also be integratedand simplified to give the form in Eqs 1 and 2The precursor column density Np now repre-sents the column density of Ot In this model sO2in Eq 2 is the ldquocross sectionrdquo for reaction of afreshly produced O with an Ot and sp9 indicatesdestruction of Ot Finally sp is the cross sectionfor production of H2 and Ot via processes dis-cussed above

Based on the sizes of the cross sections ex-tracted in the electron irradiation experimentsthe model with trapped O (Ot) as the precursoris quantitatively reasonable That is for a rangeof T both sp and sO2 might be described byevents of the type H2O R H2 1 O (whether di-rect or indirect) If these occur for 10 of theradiation-induced dissociation events as dis-cussed then the required product of cross sec-tions suggested by the sizes of s and Y` above isroughly obtained that is [sp sO2] 02ndash6 3 10234

cm2 in the low-energy electron experiments Thecross section for production of O might dependon the availability of dangling bonds at interfacesin pores or at defects as suggested by a numberof authors The importance of such bonds couldalso be consistent with the differences in the elec-tron-induced yield for initially crystalline and ini-tially amorphous ice (Orlando and Sieger 2003)

O2 production can increase in the presence ofoxygen-rich impurities such as CO2 or SO2 bothof which are known to be present at Europa Asin the MJQ model dissociation (eg CO2 R CO1 O) can provide a mobile O that can be trappedor react with a previously trapped O forming O2In addition such species are hydrogen scav-engers (Moore and Hudson 2000) This can re-duce the destruction of Ot affecting sp as seen inEq A5c

Trap density

In the simplified models above the density oftrapping sites nt was not explicitly accountedfor However in the discussion of both the ex-perimental data and models trapping sites havebeen assumed to be important It has beendemonstrated repeatedly that the density of de-fects voids and interfaces decreases as the tem-perature of formation of the ice increases How-ever the efficiency of production of O2 is theopposite it increases with increasing T There-fore the oxygen yield varies in a manner that is

the inverse of the trap density over a range oftemperatures For penetrating ions in which theenergy density deposited is high and O2 might beformed in a single track a simple explanation hasbeen given As the density of traps for O2 is re-duced and the temperature increases the O2 pro-duced can diffuse to the surface and escape moreefficiently limiting back reactions on further ir-radiation This is examined in the next sectionHowever such a process does not affect the low-energy electron yields for which the penetrationdepth is small and the formed O2 escapes read-ily at the measurement temperatures In Eqs A1cA1c9 A3c and A10 in the Appendix we againuse a simple set of rate equations for trapped andmobile O in order to determine the role of the trapdensity nt on the production of a precursorspecies The conclusions apply to any modelbased on a trapped precursor

In the MQJ model mobile O can be trappedforming Ot (described by kOt nt) and can reactwith previously trapped O to form O2 (kOOt) Inaddition it can react at other sites and Ot can bedestroyed by mobile reactive species such as H(kHOt nH) Mobile H can also react with othertrapped species (kHR nR) Incorporating these re-actions Eq A10 in the Appendix gives a steady-state yield Y` When the density of traps is largethe steady-state yield is given in Eq A5c

Y` R (sd9sd9sd) [kOOtkHR nr][kHOt kOt nt] N (5)

Here sd and sd9 are the dissociation cross sectionsthat determine the production of H (H 1 OH) andO (H2 1 O) by the incident radiation Comparingthe result in Eq 5 with Y` in the simple precur-sor model in Eq 4b the dissociation cross sec-tions in Eq 5 are now scaled by a ratio of the re-action rates

It is seen in Eq 5 that the steady-state yield Y`depends inversely on the density of traps nt Thatis there are competing destruction processes fortrapped O (Ot) Therefore in order to form O2 amobile O competes with mobile H to find atrapped O before it becomes trapped In this waythe increasing O2 yield correlates with the re-duction in the trapping sites with increasing tem-perature of formation The small activation ener-gies discussed earlier would be related to there-orientation of the lattice binding leading to thereduction in trap density At relatively high T thedensity of traps becomes small so that only the

OXIDANTS AT EUROPA 837

sample surface or grain interfaces might act astraps accounting for the observed drop in theyield at high T For this reason it is important todetermine if the yield for T greater than 150Kdepends on the incident flux f Since other quan-tities in the expression for Y` could also be tem-perature dependent experiments are neededThese would likely be pulsed radiolysis experi-ments in which the transients are studied ratherthan the steady-state yields However the avail-able data for lightly ionizing radiation do suggestthat the O2 yield depends inversely on trap den-sity as in the model above

Escape from depth

The quantity N in Eq 1 is the ldquodepthrdquo given asa molecular column density from which O2 formedbelow the surface can percolate to the surface andescape It can be converted to a physical depth bydividing by the molecular number density of theice In the case of the low-energy electrons thedepth below the surface at which ionizations andexcitations are produced is small However forfast ions that penetrate many monolayers this isnot the case Below we examine N in Eq 1

Although it has long been understood that twoexcitations events are required to produce O2(eg Brown et al 1982 Reimann et al 1984) theprecursor model suggests that at saturation(steady state) any new event can produce O2 fromthe steady-state density of trapped precursorsTherefore one might expect the yield at satura-tion to be nearly linear in the excitation cross sec-tion This is the case for the low-energy electronsand it was initially assumed to be the case for thefast ions (Brown et al 1980 Reimann et al 1984)However Baragiola et al (2003) showed that thetotal sputtering yield in the saturation region attemperatures greater than 100K which is dom-inated by decomposition (Brown et al 1982) isproportional to the square of the electronic en-ergy deposition per unit path length in the solid(dEdx)e Assuming that the production and de-struction cross sections in Y` in Eq 4b roughlyscale with (dEdx)e then their ratio [sp sO2s]also roughly varies as (dEdx)e If this is the casethen the observed quadratic dependence impliesthat the depth N from which O2 can be mobi-lized to escape must also depend on the excita-tion density (dEdx)e This is consistent with theobservation that the O2 yield depends on thick-ness for samples thinner than the ion penetration

depth (Reimann et al 1984 Benit and Brown1990) Therefore the percolation depth for escapeN is a critical factor in comparing yields for pen-etrating radiation with those for non-penetratingradiation like the low-energy electron and low-energy ion data It is also seen that percolationand trapping can be included within the simplemodel considered here

Model summary

We have examined the simple precursor modelintroduced by Sieger et al (1998) to explain theirelectron data We have shown how it can be gen-eralized and derived from an integration of theappropriate chemical rate equations with thecross sections and total column density contain-ing the reaction rates and the physics of diffusivetransport Therefore the parameters can havevery different meanings from those initially as-sumed The equations are useful when there arefast processes which go to completion in timesmuch shorter then the time between particle im-pacts and slow processes associated with speciesthat are assumed to be trapped between particleimpacts This separation allowed us to examinethose models proposed for precursor formationand destruction due to reacting mobile speciesformed from dissociation events Further fromthe size of the onset of the linearity of the yieldversus fluence we can place constraints on therole of nonlinear processes such as those thatmight be associated with the formation of a pre-cursor like peroxide

With these extensions this simple model is use-ful for approximating the full rate equations inthe temperature region for which the yields de-pend on the fluence (dose) and not on the doserate Based on the assumptions in the model thisimplies temperatures at which precursor speciesare stable over experimental times At the high-est temperatures for the yields shown in Fig 4athis might not be the case This should be testedexperimentally In this regard it is puzzling thatexperiments have shown that the absorptionbands associated with certain trapped speciessuch as OH and trapped electrons ldquodisappearrdquoat 100ndash120K (see eg Johnson and Quickenden1997) whereas the O2 yield data indicate that rel-evant precursor species must be stable to 140KTherefore the proposed O2 production processescan only be distinguished by directly determin-ing the chemical state of the irradiated ice

JOHNSON ET AL838

It may be the case that a particular precursoris not required as mentioned earlier Rather af-ter a sufficient level of oxidation a newly pro-duced O or OH can react in the oxygen-rich lat-tice producing O2 in a pseudo-first-order processThis can be tested experimentally Indeedtrapped O and peroxide may be changed into oneanother as in the matrix isolation experiments(Khriachtchev et al 2000) Therefore their rela-tive concentrations versus temperature and doseshould be studied

In the model considered here it is seen that theparameters in Y` must account for the strong tem-perature dependence Since the number of bulktrapping sites decreases with increasing T as theice sample is annealed one might expect that theprecursor concentration would decrease We sug-gest that the observed temperature dependenceis determined by the competition between theavailability of trapping sites and mean interactiondistances of a mobile species such as O and HThat is the O2 yield varies inversely with the den-sity of traps over a range of temperatures In thesimple model above this effect is included indi-rectly as shown using the more detailed equa-tions in the Appendix

EUROPA

Oxygen and peroxide production at Europa G values

The dependence of the yield on the thicknessof the ice sample for highly penetrating ions sug-gests that the O2 yield is roughly related to theamount of energy deposited in the sample There-fore the amount of O2 produced and lost fromice in steady state is often given as a G value thenumber of molecules produced per 100 eV de-posited in Table 2 In applying these G values tothe planetary environment care must be takensince the yields are temperature fluence and ex-citation density dependent They also are affectedby the presence of other proton scavengersknown to be present in Europarsquos ice

The G values for peroxide in Table 2 are thosefor production at low fluence but the G valuesfor O2 are for the production and escape of gas-phase oxygen at saturation Therefore they de-pend on the formation and destruction cross sec-tions in the model above Such G values can beused to describe the steady-state source of the am-

bient O2 atmosphere at Europa but are not help-ful in determining the ability to produce and trapO2 at depth into the icy surface Over the timescales of the experiments some O2 is seen to re-main trapped and is not destroyed by radicalsTypically there are a number of trapping sitesand recent estimates of the activation energy foran O2 molecule trapped in a substitutional site(Hori and Hondoh 2002) suggest that it can betrapped for times comparable to the time re-quired for the conversion of amorphous ice intoIh (Ayotte et al 2001) The stability of O2 inclu-sions has not been studied

To understand the relationship between the Gvalues for O2 in Table 2 we note that fast lightions and electrons lose their energy predomi-nantly by producing ionizations and excitationsThe ionization events produce secondary elec-trons that can result in further ionizations Thenumber of ionization events Ni produced by afast light ion stopping in a water ice is Ni 5EoW where Eo is the initial energy and W is theaverage energy expended per ionization event27 eV in water ice Therefore radiolysis by fastions is often described as the net effect of theshower of electrons set in motion by the incidention In this manner the G values for productionof O2 from ice by energetic light ions and elec-trons can be roughly estimated as the sum of theeffects of all the secondary electrons using theyield Ye for 10ndash100 eV electrons That is if weignore the percolation depth N we can write thenumber of O2 produced as the number of ioniza-tion events times Ye which already has in it thedestruction processes That is G NiYeEo YeW 0006 O2100 eV

This estimate of G is seen to be an order of mag-nitude larger than the keV proton value for O2 inTable 2 but is much smaller than the G value forenergetic Ne1 These differences from the simpleestimate are consistent with the observation thatthe yield in the saturation regime for penetratingions varies nonlinearly with excitation densityThe difference from the keV proton G value islikely due to the inability of O2 to escape fromdepth before destruction On the other hand thelarger G value seen using MeV Ne1 is likely dueto the high density of excitation This can resultin more efficient production of O2 increased pro-duction of damage sites for forming O2 andorincreased transport (percolation) along the tran-siently heated track That is the ability of theformed O2 to percolate to the surface during ion

OXIDANTS AT EUROPA 839

bombardment plays an essential role in deter-mining the measured yields as discussed aboveThis effect needs to be evaluated for a planetarysurface subject to particle fluxes that are many or-ders of magnitude smaller than those used in thelaboratory

In the absence of O2 yields for all of the rele-vant ions and energies an estimate of the totalsupply of O2 can be obtained using the G valuesin Table 2 For the energetic oxygen and sulfurions we use G values for neon and argon ionsThese are lower bounds as the data were obtainedat temperatures much lower than the surface tem-peratures at Europa (80ndash110K) For the fast pen-etrating electrons and protons we use the valuefor 200 keV protons in Table 2 For the heavy ionswe used the Ne1 and Ar1 values For compari-son we also give an estimate using the G valuesobtained from the low energy electron data andthe total ionization as discussed above

The G values are combined in Table 3 with theelectron and ion energy flux at Europa Althoughthere is some deflection of the flow due to in-duced fields the net energy flux to the surface isnot reduced much from these values (Paranicaset al 2002) In fact locally produced pick-up ionswhich are formed from the ejecta and acceleratedto the surface can contribute (Ip 1996) Furtheralthough G values are typically independent ofthe incident angle the loss of O2 is a surfaceprocess and is enhanced with angle That is thenumber of O2 produced within the escape depthincreases with increasing angle to the normal(steeper than the inverse cosine) This enhance-ment is 25ndash4 (Jurac et al 2001) Ignoring suchcorrections the total production rate calculated

(23ndash62 3 109 O2cm2s) is a lower bound to theactual production rate from the icy regions of Eu-roparsquos surface Recent modeling of the O2 atmo-sphere suggest that such a source rate is required(Shematovich and Johnson 2001)

Earlier estimates of the O2 source rate were in-correct because O2 production was assumed to bea fixed fraction of the H2O sputtering yield Atthe ambient temperature an ejected H2O sticks ef-ficiently (Smith and Kay 1997) to the surface ofneighboring grains in the porous regolith reduc-ing the sputtering yield for H2O (Johnson 1989)This reduction does not apply to the O2 fractionof the yield enhancing its relative importance toH2O Here we assume that all of the gas-phaseO2 produced can contribute to the ambient at-mosphere even if it is produced at a grain sur-face that is at some depth into the porous regolith

Finally Europarsquos surface is not pure ice every-where That is much of the trailing hemisphereis a hydrated sulfuric acid containing smallamounts of sulfur and SO2 and the leading hemi-sphere contains low levels of carbon CO2 andcarbonates Sodium and potassium atoms aresputtered from the icy regions (eg Leblanc et al2002) O2 production from surfaces containingthese species has not been measured C and S areboth proton scavengers and the other species canbe sources of O as discussed so that the O2 yieldfrom such regions could be enhanced Howeverthere can also be competing processes that mightreduce these rates In fact recent HST observa-tions suggest that the O2 might not be produceduniformly across Europarsquos surface (McGrath et al2000) The analysis of these observations is pre-liminary and is not simple since the observed

JOHNSON ET AL840

TABLE 3 APPLICATION TO EUROPA

OxygenImplantation Energy flux production

Species 109(cm2 s)a 109 keV(cm2 s)(1) 109(cm2 s)

H (10 keV) 0015(1) 12 0084 (072)bO (10 keV) 0015(1) 18 054cS (10 keV) 0009(1) 30 090cElectrons 62 043 (37)bPlasma (10 keV) (137 amu) 13(2) 10 03c

Total 79 23 (62)UV photons (64 eV) 46 01

aReferences are as follows 1Cooper et al (2001) [similar results in Paranicas et al (2002)] 2Mauk et al (1996)bUsing G for 200 keV protons (upper bounds in brackets are based on the low-energy electrons data as a model for

the secondary electrons produced)cUsing G for 15 MeV Ne1 at 7K a rough lower bound

emissions depend on the local plasma propertiesHowever the efficiency of O2 production in con-taminated ices needs to be studied

Porous regolith O2 interactions on grain surfaces

Although the O2 atmosphere is principally pro-duced in the near surface layers and is eventuallylost to space it also permeates Europarsquos porousregolith The porosity p (void space) of the re-golith is uncertain (Buratti 1995) It has been es-timated to be the most compact surface in theOuter Solar System p 025 on the trailing hemi-sphere and p 079 on the leading hemisphere(Buratti and Veverka 1983) and the most poroussurface p 096 globally (Domingue et al 1991Domingue and Hapke 1992) These large differ-ences are due to the uncertainties in the amountof backscattered light at small phase angles andon the photon absorption properties and grainsizes The more compact surface is likely correctGrain sizes rg of the order of 100 mm have beensuggested (Hansen and McCord 1999 2000Geissler et al 1998) Europarsquos regolith differs sig-nificantly from the lunar regolith which is asand-like surface of individual grains Analysis ofthe thermal inertia of Europarsquos surface (Grundyet al 2001) suggests that the ice grains are sin-tered forming a fairy castle-like structure Sin-tering can occur thermally or due to the incidentradiation In addition ices doped with acids andalkalis in the laboratory can be powdery solidswith the impurities primarily found at grain in-terfaces

The porous regolith affects the above estimatesof the oxygen production in three ways First forisotropic bombardment the mean distance d intothe surface at which an incident ion or photonstrikes a grain surface is d (89)rg(1 2 p) Us-ing the above value for rg then d 02 m if theporosity is very large p 96 Therefore de-sorbed molecules can come from a considerabledepth and will interact with grain surfaces Evenfor much lower porosities the effective sputteringyields can be significantly modified (Johnson1989) As discussed above the yields for ejectedH2O which sticks onto neighboring grains at theambient temperatures are reduced but those forejected O2 are not This enhances the relative im-portance of the O2 yields

Finally the surfaces and interfaces of the sin-tered grains in Europarsquos porous surface are placeswhere low temperature chemistry involving the

ambient oxygen can occur at depths that are wellbelow the nominal penetration depth of the inci-dent radiation The relative importance of suchchemistry is roughly indicated by the ratio of thetotal surface area in the regolith to the area of Eu-roparsquos visible surface Using the porosity andgrain size above and a regolith depth dr of theorder of a meter then this ratio is large 3(1 2p)drrg which is 103ndash2 3 104 for average porosi-ties of 96 and 25 respectively Therefore thegas-phase O2 molecules below the visible surfaceinteract frequently with the grain surfaces In thisway a very large volume of the subsurface couldbe altered chemically even if it is not directly ex-posed to the radiation

From the overlying structures seen in the high-resolution images of Europarsquos surface by theGalileo spacecraft (eg Prockter and Pappalardo2000) it is clear that the youngest surfaces are red-der The reddening in the visible has been attrib-uted to the presence of sulfur in the ice (McEwen1986 Johnson et al 1988 Calvin et al 1995 Carl-son et al 1999b) The amount of reddening in thespectra of the surface material decreases with ageand the surface materials appear to becomebrighter with age This might occur by the slowaccumulation of a frost over-layer or by radioly-sis (Carlson et al 2002 Johnson et al 2003) Ra-diolysis preferentially converts the sulfur in theice to sulfur dioxide or sulfuric acid both ofwhich have flat spectra in the visible In regionswhere the radiation does not directly penetratethe O2 permeating the regolith can oxidize thesulfur at the surfaces of the icy grains and at graininterfaces Similarly the carbon seen in the ice onEuroparsquos leading hemisphere (eg Carlson et al2004 Johnson et al 2003) can be oxidized by ra-diolysis in the near surface layers or by reactionson the grain surfaces at depth in the regolith bythe ambient O2 The meteoroid bombardmentthat produces the regolith also exposes fresh sur-faces for oxidation and can cause the collapse ofpores leading to the trapping of oxygen In thismanner both refractory and volatile oxidants canbe formed and trapped in the subsurface ice forpotential delivery to Europarsquos subsurface ocean

SUMMARY

The presence of a thin oxygen atmosphere andoxidants trapped in Europarsquos surface ice (egJohnson et al 2003) combined with the possible

OXIDANTS AT EUROPA 841

presence of a subsurface ocean (eg Carr et al1998) have made Europa an exciting target forfuture space exploration The molecular oxidantsproduced by radiolysis and photolysis could bespecies needed for carbon-based biochemistry inEuroparsquos putative subsurface ocean Howeverthere has been no model for the formation trap-ping and transport of relevant oxidants

In this paper we have reviewed the space ob-servations and laboratory studies related to thechemistry of formation of oxidants in ice by theincident radiation Because the laboratory dataare incomplete we examined the rate equationsfor formation of oxygen and related species byradiolysis and photolysis We showed how suchequations can be simplified to the analytic modelthat was used for fitting the fluence and temper-ature dependence of the O2 yield produced bylow-energy electrons (Sieger et al 1998) In de-riving the analytic model from the more completerate equations the competition between forma-tion and destruction processes can be studied ascan the role of trapping sites escape from depthand temperature history These equations also al-lowed us to put incident photon electron and iondata on a similar footing for the first time

The model can be used to suggest measure-ments that are needed to describe the chemicalstate of the irradiated ice and to interpret new lab-oratory data We showed that if reactions are fastthe yield data could be consistent with a precur-sor being formed by two events as is the case forperoxide However we favor trapped O as a pre-cursor as initially suggested by Matich et al(1993) for production of excited O2 Such aspecies if present could be detected spectro-scopically as can trapped O2 through bands ac-tivated by lattice perturbations (Cooper et al2003a)

The available oxygen yield data were used tocompare G values for different incident particlesThese were then used to estimate the oxygen pro-duction in the radiation environment at EuropaThe lower bounds were consistent with the pro-duction rates obtained from a recent model of Eu-roparsquos atmosphere The porosity of the regolithwas also considered It not only affects the rela-tive sputtering yields for O2 and H2O but alsoprovides an enormously increased surface areaon which the ambient gas-phase oxygen can re-act with species in the ice That this might be oc-curring is consistent with the change in re-

flectance associated with surface aging Such re-actions could also convert ambient molecularoxygen into other more refractory species that re-main trapped on subduction of Europarsquos crust

Although radiolysis produces a stable chemi-cally altered surface in the laboratory the parti-cle flux to Europarsquos surface is many orders ofmagnitude lower Therefore annealing can be ef-fective at temperatures that are much lower thanthose typically found experimentally Belowabout 120K the annealing times appear to be long(eg Baragiola 2003) so that laboratory data canoften be directly applied to Europarsquos icy surfaceHowever account has to be taken of the syner-gism between the UV and the energetic plasmaas discussed earlier (Johnson and Quickenden1997)

Transient surface melting might initiate lifeprocesses but subduction of oxidants to the pu-tative ocean is likely to be required if biochem-istry is to occur at Europa (Chyba and Hand2001) Although the temperature versus depthinto Europarsquos regolith is not well known belowthe regolith the temperature and pressure appar-ently increase with depth At some depth (ap-proximately kilometers) either very warm ice oran ocean exists (Ruiz and Tejero 2000) Thereforechemically refractory oxidants are likely to be re-quired in order to survive subduction to thewarmer subsurface regions Trapped species ifdelivered downward can become mobile andwill segregate forming inclusions (Johnson andJesser 1997) Indeed peroxide appears to segre-gate in ice (Gurman et al 1967) and these inclu-sions may remain relatively stable until the per-oxide is dissolved in the putative ocean Thevolatiles seen in reflectance such as O2 SO2 andCO2 will also segregate But at the higher tem-peratures they could also diffuse through thebulk react or percolate into the regolith atmo-sphere Therefore the observation of the trappedvolatiles in Europarsquos surface ice is of interest pri-marily because it suggests that oxidants are pre-sent The formation of oxygen-rich molecules fol-lows the preferential loss of hydrogen Howeverthe surface is in a local steady state with im-planted H O and S from the plasma and the lossto space Since the plasma flux to the surface canbe variable this balance can also vary Indeedtemporal changes in reflectance have been re-ported (Domingue and Hendrix 2004)

Based on the models for formation of O2 ex-

JOHNSON ET AL842

amined in this paper additional experiments areneeded to determine the chemical state of irra-diated ice at low fluences Of immediate inter-est is the identification of stable molecular pre-cursors required for the formation of O2 Basedon the available laboratory data individual per-oxide molecules trapped in the ice do not appearto be the principal precursors but trapped O(Matich et al 1993) remains a viable candidateAlthough the data suggest that the local hydro-gen-bonding network is important in the pro-duction of O2 the effect of the formation tem-perature and radiation damage on the structureneed to be measured It is also important to de-scribe quantitatively the relationship betweenH2 loss and the formation of O2 and H2O2 overa range of temperatures and fluences Finallyone would like to determine the steady-statedensities of biochemically interesting moleculesformed by radiolysis in an ice containing sulfurand carbon at Europarsquos surface temperaturesand then determine their stability with increas-ing temperature and pressure Such experimentsstill need to be carried out Here we have pre-sented a model that brings together the variouslaboratory data and now can be used to considerthe formation trapping and transport of relevantoxidants at Europa

APPENDIX

Introduction

In this Appendix we examine rate equationsthat describe various aspects of the radiolysis orphotolysis of ice These equations need to be un-derstood if one is to be able to calculate the for-mation trapping and transport of oxidants Theemphasis is on the formation of molecular oxy-gen We integrate these equations over the escapedepth of O2 to give the simplified rate equationsdiscussed in the text in which number densitiesare replaced by column densities (eg Eqs 1 and2) In this way we can examine the relationshipbetween the rate constants and the effective pre-cursor formation and destruction cross sectionsextracted from data We first consider a precur-sor such as H2O2 that is formed from two exci-tation events This bears on the observed linear-ity of the oxygen yield versus fluence We thenconsider the equations for trapped versus a tran-

siently mobile species We use a reaction betweena mobile and trapped O as an example althoughO 1 OH can also give oxygen Finally we ex-amine the possible dependence of the O2 forma-tion process on the density of traps (defectsvoids and grain boundaries) as a function of tem-perature The results are discussed in the text

For clarity all possible radiolytic species arenot included in the discussion below Thereforepotentially important products such as HO2 andthe charged species (eg H3O1 the trapped elec-tron etc) are missing In each section we attemptto make a point related to production of O2 andother oxidants by using a subset of the productsrather than attempt a complete description withmany unknown rate constants

Fast versus slow processes

To describe the O2 production rate from a pre-cursor such as H2O2 we first use a simple but in-structive set of rate equations to describe the for-mation of H2O2 We then relate these to thesimplified rate equation for the precursor in thetext Eq 1 and show how the postulated precur-sor formation cross section sp is related to theH2O dissociation cross section sd by the reactionrates

Using the reactions

H2O 1 radiation R H 1 OH (sd)

H 1 H R H2 (k2H)

H 1 OH R H2O (kHOH)

OH 1 OH R H2O2 (k2OH)

the reaction pathways are described by

H2

Hqk2H

1 OHH2O 1 radiation mdashRsd H

OHH2O2

OH Rk2OH

kHOH

HH2O

OXIDANTS AT EUROPA 843

The corresponding rate equations in an irradiatedvolume are

dnOHdt 5 sd f n 2 2 k2OH nOH nOH

2 kHOH nH nOH (A1a)

dnHdt 5 sd f n 2 2 k2H nH nH

2 kHOH nH nOH (A2a)

dnH2O2dt 5 k2OH nOH nOH

2 destruction processes (A3a)

Here sd is the dissociation cross section f is theradiation flux and n is the molecular numberdensity of ice with nOH and nH the density of dis-sociation products OH and H respectively Therate constants k2OH k2H and kHOH describe thereactions above In this model the H2 formed isassumed to escape for temperatures above about20K We do not distinguish between trappedspecies and mobile species although in fact thatdistinction is important as discussed in the nextsection That is any OH produced at low tem-perature by one ion would likely be trapped priorto interacting with an OH produced by a subse-quent ion At higher temperatures this can breakdown as discussed in the text We first relate theabove set of equations to Eq 1 in the text and thenconsider precursor destruction processes

Equation 1 in the text is obtained by solvingthese equations together and then by integratingover a depth Dx Here Dx is the smaller of the pen-etration depth of the particle and the O2 escape(percolation) depth Assuming the molecular den-sity of ice n is constant with depth Dx n 5 N inEq 1 Comparing Eq 1 with Eq A3a the precur-sor formation rate sp f N is equivalent to [k2OH

nOH nOH] Dx Assuming that the processes in EqsA1a and A2a are fast then the steady state is read-ily obtained In addition if k2H k2OH there ispreferential loss of H2 at short times as suggestedby the laboratory data If steady state is reachedthe production of H2 is equal to that of H2O2 us-ing only the equations above Further one obtains

[k2OH nOH nOH] Dx R [fp sd] f N (A4)

with

fp 5 05[1 1 05 kHOH(k2H k2OH)05] (A5a)

Therefore the precursor formation cross sectionsp in Eq 1 is a fraction fp of the total dissocia-

tion cross sd Since laboratory data suggest theH2O2 fraction in an irradiated ice is small thenkHOH (k2OH k2H)05 which is likely givingfp R [(k2OH k2H)05kHOH]

Trapped versus mobile species

To examine the fluence dependence discussedin the text we note that trapped OH and freshlyproduced transiently mobile OH can be treatedas different species (eg Matich et al 1993)Therefore assuming no trapping of H the OHrate equation above would be replaced by twoequations one for the trapped species nOHt andthe other for the freshly produced mobile speciesnOH Assuming the reaction of two mobile OHunits is unlikely and that peroxide is formed fromthe interaction of a trapped OH with a subse-quently produced OH an expanded set of equa-tions is obtained

OH 1 trap R OHt (kOHt)

OH 1 OHt R H2O2 (kOHOHt)

H 1 OHt R H2O (kHOHt)

then the reaction pathways are described by

qH2

H2O 1 radiation mdashRsdH

qk2H

1 OHH

trap OH

OH RkOHt

OHt kOHOHtH2O2

kHOHkHOHt

H H2O HH2O

The resulting set of equations is

dnOHdt 5 sd f n 2 kHOH nH nOH

2 kOHt nOH nt 2 kOHOHt nOH nOHt (A1b)

dnOHtdt 5 kOHt nOH nt 2 kOHOHt nOH nOHt

2 kHOHt nH nOHt (A1b9)

dnHdt 5 sd f n 2 2 k2H nH nH 2

kHOH nH nOH 2 kHOHt nH nOHt (A2b)

JOHNSON ET AL844

dnH2O2dt 5 kOHOHt nOH nOHt

2 destruction processes (A3b)

where nt is the trap density in the irradiated iceAt low T we can assume nt is large so thatdntdt 0 simplifying the set of equations If theprecursor of interest is trapped O as discussed inthe text then a similar set of equations can be con-structed for O and Ot as in the example to followOf course a full description over a broad rangeof temperatures would require the considerationof the both trapped and mobile H O and OHthe rate of change of trap density and the vari-ous destruction processes

The above equations should be solved simul-taneously but for the incident particle fluxes typ-ically used in these studies the diffusion of mo-bile species to traps is relatively fast This is trueeven at low temperatures if there is a significantdensity of trapping sites This is usually the casefor samples formed at low T Therefore in a lowtemperature solid at low dose rates there is not asignificant steady-state background of diffusingOH There are trapped or freshly produced OHunits which rapidly react Averaging over thebombardment rate this can be approximated bywriting dnOHdt 0 and then separating nOH outin Eq A1b Assuming also that H diffuses rapidlyand either reacts or forms H2 we also writednHdt 0 on the average Therefore betweenimpacts only species trapped at defects pores orgrain surfaces are assumed to be present as sug-gested by the lack of a dependence on the inci-dent beam flux In this limit the rate of change oftrapped OH roughly simplifies to

dnOHtdt 5 2 k2H nH nH 2 2 kOHOHt nOH nOHt

It is seen that the production of peroxide dependson the loss of H by formation of H2 as discussedand at steady state the production rate for per-oxide equals the rate of formation of H2 in theseequations

Treating the mobile and trapped species sepa-rately we can further simplify the rate equationsand obtain the precursor model used in the textThat is integrating over the relevant depth Dxthe densities are replaced by column densitiesThe rate constants ki are then related to effectivecross sections si In this way cross sections aregiven as an interaction length for each freshlyproduced mobile radical [ki nj] R [si f] The

trapped OH column density NOHt and peroxideproduction rate can be estimated from ldquolinearrdquorate equations

dNOHtdt lt [sd f] N 2 2 [s2OH f] NOHt

2 [sHOH f] NOHt (A6a)

dNH2O2dt lt [s2OH f] NOHt 2 s NH2O2 (A6b)

Here sd is again the dissociation cross sectionand the destruction processes in Eq A3b are con-tained in s The second and third terms on theright in Eq A6a are the loss of OH due to the pro-duction of H2O2 and H2O In the integration ofEq A1b9 [2 kOH OHt nOH] becomes [2 s2OH f] and[kHOHt nOH] becomes [sHOH f] Therefore thedensity and the interaction length of mobile H arecontained in sHOH If H2O2 is indeed the pre-cursor then in Eq 2 in the text [sp f N] is be-comes [s2OH f NOHt] from Eq A6b Accountingfor precursor destruction which can give backtwo OH and treating H2 can both be included inthe integrated model via the si values

The simplified set of equations above allows usto examine the relative time (fluence) scales forprecursor formation (in this case H2O2) SolvingEqs A6a and A6b the column density of OH andthe concentration of precursors versus fluence cpcan be obtained Using the column of H2O2 pro-duced as the precursor column Np in Eq 2 thenO2 yield can be obtained The results can be com-pared with the results from Eqs1 and 2 in thetext The average precursor (H2O2) concentrationover the column depth of water molecules N is

cp 5 fp (sds)[1 2 exp(2s F)] 2 (ssr)

[1 2 exp(2sr F)] (A7)

with

fp 5 2s2OH(sr 2 s) (A5b)

In these equations sr 5 (2s2OH 1 sHOH) andssr 1

It is seen that the precursor concentration withtime cp depends on two rates the reaction ratesthat determine sr and the destruction rates in sAs in Eq 4a the yield can be written as

Y 5 sO2 cp N (A8)

If sr and s are comparable then the yield at small

OXIDANTS AT EUROPA 845

fluence is quadratic in F However if sr sthen the linear dependence on fluence is obtainedwith a small shift in the fluence as shown in Eq4c in the text Further the steady-state yield isnow

Y` (T) 5 sO2 [fp sd] Ns (A9)

That is sp R [fp sd] as in Eq A4 and the tem-perature dependence for the incident electrondata must be contained in [sO2 fp] or in the ratio[sO2 s2OH]sHOH]

For comparison with their precursor modelSieger et al (1998) examined the possibility of ob-taining the O2 precursor by considering reactionsfor a density of diffusing dissociation products asin Eqs A1 and A2 rather than a trapped productinteracting with a freshly formed dissociationproduct Based on the above the lowest fluence atwhich O2 is first detected experimentally sets alimit on the time scale for mobilendashmobile reactions

Effect of trap density

Equations A1b A1b9 A2b and A3b can be ap-proximately solved assuming that the dnOHdtdnHdt and dntdt are very small on the averageas discussed above In order to examine the roleof the trap density nt which can depend on thetemperature and method of formation and on theradiation history of the ice we use the slightlysimpler precursor model In this model the O pro-duced directly by dissociation traps forming aprecursor O2 is then formed when a freshly pro-duced O or other species reacts with a trappedoxygen atom Ot

H2O 1 radiation R H2 1 O (sd)

O 1 trap R Ot (kOt)

O 1 Ot R O2 (kOOt)

The reaction pathways for the oxygen species are

H2O 1 radiation mdashRsd9

trap OO2

kOtOt kOOt

H2 1 OOHt

kOR H(kHO t)

ROR

We assume for simplicity both H2 and O2 escapefrom depth Dx after formation We also allow mo-bile H to react with Ot (kHOt) removing trappedO and both the mobile H and O to be removedby other reactants R (eg OH or contaminants)This leads to a simplified set of equations not di-rectly involving OH or peroxide

dnOdt 5 sd9 f n 2 kOt nO nt 2 kOR nO nR

2 kOOt nO nOt (A1c)

dnOtdt 5 kOt nt nO 2 kOOt nO nOt

2 kHOt nH nOt (A1c9)

dnHdt 5 sd f n 2 kHOt nH nOt

2 kHR nH nR (A2c)

dnO2dt 5 kOOt nO nOt (A3c)

These equations of course act in parallel with theequations above for OH and mobile OH may re-act with trapped O to form O2 Here we focusonly on mobile O

Since the diffusion of H and O are fast and thedensity of traps is large the nH and nt averagedover the bombardment rate can be taken veryroughly as time independent Further if we as-sume the number of trapped O nOt is small com-pared with the number of sites at which H canreact then nH [sd f nkHR nR] For a low ra-diation flux f nH is small on the average as as-sumed Since the O2 yield YO2 is equal to[(dnO2dt) Dxf] with Dx the escape depth thenthe steady-state yield Y` can be obtained ana-lytically

[Y`][sd9 N] 5 1 1 (q 2 1) 2 [(q 2 1)2

1 8 q d]054 d (A10)

q 5 [kOt (nt 1 nR)] (kHOt nH)(sd9 f n) kOOt

5(sdsd9) [kOt (nt 1 nR)] kHOt[kOOt kHR nR]

d 5 05 [1 1 nt(nt 1 nR)]

We note that for nt large q can be large Assum-ing that q 1 and nt nR then Eq A10 be-comes

Y` R [sd9 N]q 5 (sd9 N) (sd9sd)

[kOOt kHR nR][kHOt kOt nt] (A5c)

JOHNSON ET AL846

Comparing with Y` in the simple precursormodel in Eq 4b then s R [(kHOt nH)f] R [sd nkHOt][kHR nR] In addition the product [sO2sp] R sd9 sd9 [kOOt n][kOt nt] That is the dis-sociation cross sections are modified by reactionrates

Quite remarkably it is seen in Eq A5c that thesteady-state yield Y` can depend inversely onthe density of traps nt in the temperature andflux regimes for which the approximations arerelevant That is if there are competing destruc-tion process for trapped O then at high trap den-sity a mobile O must find a trapped O before itbecomes trapped or before mobile H destroys theOt In this way the increasing O2 yield with in-creasing T can be related to the change in the trap-ping density nt with the formation temperatureof the ice sample As the temperature increasesand the density of traps in the bulk decreases thesurface and grain interfaces become the only trap-ping sites

ACKNOWLEDGMENTS

REJ acknowledges support by a Gledden Fel-lowship in Chemistry School of Biomedical andChemical Sciences at The University of WesternAustralia and by grants from the NSF Astron-omy Program and NASArsquos Planetary Geologyand Geophysics Program PDC acknowledgessupport by an Australian Postgraduate Award

ABBREVIATIONS

HST Hubble Space Telescope MJQ MatichndashJohnsonndashQuickenden NIMS Near Infrared Map-ping Spectrometer UV ultraviolet

REFERENCES

Ayotte P Smith RS Stevenson KP Dohnarsquolek ZKimmel GA and Kay BD (2001) Effect of porosityon the adsorption desorption trapping and release ofvolatile gases by amorphous solid water J Geophys Res106(E12) 33387ndash33392

Bahr DA Famaacute M and Baragiola RA (2001) Radiol-ysis of water ice in the outer solar system sputteringand trapping of radiation products J Geophys Res 10633285ndash33290

Baragiola RA (2003) Microporous amorphous water icefilms and astronomical implications In Water in Con-

fining Geometries edited by JP Devlin and B Buch El-sevier New York in press

Baragiola RA and Bahr DA (1998) Laboratory studiesof the optical properties and stability of oxygen onGanymede J Geophys Res 103 25865ndash25872

Baragiola RA Atteberry CL and Bahr DA (1999) Ori-gin of solid oxygen on Ganymede Reply to commentby RE Johnson on ldquoLaboratory studies of the opticalproperties and stability of oxygen on Ganymederdquo JGeophys Res E104 14183ndash14187

Baragiola RA Vidal RA Svendsen W Schou J ShiM Bahr DA and Atteberrry CL (2003) Sputteringof water ice Nucl Instrum Methods Phys Res B 209294ndash303

Bar-Nun A Herman G Laufer D and Rappaport ML(1985) Trapping and release of gases by water ice andimplications for icy bodies Icarus 63 317ndash332

Baverstock KF and Burns WG (1976) Primary pro-duction of oxygen from irradiated water as an expla-nation for decreased radiobiological oxygen enhance-ment at high LET Nature 260 316ndash318

Benit J and Brown WL (1990) Electronic sputtering ofoxygen and water molecules from thin films of waterice bombarded by MeV Ne1 ions Nucl Instrum Meth-ods Phys Rev B 46 448ndash451

Benit J Bibring J-P della-Negra S LeBeyec YMendenhall M Rocard F and Standing K (1987)Erosion of ices by ion irradiation Nucl Instrum Meth-ods Phys Rev B 1920 838ndash842

Brown WL Augustyniak WM Lanzerotti LJ John-son RE and Evatt R (1980) Linear and nonlinearprocesses in the erosion of H2O ice by fast light ionsPhys Rev Lett 43 1632ndash1635

Brown WL Augustyniak W M Simmons E Mar-cantonio KJ Lanzerotti LJ Johnson RE BoringJW Reimann CT Foti G and Pirronello V (1982)Erosion and molecular formation in condensed gasfilms by electronic energy loss of fast ions Nucl In-strum Methods B 1 307ndash314

Brown WL Augustyniak WM Marcantonio KJ Sim-mons EN Boring JW Johnson RE and ReimannCT (1984) Electronic sputtering of low-temperaturemolecular solids Nucl Instrum Methods B 1 307

Buratti B (1995) Photometry and surface structure of theicy Galilean satellites J Geophys Res 100(E9) 19061ndash19066

Buratti B and Veverka J (1983) Voyager photometry ofEuropa Icarus 55 93ndash110

Calvin WM and Spencer JR (1997) Latitudinal distri-bution of O2 on Ganymede observations with the Hub-ble Space Telescope Icarus 130 505ndash516

Calvin WM Clark RN Brown RH and Spencer JA(1995) Spectra of the icy Galilean satellites from 02 to5 mm a compilation new observations and a recentsummary J Geophys Res 100 19041ndash19048

Calvin WM Johnson RE and Spencer JA (1996) O2

on Ganymede spectral characteristics and plasma for-mation mechanisms Geophys Res Lett 23 673ndash676

Calvin WM Aninich VG and Brown RH (2004) Oxy-gen temperature and phase effects Icarus (in press)

OXIDANTS AT EUROPA 847

preference for forming oxygen-rich molecules inEuroparsquos ice and can affect the efficiency of pro-duction of O2 and H2O2

MODELS FOR FORMATION OF O2

Introduction

Based on the data described above the yieldsfor incident electrons and ions are found to benearly linear in fluence at the lowest fluences forwhich O2 is detected In addition the relevant al-terations of the ice samples by radiolysis appearto be stable and independent of the beam fluxRemarkably this is the case both for low-energyelectrons that make only a single excitation foreach impact and fast ions that make a density ofexcitations along their path through the solidThis suggests that O2 is produced and trapped inthe initial events and then caused to escape fromthe solid by the subsequent radiation or that astable chemical alteration of the ice is required be-fore the radiation can produce and eject O2

Because H produced by the dissociation of wa-ter molecules can attack trapped O2 and any pre-cursor oxidants the loss of H is important Thisoccurs by the production and loss of hydrogenmolecules as discussed leaving a surface that isslightly rich in O relative to H The O2 produc-tion mechanism could be direct excitation of aparticular precursor species or it might not bespecific That is a new OH or O produced by theincident radiation could react with the H-depleted solid producing O2 This would occurmore efficiently as the ratio of O to H in the nearsurface region increased

In their early work WL Brown and colleaguesassumed that processes similar to those used todescribe radiolysis of liquid and gaseous H2Ocould be used to describe the radiolysis in ice Thelow activation energy that they found was as-sumed to be associated with diffusion of somemobile radical species in ice such as H or O orpossibly diffusion of the O2 molecule itselfBased on their studies of ejection from isotopi-cally labeled D2O samples they concluded thatthe O2 is formed in the track of a single ion andthen ejected by subsequent ions (eg Benit andBrown 1990) More recently trapped specieshave been suggested as precursors These are ei-ther subsequently excited by the incident radia-tion or attacked by newly produced mobile

species We consider a simple model that incor-porates all of these processes and can be used toorganize and interpret the experimental data Wealso discuss its relationship to more completemodels In this discussion much of which is in theAppendix we consider the standard radiationproducts that have been often described How-ever in a solid the excited electron cloud pro-duced by absorption can overlap a number of lat-tice sites Therefore their relaxation pathwaysandor product species could be very differentfrom those that have been traditionally consideredin the literature on radiolysis and photolysis

Precursor model

The column density of a precursor species Npformed in ice by incident radiation can be calcu-lated using a simple model In this model cross sec-tions are used for the production sp and destruc-tion s of the precursor O2 is then formed byexcitation of the precursor or by reaction of a newlyformed radical with the precursor This is also de-scribed using a cross section sO2 For incident elec-trons having a small penetration depth the O2 pro-duced is assumed to rapidly diffuse to the surfaceand escape (Sieger et al 1998) More generally wedefine N as the column density of H2O from whichthe incident radiation can produce and eject O2 Apair of rate equations gives both Np and the pro-duction rate of oxygen dNO2dt That is

H2O 1 radiation R precursor

precursor 1 radiation R O2

giving

dNpdt 5 sp f N 2 s f Np (1)

dNO2dt 5 sO2 f Np (2)

In Eq 1 f is the incident flux and the total pre-cursor destruction cross section can be written ass 5 sp9 1 sO2 with sp9 accounting for precursordestruction processes other than the formation ofO2 Solving Eqs 1 and 2 and assuming N is nearlyconstant (10 level as discussed) the fractionof precursors cp in the affected column after timet is obtained (Appendix)

cp 5 NpN 5 [sps] [1 2 exp(2 s F)] (3)

Here F 5 [ft] is the fluence (flux 3 time) which

JOHNSON ET AL834

is the number of particles (photons electrons orions) incident on the surface per unit area Usingcp in Eq 2 the O2 yield Y at any fluence can bewritten in terms of the yield at steady state Y`

Y 5 [dNO2dt]f 5 sO2 cp N 5 Y`

[1 2 exp(2sF)] (4a)

Y` 5 [sp sO2 s] N (4b)

If we also account for the reduction in the den-sity of water molecules in the initial column Nthese forms change only slightly The size of thetwo parameters Y` and s can be obtained by fit-ting the yield versus fluence data such as that inFig 3

For the low-energy electron data Sieger et al(1998) showed that the ratio [YY`] for a givenincident electron energy is independent of tem-perature over a narrow range of temperatureTherefore the destruction cross section s in Eq4a is roughly independent of T and can be esti-mated from the [YY`] curve Sieger et al (1998)found a precursor destruction cross section s lt08ndash30 3 10216 cm2 for incident electron energiesfrom 30 to 100 eV This is of the order of the elec-tron impact ionization cross section for H2O

Based on the above model the lack of an ob-served fluence dependence for fluences down to1017 hncm2 in the Lyman-a irradiation of ice(Westley et al 1995) can be examined The low-est fluence used implies that the precursor de-struction cross section for Lyman-a is of the or-der of the Lyman-a absorption cross section in ice(09 3 10217 cm2)

Sieger et al (1998) also assumed that sp9 5 0This would mean that the precursor is only de-stroyed by production of O2 (ie s 5 sO2) Thatis unlikely since suggested precursors can be de-stroyed by reaction with H and other radicalsTheir assumption would also mean that the crosssection for production of O2 from a precursorsO2 is large This would be surprising and ap-pear to contradict some of the data discussedabove Instead we assume that s is not equal tosO2 and the total destruction cross section s isof the order of the ionization or absorption crosssection

Having obtained s the measured temperaturedependence for production of O2 is containedonly in Y` Over a narrow range of T one canwrite Y`(T) lt Y`9 exp(2EakT) and obtain the es-

timate of the activation energies in Table 2 Ifsp9 THORN 0 as discussed then sp and sO22 in Eq 4bcannot be obtained separately from the data forY` unless an independent measure of cp is ob-tained Such a measurement is critical for de-scribing the chemical state of the irradiated sur-faces

The observed temperature dependence of Y` iscontained in sp sO2 or N Here N is the ldquodepthrdquofrom which newly formed O2 can reach the sur-face and escape given as a column density of H2OThe physical depth is obtained by dividing thecolumn density N by the molecular number den-sity of the ice Assuming that in the low-energyelectron experiments N is independent of tem-perature and is about three monolayers (Sieger etal 1998) the measured Y` gives [sO2sps]05ndash20 3 10218 cm2 over the electron impactenergy range At 110ndash120K this ratio was foundto be proportional to (E 2 Eo) for E Eo 10 eVagain assuming fixed N (Orlando et al 1998 Or-lando and Sieger 2003)

In the following section we first consider a pre-cursor like peroxide that typically forms fromtwo dissociation events We use this to show un-der what circumstances such a model does notviolate the observed linearity of the O2 yield datawith fluence Following that we relate the sim-ple precursor model above to the case wheretrapped O is the precursor We then use themodel for trapped O to examine the effect of thetrap density on the production of O2 The densityof traps (defects) is determined by the sample for-mation temperature and the subsequent radiationdamage Finally we consider the escape depthN for penetrating radiation within the precursormodel These discussions are supported by moredetailed rate equations in the Appendix

Precursor trapped peroxide

The proposed precursor peroxide is typicallyformed from products of two dissociation eventsie 2 H2O R 2 H 1 2OH R H2 1 H2O2 This wasconfirmed in low temperature ice (Gerakines etal 1996) It is also possible that a dissociation pro-ducing a trapped O followed by an excitationevent can produce peroxide 2 (H2O 1 hn) R H2

1 H2O-O 1 hn R H2O2 1 H2 (eg Khriachtchevet al 2000) The O2 yield versus fluence for inci-dent electrons is reasonably well fit by the yieldexpression above for fluences greater than1014cm2 Therefore it has been suggested that

OXIDANTS AT EUROPA 835

any precursor species must be formed in a singleevent Using a model in the Appendix in whichthe precursor is formed by two events we showthis is not necessarily the case Precursor forma-tion if it is fast could involve two dissociationevents and give the measured fluence depen-dence Although we consider peroxide here theseconclusions apply to other possible precursors

In the example in the Appendix H2O is disso-ciated to H 1 OH according to the cross sectionsd The reactions of two OH units to form H2O2(rate constant k2OH) and 2 H to form H2 (k2H)compete with the recombination of H and OH toH2O (kHOH) If k2H is much greater than k2OH then H2 loss dominates at low fluences as ex-pected (Appendix) Steady state requires not sur-prisingly that the rate for forming H2O2 isroughly equal to the rate for forming H2 Finallyit is seen in the Appendix that the low H2O2 con-centrations suggested by experiment imply thatkHOH (k2H k2OH)05 Such sizes are reasonable

The set of rate equations in the Appendix thatuse number densities of species is integrated overdepth in order to obtain the forms used in Eqs 1and 2 which are based on column densities Inthis way the effective cross sections are related tothe rate constants for the interaction of freshlyproduced mobile species with previously pro-duced trapped species We show in the Appen-dix that the O2 yield versus fluence can be writ-ten in a form similar to that in Eq 4a but with asmall offset due to the two-step precursor for-mation process

Y lt Y`(T) [1 2 exp(2sF) 2 (ssr)]

for F sr21 (4c)

Here s is again the precursor destruction crosssection but sr 5 (2s2OH 1 sHOH) is the effectivecross section for loss of OH by formation of H2O2

and H2O The corresponding expression for Y`(T)is given in the Appendix

The form for the yield versus fluence in Eq 4cis consistent with the low-energy electron mea-surements if (ssr) 1 and if sr F21 lt10214 cm2 Such sizes are indeed possible Whatis required is that at the relevant temperatures(50ndash110K) the interaction length associated withsr is more than a few molecular layers (greaterthan 10 Aring) A cross section that is likely to belarge above 20K is sHOH which describes themean distance for diffusion of mobile H before it

reacts Since s discussed earlier is of the sameorder of magnitude as the H2O dissociation crosssection sd then the precursor concentration cp(in this case H2O2) is roughly proportional to[s2OHsHOH] 1 (Appendix)

Based on the above a two-step precursor for-mation process could be consistent with the elec-tron data contrary to the statements in Sieger etal (1998) Orlando et al (1998) and Orlando andSieger (2003) Experiments are needed to deter-mine the relevant cross sections Below we con-sider a precursor formed in a single excitationevent

Precursor trapped O

O2 can be formed for instance by diffusion ofa transiently mobile O to a previously formed andtrapped O or to the production of O near the siteof a previously produced and trapped O (Matichet al 1993 Johnson and Quickenden 1997) Werefer to this as the MatichndashJohnsonndashQuickenden(MJQ) model and note that O 1 OH R O2 1 Hcan also give molecular oxygen

Trapped O as H2O-O was in fact seen to be aprecursor to O2 when attacked by a mobile O ina Kr matrix (Khriachtchev et al 2000) ThereforeO trapped as H2O-O was also suggested as a pre-cursor in ice (Johnson 2001) Before proceedingwe note that the examination of the MJQ modelin Sieger et al (1998) was incorrect They consid-ered two mobile interacting species whose life-times were long compared with the bombard-ment rate rather than a new mobile O interactingwith a previously trapped O Although this hasnot been corrected recently they added Otrapped as H2O-O as a possible precursor (Or-lando and Sieger 2003) O trapped as H2O-O isseen in matrix isolation studies as discussedabove but it has not yet been identified in iceand like OH it may trap in a number of ways

The MJQ model is essentially the followingFormation of H2 1 O is followed by loss of H2and trapping of O (Ot) A subsequent dissocia-tion or excitation event can lead to the destruc-tion of Ot by reaction with for instance H or byconversion to H2O2 However if this dissociationevent also leads to H2 1 O a mobile O is pro-duced that can react with Ot with an interactionlength that depends on T This results in the for-mation of O2 [O 1 Ot R O2] giving a net pro-duction of 2 H2 and 1 O2 In the Appendix therelevant rate equations (Eqs A1cndashA3c) are con-

JOHNSON ET AL836

sidered Assuming the mobile O processes arefast these rate equations can also be integratedand simplified to give the form in Eqs 1 and 2The precursor column density Np now repre-sents the column density of Ot In this model sO2in Eq 2 is the ldquocross sectionrdquo for reaction of afreshly produced O with an Ot and sp9 indicatesdestruction of Ot Finally sp is the cross sectionfor production of H2 and Ot via processes dis-cussed above

Based on the sizes of the cross sections ex-tracted in the electron irradiation experimentsthe model with trapped O (Ot) as the precursoris quantitatively reasonable That is for a rangeof T both sp and sO2 might be described byevents of the type H2O R H2 1 O (whether di-rect or indirect) If these occur for 10 of theradiation-induced dissociation events as dis-cussed then the required product of cross sec-tions suggested by the sizes of s and Y` above isroughly obtained that is [sp sO2] 02ndash6 3 10234

cm2 in the low-energy electron experiments Thecross section for production of O might dependon the availability of dangling bonds at interfacesin pores or at defects as suggested by a numberof authors The importance of such bonds couldalso be consistent with the differences in the elec-tron-induced yield for initially crystalline and ini-tially amorphous ice (Orlando and Sieger 2003)

O2 production can increase in the presence ofoxygen-rich impurities such as CO2 or SO2 bothof which are known to be present at Europa Asin the MJQ model dissociation (eg CO2 R CO1 O) can provide a mobile O that can be trappedor react with a previously trapped O forming O2In addition such species are hydrogen scav-engers (Moore and Hudson 2000) This can re-duce the destruction of Ot affecting sp as seen inEq A5c

Trap density

In the simplified models above the density oftrapping sites nt was not explicitly accountedfor However in the discussion of both the ex-perimental data and models trapping sites havebeen assumed to be important It has beendemonstrated repeatedly that the density of de-fects voids and interfaces decreases as the tem-perature of formation of the ice increases How-ever the efficiency of production of O2 is theopposite it increases with increasing T There-fore the oxygen yield varies in a manner that is

the inverse of the trap density over a range oftemperatures For penetrating ions in which theenergy density deposited is high and O2 might beformed in a single track a simple explanation hasbeen given As the density of traps for O2 is re-duced and the temperature increases the O2 pro-duced can diffuse to the surface and escape moreefficiently limiting back reactions on further ir-radiation This is examined in the next sectionHowever such a process does not affect the low-energy electron yields for which the penetrationdepth is small and the formed O2 escapes read-ily at the measurement temperatures In Eqs A1cA1c9 A3c and A10 in the Appendix we againuse a simple set of rate equations for trapped andmobile O in order to determine the role of the trapdensity nt on the production of a precursorspecies The conclusions apply to any modelbased on a trapped precursor

In the MQJ model mobile O can be trappedforming Ot (described by kOt nt) and can reactwith previously trapped O to form O2 (kOOt) Inaddition it can react at other sites and Ot can bedestroyed by mobile reactive species such as H(kHOt nH) Mobile H can also react with othertrapped species (kHR nR) Incorporating these re-actions Eq A10 in the Appendix gives a steady-state yield Y` When the density of traps is largethe steady-state yield is given in Eq A5c

Y` R (sd9sd9sd) [kOOtkHR nr][kHOt kOt nt] N (5)

Here sd and sd9 are the dissociation cross sectionsthat determine the production of H (H 1 OH) andO (H2 1 O) by the incident radiation Comparingthe result in Eq 5 with Y` in the simple precur-sor model in Eq 4b the dissociation cross sec-tions in Eq 5 are now scaled by a ratio of the re-action rates

It is seen in Eq 5 that the steady-state yield Y`depends inversely on the density of traps nt Thatis there are competing destruction processes fortrapped O (Ot) Therefore in order to form O2 amobile O competes with mobile H to find atrapped O before it becomes trapped In this waythe increasing O2 yield correlates with the re-duction in the trapping sites with increasing tem-perature of formation The small activation ener-gies discussed earlier would be related to there-orientation of the lattice binding leading to thereduction in trap density At relatively high T thedensity of traps becomes small so that only the

OXIDANTS AT EUROPA 837

sample surface or grain interfaces might act astraps accounting for the observed drop in theyield at high T For this reason it is important todetermine if the yield for T greater than 150Kdepends on the incident flux f Since other quan-tities in the expression for Y` could also be tem-perature dependent experiments are neededThese would likely be pulsed radiolysis experi-ments in which the transients are studied ratherthan the steady-state yields However the avail-able data for lightly ionizing radiation do suggestthat the O2 yield depends inversely on trap den-sity as in the model above

Escape from depth

The quantity N in Eq 1 is the ldquodepthrdquo given asa molecular column density from which O2 formedbelow the surface can percolate to the surface andescape It can be converted to a physical depth bydividing by the molecular number density of theice In the case of the low-energy electrons thedepth below the surface at which ionizations andexcitations are produced is small However forfast ions that penetrate many monolayers this isnot the case Below we examine N in Eq 1

Although it has long been understood that twoexcitations events are required to produce O2(eg Brown et al 1982 Reimann et al 1984) theprecursor model suggests that at saturation(steady state) any new event can produce O2 fromthe steady-state density of trapped precursorsTherefore one might expect the yield at satura-tion to be nearly linear in the excitation cross sec-tion This is the case for the low-energy electronsand it was initially assumed to be the case for thefast ions (Brown et al 1980 Reimann et al 1984)However Baragiola et al (2003) showed that thetotal sputtering yield in the saturation region attemperatures greater than 100K which is dom-inated by decomposition (Brown et al 1982) isproportional to the square of the electronic en-ergy deposition per unit path length in the solid(dEdx)e Assuming that the production and de-struction cross sections in Y` in Eq 4b roughlyscale with (dEdx)e then their ratio [sp sO2s]also roughly varies as (dEdx)e If this is the casethen the observed quadratic dependence impliesthat the depth N from which O2 can be mobi-lized to escape must also depend on the excita-tion density (dEdx)e This is consistent with theobservation that the O2 yield depends on thick-ness for samples thinner than the ion penetration

depth (Reimann et al 1984 Benit and Brown1990) Therefore the percolation depth for escapeN is a critical factor in comparing yields for pen-etrating radiation with those for non-penetratingradiation like the low-energy electron and low-energy ion data It is also seen that percolationand trapping can be included within the simplemodel considered here

Model summary

We have examined the simple precursor modelintroduced by Sieger et al (1998) to explain theirelectron data We have shown how it can be gen-eralized and derived from an integration of theappropriate chemical rate equations with thecross sections and total column density contain-ing the reaction rates and the physics of diffusivetransport Therefore the parameters can havevery different meanings from those initially as-sumed The equations are useful when there arefast processes which go to completion in timesmuch shorter then the time between particle im-pacts and slow processes associated with speciesthat are assumed to be trapped between particleimpacts This separation allowed us to examinethose models proposed for precursor formationand destruction due to reacting mobile speciesformed from dissociation events Further fromthe size of the onset of the linearity of the yieldversus fluence we can place constraints on therole of nonlinear processes such as those thatmight be associated with the formation of a pre-cursor like peroxide

With these extensions this simple model is use-ful for approximating the full rate equations inthe temperature region for which the yields de-pend on the fluence (dose) and not on the doserate Based on the assumptions in the model thisimplies temperatures at which precursor speciesare stable over experimental times At the high-est temperatures for the yields shown in Fig 4athis might not be the case This should be testedexperimentally In this regard it is puzzling thatexperiments have shown that the absorptionbands associated with certain trapped speciessuch as OH and trapped electrons ldquodisappearrdquoat 100ndash120K (see eg Johnson and Quickenden1997) whereas the O2 yield data indicate that rel-evant precursor species must be stable to 140KTherefore the proposed O2 production processescan only be distinguished by directly determin-ing the chemical state of the irradiated ice

JOHNSON ET AL838

It may be the case that a particular precursoris not required as mentioned earlier Rather af-ter a sufficient level of oxidation a newly pro-duced O or OH can react in the oxygen-rich lat-tice producing O2 in a pseudo-first-order processThis can be tested experimentally Indeedtrapped O and peroxide may be changed into oneanother as in the matrix isolation experiments(Khriachtchev et al 2000) Therefore their rela-tive concentrations versus temperature and doseshould be studied

In the model considered here it is seen that theparameters in Y` must account for the strong tem-perature dependence Since the number of bulktrapping sites decreases with increasing T as theice sample is annealed one might expect that theprecursor concentration would decrease We sug-gest that the observed temperature dependenceis determined by the competition between theavailability of trapping sites and mean interactiondistances of a mobile species such as O and HThat is the O2 yield varies inversely with the den-sity of traps over a range of temperatures In thesimple model above this effect is included indi-rectly as shown using the more detailed equa-tions in the Appendix

EUROPA

Oxygen and peroxide production at Europa G values

The dependence of the yield on the thicknessof the ice sample for highly penetrating ions sug-gests that the O2 yield is roughly related to theamount of energy deposited in the sample There-fore the amount of O2 produced and lost fromice in steady state is often given as a G value thenumber of molecules produced per 100 eV de-posited in Table 2 In applying these G values tothe planetary environment care must be takensince the yields are temperature fluence and ex-citation density dependent They also are affectedby the presence of other proton scavengersknown to be present in Europarsquos ice

The G values for peroxide in Table 2 are thosefor production at low fluence but the G valuesfor O2 are for the production and escape of gas-phase oxygen at saturation Therefore they de-pend on the formation and destruction cross sec-tions in the model above Such G values can beused to describe the steady-state source of the am-

bient O2 atmosphere at Europa but are not help-ful in determining the ability to produce and trapO2 at depth into the icy surface Over the timescales of the experiments some O2 is seen to re-main trapped and is not destroyed by radicalsTypically there are a number of trapping sitesand recent estimates of the activation energy foran O2 molecule trapped in a substitutional site(Hori and Hondoh 2002) suggest that it can betrapped for times comparable to the time re-quired for the conversion of amorphous ice intoIh (Ayotte et al 2001) The stability of O2 inclu-sions has not been studied

To understand the relationship between the Gvalues for O2 in Table 2 we note that fast lightions and electrons lose their energy predomi-nantly by producing ionizations and excitationsThe ionization events produce secondary elec-trons that can result in further ionizations Thenumber of ionization events Ni produced by afast light ion stopping in a water ice is Ni 5EoW where Eo is the initial energy and W is theaverage energy expended per ionization event27 eV in water ice Therefore radiolysis by fastions is often described as the net effect of theshower of electrons set in motion by the incidention In this manner the G values for productionof O2 from ice by energetic light ions and elec-trons can be roughly estimated as the sum of theeffects of all the secondary electrons using theyield Ye for 10ndash100 eV electrons That is if weignore the percolation depth N we can write thenumber of O2 produced as the number of ioniza-tion events times Ye which already has in it thedestruction processes That is G NiYeEo YeW 0006 O2100 eV

This estimate of G is seen to be an order of mag-nitude larger than the keV proton value for O2 inTable 2 but is much smaller than the G value forenergetic Ne1 These differences from the simpleestimate are consistent with the observation thatthe yield in the saturation regime for penetratingions varies nonlinearly with excitation densityThe difference from the keV proton G value islikely due to the inability of O2 to escape fromdepth before destruction On the other hand thelarger G value seen using MeV Ne1 is likely dueto the high density of excitation This can resultin more efficient production of O2 increased pro-duction of damage sites for forming O2 andorincreased transport (percolation) along the tran-siently heated track That is the ability of theformed O2 to percolate to the surface during ion

OXIDANTS AT EUROPA 839

bombardment plays an essential role in deter-mining the measured yields as discussed aboveThis effect needs to be evaluated for a planetarysurface subject to particle fluxes that are many or-ders of magnitude smaller than those used in thelaboratory

In the absence of O2 yields for all of the rele-vant ions and energies an estimate of the totalsupply of O2 can be obtained using the G valuesin Table 2 For the energetic oxygen and sulfurions we use G values for neon and argon ionsThese are lower bounds as the data were obtainedat temperatures much lower than the surface tem-peratures at Europa (80ndash110K) For the fast pen-etrating electrons and protons we use the valuefor 200 keV protons in Table 2 For the heavy ionswe used the Ne1 and Ar1 values For compari-son we also give an estimate using the G valuesobtained from the low energy electron data andthe total ionization as discussed above

The G values are combined in Table 3 with theelectron and ion energy flux at Europa Althoughthere is some deflection of the flow due to in-duced fields the net energy flux to the surface isnot reduced much from these values (Paranicaset al 2002) In fact locally produced pick-up ionswhich are formed from the ejecta and acceleratedto the surface can contribute (Ip 1996) Furtheralthough G values are typically independent ofthe incident angle the loss of O2 is a surfaceprocess and is enhanced with angle That is thenumber of O2 produced within the escape depthincreases with increasing angle to the normal(steeper than the inverse cosine) This enhance-ment is 25ndash4 (Jurac et al 2001) Ignoring suchcorrections the total production rate calculated

(23ndash62 3 109 O2cm2s) is a lower bound to theactual production rate from the icy regions of Eu-roparsquos surface Recent modeling of the O2 atmo-sphere suggest that such a source rate is required(Shematovich and Johnson 2001)

Earlier estimates of the O2 source rate were in-correct because O2 production was assumed to bea fixed fraction of the H2O sputtering yield Atthe ambient temperature an ejected H2O sticks ef-ficiently (Smith and Kay 1997) to the surface ofneighboring grains in the porous regolith reduc-ing the sputtering yield for H2O (Johnson 1989)This reduction does not apply to the O2 fractionof the yield enhancing its relative importance toH2O Here we assume that all of the gas-phaseO2 produced can contribute to the ambient at-mosphere even if it is produced at a grain sur-face that is at some depth into the porous regolith

Finally Europarsquos surface is not pure ice every-where That is much of the trailing hemisphereis a hydrated sulfuric acid containing smallamounts of sulfur and SO2 and the leading hemi-sphere contains low levels of carbon CO2 andcarbonates Sodium and potassium atoms aresputtered from the icy regions (eg Leblanc et al2002) O2 production from surfaces containingthese species has not been measured C and S areboth proton scavengers and the other species canbe sources of O as discussed so that the O2 yieldfrom such regions could be enhanced Howeverthere can also be competing processes that mightreduce these rates In fact recent HST observa-tions suggest that the O2 might not be produceduniformly across Europarsquos surface (McGrath et al2000) The analysis of these observations is pre-liminary and is not simple since the observed

JOHNSON ET AL840

TABLE 3 APPLICATION TO EUROPA

OxygenImplantation Energy flux production

Species 109(cm2 s)a 109 keV(cm2 s)(1) 109(cm2 s)

H (10 keV) 0015(1) 12 0084 (072)bO (10 keV) 0015(1) 18 054cS (10 keV) 0009(1) 30 090cElectrons 62 043 (37)bPlasma (10 keV) (137 amu) 13(2) 10 03c

Total 79 23 (62)UV photons (64 eV) 46 01

aReferences are as follows 1Cooper et al (2001) [similar results in Paranicas et al (2002)] 2Mauk et al (1996)bUsing G for 200 keV protons (upper bounds in brackets are based on the low-energy electrons data as a model for

the secondary electrons produced)cUsing G for 15 MeV Ne1 at 7K a rough lower bound

emissions depend on the local plasma propertiesHowever the efficiency of O2 production in con-taminated ices needs to be studied

Porous regolith O2 interactions on grain surfaces

Although the O2 atmosphere is principally pro-duced in the near surface layers and is eventuallylost to space it also permeates Europarsquos porousregolith The porosity p (void space) of the re-golith is uncertain (Buratti 1995) It has been es-timated to be the most compact surface in theOuter Solar System p 025 on the trailing hemi-sphere and p 079 on the leading hemisphere(Buratti and Veverka 1983) and the most poroussurface p 096 globally (Domingue et al 1991Domingue and Hapke 1992) These large differ-ences are due to the uncertainties in the amountof backscattered light at small phase angles andon the photon absorption properties and grainsizes The more compact surface is likely correctGrain sizes rg of the order of 100 mm have beensuggested (Hansen and McCord 1999 2000Geissler et al 1998) Europarsquos regolith differs sig-nificantly from the lunar regolith which is asand-like surface of individual grains Analysis ofthe thermal inertia of Europarsquos surface (Grundyet al 2001) suggests that the ice grains are sin-tered forming a fairy castle-like structure Sin-tering can occur thermally or due to the incidentradiation In addition ices doped with acids andalkalis in the laboratory can be powdery solidswith the impurities primarily found at grain in-terfaces

The porous regolith affects the above estimatesof the oxygen production in three ways First forisotropic bombardment the mean distance d intothe surface at which an incident ion or photonstrikes a grain surface is d (89)rg(1 2 p) Us-ing the above value for rg then d 02 m if theporosity is very large p 96 Therefore de-sorbed molecules can come from a considerabledepth and will interact with grain surfaces Evenfor much lower porosities the effective sputteringyields can be significantly modified (Johnson1989) As discussed above the yields for ejectedH2O which sticks onto neighboring grains at theambient temperatures are reduced but those forejected O2 are not This enhances the relative im-portance of the O2 yields

Finally the surfaces and interfaces of the sin-tered grains in Europarsquos porous surface are placeswhere low temperature chemistry involving the

ambient oxygen can occur at depths that are wellbelow the nominal penetration depth of the inci-dent radiation The relative importance of suchchemistry is roughly indicated by the ratio of thetotal surface area in the regolith to the area of Eu-roparsquos visible surface Using the porosity andgrain size above and a regolith depth dr of theorder of a meter then this ratio is large 3(1 2p)drrg which is 103ndash2 3 104 for average porosi-ties of 96 and 25 respectively Therefore thegas-phase O2 molecules below the visible surfaceinteract frequently with the grain surfaces In thisway a very large volume of the subsurface couldbe altered chemically even if it is not directly ex-posed to the radiation

From the overlying structures seen in the high-resolution images of Europarsquos surface by theGalileo spacecraft (eg Prockter and Pappalardo2000) it is clear that the youngest surfaces are red-der The reddening in the visible has been attrib-uted to the presence of sulfur in the ice (McEwen1986 Johnson et al 1988 Calvin et al 1995 Carl-son et al 1999b) The amount of reddening in thespectra of the surface material decreases with ageand the surface materials appear to becomebrighter with age This might occur by the slowaccumulation of a frost over-layer or by radioly-sis (Carlson et al 2002 Johnson et al 2003) Ra-diolysis preferentially converts the sulfur in theice to sulfur dioxide or sulfuric acid both ofwhich have flat spectra in the visible In regionswhere the radiation does not directly penetratethe O2 permeating the regolith can oxidize thesulfur at the surfaces of the icy grains and at graininterfaces Similarly the carbon seen in the ice onEuroparsquos leading hemisphere (eg Carlson et al2004 Johnson et al 2003) can be oxidized by ra-diolysis in the near surface layers or by reactionson the grain surfaces at depth in the regolith bythe ambient O2 The meteoroid bombardmentthat produces the regolith also exposes fresh sur-faces for oxidation and can cause the collapse ofpores leading to the trapping of oxygen In thismanner both refractory and volatile oxidants canbe formed and trapped in the subsurface ice forpotential delivery to Europarsquos subsurface ocean

SUMMARY

The presence of a thin oxygen atmosphere andoxidants trapped in Europarsquos surface ice (egJohnson et al 2003) combined with the possible

OXIDANTS AT EUROPA 841

presence of a subsurface ocean (eg Carr et al1998) have made Europa an exciting target forfuture space exploration The molecular oxidantsproduced by radiolysis and photolysis could bespecies needed for carbon-based biochemistry inEuroparsquos putative subsurface ocean Howeverthere has been no model for the formation trap-ping and transport of relevant oxidants

In this paper we have reviewed the space ob-servations and laboratory studies related to thechemistry of formation of oxidants in ice by theincident radiation Because the laboratory dataare incomplete we examined the rate equationsfor formation of oxygen and related species byradiolysis and photolysis We showed how suchequations can be simplified to the analytic modelthat was used for fitting the fluence and temper-ature dependence of the O2 yield produced bylow-energy electrons (Sieger et al 1998) In de-riving the analytic model from the more completerate equations the competition between forma-tion and destruction processes can be studied ascan the role of trapping sites escape from depthand temperature history These equations also al-lowed us to put incident photon electron and iondata on a similar footing for the first time

The model can be used to suggest measure-ments that are needed to describe the chemicalstate of the irradiated ice and to interpret new lab-oratory data We showed that if reactions are fastthe yield data could be consistent with a precur-sor being formed by two events as is the case forperoxide However we favor trapped O as a pre-cursor as initially suggested by Matich et al(1993) for production of excited O2 Such aspecies if present could be detected spectro-scopically as can trapped O2 through bands ac-tivated by lattice perturbations (Cooper et al2003a)

The available oxygen yield data were used tocompare G values for different incident particlesThese were then used to estimate the oxygen pro-duction in the radiation environment at EuropaThe lower bounds were consistent with the pro-duction rates obtained from a recent model of Eu-roparsquos atmosphere The porosity of the regolithwas also considered It not only affects the rela-tive sputtering yields for O2 and H2O but alsoprovides an enormously increased surface areaon which the ambient gas-phase oxygen can re-act with species in the ice That this might be oc-curring is consistent with the change in re-

flectance associated with surface aging Such re-actions could also convert ambient molecularoxygen into other more refractory species that re-main trapped on subduction of Europarsquos crust

Although radiolysis produces a stable chemi-cally altered surface in the laboratory the parti-cle flux to Europarsquos surface is many orders ofmagnitude lower Therefore annealing can be ef-fective at temperatures that are much lower thanthose typically found experimentally Belowabout 120K the annealing times appear to be long(eg Baragiola 2003) so that laboratory data canoften be directly applied to Europarsquos icy surfaceHowever account has to be taken of the syner-gism between the UV and the energetic plasmaas discussed earlier (Johnson and Quickenden1997)

Transient surface melting might initiate lifeprocesses but subduction of oxidants to the pu-tative ocean is likely to be required if biochem-istry is to occur at Europa (Chyba and Hand2001) Although the temperature versus depthinto Europarsquos regolith is not well known belowthe regolith the temperature and pressure appar-ently increase with depth At some depth (ap-proximately kilometers) either very warm ice oran ocean exists (Ruiz and Tejero 2000) Thereforechemically refractory oxidants are likely to be re-quired in order to survive subduction to thewarmer subsurface regions Trapped species ifdelivered downward can become mobile andwill segregate forming inclusions (Johnson andJesser 1997) Indeed peroxide appears to segre-gate in ice (Gurman et al 1967) and these inclu-sions may remain relatively stable until the per-oxide is dissolved in the putative ocean Thevolatiles seen in reflectance such as O2 SO2 andCO2 will also segregate But at the higher tem-peratures they could also diffuse through thebulk react or percolate into the regolith atmo-sphere Therefore the observation of the trappedvolatiles in Europarsquos surface ice is of interest pri-marily because it suggests that oxidants are pre-sent The formation of oxygen-rich molecules fol-lows the preferential loss of hydrogen Howeverthe surface is in a local steady state with im-planted H O and S from the plasma and the lossto space Since the plasma flux to the surface canbe variable this balance can also vary Indeedtemporal changes in reflectance have been re-ported (Domingue and Hendrix 2004)

Based on the models for formation of O2 ex-

JOHNSON ET AL842

amined in this paper additional experiments areneeded to determine the chemical state of irra-diated ice at low fluences Of immediate inter-est is the identification of stable molecular pre-cursors required for the formation of O2 Basedon the available laboratory data individual per-oxide molecules trapped in the ice do not appearto be the principal precursors but trapped O(Matich et al 1993) remains a viable candidateAlthough the data suggest that the local hydro-gen-bonding network is important in the pro-duction of O2 the effect of the formation tem-perature and radiation damage on the structureneed to be measured It is also important to de-scribe quantitatively the relationship betweenH2 loss and the formation of O2 and H2O2 overa range of temperatures and fluences Finallyone would like to determine the steady-statedensities of biochemically interesting moleculesformed by radiolysis in an ice containing sulfurand carbon at Europarsquos surface temperaturesand then determine their stability with increas-ing temperature and pressure Such experimentsstill need to be carried out Here we have pre-sented a model that brings together the variouslaboratory data and now can be used to considerthe formation trapping and transport of relevantoxidants at Europa

APPENDIX

Introduction

In this Appendix we examine rate equationsthat describe various aspects of the radiolysis orphotolysis of ice These equations need to be un-derstood if one is to be able to calculate the for-mation trapping and transport of oxidants Theemphasis is on the formation of molecular oxy-gen We integrate these equations over the escapedepth of O2 to give the simplified rate equationsdiscussed in the text in which number densitiesare replaced by column densities (eg Eqs 1 and2) In this way we can examine the relationshipbetween the rate constants and the effective pre-cursor formation and destruction cross sectionsextracted from data We first consider a precur-sor such as H2O2 that is formed from two exci-tation events This bears on the observed linear-ity of the oxygen yield versus fluence We thenconsider the equations for trapped versus a tran-

siently mobile species We use a reaction betweena mobile and trapped O as an example althoughO 1 OH can also give oxygen Finally we ex-amine the possible dependence of the O2 forma-tion process on the density of traps (defectsvoids and grain boundaries) as a function of tem-perature The results are discussed in the text

For clarity all possible radiolytic species arenot included in the discussion below Thereforepotentially important products such as HO2 andthe charged species (eg H3O1 the trapped elec-tron etc) are missing In each section we attemptto make a point related to production of O2 andother oxidants by using a subset of the productsrather than attempt a complete description withmany unknown rate constants

Fast versus slow processes

To describe the O2 production rate from a pre-cursor such as H2O2 we first use a simple but in-structive set of rate equations to describe the for-mation of H2O2 We then relate these to thesimplified rate equation for the precursor in thetext Eq 1 and show how the postulated precur-sor formation cross section sp is related to theH2O dissociation cross section sd by the reactionrates

Using the reactions

H2O 1 radiation R H 1 OH (sd)

H 1 H R H2 (k2H)

H 1 OH R H2O (kHOH)

OH 1 OH R H2O2 (k2OH)

the reaction pathways are described by

H2

Hqk2H

1 OHH2O 1 radiation mdashRsd H

OHH2O2

OH Rk2OH

kHOH

HH2O

OXIDANTS AT EUROPA 843

The corresponding rate equations in an irradiatedvolume are

dnOHdt 5 sd f n 2 2 k2OH nOH nOH

2 kHOH nH nOH (A1a)

dnHdt 5 sd f n 2 2 k2H nH nH

2 kHOH nH nOH (A2a)

dnH2O2dt 5 k2OH nOH nOH

2 destruction processes (A3a)

Here sd is the dissociation cross section f is theradiation flux and n is the molecular numberdensity of ice with nOH and nH the density of dis-sociation products OH and H respectively Therate constants k2OH k2H and kHOH describe thereactions above In this model the H2 formed isassumed to escape for temperatures above about20K We do not distinguish between trappedspecies and mobile species although in fact thatdistinction is important as discussed in the nextsection That is any OH produced at low tem-perature by one ion would likely be trapped priorto interacting with an OH produced by a subse-quent ion At higher temperatures this can breakdown as discussed in the text We first relate theabove set of equations to Eq 1 in the text and thenconsider precursor destruction processes

Equation 1 in the text is obtained by solvingthese equations together and then by integratingover a depth Dx Here Dx is the smaller of the pen-etration depth of the particle and the O2 escape(percolation) depth Assuming the molecular den-sity of ice n is constant with depth Dx n 5 N inEq 1 Comparing Eq 1 with Eq A3a the precur-sor formation rate sp f N is equivalent to [k2OH

nOH nOH] Dx Assuming that the processes in EqsA1a and A2a are fast then the steady state is read-ily obtained In addition if k2H k2OH there ispreferential loss of H2 at short times as suggestedby the laboratory data If steady state is reachedthe production of H2 is equal to that of H2O2 us-ing only the equations above Further one obtains

[k2OH nOH nOH] Dx R [fp sd] f N (A4)

with

fp 5 05[1 1 05 kHOH(k2H k2OH)05] (A5a)

Therefore the precursor formation cross sectionsp in Eq 1 is a fraction fp of the total dissocia-

tion cross sd Since laboratory data suggest theH2O2 fraction in an irradiated ice is small thenkHOH (k2OH k2H)05 which is likely givingfp R [(k2OH k2H)05kHOH]

Trapped versus mobile species

To examine the fluence dependence discussedin the text we note that trapped OH and freshlyproduced transiently mobile OH can be treatedas different species (eg Matich et al 1993)Therefore assuming no trapping of H the OHrate equation above would be replaced by twoequations one for the trapped species nOHt andthe other for the freshly produced mobile speciesnOH Assuming the reaction of two mobile OHunits is unlikely and that peroxide is formed fromthe interaction of a trapped OH with a subse-quently produced OH an expanded set of equa-tions is obtained

OH 1 trap R OHt (kOHt)

OH 1 OHt R H2O2 (kOHOHt)

H 1 OHt R H2O (kHOHt)

then the reaction pathways are described by

qH2

H2O 1 radiation mdashRsdH

qk2H

1 OHH

trap OH

OH RkOHt

OHt kOHOHtH2O2

kHOHkHOHt

H H2O HH2O

The resulting set of equations is

dnOHdt 5 sd f n 2 kHOH nH nOH

2 kOHt nOH nt 2 kOHOHt nOH nOHt (A1b)

dnOHtdt 5 kOHt nOH nt 2 kOHOHt nOH nOHt

2 kHOHt nH nOHt (A1b9)

dnHdt 5 sd f n 2 2 k2H nH nH 2

kHOH nH nOH 2 kHOHt nH nOHt (A2b)

JOHNSON ET AL844

dnH2O2dt 5 kOHOHt nOH nOHt

2 destruction processes (A3b)

where nt is the trap density in the irradiated iceAt low T we can assume nt is large so thatdntdt 0 simplifying the set of equations If theprecursor of interest is trapped O as discussed inthe text then a similar set of equations can be con-structed for O and Ot as in the example to followOf course a full description over a broad rangeof temperatures would require the considerationof the both trapped and mobile H O and OHthe rate of change of trap density and the vari-ous destruction processes

The above equations should be solved simul-taneously but for the incident particle fluxes typ-ically used in these studies the diffusion of mo-bile species to traps is relatively fast This is trueeven at low temperatures if there is a significantdensity of trapping sites This is usually the casefor samples formed at low T Therefore in a lowtemperature solid at low dose rates there is not asignificant steady-state background of diffusingOH There are trapped or freshly produced OHunits which rapidly react Averaging over thebombardment rate this can be approximated bywriting dnOHdt 0 and then separating nOH outin Eq A1b Assuming also that H diffuses rapidlyand either reacts or forms H2 we also writednHdt 0 on the average Therefore betweenimpacts only species trapped at defects pores orgrain surfaces are assumed to be present as sug-gested by the lack of a dependence on the inci-dent beam flux In this limit the rate of change oftrapped OH roughly simplifies to

dnOHtdt 5 2 k2H nH nH 2 2 kOHOHt nOH nOHt

It is seen that the production of peroxide dependson the loss of H by formation of H2 as discussedand at steady state the production rate for per-oxide equals the rate of formation of H2 in theseequations

Treating the mobile and trapped species sepa-rately we can further simplify the rate equationsand obtain the precursor model used in the textThat is integrating over the relevant depth Dxthe densities are replaced by column densitiesThe rate constants ki are then related to effectivecross sections si In this way cross sections aregiven as an interaction length for each freshlyproduced mobile radical [ki nj] R [si f] The

trapped OH column density NOHt and peroxideproduction rate can be estimated from ldquolinearrdquorate equations

dNOHtdt lt [sd f] N 2 2 [s2OH f] NOHt

2 [sHOH f] NOHt (A6a)

dNH2O2dt lt [s2OH f] NOHt 2 s NH2O2 (A6b)

Here sd is again the dissociation cross sectionand the destruction processes in Eq A3b are con-tained in s The second and third terms on theright in Eq A6a are the loss of OH due to the pro-duction of H2O2 and H2O In the integration ofEq A1b9 [2 kOH OHt nOH] becomes [2 s2OH f] and[kHOHt nOH] becomes [sHOH f] Therefore thedensity and the interaction length of mobile H arecontained in sHOH If H2O2 is indeed the pre-cursor then in Eq 2 in the text [sp f N] is be-comes [s2OH f NOHt] from Eq A6b Accountingfor precursor destruction which can give backtwo OH and treating H2 can both be included inthe integrated model via the si values

The simplified set of equations above allows usto examine the relative time (fluence) scales forprecursor formation (in this case H2O2) SolvingEqs A6a and A6b the column density of OH andthe concentration of precursors versus fluence cpcan be obtained Using the column of H2O2 pro-duced as the precursor column Np in Eq 2 thenO2 yield can be obtained The results can be com-pared with the results from Eqs1 and 2 in thetext The average precursor (H2O2) concentrationover the column depth of water molecules N is

cp 5 fp (sds)[1 2 exp(2s F)] 2 (ssr)

[1 2 exp(2sr F)] (A7)

with

fp 5 2s2OH(sr 2 s) (A5b)

In these equations sr 5 (2s2OH 1 sHOH) andssr 1

It is seen that the precursor concentration withtime cp depends on two rates the reaction ratesthat determine sr and the destruction rates in sAs in Eq 4a the yield can be written as

Y 5 sO2 cp N (A8)

If sr and s are comparable then the yield at small

OXIDANTS AT EUROPA 845

fluence is quadratic in F However if sr sthen the linear dependence on fluence is obtainedwith a small shift in the fluence as shown in Eq4c in the text Further the steady-state yield isnow

Y` (T) 5 sO2 [fp sd] Ns (A9)

That is sp R [fp sd] as in Eq A4 and the tem-perature dependence for the incident electrondata must be contained in [sO2 fp] or in the ratio[sO2 s2OH]sHOH]

For comparison with their precursor modelSieger et al (1998) examined the possibility of ob-taining the O2 precursor by considering reactionsfor a density of diffusing dissociation products asin Eqs A1 and A2 rather than a trapped productinteracting with a freshly formed dissociationproduct Based on the above the lowest fluence atwhich O2 is first detected experimentally sets alimit on the time scale for mobilendashmobile reactions

Effect of trap density

Equations A1b A1b9 A2b and A3b can be ap-proximately solved assuming that the dnOHdtdnHdt and dntdt are very small on the averageas discussed above In order to examine the roleof the trap density nt which can depend on thetemperature and method of formation and on theradiation history of the ice we use the slightlysimpler precursor model In this model the O pro-duced directly by dissociation traps forming aprecursor O2 is then formed when a freshly pro-duced O or other species reacts with a trappedoxygen atom Ot

H2O 1 radiation R H2 1 O (sd)

O 1 trap R Ot (kOt)

O 1 Ot R O2 (kOOt)

The reaction pathways for the oxygen species are

H2O 1 radiation mdashRsd9

trap OO2

kOtOt kOOt

H2 1 OOHt

kOR H(kHO t)

ROR

We assume for simplicity both H2 and O2 escapefrom depth Dx after formation We also allow mo-bile H to react with Ot (kHOt) removing trappedO and both the mobile H and O to be removedby other reactants R (eg OH or contaminants)This leads to a simplified set of equations not di-rectly involving OH or peroxide

dnOdt 5 sd9 f n 2 kOt nO nt 2 kOR nO nR

2 kOOt nO nOt (A1c)

dnOtdt 5 kOt nt nO 2 kOOt nO nOt

2 kHOt nH nOt (A1c9)

dnHdt 5 sd f n 2 kHOt nH nOt

2 kHR nH nR (A2c)

dnO2dt 5 kOOt nO nOt (A3c)

These equations of course act in parallel with theequations above for OH and mobile OH may re-act with trapped O to form O2 Here we focusonly on mobile O

Since the diffusion of H and O are fast and thedensity of traps is large the nH and nt averagedover the bombardment rate can be taken veryroughly as time independent Further if we as-sume the number of trapped O nOt is small com-pared with the number of sites at which H canreact then nH [sd f nkHR nR] For a low ra-diation flux f nH is small on the average as as-sumed Since the O2 yield YO2 is equal to[(dnO2dt) Dxf] with Dx the escape depth thenthe steady-state yield Y` can be obtained ana-lytically

[Y`][sd9 N] 5 1 1 (q 2 1) 2 [(q 2 1)2

1 8 q d]054 d (A10)

q 5 [kOt (nt 1 nR)] (kHOt nH)(sd9 f n) kOOt

5(sdsd9) [kOt (nt 1 nR)] kHOt[kOOt kHR nR]

d 5 05 [1 1 nt(nt 1 nR)]

We note that for nt large q can be large Assum-ing that q 1 and nt nR then Eq A10 be-comes

Y` R [sd9 N]q 5 (sd9 N) (sd9sd)

[kOOt kHR nR][kHOt kOt nt] (A5c)

JOHNSON ET AL846

Comparing with Y` in the simple precursormodel in Eq 4b then s R [(kHOt nH)f] R [sd nkHOt][kHR nR] In addition the product [sO2sp] R sd9 sd9 [kOOt n][kOt nt] That is the dis-sociation cross sections are modified by reactionrates

Quite remarkably it is seen in Eq A5c that thesteady-state yield Y` can depend inversely onthe density of traps nt in the temperature andflux regimes for which the approximations arerelevant That is if there are competing destruc-tion process for trapped O then at high trap den-sity a mobile O must find a trapped O before itbecomes trapped or before mobile H destroys theOt In this way the increasing O2 yield with in-creasing T can be related to the change in the trap-ping density nt with the formation temperatureof the ice sample As the temperature increasesand the density of traps in the bulk decreases thesurface and grain interfaces become the only trap-ping sites

ACKNOWLEDGMENTS

REJ acknowledges support by a Gledden Fel-lowship in Chemistry School of Biomedical andChemical Sciences at The University of WesternAustralia and by grants from the NSF Astron-omy Program and NASArsquos Planetary Geologyand Geophysics Program PDC acknowledgessupport by an Australian Postgraduate Award

ABBREVIATIONS

HST Hubble Space Telescope MJQ MatichndashJohnsonndashQuickenden NIMS Near Infrared Map-ping Spectrometer UV ultraviolet

REFERENCES

Ayotte P Smith RS Stevenson KP Dohnarsquolek ZKimmel GA and Kay BD (2001) Effect of porosityon the adsorption desorption trapping and release ofvolatile gases by amorphous solid water J Geophys Res106(E12) 33387ndash33392

Bahr DA Famaacute M and Baragiola RA (2001) Radiol-ysis of water ice in the outer solar system sputteringand trapping of radiation products J Geophys Res 10633285ndash33290

Baragiola RA (2003) Microporous amorphous water icefilms and astronomical implications In Water in Con-

fining Geometries edited by JP Devlin and B Buch El-sevier New York in press

Baragiola RA and Bahr DA (1998) Laboratory studiesof the optical properties and stability of oxygen onGanymede J Geophys Res 103 25865ndash25872

Baragiola RA Atteberry CL and Bahr DA (1999) Ori-gin of solid oxygen on Ganymede Reply to commentby RE Johnson on ldquoLaboratory studies of the opticalproperties and stability of oxygen on Ganymederdquo JGeophys Res E104 14183ndash14187

Baragiola RA Vidal RA Svendsen W Schou J ShiM Bahr DA and Atteberrry CL (2003) Sputteringof water ice Nucl Instrum Methods Phys Res B 209294ndash303

Bar-Nun A Herman G Laufer D and Rappaport ML(1985) Trapping and release of gases by water ice andimplications for icy bodies Icarus 63 317ndash332

Baverstock KF and Burns WG (1976) Primary pro-duction of oxygen from irradiated water as an expla-nation for decreased radiobiological oxygen enhance-ment at high LET Nature 260 316ndash318

Benit J and Brown WL (1990) Electronic sputtering ofoxygen and water molecules from thin films of waterice bombarded by MeV Ne1 ions Nucl Instrum Meth-ods Phys Rev B 46 448ndash451

Benit J Bibring J-P della-Negra S LeBeyec YMendenhall M Rocard F and Standing K (1987)Erosion of ices by ion irradiation Nucl Instrum Meth-ods Phys Rev B 1920 838ndash842

Brown WL Augustyniak WM Lanzerotti LJ John-son RE and Evatt R (1980) Linear and nonlinearprocesses in the erosion of H2O ice by fast light ionsPhys Rev Lett 43 1632ndash1635

Brown WL Augustyniak W M Simmons E Mar-cantonio KJ Lanzerotti LJ Johnson RE BoringJW Reimann CT Foti G and Pirronello V (1982)Erosion and molecular formation in condensed gasfilms by electronic energy loss of fast ions Nucl In-strum Methods B 1 307ndash314

Brown WL Augustyniak WM Marcantonio KJ Sim-mons EN Boring JW Johnson RE and ReimannCT (1984) Electronic sputtering of low-temperaturemolecular solids Nucl Instrum Methods B 1 307

Buratti B (1995) Photometry and surface structure of theicy Galilean satellites J Geophys Res 100(E9) 19061ndash19066

Buratti B and Veverka J (1983) Voyager photometry ofEuropa Icarus 55 93ndash110

Calvin WM and Spencer JR (1997) Latitudinal distri-bution of O2 on Ganymede observations with the Hub-ble Space Telescope Icarus 130 505ndash516

Calvin WM Clark RN Brown RH and Spencer JA(1995) Spectra of the icy Galilean satellites from 02 to5 mm a compilation new observations and a recentsummary J Geophys Res 100 19041ndash19048

Calvin WM Johnson RE and Spencer JA (1996) O2

on Ganymede spectral characteristics and plasma for-mation mechanisms Geophys Res Lett 23 673ndash676

Calvin WM Aninich VG and Brown RH (2004) Oxy-gen temperature and phase effects Icarus (in press)

OXIDANTS AT EUROPA 847

is the number of particles (photons electrons orions) incident on the surface per unit area Usingcp in Eq 2 the O2 yield Y at any fluence can bewritten in terms of the yield at steady state Y`

Y 5 [dNO2dt]f 5 sO2 cp N 5 Y`

[1 2 exp(2sF)] (4a)

Y` 5 [sp sO2 s] N (4b)

If we also account for the reduction in the den-sity of water molecules in the initial column Nthese forms change only slightly The size of thetwo parameters Y` and s can be obtained by fit-ting the yield versus fluence data such as that inFig 3

For the low-energy electron data Sieger et al(1998) showed that the ratio [YY`] for a givenincident electron energy is independent of tem-perature over a narrow range of temperatureTherefore the destruction cross section s in Eq4a is roughly independent of T and can be esti-mated from the [YY`] curve Sieger et al (1998)found a precursor destruction cross section s lt08ndash30 3 10216 cm2 for incident electron energiesfrom 30 to 100 eV This is of the order of the elec-tron impact ionization cross section for H2O

Based on the above model the lack of an ob-served fluence dependence for fluences down to1017 hncm2 in the Lyman-a irradiation of ice(Westley et al 1995) can be examined The low-est fluence used implies that the precursor de-struction cross section for Lyman-a is of the or-der of the Lyman-a absorption cross section in ice(09 3 10217 cm2)

Sieger et al (1998) also assumed that sp9 5 0This would mean that the precursor is only de-stroyed by production of O2 (ie s 5 sO2) Thatis unlikely since suggested precursors can be de-stroyed by reaction with H and other radicalsTheir assumption would also mean that the crosssection for production of O2 from a precursorsO2 is large This would be surprising and ap-pear to contradict some of the data discussedabove Instead we assume that s is not equal tosO2 and the total destruction cross section s isof the order of the ionization or absorption crosssection

Having obtained s the measured temperaturedependence for production of O2 is containedonly in Y` Over a narrow range of T one canwrite Y`(T) lt Y`9 exp(2EakT) and obtain the es-

timate of the activation energies in Table 2 Ifsp9 THORN 0 as discussed then sp and sO22 in Eq 4bcannot be obtained separately from the data forY` unless an independent measure of cp is ob-tained Such a measurement is critical for de-scribing the chemical state of the irradiated sur-faces

The observed temperature dependence of Y` iscontained in sp sO2 or N Here N is the ldquodepthrdquofrom which newly formed O2 can reach the sur-face and escape given as a column density of H2OThe physical depth is obtained by dividing thecolumn density N by the molecular number den-sity of the ice Assuming that in the low-energyelectron experiments N is independent of tem-perature and is about three monolayers (Sieger etal 1998) the measured Y` gives [sO2sps]05ndash20 3 10218 cm2 over the electron impactenergy range At 110ndash120K this ratio was foundto be proportional to (E 2 Eo) for E Eo 10 eVagain assuming fixed N (Orlando et al 1998 Or-lando and Sieger 2003)

In the following section we first consider a pre-cursor like peroxide that typically forms fromtwo dissociation events We use this to show un-der what circumstances such a model does notviolate the observed linearity of the O2 yield datawith fluence Following that we relate the sim-ple precursor model above to the case wheretrapped O is the precursor We then use themodel for trapped O to examine the effect of thetrap density on the production of O2 The densityof traps (defects) is determined by the sample for-mation temperature and the subsequent radiationdamage Finally we consider the escape depthN for penetrating radiation within the precursormodel These discussions are supported by moredetailed rate equations in the Appendix

Precursor trapped peroxide

The proposed precursor peroxide is typicallyformed from products of two dissociation eventsie 2 H2O R 2 H 1 2OH R H2 1 H2O2 This wasconfirmed in low temperature ice (Gerakines etal 1996) It is also possible that a dissociation pro-ducing a trapped O followed by an excitationevent can produce peroxide 2 (H2O 1 hn) R H2

1 H2O-O 1 hn R H2O2 1 H2 (eg Khriachtchevet al 2000) The O2 yield versus fluence for inci-dent electrons is reasonably well fit by the yieldexpression above for fluences greater than1014cm2 Therefore it has been suggested that

OXIDANTS AT EUROPA 835

any precursor species must be formed in a singleevent Using a model in the Appendix in whichthe precursor is formed by two events we showthis is not necessarily the case Precursor forma-tion if it is fast could involve two dissociationevents and give the measured fluence depen-dence Although we consider peroxide here theseconclusions apply to other possible precursors

In the example in the Appendix H2O is disso-ciated to H 1 OH according to the cross sectionsd The reactions of two OH units to form H2O2(rate constant k2OH) and 2 H to form H2 (k2H)compete with the recombination of H and OH toH2O (kHOH) If k2H is much greater than k2OH then H2 loss dominates at low fluences as ex-pected (Appendix) Steady state requires not sur-prisingly that the rate for forming H2O2 isroughly equal to the rate for forming H2 Finallyit is seen in the Appendix that the low H2O2 con-centrations suggested by experiment imply thatkHOH (k2H k2OH)05 Such sizes are reasonable

The set of rate equations in the Appendix thatuse number densities of species is integrated overdepth in order to obtain the forms used in Eqs 1and 2 which are based on column densities Inthis way the effective cross sections are related tothe rate constants for the interaction of freshlyproduced mobile species with previously pro-duced trapped species We show in the Appen-dix that the O2 yield versus fluence can be writ-ten in a form similar to that in Eq 4a but with asmall offset due to the two-step precursor for-mation process

Y lt Y`(T) [1 2 exp(2sF) 2 (ssr)]

for F sr21 (4c)

Here s is again the precursor destruction crosssection but sr 5 (2s2OH 1 sHOH) is the effectivecross section for loss of OH by formation of H2O2

and H2O The corresponding expression for Y`(T)is given in the Appendix

The form for the yield versus fluence in Eq 4cis consistent with the low-energy electron mea-surements if (ssr) 1 and if sr F21 lt10214 cm2 Such sizes are indeed possible Whatis required is that at the relevant temperatures(50ndash110K) the interaction length associated withsr is more than a few molecular layers (greaterthan 10 Aring) A cross section that is likely to belarge above 20K is sHOH which describes themean distance for diffusion of mobile H before it

reacts Since s discussed earlier is of the sameorder of magnitude as the H2O dissociation crosssection sd then the precursor concentration cp(in this case H2O2) is roughly proportional to[s2OHsHOH] 1 (Appendix)

Based on the above a two-step precursor for-mation process could be consistent with the elec-tron data contrary to the statements in Sieger etal (1998) Orlando et al (1998) and Orlando andSieger (2003) Experiments are needed to deter-mine the relevant cross sections Below we con-sider a precursor formed in a single excitationevent

Precursor trapped O

O2 can be formed for instance by diffusion ofa transiently mobile O to a previously formed andtrapped O or to the production of O near the siteof a previously produced and trapped O (Matichet al 1993 Johnson and Quickenden 1997) Werefer to this as the MatichndashJohnsonndashQuickenden(MJQ) model and note that O 1 OH R O2 1 Hcan also give molecular oxygen

Trapped O as H2O-O was in fact seen to be aprecursor to O2 when attacked by a mobile O ina Kr matrix (Khriachtchev et al 2000) ThereforeO trapped as H2O-O was also suggested as a pre-cursor in ice (Johnson 2001) Before proceedingwe note that the examination of the MJQ modelin Sieger et al (1998) was incorrect They consid-ered two mobile interacting species whose life-times were long compared with the bombard-ment rate rather than a new mobile O interactingwith a previously trapped O Although this hasnot been corrected recently they added Otrapped as H2O-O as a possible precursor (Or-lando and Sieger 2003) O trapped as H2O-O isseen in matrix isolation studies as discussedabove but it has not yet been identified in iceand like OH it may trap in a number of ways

The MJQ model is essentially the followingFormation of H2 1 O is followed by loss of H2and trapping of O (Ot) A subsequent dissocia-tion or excitation event can lead to the destruc-tion of Ot by reaction with for instance H or byconversion to H2O2 However if this dissociationevent also leads to H2 1 O a mobile O is pro-duced that can react with Ot with an interactionlength that depends on T This results in the for-mation of O2 [O 1 Ot R O2] giving a net pro-duction of 2 H2 and 1 O2 In the Appendix therelevant rate equations (Eqs A1cndashA3c) are con-

JOHNSON ET AL836

sidered Assuming the mobile O processes arefast these rate equations can also be integratedand simplified to give the form in Eqs 1 and 2The precursor column density Np now repre-sents the column density of Ot In this model sO2in Eq 2 is the ldquocross sectionrdquo for reaction of afreshly produced O with an Ot and sp9 indicatesdestruction of Ot Finally sp is the cross sectionfor production of H2 and Ot via processes dis-cussed above

Based on the sizes of the cross sections ex-tracted in the electron irradiation experimentsthe model with trapped O (Ot) as the precursoris quantitatively reasonable That is for a rangeof T both sp and sO2 might be described byevents of the type H2O R H2 1 O (whether di-rect or indirect) If these occur for 10 of theradiation-induced dissociation events as dis-cussed then the required product of cross sec-tions suggested by the sizes of s and Y` above isroughly obtained that is [sp sO2] 02ndash6 3 10234

cm2 in the low-energy electron experiments Thecross section for production of O might dependon the availability of dangling bonds at interfacesin pores or at defects as suggested by a numberof authors The importance of such bonds couldalso be consistent with the differences in the elec-tron-induced yield for initially crystalline and ini-tially amorphous ice (Orlando and Sieger 2003)

O2 production can increase in the presence ofoxygen-rich impurities such as CO2 or SO2 bothof which are known to be present at Europa Asin the MJQ model dissociation (eg CO2 R CO1 O) can provide a mobile O that can be trappedor react with a previously trapped O forming O2In addition such species are hydrogen scav-engers (Moore and Hudson 2000) This can re-duce the destruction of Ot affecting sp as seen inEq A5c

Trap density

In the simplified models above the density oftrapping sites nt was not explicitly accountedfor However in the discussion of both the ex-perimental data and models trapping sites havebeen assumed to be important It has beendemonstrated repeatedly that the density of de-fects voids and interfaces decreases as the tem-perature of formation of the ice increases How-ever the efficiency of production of O2 is theopposite it increases with increasing T There-fore the oxygen yield varies in a manner that is

the inverse of the trap density over a range oftemperatures For penetrating ions in which theenergy density deposited is high and O2 might beformed in a single track a simple explanation hasbeen given As the density of traps for O2 is re-duced and the temperature increases the O2 pro-duced can diffuse to the surface and escape moreefficiently limiting back reactions on further ir-radiation This is examined in the next sectionHowever such a process does not affect the low-energy electron yields for which the penetrationdepth is small and the formed O2 escapes read-ily at the measurement temperatures In Eqs A1cA1c9 A3c and A10 in the Appendix we againuse a simple set of rate equations for trapped andmobile O in order to determine the role of the trapdensity nt on the production of a precursorspecies The conclusions apply to any modelbased on a trapped precursor

In the MQJ model mobile O can be trappedforming Ot (described by kOt nt) and can reactwith previously trapped O to form O2 (kOOt) Inaddition it can react at other sites and Ot can bedestroyed by mobile reactive species such as H(kHOt nH) Mobile H can also react with othertrapped species (kHR nR) Incorporating these re-actions Eq A10 in the Appendix gives a steady-state yield Y` When the density of traps is largethe steady-state yield is given in Eq A5c

Y` R (sd9sd9sd) [kOOtkHR nr][kHOt kOt nt] N (5)

Here sd and sd9 are the dissociation cross sectionsthat determine the production of H (H 1 OH) andO (H2 1 O) by the incident radiation Comparingthe result in Eq 5 with Y` in the simple precur-sor model in Eq 4b the dissociation cross sec-tions in Eq 5 are now scaled by a ratio of the re-action rates

It is seen in Eq 5 that the steady-state yield Y`depends inversely on the density of traps nt Thatis there are competing destruction processes fortrapped O (Ot) Therefore in order to form O2 amobile O competes with mobile H to find atrapped O before it becomes trapped In this waythe increasing O2 yield correlates with the re-duction in the trapping sites with increasing tem-perature of formation The small activation ener-gies discussed earlier would be related to there-orientation of the lattice binding leading to thereduction in trap density At relatively high T thedensity of traps becomes small so that only the

OXIDANTS AT EUROPA 837

sample surface or grain interfaces might act astraps accounting for the observed drop in theyield at high T For this reason it is important todetermine if the yield for T greater than 150Kdepends on the incident flux f Since other quan-tities in the expression for Y` could also be tem-perature dependent experiments are neededThese would likely be pulsed radiolysis experi-ments in which the transients are studied ratherthan the steady-state yields However the avail-able data for lightly ionizing radiation do suggestthat the O2 yield depends inversely on trap den-sity as in the model above

Escape from depth

The quantity N in Eq 1 is the ldquodepthrdquo given asa molecular column density from which O2 formedbelow the surface can percolate to the surface andescape It can be converted to a physical depth bydividing by the molecular number density of theice In the case of the low-energy electrons thedepth below the surface at which ionizations andexcitations are produced is small However forfast ions that penetrate many monolayers this isnot the case Below we examine N in Eq 1

Although it has long been understood that twoexcitations events are required to produce O2(eg Brown et al 1982 Reimann et al 1984) theprecursor model suggests that at saturation(steady state) any new event can produce O2 fromthe steady-state density of trapped precursorsTherefore one might expect the yield at satura-tion to be nearly linear in the excitation cross sec-tion This is the case for the low-energy electronsand it was initially assumed to be the case for thefast ions (Brown et al 1980 Reimann et al 1984)However Baragiola et al (2003) showed that thetotal sputtering yield in the saturation region attemperatures greater than 100K which is dom-inated by decomposition (Brown et al 1982) isproportional to the square of the electronic en-ergy deposition per unit path length in the solid(dEdx)e Assuming that the production and de-struction cross sections in Y` in Eq 4b roughlyscale with (dEdx)e then their ratio [sp sO2s]also roughly varies as (dEdx)e If this is the casethen the observed quadratic dependence impliesthat the depth N from which O2 can be mobi-lized to escape must also depend on the excita-tion density (dEdx)e This is consistent with theobservation that the O2 yield depends on thick-ness for samples thinner than the ion penetration

depth (Reimann et al 1984 Benit and Brown1990) Therefore the percolation depth for escapeN is a critical factor in comparing yields for pen-etrating radiation with those for non-penetratingradiation like the low-energy electron and low-energy ion data It is also seen that percolationand trapping can be included within the simplemodel considered here

Model summary

We have examined the simple precursor modelintroduced by Sieger et al (1998) to explain theirelectron data We have shown how it can be gen-eralized and derived from an integration of theappropriate chemical rate equations with thecross sections and total column density contain-ing the reaction rates and the physics of diffusivetransport Therefore the parameters can havevery different meanings from those initially as-sumed The equations are useful when there arefast processes which go to completion in timesmuch shorter then the time between particle im-pacts and slow processes associated with speciesthat are assumed to be trapped between particleimpacts This separation allowed us to examinethose models proposed for precursor formationand destruction due to reacting mobile speciesformed from dissociation events Further fromthe size of the onset of the linearity of the yieldversus fluence we can place constraints on therole of nonlinear processes such as those thatmight be associated with the formation of a pre-cursor like peroxide

With these extensions this simple model is use-ful for approximating the full rate equations inthe temperature region for which the yields de-pend on the fluence (dose) and not on the doserate Based on the assumptions in the model thisimplies temperatures at which precursor speciesare stable over experimental times At the high-est temperatures for the yields shown in Fig 4athis might not be the case This should be testedexperimentally In this regard it is puzzling thatexperiments have shown that the absorptionbands associated with certain trapped speciessuch as OH and trapped electrons ldquodisappearrdquoat 100ndash120K (see eg Johnson and Quickenden1997) whereas the O2 yield data indicate that rel-evant precursor species must be stable to 140KTherefore the proposed O2 production processescan only be distinguished by directly determin-ing the chemical state of the irradiated ice

JOHNSON ET AL838

It may be the case that a particular precursoris not required as mentioned earlier Rather af-ter a sufficient level of oxidation a newly pro-duced O or OH can react in the oxygen-rich lat-tice producing O2 in a pseudo-first-order processThis can be tested experimentally Indeedtrapped O and peroxide may be changed into oneanother as in the matrix isolation experiments(Khriachtchev et al 2000) Therefore their rela-tive concentrations versus temperature and doseshould be studied

In the model considered here it is seen that theparameters in Y` must account for the strong tem-perature dependence Since the number of bulktrapping sites decreases with increasing T as theice sample is annealed one might expect that theprecursor concentration would decrease We sug-gest that the observed temperature dependenceis determined by the competition between theavailability of trapping sites and mean interactiondistances of a mobile species such as O and HThat is the O2 yield varies inversely with the den-sity of traps over a range of temperatures In thesimple model above this effect is included indi-rectly as shown using the more detailed equa-tions in the Appendix

EUROPA

Oxygen and peroxide production at Europa G values

The dependence of the yield on the thicknessof the ice sample for highly penetrating ions sug-gests that the O2 yield is roughly related to theamount of energy deposited in the sample There-fore the amount of O2 produced and lost fromice in steady state is often given as a G value thenumber of molecules produced per 100 eV de-posited in Table 2 In applying these G values tothe planetary environment care must be takensince the yields are temperature fluence and ex-citation density dependent They also are affectedby the presence of other proton scavengersknown to be present in Europarsquos ice

The G values for peroxide in Table 2 are thosefor production at low fluence but the G valuesfor O2 are for the production and escape of gas-phase oxygen at saturation Therefore they de-pend on the formation and destruction cross sec-tions in the model above Such G values can beused to describe the steady-state source of the am-

bient O2 atmosphere at Europa but are not help-ful in determining the ability to produce and trapO2 at depth into the icy surface Over the timescales of the experiments some O2 is seen to re-main trapped and is not destroyed by radicalsTypically there are a number of trapping sitesand recent estimates of the activation energy foran O2 molecule trapped in a substitutional site(Hori and Hondoh 2002) suggest that it can betrapped for times comparable to the time re-quired for the conversion of amorphous ice intoIh (Ayotte et al 2001) The stability of O2 inclu-sions has not been studied

To understand the relationship between the Gvalues for O2 in Table 2 we note that fast lightions and electrons lose their energy predomi-nantly by producing ionizations and excitationsThe ionization events produce secondary elec-trons that can result in further ionizations Thenumber of ionization events Ni produced by afast light ion stopping in a water ice is Ni 5EoW where Eo is the initial energy and W is theaverage energy expended per ionization event27 eV in water ice Therefore radiolysis by fastions is often described as the net effect of theshower of electrons set in motion by the incidention In this manner the G values for productionof O2 from ice by energetic light ions and elec-trons can be roughly estimated as the sum of theeffects of all the secondary electrons using theyield Ye for 10ndash100 eV electrons That is if weignore the percolation depth N we can write thenumber of O2 produced as the number of ioniza-tion events times Ye which already has in it thedestruction processes That is G NiYeEo YeW 0006 O2100 eV

This estimate of G is seen to be an order of mag-nitude larger than the keV proton value for O2 inTable 2 but is much smaller than the G value forenergetic Ne1 These differences from the simpleestimate are consistent with the observation thatthe yield in the saturation regime for penetratingions varies nonlinearly with excitation densityThe difference from the keV proton G value islikely due to the inability of O2 to escape fromdepth before destruction On the other hand thelarger G value seen using MeV Ne1 is likely dueto the high density of excitation This can resultin more efficient production of O2 increased pro-duction of damage sites for forming O2 andorincreased transport (percolation) along the tran-siently heated track That is the ability of theformed O2 to percolate to the surface during ion

OXIDANTS AT EUROPA 839

bombardment plays an essential role in deter-mining the measured yields as discussed aboveThis effect needs to be evaluated for a planetarysurface subject to particle fluxes that are many or-ders of magnitude smaller than those used in thelaboratory

In the absence of O2 yields for all of the rele-vant ions and energies an estimate of the totalsupply of O2 can be obtained using the G valuesin Table 2 For the energetic oxygen and sulfurions we use G values for neon and argon ionsThese are lower bounds as the data were obtainedat temperatures much lower than the surface tem-peratures at Europa (80ndash110K) For the fast pen-etrating electrons and protons we use the valuefor 200 keV protons in Table 2 For the heavy ionswe used the Ne1 and Ar1 values For compari-son we also give an estimate using the G valuesobtained from the low energy electron data andthe total ionization as discussed above

The G values are combined in Table 3 with theelectron and ion energy flux at Europa Althoughthere is some deflection of the flow due to in-duced fields the net energy flux to the surface isnot reduced much from these values (Paranicaset al 2002) In fact locally produced pick-up ionswhich are formed from the ejecta and acceleratedto the surface can contribute (Ip 1996) Furtheralthough G values are typically independent ofthe incident angle the loss of O2 is a surfaceprocess and is enhanced with angle That is thenumber of O2 produced within the escape depthincreases with increasing angle to the normal(steeper than the inverse cosine) This enhance-ment is 25ndash4 (Jurac et al 2001) Ignoring suchcorrections the total production rate calculated

(23ndash62 3 109 O2cm2s) is a lower bound to theactual production rate from the icy regions of Eu-roparsquos surface Recent modeling of the O2 atmo-sphere suggest that such a source rate is required(Shematovich and Johnson 2001)

Earlier estimates of the O2 source rate were in-correct because O2 production was assumed to bea fixed fraction of the H2O sputtering yield Atthe ambient temperature an ejected H2O sticks ef-ficiently (Smith and Kay 1997) to the surface ofneighboring grains in the porous regolith reduc-ing the sputtering yield for H2O (Johnson 1989)This reduction does not apply to the O2 fractionof the yield enhancing its relative importance toH2O Here we assume that all of the gas-phaseO2 produced can contribute to the ambient at-mosphere even if it is produced at a grain sur-face that is at some depth into the porous regolith

Finally Europarsquos surface is not pure ice every-where That is much of the trailing hemisphereis a hydrated sulfuric acid containing smallamounts of sulfur and SO2 and the leading hemi-sphere contains low levels of carbon CO2 andcarbonates Sodium and potassium atoms aresputtered from the icy regions (eg Leblanc et al2002) O2 production from surfaces containingthese species has not been measured C and S areboth proton scavengers and the other species canbe sources of O as discussed so that the O2 yieldfrom such regions could be enhanced Howeverthere can also be competing processes that mightreduce these rates In fact recent HST observa-tions suggest that the O2 might not be produceduniformly across Europarsquos surface (McGrath et al2000) The analysis of these observations is pre-liminary and is not simple since the observed

JOHNSON ET AL840

TABLE 3 APPLICATION TO EUROPA

OxygenImplantation Energy flux production

Species 109(cm2 s)a 109 keV(cm2 s)(1) 109(cm2 s)

H (10 keV) 0015(1) 12 0084 (072)bO (10 keV) 0015(1) 18 054cS (10 keV) 0009(1) 30 090cElectrons 62 043 (37)bPlasma (10 keV) (137 amu) 13(2) 10 03c

Total 79 23 (62)UV photons (64 eV) 46 01

aReferences are as follows 1Cooper et al (2001) [similar results in Paranicas et al (2002)] 2Mauk et al (1996)bUsing G for 200 keV protons (upper bounds in brackets are based on the low-energy electrons data as a model for

the secondary electrons produced)cUsing G for 15 MeV Ne1 at 7K a rough lower bound

emissions depend on the local plasma propertiesHowever the efficiency of O2 production in con-taminated ices needs to be studied

Porous regolith O2 interactions on grain surfaces

Although the O2 atmosphere is principally pro-duced in the near surface layers and is eventuallylost to space it also permeates Europarsquos porousregolith The porosity p (void space) of the re-golith is uncertain (Buratti 1995) It has been es-timated to be the most compact surface in theOuter Solar System p 025 on the trailing hemi-sphere and p 079 on the leading hemisphere(Buratti and Veverka 1983) and the most poroussurface p 096 globally (Domingue et al 1991Domingue and Hapke 1992) These large differ-ences are due to the uncertainties in the amountof backscattered light at small phase angles andon the photon absorption properties and grainsizes The more compact surface is likely correctGrain sizes rg of the order of 100 mm have beensuggested (Hansen and McCord 1999 2000Geissler et al 1998) Europarsquos regolith differs sig-nificantly from the lunar regolith which is asand-like surface of individual grains Analysis ofthe thermal inertia of Europarsquos surface (Grundyet al 2001) suggests that the ice grains are sin-tered forming a fairy castle-like structure Sin-tering can occur thermally or due to the incidentradiation In addition ices doped with acids andalkalis in the laboratory can be powdery solidswith the impurities primarily found at grain in-terfaces

The porous regolith affects the above estimatesof the oxygen production in three ways First forisotropic bombardment the mean distance d intothe surface at which an incident ion or photonstrikes a grain surface is d (89)rg(1 2 p) Us-ing the above value for rg then d 02 m if theporosity is very large p 96 Therefore de-sorbed molecules can come from a considerabledepth and will interact with grain surfaces Evenfor much lower porosities the effective sputteringyields can be significantly modified (Johnson1989) As discussed above the yields for ejectedH2O which sticks onto neighboring grains at theambient temperatures are reduced but those forejected O2 are not This enhances the relative im-portance of the O2 yields

Finally the surfaces and interfaces of the sin-tered grains in Europarsquos porous surface are placeswhere low temperature chemistry involving the

ambient oxygen can occur at depths that are wellbelow the nominal penetration depth of the inci-dent radiation The relative importance of suchchemistry is roughly indicated by the ratio of thetotal surface area in the regolith to the area of Eu-roparsquos visible surface Using the porosity andgrain size above and a regolith depth dr of theorder of a meter then this ratio is large 3(1 2p)drrg which is 103ndash2 3 104 for average porosi-ties of 96 and 25 respectively Therefore thegas-phase O2 molecules below the visible surfaceinteract frequently with the grain surfaces In thisway a very large volume of the subsurface couldbe altered chemically even if it is not directly ex-posed to the radiation

From the overlying structures seen in the high-resolution images of Europarsquos surface by theGalileo spacecraft (eg Prockter and Pappalardo2000) it is clear that the youngest surfaces are red-der The reddening in the visible has been attrib-uted to the presence of sulfur in the ice (McEwen1986 Johnson et al 1988 Calvin et al 1995 Carl-son et al 1999b) The amount of reddening in thespectra of the surface material decreases with ageand the surface materials appear to becomebrighter with age This might occur by the slowaccumulation of a frost over-layer or by radioly-sis (Carlson et al 2002 Johnson et al 2003) Ra-diolysis preferentially converts the sulfur in theice to sulfur dioxide or sulfuric acid both ofwhich have flat spectra in the visible In regionswhere the radiation does not directly penetratethe O2 permeating the regolith can oxidize thesulfur at the surfaces of the icy grains and at graininterfaces Similarly the carbon seen in the ice onEuroparsquos leading hemisphere (eg Carlson et al2004 Johnson et al 2003) can be oxidized by ra-diolysis in the near surface layers or by reactionson the grain surfaces at depth in the regolith bythe ambient O2 The meteoroid bombardmentthat produces the regolith also exposes fresh sur-faces for oxidation and can cause the collapse ofpores leading to the trapping of oxygen In thismanner both refractory and volatile oxidants canbe formed and trapped in the subsurface ice forpotential delivery to Europarsquos subsurface ocean

SUMMARY

The presence of a thin oxygen atmosphere andoxidants trapped in Europarsquos surface ice (egJohnson et al 2003) combined with the possible

OXIDANTS AT EUROPA 841

presence of a subsurface ocean (eg Carr et al1998) have made Europa an exciting target forfuture space exploration The molecular oxidantsproduced by radiolysis and photolysis could bespecies needed for carbon-based biochemistry inEuroparsquos putative subsurface ocean Howeverthere has been no model for the formation trap-ping and transport of relevant oxidants

In this paper we have reviewed the space ob-servations and laboratory studies related to thechemistry of formation of oxidants in ice by theincident radiation Because the laboratory dataare incomplete we examined the rate equationsfor formation of oxygen and related species byradiolysis and photolysis We showed how suchequations can be simplified to the analytic modelthat was used for fitting the fluence and temper-ature dependence of the O2 yield produced bylow-energy electrons (Sieger et al 1998) In de-riving the analytic model from the more completerate equations the competition between forma-tion and destruction processes can be studied ascan the role of trapping sites escape from depthand temperature history These equations also al-lowed us to put incident photon electron and iondata on a similar footing for the first time

The model can be used to suggest measure-ments that are needed to describe the chemicalstate of the irradiated ice and to interpret new lab-oratory data We showed that if reactions are fastthe yield data could be consistent with a precur-sor being formed by two events as is the case forperoxide However we favor trapped O as a pre-cursor as initially suggested by Matich et al(1993) for production of excited O2 Such aspecies if present could be detected spectro-scopically as can trapped O2 through bands ac-tivated by lattice perturbations (Cooper et al2003a)

The available oxygen yield data were used tocompare G values for different incident particlesThese were then used to estimate the oxygen pro-duction in the radiation environment at EuropaThe lower bounds were consistent with the pro-duction rates obtained from a recent model of Eu-roparsquos atmosphere The porosity of the regolithwas also considered It not only affects the rela-tive sputtering yields for O2 and H2O but alsoprovides an enormously increased surface areaon which the ambient gas-phase oxygen can re-act with species in the ice That this might be oc-curring is consistent with the change in re-

flectance associated with surface aging Such re-actions could also convert ambient molecularoxygen into other more refractory species that re-main trapped on subduction of Europarsquos crust

Although radiolysis produces a stable chemi-cally altered surface in the laboratory the parti-cle flux to Europarsquos surface is many orders ofmagnitude lower Therefore annealing can be ef-fective at temperatures that are much lower thanthose typically found experimentally Belowabout 120K the annealing times appear to be long(eg Baragiola 2003) so that laboratory data canoften be directly applied to Europarsquos icy surfaceHowever account has to be taken of the syner-gism between the UV and the energetic plasmaas discussed earlier (Johnson and Quickenden1997)

Transient surface melting might initiate lifeprocesses but subduction of oxidants to the pu-tative ocean is likely to be required if biochem-istry is to occur at Europa (Chyba and Hand2001) Although the temperature versus depthinto Europarsquos regolith is not well known belowthe regolith the temperature and pressure appar-ently increase with depth At some depth (ap-proximately kilometers) either very warm ice oran ocean exists (Ruiz and Tejero 2000) Thereforechemically refractory oxidants are likely to be re-quired in order to survive subduction to thewarmer subsurface regions Trapped species ifdelivered downward can become mobile andwill segregate forming inclusions (Johnson andJesser 1997) Indeed peroxide appears to segre-gate in ice (Gurman et al 1967) and these inclu-sions may remain relatively stable until the per-oxide is dissolved in the putative ocean Thevolatiles seen in reflectance such as O2 SO2 andCO2 will also segregate But at the higher tem-peratures they could also diffuse through thebulk react or percolate into the regolith atmo-sphere Therefore the observation of the trappedvolatiles in Europarsquos surface ice is of interest pri-marily because it suggests that oxidants are pre-sent The formation of oxygen-rich molecules fol-lows the preferential loss of hydrogen Howeverthe surface is in a local steady state with im-planted H O and S from the plasma and the lossto space Since the plasma flux to the surface canbe variable this balance can also vary Indeedtemporal changes in reflectance have been re-ported (Domingue and Hendrix 2004)

Based on the models for formation of O2 ex-

JOHNSON ET AL842

amined in this paper additional experiments areneeded to determine the chemical state of irra-diated ice at low fluences Of immediate inter-est is the identification of stable molecular pre-cursors required for the formation of O2 Basedon the available laboratory data individual per-oxide molecules trapped in the ice do not appearto be the principal precursors but trapped O(Matich et al 1993) remains a viable candidateAlthough the data suggest that the local hydro-gen-bonding network is important in the pro-duction of O2 the effect of the formation tem-perature and radiation damage on the structureneed to be measured It is also important to de-scribe quantitatively the relationship betweenH2 loss and the formation of O2 and H2O2 overa range of temperatures and fluences Finallyone would like to determine the steady-statedensities of biochemically interesting moleculesformed by radiolysis in an ice containing sulfurand carbon at Europarsquos surface temperaturesand then determine their stability with increas-ing temperature and pressure Such experimentsstill need to be carried out Here we have pre-sented a model that brings together the variouslaboratory data and now can be used to considerthe formation trapping and transport of relevantoxidants at Europa

APPENDIX

Introduction

In this Appendix we examine rate equationsthat describe various aspects of the radiolysis orphotolysis of ice These equations need to be un-derstood if one is to be able to calculate the for-mation trapping and transport of oxidants Theemphasis is on the formation of molecular oxy-gen We integrate these equations over the escapedepth of O2 to give the simplified rate equationsdiscussed in the text in which number densitiesare replaced by column densities (eg Eqs 1 and2) In this way we can examine the relationshipbetween the rate constants and the effective pre-cursor formation and destruction cross sectionsextracted from data We first consider a precur-sor such as H2O2 that is formed from two exci-tation events This bears on the observed linear-ity of the oxygen yield versus fluence We thenconsider the equations for trapped versus a tran-

siently mobile species We use a reaction betweena mobile and trapped O as an example althoughO 1 OH can also give oxygen Finally we ex-amine the possible dependence of the O2 forma-tion process on the density of traps (defectsvoids and grain boundaries) as a function of tem-perature The results are discussed in the text

For clarity all possible radiolytic species arenot included in the discussion below Thereforepotentially important products such as HO2 andthe charged species (eg H3O1 the trapped elec-tron etc) are missing In each section we attemptto make a point related to production of O2 andother oxidants by using a subset of the productsrather than attempt a complete description withmany unknown rate constants

Fast versus slow processes

To describe the O2 production rate from a pre-cursor such as H2O2 we first use a simple but in-structive set of rate equations to describe the for-mation of H2O2 We then relate these to thesimplified rate equation for the precursor in thetext Eq 1 and show how the postulated precur-sor formation cross section sp is related to theH2O dissociation cross section sd by the reactionrates

Using the reactions

H2O 1 radiation R H 1 OH (sd)

H 1 H R H2 (k2H)

H 1 OH R H2O (kHOH)

OH 1 OH R H2O2 (k2OH)

the reaction pathways are described by

H2

Hqk2H

1 OHH2O 1 radiation mdashRsd H

OHH2O2

OH Rk2OH

kHOH

HH2O

OXIDANTS AT EUROPA 843

The corresponding rate equations in an irradiatedvolume are

dnOHdt 5 sd f n 2 2 k2OH nOH nOH

2 kHOH nH nOH (A1a)

dnHdt 5 sd f n 2 2 k2H nH nH

2 kHOH nH nOH (A2a)

dnH2O2dt 5 k2OH nOH nOH

2 destruction processes (A3a)

Here sd is the dissociation cross section f is theradiation flux and n is the molecular numberdensity of ice with nOH and nH the density of dis-sociation products OH and H respectively Therate constants k2OH k2H and kHOH describe thereactions above In this model the H2 formed isassumed to escape for temperatures above about20K We do not distinguish between trappedspecies and mobile species although in fact thatdistinction is important as discussed in the nextsection That is any OH produced at low tem-perature by one ion would likely be trapped priorto interacting with an OH produced by a subse-quent ion At higher temperatures this can breakdown as discussed in the text We first relate theabove set of equations to Eq 1 in the text and thenconsider precursor destruction processes

Equation 1 in the text is obtained by solvingthese equations together and then by integratingover a depth Dx Here Dx is the smaller of the pen-etration depth of the particle and the O2 escape(percolation) depth Assuming the molecular den-sity of ice n is constant with depth Dx n 5 N inEq 1 Comparing Eq 1 with Eq A3a the precur-sor formation rate sp f N is equivalent to [k2OH

nOH nOH] Dx Assuming that the processes in EqsA1a and A2a are fast then the steady state is read-ily obtained In addition if k2H k2OH there ispreferential loss of H2 at short times as suggestedby the laboratory data If steady state is reachedthe production of H2 is equal to that of H2O2 us-ing only the equations above Further one obtains

[k2OH nOH nOH] Dx R [fp sd] f N (A4)

with

fp 5 05[1 1 05 kHOH(k2H k2OH)05] (A5a)

Therefore the precursor formation cross sectionsp in Eq 1 is a fraction fp of the total dissocia-

tion cross sd Since laboratory data suggest theH2O2 fraction in an irradiated ice is small thenkHOH (k2OH k2H)05 which is likely givingfp R [(k2OH k2H)05kHOH]

Trapped versus mobile species

To examine the fluence dependence discussedin the text we note that trapped OH and freshlyproduced transiently mobile OH can be treatedas different species (eg Matich et al 1993)Therefore assuming no trapping of H the OHrate equation above would be replaced by twoequations one for the trapped species nOHt andthe other for the freshly produced mobile speciesnOH Assuming the reaction of two mobile OHunits is unlikely and that peroxide is formed fromthe interaction of a trapped OH with a subse-quently produced OH an expanded set of equa-tions is obtained

OH 1 trap R OHt (kOHt)

OH 1 OHt R H2O2 (kOHOHt)

H 1 OHt R H2O (kHOHt)

then the reaction pathways are described by

qH2

H2O 1 radiation mdashRsdH

qk2H

1 OHH

trap OH

OH RkOHt

OHt kOHOHtH2O2

kHOHkHOHt

H H2O HH2O

The resulting set of equations is

dnOHdt 5 sd f n 2 kHOH nH nOH

2 kOHt nOH nt 2 kOHOHt nOH nOHt (A1b)

dnOHtdt 5 kOHt nOH nt 2 kOHOHt nOH nOHt

2 kHOHt nH nOHt (A1b9)

dnHdt 5 sd f n 2 2 k2H nH nH 2

kHOH nH nOH 2 kHOHt nH nOHt (A2b)

JOHNSON ET AL844

dnH2O2dt 5 kOHOHt nOH nOHt

2 destruction processes (A3b)

where nt is the trap density in the irradiated iceAt low T we can assume nt is large so thatdntdt 0 simplifying the set of equations If theprecursor of interest is trapped O as discussed inthe text then a similar set of equations can be con-structed for O and Ot as in the example to followOf course a full description over a broad rangeof temperatures would require the considerationof the both trapped and mobile H O and OHthe rate of change of trap density and the vari-ous destruction processes

The above equations should be solved simul-taneously but for the incident particle fluxes typ-ically used in these studies the diffusion of mo-bile species to traps is relatively fast This is trueeven at low temperatures if there is a significantdensity of trapping sites This is usually the casefor samples formed at low T Therefore in a lowtemperature solid at low dose rates there is not asignificant steady-state background of diffusingOH There are trapped or freshly produced OHunits which rapidly react Averaging over thebombardment rate this can be approximated bywriting dnOHdt 0 and then separating nOH outin Eq A1b Assuming also that H diffuses rapidlyand either reacts or forms H2 we also writednHdt 0 on the average Therefore betweenimpacts only species trapped at defects pores orgrain surfaces are assumed to be present as sug-gested by the lack of a dependence on the inci-dent beam flux In this limit the rate of change oftrapped OH roughly simplifies to

dnOHtdt 5 2 k2H nH nH 2 2 kOHOHt nOH nOHt

It is seen that the production of peroxide dependson the loss of H by formation of H2 as discussedand at steady state the production rate for per-oxide equals the rate of formation of H2 in theseequations

Treating the mobile and trapped species sepa-rately we can further simplify the rate equationsand obtain the precursor model used in the textThat is integrating over the relevant depth Dxthe densities are replaced by column densitiesThe rate constants ki are then related to effectivecross sections si In this way cross sections aregiven as an interaction length for each freshlyproduced mobile radical [ki nj] R [si f] The

trapped OH column density NOHt and peroxideproduction rate can be estimated from ldquolinearrdquorate equations

dNOHtdt lt [sd f] N 2 2 [s2OH f] NOHt

2 [sHOH f] NOHt (A6a)

dNH2O2dt lt [s2OH f] NOHt 2 s NH2O2 (A6b)

Here sd is again the dissociation cross sectionand the destruction processes in Eq A3b are con-tained in s The second and third terms on theright in Eq A6a are the loss of OH due to the pro-duction of H2O2 and H2O In the integration ofEq A1b9 [2 kOH OHt nOH] becomes [2 s2OH f] and[kHOHt nOH] becomes [sHOH f] Therefore thedensity and the interaction length of mobile H arecontained in sHOH If H2O2 is indeed the pre-cursor then in Eq 2 in the text [sp f N] is be-comes [s2OH f NOHt] from Eq A6b Accountingfor precursor destruction which can give backtwo OH and treating H2 can both be included inthe integrated model via the si values

The simplified set of equations above allows usto examine the relative time (fluence) scales forprecursor formation (in this case H2O2) SolvingEqs A6a and A6b the column density of OH andthe concentration of precursors versus fluence cpcan be obtained Using the column of H2O2 pro-duced as the precursor column Np in Eq 2 thenO2 yield can be obtained The results can be com-pared with the results from Eqs1 and 2 in thetext The average precursor (H2O2) concentrationover the column depth of water molecules N is

cp 5 fp (sds)[1 2 exp(2s F)] 2 (ssr)

[1 2 exp(2sr F)] (A7)

with

fp 5 2s2OH(sr 2 s) (A5b)

In these equations sr 5 (2s2OH 1 sHOH) andssr 1

It is seen that the precursor concentration withtime cp depends on two rates the reaction ratesthat determine sr and the destruction rates in sAs in Eq 4a the yield can be written as

Y 5 sO2 cp N (A8)

If sr and s are comparable then the yield at small

OXIDANTS AT EUROPA 845

fluence is quadratic in F However if sr sthen the linear dependence on fluence is obtainedwith a small shift in the fluence as shown in Eq4c in the text Further the steady-state yield isnow

Y` (T) 5 sO2 [fp sd] Ns (A9)

That is sp R [fp sd] as in Eq A4 and the tem-perature dependence for the incident electrondata must be contained in [sO2 fp] or in the ratio[sO2 s2OH]sHOH]

For comparison with their precursor modelSieger et al (1998) examined the possibility of ob-taining the O2 precursor by considering reactionsfor a density of diffusing dissociation products asin Eqs A1 and A2 rather than a trapped productinteracting with a freshly formed dissociationproduct Based on the above the lowest fluence atwhich O2 is first detected experimentally sets alimit on the time scale for mobilendashmobile reactions

Effect of trap density

Equations A1b A1b9 A2b and A3b can be ap-proximately solved assuming that the dnOHdtdnHdt and dntdt are very small on the averageas discussed above In order to examine the roleof the trap density nt which can depend on thetemperature and method of formation and on theradiation history of the ice we use the slightlysimpler precursor model In this model the O pro-duced directly by dissociation traps forming aprecursor O2 is then formed when a freshly pro-duced O or other species reacts with a trappedoxygen atom Ot

H2O 1 radiation R H2 1 O (sd)

O 1 trap R Ot (kOt)

O 1 Ot R O2 (kOOt)

The reaction pathways for the oxygen species are

H2O 1 radiation mdashRsd9

trap OO2

kOtOt kOOt

H2 1 OOHt

kOR H(kHO t)

ROR

We assume for simplicity both H2 and O2 escapefrom depth Dx after formation We also allow mo-bile H to react with Ot (kHOt) removing trappedO and both the mobile H and O to be removedby other reactants R (eg OH or contaminants)This leads to a simplified set of equations not di-rectly involving OH or peroxide

dnOdt 5 sd9 f n 2 kOt nO nt 2 kOR nO nR

2 kOOt nO nOt (A1c)

dnOtdt 5 kOt nt nO 2 kOOt nO nOt

2 kHOt nH nOt (A1c9)

dnHdt 5 sd f n 2 kHOt nH nOt

2 kHR nH nR (A2c)

dnO2dt 5 kOOt nO nOt (A3c)

These equations of course act in parallel with theequations above for OH and mobile OH may re-act with trapped O to form O2 Here we focusonly on mobile O

Since the diffusion of H and O are fast and thedensity of traps is large the nH and nt averagedover the bombardment rate can be taken veryroughly as time independent Further if we as-sume the number of trapped O nOt is small com-pared with the number of sites at which H canreact then nH [sd f nkHR nR] For a low ra-diation flux f nH is small on the average as as-sumed Since the O2 yield YO2 is equal to[(dnO2dt) Dxf] with Dx the escape depth thenthe steady-state yield Y` can be obtained ana-lytically

[Y`][sd9 N] 5 1 1 (q 2 1) 2 [(q 2 1)2

1 8 q d]054 d (A10)

q 5 [kOt (nt 1 nR)] (kHOt nH)(sd9 f n) kOOt

5(sdsd9) [kOt (nt 1 nR)] kHOt[kOOt kHR nR]

d 5 05 [1 1 nt(nt 1 nR)]

We note that for nt large q can be large Assum-ing that q 1 and nt nR then Eq A10 be-comes

Y` R [sd9 N]q 5 (sd9 N) (sd9sd)

[kOOt kHR nR][kHOt kOt nt] (A5c)

JOHNSON ET AL846

Comparing with Y` in the simple precursormodel in Eq 4b then s R [(kHOt nH)f] R [sd nkHOt][kHR nR] In addition the product [sO2sp] R sd9 sd9 [kOOt n][kOt nt] That is the dis-sociation cross sections are modified by reactionrates

Quite remarkably it is seen in Eq A5c that thesteady-state yield Y` can depend inversely onthe density of traps nt in the temperature andflux regimes for which the approximations arerelevant That is if there are competing destruc-tion process for trapped O then at high trap den-sity a mobile O must find a trapped O before itbecomes trapped or before mobile H destroys theOt In this way the increasing O2 yield with in-creasing T can be related to the change in the trap-ping density nt with the formation temperatureof the ice sample As the temperature increasesand the density of traps in the bulk decreases thesurface and grain interfaces become the only trap-ping sites

ACKNOWLEDGMENTS

REJ acknowledges support by a Gledden Fel-lowship in Chemistry School of Biomedical andChemical Sciences at The University of WesternAustralia and by grants from the NSF Astron-omy Program and NASArsquos Planetary Geologyand Geophysics Program PDC acknowledgessupport by an Australian Postgraduate Award

ABBREVIATIONS

HST Hubble Space Telescope MJQ MatichndashJohnsonndashQuickenden NIMS Near Infrared Map-ping Spectrometer UV ultraviolet

REFERENCES

Ayotte P Smith RS Stevenson KP Dohnarsquolek ZKimmel GA and Kay BD (2001) Effect of porosityon the adsorption desorption trapping and release ofvolatile gases by amorphous solid water J Geophys Res106(E12) 33387ndash33392

Bahr DA Famaacute M and Baragiola RA (2001) Radiol-ysis of water ice in the outer solar system sputteringand trapping of radiation products J Geophys Res 10633285ndash33290

Baragiola RA (2003) Microporous amorphous water icefilms and astronomical implications In Water in Con-

fining Geometries edited by JP Devlin and B Buch El-sevier New York in press

Baragiola RA and Bahr DA (1998) Laboratory studiesof the optical properties and stability of oxygen onGanymede J Geophys Res 103 25865ndash25872

Baragiola RA Atteberry CL and Bahr DA (1999) Ori-gin of solid oxygen on Ganymede Reply to commentby RE Johnson on ldquoLaboratory studies of the opticalproperties and stability of oxygen on Ganymederdquo JGeophys Res E104 14183ndash14187

Baragiola RA Vidal RA Svendsen W Schou J ShiM Bahr DA and Atteberrry CL (2003) Sputteringof water ice Nucl Instrum Methods Phys Res B 209294ndash303

Bar-Nun A Herman G Laufer D and Rappaport ML(1985) Trapping and release of gases by water ice andimplications for icy bodies Icarus 63 317ndash332

Baverstock KF and Burns WG (1976) Primary pro-duction of oxygen from irradiated water as an expla-nation for decreased radiobiological oxygen enhance-ment at high LET Nature 260 316ndash318

Benit J and Brown WL (1990) Electronic sputtering ofoxygen and water molecules from thin films of waterice bombarded by MeV Ne1 ions Nucl Instrum Meth-ods Phys Rev B 46 448ndash451

Benit J Bibring J-P della-Negra S LeBeyec YMendenhall M Rocard F and Standing K (1987)Erosion of ices by ion irradiation Nucl Instrum Meth-ods Phys Rev B 1920 838ndash842

Brown WL Augustyniak WM Lanzerotti LJ John-son RE and Evatt R (1980) Linear and nonlinearprocesses in the erosion of H2O ice by fast light ionsPhys Rev Lett 43 1632ndash1635

Brown WL Augustyniak W M Simmons E Mar-cantonio KJ Lanzerotti LJ Johnson RE BoringJW Reimann CT Foti G and Pirronello V (1982)Erosion and molecular formation in condensed gasfilms by electronic energy loss of fast ions Nucl In-strum Methods B 1 307ndash314

Brown WL Augustyniak WM Marcantonio KJ Sim-mons EN Boring JW Johnson RE and ReimannCT (1984) Electronic sputtering of low-temperaturemolecular solids Nucl Instrum Methods B 1 307

Buratti B (1995) Photometry and surface structure of theicy Galilean satellites J Geophys Res 100(E9) 19061ndash19066

Buratti B and Veverka J (1983) Voyager photometry ofEuropa Icarus 55 93ndash110

Calvin WM and Spencer JR (1997) Latitudinal distri-bution of O2 on Ganymede observations with the Hub-ble Space Telescope Icarus 130 505ndash516

Calvin WM Clark RN Brown RH and Spencer JA(1995) Spectra of the icy Galilean satellites from 02 to5 mm a compilation new observations and a recentsummary J Geophys Res 100 19041ndash19048

Calvin WM Johnson RE and Spencer JA (1996) O2

on Ganymede spectral characteristics and plasma for-mation mechanisms Geophys Res Lett 23 673ndash676

Calvin WM Aninich VG and Brown RH (2004) Oxy-gen temperature and phase effects Icarus (in press)

OXIDANTS AT EUROPA 847

any precursor species must be formed in a singleevent Using a model in the Appendix in whichthe precursor is formed by two events we showthis is not necessarily the case Precursor forma-tion if it is fast could involve two dissociationevents and give the measured fluence depen-dence Although we consider peroxide here theseconclusions apply to other possible precursors

In the example in the Appendix H2O is disso-ciated to H 1 OH according to the cross sectionsd The reactions of two OH units to form H2O2(rate constant k2OH) and 2 H to form H2 (k2H)compete with the recombination of H and OH toH2O (kHOH) If k2H is much greater than k2OH then H2 loss dominates at low fluences as ex-pected (Appendix) Steady state requires not sur-prisingly that the rate for forming H2O2 isroughly equal to the rate for forming H2 Finallyit is seen in the Appendix that the low H2O2 con-centrations suggested by experiment imply thatkHOH (k2H k2OH)05 Such sizes are reasonable

The set of rate equations in the Appendix thatuse number densities of species is integrated overdepth in order to obtain the forms used in Eqs 1and 2 which are based on column densities Inthis way the effective cross sections are related tothe rate constants for the interaction of freshlyproduced mobile species with previously pro-duced trapped species We show in the Appen-dix that the O2 yield versus fluence can be writ-ten in a form similar to that in Eq 4a but with asmall offset due to the two-step precursor for-mation process

Y lt Y`(T) [1 2 exp(2sF) 2 (ssr)]

for F sr21 (4c)

Here s is again the precursor destruction crosssection but sr 5 (2s2OH 1 sHOH) is the effectivecross section for loss of OH by formation of H2O2

and H2O The corresponding expression for Y`(T)is given in the Appendix

The form for the yield versus fluence in Eq 4cis consistent with the low-energy electron mea-surements if (ssr) 1 and if sr F21 lt10214 cm2 Such sizes are indeed possible Whatis required is that at the relevant temperatures(50ndash110K) the interaction length associated withsr is more than a few molecular layers (greaterthan 10 Aring) A cross section that is likely to belarge above 20K is sHOH which describes themean distance for diffusion of mobile H before it

reacts Since s discussed earlier is of the sameorder of magnitude as the H2O dissociation crosssection sd then the precursor concentration cp(in this case H2O2) is roughly proportional to[s2OHsHOH] 1 (Appendix)

Based on the above a two-step precursor for-mation process could be consistent with the elec-tron data contrary to the statements in Sieger etal (1998) Orlando et al (1998) and Orlando andSieger (2003) Experiments are needed to deter-mine the relevant cross sections Below we con-sider a precursor formed in a single excitationevent

Precursor trapped O

O2 can be formed for instance by diffusion ofa transiently mobile O to a previously formed andtrapped O or to the production of O near the siteof a previously produced and trapped O (Matichet al 1993 Johnson and Quickenden 1997) Werefer to this as the MatichndashJohnsonndashQuickenden(MJQ) model and note that O 1 OH R O2 1 Hcan also give molecular oxygen

Trapped O as H2O-O was in fact seen to be aprecursor to O2 when attacked by a mobile O ina Kr matrix (Khriachtchev et al 2000) ThereforeO trapped as H2O-O was also suggested as a pre-cursor in ice (Johnson 2001) Before proceedingwe note that the examination of the MJQ modelin Sieger et al (1998) was incorrect They consid-ered two mobile interacting species whose life-times were long compared with the bombard-ment rate rather than a new mobile O interactingwith a previously trapped O Although this hasnot been corrected recently they added Otrapped as H2O-O as a possible precursor (Or-lando and Sieger 2003) O trapped as H2O-O isseen in matrix isolation studies as discussedabove but it has not yet been identified in iceand like OH it may trap in a number of ways

The MJQ model is essentially the followingFormation of H2 1 O is followed by loss of H2and trapping of O (Ot) A subsequent dissocia-tion or excitation event can lead to the destruc-tion of Ot by reaction with for instance H or byconversion to H2O2 However if this dissociationevent also leads to H2 1 O a mobile O is pro-duced that can react with Ot with an interactionlength that depends on T This results in the for-mation of O2 [O 1 Ot R O2] giving a net pro-duction of 2 H2 and 1 O2 In the Appendix therelevant rate equations (Eqs A1cndashA3c) are con-

JOHNSON ET AL836

sidered Assuming the mobile O processes arefast these rate equations can also be integratedand simplified to give the form in Eqs 1 and 2The precursor column density Np now repre-sents the column density of Ot In this model sO2in Eq 2 is the ldquocross sectionrdquo for reaction of afreshly produced O with an Ot and sp9 indicatesdestruction of Ot Finally sp is the cross sectionfor production of H2 and Ot via processes dis-cussed above

Based on the sizes of the cross sections ex-tracted in the electron irradiation experimentsthe model with trapped O (Ot) as the precursoris quantitatively reasonable That is for a rangeof T both sp and sO2 might be described byevents of the type H2O R H2 1 O (whether di-rect or indirect) If these occur for 10 of theradiation-induced dissociation events as dis-cussed then the required product of cross sec-tions suggested by the sizes of s and Y` above isroughly obtained that is [sp sO2] 02ndash6 3 10234

cm2 in the low-energy electron experiments Thecross section for production of O might dependon the availability of dangling bonds at interfacesin pores or at defects as suggested by a numberof authors The importance of such bonds couldalso be consistent with the differences in the elec-tron-induced yield for initially crystalline and ini-tially amorphous ice (Orlando and Sieger 2003)

O2 production can increase in the presence ofoxygen-rich impurities such as CO2 or SO2 bothof which are known to be present at Europa Asin the MJQ model dissociation (eg CO2 R CO1 O) can provide a mobile O that can be trappedor react with a previously trapped O forming O2In addition such species are hydrogen scav-engers (Moore and Hudson 2000) This can re-duce the destruction of Ot affecting sp as seen inEq A5c

Trap density

In the simplified models above the density oftrapping sites nt was not explicitly accountedfor However in the discussion of both the ex-perimental data and models trapping sites havebeen assumed to be important It has beendemonstrated repeatedly that the density of de-fects voids and interfaces decreases as the tem-perature of formation of the ice increases How-ever the efficiency of production of O2 is theopposite it increases with increasing T There-fore the oxygen yield varies in a manner that is

the inverse of the trap density over a range oftemperatures For penetrating ions in which theenergy density deposited is high and O2 might beformed in a single track a simple explanation hasbeen given As the density of traps for O2 is re-duced and the temperature increases the O2 pro-duced can diffuse to the surface and escape moreefficiently limiting back reactions on further ir-radiation This is examined in the next sectionHowever such a process does not affect the low-energy electron yields for which the penetrationdepth is small and the formed O2 escapes read-ily at the measurement temperatures In Eqs A1cA1c9 A3c and A10 in the Appendix we againuse a simple set of rate equations for trapped andmobile O in order to determine the role of the trapdensity nt on the production of a precursorspecies The conclusions apply to any modelbased on a trapped precursor

In the MQJ model mobile O can be trappedforming Ot (described by kOt nt) and can reactwith previously trapped O to form O2 (kOOt) Inaddition it can react at other sites and Ot can bedestroyed by mobile reactive species such as H(kHOt nH) Mobile H can also react with othertrapped species (kHR nR) Incorporating these re-actions Eq A10 in the Appendix gives a steady-state yield Y` When the density of traps is largethe steady-state yield is given in Eq A5c

Y` R (sd9sd9sd) [kOOtkHR nr][kHOt kOt nt] N (5)

Here sd and sd9 are the dissociation cross sectionsthat determine the production of H (H 1 OH) andO (H2 1 O) by the incident radiation Comparingthe result in Eq 5 with Y` in the simple precur-sor model in Eq 4b the dissociation cross sec-tions in Eq 5 are now scaled by a ratio of the re-action rates

It is seen in Eq 5 that the steady-state yield Y`depends inversely on the density of traps nt Thatis there are competing destruction processes fortrapped O (Ot) Therefore in order to form O2 amobile O competes with mobile H to find atrapped O before it becomes trapped In this waythe increasing O2 yield correlates with the re-duction in the trapping sites with increasing tem-perature of formation The small activation ener-gies discussed earlier would be related to there-orientation of the lattice binding leading to thereduction in trap density At relatively high T thedensity of traps becomes small so that only the

OXIDANTS AT EUROPA 837

sample surface or grain interfaces might act astraps accounting for the observed drop in theyield at high T For this reason it is important todetermine if the yield for T greater than 150Kdepends on the incident flux f Since other quan-tities in the expression for Y` could also be tem-perature dependent experiments are neededThese would likely be pulsed radiolysis experi-ments in which the transients are studied ratherthan the steady-state yields However the avail-able data for lightly ionizing radiation do suggestthat the O2 yield depends inversely on trap den-sity as in the model above

Escape from depth

The quantity N in Eq 1 is the ldquodepthrdquo given asa molecular column density from which O2 formedbelow the surface can percolate to the surface andescape It can be converted to a physical depth bydividing by the molecular number density of theice In the case of the low-energy electrons thedepth below the surface at which ionizations andexcitations are produced is small However forfast ions that penetrate many monolayers this isnot the case Below we examine N in Eq 1

Although it has long been understood that twoexcitations events are required to produce O2(eg Brown et al 1982 Reimann et al 1984) theprecursor model suggests that at saturation(steady state) any new event can produce O2 fromthe steady-state density of trapped precursorsTherefore one might expect the yield at satura-tion to be nearly linear in the excitation cross sec-tion This is the case for the low-energy electronsand it was initially assumed to be the case for thefast ions (Brown et al 1980 Reimann et al 1984)However Baragiola et al (2003) showed that thetotal sputtering yield in the saturation region attemperatures greater than 100K which is dom-inated by decomposition (Brown et al 1982) isproportional to the square of the electronic en-ergy deposition per unit path length in the solid(dEdx)e Assuming that the production and de-struction cross sections in Y` in Eq 4b roughlyscale with (dEdx)e then their ratio [sp sO2s]also roughly varies as (dEdx)e If this is the casethen the observed quadratic dependence impliesthat the depth N from which O2 can be mobi-lized to escape must also depend on the excita-tion density (dEdx)e This is consistent with theobservation that the O2 yield depends on thick-ness for samples thinner than the ion penetration

depth (Reimann et al 1984 Benit and Brown1990) Therefore the percolation depth for escapeN is a critical factor in comparing yields for pen-etrating radiation with those for non-penetratingradiation like the low-energy electron and low-energy ion data It is also seen that percolationand trapping can be included within the simplemodel considered here

Model summary

We have examined the simple precursor modelintroduced by Sieger et al (1998) to explain theirelectron data We have shown how it can be gen-eralized and derived from an integration of theappropriate chemical rate equations with thecross sections and total column density contain-ing the reaction rates and the physics of diffusivetransport Therefore the parameters can havevery different meanings from those initially as-sumed The equations are useful when there arefast processes which go to completion in timesmuch shorter then the time between particle im-pacts and slow processes associated with speciesthat are assumed to be trapped between particleimpacts This separation allowed us to examinethose models proposed for precursor formationand destruction due to reacting mobile speciesformed from dissociation events Further fromthe size of the onset of the linearity of the yieldversus fluence we can place constraints on therole of nonlinear processes such as those thatmight be associated with the formation of a pre-cursor like peroxide

With these extensions this simple model is use-ful for approximating the full rate equations inthe temperature region for which the yields de-pend on the fluence (dose) and not on the doserate Based on the assumptions in the model thisimplies temperatures at which precursor speciesare stable over experimental times At the high-est temperatures for the yields shown in Fig 4athis might not be the case This should be testedexperimentally In this regard it is puzzling thatexperiments have shown that the absorptionbands associated with certain trapped speciessuch as OH and trapped electrons ldquodisappearrdquoat 100ndash120K (see eg Johnson and Quickenden1997) whereas the O2 yield data indicate that rel-evant precursor species must be stable to 140KTherefore the proposed O2 production processescan only be distinguished by directly determin-ing the chemical state of the irradiated ice

JOHNSON ET AL838

It may be the case that a particular precursoris not required as mentioned earlier Rather af-ter a sufficient level of oxidation a newly pro-duced O or OH can react in the oxygen-rich lat-tice producing O2 in a pseudo-first-order processThis can be tested experimentally Indeedtrapped O and peroxide may be changed into oneanother as in the matrix isolation experiments(Khriachtchev et al 2000) Therefore their rela-tive concentrations versus temperature and doseshould be studied

In the model considered here it is seen that theparameters in Y` must account for the strong tem-perature dependence Since the number of bulktrapping sites decreases with increasing T as theice sample is annealed one might expect that theprecursor concentration would decrease We sug-gest that the observed temperature dependenceis determined by the competition between theavailability of trapping sites and mean interactiondistances of a mobile species such as O and HThat is the O2 yield varies inversely with the den-sity of traps over a range of temperatures In thesimple model above this effect is included indi-rectly as shown using the more detailed equa-tions in the Appendix

EUROPA

Oxygen and peroxide production at Europa G values

The dependence of the yield on the thicknessof the ice sample for highly penetrating ions sug-gests that the O2 yield is roughly related to theamount of energy deposited in the sample There-fore the amount of O2 produced and lost fromice in steady state is often given as a G value thenumber of molecules produced per 100 eV de-posited in Table 2 In applying these G values tothe planetary environment care must be takensince the yields are temperature fluence and ex-citation density dependent They also are affectedby the presence of other proton scavengersknown to be present in Europarsquos ice

The G values for peroxide in Table 2 are thosefor production at low fluence but the G valuesfor O2 are for the production and escape of gas-phase oxygen at saturation Therefore they de-pend on the formation and destruction cross sec-tions in the model above Such G values can beused to describe the steady-state source of the am-

bient O2 atmosphere at Europa but are not help-ful in determining the ability to produce and trapO2 at depth into the icy surface Over the timescales of the experiments some O2 is seen to re-main trapped and is not destroyed by radicalsTypically there are a number of trapping sitesand recent estimates of the activation energy foran O2 molecule trapped in a substitutional site(Hori and Hondoh 2002) suggest that it can betrapped for times comparable to the time re-quired for the conversion of amorphous ice intoIh (Ayotte et al 2001) The stability of O2 inclu-sions has not been studied

To understand the relationship between the Gvalues for O2 in Table 2 we note that fast lightions and electrons lose their energy predomi-nantly by producing ionizations and excitationsThe ionization events produce secondary elec-trons that can result in further ionizations Thenumber of ionization events Ni produced by afast light ion stopping in a water ice is Ni 5EoW where Eo is the initial energy and W is theaverage energy expended per ionization event27 eV in water ice Therefore radiolysis by fastions is often described as the net effect of theshower of electrons set in motion by the incidention In this manner the G values for productionof O2 from ice by energetic light ions and elec-trons can be roughly estimated as the sum of theeffects of all the secondary electrons using theyield Ye for 10ndash100 eV electrons That is if weignore the percolation depth N we can write thenumber of O2 produced as the number of ioniza-tion events times Ye which already has in it thedestruction processes That is G NiYeEo YeW 0006 O2100 eV

This estimate of G is seen to be an order of mag-nitude larger than the keV proton value for O2 inTable 2 but is much smaller than the G value forenergetic Ne1 These differences from the simpleestimate are consistent with the observation thatthe yield in the saturation regime for penetratingions varies nonlinearly with excitation densityThe difference from the keV proton G value islikely due to the inability of O2 to escape fromdepth before destruction On the other hand thelarger G value seen using MeV Ne1 is likely dueto the high density of excitation This can resultin more efficient production of O2 increased pro-duction of damage sites for forming O2 andorincreased transport (percolation) along the tran-siently heated track That is the ability of theformed O2 to percolate to the surface during ion

OXIDANTS AT EUROPA 839

bombardment plays an essential role in deter-mining the measured yields as discussed aboveThis effect needs to be evaluated for a planetarysurface subject to particle fluxes that are many or-ders of magnitude smaller than those used in thelaboratory

In the absence of O2 yields for all of the rele-vant ions and energies an estimate of the totalsupply of O2 can be obtained using the G valuesin Table 2 For the energetic oxygen and sulfurions we use G values for neon and argon ionsThese are lower bounds as the data were obtainedat temperatures much lower than the surface tem-peratures at Europa (80ndash110K) For the fast pen-etrating electrons and protons we use the valuefor 200 keV protons in Table 2 For the heavy ionswe used the Ne1 and Ar1 values For compari-son we also give an estimate using the G valuesobtained from the low energy electron data andthe total ionization as discussed above

The G values are combined in Table 3 with theelectron and ion energy flux at Europa Althoughthere is some deflection of the flow due to in-duced fields the net energy flux to the surface isnot reduced much from these values (Paranicaset al 2002) In fact locally produced pick-up ionswhich are formed from the ejecta and acceleratedto the surface can contribute (Ip 1996) Furtheralthough G values are typically independent ofthe incident angle the loss of O2 is a surfaceprocess and is enhanced with angle That is thenumber of O2 produced within the escape depthincreases with increasing angle to the normal(steeper than the inverse cosine) This enhance-ment is 25ndash4 (Jurac et al 2001) Ignoring suchcorrections the total production rate calculated

(23ndash62 3 109 O2cm2s) is a lower bound to theactual production rate from the icy regions of Eu-roparsquos surface Recent modeling of the O2 atmo-sphere suggest that such a source rate is required(Shematovich and Johnson 2001)

Earlier estimates of the O2 source rate were in-correct because O2 production was assumed to bea fixed fraction of the H2O sputtering yield Atthe ambient temperature an ejected H2O sticks ef-ficiently (Smith and Kay 1997) to the surface ofneighboring grains in the porous regolith reduc-ing the sputtering yield for H2O (Johnson 1989)This reduction does not apply to the O2 fractionof the yield enhancing its relative importance toH2O Here we assume that all of the gas-phaseO2 produced can contribute to the ambient at-mosphere even if it is produced at a grain sur-face that is at some depth into the porous regolith

Finally Europarsquos surface is not pure ice every-where That is much of the trailing hemisphereis a hydrated sulfuric acid containing smallamounts of sulfur and SO2 and the leading hemi-sphere contains low levels of carbon CO2 andcarbonates Sodium and potassium atoms aresputtered from the icy regions (eg Leblanc et al2002) O2 production from surfaces containingthese species has not been measured C and S areboth proton scavengers and the other species canbe sources of O as discussed so that the O2 yieldfrom such regions could be enhanced Howeverthere can also be competing processes that mightreduce these rates In fact recent HST observa-tions suggest that the O2 might not be produceduniformly across Europarsquos surface (McGrath et al2000) The analysis of these observations is pre-liminary and is not simple since the observed

JOHNSON ET AL840

TABLE 3 APPLICATION TO EUROPA

OxygenImplantation Energy flux production

Species 109(cm2 s)a 109 keV(cm2 s)(1) 109(cm2 s)

H (10 keV) 0015(1) 12 0084 (072)bO (10 keV) 0015(1) 18 054cS (10 keV) 0009(1) 30 090cElectrons 62 043 (37)bPlasma (10 keV) (137 amu) 13(2) 10 03c

Total 79 23 (62)UV photons (64 eV) 46 01

aReferences are as follows 1Cooper et al (2001) [similar results in Paranicas et al (2002)] 2Mauk et al (1996)bUsing G for 200 keV protons (upper bounds in brackets are based on the low-energy electrons data as a model for

the secondary electrons produced)cUsing G for 15 MeV Ne1 at 7K a rough lower bound

emissions depend on the local plasma propertiesHowever the efficiency of O2 production in con-taminated ices needs to be studied

Porous regolith O2 interactions on grain surfaces

Although the O2 atmosphere is principally pro-duced in the near surface layers and is eventuallylost to space it also permeates Europarsquos porousregolith The porosity p (void space) of the re-golith is uncertain (Buratti 1995) It has been es-timated to be the most compact surface in theOuter Solar System p 025 on the trailing hemi-sphere and p 079 on the leading hemisphere(Buratti and Veverka 1983) and the most poroussurface p 096 globally (Domingue et al 1991Domingue and Hapke 1992) These large differ-ences are due to the uncertainties in the amountof backscattered light at small phase angles andon the photon absorption properties and grainsizes The more compact surface is likely correctGrain sizes rg of the order of 100 mm have beensuggested (Hansen and McCord 1999 2000Geissler et al 1998) Europarsquos regolith differs sig-nificantly from the lunar regolith which is asand-like surface of individual grains Analysis ofthe thermal inertia of Europarsquos surface (Grundyet al 2001) suggests that the ice grains are sin-tered forming a fairy castle-like structure Sin-tering can occur thermally or due to the incidentradiation In addition ices doped with acids andalkalis in the laboratory can be powdery solidswith the impurities primarily found at grain in-terfaces

The porous regolith affects the above estimatesof the oxygen production in three ways First forisotropic bombardment the mean distance d intothe surface at which an incident ion or photonstrikes a grain surface is d (89)rg(1 2 p) Us-ing the above value for rg then d 02 m if theporosity is very large p 96 Therefore de-sorbed molecules can come from a considerabledepth and will interact with grain surfaces Evenfor much lower porosities the effective sputteringyields can be significantly modified (Johnson1989) As discussed above the yields for ejectedH2O which sticks onto neighboring grains at theambient temperatures are reduced but those forejected O2 are not This enhances the relative im-portance of the O2 yields

Finally the surfaces and interfaces of the sin-tered grains in Europarsquos porous surface are placeswhere low temperature chemistry involving the

ambient oxygen can occur at depths that are wellbelow the nominal penetration depth of the inci-dent radiation The relative importance of suchchemistry is roughly indicated by the ratio of thetotal surface area in the regolith to the area of Eu-roparsquos visible surface Using the porosity andgrain size above and a regolith depth dr of theorder of a meter then this ratio is large 3(1 2p)drrg which is 103ndash2 3 104 for average porosi-ties of 96 and 25 respectively Therefore thegas-phase O2 molecules below the visible surfaceinteract frequently with the grain surfaces In thisway a very large volume of the subsurface couldbe altered chemically even if it is not directly ex-posed to the radiation

From the overlying structures seen in the high-resolution images of Europarsquos surface by theGalileo spacecraft (eg Prockter and Pappalardo2000) it is clear that the youngest surfaces are red-der The reddening in the visible has been attrib-uted to the presence of sulfur in the ice (McEwen1986 Johnson et al 1988 Calvin et al 1995 Carl-son et al 1999b) The amount of reddening in thespectra of the surface material decreases with ageand the surface materials appear to becomebrighter with age This might occur by the slowaccumulation of a frost over-layer or by radioly-sis (Carlson et al 2002 Johnson et al 2003) Ra-diolysis preferentially converts the sulfur in theice to sulfur dioxide or sulfuric acid both ofwhich have flat spectra in the visible In regionswhere the radiation does not directly penetratethe O2 permeating the regolith can oxidize thesulfur at the surfaces of the icy grains and at graininterfaces Similarly the carbon seen in the ice onEuroparsquos leading hemisphere (eg Carlson et al2004 Johnson et al 2003) can be oxidized by ra-diolysis in the near surface layers or by reactionson the grain surfaces at depth in the regolith bythe ambient O2 The meteoroid bombardmentthat produces the regolith also exposes fresh sur-faces for oxidation and can cause the collapse ofpores leading to the trapping of oxygen In thismanner both refractory and volatile oxidants canbe formed and trapped in the subsurface ice forpotential delivery to Europarsquos subsurface ocean

SUMMARY

The presence of a thin oxygen atmosphere andoxidants trapped in Europarsquos surface ice (egJohnson et al 2003) combined with the possible

OXIDANTS AT EUROPA 841

presence of a subsurface ocean (eg Carr et al1998) have made Europa an exciting target forfuture space exploration The molecular oxidantsproduced by radiolysis and photolysis could bespecies needed for carbon-based biochemistry inEuroparsquos putative subsurface ocean Howeverthere has been no model for the formation trap-ping and transport of relevant oxidants

In this paper we have reviewed the space ob-servations and laboratory studies related to thechemistry of formation of oxidants in ice by theincident radiation Because the laboratory dataare incomplete we examined the rate equationsfor formation of oxygen and related species byradiolysis and photolysis We showed how suchequations can be simplified to the analytic modelthat was used for fitting the fluence and temper-ature dependence of the O2 yield produced bylow-energy electrons (Sieger et al 1998) In de-riving the analytic model from the more completerate equations the competition between forma-tion and destruction processes can be studied ascan the role of trapping sites escape from depthand temperature history These equations also al-lowed us to put incident photon electron and iondata on a similar footing for the first time

The model can be used to suggest measure-ments that are needed to describe the chemicalstate of the irradiated ice and to interpret new lab-oratory data We showed that if reactions are fastthe yield data could be consistent with a precur-sor being formed by two events as is the case forperoxide However we favor trapped O as a pre-cursor as initially suggested by Matich et al(1993) for production of excited O2 Such aspecies if present could be detected spectro-scopically as can trapped O2 through bands ac-tivated by lattice perturbations (Cooper et al2003a)

The available oxygen yield data were used tocompare G values for different incident particlesThese were then used to estimate the oxygen pro-duction in the radiation environment at EuropaThe lower bounds were consistent with the pro-duction rates obtained from a recent model of Eu-roparsquos atmosphere The porosity of the regolithwas also considered It not only affects the rela-tive sputtering yields for O2 and H2O but alsoprovides an enormously increased surface areaon which the ambient gas-phase oxygen can re-act with species in the ice That this might be oc-curring is consistent with the change in re-

flectance associated with surface aging Such re-actions could also convert ambient molecularoxygen into other more refractory species that re-main trapped on subduction of Europarsquos crust

Although radiolysis produces a stable chemi-cally altered surface in the laboratory the parti-cle flux to Europarsquos surface is many orders ofmagnitude lower Therefore annealing can be ef-fective at temperatures that are much lower thanthose typically found experimentally Belowabout 120K the annealing times appear to be long(eg Baragiola 2003) so that laboratory data canoften be directly applied to Europarsquos icy surfaceHowever account has to be taken of the syner-gism between the UV and the energetic plasmaas discussed earlier (Johnson and Quickenden1997)

Transient surface melting might initiate lifeprocesses but subduction of oxidants to the pu-tative ocean is likely to be required if biochem-istry is to occur at Europa (Chyba and Hand2001) Although the temperature versus depthinto Europarsquos regolith is not well known belowthe regolith the temperature and pressure appar-ently increase with depth At some depth (ap-proximately kilometers) either very warm ice oran ocean exists (Ruiz and Tejero 2000) Thereforechemically refractory oxidants are likely to be re-quired in order to survive subduction to thewarmer subsurface regions Trapped species ifdelivered downward can become mobile andwill segregate forming inclusions (Johnson andJesser 1997) Indeed peroxide appears to segre-gate in ice (Gurman et al 1967) and these inclu-sions may remain relatively stable until the per-oxide is dissolved in the putative ocean Thevolatiles seen in reflectance such as O2 SO2 andCO2 will also segregate But at the higher tem-peratures they could also diffuse through thebulk react or percolate into the regolith atmo-sphere Therefore the observation of the trappedvolatiles in Europarsquos surface ice is of interest pri-marily because it suggests that oxidants are pre-sent The formation of oxygen-rich molecules fol-lows the preferential loss of hydrogen Howeverthe surface is in a local steady state with im-planted H O and S from the plasma and the lossto space Since the plasma flux to the surface canbe variable this balance can also vary Indeedtemporal changes in reflectance have been re-ported (Domingue and Hendrix 2004)

Based on the models for formation of O2 ex-

JOHNSON ET AL842

amined in this paper additional experiments areneeded to determine the chemical state of irra-diated ice at low fluences Of immediate inter-est is the identification of stable molecular pre-cursors required for the formation of O2 Basedon the available laboratory data individual per-oxide molecules trapped in the ice do not appearto be the principal precursors but trapped O(Matich et al 1993) remains a viable candidateAlthough the data suggest that the local hydro-gen-bonding network is important in the pro-duction of O2 the effect of the formation tem-perature and radiation damage on the structureneed to be measured It is also important to de-scribe quantitatively the relationship betweenH2 loss and the formation of O2 and H2O2 overa range of temperatures and fluences Finallyone would like to determine the steady-statedensities of biochemically interesting moleculesformed by radiolysis in an ice containing sulfurand carbon at Europarsquos surface temperaturesand then determine their stability with increas-ing temperature and pressure Such experimentsstill need to be carried out Here we have pre-sented a model that brings together the variouslaboratory data and now can be used to considerthe formation trapping and transport of relevantoxidants at Europa

APPENDIX

Introduction

In this Appendix we examine rate equationsthat describe various aspects of the radiolysis orphotolysis of ice These equations need to be un-derstood if one is to be able to calculate the for-mation trapping and transport of oxidants Theemphasis is on the formation of molecular oxy-gen We integrate these equations over the escapedepth of O2 to give the simplified rate equationsdiscussed in the text in which number densitiesare replaced by column densities (eg Eqs 1 and2) In this way we can examine the relationshipbetween the rate constants and the effective pre-cursor formation and destruction cross sectionsextracted from data We first consider a precur-sor such as H2O2 that is formed from two exci-tation events This bears on the observed linear-ity of the oxygen yield versus fluence We thenconsider the equations for trapped versus a tran-

siently mobile species We use a reaction betweena mobile and trapped O as an example althoughO 1 OH can also give oxygen Finally we ex-amine the possible dependence of the O2 forma-tion process on the density of traps (defectsvoids and grain boundaries) as a function of tem-perature The results are discussed in the text

For clarity all possible radiolytic species arenot included in the discussion below Thereforepotentially important products such as HO2 andthe charged species (eg H3O1 the trapped elec-tron etc) are missing In each section we attemptto make a point related to production of O2 andother oxidants by using a subset of the productsrather than attempt a complete description withmany unknown rate constants

Fast versus slow processes

To describe the O2 production rate from a pre-cursor such as H2O2 we first use a simple but in-structive set of rate equations to describe the for-mation of H2O2 We then relate these to thesimplified rate equation for the precursor in thetext Eq 1 and show how the postulated precur-sor formation cross section sp is related to theH2O dissociation cross section sd by the reactionrates

Using the reactions

H2O 1 radiation R H 1 OH (sd)

H 1 H R H2 (k2H)

H 1 OH R H2O (kHOH)

OH 1 OH R H2O2 (k2OH)

the reaction pathways are described by

H2

Hqk2H

1 OHH2O 1 radiation mdashRsd H

OHH2O2

OH Rk2OH

kHOH

HH2O

OXIDANTS AT EUROPA 843

The corresponding rate equations in an irradiatedvolume are

dnOHdt 5 sd f n 2 2 k2OH nOH nOH

2 kHOH nH nOH (A1a)

dnHdt 5 sd f n 2 2 k2H nH nH

2 kHOH nH nOH (A2a)

dnH2O2dt 5 k2OH nOH nOH

2 destruction processes (A3a)

Here sd is the dissociation cross section f is theradiation flux and n is the molecular numberdensity of ice with nOH and nH the density of dis-sociation products OH and H respectively Therate constants k2OH k2H and kHOH describe thereactions above In this model the H2 formed isassumed to escape for temperatures above about20K We do not distinguish between trappedspecies and mobile species although in fact thatdistinction is important as discussed in the nextsection That is any OH produced at low tem-perature by one ion would likely be trapped priorto interacting with an OH produced by a subse-quent ion At higher temperatures this can breakdown as discussed in the text We first relate theabove set of equations to Eq 1 in the text and thenconsider precursor destruction processes

Equation 1 in the text is obtained by solvingthese equations together and then by integratingover a depth Dx Here Dx is the smaller of the pen-etration depth of the particle and the O2 escape(percolation) depth Assuming the molecular den-sity of ice n is constant with depth Dx n 5 N inEq 1 Comparing Eq 1 with Eq A3a the precur-sor formation rate sp f N is equivalent to [k2OH

nOH nOH] Dx Assuming that the processes in EqsA1a and A2a are fast then the steady state is read-ily obtained In addition if k2H k2OH there ispreferential loss of H2 at short times as suggestedby the laboratory data If steady state is reachedthe production of H2 is equal to that of H2O2 us-ing only the equations above Further one obtains

[k2OH nOH nOH] Dx R [fp sd] f N (A4)

with

fp 5 05[1 1 05 kHOH(k2H k2OH)05] (A5a)

Therefore the precursor formation cross sectionsp in Eq 1 is a fraction fp of the total dissocia-

tion cross sd Since laboratory data suggest theH2O2 fraction in an irradiated ice is small thenkHOH (k2OH k2H)05 which is likely givingfp R [(k2OH k2H)05kHOH]

Trapped versus mobile species

To examine the fluence dependence discussedin the text we note that trapped OH and freshlyproduced transiently mobile OH can be treatedas different species (eg Matich et al 1993)Therefore assuming no trapping of H the OHrate equation above would be replaced by twoequations one for the trapped species nOHt andthe other for the freshly produced mobile speciesnOH Assuming the reaction of two mobile OHunits is unlikely and that peroxide is formed fromthe interaction of a trapped OH with a subse-quently produced OH an expanded set of equa-tions is obtained

OH 1 trap R OHt (kOHt)

OH 1 OHt R H2O2 (kOHOHt)

H 1 OHt R H2O (kHOHt)

then the reaction pathways are described by

qH2

H2O 1 radiation mdashRsdH

qk2H

1 OHH

trap OH

OH RkOHt

OHt kOHOHtH2O2

kHOHkHOHt

H H2O HH2O

The resulting set of equations is

dnOHdt 5 sd f n 2 kHOH nH nOH

2 kOHt nOH nt 2 kOHOHt nOH nOHt (A1b)

dnOHtdt 5 kOHt nOH nt 2 kOHOHt nOH nOHt

2 kHOHt nH nOHt (A1b9)

dnHdt 5 sd f n 2 2 k2H nH nH 2

kHOH nH nOH 2 kHOHt nH nOHt (A2b)

JOHNSON ET AL844

dnH2O2dt 5 kOHOHt nOH nOHt

2 destruction processes (A3b)

where nt is the trap density in the irradiated iceAt low T we can assume nt is large so thatdntdt 0 simplifying the set of equations If theprecursor of interest is trapped O as discussed inthe text then a similar set of equations can be con-structed for O and Ot as in the example to followOf course a full description over a broad rangeof temperatures would require the considerationof the both trapped and mobile H O and OHthe rate of change of trap density and the vari-ous destruction processes

The above equations should be solved simul-taneously but for the incident particle fluxes typ-ically used in these studies the diffusion of mo-bile species to traps is relatively fast This is trueeven at low temperatures if there is a significantdensity of trapping sites This is usually the casefor samples formed at low T Therefore in a lowtemperature solid at low dose rates there is not asignificant steady-state background of diffusingOH There are trapped or freshly produced OHunits which rapidly react Averaging over thebombardment rate this can be approximated bywriting dnOHdt 0 and then separating nOH outin Eq A1b Assuming also that H diffuses rapidlyand either reacts or forms H2 we also writednHdt 0 on the average Therefore betweenimpacts only species trapped at defects pores orgrain surfaces are assumed to be present as sug-gested by the lack of a dependence on the inci-dent beam flux In this limit the rate of change oftrapped OH roughly simplifies to

dnOHtdt 5 2 k2H nH nH 2 2 kOHOHt nOH nOHt

It is seen that the production of peroxide dependson the loss of H by formation of H2 as discussedand at steady state the production rate for per-oxide equals the rate of formation of H2 in theseequations

Treating the mobile and trapped species sepa-rately we can further simplify the rate equationsand obtain the precursor model used in the textThat is integrating over the relevant depth Dxthe densities are replaced by column densitiesThe rate constants ki are then related to effectivecross sections si In this way cross sections aregiven as an interaction length for each freshlyproduced mobile radical [ki nj] R [si f] The

trapped OH column density NOHt and peroxideproduction rate can be estimated from ldquolinearrdquorate equations

dNOHtdt lt [sd f] N 2 2 [s2OH f] NOHt

2 [sHOH f] NOHt (A6a)

dNH2O2dt lt [s2OH f] NOHt 2 s NH2O2 (A6b)

Here sd is again the dissociation cross sectionand the destruction processes in Eq A3b are con-tained in s The second and third terms on theright in Eq A6a are the loss of OH due to the pro-duction of H2O2 and H2O In the integration ofEq A1b9 [2 kOH OHt nOH] becomes [2 s2OH f] and[kHOHt nOH] becomes [sHOH f] Therefore thedensity and the interaction length of mobile H arecontained in sHOH If H2O2 is indeed the pre-cursor then in Eq 2 in the text [sp f N] is be-comes [s2OH f NOHt] from Eq A6b Accountingfor precursor destruction which can give backtwo OH and treating H2 can both be included inthe integrated model via the si values

The simplified set of equations above allows usto examine the relative time (fluence) scales forprecursor formation (in this case H2O2) SolvingEqs A6a and A6b the column density of OH andthe concentration of precursors versus fluence cpcan be obtained Using the column of H2O2 pro-duced as the precursor column Np in Eq 2 thenO2 yield can be obtained The results can be com-pared with the results from Eqs1 and 2 in thetext The average precursor (H2O2) concentrationover the column depth of water molecules N is

cp 5 fp (sds)[1 2 exp(2s F)] 2 (ssr)

[1 2 exp(2sr F)] (A7)

with

fp 5 2s2OH(sr 2 s) (A5b)

In these equations sr 5 (2s2OH 1 sHOH) andssr 1

It is seen that the precursor concentration withtime cp depends on two rates the reaction ratesthat determine sr and the destruction rates in sAs in Eq 4a the yield can be written as

Y 5 sO2 cp N (A8)

If sr and s are comparable then the yield at small

OXIDANTS AT EUROPA 845

fluence is quadratic in F However if sr sthen the linear dependence on fluence is obtainedwith a small shift in the fluence as shown in Eq4c in the text Further the steady-state yield isnow

Y` (T) 5 sO2 [fp sd] Ns (A9)

That is sp R [fp sd] as in Eq A4 and the tem-perature dependence for the incident electrondata must be contained in [sO2 fp] or in the ratio[sO2 s2OH]sHOH]

For comparison with their precursor modelSieger et al (1998) examined the possibility of ob-taining the O2 precursor by considering reactionsfor a density of diffusing dissociation products asin Eqs A1 and A2 rather than a trapped productinteracting with a freshly formed dissociationproduct Based on the above the lowest fluence atwhich O2 is first detected experimentally sets alimit on the time scale for mobilendashmobile reactions

Effect of trap density

Equations A1b A1b9 A2b and A3b can be ap-proximately solved assuming that the dnOHdtdnHdt and dntdt are very small on the averageas discussed above In order to examine the roleof the trap density nt which can depend on thetemperature and method of formation and on theradiation history of the ice we use the slightlysimpler precursor model In this model the O pro-duced directly by dissociation traps forming aprecursor O2 is then formed when a freshly pro-duced O or other species reacts with a trappedoxygen atom Ot

H2O 1 radiation R H2 1 O (sd)

O 1 trap R Ot (kOt)

O 1 Ot R O2 (kOOt)

The reaction pathways for the oxygen species are

H2O 1 radiation mdashRsd9

trap OO2

kOtOt kOOt

H2 1 OOHt

kOR H(kHO t)

ROR

We assume for simplicity both H2 and O2 escapefrom depth Dx after formation We also allow mo-bile H to react with Ot (kHOt) removing trappedO and both the mobile H and O to be removedby other reactants R (eg OH or contaminants)This leads to a simplified set of equations not di-rectly involving OH or peroxide

dnOdt 5 sd9 f n 2 kOt nO nt 2 kOR nO nR

2 kOOt nO nOt (A1c)

dnOtdt 5 kOt nt nO 2 kOOt nO nOt

2 kHOt nH nOt (A1c9)

dnHdt 5 sd f n 2 kHOt nH nOt

2 kHR nH nR (A2c)

dnO2dt 5 kOOt nO nOt (A3c)

These equations of course act in parallel with theequations above for OH and mobile OH may re-act with trapped O to form O2 Here we focusonly on mobile O

Since the diffusion of H and O are fast and thedensity of traps is large the nH and nt averagedover the bombardment rate can be taken veryroughly as time independent Further if we as-sume the number of trapped O nOt is small com-pared with the number of sites at which H canreact then nH [sd f nkHR nR] For a low ra-diation flux f nH is small on the average as as-sumed Since the O2 yield YO2 is equal to[(dnO2dt) Dxf] with Dx the escape depth thenthe steady-state yield Y` can be obtained ana-lytically

[Y`][sd9 N] 5 1 1 (q 2 1) 2 [(q 2 1)2

1 8 q d]054 d (A10)

q 5 [kOt (nt 1 nR)] (kHOt nH)(sd9 f n) kOOt

5(sdsd9) [kOt (nt 1 nR)] kHOt[kOOt kHR nR]

d 5 05 [1 1 nt(nt 1 nR)]

We note that for nt large q can be large Assum-ing that q 1 and nt nR then Eq A10 be-comes

Y` R [sd9 N]q 5 (sd9 N) (sd9sd)

[kOOt kHR nR][kHOt kOt nt] (A5c)

JOHNSON ET AL846

Comparing with Y` in the simple precursormodel in Eq 4b then s R [(kHOt nH)f] R [sd nkHOt][kHR nR] In addition the product [sO2sp] R sd9 sd9 [kOOt n][kOt nt] That is the dis-sociation cross sections are modified by reactionrates

Quite remarkably it is seen in Eq A5c that thesteady-state yield Y` can depend inversely onthe density of traps nt in the temperature andflux regimes for which the approximations arerelevant That is if there are competing destruc-tion process for trapped O then at high trap den-sity a mobile O must find a trapped O before itbecomes trapped or before mobile H destroys theOt In this way the increasing O2 yield with in-creasing T can be related to the change in the trap-ping density nt with the formation temperatureof the ice sample As the temperature increasesand the density of traps in the bulk decreases thesurface and grain interfaces become the only trap-ping sites

ACKNOWLEDGMENTS

REJ acknowledges support by a Gledden Fel-lowship in Chemistry School of Biomedical andChemical Sciences at The University of WesternAustralia and by grants from the NSF Astron-omy Program and NASArsquos Planetary Geologyand Geophysics Program PDC acknowledgessupport by an Australian Postgraduate Award

ABBREVIATIONS

HST Hubble Space Telescope MJQ MatichndashJohnsonndashQuickenden NIMS Near Infrared Map-ping Spectrometer UV ultraviolet

REFERENCES

Ayotte P Smith RS Stevenson KP Dohnarsquolek ZKimmel GA and Kay BD (2001) Effect of porosityon the adsorption desorption trapping and release ofvolatile gases by amorphous solid water J Geophys Res106(E12) 33387ndash33392

Bahr DA Famaacute M and Baragiola RA (2001) Radiol-ysis of water ice in the outer solar system sputteringand trapping of radiation products J Geophys Res 10633285ndash33290

Baragiola RA (2003) Microporous amorphous water icefilms and astronomical implications In Water in Con-

fining Geometries edited by JP Devlin and B Buch El-sevier New York in press

Baragiola RA and Bahr DA (1998) Laboratory studiesof the optical properties and stability of oxygen onGanymede J Geophys Res 103 25865ndash25872

Baragiola RA Atteberry CL and Bahr DA (1999) Ori-gin of solid oxygen on Ganymede Reply to commentby RE Johnson on ldquoLaboratory studies of the opticalproperties and stability of oxygen on Ganymederdquo JGeophys Res E104 14183ndash14187

Baragiola RA Vidal RA Svendsen W Schou J ShiM Bahr DA and Atteberrry CL (2003) Sputteringof water ice Nucl Instrum Methods Phys Res B 209294ndash303

Bar-Nun A Herman G Laufer D and Rappaport ML(1985) Trapping and release of gases by water ice andimplications for icy bodies Icarus 63 317ndash332

Baverstock KF and Burns WG (1976) Primary pro-duction of oxygen from irradiated water as an expla-nation for decreased radiobiological oxygen enhance-ment at high LET Nature 260 316ndash318

Benit J and Brown WL (1990) Electronic sputtering ofoxygen and water molecules from thin films of waterice bombarded by MeV Ne1 ions Nucl Instrum Meth-ods Phys Rev B 46 448ndash451

Benit J Bibring J-P della-Negra S LeBeyec YMendenhall M Rocard F and Standing K (1987)Erosion of ices by ion irradiation Nucl Instrum Meth-ods Phys Rev B 1920 838ndash842

Brown WL Augustyniak WM Lanzerotti LJ John-son RE and Evatt R (1980) Linear and nonlinearprocesses in the erosion of H2O ice by fast light ionsPhys Rev Lett 43 1632ndash1635

Brown WL Augustyniak W M Simmons E Mar-cantonio KJ Lanzerotti LJ Johnson RE BoringJW Reimann CT Foti G and Pirronello V (1982)Erosion and molecular formation in condensed gasfilms by electronic energy loss of fast ions Nucl In-strum Methods B 1 307ndash314

Brown WL Augustyniak WM Marcantonio KJ Sim-mons EN Boring JW Johnson RE and ReimannCT (1984) Electronic sputtering of low-temperaturemolecular solids Nucl Instrum Methods B 1 307

Buratti B (1995) Photometry and surface structure of theicy Galilean satellites J Geophys Res 100(E9) 19061ndash19066

Buratti B and Veverka J (1983) Voyager photometry ofEuropa Icarus 55 93ndash110

Calvin WM and Spencer JR (1997) Latitudinal distri-bution of O2 on Ganymede observations with the Hub-ble Space Telescope Icarus 130 505ndash516

Calvin WM Clark RN Brown RH and Spencer JA(1995) Spectra of the icy Galilean satellites from 02 to5 mm a compilation new observations and a recentsummary J Geophys Res 100 19041ndash19048

Calvin WM Johnson RE and Spencer JA (1996) O2

on Ganymede spectral characteristics and plasma for-mation mechanisms Geophys Res Lett 23 673ndash676

Calvin WM Aninich VG and Brown RH (2004) Oxy-gen temperature and phase effects Icarus (in press)

OXIDANTS AT EUROPA 847

sidered Assuming the mobile O processes arefast these rate equations can also be integratedand simplified to give the form in Eqs 1 and 2The precursor column density Np now repre-sents the column density of Ot In this model sO2in Eq 2 is the ldquocross sectionrdquo for reaction of afreshly produced O with an Ot and sp9 indicatesdestruction of Ot Finally sp is the cross sectionfor production of H2 and Ot via processes dis-cussed above

Based on the sizes of the cross sections ex-tracted in the electron irradiation experimentsthe model with trapped O (Ot) as the precursoris quantitatively reasonable That is for a rangeof T both sp and sO2 might be described byevents of the type H2O R H2 1 O (whether di-rect or indirect) If these occur for 10 of theradiation-induced dissociation events as dis-cussed then the required product of cross sec-tions suggested by the sizes of s and Y` above isroughly obtained that is [sp sO2] 02ndash6 3 10234

cm2 in the low-energy electron experiments Thecross section for production of O might dependon the availability of dangling bonds at interfacesin pores or at defects as suggested by a numberof authors The importance of such bonds couldalso be consistent with the differences in the elec-tron-induced yield for initially crystalline and ini-tially amorphous ice (Orlando and Sieger 2003)

O2 production can increase in the presence ofoxygen-rich impurities such as CO2 or SO2 bothof which are known to be present at Europa Asin the MJQ model dissociation (eg CO2 R CO1 O) can provide a mobile O that can be trappedor react with a previously trapped O forming O2In addition such species are hydrogen scav-engers (Moore and Hudson 2000) This can re-duce the destruction of Ot affecting sp as seen inEq A5c

Trap density

In the simplified models above the density oftrapping sites nt was not explicitly accountedfor However in the discussion of both the ex-perimental data and models trapping sites havebeen assumed to be important It has beendemonstrated repeatedly that the density of de-fects voids and interfaces decreases as the tem-perature of formation of the ice increases How-ever the efficiency of production of O2 is theopposite it increases with increasing T There-fore the oxygen yield varies in a manner that is

the inverse of the trap density over a range oftemperatures For penetrating ions in which theenergy density deposited is high and O2 might beformed in a single track a simple explanation hasbeen given As the density of traps for O2 is re-duced and the temperature increases the O2 pro-duced can diffuse to the surface and escape moreefficiently limiting back reactions on further ir-radiation This is examined in the next sectionHowever such a process does not affect the low-energy electron yields for which the penetrationdepth is small and the formed O2 escapes read-ily at the measurement temperatures In Eqs A1cA1c9 A3c and A10 in the Appendix we againuse a simple set of rate equations for trapped andmobile O in order to determine the role of the trapdensity nt on the production of a precursorspecies The conclusions apply to any modelbased on a trapped precursor

In the MQJ model mobile O can be trappedforming Ot (described by kOt nt) and can reactwith previously trapped O to form O2 (kOOt) Inaddition it can react at other sites and Ot can bedestroyed by mobile reactive species such as H(kHOt nH) Mobile H can also react with othertrapped species (kHR nR) Incorporating these re-actions Eq A10 in the Appendix gives a steady-state yield Y` When the density of traps is largethe steady-state yield is given in Eq A5c

Y` R (sd9sd9sd) [kOOtkHR nr][kHOt kOt nt] N (5)

Here sd and sd9 are the dissociation cross sectionsthat determine the production of H (H 1 OH) andO (H2 1 O) by the incident radiation Comparingthe result in Eq 5 with Y` in the simple precur-sor model in Eq 4b the dissociation cross sec-tions in Eq 5 are now scaled by a ratio of the re-action rates

It is seen in Eq 5 that the steady-state yield Y`depends inversely on the density of traps nt Thatis there are competing destruction processes fortrapped O (Ot) Therefore in order to form O2 amobile O competes with mobile H to find atrapped O before it becomes trapped In this waythe increasing O2 yield correlates with the re-duction in the trapping sites with increasing tem-perature of formation The small activation ener-gies discussed earlier would be related to there-orientation of the lattice binding leading to thereduction in trap density At relatively high T thedensity of traps becomes small so that only the

OXIDANTS AT EUROPA 837

sample surface or grain interfaces might act astraps accounting for the observed drop in theyield at high T For this reason it is important todetermine if the yield for T greater than 150Kdepends on the incident flux f Since other quan-tities in the expression for Y` could also be tem-perature dependent experiments are neededThese would likely be pulsed radiolysis experi-ments in which the transients are studied ratherthan the steady-state yields However the avail-able data for lightly ionizing radiation do suggestthat the O2 yield depends inversely on trap den-sity as in the model above

Escape from depth

The quantity N in Eq 1 is the ldquodepthrdquo given asa molecular column density from which O2 formedbelow the surface can percolate to the surface andescape It can be converted to a physical depth bydividing by the molecular number density of theice In the case of the low-energy electrons thedepth below the surface at which ionizations andexcitations are produced is small However forfast ions that penetrate many monolayers this isnot the case Below we examine N in Eq 1

Although it has long been understood that twoexcitations events are required to produce O2(eg Brown et al 1982 Reimann et al 1984) theprecursor model suggests that at saturation(steady state) any new event can produce O2 fromthe steady-state density of trapped precursorsTherefore one might expect the yield at satura-tion to be nearly linear in the excitation cross sec-tion This is the case for the low-energy electronsand it was initially assumed to be the case for thefast ions (Brown et al 1980 Reimann et al 1984)However Baragiola et al (2003) showed that thetotal sputtering yield in the saturation region attemperatures greater than 100K which is dom-inated by decomposition (Brown et al 1982) isproportional to the square of the electronic en-ergy deposition per unit path length in the solid(dEdx)e Assuming that the production and de-struction cross sections in Y` in Eq 4b roughlyscale with (dEdx)e then their ratio [sp sO2s]also roughly varies as (dEdx)e If this is the casethen the observed quadratic dependence impliesthat the depth N from which O2 can be mobi-lized to escape must also depend on the excita-tion density (dEdx)e This is consistent with theobservation that the O2 yield depends on thick-ness for samples thinner than the ion penetration

depth (Reimann et al 1984 Benit and Brown1990) Therefore the percolation depth for escapeN is a critical factor in comparing yields for pen-etrating radiation with those for non-penetratingradiation like the low-energy electron and low-energy ion data It is also seen that percolationand trapping can be included within the simplemodel considered here

Model summary

We have examined the simple precursor modelintroduced by Sieger et al (1998) to explain theirelectron data We have shown how it can be gen-eralized and derived from an integration of theappropriate chemical rate equations with thecross sections and total column density contain-ing the reaction rates and the physics of diffusivetransport Therefore the parameters can havevery different meanings from those initially as-sumed The equations are useful when there arefast processes which go to completion in timesmuch shorter then the time between particle im-pacts and slow processes associated with speciesthat are assumed to be trapped between particleimpacts This separation allowed us to examinethose models proposed for precursor formationand destruction due to reacting mobile speciesformed from dissociation events Further fromthe size of the onset of the linearity of the yieldversus fluence we can place constraints on therole of nonlinear processes such as those thatmight be associated with the formation of a pre-cursor like peroxide

With these extensions this simple model is use-ful for approximating the full rate equations inthe temperature region for which the yields de-pend on the fluence (dose) and not on the doserate Based on the assumptions in the model thisimplies temperatures at which precursor speciesare stable over experimental times At the high-est temperatures for the yields shown in Fig 4athis might not be the case This should be testedexperimentally In this regard it is puzzling thatexperiments have shown that the absorptionbands associated with certain trapped speciessuch as OH and trapped electrons ldquodisappearrdquoat 100ndash120K (see eg Johnson and Quickenden1997) whereas the O2 yield data indicate that rel-evant precursor species must be stable to 140KTherefore the proposed O2 production processescan only be distinguished by directly determin-ing the chemical state of the irradiated ice

JOHNSON ET AL838

It may be the case that a particular precursoris not required as mentioned earlier Rather af-ter a sufficient level of oxidation a newly pro-duced O or OH can react in the oxygen-rich lat-tice producing O2 in a pseudo-first-order processThis can be tested experimentally Indeedtrapped O and peroxide may be changed into oneanother as in the matrix isolation experiments(Khriachtchev et al 2000) Therefore their rela-tive concentrations versus temperature and doseshould be studied

In the model considered here it is seen that theparameters in Y` must account for the strong tem-perature dependence Since the number of bulktrapping sites decreases with increasing T as theice sample is annealed one might expect that theprecursor concentration would decrease We sug-gest that the observed temperature dependenceis determined by the competition between theavailability of trapping sites and mean interactiondistances of a mobile species such as O and HThat is the O2 yield varies inversely with the den-sity of traps over a range of temperatures In thesimple model above this effect is included indi-rectly as shown using the more detailed equa-tions in the Appendix

EUROPA

Oxygen and peroxide production at Europa G values

The dependence of the yield on the thicknessof the ice sample for highly penetrating ions sug-gests that the O2 yield is roughly related to theamount of energy deposited in the sample There-fore the amount of O2 produced and lost fromice in steady state is often given as a G value thenumber of molecules produced per 100 eV de-posited in Table 2 In applying these G values tothe planetary environment care must be takensince the yields are temperature fluence and ex-citation density dependent They also are affectedby the presence of other proton scavengersknown to be present in Europarsquos ice

The G values for peroxide in Table 2 are thosefor production at low fluence but the G valuesfor O2 are for the production and escape of gas-phase oxygen at saturation Therefore they de-pend on the formation and destruction cross sec-tions in the model above Such G values can beused to describe the steady-state source of the am-

bient O2 atmosphere at Europa but are not help-ful in determining the ability to produce and trapO2 at depth into the icy surface Over the timescales of the experiments some O2 is seen to re-main trapped and is not destroyed by radicalsTypically there are a number of trapping sitesand recent estimates of the activation energy foran O2 molecule trapped in a substitutional site(Hori and Hondoh 2002) suggest that it can betrapped for times comparable to the time re-quired for the conversion of amorphous ice intoIh (Ayotte et al 2001) The stability of O2 inclu-sions has not been studied

To understand the relationship between the Gvalues for O2 in Table 2 we note that fast lightions and electrons lose their energy predomi-nantly by producing ionizations and excitationsThe ionization events produce secondary elec-trons that can result in further ionizations Thenumber of ionization events Ni produced by afast light ion stopping in a water ice is Ni 5EoW where Eo is the initial energy and W is theaverage energy expended per ionization event27 eV in water ice Therefore radiolysis by fastions is often described as the net effect of theshower of electrons set in motion by the incidention In this manner the G values for productionof O2 from ice by energetic light ions and elec-trons can be roughly estimated as the sum of theeffects of all the secondary electrons using theyield Ye for 10ndash100 eV electrons That is if weignore the percolation depth N we can write thenumber of O2 produced as the number of ioniza-tion events times Ye which already has in it thedestruction processes That is G NiYeEo YeW 0006 O2100 eV

This estimate of G is seen to be an order of mag-nitude larger than the keV proton value for O2 inTable 2 but is much smaller than the G value forenergetic Ne1 These differences from the simpleestimate are consistent with the observation thatthe yield in the saturation regime for penetratingions varies nonlinearly with excitation densityThe difference from the keV proton G value islikely due to the inability of O2 to escape fromdepth before destruction On the other hand thelarger G value seen using MeV Ne1 is likely dueto the high density of excitation This can resultin more efficient production of O2 increased pro-duction of damage sites for forming O2 andorincreased transport (percolation) along the tran-siently heated track That is the ability of theformed O2 to percolate to the surface during ion

OXIDANTS AT EUROPA 839

bombardment plays an essential role in deter-mining the measured yields as discussed aboveThis effect needs to be evaluated for a planetarysurface subject to particle fluxes that are many or-ders of magnitude smaller than those used in thelaboratory

In the absence of O2 yields for all of the rele-vant ions and energies an estimate of the totalsupply of O2 can be obtained using the G valuesin Table 2 For the energetic oxygen and sulfurions we use G values for neon and argon ionsThese are lower bounds as the data were obtainedat temperatures much lower than the surface tem-peratures at Europa (80ndash110K) For the fast pen-etrating electrons and protons we use the valuefor 200 keV protons in Table 2 For the heavy ionswe used the Ne1 and Ar1 values For compari-son we also give an estimate using the G valuesobtained from the low energy electron data andthe total ionization as discussed above

The G values are combined in Table 3 with theelectron and ion energy flux at Europa Althoughthere is some deflection of the flow due to in-duced fields the net energy flux to the surface isnot reduced much from these values (Paranicaset al 2002) In fact locally produced pick-up ionswhich are formed from the ejecta and acceleratedto the surface can contribute (Ip 1996) Furtheralthough G values are typically independent ofthe incident angle the loss of O2 is a surfaceprocess and is enhanced with angle That is thenumber of O2 produced within the escape depthincreases with increasing angle to the normal(steeper than the inverse cosine) This enhance-ment is 25ndash4 (Jurac et al 2001) Ignoring suchcorrections the total production rate calculated

(23ndash62 3 109 O2cm2s) is a lower bound to theactual production rate from the icy regions of Eu-roparsquos surface Recent modeling of the O2 atmo-sphere suggest that such a source rate is required(Shematovich and Johnson 2001)

Earlier estimates of the O2 source rate were in-correct because O2 production was assumed to bea fixed fraction of the H2O sputtering yield Atthe ambient temperature an ejected H2O sticks ef-ficiently (Smith and Kay 1997) to the surface ofneighboring grains in the porous regolith reduc-ing the sputtering yield for H2O (Johnson 1989)This reduction does not apply to the O2 fractionof the yield enhancing its relative importance toH2O Here we assume that all of the gas-phaseO2 produced can contribute to the ambient at-mosphere even if it is produced at a grain sur-face that is at some depth into the porous regolith

Finally Europarsquos surface is not pure ice every-where That is much of the trailing hemisphereis a hydrated sulfuric acid containing smallamounts of sulfur and SO2 and the leading hemi-sphere contains low levels of carbon CO2 andcarbonates Sodium and potassium atoms aresputtered from the icy regions (eg Leblanc et al2002) O2 production from surfaces containingthese species has not been measured C and S areboth proton scavengers and the other species canbe sources of O as discussed so that the O2 yieldfrom such regions could be enhanced Howeverthere can also be competing processes that mightreduce these rates In fact recent HST observa-tions suggest that the O2 might not be produceduniformly across Europarsquos surface (McGrath et al2000) The analysis of these observations is pre-liminary and is not simple since the observed

JOHNSON ET AL840

TABLE 3 APPLICATION TO EUROPA

OxygenImplantation Energy flux production

Species 109(cm2 s)a 109 keV(cm2 s)(1) 109(cm2 s)

H (10 keV) 0015(1) 12 0084 (072)bO (10 keV) 0015(1) 18 054cS (10 keV) 0009(1) 30 090cElectrons 62 043 (37)bPlasma (10 keV) (137 amu) 13(2) 10 03c

Total 79 23 (62)UV photons (64 eV) 46 01

aReferences are as follows 1Cooper et al (2001) [similar results in Paranicas et al (2002)] 2Mauk et al (1996)bUsing G for 200 keV protons (upper bounds in brackets are based on the low-energy electrons data as a model for

the secondary electrons produced)cUsing G for 15 MeV Ne1 at 7K a rough lower bound

emissions depend on the local plasma propertiesHowever the efficiency of O2 production in con-taminated ices needs to be studied

Porous regolith O2 interactions on grain surfaces

Although the O2 atmosphere is principally pro-duced in the near surface layers and is eventuallylost to space it also permeates Europarsquos porousregolith The porosity p (void space) of the re-golith is uncertain (Buratti 1995) It has been es-timated to be the most compact surface in theOuter Solar System p 025 on the trailing hemi-sphere and p 079 on the leading hemisphere(Buratti and Veverka 1983) and the most poroussurface p 096 globally (Domingue et al 1991Domingue and Hapke 1992) These large differ-ences are due to the uncertainties in the amountof backscattered light at small phase angles andon the photon absorption properties and grainsizes The more compact surface is likely correctGrain sizes rg of the order of 100 mm have beensuggested (Hansen and McCord 1999 2000Geissler et al 1998) Europarsquos regolith differs sig-nificantly from the lunar regolith which is asand-like surface of individual grains Analysis ofthe thermal inertia of Europarsquos surface (Grundyet al 2001) suggests that the ice grains are sin-tered forming a fairy castle-like structure Sin-tering can occur thermally or due to the incidentradiation In addition ices doped with acids andalkalis in the laboratory can be powdery solidswith the impurities primarily found at grain in-terfaces

The porous regolith affects the above estimatesof the oxygen production in three ways First forisotropic bombardment the mean distance d intothe surface at which an incident ion or photonstrikes a grain surface is d (89)rg(1 2 p) Us-ing the above value for rg then d 02 m if theporosity is very large p 96 Therefore de-sorbed molecules can come from a considerabledepth and will interact with grain surfaces Evenfor much lower porosities the effective sputteringyields can be significantly modified (Johnson1989) As discussed above the yields for ejectedH2O which sticks onto neighboring grains at theambient temperatures are reduced but those forejected O2 are not This enhances the relative im-portance of the O2 yields

Finally the surfaces and interfaces of the sin-tered grains in Europarsquos porous surface are placeswhere low temperature chemistry involving the

ambient oxygen can occur at depths that are wellbelow the nominal penetration depth of the inci-dent radiation The relative importance of suchchemistry is roughly indicated by the ratio of thetotal surface area in the regolith to the area of Eu-roparsquos visible surface Using the porosity andgrain size above and a regolith depth dr of theorder of a meter then this ratio is large 3(1 2p)drrg which is 103ndash2 3 104 for average porosi-ties of 96 and 25 respectively Therefore thegas-phase O2 molecules below the visible surfaceinteract frequently with the grain surfaces In thisway a very large volume of the subsurface couldbe altered chemically even if it is not directly ex-posed to the radiation

From the overlying structures seen in the high-resolution images of Europarsquos surface by theGalileo spacecraft (eg Prockter and Pappalardo2000) it is clear that the youngest surfaces are red-der The reddening in the visible has been attrib-uted to the presence of sulfur in the ice (McEwen1986 Johnson et al 1988 Calvin et al 1995 Carl-son et al 1999b) The amount of reddening in thespectra of the surface material decreases with ageand the surface materials appear to becomebrighter with age This might occur by the slowaccumulation of a frost over-layer or by radioly-sis (Carlson et al 2002 Johnson et al 2003) Ra-diolysis preferentially converts the sulfur in theice to sulfur dioxide or sulfuric acid both ofwhich have flat spectra in the visible In regionswhere the radiation does not directly penetratethe O2 permeating the regolith can oxidize thesulfur at the surfaces of the icy grains and at graininterfaces Similarly the carbon seen in the ice onEuroparsquos leading hemisphere (eg Carlson et al2004 Johnson et al 2003) can be oxidized by ra-diolysis in the near surface layers or by reactionson the grain surfaces at depth in the regolith bythe ambient O2 The meteoroid bombardmentthat produces the regolith also exposes fresh sur-faces for oxidation and can cause the collapse ofpores leading to the trapping of oxygen In thismanner both refractory and volatile oxidants canbe formed and trapped in the subsurface ice forpotential delivery to Europarsquos subsurface ocean

SUMMARY

The presence of a thin oxygen atmosphere andoxidants trapped in Europarsquos surface ice (egJohnson et al 2003) combined with the possible

OXIDANTS AT EUROPA 841

presence of a subsurface ocean (eg Carr et al1998) have made Europa an exciting target forfuture space exploration The molecular oxidantsproduced by radiolysis and photolysis could bespecies needed for carbon-based biochemistry inEuroparsquos putative subsurface ocean Howeverthere has been no model for the formation trap-ping and transport of relevant oxidants

In this paper we have reviewed the space ob-servations and laboratory studies related to thechemistry of formation of oxidants in ice by theincident radiation Because the laboratory dataare incomplete we examined the rate equationsfor formation of oxygen and related species byradiolysis and photolysis We showed how suchequations can be simplified to the analytic modelthat was used for fitting the fluence and temper-ature dependence of the O2 yield produced bylow-energy electrons (Sieger et al 1998) In de-riving the analytic model from the more completerate equations the competition between forma-tion and destruction processes can be studied ascan the role of trapping sites escape from depthand temperature history These equations also al-lowed us to put incident photon electron and iondata on a similar footing for the first time

The model can be used to suggest measure-ments that are needed to describe the chemicalstate of the irradiated ice and to interpret new lab-oratory data We showed that if reactions are fastthe yield data could be consistent with a precur-sor being formed by two events as is the case forperoxide However we favor trapped O as a pre-cursor as initially suggested by Matich et al(1993) for production of excited O2 Such aspecies if present could be detected spectro-scopically as can trapped O2 through bands ac-tivated by lattice perturbations (Cooper et al2003a)

The available oxygen yield data were used tocompare G values for different incident particlesThese were then used to estimate the oxygen pro-duction in the radiation environment at EuropaThe lower bounds were consistent with the pro-duction rates obtained from a recent model of Eu-roparsquos atmosphere The porosity of the regolithwas also considered It not only affects the rela-tive sputtering yields for O2 and H2O but alsoprovides an enormously increased surface areaon which the ambient gas-phase oxygen can re-act with species in the ice That this might be oc-curring is consistent with the change in re-

flectance associated with surface aging Such re-actions could also convert ambient molecularoxygen into other more refractory species that re-main trapped on subduction of Europarsquos crust

Although radiolysis produces a stable chemi-cally altered surface in the laboratory the parti-cle flux to Europarsquos surface is many orders ofmagnitude lower Therefore annealing can be ef-fective at temperatures that are much lower thanthose typically found experimentally Belowabout 120K the annealing times appear to be long(eg Baragiola 2003) so that laboratory data canoften be directly applied to Europarsquos icy surfaceHowever account has to be taken of the syner-gism between the UV and the energetic plasmaas discussed earlier (Johnson and Quickenden1997)

Transient surface melting might initiate lifeprocesses but subduction of oxidants to the pu-tative ocean is likely to be required if biochem-istry is to occur at Europa (Chyba and Hand2001) Although the temperature versus depthinto Europarsquos regolith is not well known belowthe regolith the temperature and pressure appar-ently increase with depth At some depth (ap-proximately kilometers) either very warm ice oran ocean exists (Ruiz and Tejero 2000) Thereforechemically refractory oxidants are likely to be re-quired in order to survive subduction to thewarmer subsurface regions Trapped species ifdelivered downward can become mobile andwill segregate forming inclusions (Johnson andJesser 1997) Indeed peroxide appears to segre-gate in ice (Gurman et al 1967) and these inclu-sions may remain relatively stable until the per-oxide is dissolved in the putative ocean Thevolatiles seen in reflectance such as O2 SO2 andCO2 will also segregate But at the higher tem-peratures they could also diffuse through thebulk react or percolate into the regolith atmo-sphere Therefore the observation of the trappedvolatiles in Europarsquos surface ice is of interest pri-marily because it suggests that oxidants are pre-sent The formation of oxygen-rich molecules fol-lows the preferential loss of hydrogen Howeverthe surface is in a local steady state with im-planted H O and S from the plasma and the lossto space Since the plasma flux to the surface canbe variable this balance can also vary Indeedtemporal changes in reflectance have been re-ported (Domingue and Hendrix 2004)

Based on the models for formation of O2 ex-

JOHNSON ET AL842

amined in this paper additional experiments areneeded to determine the chemical state of irra-diated ice at low fluences Of immediate inter-est is the identification of stable molecular pre-cursors required for the formation of O2 Basedon the available laboratory data individual per-oxide molecules trapped in the ice do not appearto be the principal precursors but trapped O(Matich et al 1993) remains a viable candidateAlthough the data suggest that the local hydro-gen-bonding network is important in the pro-duction of O2 the effect of the formation tem-perature and radiation damage on the structureneed to be measured It is also important to de-scribe quantitatively the relationship betweenH2 loss and the formation of O2 and H2O2 overa range of temperatures and fluences Finallyone would like to determine the steady-statedensities of biochemically interesting moleculesformed by radiolysis in an ice containing sulfurand carbon at Europarsquos surface temperaturesand then determine their stability with increas-ing temperature and pressure Such experimentsstill need to be carried out Here we have pre-sented a model that brings together the variouslaboratory data and now can be used to considerthe formation trapping and transport of relevantoxidants at Europa

APPENDIX

Introduction

In this Appendix we examine rate equationsthat describe various aspects of the radiolysis orphotolysis of ice These equations need to be un-derstood if one is to be able to calculate the for-mation trapping and transport of oxidants Theemphasis is on the formation of molecular oxy-gen We integrate these equations over the escapedepth of O2 to give the simplified rate equationsdiscussed in the text in which number densitiesare replaced by column densities (eg Eqs 1 and2) In this way we can examine the relationshipbetween the rate constants and the effective pre-cursor formation and destruction cross sectionsextracted from data We first consider a precur-sor such as H2O2 that is formed from two exci-tation events This bears on the observed linear-ity of the oxygen yield versus fluence We thenconsider the equations for trapped versus a tran-

siently mobile species We use a reaction betweena mobile and trapped O as an example althoughO 1 OH can also give oxygen Finally we ex-amine the possible dependence of the O2 forma-tion process on the density of traps (defectsvoids and grain boundaries) as a function of tem-perature The results are discussed in the text

For clarity all possible radiolytic species arenot included in the discussion below Thereforepotentially important products such as HO2 andthe charged species (eg H3O1 the trapped elec-tron etc) are missing In each section we attemptto make a point related to production of O2 andother oxidants by using a subset of the productsrather than attempt a complete description withmany unknown rate constants

Fast versus slow processes

To describe the O2 production rate from a pre-cursor such as H2O2 we first use a simple but in-structive set of rate equations to describe the for-mation of H2O2 We then relate these to thesimplified rate equation for the precursor in thetext Eq 1 and show how the postulated precur-sor formation cross section sp is related to theH2O dissociation cross section sd by the reactionrates

Using the reactions

H2O 1 radiation R H 1 OH (sd)

H 1 H R H2 (k2H)

H 1 OH R H2O (kHOH)

OH 1 OH R H2O2 (k2OH)

the reaction pathways are described by

H2

Hqk2H

1 OHH2O 1 radiation mdashRsd H

OHH2O2

OH Rk2OH

kHOH

HH2O

OXIDANTS AT EUROPA 843

The corresponding rate equations in an irradiatedvolume are

dnOHdt 5 sd f n 2 2 k2OH nOH nOH

2 kHOH nH nOH (A1a)

dnHdt 5 sd f n 2 2 k2H nH nH

2 kHOH nH nOH (A2a)

dnH2O2dt 5 k2OH nOH nOH

2 destruction processes (A3a)

Here sd is the dissociation cross section f is theradiation flux and n is the molecular numberdensity of ice with nOH and nH the density of dis-sociation products OH and H respectively Therate constants k2OH k2H and kHOH describe thereactions above In this model the H2 formed isassumed to escape for temperatures above about20K We do not distinguish between trappedspecies and mobile species although in fact thatdistinction is important as discussed in the nextsection That is any OH produced at low tem-perature by one ion would likely be trapped priorto interacting with an OH produced by a subse-quent ion At higher temperatures this can breakdown as discussed in the text We first relate theabove set of equations to Eq 1 in the text and thenconsider precursor destruction processes

Equation 1 in the text is obtained by solvingthese equations together and then by integratingover a depth Dx Here Dx is the smaller of the pen-etration depth of the particle and the O2 escape(percolation) depth Assuming the molecular den-sity of ice n is constant with depth Dx n 5 N inEq 1 Comparing Eq 1 with Eq A3a the precur-sor formation rate sp f N is equivalent to [k2OH

nOH nOH] Dx Assuming that the processes in EqsA1a and A2a are fast then the steady state is read-ily obtained In addition if k2H k2OH there ispreferential loss of H2 at short times as suggestedby the laboratory data If steady state is reachedthe production of H2 is equal to that of H2O2 us-ing only the equations above Further one obtains

[k2OH nOH nOH] Dx R [fp sd] f N (A4)

with

fp 5 05[1 1 05 kHOH(k2H k2OH)05] (A5a)

Therefore the precursor formation cross sectionsp in Eq 1 is a fraction fp of the total dissocia-

tion cross sd Since laboratory data suggest theH2O2 fraction in an irradiated ice is small thenkHOH (k2OH k2H)05 which is likely givingfp R [(k2OH k2H)05kHOH]

Trapped versus mobile species

To examine the fluence dependence discussedin the text we note that trapped OH and freshlyproduced transiently mobile OH can be treatedas different species (eg Matich et al 1993)Therefore assuming no trapping of H the OHrate equation above would be replaced by twoequations one for the trapped species nOHt andthe other for the freshly produced mobile speciesnOH Assuming the reaction of two mobile OHunits is unlikely and that peroxide is formed fromthe interaction of a trapped OH with a subse-quently produced OH an expanded set of equa-tions is obtained

OH 1 trap R OHt (kOHt)

OH 1 OHt R H2O2 (kOHOHt)

H 1 OHt R H2O (kHOHt)

then the reaction pathways are described by

qH2

H2O 1 radiation mdashRsdH

qk2H

1 OHH

trap OH

OH RkOHt

OHt kOHOHtH2O2

kHOHkHOHt

H H2O HH2O

The resulting set of equations is

dnOHdt 5 sd f n 2 kHOH nH nOH

2 kOHt nOH nt 2 kOHOHt nOH nOHt (A1b)

dnOHtdt 5 kOHt nOH nt 2 kOHOHt nOH nOHt

2 kHOHt nH nOHt (A1b9)

dnHdt 5 sd f n 2 2 k2H nH nH 2

kHOH nH nOH 2 kHOHt nH nOHt (A2b)

JOHNSON ET AL844

dnH2O2dt 5 kOHOHt nOH nOHt

2 destruction processes (A3b)

where nt is the trap density in the irradiated iceAt low T we can assume nt is large so thatdntdt 0 simplifying the set of equations If theprecursor of interest is trapped O as discussed inthe text then a similar set of equations can be con-structed for O and Ot as in the example to followOf course a full description over a broad rangeof temperatures would require the considerationof the both trapped and mobile H O and OHthe rate of change of trap density and the vari-ous destruction processes

The above equations should be solved simul-taneously but for the incident particle fluxes typ-ically used in these studies the diffusion of mo-bile species to traps is relatively fast This is trueeven at low temperatures if there is a significantdensity of trapping sites This is usually the casefor samples formed at low T Therefore in a lowtemperature solid at low dose rates there is not asignificant steady-state background of diffusingOH There are trapped or freshly produced OHunits which rapidly react Averaging over thebombardment rate this can be approximated bywriting dnOHdt 0 and then separating nOH outin Eq A1b Assuming also that H diffuses rapidlyand either reacts or forms H2 we also writednHdt 0 on the average Therefore betweenimpacts only species trapped at defects pores orgrain surfaces are assumed to be present as sug-gested by the lack of a dependence on the inci-dent beam flux In this limit the rate of change oftrapped OH roughly simplifies to

dnOHtdt 5 2 k2H nH nH 2 2 kOHOHt nOH nOHt

It is seen that the production of peroxide dependson the loss of H by formation of H2 as discussedand at steady state the production rate for per-oxide equals the rate of formation of H2 in theseequations

Treating the mobile and trapped species sepa-rately we can further simplify the rate equationsand obtain the precursor model used in the textThat is integrating over the relevant depth Dxthe densities are replaced by column densitiesThe rate constants ki are then related to effectivecross sections si In this way cross sections aregiven as an interaction length for each freshlyproduced mobile radical [ki nj] R [si f] The

trapped OH column density NOHt and peroxideproduction rate can be estimated from ldquolinearrdquorate equations

dNOHtdt lt [sd f] N 2 2 [s2OH f] NOHt

2 [sHOH f] NOHt (A6a)

dNH2O2dt lt [s2OH f] NOHt 2 s NH2O2 (A6b)

Here sd is again the dissociation cross sectionand the destruction processes in Eq A3b are con-tained in s The second and third terms on theright in Eq A6a are the loss of OH due to the pro-duction of H2O2 and H2O In the integration ofEq A1b9 [2 kOH OHt nOH] becomes [2 s2OH f] and[kHOHt nOH] becomes [sHOH f] Therefore thedensity and the interaction length of mobile H arecontained in sHOH If H2O2 is indeed the pre-cursor then in Eq 2 in the text [sp f N] is be-comes [s2OH f NOHt] from Eq A6b Accountingfor precursor destruction which can give backtwo OH and treating H2 can both be included inthe integrated model via the si values

The simplified set of equations above allows usto examine the relative time (fluence) scales forprecursor formation (in this case H2O2) SolvingEqs A6a and A6b the column density of OH andthe concentration of precursors versus fluence cpcan be obtained Using the column of H2O2 pro-duced as the precursor column Np in Eq 2 thenO2 yield can be obtained The results can be com-pared with the results from Eqs1 and 2 in thetext The average precursor (H2O2) concentrationover the column depth of water molecules N is

cp 5 fp (sds)[1 2 exp(2s F)] 2 (ssr)

[1 2 exp(2sr F)] (A7)

with

fp 5 2s2OH(sr 2 s) (A5b)

In these equations sr 5 (2s2OH 1 sHOH) andssr 1

It is seen that the precursor concentration withtime cp depends on two rates the reaction ratesthat determine sr and the destruction rates in sAs in Eq 4a the yield can be written as

Y 5 sO2 cp N (A8)

If sr and s are comparable then the yield at small

OXIDANTS AT EUROPA 845

fluence is quadratic in F However if sr sthen the linear dependence on fluence is obtainedwith a small shift in the fluence as shown in Eq4c in the text Further the steady-state yield isnow

Y` (T) 5 sO2 [fp sd] Ns (A9)

That is sp R [fp sd] as in Eq A4 and the tem-perature dependence for the incident electrondata must be contained in [sO2 fp] or in the ratio[sO2 s2OH]sHOH]

For comparison with their precursor modelSieger et al (1998) examined the possibility of ob-taining the O2 precursor by considering reactionsfor a density of diffusing dissociation products asin Eqs A1 and A2 rather than a trapped productinteracting with a freshly formed dissociationproduct Based on the above the lowest fluence atwhich O2 is first detected experimentally sets alimit on the time scale for mobilendashmobile reactions

Effect of trap density

Equations A1b A1b9 A2b and A3b can be ap-proximately solved assuming that the dnOHdtdnHdt and dntdt are very small on the averageas discussed above In order to examine the roleof the trap density nt which can depend on thetemperature and method of formation and on theradiation history of the ice we use the slightlysimpler precursor model In this model the O pro-duced directly by dissociation traps forming aprecursor O2 is then formed when a freshly pro-duced O or other species reacts with a trappedoxygen atom Ot

H2O 1 radiation R H2 1 O (sd)

O 1 trap R Ot (kOt)

O 1 Ot R O2 (kOOt)

The reaction pathways for the oxygen species are

H2O 1 radiation mdashRsd9

trap OO2

kOtOt kOOt

H2 1 OOHt

kOR H(kHO t)

ROR

We assume for simplicity both H2 and O2 escapefrom depth Dx after formation We also allow mo-bile H to react with Ot (kHOt) removing trappedO and both the mobile H and O to be removedby other reactants R (eg OH or contaminants)This leads to a simplified set of equations not di-rectly involving OH or peroxide

dnOdt 5 sd9 f n 2 kOt nO nt 2 kOR nO nR

2 kOOt nO nOt (A1c)

dnOtdt 5 kOt nt nO 2 kOOt nO nOt

2 kHOt nH nOt (A1c9)

dnHdt 5 sd f n 2 kHOt nH nOt

2 kHR nH nR (A2c)

dnO2dt 5 kOOt nO nOt (A3c)

These equations of course act in parallel with theequations above for OH and mobile OH may re-act with trapped O to form O2 Here we focusonly on mobile O

Since the diffusion of H and O are fast and thedensity of traps is large the nH and nt averagedover the bombardment rate can be taken veryroughly as time independent Further if we as-sume the number of trapped O nOt is small com-pared with the number of sites at which H canreact then nH [sd f nkHR nR] For a low ra-diation flux f nH is small on the average as as-sumed Since the O2 yield YO2 is equal to[(dnO2dt) Dxf] with Dx the escape depth thenthe steady-state yield Y` can be obtained ana-lytically

[Y`][sd9 N] 5 1 1 (q 2 1) 2 [(q 2 1)2

1 8 q d]054 d (A10)

q 5 [kOt (nt 1 nR)] (kHOt nH)(sd9 f n) kOOt

5(sdsd9) [kOt (nt 1 nR)] kHOt[kOOt kHR nR]

d 5 05 [1 1 nt(nt 1 nR)]

We note that for nt large q can be large Assum-ing that q 1 and nt nR then Eq A10 be-comes

Y` R [sd9 N]q 5 (sd9 N) (sd9sd)

[kOOt kHR nR][kHOt kOt nt] (A5c)

JOHNSON ET AL846

Comparing with Y` in the simple precursormodel in Eq 4b then s R [(kHOt nH)f] R [sd nkHOt][kHR nR] In addition the product [sO2sp] R sd9 sd9 [kOOt n][kOt nt] That is the dis-sociation cross sections are modified by reactionrates

Quite remarkably it is seen in Eq A5c that thesteady-state yield Y` can depend inversely onthe density of traps nt in the temperature andflux regimes for which the approximations arerelevant That is if there are competing destruc-tion process for trapped O then at high trap den-sity a mobile O must find a trapped O before itbecomes trapped or before mobile H destroys theOt In this way the increasing O2 yield with in-creasing T can be related to the change in the trap-ping density nt with the formation temperatureof the ice sample As the temperature increasesand the density of traps in the bulk decreases thesurface and grain interfaces become the only trap-ping sites

ACKNOWLEDGMENTS

REJ acknowledges support by a Gledden Fel-lowship in Chemistry School of Biomedical andChemical Sciences at The University of WesternAustralia and by grants from the NSF Astron-omy Program and NASArsquos Planetary Geologyand Geophysics Program PDC acknowledgessupport by an Australian Postgraduate Award

ABBREVIATIONS

HST Hubble Space Telescope MJQ MatichndashJohnsonndashQuickenden NIMS Near Infrared Map-ping Spectrometer UV ultraviolet

REFERENCES

Ayotte P Smith RS Stevenson KP Dohnarsquolek ZKimmel GA and Kay BD (2001) Effect of porosityon the adsorption desorption trapping and release ofvolatile gases by amorphous solid water J Geophys Res106(E12) 33387ndash33392

Bahr DA Famaacute M and Baragiola RA (2001) Radiol-ysis of water ice in the outer solar system sputteringand trapping of radiation products J Geophys Res 10633285ndash33290

Baragiola RA (2003) Microporous amorphous water icefilms and astronomical implications In Water in Con-

fining Geometries edited by JP Devlin and B Buch El-sevier New York in press

Baragiola RA and Bahr DA (1998) Laboratory studiesof the optical properties and stability of oxygen onGanymede J Geophys Res 103 25865ndash25872

Baragiola RA Atteberry CL and Bahr DA (1999) Ori-gin of solid oxygen on Ganymede Reply to commentby RE Johnson on ldquoLaboratory studies of the opticalproperties and stability of oxygen on Ganymederdquo JGeophys Res E104 14183ndash14187

Baragiola RA Vidal RA Svendsen W Schou J ShiM Bahr DA and Atteberrry CL (2003) Sputteringof water ice Nucl Instrum Methods Phys Res B 209294ndash303

Bar-Nun A Herman G Laufer D and Rappaport ML(1985) Trapping and release of gases by water ice andimplications for icy bodies Icarus 63 317ndash332

Baverstock KF and Burns WG (1976) Primary pro-duction of oxygen from irradiated water as an expla-nation for decreased radiobiological oxygen enhance-ment at high LET Nature 260 316ndash318

Benit J and Brown WL (1990) Electronic sputtering ofoxygen and water molecules from thin films of waterice bombarded by MeV Ne1 ions Nucl Instrum Meth-ods Phys Rev B 46 448ndash451

Benit J Bibring J-P della-Negra S LeBeyec YMendenhall M Rocard F and Standing K (1987)Erosion of ices by ion irradiation Nucl Instrum Meth-ods Phys Rev B 1920 838ndash842

Brown WL Augustyniak WM Lanzerotti LJ John-son RE and Evatt R (1980) Linear and nonlinearprocesses in the erosion of H2O ice by fast light ionsPhys Rev Lett 43 1632ndash1635

Brown WL Augustyniak W M Simmons E Mar-cantonio KJ Lanzerotti LJ Johnson RE BoringJW Reimann CT Foti G and Pirronello V (1982)Erosion and molecular formation in condensed gasfilms by electronic energy loss of fast ions Nucl In-strum Methods B 1 307ndash314

Brown WL Augustyniak WM Marcantonio KJ Sim-mons EN Boring JW Johnson RE and ReimannCT (1984) Electronic sputtering of low-temperaturemolecular solids Nucl Instrum Methods B 1 307

Buratti B (1995) Photometry and surface structure of theicy Galilean satellites J Geophys Res 100(E9) 19061ndash19066

Buratti B and Veverka J (1983) Voyager photometry ofEuropa Icarus 55 93ndash110

Calvin WM and Spencer JR (1997) Latitudinal distri-bution of O2 on Ganymede observations with the Hub-ble Space Telescope Icarus 130 505ndash516

Calvin WM Clark RN Brown RH and Spencer JA(1995) Spectra of the icy Galilean satellites from 02 to5 mm a compilation new observations and a recentsummary J Geophys Res 100 19041ndash19048

Calvin WM Johnson RE and Spencer JA (1996) O2

on Ganymede spectral characteristics and plasma for-mation mechanisms Geophys Res Lett 23 673ndash676

Calvin WM Aninich VG and Brown RH (2004) Oxy-gen temperature and phase effects Icarus (in press)

OXIDANTS AT EUROPA 847

sample surface or grain interfaces might act astraps accounting for the observed drop in theyield at high T For this reason it is important todetermine if the yield for T greater than 150Kdepends on the incident flux f Since other quan-tities in the expression for Y` could also be tem-perature dependent experiments are neededThese would likely be pulsed radiolysis experi-ments in which the transients are studied ratherthan the steady-state yields However the avail-able data for lightly ionizing radiation do suggestthat the O2 yield depends inversely on trap den-sity as in the model above

Escape from depth

The quantity N in Eq 1 is the ldquodepthrdquo given asa molecular column density from which O2 formedbelow the surface can percolate to the surface andescape It can be converted to a physical depth bydividing by the molecular number density of theice In the case of the low-energy electrons thedepth below the surface at which ionizations andexcitations are produced is small However forfast ions that penetrate many monolayers this isnot the case Below we examine N in Eq 1

Although it has long been understood that twoexcitations events are required to produce O2(eg Brown et al 1982 Reimann et al 1984) theprecursor model suggests that at saturation(steady state) any new event can produce O2 fromthe steady-state density of trapped precursorsTherefore one might expect the yield at satura-tion to be nearly linear in the excitation cross sec-tion This is the case for the low-energy electronsand it was initially assumed to be the case for thefast ions (Brown et al 1980 Reimann et al 1984)However Baragiola et al (2003) showed that thetotal sputtering yield in the saturation region attemperatures greater than 100K which is dom-inated by decomposition (Brown et al 1982) isproportional to the square of the electronic en-ergy deposition per unit path length in the solid(dEdx)e Assuming that the production and de-struction cross sections in Y` in Eq 4b roughlyscale with (dEdx)e then their ratio [sp sO2s]also roughly varies as (dEdx)e If this is the casethen the observed quadratic dependence impliesthat the depth N from which O2 can be mobi-lized to escape must also depend on the excita-tion density (dEdx)e This is consistent with theobservation that the O2 yield depends on thick-ness for samples thinner than the ion penetration

depth (Reimann et al 1984 Benit and Brown1990) Therefore the percolation depth for escapeN is a critical factor in comparing yields for pen-etrating radiation with those for non-penetratingradiation like the low-energy electron and low-energy ion data It is also seen that percolationand trapping can be included within the simplemodel considered here

Model summary

We have examined the simple precursor modelintroduced by Sieger et al (1998) to explain theirelectron data We have shown how it can be gen-eralized and derived from an integration of theappropriate chemical rate equations with thecross sections and total column density contain-ing the reaction rates and the physics of diffusivetransport Therefore the parameters can havevery different meanings from those initially as-sumed The equations are useful when there arefast processes which go to completion in timesmuch shorter then the time between particle im-pacts and slow processes associated with speciesthat are assumed to be trapped between particleimpacts This separation allowed us to examinethose models proposed for precursor formationand destruction due to reacting mobile speciesformed from dissociation events Further fromthe size of the onset of the linearity of the yieldversus fluence we can place constraints on therole of nonlinear processes such as those thatmight be associated with the formation of a pre-cursor like peroxide

With these extensions this simple model is use-ful for approximating the full rate equations inthe temperature region for which the yields de-pend on the fluence (dose) and not on the doserate Based on the assumptions in the model thisimplies temperatures at which precursor speciesare stable over experimental times At the high-est temperatures for the yields shown in Fig 4athis might not be the case This should be testedexperimentally In this regard it is puzzling thatexperiments have shown that the absorptionbands associated with certain trapped speciessuch as OH and trapped electrons ldquodisappearrdquoat 100ndash120K (see eg Johnson and Quickenden1997) whereas the O2 yield data indicate that rel-evant precursor species must be stable to 140KTherefore the proposed O2 production processescan only be distinguished by directly determin-ing the chemical state of the irradiated ice

JOHNSON ET AL838

It may be the case that a particular precursoris not required as mentioned earlier Rather af-ter a sufficient level of oxidation a newly pro-duced O or OH can react in the oxygen-rich lat-tice producing O2 in a pseudo-first-order processThis can be tested experimentally Indeedtrapped O and peroxide may be changed into oneanother as in the matrix isolation experiments(Khriachtchev et al 2000) Therefore their rela-tive concentrations versus temperature and doseshould be studied

In the model considered here it is seen that theparameters in Y` must account for the strong tem-perature dependence Since the number of bulktrapping sites decreases with increasing T as theice sample is annealed one might expect that theprecursor concentration would decrease We sug-gest that the observed temperature dependenceis determined by the competition between theavailability of trapping sites and mean interactiondistances of a mobile species such as O and HThat is the O2 yield varies inversely with the den-sity of traps over a range of temperatures In thesimple model above this effect is included indi-rectly as shown using the more detailed equa-tions in the Appendix

EUROPA

Oxygen and peroxide production at Europa G values

The dependence of the yield on the thicknessof the ice sample for highly penetrating ions sug-gests that the O2 yield is roughly related to theamount of energy deposited in the sample There-fore the amount of O2 produced and lost fromice in steady state is often given as a G value thenumber of molecules produced per 100 eV de-posited in Table 2 In applying these G values tothe planetary environment care must be takensince the yields are temperature fluence and ex-citation density dependent They also are affectedby the presence of other proton scavengersknown to be present in Europarsquos ice

The G values for peroxide in Table 2 are thosefor production at low fluence but the G valuesfor O2 are for the production and escape of gas-phase oxygen at saturation Therefore they de-pend on the formation and destruction cross sec-tions in the model above Such G values can beused to describe the steady-state source of the am-

bient O2 atmosphere at Europa but are not help-ful in determining the ability to produce and trapO2 at depth into the icy surface Over the timescales of the experiments some O2 is seen to re-main trapped and is not destroyed by radicalsTypically there are a number of trapping sitesand recent estimates of the activation energy foran O2 molecule trapped in a substitutional site(Hori and Hondoh 2002) suggest that it can betrapped for times comparable to the time re-quired for the conversion of amorphous ice intoIh (Ayotte et al 2001) The stability of O2 inclu-sions has not been studied

To understand the relationship between the Gvalues for O2 in Table 2 we note that fast lightions and electrons lose their energy predomi-nantly by producing ionizations and excitationsThe ionization events produce secondary elec-trons that can result in further ionizations Thenumber of ionization events Ni produced by afast light ion stopping in a water ice is Ni 5EoW where Eo is the initial energy and W is theaverage energy expended per ionization event27 eV in water ice Therefore radiolysis by fastions is often described as the net effect of theshower of electrons set in motion by the incidention In this manner the G values for productionof O2 from ice by energetic light ions and elec-trons can be roughly estimated as the sum of theeffects of all the secondary electrons using theyield Ye for 10ndash100 eV electrons That is if weignore the percolation depth N we can write thenumber of O2 produced as the number of ioniza-tion events times Ye which already has in it thedestruction processes That is G NiYeEo YeW 0006 O2100 eV

This estimate of G is seen to be an order of mag-nitude larger than the keV proton value for O2 inTable 2 but is much smaller than the G value forenergetic Ne1 These differences from the simpleestimate are consistent with the observation thatthe yield in the saturation regime for penetratingions varies nonlinearly with excitation densityThe difference from the keV proton G value islikely due to the inability of O2 to escape fromdepth before destruction On the other hand thelarger G value seen using MeV Ne1 is likely dueto the high density of excitation This can resultin more efficient production of O2 increased pro-duction of damage sites for forming O2 andorincreased transport (percolation) along the tran-siently heated track That is the ability of theformed O2 to percolate to the surface during ion

OXIDANTS AT EUROPA 839

bombardment plays an essential role in deter-mining the measured yields as discussed aboveThis effect needs to be evaluated for a planetarysurface subject to particle fluxes that are many or-ders of magnitude smaller than those used in thelaboratory

In the absence of O2 yields for all of the rele-vant ions and energies an estimate of the totalsupply of O2 can be obtained using the G valuesin Table 2 For the energetic oxygen and sulfurions we use G values for neon and argon ionsThese are lower bounds as the data were obtainedat temperatures much lower than the surface tem-peratures at Europa (80ndash110K) For the fast pen-etrating electrons and protons we use the valuefor 200 keV protons in Table 2 For the heavy ionswe used the Ne1 and Ar1 values For compari-son we also give an estimate using the G valuesobtained from the low energy electron data andthe total ionization as discussed above

The G values are combined in Table 3 with theelectron and ion energy flux at Europa Althoughthere is some deflection of the flow due to in-duced fields the net energy flux to the surface isnot reduced much from these values (Paranicaset al 2002) In fact locally produced pick-up ionswhich are formed from the ejecta and acceleratedto the surface can contribute (Ip 1996) Furtheralthough G values are typically independent ofthe incident angle the loss of O2 is a surfaceprocess and is enhanced with angle That is thenumber of O2 produced within the escape depthincreases with increasing angle to the normal(steeper than the inverse cosine) This enhance-ment is 25ndash4 (Jurac et al 2001) Ignoring suchcorrections the total production rate calculated

(23ndash62 3 109 O2cm2s) is a lower bound to theactual production rate from the icy regions of Eu-roparsquos surface Recent modeling of the O2 atmo-sphere suggest that such a source rate is required(Shematovich and Johnson 2001)

Earlier estimates of the O2 source rate were in-correct because O2 production was assumed to bea fixed fraction of the H2O sputtering yield Atthe ambient temperature an ejected H2O sticks ef-ficiently (Smith and Kay 1997) to the surface ofneighboring grains in the porous regolith reduc-ing the sputtering yield for H2O (Johnson 1989)This reduction does not apply to the O2 fractionof the yield enhancing its relative importance toH2O Here we assume that all of the gas-phaseO2 produced can contribute to the ambient at-mosphere even if it is produced at a grain sur-face that is at some depth into the porous regolith

Finally Europarsquos surface is not pure ice every-where That is much of the trailing hemisphereis a hydrated sulfuric acid containing smallamounts of sulfur and SO2 and the leading hemi-sphere contains low levels of carbon CO2 andcarbonates Sodium and potassium atoms aresputtered from the icy regions (eg Leblanc et al2002) O2 production from surfaces containingthese species has not been measured C and S areboth proton scavengers and the other species canbe sources of O as discussed so that the O2 yieldfrom such regions could be enhanced Howeverthere can also be competing processes that mightreduce these rates In fact recent HST observa-tions suggest that the O2 might not be produceduniformly across Europarsquos surface (McGrath et al2000) The analysis of these observations is pre-liminary and is not simple since the observed

JOHNSON ET AL840

TABLE 3 APPLICATION TO EUROPA

OxygenImplantation Energy flux production

Species 109(cm2 s)a 109 keV(cm2 s)(1) 109(cm2 s)

H (10 keV) 0015(1) 12 0084 (072)bO (10 keV) 0015(1) 18 054cS (10 keV) 0009(1) 30 090cElectrons 62 043 (37)bPlasma (10 keV) (137 amu) 13(2) 10 03c

Total 79 23 (62)UV photons (64 eV) 46 01

aReferences are as follows 1Cooper et al (2001) [similar results in Paranicas et al (2002)] 2Mauk et al (1996)bUsing G for 200 keV protons (upper bounds in brackets are based on the low-energy electrons data as a model for

the secondary electrons produced)cUsing G for 15 MeV Ne1 at 7K a rough lower bound

emissions depend on the local plasma propertiesHowever the efficiency of O2 production in con-taminated ices needs to be studied

Porous regolith O2 interactions on grain surfaces

Although the O2 atmosphere is principally pro-duced in the near surface layers and is eventuallylost to space it also permeates Europarsquos porousregolith The porosity p (void space) of the re-golith is uncertain (Buratti 1995) It has been es-timated to be the most compact surface in theOuter Solar System p 025 on the trailing hemi-sphere and p 079 on the leading hemisphere(Buratti and Veverka 1983) and the most poroussurface p 096 globally (Domingue et al 1991Domingue and Hapke 1992) These large differ-ences are due to the uncertainties in the amountof backscattered light at small phase angles andon the photon absorption properties and grainsizes The more compact surface is likely correctGrain sizes rg of the order of 100 mm have beensuggested (Hansen and McCord 1999 2000Geissler et al 1998) Europarsquos regolith differs sig-nificantly from the lunar regolith which is asand-like surface of individual grains Analysis ofthe thermal inertia of Europarsquos surface (Grundyet al 2001) suggests that the ice grains are sin-tered forming a fairy castle-like structure Sin-tering can occur thermally or due to the incidentradiation In addition ices doped with acids andalkalis in the laboratory can be powdery solidswith the impurities primarily found at grain in-terfaces

The porous regolith affects the above estimatesof the oxygen production in three ways First forisotropic bombardment the mean distance d intothe surface at which an incident ion or photonstrikes a grain surface is d (89)rg(1 2 p) Us-ing the above value for rg then d 02 m if theporosity is very large p 96 Therefore de-sorbed molecules can come from a considerabledepth and will interact with grain surfaces Evenfor much lower porosities the effective sputteringyields can be significantly modified (Johnson1989) As discussed above the yields for ejectedH2O which sticks onto neighboring grains at theambient temperatures are reduced but those forejected O2 are not This enhances the relative im-portance of the O2 yields

Finally the surfaces and interfaces of the sin-tered grains in Europarsquos porous surface are placeswhere low temperature chemistry involving the

ambient oxygen can occur at depths that are wellbelow the nominal penetration depth of the inci-dent radiation The relative importance of suchchemistry is roughly indicated by the ratio of thetotal surface area in the regolith to the area of Eu-roparsquos visible surface Using the porosity andgrain size above and a regolith depth dr of theorder of a meter then this ratio is large 3(1 2p)drrg which is 103ndash2 3 104 for average porosi-ties of 96 and 25 respectively Therefore thegas-phase O2 molecules below the visible surfaceinteract frequently with the grain surfaces In thisway a very large volume of the subsurface couldbe altered chemically even if it is not directly ex-posed to the radiation

From the overlying structures seen in the high-resolution images of Europarsquos surface by theGalileo spacecraft (eg Prockter and Pappalardo2000) it is clear that the youngest surfaces are red-der The reddening in the visible has been attrib-uted to the presence of sulfur in the ice (McEwen1986 Johnson et al 1988 Calvin et al 1995 Carl-son et al 1999b) The amount of reddening in thespectra of the surface material decreases with ageand the surface materials appear to becomebrighter with age This might occur by the slowaccumulation of a frost over-layer or by radioly-sis (Carlson et al 2002 Johnson et al 2003) Ra-diolysis preferentially converts the sulfur in theice to sulfur dioxide or sulfuric acid both ofwhich have flat spectra in the visible In regionswhere the radiation does not directly penetratethe O2 permeating the regolith can oxidize thesulfur at the surfaces of the icy grains and at graininterfaces Similarly the carbon seen in the ice onEuroparsquos leading hemisphere (eg Carlson et al2004 Johnson et al 2003) can be oxidized by ra-diolysis in the near surface layers or by reactionson the grain surfaces at depth in the regolith bythe ambient O2 The meteoroid bombardmentthat produces the regolith also exposes fresh sur-faces for oxidation and can cause the collapse ofpores leading to the trapping of oxygen In thismanner both refractory and volatile oxidants canbe formed and trapped in the subsurface ice forpotential delivery to Europarsquos subsurface ocean

SUMMARY

The presence of a thin oxygen atmosphere andoxidants trapped in Europarsquos surface ice (egJohnson et al 2003) combined with the possible

OXIDANTS AT EUROPA 841

presence of a subsurface ocean (eg Carr et al1998) have made Europa an exciting target forfuture space exploration The molecular oxidantsproduced by radiolysis and photolysis could bespecies needed for carbon-based biochemistry inEuroparsquos putative subsurface ocean Howeverthere has been no model for the formation trap-ping and transport of relevant oxidants

In this paper we have reviewed the space ob-servations and laboratory studies related to thechemistry of formation of oxidants in ice by theincident radiation Because the laboratory dataare incomplete we examined the rate equationsfor formation of oxygen and related species byradiolysis and photolysis We showed how suchequations can be simplified to the analytic modelthat was used for fitting the fluence and temper-ature dependence of the O2 yield produced bylow-energy electrons (Sieger et al 1998) In de-riving the analytic model from the more completerate equations the competition between forma-tion and destruction processes can be studied ascan the role of trapping sites escape from depthand temperature history These equations also al-lowed us to put incident photon electron and iondata on a similar footing for the first time

The model can be used to suggest measure-ments that are needed to describe the chemicalstate of the irradiated ice and to interpret new lab-oratory data We showed that if reactions are fastthe yield data could be consistent with a precur-sor being formed by two events as is the case forperoxide However we favor trapped O as a pre-cursor as initially suggested by Matich et al(1993) for production of excited O2 Such aspecies if present could be detected spectro-scopically as can trapped O2 through bands ac-tivated by lattice perturbations (Cooper et al2003a)

The available oxygen yield data were used tocompare G values for different incident particlesThese were then used to estimate the oxygen pro-duction in the radiation environment at EuropaThe lower bounds were consistent with the pro-duction rates obtained from a recent model of Eu-roparsquos atmosphere The porosity of the regolithwas also considered It not only affects the rela-tive sputtering yields for O2 and H2O but alsoprovides an enormously increased surface areaon which the ambient gas-phase oxygen can re-act with species in the ice That this might be oc-curring is consistent with the change in re-

flectance associated with surface aging Such re-actions could also convert ambient molecularoxygen into other more refractory species that re-main trapped on subduction of Europarsquos crust

Although radiolysis produces a stable chemi-cally altered surface in the laboratory the parti-cle flux to Europarsquos surface is many orders ofmagnitude lower Therefore annealing can be ef-fective at temperatures that are much lower thanthose typically found experimentally Belowabout 120K the annealing times appear to be long(eg Baragiola 2003) so that laboratory data canoften be directly applied to Europarsquos icy surfaceHowever account has to be taken of the syner-gism between the UV and the energetic plasmaas discussed earlier (Johnson and Quickenden1997)

Transient surface melting might initiate lifeprocesses but subduction of oxidants to the pu-tative ocean is likely to be required if biochem-istry is to occur at Europa (Chyba and Hand2001) Although the temperature versus depthinto Europarsquos regolith is not well known belowthe regolith the temperature and pressure appar-ently increase with depth At some depth (ap-proximately kilometers) either very warm ice oran ocean exists (Ruiz and Tejero 2000) Thereforechemically refractory oxidants are likely to be re-quired in order to survive subduction to thewarmer subsurface regions Trapped species ifdelivered downward can become mobile andwill segregate forming inclusions (Johnson andJesser 1997) Indeed peroxide appears to segre-gate in ice (Gurman et al 1967) and these inclu-sions may remain relatively stable until the per-oxide is dissolved in the putative ocean Thevolatiles seen in reflectance such as O2 SO2 andCO2 will also segregate But at the higher tem-peratures they could also diffuse through thebulk react or percolate into the regolith atmo-sphere Therefore the observation of the trappedvolatiles in Europarsquos surface ice is of interest pri-marily because it suggests that oxidants are pre-sent The formation of oxygen-rich molecules fol-lows the preferential loss of hydrogen Howeverthe surface is in a local steady state with im-planted H O and S from the plasma and the lossto space Since the plasma flux to the surface canbe variable this balance can also vary Indeedtemporal changes in reflectance have been re-ported (Domingue and Hendrix 2004)

Based on the models for formation of O2 ex-

JOHNSON ET AL842

amined in this paper additional experiments areneeded to determine the chemical state of irra-diated ice at low fluences Of immediate inter-est is the identification of stable molecular pre-cursors required for the formation of O2 Basedon the available laboratory data individual per-oxide molecules trapped in the ice do not appearto be the principal precursors but trapped O(Matich et al 1993) remains a viable candidateAlthough the data suggest that the local hydro-gen-bonding network is important in the pro-duction of O2 the effect of the formation tem-perature and radiation damage on the structureneed to be measured It is also important to de-scribe quantitatively the relationship betweenH2 loss and the formation of O2 and H2O2 overa range of temperatures and fluences Finallyone would like to determine the steady-statedensities of biochemically interesting moleculesformed by radiolysis in an ice containing sulfurand carbon at Europarsquos surface temperaturesand then determine their stability with increas-ing temperature and pressure Such experimentsstill need to be carried out Here we have pre-sented a model that brings together the variouslaboratory data and now can be used to considerthe formation trapping and transport of relevantoxidants at Europa

APPENDIX

Introduction

In this Appendix we examine rate equationsthat describe various aspects of the radiolysis orphotolysis of ice These equations need to be un-derstood if one is to be able to calculate the for-mation trapping and transport of oxidants Theemphasis is on the formation of molecular oxy-gen We integrate these equations over the escapedepth of O2 to give the simplified rate equationsdiscussed in the text in which number densitiesare replaced by column densities (eg Eqs 1 and2) In this way we can examine the relationshipbetween the rate constants and the effective pre-cursor formation and destruction cross sectionsextracted from data We first consider a precur-sor such as H2O2 that is formed from two exci-tation events This bears on the observed linear-ity of the oxygen yield versus fluence We thenconsider the equations for trapped versus a tran-

siently mobile species We use a reaction betweena mobile and trapped O as an example althoughO 1 OH can also give oxygen Finally we ex-amine the possible dependence of the O2 forma-tion process on the density of traps (defectsvoids and grain boundaries) as a function of tem-perature The results are discussed in the text

For clarity all possible radiolytic species arenot included in the discussion below Thereforepotentially important products such as HO2 andthe charged species (eg H3O1 the trapped elec-tron etc) are missing In each section we attemptto make a point related to production of O2 andother oxidants by using a subset of the productsrather than attempt a complete description withmany unknown rate constants

Fast versus slow processes

To describe the O2 production rate from a pre-cursor such as H2O2 we first use a simple but in-structive set of rate equations to describe the for-mation of H2O2 We then relate these to thesimplified rate equation for the precursor in thetext Eq 1 and show how the postulated precur-sor formation cross section sp is related to theH2O dissociation cross section sd by the reactionrates

Using the reactions

H2O 1 radiation R H 1 OH (sd)

H 1 H R H2 (k2H)

H 1 OH R H2O (kHOH)

OH 1 OH R H2O2 (k2OH)

the reaction pathways are described by

H2

Hqk2H

1 OHH2O 1 radiation mdashRsd H

OHH2O2

OH Rk2OH

kHOH

HH2O

OXIDANTS AT EUROPA 843

The corresponding rate equations in an irradiatedvolume are

dnOHdt 5 sd f n 2 2 k2OH nOH nOH

2 kHOH nH nOH (A1a)

dnHdt 5 sd f n 2 2 k2H nH nH

2 kHOH nH nOH (A2a)

dnH2O2dt 5 k2OH nOH nOH

2 destruction processes (A3a)

Here sd is the dissociation cross section f is theradiation flux and n is the molecular numberdensity of ice with nOH and nH the density of dis-sociation products OH and H respectively Therate constants k2OH k2H and kHOH describe thereactions above In this model the H2 formed isassumed to escape for temperatures above about20K We do not distinguish between trappedspecies and mobile species although in fact thatdistinction is important as discussed in the nextsection That is any OH produced at low tem-perature by one ion would likely be trapped priorto interacting with an OH produced by a subse-quent ion At higher temperatures this can breakdown as discussed in the text We first relate theabove set of equations to Eq 1 in the text and thenconsider precursor destruction processes

Equation 1 in the text is obtained by solvingthese equations together and then by integratingover a depth Dx Here Dx is the smaller of the pen-etration depth of the particle and the O2 escape(percolation) depth Assuming the molecular den-sity of ice n is constant with depth Dx n 5 N inEq 1 Comparing Eq 1 with Eq A3a the precur-sor formation rate sp f N is equivalent to [k2OH

nOH nOH] Dx Assuming that the processes in EqsA1a and A2a are fast then the steady state is read-ily obtained In addition if k2H k2OH there ispreferential loss of H2 at short times as suggestedby the laboratory data If steady state is reachedthe production of H2 is equal to that of H2O2 us-ing only the equations above Further one obtains

[k2OH nOH nOH] Dx R [fp sd] f N (A4)

with

fp 5 05[1 1 05 kHOH(k2H k2OH)05] (A5a)

Therefore the precursor formation cross sectionsp in Eq 1 is a fraction fp of the total dissocia-

tion cross sd Since laboratory data suggest theH2O2 fraction in an irradiated ice is small thenkHOH (k2OH k2H)05 which is likely givingfp R [(k2OH k2H)05kHOH]

Trapped versus mobile species

To examine the fluence dependence discussedin the text we note that trapped OH and freshlyproduced transiently mobile OH can be treatedas different species (eg Matich et al 1993)Therefore assuming no trapping of H the OHrate equation above would be replaced by twoequations one for the trapped species nOHt andthe other for the freshly produced mobile speciesnOH Assuming the reaction of two mobile OHunits is unlikely and that peroxide is formed fromthe interaction of a trapped OH with a subse-quently produced OH an expanded set of equa-tions is obtained

OH 1 trap R OHt (kOHt)

OH 1 OHt R H2O2 (kOHOHt)

H 1 OHt R H2O (kHOHt)

then the reaction pathways are described by

qH2

H2O 1 radiation mdashRsdH

qk2H

1 OHH

trap OH

OH RkOHt

OHt kOHOHtH2O2

kHOHkHOHt

H H2O HH2O

The resulting set of equations is

dnOHdt 5 sd f n 2 kHOH nH nOH

2 kOHt nOH nt 2 kOHOHt nOH nOHt (A1b)

dnOHtdt 5 kOHt nOH nt 2 kOHOHt nOH nOHt

2 kHOHt nH nOHt (A1b9)

dnHdt 5 sd f n 2 2 k2H nH nH 2

kHOH nH nOH 2 kHOHt nH nOHt (A2b)

JOHNSON ET AL844

dnH2O2dt 5 kOHOHt nOH nOHt

2 destruction processes (A3b)

where nt is the trap density in the irradiated iceAt low T we can assume nt is large so thatdntdt 0 simplifying the set of equations If theprecursor of interest is trapped O as discussed inthe text then a similar set of equations can be con-structed for O and Ot as in the example to followOf course a full description over a broad rangeof temperatures would require the considerationof the both trapped and mobile H O and OHthe rate of change of trap density and the vari-ous destruction processes

The above equations should be solved simul-taneously but for the incident particle fluxes typ-ically used in these studies the diffusion of mo-bile species to traps is relatively fast This is trueeven at low temperatures if there is a significantdensity of trapping sites This is usually the casefor samples formed at low T Therefore in a lowtemperature solid at low dose rates there is not asignificant steady-state background of diffusingOH There are trapped or freshly produced OHunits which rapidly react Averaging over thebombardment rate this can be approximated bywriting dnOHdt 0 and then separating nOH outin Eq A1b Assuming also that H diffuses rapidlyand either reacts or forms H2 we also writednHdt 0 on the average Therefore betweenimpacts only species trapped at defects pores orgrain surfaces are assumed to be present as sug-gested by the lack of a dependence on the inci-dent beam flux In this limit the rate of change oftrapped OH roughly simplifies to

dnOHtdt 5 2 k2H nH nH 2 2 kOHOHt nOH nOHt

It is seen that the production of peroxide dependson the loss of H by formation of H2 as discussedand at steady state the production rate for per-oxide equals the rate of formation of H2 in theseequations

Treating the mobile and trapped species sepa-rately we can further simplify the rate equationsand obtain the precursor model used in the textThat is integrating over the relevant depth Dxthe densities are replaced by column densitiesThe rate constants ki are then related to effectivecross sections si In this way cross sections aregiven as an interaction length for each freshlyproduced mobile radical [ki nj] R [si f] The

trapped OH column density NOHt and peroxideproduction rate can be estimated from ldquolinearrdquorate equations

dNOHtdt lt [sd f] N 2 2 [s2OH f] NOHt

2 [sHOH f] NOHt (A6a)

dNH2O2dt lt [s2OH f] NOHt 2 s NH2O2 (A6b)

Here sd is again the dissociation cross sectionand the destruction processes in Eq A3b are con-tained in s The second and third terms on theright in Eq A6a are the loss of OH due to the pro-duction of H2O2 and H2O In the integration ofEq A1b9 [2 kOH OHt nOH] becomes [2 s2OH f] and[kHOHt nOH] becomes [sHOH f] Therefore thedensity and the interaction length of mobile H arecontained in sHOH If H2O2 is indeed the pre-cursor then in Eq 2 in the text [sp f N] is be-comes [s2OH f NOHt] from Eq A6b Accountingfor precursor destruction which can give backtwo OH and treating H2 can both be included inthe integrated model via the si values

The simplified set of equations above allows usto examine the relative time (fluence) scales forprecursor formation (in this case H2O2) SolvingEqs A6a and A6b the column density of OH andthe concentration of precursors versus fluence cpcan be obtained Using the column of H2O2 pro-duced as the precursor column Np in Eq 2 thenO2 yield can be obtained The results can be com-pared with the results from Eqs1 and 2 in thetext The average precursor (H2O2) concentrationover the column depth of water molecules N is

cp 5 fp (sds)[1 2 exp(2s F)] 2 (ssr)

[1 2 exp(2sr F)] (A7)

with

fp 5 2s2OH(sr 2 s) (A5b)

In these equations sr 5 (2s2OH 1 sHOH) andssr 1

It is seen that the precursor concentration withtime cp depends on two rates the reaction ratesthat determine sr and the destruction rates in sAs in Eq 4a the yield can be written as

Y 5 sO2 cp N (A8)

If sr and s are comparable then the yield at small

OXIDANTS AT EUROPA 845

fluence is quadratic in F However if sr sthen the linear dependence on fluence is obtainedwith a small shift in the fluence as shown in Eq4c in the text Further the steady-state yield isnow

Y` (T) 5 sO2 [fp sd] Ns (A9)

That is sp R [fp sd] as in Eq A4 and the tem-perature dependence for the incident electrondata must be contained in [sO2 fp] or in the ratio[sO2 s2OH]sHOH]

For comparison with their precursor modelSieger et al (1998) examined the possibility of ob-taining the O2 precursor by considering reactionsfor a density of diffusing dissociation products asin Eqs A1 and A2 rather than a trapped productinteracting with a freshly formed dissociationproduct Based on the above the lowest fluence atwhich O2 is first detected experimentally sets alimit on the time scale for mobilendashmobile reactions

Effect of trap density

Equations A1b A1b9 A2b and A3b can be ap-proximately solved assuming that the dnOHdtdnHdt and dntdt are very small on the averageas discussed above In order to examine the roleof the trap density nt which can depend on thetemperature and method of formation and on theradiation history of the ice we use the slightlysimpler precursor model In this model the O pro-duced directly by dissociation traps forming aprecursor O2 is then formed when a freshly pro-duced O or other species reacts with a trappedoxygen atom Ot

H2O 1 radiation R H2 1 O (sd)

O 1 trap R Ot (kOt)

O 1 Ot R O2 (kOOt)

The reaction pathways for the oxygen species are

H2O 1 radiation mdashRsd9

trap OO2

kOtOt kOOt

H2 1 OOHt

kOR H(kHO t)

ROR

We assume for simplicity both H2 and O2 escapefrom depth Dx after formation We also allow mo-bile H to react with Ot (kHOt) removing trappedO and both the mobile H and O to be removedby other reactants R (eg OH or contaminants)This leads to a simplified set of equations not di-rectly involving OH or peroxide

dnOdt 5 sd9 f n 2 kOt nO nt 2 kOR nO nR

2 kOOt nO nOt (A1c)

dnOtdt 5 kOt nt nO 2 kOOt nO nOt

2 kHOt nH nOt (A1c9)

dnHdt 5 sd f n 2 kHOt nH nOt

2 kHR nH nR (A2c)

dnO2dt 5 kOOt nO nOt (A3c)

These equations of course act in parallel with theequations above for OH and mobile OH may re-act with trapped O to form O2 Here we focusonly on mobile O

Since the diffusion of H and O are fast and thedensity of traps is large the nH and nt averagedover the bombardment rate can be taken veryroughly as time independent Further if we as-sume the number of trapped O nOt is small com-pared with the number of sites at which H canreact then nH [sd f nkHR nR] For a low ra-diation flux f nH is small on the average as as-sumed Since the O2 yield YO2 is equal to[(dnO2dt) Dxf] with Dx the escape depth thenthe steady-state yield Y` can be obtained ana-lytically

[Y`][sd9 N] 5 1 1 (q 2 1) 2 [(q 2 1)2

1 8 q d]054 d (A10)

q 5 [kOt (nt 1 nR)] (kHOt nH)(sd9 f n) kOOt

5(sdsd9) [kOt (nt 1 nR)] kHOt[kOOt kHR nR]

d 5 05 [1 1 nt(nt 1 nR)]

We note that for nt large q can be large Assum-ing that q 1 and nt nR then Eq A10 be-comes

Y` R [sd9 N]q 5 (sd9 N) (sd9sd)

[kOOt kHR nR][kHOt kOt nt] (A5c)

JOHNSON ET AL846

Comparing with Y` in the simple precursormodel in Eq 4b then s R [(kHOt nH)f] R [sd nkHOt][kHR nR] In addition the product [sO2sp] R sd9 sd9 [kOOt n][kOt nt] That is the dis-sociation cross sections are modified by reactionrates

Quite remarkably it is seen in Eq A5c that thesteady-state yield Y` can depend inversely onthe density of traps nt in the temperature andflux regimes for which the approximations arerelevant That is if there are competing destruc-tion process for trapped O then at high trap den-sity a mobile O must find a trapped O before itbecomes trapped or before mobile H destroys theOt In this way the increasing O2 yield with in-creasing T can be related to the change in the trap-ping density nt with the formation temperatureof the ice sample As the temperature increasesand the density of traps in the bulk decreases thesurface and grain interfaces become the only trap-ping sites

ACKNOWLEDGMENTS

REJ acknowledges support by a Gledden Fel-lowship in Chemistry School of Biomedical andChemical Sciences at The University of WesternAustralia and by grants from the NSF Astron-omy Program and NASArsquos Planetary Geologyand Geophysics Program PDC acknowledgessupport by an Australian Postgraduate Award

ABBREVIATIONS

HST Hubble Space Telescope MJQ MatichndashJohnsonndashQuickenden NIMS Near Infrared Map-ping Spectrometer UV ultraviolet

REFERENCES

Ayotte P Smith RS Stevenson KP Dohnarsquolek ZKimmel GA and Kay BD (2001) Effect of porosityon the adsorption desorption trapping and release ofvolatile gases by amorphous solid water J Geophys Res106(E12) 33387ndash33392

Bahr DA Famaacute M and Baragiola RA (2001) Radiol-ysis of water ice in the outer solar system sputteringand trapping of radiation products J Geophys Res 10633285ndash33290

Baragiola RA (2003) Microporous amorphous water icefilms and astronomical implications In Water in Con-

fining Geometries edited by JP Devlin and B Buch El-sevier New York in press

Baragiola RA and Bahr DA (1998) Laboratory studiesof the optical properties and stability of oxygen onGanymede J Geophys Res 103 25865ndash25872

Baragiola RA Atteberry CL and Bahr DA (1999) Ori-gin of solid oxygen on Ganymede Reply to commentby RE Johnson on ldquoLaboratory studies of the opticalproperties and stability of oxygen on Ganymederdquo JGeophys Res E104 14183ndash14187

Baragiola RA Vidal RA Svendsen W Schou J ShiM Bahr DA and Atteberrry CL (2003) Sputteringof water ice Nucl Instrum Methods Phys Res B 209294ndash303

Bar-Nun A Herman G Laufer D and Rappaport ML(1985) Trapping and release of gases by water ice andimplications for icy bodies Icarus 63 317ndash332

Baverstock KF and Burns WG (1976) Primary pro-duction of oxygen from irradiated water as an expla-nation for decreased radiobiological oxygen enhance-ment at high LET Nature 260 316ndash318

Benit J and Brown WL (1990) Electronic sputtering ofoxygen and water molecules from thin films of waterice bombarded by MeV Ne1 ions Nucl Instrum Meth-ods Phys Rev B 46 448ndash451

Benit J Bibring J-P della-Negra S LeBeyec YMendenhall M Rocard F and Standing K (1987)Erosion of ices by ion irradiation Nucl Instrum Meth-ods Phys Rev B 1920 838ndash842

Brown WL Augustyniak WM Lanzerotti LJ John-son RE and Evatt R (1980) Linear and nonlinearprocesses in the erosion of H2O ice by fast light ionsPhys Rev Lett 43 1632ndash1635

Brown WL Augustyniak W M Simmons E Mar-cantonio KJ Lanzerotti LJ Johnson RE BoringJW Reimann CT Foti G and Pirronello V (1982)Erosion and molecular formation in condensed gasfilms by electronic energy loss of fast ions Nucl In-strum Methods B 1 307ndash314

Brown WL Augustyniak WM Marcantonio KJ Sim-mons EN Boring JW Johnson RE and ReimannCT (1984) Electronic sputtering of low-temperaturemolecular solids Nucl Instrum Methods B 1 307

Buratti B (1995) Photometry and surface structure of theicy Galilean satellites J Geophys Res 100(E9) 19061ndash19066

Buratti B and Veverka J (1983) Voyager photometry ofEuropa Icarus 55 93ndash110

Calvin WM and Spencer JR (1997) Latitudinal distri-bution of O2 on Ganymede observations with the Hub-ble Space Telescope Icarus 130 505ndash516

Calvin WM Clark RN Brown RH and Spencer JA(1995) Spectra of the icy Galilean satellites from 02 to5 mm a compilation new observations and a recentsummary J Geophys Res 100 19041ndash19048

Calvin WM Johnson RE and Spencer JA (1996) O2

on Ganymede spectral characteristics and plasma for-mation mechanisms Geophys Res Lett 23 673ndash676

Calvin WM Aninich VG and Brown RH (2004) Oxy-gen temperature and phase effects Icarus (in press)

OXIDANTS AT EUROPA 847

It may be the case that a particular precursoris not required as mentioned earlier Rather af-ter a sufficient level of oxidation a newly pro-duced O or OH can react in the oxygen-rich lat-tice producing O2 in a pseudo-first-order processThis can be tested experimentally Indeedtrapped O and peroxide may be changed into oneanother as in the matrix isolation experiments(Khriachtchev et al 2000) Therefore their rela-tive concentrations versus temperature and doseshould be studied

In the model considered here it is seen that theparameters in Y` must account for the strong tem-perature dependence Since the number of bulktrapping sites decreases with increasing T as theice sample is annealed one might expect that theprecursor concentration would decrease We sug-gest that the observed temperature dependenceis determined by the competition between theavailability of trapping sites and mean interactiondistances of a mobile species such as O and HThat is the O2 yield varies inversely with the den-sity of traps over a range of temperatures In thesimple model above this effect is included indi-rectly as shown using the more detailed equa-tions in the Appendix

EUROPA

Oxygen and peroxide production at Europa G values

The dependence of the yield on the thicknessof the ice sample for highly penetrating ions sug-gests that the O2 yield is roughly related to theamount of energy deposited in the sample There-fore the amount of O2 produced and lost fromice in steady state is often given as a G value thenumber of molecules produced per 100 eV de-posited in Table 2 In applying these G values tothe planetary environment care must be takensince the yields are temperature fluence and ex-citation density dependent They also are affectedby the presence of other proton scavengersknown to be present in Europarsquos ice

The G values for peroxide in Table 2 are thosefor production at low fluence but the G valuesfor O2 are for the production and escape of gas-phase oxygen at saturation Therefore they de-pend on the formation and destruction cross sec-tions in the model above Such G values can beused to describe the steady-state source of the am-

bient O2 atmosphere at Europa but are not help-ful in determining the ability to produce and trapO2 at depth into the icy surface Over the timescales of the experiments some O2 is seen to re-main trapped and is not destroyed by radicalsTypically there are a number of trapping sitesand recent estimates of the activation energy foran O2 molecule trapped in a substitutional site(Hori and Hondoh 2002) suggest that it can betrapped for times comparable to the time re-quired for the conversion of amorphous ice intoIh (Ayotte et al 2001) The stability of O2 inclu-sions has not been studied

To understand the relationship between the Gvalues for O2 in Table 2 we note that fast lightions and electrons lose their energy predomi-nantly by producing ionizations and excitationsThe ionization events produce secondary elec-trons that can result in further ionizations Thenumber of ionization events Ni produced by afast light ion stopping in a water ice is Ni 5EoW where Eo is the initial energy and W is theaverage energy expended per ionization event27 eV in water ice Therefore radiolysis by fastions is often described as the net effect of theshower of electrons set in motion by the incidention In this manner the G values for productionof O2 from ice by energetic light ions and elec-trons can be roughly estimated as the sum of theeffects of all the secondary electrons using theyield Ye for 10ndash100 eV electrons That is if weignore the percolation depth N we can write thenumber of O2 produced as the number of ioniza-tion events times Ye which already has in it thedestruction processes That is G NiYeEo YeW 0006 O2100 eV

This estimate of G is seen to be an order of mag-nitude larger than the keV proton value for O2 inTable 2 but is much smaller than the G value forenergetic Ne1 These differences from the simpleestimate are consistent with the observation thatthe yield in the saturation regime for penetratingions varies nonlinearly with excitation densityThe difference from the keV proton G value islikely due to the inability of O2 to escape fromdepth before destruction On the other hand thelarger G value seen using MeV Ne1 is likely dueto the high density of excitation This can resultin more efficient production of O2 increased pro-duction of damage sites for forming O2 andorincreased transport (percolation) along the tran-siently heated track That is the ability of theformed O2 to percolate to the surface during ion

OXIDANTS AT EUROPA 839

bombardment plays an essential role in deter-mining the measured yields as discussed aboveThis effect needs to be evaluated for a planetarysurface subject to particle fluxes that are many or-ders of magnitude smaller than those used in thelaboratory

In the absence of O2 yields for all of the rele-vant ions and energies an estimate of the totalsupply of O2 can be obtained using the G valuesin Table 2 For the energetic oxygen and sulfurions we use G values for neon and argon ionsThese are lower bounds as the data were obtainedat temperatures much lower than the surface tem-peratures at Europa (80ndash110K) For the fast pen-etrating electrons and protons we use the valuefor 200 keV protons in Table 2 For the heavy ionswe used the Ne1 and Ar1 values For compari-son we also give an estimate using the G valuesobtained from the low energy electron data andthe total ionization as discussed above

The G values are combined in Table 3 with theelectron and ion energy flux at Europa Althoughthere is some deflection of the flow due to in-duced fields the net energy flux to the surface isnot reduced much from these values (Paranicaset al 2002) In fact locally produced pick-up ionswhich are formed from the ejecta and acceleratedto the surface can contribute (Ip 1996) Furtheralthough G values are typically independent ofthe incident angle the loss of O2 is a surfaceprocess and is enhanced with angle That is thenumber of O2 produced within the escape depthincreases with increasing angle to the normal(steeper than the inverse cosine) This enhance-ment is 25ndash4 (Jurac et al 2001) Ignoring suchcorrections the total production rate calculated

(23ndash62 3 109 O2cm2s) is a lower bound to theactual production rate from the icy regions of Eu-roparsquos surface Recent modeling of the O2 atmo-sphere suggest that such a source rate is required(Shematovich and Johnson 2001)

Earlier estimates of the O2 source rate were in-correct because O2 production was assumed to bea fixed fraction of the H2O sputtering yield Atthe ambient temperature an ejected H2O sticks ef-ficiently (Smith and Kay 1997) to the surface ofneighboring grains in the porous regolith reduc-ing the sputtering yield for H2O (Johnson 1989)This reduction does not apply to the O2 fractionof the yield enhancing its relative importance toH2O Here we assume that all of the gas-phaseO2 produced can contribute to the ambient at-mosphere even if it is produced at a grain sur-face that is at some depth into the porous regolith

Finally Europarsquos surface is not pure ice every-where That is much of the trailing hemisphereis a hydrated sulfuric acid containing smallamounts of sulfur and SO2 and the leading hemi-sphere contains low levels of carbon CO2 andcarbonates Sodium and potassium atoms aresputtered from the icy regions (eg Leblanc et al2002) O2 production from surfaces containingthese species has not been measured C and S areboth proton scavengers and the other species canbe sources of O as discussed so that the O2 yieldfrom such regions could be enhanced Howeverthere can also be competing processes that mightreduce these rates In fact recent HST observa-tions suggest that the O2 might not be produceduniformly across Europarsquos surface (McGrath et al2000) The analysis of these observations is pre-liminary and is not simple since the observed

JOHNSON ET AL840

TABLE 3 APPLICATION TO EUROPA

OxygenImplantation Energy flux production

Species 109(cm2 s)a 109 keV(cm2 s)(1) 109(cm2 s)

H (10 keV) 0015(1) 12 0084 (072)bO (10 keV) 0015(1) 18 054cS (10 keV) 0009(1) 30 090cElectrons 62 043 (37)bPlasma (10 keV) (137 amu) 13(2) 10 03c

Total 79 23 (62)UV photons (64 eV) 46 01

aReferences are as follows 1Cooper et al (2001) [similar results in Paranicas et al (2002)] 2Mauk et al (1996)bUsing G for 200 keV protons (upper bounds in brackets are based on the low-energy electrons data as a model for

the secondary electrons produced)cUsing G for 15 MeV Ne1 at 7K a rough lower bound

emissions depend on the local plasma propertiesHowever the efficiency of O2 production in con-taminated ices needs to be studied

Porous regolith O2 interactions on grain surfaces

Although the O2 atmosphere is principally pro-duced in the near surface layers and is eventuallylost to space it also permeates Europarsquos porousregolith The porosity p (void space) of the re-golith is uncertain (Buratti 1995) It has been es-timated to be the most compact surface in theOuter Solar System p 025 on the trailing hemi-sphere and p 079 on the leading hemisphere(Buratti and Veverka 1983) and the most poroussurface p 096 globally (Domingue et al 1991Domingue and Hapke 1992) These large differ-ences are due to the uncertainties in the amountof backscattered light at small phase angles andon the photon absorption properties and grainsizes The more compact surface is likely correctGrain sizes rg of the order of 100 mm have beensuggested (Hansen and McCord 1999 2000Geissler et al 1998) Europarsquos regolith differs sig-nificantly from the lunar regolith which is asand-like surface of individual grains Analysis ofthe thermal inertia of Europarsquos surface (Grundyet al 2001) suggests that the ice grains are sin-tered forming a fairy castle-like structure Sin-tering can occur thermally or due to the incidentradiation In addition ices doped with acids andalkalis in the laboratory can be powdery solidswith the impurities primarily found at grain in-terfaces

The porous regolith affects the above estimatesof the oxygen production in three ways First forisotropic bombardment the mean distance d intothe surface at which an incident ion or photonstrikes a grain surface is d (89)rg(1 2 p) Us-ing the above value for rg then d 02 m if theporosity is very large p 96 Therefore de-sorbed molecules can come from a considerabledepth and will interact with grain surfaces Evenfor much lower porosities the effective sputteringyields can be significantly modified (Johnson1989) As discussed above the yields for ejectedH2O which sticks onto neighboring grains at theambient temperatures are reduced but those forejected O2 are not This enhances the relative im-portance of the O2 yields

Finally the surfaces and interfaces of the sin-tered grains in Europarsquos porous surface are placeswhere low temperature chemistry involving the

ambient oxygen can occur at depths that are wellbelow the nominal penetration depth of the inci-dent radiation The relative importance of suchchemistry is roughly indicated by the ratio of thetotal surface area in the regolith to the area of Eu-roparsquos visible surface Using the porosity andgrain size above and a regolith depth dr of theorder of a meter then this ratio is large 3(1 2p)drrg which is 103ndash2 3 104 for average porosi-ties of 96 and 25 respectively Therefore thegas-phase O2 molecules below the visible surfaceinteract frequently with the grain surfaces In thisway a very large volume of the subsurface couldbe altered chemically even if it is not directly ex-posed to the radiation

From the overlying structures seen in the high-resolution images of Europarsquos surface by theGalileo spacecraft (eg Prockter and Pappalardo2000) it is clear that the youngest surfaces are red-der The reddening in the visible has been attrib-uted to the presence of sulfur in the ice (McEwen1986 Johnson et al 1988 Calvin et al 1995 Carl-son et al 1999b) The amount of reddening in thespectra of the surface material decreases with ageand the surface materials appear to becomebrighter with age This might occur by the slowaccumulation of a frost over-layer or by radioly-sis (Carlson et al 2002 Johnson et al 2003) Ra-diolysis preferentially converts the sulfur in theice to sulfur dioxide or sulfuric acid both ofwhich have flat spectra in the visible In regionswhere the radiation does not directly penetratethe O2 permeating the regolith can oxidize thesulfur at the surfaces of the icy grains and at graininterfaces Similarly the carbon seen in the ice onEuroparsquos leading hemisphere (eg Carlson et al2004 Johnson et al 2003) can be oxidized by ra-diolysis in the near surface layers or by reactionson the grain surfaces at depth in the regolith bythe ambient O2 The meteoroid bombardmentthat produces the regolith also exposes fresh sur-faces for oxidation and can cause the collapse ofpores leading to the trapping of oxygen In thismanner both refractory and volatile oxidants canbe formed and trapped in the subsurface ice forpotential delivery to Europarsquos subsurface ocean

SUMMARY

The presence of a thin oxygen atmosphere andoxidants trapped in Europarsquos surface ice (egJohnson et al 2003) combined with the possible

OXIDANTS AT EUROPA 841

presence of a subsurface ocean (eg Carr et al1998) have made Europa an exciting target forfuture space exploration The molecular oxidantsproduced by radiolysis and photolysis could bespecies needed for carbon-based biochemistry inEuroparsquos putative subsurface ocean Howeverthere has been no model for the formation trap-ping and transport of relevant oxidants

In this paper we have reviewed the space ob-servations and laboratory studies related to thechemistry of formation of oxidants in ice by theincident radiation Because the laboratory dataare incomplete we examined the rate equationsfor formation of oxygen and related species byradiolysis and photolysis We showed how suchequations can be simplified to the analytic modelthat was used for fitting the fluence and temper-ature dependence of the O2 yield produced bylow-energy electrons (Sieger et al 1998) In de-riving the analytic model from the more completerate equations the competition between forma-tion and destruction processes can be studied ascan the role of trapping sites escape from depthand temperature history These equations also al-lowed us to put incident photon electron and iondata on a similar footing for the first time

The model can be used to suggest measure-ments that are needed to describe the chemicalstate of the irradiated ice and to interpret new lab-oratory data We showed that if reactions are fastthe yield data could be consistent with a precur-sor being formed by two events as is the case forperoxide However we favor trapped O as a pre-cursor as initially suggested by Matich et al(1993) for production of excited O2 Such aspecies if present could be detected spectro-scopically as can trapped O2 through bands ac-tivated by lattice perturbations (Cooper et al2003a)

The available oxygen yield data were used tocompare G values for different incident particlesThese were then used to estimate the oxygen pro-duction in the radiation environment at EuropaThe lower bounds were consistent with the pro-duction rates obtained from a recent model of Eu-roparsquos atmosphere The porosity of the regolithwas also considered It not only affects the rela-tive sputtering yields for O2 and H2O but alsoprovides an enormously increased surface areaon which the ambient gas-phase oxygen can re-act with species in the ice That this might be oc-curring is consistent with the change in re-

flectance associated with surface aging Such re-actions could also convert ambient molecularoxygen into other more refractory species that re-main trapped on subduction of Europarsquos crust

Although radiolysis produces a stable chemi-cally altered surface in the laboratory the parti-cle flux to Europarsquos surface is many orders ofmagnitude lower Therefore annealing can be ef-fective at temperatures that are much lower thanthose typically found experimentally Belowabout 120K the annealing times appear to be long(eg Baragiola 2003) so that laboratory data canoften be directly applied to Europarsquos icy surfaceHowever account has to be taken of the syner-gism between the UV and the energetic plasmaas discussed earlier (Johnson and Quickenden1997)

Transient surface melting might initiate lifeprocesses but subduction of oxidants to the pu-tative ocean is likely to be required if biochem-istry is to occur at Europa (Chyba and Hand2001) Although the temperature versus depthinto Europarsquos regolith is not well known belowthe regolith the temperature and pressure appar-ently increase with depth At some depth (ap-proximately kilometers) either very warm ice oran ocean exists (Ruiz and Tejero 2000) Thereforechemically refractory oxidants are likely to be re-quired in order to survive subduction to thewarmer subsurface regions Trapped species ifdelivered downward can become mobile andwill segregate forming inclusions (Johnson andJesser 1997) Indeed peroxide appears to segre-gate in ice (Gurman et al 1967) and these inclu-sions may remain relatively stable until the per-oxide is dissolved in the putative ocean Thevolatiles seen in reflectance such as O2 SO2 andCO2 will also segregate But at the higher tem-peratures they could also diffuse through thebulk react or percolate into the regolith atmo-sphere Therefore the observation of the trappedvolatiles in Europarsquos surface ice is of interest pri-marily because it suggests that oxidants are pre-sent The formation of oxygen-rich molecules fol-lows the preferential loss of hydrogen Howeverthe surface is in a local steady state with im-planted H O and S from the plasma and the lossto space Since the plasma flux to the surface canbe variable this balance can also vary Indeedtemporal changes in reflectance have been re-ported (Domingue and Hendrix 2004)

Based on the models for formation of O2 ex-

JOHNSON ET AL842

amined in this paper additional experiments areneeded to determine the chemical state of irra-diated ice at low fluences Of immediate inter-est is the identification of stable molecular pre-cursors required for the formation of O2 Basedon the available laboratory data individual per-oxide molecules trapped in the ice do not appearto be the principal precursors but trapped O(Matich et al 1993) remains a viable candidateAlthough the data suggest that the local hydro-gen-bonding network is important in the pro-duction of O2 the effect of the formation tem-perature and radiation damage on the structureneed to be measured It is also important to de-scribe quantitatively the relationship betweenH2 loss and the formation of O2 and H2O2 overa range of temperatures and fluences Finallyone would like to determine the steady-statedensities of biochemically interesting moleculesformed by radiolysis in an ice containing sulfurand carbon at Europarsquos surface temperaturesand then determine their stability with increas-ing temperature and pressure Such experimentsstill need to be carried out Here we have pre-sented a model that brings together the variouslaboratory data and now can be used to considerthe formation trapping and transport of relevantoxidants at Europa

APPENDIX

Introduction

In this Appendix we examine rate equationsthat describe various aspects of the radiolysis orphotolysis of ice These equations need to be un-derstood if one is to be able to calculate the for-mation trapping and transport of oxidants Theemphasis is on the formation of molecular oxy-gen We integrate these equations over the escapedepth of O2 to give the simplified rate equationsdiscussed in the text in which number densitiesare replaced by column densities (eg Eqs 1 and2) In this way we can examine the relationshipbetween the rate constants and the effective pre-cursor formation and destruction cross sectionsextracted from data We first consider a precur-sor such as H2O2 that is formed from two exci-tation events This bears on the observed linear-ity of the oxygen yield versus fluence We thenconsider the equations for trapped versus a tran-

siently mobile species We use a reaction betweena mobile and trapped O as an example althoughO 1 OH can also give oxygen Finally we ex-amine the possible dependence of the O2 forma-tion process on the density of traps (defectsvoids and grain boundaries) as a function of tem-perature The results are discussed in the text

For clarity all possible radiolytic species arenot included in the discussion below Thereforepotentially important products such as HO2 andthe charged species (eg H3O1 the trapped elec-tron etc) are missing In each section we attemptto make a point related to production of O2 andother oxidants by using a subset of the productsrather than attempt a complete description withmany unknown rate constants

Fast versus slow processes

To describe the O2 production rate from a pre-cursor such as H2O2 we first use a simple but in-structive set of rate equations to describe the for-mation of H2O2 We then relate these to thesimplified rate equation for the precursor in thetext Eq 1 and show how the postulated precur-sor formation cross section sp is related to theH2O dissociation cross section sd by the reactionrates

Using the reactions

H2O 1 radiation R H 1 OH (sd)

H 1 H R H2 (k2H)

H 1 OH R H2O (kHOH)

OH 1 OH R H2O2 (k2OH)

the reaction pathways are described by

H2

Hqk2H

1 OHH2O 1 radiation mdashRsd H

OHH2O2

OH Rk2OH

kHOH

HH2O

OXIDANTS AT EUROPA 843

The corresponding rate equations in an irradiatedvolume are

dnOHdt 5 sd f n 2 2 k2OH nOH nOH

2 kHOH nH nOH (A1a)

dnHdt 5 sd f n 2 2 k2H nH nH

2 kHOH nH nOH (A2a)

dnH2O2dt 5 k2OH nOH nOH

2 destruction processes (A3a)

Here sd is the dissociation cross section f is theradiation flux and n is the molecular numberdensity of ice with nOH and nH the density of dis-sociation products OH and H respectively Therate constants k2OH k2H and kHOH describe thereactions above In this model the H2 formed isassumed to escape for temperatures above about20K We do not distinguish between trappedspecies and mobile species although in fact thatdistinction is important as discussed in the nextsection That is any OH produced at low tem-perature by one ion would likely be trapped priorto interacting with an OH produced by a subse-quent ion At higher temperatures this can breakdown as discussed in the text We first relate theabove set of equations to Eq 1 in the text and thenconsider precursor destruction processes

Equation 1 in the text is obtained by solvingthese equations together and then by integratingover a depth Dx Here Dx is the smaller of the pen-etration depth of the particle and the O2 escape(percolation) depth Assuming the molecular den-sity of ice n is constant with depth Dx n 5 N inEq 1 Comparing Eq 1 with Eq A3a the precur-sor formation rate sp f N is equivalent to [k2OH

nOH nOH] Dx Assuming that the processes in EqsA1a and A2a are fast then the steady state is read-ily obtained In addition if k2H k2OH there ispreferential loss of H2 at short times as suggestedby the laboratory data If steady state is reachedthe production of H2 is equal to that of H2O2 us-ing only the equations above Further one obtains

[k2OH nOH nOH] Dx R [fp sd] f N (A4)

with

fp 5 05[1 1 05 kHOH(k2H k2OH)05] (A5a)

Therefore the precursor formation cross sectionsp in Eq 1 is a fraction fp of the total dissocia-

tion cross sd Since laboratory data suggest theH2O2 fraction in an irradiated ice is small thenkHOH (k2OH k2H)05 which is likely givingfp R [(k2OH k2H)05kHOH]

Trapped versus mobile species

To examine the fluence dependence discussedin the text we note that trapped OH and freshlyproduced transiently mobile OH can be treatedas different species (eg Matich et al 1993)Therefore assuming no trapping of H the OHrate equation above would be replaced by twoequations one for the trapped species nOHt andthe other for the freshly produced mobile speciesnOH Assuming the reaction of two mobile OHunits is unlikely and that peroxide is formed fromthe interaction of a trapped OH with a subse-quently produced OH an expanded set of equa-tions is obtained

OH 1 trap R OHt (kOHt)

OH 1 OHt R H2O2 (kOHOHt)

H 1 OHt R H2O (kHOHt)

then the reaction pathways are described by

qH2

H2O 1 radiation mdashRsdH

qk2H

1 OHH

trap OH

OH RkOHt

OHt kOHOHtH2O2

kHOHkHOHt

H H2O HH2O

The resulting set of equations is

dnOHdt 5 sd f n 2 kHOH nH nOH

2 kOHt nOH nt 2 kOHOHt nOH nOHt (A1b)

dnOHtdt 5 kOHt nOH nt 2 kOHOHt nOH nOHt

2 kHOHt nH nOHt (A1b9)

dnHdt 5 sd f n 2 2 k2H nH nH 2

kHOH nH nOH 2 kHOHt nH nOHt (A2b)

JOHNSON ET AL844

dnH2O2dt 5 kOHOHt nOH nOHt

2 destruction processes (A3b)

where nt is the trap density in the irradiated iceAt low T we can assume nt is large so thatdntdt 0 simplifying the set of equations If theprecursor of interest is trapped O as discussed inthe text then a similar set of equations can be con-structed for O and Ot as in the example to followOf course a full description over a broad rangeof temperatures would require the considerationof the both trapped and mobile H O and OHthe rate of change of trap density and the vari-ous destruction processes

The above equations should be solved simul-taneously but for the incident particle fluxes typ-ically used in these studies the diffusion of mo-bile species to traps is relatively fast This is trueeven at low temperatures if there is a significantdensity of trapping sites This is usually the casefor samples formed at low T Therefore in a lowtemperature solid at low dose rates there is not asignificant steady-state background of diffusingOH There are trapped or freshly produced OHunits which rapidly react Averaging over thebombardment rate this can be approximated bywriting dnOHdt 0 and then separating nOH outin Eq A1b Assuming also that H diffuses rapidlyand either reacts or forms H2 we also writednHdt 0 on the average Therefore betweenimpacts only species trapped at defects pores orgrain surfaces are assumed to be present as sug-gested by the lack of a dependence on the inci-dent beam flux In this limit the rate of change oftrapped OH roughly simplifies to

dnOHtdt 5 2 k2H nH nH 2 2 kOHOHt nOH nOHt

It is seen that the production of peroxide dependson the loss of H by formation of H2 as discussedand at steady state the production rate for per-oxide equals the rate of formation of H2 in theseequations

Treating the mobile and trapped species sepa-rately we can further simplify the rate equationsand obtain the precursor model used in the textThat is integrating over the relevant depth Dxthe densities are replaced by column densitiesThe rate constants ki are then related to effectivecross sections si In this way cross sections aregiven as an interaction length for each freshlyproduced mobile radical [ki nj] R [si f] The

trapped OH column density NOHt and peroxideproduction rate can be estimated from ldquolinearrdquorate equations

dNOHtdt lt [sd f] N 2 2 [s2OH f] NOHt

2 [sHOH f] NOHt (A6a)

dNH2O2dt lt [s2OH f] NOHt 2 s NH2O2 (A6b)

Here sd is again the dissociation cross sectionand the destruction processes in Eq A3b are con-tained in s The second and third terms on theright in Eq A6a are the loss of OH due to the pro-duction of H2O2 and H2O In the integration ofEq A1b9 [2 kOH OHt nOH] becomes [2 s2OH f] and[kHOHt nOH] becomes [sHOH f] Therefore thedensity and the interaction length of mobile H arecontained in sHOH If H2O2 is indeed the pre-cursor then in Eq 2 in the text [sp f N] is be-comes [s2OH f NOHt] from Eq A6b Accountingfor precursor destruction which can give backtwo OH and treating H2 can both be included inthe integrated model via the si values

The simplified set of equations above allows usto examine the relative time (fluence) scales forprecursor formation (in this case H2O2) SolvingEqs A6a and A6b the column density of OH andthe concentration of precursors versus fluence cpcan be obtained Using the column of H2O2 pro-duced as the precursor column Np in Eq 2 thenO2 yield can be obtained The results can be com-pared with the results from Eqs1 and 2 in thetext The average precursor (H2O2) concentrationover the column depth of water molecules N is

cp 5 fp (sds)[1 2 exp(2s F)] 2 (ssr)

[1 2 exp(2sr F)] (A7)

with

fp 5 2s2OH(sr 2 s) (A5b)

In these equations sr 5 (2s2OH 1 sHOH) andssr 1

It is seen that the precursor concentration withtime cp depends on two rates the reaction ratesthat determine sr and the destruction rates in sAs in Eq 4a the yield can be written as

Y 5 sO2 cp N (A8)

If sr and s are comparable then the yield at small

OXIDANTS AT EUROPA 845

fluence is quadratic in F However if sr sthen the linear dependence on fluence is obtainedwith a small shift in the fluence as shown in Eq4c in the text Further the steady-state yield isnow

Y` (T) 5 sO2 [fp sd] Ns (A9)

That is sp R [fp sd] as in Eq A4 and the tem-perature dependence for the incident electrondata must be contained in [sO2 fp] or in the ratio[sO2 s2OH]sHOH]

For comparison with their precursor modelSieger et al (1998) examined the possibility of ob-taining the O2 precursor by considering reactionsfor a density of diffusing dissociation products asin Eqs A1 and A2 rather than a trapped productinteracting with a freshly formed dissociationproduct Based on the above the lowest fluence atwhich O2 is first detected experimentally sets alimit on the time scale for mobilendashmobile reactions

Effect of trap density

Equations A1b A1b9 A2b and A3b can be ap-proximately solved assuming that the dnOHdtdnHdt and dntdt are very small on the averageas discussed above In order to examine the roleof the trap density nt which can depend on thetemperature and method of formation and on theradiation history of the ice we use the slightlysimpler precursor model In this model the O pro-duced directly by dissociation traps forming aprecursor O2 is then formed when a freshly pro-duced O or other species reacts with a trappedoxygen atom Ot

H2O 1 radiation R H2 1 O (sd)

O 1 trap R Ot (kOt)

O 1 Ot R O2 (kOOt)

The reaction pathways for the oxygen species are

H2O 1 radiation mdashRsd9

trap OO2

kOtOt kOOt

H2 1 OOHt

kOR H(kHO t)

ROR

We assume for simplicity both H2 and O2 escapefrom depth Dx after formation We also allow mo-bile H to react with Ot (kHOt) removing trappedO and both the mobile H and O to be removedby other reactants R (eg OH or contaminants)This leads to a simplified set of equations not di-rectly involving OH or peroxide

dnOdt 5 sd9 f n 2 kOt nO nt 2 kOR nO nR

2 kOOt nO nOt (A1c)

dnOtdt 5 kOt nt nO 2 kOOt nO nOt

2 kHOt nH nOt (A1c9)

dnHdt 5 sd f n 2 kHOt nH nOt

2 kHR nH nR (A2c)

dnO2dt 5 kOOt nO nOt (A3c)

These equations of course act in parallel with theequations above for OH and mobile OH may re-act with trapped O to form O2 Here we focusonly on mobile O

Since the diffusion of H and O are fast and thedensity of traps is large the nH and nt averagedover the bombardment rate can be taken veryroughly as time independent Further if we as-sume the number of trapped O nOt is small com-pared with the number of sites at which H canreact then nH [sd f nkHR nR] For a low ra-diation flux f nH is small on the average as as-sumed Since the O2 yield YO2 is equal to[(dnO2dt) Dxf] with Dx the escape depth thenthe steady-state yield Y` can be obtained ana-lytically

[Y`][sd9 N] 5 1 1 (q 2 1) 2 [(q 2 1)2

1 8 q d]054 d (A10)

q 5 [kOt (nt 1 nR)] (kHOt nH)(sd9 f n) kOOt

5(sdsd9) [kOt (nt 1 nR)] kHOt[kOOt kHR nR]

d 5 05 [1 1 nt(nt 1 nR)]

We note that for nt large q can be large Assum-ing that q 1 and nt nR then Eq A10 be-comes

Y` R [sd9 N]q 5 (sd9 N) (sd9sd)

[kOOt kHR nR][kHOt kOt nt] (A5c)

JOHNSON ET AL846

Comparing with Y` in the simple precursormodel in Eq 4b then s R [(kHOt nH)f] R [sd nkHOt][kHR nR] In addition the product [sO2sp] R sd9 sd9 [kOOt n][kOt nt] That is the dis-sociation cross sections are modified by reactionrates

Quite remarkably it is seen in Eq A5c that thesteady-state yield Y` can depend inversely onthe density of traps nt in the temperature andflux regimes for which the approximations arerelevant That is if there are competing destruc-tion process for trapped O then at high trap den-sity a mobile O must find a trapped O before itbecomes trapped or before mobile H destroys theOt In this way the increasing O2 yield with in-creasing T can be related to the change in the trap-ping density nt with the formation temperatureof the ice sample As the temperature increasesand the density of traps in the bulk decreases thesurface and grain interfaces become the only trap-ping sites

ACKNOWLEDGMENTS

REJ acknowledges support by a Gledden Fel-lowship in Chemistry School of Biomedical andChemical Sciences at The University of WesternAustralia and by grants from the NSF Astron-omy Program and NASArsquos Planetary Geologyand Geophysics Program PDC acknowledgessupport by an Australian Postgraduate Award

ABBREVIATIONS

HST Hubble Space Telescope MJQ MatichndashJohnsonndashQuickenden NIMS Near Infrared Map-ping Spectrometer UV ultraviolet

REFERENCES

Ayotte P Smith RS Stevenson KP Dohnarsquolek ZKimmel GA and Kay BD (2001) Effect of porosityon the adsorption desorption trapping and release ofvolatile gases by amorphous solid water J Geophys Res106(E12) 33387ndash33392

Bahr DA Famaacute M and Baragiola RA (2001) Radiol-ysis of water ice in the outer solar system sputteringand trapping of radiation products J Geophys Res 10633285ndash33290

Baragiola RA (2003) Microporous amorphous water icefilms and astronomical implications In Water in Con-

fining Geometries edited by JP Devlin and B Buch El-sevier New York in press

Baragiola RA and Bahr DA (1998) Laboratory studiesof the optical properties and stability of oxygen onGanymede J Geophys Res 103 25865ndash25872

Baragiola RA Atteberry CL and Bahr DA (1999) Ori-gin of solid oxygen on Ganymede Reply to commentby RE Johnson on ldquoLaboratory studies of the opticalproperties and stability of oxygen on Ganymederdquo JGeophys Res E104 14183ndash14187

Baragiola RA Vidal RA Svendsen W Schou J ShiM Bahr DA and Atteberrry CL (2003) Sputteringof water ice Nucl Instrum Methods Phys Res B 209294ndash303

Bar-Nun A Herman G Laufer D and Rappaport ML(1985) Trapping and release of gases by water ice andimplications for icy bodies Icarus 63 317ndash332

Baverstock KF and Burns WG (1976) Primary pro-duction of oxygen from irradiated water as an expla-nation for decreased radiobiological oxygen enhance-ment at high LET Nature 260 316ndash318

Benit J and Brown WL (1990) Electronic sputtering ofoxygen and water molecules from thin films of waterice bombarded by MeV Ne1 ions Nucl Instrum Meth-ods Phys Rev B 46 448ndash451

Benit J Bibring J-P della-Negra S LeBeyec YMendenhall M Rocard F and Standing K (1987)Erosion of ices by ion irradiation Nucl Instrum Meth-ods Phys Rev B 1920 838ndash842

Brown WL Augustyniak WM Lanzerotti LJ John-son RE and Evatt R (1980) Linear and nonlinearprocesses in the erosion of H2O ice by fast light ionsPhys Rev Lett 43 1632ndash1635

Brown WL Augustyniak W M Simmons E Mar-cantonio KJ Lanzerotti LJ Johnson RE BoringJW Reimann CT Foti G and Pirronello V (1982)Erosion and molecular formation in condensed gasfilms by electronic energy loss of fast ions Nucl In-strum Methods B 1 307ndash314

Brown WL Augustyniak WM Marcantonio KJ Sim-mons EN Boring JW Johnson RE and ReimannCT (1984) Electronic sputtering of low-temperaturemolecular solids Nucl Instrum Methods B 1 307

Buratti B (1995) Photometry and surface structure of theicy Galilean satellites J Geophys Res 100(E9) 19061ndash19066

Buratti B and Veverka J (1983) Voyager photometry ofEuropa Icarus 55 93ndash110

Calvin WM and Spencer JR (1997) Latitudinal distri-bution of O2 on Ganymede observations with the Hub-ble Space Telescope Icarus 130 505ndash516

Calvin WM Clark RN Brown RH and Spencer JA(1995) Spectra of the icy Galilean satellites from 02 to5 mm a compilation new observations and a recentsummary J Geophys Res 100 19041ndash19048

Calvin WM Johnson RE and Spencer JA (1996) O2

on Ganymede spectral characteristics and plasma for-mation mechanisms Geophys Res Lett 23 673ndash676

Calvin WM Aninich VG and Brown RH (2004) Oxy-gen temperature and phase effects Icarus (in press)

OXIDANTS AT EUROPA 847

bombardment plays an essential role in deter-mining the measured yields as discussed aboveThis effect needs to be evaluated for a planetarysurface subject to particle fluxes that are many or-ders of magnitude smaller than those used in thelaboratory

In the absence of O2 yields for all of the rele-vant ions and energies an estimate of the totalsupply of O2 can be obtained using the G valuesin Table 2 For the energetic oxygen and sulfurions we use G values for neon and argon ionsThese are lower bounds as the data were obtainedat temperatures much lower than the surface tem-peratures at Europa (80ndash110K) For the fast pen-etrating electrons and protons we use the valuefor 200 keV protons in Table 2 For the heavy ionswe used the Ne1 and Ar1 values For compari-son we also give an estimate using the G valuesobtained from the low energy electron data andthe total ionization as discussed above

The G values are combined in Table 3 with theelectron and ion energy flux at Europa Althoughthere is some deflection of the flow due to in-duced fields the net energy flux to the surface isnot reduced much from these values (Paranicaset al 2002) In fact locally produced pick-up ionswhich are formed from the ejecta and acceleratedto the surface can contribute (Ip 1996) Furtheralthough G values are typically independent ofthe incident angle the loss of O2 is a surfaceprocess and is enhanced with angle That is thenumber of O2 produced within the escape depthincreases with increasing angle to the normal(steeper than the inverse cosine) This enhance-ment is 25ndash4 (Jurac et al 2001) Ignoring suchcorrections the total production rate calculated

(23ndash62 3 109 O2cm2s) is a lower bound to theactual production rate from the icy regions of Eu-roparsquos surface Recent modeling of the O2 atmo-sphere suggest that such a source rate is required(Shematovich and Johnson 2001)

Earlier estimates of the O2 source rate were in-correct because O2 production was assumed to bea fixed fraction of the H2O sputtering yield Atthe ambient temperature an ejected H2O sticks ef-ficiently (Smith and Kay 1997) to the surface ofneighboring grains in the porous regolith reduc-ing the sputtering yield for H2O (Johnson 1989)This reduction does not apply to the O2 fractionof the yield enhancing its relative importance toH2O Here we assume that all of the gas-phaseO2 produced can contribute to the ambient at-mosphere even if it is produced at a grain sur-face that is at some depth into the porous regolith

Finally Europarsquos surface is not pure ice every-where That is much of the trailing hemisphereis a hydrated sulfuric acid containing smallamounts of sulfur and SO2 and the leading hemi-sphere contains low levels of carbon CO2 andcarbonates Sodium and potassium atoms aresputtered from the icy regions (eg Leblanc et al2002) O2 production from surfaces containingthese species has not been measured C and S areboth proton scavengers and the other species canbe sources of O as discussed so that the O2 yieldfrom such regions could be enhanced Howeverthere can also be competing processes that mightreduce these rates In fact recent HST observa-tions suggest that the O2 might not be produceduniformly across Europarsquos surface (McGrath et al2000) The analysis of these observations is pre-liminary and is not simple since the observed

JOHNSON ET AL840

TABLE 3 APPLICATION TO EUROPA

OxygenImplantation Energy flux production

Species 109(cm2 s)a 109 keV(cm2 s)(1) 109(cm2 s)

H (10 keV) 0015(1) 12 0084 (072)bO (10 keV) 0015(1) 18 054cS (10 keV) 0009(1) 30 090cElectrons 62 043 (37)bPlasma (10 keV) (137 amu) 13(2) 10 03c

Total 79 23 (62)UV photons (64 eV) 46 01

aReferences are as follows 1Cooper et al (2001) [similar results in Paranicas et al (2002)] 2Mauk et al (1996)bUsing G for 200 keV protons (upper bounds in brackets are based on the low-energy electrons data as a model for

the secondary electrons produced)cUsing G for 15 MeV Ne1 at 7K a rough lower bound

emissions depend on the local plasma propertiesHowever the efficiency of O2 production in con-taminated ices needs to be studied

Porous regolith O2 interactions on grain surfaces

Although the O2 atmosphere is principally pro-duced in the near surface layers and is eventuallylost to space it also permeates Europarsquos porousregolith The porosity p (void space) of the re-golith is uncertain (Buratti 1995) It has been es-timated to be the most compact surface in theOuter Solar System p 025 on the trailing hemi-sphere and p 079 on the leading hemisphere(Buratti and Veverka 1983) and the most poroussurface p 096 globally (Domingue et al 1991Domingue and Hapke 1992) These large differ-ences are due to the uncertainties in the amountof backscattered light at small phase angles andon the photon absorption properties and grainsizes The more compact surface is likely correctGrain sizes rg of the order of 100 mm have beensuggested (Hansen and McCord 1999 2000Geissler et al 1998) Europarsquos regolith differs sig-nificantly from the lunar regolith which is asand-like surface of individual grains Analysis ofthe thermal inertia of Europarsquos surface (Grundyet al 2001) suggests that the ice grains are sin-tered forming a fairy castle-like structure Sin-tering can occur thermally or due to the incidentradiation In addition ices doped with acids andalkalis in the laboratory can be powdery solidswith the impurities primarily found at grain in-terfaces

The porous regolith affects the above estimatesof the oxygen production in three ways First forisotropic bombardment the mean distance d intothe surface at which an incident ion or photonstrikes a grain surface is d (89)rg(1 2 p) Us-ing the above value for rg then d 02 m if theporosity is very large p 96 Therefore de-sorbed molecules can come from a considerabledepth and will interact with grain surfaces Evenfor much lower porosities the effective sputteringyields can be significantly modified (Johnson1989) As discussed above the yields for ejectedH2O which sticks onto neighboring grains at theambient temperatures are reduced but those forejected O2 are not This enhances the relative im-portance of the O2 yields

Finally the surfaces and interfaces of the sin-tered grains in Europarsquos porous surface are placeswhere low temperature chemistry involving the

ambient oxygen can occur at depths that are wellbelow the nominal penetration depth of the inci-dent radiation The relative importance of suchchemistry is roughly indicated by the ratio of thetotal surface area in the regolith to the area of Eu-roparsquos visible surface Using the porosity andgrain size above and a regolith depth dr of theorder of a meter then this ratio is large 3(1 2p)drrg which is 103ndash2 3 104 for average porosi-ties of 96 and 25 respectively Therefore thegas-phase O2 molecules below the visible surfaceinteract frequently with the grain surfaces In thisway a very large volume of the subsurface couldbe altered chemically even if it is not directly ex-posed to the radiation

From the overlying structures seen in the high-resolution images of Europarsquos surface by theGalileo spacecraft (eg Prockter and Pappalardo2000) it is clear that the youngest surfaces are red-der The reddening in the visible has been attrib-uted to the presence of sulfur in the ice (McEwen1986 Johnson et al 1988 Calvin et al 1995 Carl-son et al 1999b) The amount of reddening in thespectra of the surface material decreases with ageand the surface materials appear to becomebrighter with age This might occur by the slowaccumulation of a frost over-layer or by radioly-sis (Carlson et al 2002 Johnson et al 2003) Ra-diolysis preferentially converts the sulfur in theice to sulfur dioxide or sulfuric acid both ofwhich have flat spectra in the visible In regionswhere the radiation does not directly penetratethe O2 permeating the regolith can oxidize thesulfur at the surfaces of the icy grains and at graininterfaces Similarly the carbon seen in the ice onEuroparsquos leading hemisphere (eg Carlson et al2004 Johnson et al 2003) can be oxidized by ra-diolysis in the near surface layers or by reactionson the grain surfaces at depth in the regolith bythe ambient O2 The meteoroid bombardmentthat produces the regolith also exposes fresh sur-faces for oxidation and can cause the collapse ofpores leading to the trapping of oxygen In thismanner both refractory and volatile oxidants canbe formed and trapped in the subsurface ice forpotential delivery to Europarsquos subsurface ocean

SUMMARY

The presence of a thin oxygen atmosphere andoxidants trapped in Europarsquos surface ice (egJohnson et al 2003) combined with the possible

OXIDANTS AT EUROPA 841

presence of a subsurface ocean (eg Carr et al1998) have made Europa an exciting target forfuture space exploration The molecular oxidantsproduced by radiolysis and photolysis could bespecies needed for carbon-based biochemistry inEuroparsquos putative subsurface ocean Howeverthere has been no model for the formation trap-ping and transport of relevant oxidants

In this paper we have reviewed the space ob-servations and laboratory studies related to thechemistry of formation of oxidants in ice by theincident radiation Because the laboratory dataare incomplete we examined the rate equationsfor formation of oxygen and related species byradiolysis and photolysis We showed how suchequations can be simplified to the analytic modelthat was used for fitting the fluence and temper-ature dependence of the O2 yield produced bylow-energy electrons (Sieger et al 1998) In de-riving the analytic model from the more completerate equations the competition between forma-tion and destruction processes can be studied ascan the role of trapping sites escape from depthand temperature history These equations also al-lowed us to put incident photon electron and iondata on a similar footing for the first time

The model can be used to suggest measure-ments that are needed to describe the chemicalstate of the irradiated ice and to interpret new lab-oratory data We showed that if reactions are fastthe yield data could be consistent with a precur-sor being formed by two events as is the case forperoxide However we favor trapped O as a pre-cursor as initially suggested by Matich et al(1993) for production of excited O2 Such aspecies if present could be detected spectro-scopically as can trapped O2 through bands ac-tivated by lattice perturbations (Cooper et al2003a)

The available oxygen yield data were used tocompare G values for different incident particlesThese were then used to estimate the oxygen pro-duction in the radiation environment at EuropaThe lower bounds were consistent with the pro-duction rates obtained from a recent model of Eu-roparsquos atmosphere The porosity of the regolithwas also considered It not only affects the rela-tive sputtering yields for O2 and H2O but alsoprovides an enormously increased surface areaon which the ambient gas-phase oxygen can re-act with species in the ice That this might be oc-curring is consistent with the change in re-

flectance associated with surface aging Such re-actions could also convert ambient molecularoxygen into other more refractory species that re-main trapped on subduction of Europarsquos crust

Although radiolysis produces a stable chemi-cally altered surface in the laboratory the parti-cle flux to Europarsquos surface is many orders ofmagnitude lower Therefore annealing can be ef-fective at temperatures that are much lower thanthose typically found experimentally Belowabout 120K the annealing times appear to be long(eg Baragiola 2003) so that laboratory data canoften be directly applied to Europarsquos icy surfaceHowever account has to be taken of the syner-gism between the UV and the energetic plasmaas discussed earlier (Johnson and Quickenden1997)

Transient surface melting might initiate lifeprocesses but subduction of oxidants to the pu-tative ocean is likely to be required if biochem-istry is to occur at Europa (Chyba and Hand2001) Although the temperature versus depthinto Europarsquos regolith is not well known belowthe regolith the temperature and pressure appar-ently increase with depth At some depth (ap-proximately kilometers) either very warm ice oran ocean exists (Ruiz and Tejero 2000) Thereforechemically refractory oxidants are likely to be re-quired in order to survive subduction to thewarmer subsurface regions Trapped species ifdelivered downward can become mobile andwill segregate forming inclusions (Johnson andJesser 1997) Indeed peroxide appears to segre-gate in ice (Gurman et al 1967) and these inclu-sions may remain relatively stable until the per-oxide is dissolved in the putative ocean Thevolatiles seen in reflectance such as O2 SO2 andCO2 will also segregate But at the higher tem-peratures they could also diffuse through thebulk react or percolate into the regolith atmo-sphere Therefore the observation of the trappedvolatiles in Europarsquos surface ice is of interest pri-marily because it suggests that oxidants are pre-sent The formation of oxygen-rich molecules fol-lows the preferential loss of hydrogen Howeverthe surface is in a local steady state with im-planted H O and S from the plasma and the lossto space Since the plasma flux to the surface canbe variable this balance can also vary Indeedtemporal changes in reflectance have been re-ported (Domingue and Hendrix 2004)

Based on the models for formation of O2 ex-

JOHNSON ET AL842

amined in this paper additional experiments areneeded to determine the chemical state of irra-diated ice at low fluences Of immediate inter-est is the identification of stable molecular pre-cursors required for the formation of O2 Basedon the available laboratory data individual per-oxide molecules trapped in the ice do not appearto be the principal precursors but trapped O(Matich et al 1993) remains a viable candidateAlthough the data suggest that the local hydro-gen-bonding network is important in the pro-duction of O2 the effect of the formation tem-perature and radiation damage on the structureneed to be measured It is also important to de-scribe quantitatively the relationship betweenH2 loss and the formation of O2 and H2O2 overa range of temperatures and fluences Finallyone would like to determine the steady-statedensities of biochemically interesting moleculesformed by radiolysis in an ice containing sulfurand carbon at Europarsquos surface temperaturesand then determine their stability with increas-ing temperature and pressure Such experimentsstill need to be carried out Here we have pre-sented a model that brings together the variouslaboratory data and now can be used to considerthe formation trapping and transport of relevantoxidants at Europa

APPENDIX

Introduction

In this Appendix we examine rate equationsthat describe various aspects of the radiolysis orphotolysis of ice These equations need to be un-derstood if one is to be able to calculate the for-mation trapping and transport of oxidants Theemphasis is on the formation of molecular oxy-gen We integrate these equations over the escapedepth of O2 to give the simplified rate equationsdiscussed in the text in which number densitiesare replaced by column densities (eg Eqs 1 and2) In this way we can examine the relationshipbetween the rate constants and the effective pre-cursor formation and destruction cross sectionsextracted from data We first consider a precur-sor such as H2O2 that is formed from two exci-tation events This bears on the observed linear-ity of the oxygen yield versus fluence We thenconsider the equations for trapped versus a tran-

siently mobile species We use a reaction betweena mobile and trapped O as an example althoughO 1 OH can also give oxygen Finally we ex-amine the possible dependence of the O2 forma-tion process on the density of traps (defectsvoids and grain boundaries) as a function of tem-perature The results are discussed in the text

For clarity all possible radiolytic species arenot included in the discussion below Thereforepotentially important products such as HO2 andthe charged species (eg H3O1 the trapped elec-tron etc) are missing In each section we attemptto make a point related to production of O2 andother oxidants by using a subset of the productsrather than attempt a complete description withmany unknown rate constants

Fast versus slow processes

To describe the O2 production rate from a pre-cursor such as H2O2 we first use a simple but in-structive set of rate equations to describe the for-mation of H2O2 We then relate these to thesimplified rate equation for the precursor in thetext Eq 1 and show how the postulated precur-sor formation cross section sp is related to theH2O dissociation cross section sd by the reactionrates

Using the reactions

H2O 1 radiation R H 1 OH (sd)

H 1 H R H2 (k2H)

H 1 OH R H2O (kHOH)

OH 1 OH R H2O2 (k2OH)

the reaction pathways are described by

H2

Hqk2H

1 OHH2O 1 radiation mdashRsd H

OHH2O2

OH Rk2OH

kHOH

HH2O

OXIDANTS AT EUROPA 843

The corresponding rate equations in an irradiatedvolume are

dnOHdt 5 sd f n 2 2 k2OH nOH nOH

2 kHOH nH nOH (A1a)

dnHdt 5 sd f n 2 2 k2H nH nH

2 kHOH nH nOH (A2a)

dnH2O2dt 5 k2OH nOH nOH

2 destruction processes (A3a)

Here sd is the dissociation cross section f is theradiation flux and n is the molecular numberdensity of ice with nOH and nH the density of dis-sociation products OH and H respectively Therate constants k2OH k2H and kHOH describe thereactions above In this model the H2 formed isassumed to escape for temperatures above about20K We do not distinguish between trappedspecies and mobile species although in fact thatdistinction is important as discussed in the nextsection That is any OH produced at low tem-perature by one ion would likely be trapped priorto interacting with an OH produced by a subse-quent ion At higher temperatures this can breakdown as discussed in the text We first relate theabove set of equations to Eq 1 in the text and thenconsider precursor destruction processes

Equation 1 in the text is obtained by solvingthese equations together and then by integratingover a depth Dx Here Dx is the smaller of the pen-etration depth of the particle and the O2 escape(percolation) depth Assuming the molecular den-sity of ice n is constant with depth Dx n 5 N inEq 1 Comparing Eq 1 with Eq A3a the precur-sor formation rate sp f N is equivalent to [k2OH

nOH nOH] Dx Assuming that the processes in EqsA1a and A2a are fast then the steady state is read-ily obtained In addition if k2H k2OH there ispreferential loss of H2 at short times as suggestedby the laboratory data If steady state is reachedthe production of H2 is equal to that of H2O2 us-ing only the equations above Further one obtains

[k2OH nOH nOH] Dx R [fp sd] f N (A4)

with

fp 5 05[1 1 05 kHOH(k2H k2OH)05] (A5a)

Therefore the precursor formation cross sectionsp in Eq 1 is a fraction fp of the total dissocia-

tion cross sd Since laboratory data suggest theH2O2 fraction in an irradiated ice is small thenkHOH (k2OH k2H)05 which is likely givingfp R [(k2OH k2H)05kHOH]

Trapped versus mobile species

To examine the fluence dependence discussedin the text we note that trapped OH and freshlyproduced transiently mobile OH can be treatedas different species (eg Matich et al 1993)Therefore assuming no trapping of H the OHrate equation above would be replaced by twoequations one for the trapped species nOHt andthe other for the freshly produced mobile speciesnOH Assuming the reaction of two mobile OHunits is unlikely and that peroxide is formed fromthe interaction of a trapped OH with a subse-quently produced OH an expanded set of equa-tions is obtained

OH 1 trap R OHt (kOHt)

OH 1 OHt R H2O2 (kOHOHt)

H 1 OHt R H2O (kHOHt)

then the reaction pathways are described by

qH2

H2O 1 radiation mdashRsdH

qk2H

1 OHH

trap OH

OH RkOHt

OHt kOHOHtH2O2

kHOHkHOHt

H H2O HH2O

The resulting set of equations is

dnOHdt 5 sd f n 2 kHOH nH nOH

2 kOHt nOH nt 2 kOHOHt nOH nOHt (A1b)

dnOHtdt 5 kOHt nOH nt 2 kOHOHt nOH nOHt

2 kHOHt nH nOHt (A1b9)

dnHdt 5 sd f n 2 2 k2H nH nH 2

kHOH nH nOH 2 kHOHt nH nOHt (A2b)

JOHNSON ET AL844

dnH2O2dt 5 kOHOHt nOH nOHt

2 destruction processes (A3b)

where nt is the trap density in the irradiated iceAt low T we can assume nt is large so thatdntdt 0 simplifying the set of equations If theprecursor of interest is trapped O as discussed inthe text then a similar set of equations can be con-structed for O and Ot as in the example to followOf course a full description over a broad rangeof temperatures would require the considerationof the both trapped and mobile H O and OHthe rate of change of trap density and the vari-ous destruction processes

The above equations should be solved simul-taneously but for the incident particle fluxes typ-ically used in these studies the diffusion of mo-bile species to traps is relatively fast This is trueeven at low temperatures if there is a significantdensity of trapping sites This is usually the casefor samples formed at low T Therefore in a lowtemperature solid at low dose rates there is not asignificant steady-state background of diffusingOH There are trapped or freshly produced OHunits which rapidly react Averaging over thebombardment rate this can be approximated bywriting dnOHdt 0 and then separating nOH outin Eq A1b Assuming also that H diffuses rapidlyand either reacts or forms H2 we also writednHdt 0 on the average Therefore betweenimpacts only species trapped at defects pores orgrain surfaces are assumed to be present as sug-gested by the lack of a dependence on the inci-dent beam flux In this limit the rate of change oftrapped OH roughly simplifies to

dnOHtdt 5 2 k2H nH nH 2 2 kOHOHt nOH nOHt

It is seen that the production of peroxide dependson the loss of H by formation of H2 as discussedand at steady state the production rate for per-oxide equals the rate of formation of H2 in theseequations

Treating the mobile and trapped species sepa-rately we can further simplify the rate equationsand obtain the precursor model used in the textThat is integrating over the relevant depth Dxthe densities are replaced by column densitiesThe rate constants ki are then related to effectivecross sections si In this way cross sections aregiven as an interaction length for each freshlyproduced mobile radical [ki nj] R [si f] The

trapped OH column density NOHt and peroxideproduction rate can be estimated from ldquolinearrdquorate equations

dNOHtdt lt [sd f] N 2 2 [s2OH f] NOHt

2 [sHOH f] NOHt (A6a)

dNH2O2dt lt [s2OH f] NOHt 2 s NH2O2 (A6b)

Here sd is again the dissociation cross sectionand the destruction processes in Eq A3b are con-tained in s The second and third terms on theright in Eq A6a are the loss of OH due to the pro-duction of H2O2 and H2O In the integration ofEq A1b9 [2 kOH OHt nOH] becomes [2 s2OH f] and[kHOHt nOH] becomes [sHOH f] Therefore thedensity and the interaction length of mobile H arecontained in sHOH If H2O2 is indeed the pre-cursor then in Eq 2 in the text [sp f N] is be-comes [s2OH f NOHt] from Eq A6b Accountingfor precursor destruction which can give backtwo OH and treating H2 can both be included inthe integrated model via the si values

The simplified set of equations above allows usto examine the relative time (fluence) scales forprecursor formation (in this case H2O2) SolvingEqs A6a and A6b the column density of OH andthe concentration of precursors versus fluence cpcan be obtained Using the column of H2O2 pro-duced as the precursor column Np in Eq 2 thenO2 yield can be obtained The results can be com-pared with the results from Eqs1 and 2 in thetext The average precursor (H2O2) concentrationover the column depth of water molecules N is

cp 5 fp (sds)[1 2 exp(2s F)] 2 (ssr)

[1 2 exp(2sr F)] (A7)

with

fp 5 2s2OH(sr 2 s) (A5b)

In these equations sr 5 (2s2OH 1 sHOH) andssr 1

It is seen that the precursor concentration withtime cp depends on two rates the reaction ratesthat determine sr and the destruction rates in sAs in Eq 4a the yield can be written as

Y 5 sO2 cp N (A8)

If sr and s are comparable then the yield at small

OXIDANTS AT EUROPA 845

fluence is quadratic in F However if sr sthen the linear dependence on fluence is obtainedwith a small shift in the fluence as shown in Eq4c in the text Further the steady-state yield isnow

Y` (T) 5 sO2 [fp sd] Ns (A9)

That is sp R [fp sd] as in Eq A4 and the tem-perature dependence for the incident electrondata must be contained in [sO2 fp] or in the ratio[sO2 s2OH]sHOH]

For comparison with their precursor modelSieger et al (1998) examined the possibility of ob-taining the O2 precursor by considering reactionsfor a density of diffusing dissociation products asin Eqs A1 and A2 rather than a trapped productinteracting with a freshly formed dissociationproduct Based on the above the lowest fluence atwhich O2 is first detected experimentally sets alimit on the time scale for mobilendashmobile reactions

Effect of trap density

Equations A1b A1b9 A2b and A3b can be ap-proximately solved assuming that the dnOHdtdnHdt and dntdt are very small on the averageas discussed above In order to examine the roleof the trap density nt which can depend on thetemperature and method of formation and on theradiation history of the ice we use the slightlysimpler precursor model In this model the O pro-duced directly by dissociation traps forming aprecursor O2 is then formed when a freshly pro-duced O or other species reacts with a trappedoxygen atom Ot

H2O 1 radiation R H2 1 O (sd)

O 1 trap R Ot (kOt)

O 1 Ot R O2 (kOOt)

The reaction pathways for the oxygen species are

H2O 1 radiation mdashRsd9

trap OO2

kOtOt kOOt

H2 1 OOHt

kOR H(kHO t)

ROR

We assume for simplicity both H2 and O2 escapefrom depth Dx after formation We also allow mo-bile H to react with Ot (kHOt) removing trappedO and both the mobile H and O to be removedby other reactants R (eg OH or contaminants)This leads to a simplified set of equations not di-rectly involving OH or peroxide

dnOdt 5 sd9 f n 2 kOt nO nt 2 kOR nO nR

2 kOOt nO nOt (A1c)

dnOtdt 5 kOt nt nO 2 kOOt nO nOt

2 kHOt nH nOt (A1c9)

dnHdt 5 sd f n 2 kHOt nH nOt

2 kHR nH nR (A2c)

dnO2dt 5 kOOt nO nOt (A3c)

These equations of course act in parallel with theequations above for OH and mobile OH may re-act with trapped O to form O2 Here we focusonly on mobile O

Since the diffusion of H and O are fast and thedensity of traps is large the nH and nt averagedover the bombardment rate can be taken veryroughly as time independent Further if we as-sume the number of trapped O nOt is small com-pared with the number of sites at which H canreact then nH [sd f nkHR nR] For a low ra-diation flux f nH is small on the average as as-sumed Since the O2 yield YO2 is equal to[(dnO2dt) Dxf] with Dx the escape depth thenthe steady-state yield Y` can be obtained ana-lytically

[Y`][sd9 N] 5 1 1 (q 2 1) 2 [(q 2 1)2

1 8 q d]054 d (A10)

q 5 [kOt (nt 1 nR)] (kHOt nH)(sd9 f n) kOOt

5(sdsd9) [kOt (nt 1 nR)] kHOt[kOOt kHR nR]

d 5 05 [1 1 nt(nt 1 nR)]

We note that for nt large q can be large Assum-ing that q 1 and nt nR then Eq A10 be-comes

Y` R [sd9 N]q 5 (sd9 N) (sd9sd)

[kOOt kHR nR][kHOt kOt nt] (A5c)

JOHNSON ET AL846

Comparing with Y` in the simple precursormodel in Eq 4b then s R [(kHOt nH)f] R [sd nkHOt][kHR nR] In addition the product [sO2sp] R sd9 sd9 [kOOt n][kOt nt] That is the dis-sociation cross sections are modified by reactionrates

Quite remarkably it is seen in Eq A5c that thesteady-state yield Y` can depend inversely onthe density of traps nt in the temperature andflux regimes for which the approximations arerelevant That is if there are competing destruc-tion process for trapped O then at high trap den-sity a mobile O must find a trapped O before itbecomes trapped or before mobile H destroys theOt In this way the increasing O2 yield with in-creasing T can be related to the change in the trap-ping density nt with the formation temperatureof the ice sample As the temperature increasesand the density of traps in the bulk decreases thesurface and grain interfaces become the only trap-ping sites

ACKNOWLEDGMENTS

REJ acknowledges support by a Gledden Fel-lowship in Chemistry School of Biomedical andChemical Sciences at The University of WesternAustralia and by grants from the NSF Astron-omy Program and NASArsquos Planetary Geologyand Geophysics Program PDC acknowledgessupport by an Australian Postgraduate Award

ABBREVIATIONS

HST Hubble Space Telescope MJQ MatichndashJohnsonndashQuickenden NIMS Near Infrared Map-ping Spectrometer UV ultraviolet

REFERENCES

Ayotte P Smith RS Stevenson KP Dohnarsquolek ZKimmel GA and Kay BD (2001) Effect of porosityon the adsorption desorption trapping and release ofvolatile gases by amorphous solid water J Geophys Res106(E12) 33387ndash33392

Bahr DA Famaacute M and Baragiola RA (2001) Radiol-ysis of water ice in the outer solar system sputteringand trapping of radiation products J Geophys Res 10633285ndash33290

Baragiola RA (2003) Microporous amorphous water icefilms and astronomical implications In Water in Con-

fining Geometries edited by JP Devlin and B Buch El-sevier New York in press

Baragiola RA and Bahr DA (1998) Laboratory studiesof the optical properties and stability of oxygen onGanymede J Geophys Res 103 25865ndash25872

Baragiola RA Atteberry CL and Bahr DA (1999) Ori-gin of solid oxygen on Ganymede Reply to commentby RE Johnson on ldquoLaboratory studies of the opticalproperties and stability of oxygen on Ganymederdquo JGeophys Res E104 14183ndash14187

Baragiola RA Vidal RA Svendsen W Schou J ShiM Bahr DA and Atteberrry CL (2003) Sputteringof water ice Nucl Instrum Methods Phys Res B 209294ndash303

Bar-Nun A Herman G Laufer D and Rappaport ML(1985) Trapping and release of gases by water ice andimplications for icy bodies Icarus 63 317ndash332

Baverstock KF and Burns WG (1976) Primary pro-duction of oxygen from irradiated water as an expla-nation for decreased radiobiological oxygen enhance-ment at high LET Nature 260 316ndash318

Benit J and Brown WL (1990) Electronic sputtering ofoxygen and water molecules from thin films of waterice bombarded by MeV Ne1 ions Nucl Instrum Meth-ods Phys Rev B 46 448ndash451

Benit J Bibring J-P della-Negra S LeBeyec YMendenhall M Rocard F and Standing K (1987)Erosion of ices by ion irradiation Nucl Instrum Meth-ods Phys Rev B 1920 838ndash842

Brown WL Augustyniak WM Lanzerotti LJ John-son RE and Evatt R (1980) Linear and nonlinearprocesses in the erosion of H2O ice by fast light ionsPhys Rev Lett 43 1632ndash1635

Brown WL Augustyniak W M Simmons E Mar-cantonio KJ Lanzerotti LJ Johnson RE BoringJW Reimann CT Foti G and Pirronello V (1982)Erosion and molecular formation in condensed gasfilms by electronic energy loss of fast ions Nucl In-strum Methods B 1 307ndash314

Brown WL Augustyniak WM Marcantonio KJ Sim-mons EN Boring JW Johnson RE and ReimannCT (1984) Electronic sputtering of low-temperaturemolecular solids Nucl Instrum Methods B 1 307

Buratti B (1995) Photometry and surface structure of theicy Galilean satellites J Geophys Res 100(E9) 19061ndash19066

Buratti B and Veverka J (1983) Voyager photometry ofEuropa Icarus 55 93ndash110

Calvin WM and Spencer JR (1997) Latitudinal distri-bution of O2 on Ganymede observations with the Hub-ble Space Telescope Icarus 130 505ndash516

Calvin WM Clark RN Brown RH and Spencer JA(1995) Spectra of the icy Galilean satellites from 02 to5 mm a compilation new observations and a recentsummary J Geophys Res 100 19041ndash19048

Calvin WM Johnson RE and Spencer JA (1996) O2

on Ganymede spectral characteristics and plasma for-mation mechanisms Geophys Res Lett 23 673ndash676

Calvin WM Aninich VG and Brown RH (2004) Oxy-gen temperature and phase effects Icarus (in press)

OXIDANTS AT EUROPA 847

emissions depend on the local plasma propertiesHowever the efficiency of O2 production in con-taminated ices needs to be studied

Porous regolith O2 interactions on grain surfaces

Although the O2 atmosphere is principally pro-duced in the near surface layers and is eventuallylost to space it also permeates Europarsquos porousregolith The porosity p (void space) of the re-golith is uncertain (Buratti 1995) It has been es-timated to be the most compact surface in theOuter Solar System p 025 on the trailing hemi-sphere and p 079 on the leading hemisphere(Buratti and Veverka 1983) and the most poroussurface p 096 globally (Domingue et al 1991Domingue and Hapke 1992) These large differ-ences are due to the uncertainties in the amountof backscattered light at small phase angles andon the photon absorption properties and grainsizes The more compact surface is likely correctGrain sizes rg of the order of 100 mm have beensuggested (Hansen and McCord 1999 2000Geissler et al 1998) Europarsquos regolith differs sig-nificantly from the lunar regolith which is asand-like surface of individual grains Analysis ofthe thermal inertia of Europarsquos surface (Grundyet al 2001) suggests that the ice grains are sin-tered forming a fairy castle-like structure Sin-tering can occur thermally or due to the incidentradiation In addition ices doped with acids andalkalis in the laboratory can be powdery solidswith the impurities primarily found at grain in-terfaces

The porous regolith affects the above estimatesof the oxygen production in three ways First forisotropic bombardment the mean distance d intothe surface at which an incident ion or photonstrikes a grain surface is d (89)rg(1 2 p) Us-ing the above value for rg then d 02 m if theporosity is very large p 96 Therefore de-sorbed molecules can come from a considerabledepth and will interact with grain surfaces Evenfor much lower porosities the effective sputteringyields can be significantly modified (Johnson1989) As discussed above the yields for ejectedH2O which sticks onto neighboring grains at theambient temperatures are reduced but those forejected O2 are not This enhances the relative im-portance of the O2 yields

Finally the surfaces and interfaces of the sin-tered grains in Europarsquos porous surface are placeswhere low temperature chemistry involving the

ambient oxygen can occur at depths that are wellbelow the nominal penetration depth of the inci-dent radiation The relative importance of suchchemistry is roughly indicated by the ratio of thetotal surface area in the regolith to the area of Eu-roparsquos visible surface Using the porosity andgrain size above and a regolith depth dr of theorder of a meter then this ratio is large 3(1 2p)drrg which is 103ndash2 3 104 for average porosi-ties of 96 and 25 respectively Therefore thegas-phase O2 molecules below the visible surfaceinteract frequently with the grain surfaces In thisway a very large volume of the subsurface couldbe altered chemically even if it is not directly ex-posed to the radiation

From the overlying structures seen in the high-resolution images of Europarsquos surface by theGalileo spacecraft (eg Prockter and Pappalardo2000) it is clear that the youngest surfaces are red-der The reddening in the visible has been attrib-uted to the presence of sulfur in the ice (McEwen1986 Johnson et al 1988 Calvin et al 1995 Carl-son et al 1999b) The amount of reddening in thespectra of the surface material decreases with ageand the surface materials appear to becomebrighter with age This might occur by the slowaccumulation of a frost over-layer or by radioly-sis (Carlson et al 2002 Johnson et al 2003) Ra-diolysis preferentially converts the sulfur in theice to sulfur dioxide or sulfuric acid both ofwhich have flat spectra in the visible In regionswhere the radiation does not directly penetratethe O2 permeating the regolith can oxidize thesulfur at the surfaces of the icy grains and at graininterfaces Similarly the carbon seen in the ice onEuroparsquos leading hemisphere (eg Carlson et al2004 Johnson et al 2003) can be oxidized by ra-diolysis in the near surface layers or by reactionson the grain surfaces at depth in the regolith bythe ambient O2 The meteoroid bombardmentthat produces the regolith also exposes fresh sur-faces for oxidation and can cause the collapse ofpores leading to the trapping of oxygen In thismanner both refractory and volatile oxidants canbe formed and trapped in the subsurface ice forpotential delivery to Europarsquos subsurface ocean

SUMMARY

The presence of a thin oxygen atmosphere andoxidants trapped in Europarsquos surface ice (egJohnson et al 2003) combined with the possible

OXIDANTS AT EUROPA 841

presence of a subsurface ocean (eg Carr et al1998) have made Europa an exciting target forfuture space exploration The molecular oxidantsproduced by radiolysis and photolysis could bespecies needed for carbon-based biochemistry inEuroparsquos putative subsurface ocean Howeverthere has been no model for the formation trap-ping and transport of relevant oxidants

In this paper we have reviewed the space ob-servations and laboratory studies related to thechemistry of formation of oxidants in ice by theincident radiation Because the laboratory dataare incomplete we examined the rate equationsfor formation of oxygen and related species byradiolysis and photolysis We showed how suchequations can be simplified to the analytic modelthat was used for fitting the fluence and temper-ature dependence of the O2 yield produced bylow-energy electrons (Sieger et al 1998) In de-riving the analytic model from the more completerate equations the competition between forma-tion and destruction processes can be studied ascan the role of trapping sites escape from depthand temperature history These equations also al-lowed us to put incident photon electron and iondata on a similar footing for the first time

The model can be used to suggest measure-ments that are needed to describe the chemicalstate of the irradiated ice and to interpret new lab-oratory data We showed that if reactions are fastthe yield data could be consistent with a precur-sor being formed by two events as is the case forperoxide However we favor trapped O as a pre-cursor as initially suggested by Matich et al(1993) for production of excited O2 Such aspecies if present could be detected spectro-scopically as can trapped O2 through bands ac-tivated by lattice perturbations (Cooper et al2003a)

The available oxygen yield data were used tocompare G values for different incident particlesThese were then used to estimate the oxygen pro-duction in the radiation environment at EuropaThe lower bounds were consistent with the pro-duction rates obtained from a recent model of Eu-roparsquos atmosphere The porosity of the regolithwas also considered It not only affects the rela-tive sputtering yields for O2 and H2O but alsoprovides an enormously increased surface areaon which the ambient gas-phase oxygen can re-act with species in the ice That this might be oc-curring is consistent with the change in re-

flectance associated with surface aging Such re-actions could also convert ambient molecularoxygen into other more refractory species that re-main trapped on subduction of Europarsquos crust

Although radiolysis produces a stable chemi-cally altered surface in the laboratory the parti-cle flux to Europarsquos surface is many orders ofmagnitude lower Therefore annealing can be ef-fective at temperatures that are much lower thanthose typically found experimentally Belowabout 120K the annealing times appear to be long(eg Baragiola 2003) so that laboratory data canoften be directly applied to Europarsquos icy surfaceHowever account has to be taken of the syner-gism between the UV and the energetic plasmaas discussed earlier (Johnson and Quickenden1997)

Transient surface melting might initiate lifeprocesses but subduction of oxidants to the pu-tative ocean is likely to be required if biochem-istry is to occur at Europa (Chyba and Hand2001) Although the temperature versus depthinto Europarsquos regolith is not well known belowthe regolith the temperature and pressure appar-ently increase with depth At some depth (ap-proximately kilometers) either very warm ice oran ocean exists (Ruiz and Tejero 2000) Thereforechemically refractory oxidants are likely to be re-quired in order to survive subduction to thewarmer subsurface regions Trapped species ifdelivered downward can become mobile andwill segregate forming inclusions (Johnson andJesser 1997) Indeed peroxide appears to segre-gate in ice (Gurman et al 1967) and these inclu-sions may remain relatively stable until the per-oxide is dissolved in the putative ocean Thevolatiles seen in reflectance such as O2 SO2 andCO2 will also segregate But at the higher tem-peratures they could also diffuse through thebulk react or percolate into the regolith atmo-sphere Therefore the observation of the trappedvolatiles in Europarsquos surface ice is of interest pri-marily because it suggests that oxidants are pre-sent The formation of oxygen-rich molecules fol-lows the preferential loss of hydrogen Howeverthe surface is in a local steady state with im-planted H O and S from the plasma and the lossto space Since the plasma flux to the surface canbe variable this balance can also vary Indeedtemporal changes in reflectance have been re-ported (Domingue and Hendrix 2004)

Based on the models for formation of O2 ex-

JOHNSON ET AL842

amined in this paper additional experiments areneeded to determine the chemical state of irra-diated ice at low fluences Of immediate inter-est is the identification of stable molecular pre-cursors required for the formation of O2 Basedon the available laboratory data individual per-oxide molecules trapped in the ice do not appearto be the principal precursors but trapped O(Matich et al 1993) remains a viable candidateAlthough the data suggest that the local hydro-gen-bonding network is important in the pro-duction of O2 the effect of the formation tem-perature and radiation damage on the structureneed to be measured It is also important to de-scribe quantitatively the relationship betweenH2 loss and the formation of O2 and H2O2 overa range of temperatures and fluences Finallyone would like to determine the steady-statedensities of biochemically interesting moleculesformed by radiolysis in an ice containing sulfurand carbon at Europarsquos surface temperaturesand then determine their stability with increas-ing temperature and pressure Such experimentsstill need to be carried out Here we have pre-sented a model that brings together the variouslaboratory data and now can be used to considerthe formation trapping and transport of relevantoxidants at Europa

APPENDIX

Introduction

In this Appendix we examine rate equationsthat describe various aspects of the radiolysis orphotolysis of ice These equations need to be un-derstood if one is to be able to calculate the for-mation trapping and transport of oxidants Theemphasis is on the formation of molecular oxy-gen We integrate these equations over the escapedepth of O2 to give the simplified rate equationsdiscussed in the text in which number densitiesare replaced by column densities (eg Eqs 1 and2) In this way we can examine the relationshipbetween the rate constants and the effective pre-cursor formation and destruction cross sectionsextracted from data We first consider a precur-sor such as H2O2 that is formed from two exci-tation events This bears on the observed linear-ity of the oxygen yield versus fluence We thenconsider the equations for trapped versus a tran-

siently mobile species We use a reaction betweena mobile and trapped O as an example althoughO 1 OH can also give oxygen Finally we ex-amine the possible dependence of the O2 forma-tion process on the density of traps (defectsvoids and grain boundaries) as a function of tem-perature The results are discussed in the text

For clarity all possible radiolytic species arenot included in the discussion below Thereforepotentially important products such as HO2 andthe charged species (eg H3O1 the trapped elec-tron etc) are missing In each section we attemptto make a point related to production of O2 andother oxidants by using a subset of the productsrather than attempt a complete description withmany unknown rate constants

Fast versus slow processes

To describe the O2 production rate from a pre-cursor such as H2O2 we first use a simple but in-structive set of rate equations to describe the for-mation of H2O2 We then relate these to thesimplified rate equation for the precursor in thetext Eq 1 and show how the postulated precur-sor formation cross section sp is related to theH2O dissociation cross section sd by the reactionrates

Using the reactions

H2O 1 radiation R H 1 OH (sd)

H 1 H R H2 (k2H)

H 1 OH R H2O (kHOH)

OH 1 OH R H2O2 (k2OH)

the reaction pathways are described by

H2

Hqk2H

1 OHH2O 1 radiation mdashRsd H

OHH2O2

OH Rk2OH

kHOH

HH2O

OXIDANTS AT EUROPA 843

The corresponding rate equations in an irradiatedvolume are

dnOHdt 5 sd f n 2 2 k2OH nOH nOH

2 kHOH nH nOH (A1a)

dnHdt 5 sd f n 2 2 k2H nH nH

2 kHOH nH nOH (A2a)

dnH2O2dt 5 k2OH nOH nOH

2 destruction processes (A3a)

Here sd is the dissociation cross section f is theradiation flux and n is the molecular numberdensity of ice with nOH and nH the density of dis-sociation products OH and H respectively Therate constants k2OH k2H and kHOH describe thereactions above In this model the H2 formed isassumed to escape for temperatures above about20K We do not distinguish between trappedspecies and mobile species although in fact thatdistinction is important as discussed in the nextsection That is any OH produced at low tem-perature by one ion would likely be trapped priorto interacting with an OH produced by a subse-quent ion At higher temperatures this can breakdown as discussed in the text We first relate theabove set of equations to Eq 1 in the text and thenconsider precursor destruction processes

Equation 1 in the text is obtained by solvingthese equations together and then by integratingover a depth Dx Here Dx is the smaller of the pen-etration depth of the particle and the O2 escape(percolation) depth Assuming the molecular den-sity of ice n is constant with depth Dx n 5 N inEq 1 Comparing Eq 1 with Eq A3a the precur-sor formation rate sp f N is equivalent to [k2OH

nOH nOH] Dx Assuming that the processes in EqsA1a and A2a are fast then the steady state is read-ily obtained In addition if k2H k2OH there ispreferential loss of H2 at short times as suggestedby the laboratory data If steady state is reachedthe production of H2 is equal to that of H2O2 us-ing only the equations above Further one obtains

[k2OH nOH nOH] Dx R [fp sd] f N (A4)

with

fp 5 05[1 1 05 kHOH(k2H k2OH)05] (A5a)

Therefore the precursor formation cross sectionsp in Eq 1 is a fraction fp of the total dissocia-

tion cross sd Since laboratory data suggest theH2O2 fraction in an irradiated ice is small thenkHOH (k2OH k2H)05 which is likely givingfp R [(k2OH k2H)05kHOH]

Trapped versus mobile species

To examine the fluence dependence discussedin the text we note that trapped OH and freshlyproduced transiently mobile OH can be treatedas different species (eg Matich et al 1993)Therefore assuming no trapping of H the OHrate equation above would be replaced by twoequations one for the trapped species nOHt andthe other for the freshly produced mobile speciesnOH Assuming the reaction of two mobile OHunits is unlikely and that peroxide is formed fromthe interaction of a trapped OH with a subse-quently produced OH an expanded set of equa-tions is obtained

OH 1 trap R OHt (kOHt)

OH 1 OHt R H2O2 (kOHOHt)

H 1 OHt R H2O (kHOHt)

then the reaction pathways are described by

qH2

H2O 1 radiation mdashRsdH

qk2H

1 OHH

trap OH

OH RkOHt

OHt kOHOHtH2O2

kHOHkHOHt

H H2O HH2O

The resulting set of equations is

dnOHdt 5 sd f n 2 kHOH nH nOH

2 kOHt nOH nt 2 kOHOHt nOH nOHt (A1b)

dnOHtdt 5 kOHt nOH nt 2 kOHOHt nOH nOHt

2 kHOHt nH nOHt (A1b9)

dnHdt 5 sd f n 2 2 k2H nH nH 2

kHOH nH nOH 2 kHOHt nH nOHt (A2b)

JOHNSON ET AL844

dnH2O2dt 5 kOHOHt nOH nOHt

2 destruction processes (A3b)

where nt is the trap density in the irradiated iceAt low T we can assume nt is large so thatdntdt 0 simplifying the set of equations If theprecursor of interest is trapped O as discussed inthe text then a similar set of equations can be con-structed for O and Ot as in the example to followOf course a full description over a broad rangeof temperatures would require the considerationof the both trapped and mobile H O and OHthe rate of change of trap density and the vari-ous destruction processes

The above equations should be solved simul-taneously but for the incident particle fluxes typ-ically used in these studies the diffusion of mo-bile species to traps is relatively fast This is trueeven at low temperatures if there is a significantdensity of trapping sites This is usually the casefor samples formed at low T Therefore in a lowtemperature solid at low dose rates there is not asignificant steady-state background of diffusingOH There are trapped or freshly produced OHunits which rapidly react Averaging over thebombardment rate this can be approximated bywriting dnOHdt 0 and then separating nOH outin Eq A1b Assuming also that H diffuses rapidlyand either reacts or forms H2 we also writednHdt 0 on the average Therefore betweenimpacts only species trapped at defects pores orgrain surfaces are assumed to be present as sug-gested by the lack of a dependence on the inci-dent beam flux In this limit the rate of change oftrapped OH roughly simplifies to

dnOHtdt 5 2 k2H nH nH 2 2 kOHOHt nOH nOHt

It is seen that the production of peroxide dependson the loss of H by formation of H2 as discussedand at steady state the production rate for per-oxide equals the rate of formation of H2 in theseequations

Treating the mobile and trapped species sepa-rately we can further simplify the rate equationsand obtain the precursor model used in the textThat is integrating over the relevant depth Dxthe densities are replaced by column densitiesThe rate constants ki are then related to effectivecross sections si In this way cross sections aregiven as an interaction length for each freshlyproduced mobile radical [ki nj] R [si f] The

trapped OH column density NOHt and peroxideproduction rate can be estimated from ldquolinearrdquorate equations

dNOHtdt lt [sd f] N 2 2 [s2OH f] NOHt

2 [sHOH f] NOHt (A6a)

dNH2O2dt lt [s2OH f] NOHt 2 s NH2O2 (A6b)

Here sd is again the dissociation cross sectionand the destruction processes in Eq A3b are con-tained in s The second and third terms on theright in Eq A6a are the loss of OH due to the pro-duction of H2O2 and H2O In the integration ofEq A1b9 [2 kOH OHt nOH] becomes [2 s2OH f] and[kHOHt nOH] becomes [sHOH f] Therefore thedensity and the interaction length of mobile H arecontained in sHOH If H2O2 is indeed the pre-cursor then in Eq 2 in the text [sp f N] is be-comes [s2OH f NOHt] from Eq A6b Accountingfor precursor destruction which can give backtwo OH and treating H2 can both be included inthe integrated model via the si values

The simplified set of equations above allows usto examine the relative time (fluence) scales forprecursor formation (in this case H2O2) SolvingEqs A6a and A6b the column density of OH andthe concentration of precursors versus fluence cpcan be obtained Using the column of H2O2 pro-duced as the precursor column Np in Eq 2 thenO2 yield can be obtained The results can be com-pared with the results from Eqs1 and 2 in thetext The average precursor (H2O2) concentrationover the column depth of water molecules N is

cp 5 fp (sds)[1 2 exp(2s F)] 2 (ssr)

[1 2 exp(2sr F)] (A7)

with

fp 5 2s2OH(sr 2 s) (A5b)

In these equations sr 5 (2s2OH 1 sHOH) andssr 1

It is seen that the precursor concentration withtime cp depends on two rates the reaction ratesthat determine sr and the destruction rates in sAs in Eq 4a the yield can be written as

Y 5 sO2 cp N (A8)

If sr and s are comparable then the yield at small

OXIDANTS AT EUROPA 845

fluence is quadratic in F However if sr sthen the linear dependence on fluence is obtainedwith a small shift in the fluence as shown in Eq4c in the text Further the steady-state yield isnow

Y` (T) 5 sO2 [fp sd] Ns (A9)

That is sp R [fp sd] as in Eq A4 and the tem-perature dependence for the incident electrondata must be contained in [sO2 fp] or in the ratio[sO2 s2OH]sHOH]

For comparison with their precursor modelSieger et al (1998) examined the possibility of ob-taining the O2 precursor by considering reactionsfor a density of diffusing dissociation products asin Eqs A1 and A2 rather than a trapped productinteracting with a freshly formed dissociationproduct Based on the above the lowest fluence atwhich O2 is first detected experimentally sets alimit on the time scale for mobilendashmobile reactions

Effect of trap density

Equations A1b A1b9 A2b and A3b can be ap-proximately solved assuming that the dnOHdtdnHdt and dntdt are very small on the averageas discussed above In order to examine the roleof the trap density nt which can depend on thetemperature and method of formation and on theradiation history of the ice we use the slightlysimpler precursor model In this model the O pro-duced directly by dissociation traps forming aprecursor O2 is then formed when a freshly pro-duced O or other species reacts with a trappedoxygen atom Ot

H2O 1 radiation R H2 1 O (sd)

O 1 trap R Ot (kOt)

O 1 Ot R O2 (kOOt)

The reaction pathways for the oxygen species are

H2O 1 radiation mdashRsd9

trap OO2

kOtOt kOOt

H2 1 OOHt

kOR H(kHO t)

ROR

We assume for simplicity both H2 and O2 escapefrom depth Dx after formation We also allow mo-bile H to react with Ot (kHOt) removing trappedO and both the mobile H and O to be removedby other reactants R (eg OH or contaminants)This leads to a simplified set of equations not di-rectly involving OH or peroxide

dnOdt 5 sd9 f n 2 kOt nO nt 2 kOR nO nR

2 kOOt nO nOt (A1c)

dnOtdt 5 kOt nt nO 2 kOOt nO nOt

2 kHOt nH nOt (A1c9)

dnHdt 5 sd f n 2 kHOt nH nOt

2 kHR nH nR (A2c)

dnO2dt 5 kOOt nO nOt (A3c)

These equations of course act in parallel with theequations above for OH and mobile OH may re-act with trapped O to form O2 Here we focusonly on mobile O

Since the diffusion of H and O are fast and thedensity of traps is large the nH and nt averagedover the bombardment rate can be taken veryroughly as time independent Further if we as-sume the number of trapped O nOt is small com-pared with the number of sites at which H canreact then nH [sd f nkHR nR] For a low ra-diation flux f nH is small on the average as as-sumed Since the O2 yield YO2 is equal to[(dnO2dt) Dxf] with Dx the escape depth thenthe steady-state yield Y` can be obtained ana-lytically

[Y`][sd9 N] 5 1 1 (q 2 1) 2 [(q 2 1)2

1 8 q d]054 d (A10)

q 5 [kOt (nt 1 nR)] (kHOt nH)(sd9 f n) kOOt

5(sdsd9) [kOt (nt 1 nR)] kHOt[kOOt kHR nR]

d 5 05 [1 1 nt(nt 1 nR)]

We note that for nt large q can be large Assum-ing that q 1 and nt nR then Eq A10 be-comes

Y` R [sd9 N]q 5 (sd9 N) (sd9sd)

[kOOt kHR nR][kHOt kOt nt] (A5c)

JOHNSON ET AL846

Comparing with Y` in the simple precursormodel in Eq 4b then s R [(kHOt nH)f] R [sd nkHOt][kHR nR] In addition the product [sO2sp] R sd9 sd9 [kOOt n][kOt nt] That is the dis-sociation cross sections are modified by reactionrates

Quite remarkably it is seen in Eq A5c that thesteady-state yield Y` can depend inversely onthe density of traps nt in the temperature andflux regimes for which the approximations arerelevant That is if there are competing destruc-tion process for trapped O then at high trap den-sity a mobile O must find a trapped O before itbecomes trapped or before mobile H destroys theOt In this way the increasing O2 yield with in-creasing T can be related to the change in the trap-ping density nt with the formation temperatureof the ice sample As the temperature increasesand the density of traps in the bulk decreases thesurface and grain interfaces become the only trap-ping sites

ACKNOWLEDGMENTS

REJ acknowledges support by a Gledden Fel-lowship in Chemistry School of Biomedical andChemical Sciences at The University of WesternAustralia and by grants from the NSF Astron-omy Program and NASArsquos Planetary Geologyand Geophysics Program PDC acknowledgessupport by an Australian Postgraduate Award

ABBREVIATIONS

HST Hubble Space Telescope MJQ MatichndashJohnsonndashQuickenden NIMS Near Infrared Map-ping Spectrometer UV ultraviolet

REFERENCES

Ayotte P Smith RS Stevenson KP Dohnarsquolek ZKimmel GA and Kay BD (2001) Effect of porosityon the adsorption desorption trapping and release ofvolatile gases by amorphous solid water J Geophys Res106(E12) 33387ndash33392

Bahr DA Famaacute M and Baragiola RA (2001) Radiol-ysis of water ice in the outer solar system sputteringand trapping of radiation products J Geophys Res 10633285ndash33290

Baragiola RA (2003) Microporous amorphous water icefilms and astronomical implications In Water in Con-

fining Geometries edited by JP Devlin and B Buch El-sevier New York in press

Baragiola RA and Bahr DA (1998) Laboratory studiesof the optical properties and stability of oxygen onGanymede J Geophys Res 103 25865ndash25872

Baragiola RA Atteberry CL and Bahr DA (1999) Ori-gin of solid oxygen on Ganymede Reply to commentby RE Johnson on ldquoLaboratory studies of the opticalproperties and stability of oxygen on Ganymederdquo JGeophys Res E104 14183ndash14187

Baragiola RA Vidal RA Svendsen W Schou J ShiM Bahr DA and Atteberrry CL (2003) Sputteringof water ice Nucl Instrum Methods Phys Res B 209294ndash303

Bar-Nun A Herman G Laufer D and Rappaport ML(1985) Trapping and release of gases by water ice andimplications for icy bodies Icarus 63 317ndash332

Baverstock KF and Burns WG (1976) Primary pro-duction of oxygen from irradiated water as an expla-nation for decreased radiobiological oxygen enhance-ment at high LET Nature 260 316ndash318

Benit J and Brown WL (1990) Electronic sputtering ofoxygen and water molecules from thin films of waterice bombarded by MeV Ne1 ions Nucl Instrum Meth-ods Phys Rev B 46 448ndash451

Benit J Bibring J-P della-Negra S LeBeyec YMendenhall M Rocard F and Standing K (1987)Erosion of ices by ion irradiation Nucl Instrum Meth-ods Phys Rev B 1920 838ndash842

Brown WL Augustyniak WM Lanzerotti LJ John-son RE and Evatt R (1980) Linear and nonlinearprocesses in the erosion of H2O ice by fast light ionsPhys Rev Lett 43 1632ndash1635

Brown WL Augustyniak W M Simmons E Mar-cantonio KJ Lanzerotti LJ Johnson RE BoringJW Reimann CT Foti G and Pirronello V (1982)Erosion and molecular formation in condensed gasfilms by electronic energy loss of fast ions Nucl In-strum Methods B 1 307ndash314

Brown WL Augustyniak WM Marcantonio KJ Sim-mons EN Boring JW Johnson RE and ReimannCT (1984) Electronic sputtering of low-temperaturemolecular solids Nucl Instrum Methods B 1 307

Buratti B (1995) Photometry and surface structure of theicy Galilean satellites J Geophys Res 100(E9) 19061ndash19066

Buratti B and Veverka J (1983) Voyager photometry ofEuropa Icarus 55 93ndash110

Calvin WM and Spencer JR (1997) Latitudinal distri-bution of O2 on Ganymede observations with the Hub-ble Space Telescope Icarus 130 505ndash516

Calvin WM Clark RN Brown RH and Spencer JA(1995) Spectra of the icy Galilean satellites from 02 to5 mm a compilation new observations and a recentsummary J Geophys Res 100 19041ndash19048

Calvin WM Johnson RE and Spencer JA (1996) O2

on Ganymede spectral characteristics and plasma for-mation mechanisms Geophys Res Lett 23 673ndash676

Calvin WM Aninich VG and Brown RH (2004) Oxy-gen temperature and phase effects Icarus (in press)

OXIDANTS AT EUROPA 847

presence of a subsurface ocean (eg Carr et al1998) have made Europa an exciting target forfuture space exploration The molecular oxidantsproduced by radiolysis and photolysis could bespecies needed for carbon-based biochemistry inEuroparsquos putative subsurface ocean Howeverthere has been no model for the formation trap-ping and transport of relevant oxidants

In this paper we have reviewed the space ob-servations and laboratory studies related to thechemistry of formation of oxidants in ice by theincident radiation Because the laboratory dataare incomplete we examined the rate equationsfor formation of oxygen and related species byradiolysis and photolysis We showed how suchequations can be simplified to the analytic modelthat was used for fitting the fluence and temper-ature dependence of the O2 yield produced bylow-energy electrons (Sieger et al 1998) In de-riving the analytic model from the more completerate equations the competition between forma-tion and destruction processes can be studied ascan the role of trapping sites escape from depthand temperature history These equations also al-lowed us to put incident photon electron and iondata on a similar footing for the first time

The model can be used to suggest measure-ments that are needed to describe the chemicalstate of the irradiated ice and to interpret new lab-oratory data We showed that if reactions are fastthe yield data could be consistent with a precur-sor being formed by two events as is the case forperoxide However we favor trapped O as a pre-cursor as initially suggested by Matich et al(1993) for production of excited O2 Such aspecies if present could be detected spectro-scopically as can trapped O2 through bands ac-tivated by lattice perturbations (Cooper et al2003a)

The available oxygen yield data were used tocompare G values for different incident particlesThese were then used to estimate the oxygen pro-duction in the radiation environment at EuropaThe lower bounds were consistent with the pro-duction rates obtained from a recent model of Eu-roparsquos atmosphere The porosity of the regolithwas also considered It not only affects the rela-tive sputtering yields for O2 and H2O but alsoprovides an enormously increased surface areaon which the ambient gas-phase oxygen can re-act with species in the ice That this might be oc-curring is consistent with the change in re-

flectance associated with surface aging Such re-actions could also convert ambient molecularoxygen into other more refractory species that re-main trapped on subduction of Europarsquos crust

Although radiolysis produces a stable chemi-cally altered surface in the laboratory the parti-cle flux to Europarsquos surface is many orders ofmagnitude lower Therefore annealing can be ef-fective at temperatures that are much lower thanthose typically found experimentally Belowabout 120K the annealing times appear to be long(eg Baragiola 2003) so that laboratory data canoften be directly applied to Europarsquos icy surfaceHowever account has to be taken of the syner-gism between the UV and the energetic plasmaas discussed earlier (Johnson and Quickenden1997)

Transient surface melting might initiate lifeprocesses but subduction of oxidants to the pu-tative ocean is likely to be required if biochem-istry is to occur at Europa (Chyba and Hand2001) Although the temperature versus depthinto Europarsquos regolith is not well known belowthe regolith the temperature and pressure appar-ently increase with depth At some depth (ap-proximately kilometers) either very warm ice oran ocean exists (Ruiz and Tejero 2000) Thereforechemically refractory oxidants are likely to be re-quired in order to survive subduction to thewarmer subsurface regions Trapped species ifdelivered downward can become mobile andwill segregate forming inclusions (Johnson andJesser 1997) Indeed peroxide appears to segre-gate in ice (Gurman et al 1967) and these inclu-sions may remain relatively stable until the per-oxide is dissolved in the putative ocean Thevolatiles seen in reflectance such as O2 SO2 andCO2 will also segregate But at the higher tem-peratures they could also diffuse through thebulk react or percolate into the regolith atmo-sphere Therefore the observation of the trappedvolatiles in Europarsquos surface ice is of interest pri-marily because it suggests that oxidants are pre-sent The formation of oxygen-rich molecules fol-lows the preferential loss of hydrogen Howeverthe surface is in a local steady state with im-planted H O and S from the plasma and the lossto space Since the plasma flux to the surface canbe variable this balance can also vary Indeedtemporal changes in reflectance have been re-ported (Domingue and Hendrix 2004)

Based on the models for formation of O2 ex-

JOHNSON ET AL842

amined in this paper additional experiments areneeded to determine the chemical state of irra-diated ice at low fluences Of immediate inter-est is the identification of stable molecular pre-cursors required for the formation of O2 Basedon the available laboratory data individual per-oxide molecules trapped in the ice do not appearto be the principal precursors but trapped O(Matich et al 1993) remains a viable candidateAlthough the data suggest that the local hydro-gen-bonding network is important in the pro-duction of O2 the effect of the formation tem-perature and radiation damage on the structureneed to be measured It is also important to de-scribe quantitatively the relationship betweenH2 loss and the formation of O2 and H2O2 overa range of temperatures and fluences Finallyone would like to determine the steady-statedensities of biochemically interesting moleculesformed by radiolysis in an ice containing sulfurand carbon at Europarsquos surface temperaturesand then determine their stability with increas-ing temperature and pressure Such experimentsstill need to be carried out Here we have pre-sented a model that brings together the variouslaboratory data and now can be used to considerthe formation trapping and transport of relevantoxidants at Europa

APPENDIX

Introduction

In this Appendix we examine rate equationsthat describe various aspects of the radiolysis orphotolysis of ice These equations need to be un-derstood if one is to be able to calculate the for-mation trapping and transport of oxidants Theemphasis is on the formation of molecular oxy-gen We integrate these equations over the escapedepth of O2 to give the simplified rate equationsdiscussed in the text in which number densitiesare replaced by column densities (eg Eqs 1 and2) In this way we can examine the relationshipbetween the rate constants and the effective pre-cursor formation and destruction cross sectionsextracted from data We first consider a precur-sor such as H2O2 that is formed from two exci-tation events This bears on the observed linear-ity of the oxygen yield versus fluence We thenconsider the equations for trapped versus a tran-

siently mobile species We use a reaction betweena mobile and trapped O as an example althoughO 1 OH can also give oxygen Finally we ex-amine the possible dependence of the O2 forma-tion process on the density of traps (defectsvoids and grain boundaries) as a function of tem-perature The results are discussed in the text

For clarity all possible radiolytic species arenot included in the discussion below Thereforepotentially important products such as HO2 andthe charged species (eg H3O1 the trapped elec-tron etc) are missing In each section we attemptto make a point related to production of O2 andother oxidants by using a subset of the productsrather than attempt a complete description withmany unknown rate constants

Fast versus slow processes

To describe the O2 production rate from a pre-cursor such as H2O2 we first use a simple but in-structive set of rate equations to describe the for-mation of H2O2 We then relate these to thesimplified rate equation for the precursor in thetext Eq 1 and show how the postulated precur-sor formation cross section sp is related to theH2O dissociation cross section sd by the reactionrates

Using the reactions

H2O 1 radiation R H 1 OH (sd)

H 1 H R H2 (k2H)

H 1 OH R H2O (kHOH)

OH 1 OH R H2O2 (k2OH)

the reaction pathways are described by

H2

Hqk2H

1 OHH2O 1 radiation mdashRsd H

OHH2O2

OH Rk2OH

kHOH

HH2O

OXIDANTS AT EUROPA 843

The corresponding rate equations in an irradiatedvolume are

dnOHdt 5 sd f n 2 2 k2OH nOH nOH

2 kHOH nH nOH (A1a)

dnHdt 5 sd f n 2 2 k2H nH nH

2 kHOH nH nOH (A2a)

dnH2O2dt 5 k2OH nOH nOH

2 destruction processes (A3a)

Here sd is the dissociation cross section f is theradiation flux and n is the molecular numberdensity of ice with nOH and nH the density of dis-sociation products OH and H respectively Therate constants k2OH k2H and kHOH describe thereactions above In this model the H2 formed isassumed to escape for temperatures above about20K We do not distinguish between trappedspecies and mobile species although in fact thatdistinction is important as discussed in the nextsection That is any OH produced at low tem-perature by one ion would likely be trapped priorto interacting with an OH produced by a subse-quent ion At higher temperatures this can breakdown as discussed in the text We first relate theabove set of equations to Eq 1 in the text and thenconsider precursor destruction processes

Equation 1 in the text is obtained by solvingthese equations together and then by integratingover a depth Dx Here Dx is the smaller of the pen-etration depth of the particle and the O2 escape(percolation) depth Assuming the molecular den-sity of ice n is constant with depth Dx n 5 N inEq 1 Comparing Eq 1 with Eq A3a the precur-sor formation rate sp f N is equivalent to [k2OH

nOH nOH] Dx Assuming that the processes in EqsA1a and A2a are fast then the steady state is read-ily obtained In addition if k2H k2OH there ispreferential loss of H2 at short times as suggestedby the laboratory data If steady state is reachedthe production of H2 is equal to that of H2O2 us-ing only the equations above Further one obtains

[k2OH nOH nOH] Dx R [fp sd] f N (A4)

with

fp 5 05[1 1 05 kHOH(k2H k2OH)05] (A5a)

Therefore the precursor formation cross sectionsp in Eq 1 is a fraction fp of the total dissocia-

tion cross sd Since laboratory data suggest theH2O2 fraction in an irradiated ice is small thenkHOH (k2OH k2H)05 which is likely givingfp R [(k2OH k2H)05kHOH]

Trapped versus mobile species

To examine the fluence dependence discussedin the text we note that trapped OH and freshlyproduced transiently mobile OH can be treatedas different species (eg Matich et al 1993)Therefore assuming no trapping of H the OHrate equation above would be replaced by twoequations one for the trapped species nOHt andthe other for the freshly produced mobile speciesnOH Assuming the reaction of two mobile OHunits is unlikely and that peroxide is formed fromthe interaction of a trapped OH with a subse-quently produced OH an expanded set of equa-tions is obtained

OH 1 trap R OHt (kOHt)

OH 1 OHt R H2O2 (kOHOHt)

H 1 OHt R H2O (kHOHt)

then the reaction pathways are described by

qH2

H2O 1 radiation mdashRsdH

qk2H

1 OHH

trap OH

OH RkOHt

OHt kOHOHtH2O2

kHOHkHOHt

H H2O HH2O

The resulting set of equations is

dnOHdt 5 sd f n 2 kHOH nH nOH

2 kOHt nOH nt 2 kOHOHt nOH nOHt (A1b)

dnOHtdt 5 kOHt nOH nt 2 kOHOHt nOH nOHt

2 kHOHt nH nOHt (A1b9)

dnHdt 5 sd f n 2 2 k2H nH nH 2

kHOH nH nOH 2 kHOHt nH nOHt (A2b)

JOHNSON ET AL844

dnH2O2dt 5 kOHOHt nOH nOHt

2 destruction processes (A3b)

where nt is the trap density in the irradiated iceAt low T we can assume nt is large so thatdntdt 0 simplifying the set of equations If theprecursor of interest is trapped O as discussed inthe text then a similar set of equations can be con-structed for O and Ot as in the example to followOf course a full description over a broad rangeof temperatures would require the considerationof the both trapped and mobile H O and OHthe rate of change of trap density and the vari-ous destruction processes

The above equations should be solved simul-taneously but for the incident particle fluxes typ-ically used in these studies the diffusion of mo-bile species to traps is relatively fast This is trueeven at low temperatures if there is a significantdensity of trapping sites This is usually the casefor samples formed at low T Therefore in a lowtemperature solid at low dose rates there is not asignificant steady-state background of diffusingOH There are trapped or freshly produced OHunits which rapidly react Averaging over thebombardment rate this can be approximated bywriting dnOHdt 0 and then separating nOH outin Eq A1b Assuming also that H diffuses rapidlyand either reacts or forms H2 we also writednHdt 0 on the average Therefore betweenimpacts only species trapped at defects pores orgrain surfaces are assumed to be present as sug-gested by the lack of a dependence on the inci-dent beam flux In this limit the rate of change oftrapped OH roughly simplifies to

dnOHtdt 5 2 k2H nH nH 2 2 kOHOHt nOH nOHt

It is seen that the production of peroxide dependson the loss of H by formation of H2 as discussedand at steady state the production rate for per-oxide equals the rate of formation of H2 in theseequations

Treating the mobile and trapped species sepa-rately we can further simplify the rate equationsand obtain the precursor model used in the textThat is integrating over the relevant depth Dxthe densities are replaced by column densitiesThe rate constants ki are then related to effectivecross sections si In this way cross sections aregiven as an interaction length for each freshlyproduced mobile radical [ki nj] R [si f] The

trapped OH column density NOHt and peroxideproduction rate can be estimated from ldquolinearrdquorate equations

dNOHtdt lt [sd f] N 2 2 [s2OH f] NOHt

2 [sHOH f] NOHt (A6a)

dNH2O2dt lt [s2OH f] NOHt 2 s NH2O2 (A6b)

Here sd is again the dissociation cross sectionand the destruction processes in Eq A3b are con-tained in s The second and third terms on theright in Eq A6a are the loss of OH due to the pro-duction of H2O2 and H2O In the integration ofEq A1b9 [2 kOH OHt nOH] becomes [2 s2OH f] and[kHOHt nOH] becomes [sHOH f] Therefore thedensity and the interaction length of mobile H arecontained in sHOH If H2O2 is indeed the pre-cursor then in Eq 2 in the text [sp f N] is be-comes [s2OH f NOHt] from Eq A6b Accountingfor precursor destruction which can give backtwo OH and treating H2 can both be included inthe integrated model via the si values

The simplified set of equations above allows usto examine the relative time (fluence) scales forprecursor formation (in this case H2O2) SolvingEqs A6a and A6b the column density of OH andthe concentration of precursors versus fluence cpcan be obtained Using the column of H2O2 pro-duced as the precursor column Np in Eq 2 thenO2 yield can be obtained The results can be com-pared with the results from Eqs1 and 2 in thetext The average precursor (H2O2) concentrationover the column depth of water molecules N is

cp 5 fp (sds)[1 2 exp(2s F)] 2 (ssr)

[1 2 exp(2sr F)] (A7)

with

fp 5 2s2OH(sr 2 s) (A5b)

In these equations sr 5 (2s2OH 1 sHOH) andssr 1

It is seen that the precursor concentration withtime cp depends on two rates the reaction ratesthat determine sr and the destruction rates in sAs in Eq 4a the yield can be written as

Y 5 sO2 cp N (A8)

If sr and s are comparable then the yield at small

OXIDANTS AT EUROPA 845

fluence is quadratic in F However if sr sthen the linear dependence on fluence is obtainedwith a small shift in the fluence as shown in Eq4c in the text Further the steady-state yield isnow

Y` (T) 5 sO2 [fp sd] Ns (A9)

That is sp R [fp sd] as in Eq A4 and the tem-perature dependence for the incident electrondata must be contained in [sO2 fp] or in the ratio[sO2 s2OH]sHOH]

For comparison with their precursor modelSieger et al (1998) examined the possibility of ob-taining the O2 precursor by considering reactionsfor a density of diffusing dissociation products asin Eqs A1 and A2 rather than a trapped productinteracting with a freshly formed dissociationproduct Based on the above the lowest fluence atwhich O2 is first detected experimentally sets alimit on the time scale for mobilendashmobile reactions

Effect of trap density

Equations A1b A1b9 A2b and A3b can be ap-proximately solved assuming that the dnOHdtdnHdt and dntdt are very small on the averageas discussed above In order to examine the roleof the trap density nt which can depend on thetemperature and method of formation and on theradiation history of the ice we use the slightlysimpler precursor model In this model the O pro-duced directly by dissociation traps forming aprecursor O2 is then formed when a freshly pro-duced O or other species reacts with a trappedoxygen atom Ot

H2O 1 radiation R H2 1 O (sd)

O 1 trap R Ot (kOt)

O 1 Ot R O2 (kOOt)

The reaction pathways for the oxygen species are

H2O 1 radiation mdashRsd9

trap OO2

kOtOt kOOt

H2 1 OOHt

kOR H(kHO t)

ROR

We assume for simplicity both H2 and O2 escapefrom depth Dx after formation We also allow mo-bile H to react with Ot (kHOt) removing trappedO and both the mobile H and O to be removedby other reactants R (eg OH or contaminants)This leads to a simplified set of equations not di-rectly involving OH or peroxide

dnOdt 5 sd9 f n 2 kOt nO nt 2 kOR nO nR

2 kOOt nO nOt (A1c)

dnOtdt 5 kOt nt nO 2 kOOt nO nOt

2 kHOt nH nOt (A1c9)

dnHdt 5 sd f n 2 kHOt nH nOt

2 kHR nH nR (A2c)

dnO2dt 5 kOOt nO nOt (A3c)

These equations of course act in parallel with theequations above for OH and mobile OH may re-act with trapped O to form O2 Here we focusonly on mobile O

Since the diffusion of H and O are fast and thedensity of traps is large the nH and nt averagedover the bombardment rate can be taken veryroughly as time independent Further if we as-sume the number of trapped O nOt is small com-pared with the number of sites at which H canreact then nH [sd f nkHR nR] For a low ra-diation flux f nH is small on the average as as-sumed Since the O2 yield YO2 is equal to[(dnO2dt) Dxf] with Dx the escape depth thenthe steady-state yield Y` can be obtained ana-lytically

[Y`][sd9 N] 5 1 1 (q 2 1) 2 [(q 2 1)2

1 8 q d]054 d (A10)

q 5 [kOt (nt 1 nR)] (kHOt nH)(sd9 f n) kOOt

5(sdsd9) [kOt (nt 1 nR)] kHOt[kOOt kHR nR]

d 5 05 [1 1 nt(nt 1 nR)]

We note that for nt large q can be large Assum-ing that q 1 and nt nR then Eq A10 be-comes

Y` R [sd9 N]q 5 (sd9 N) (sd9sd)

[kOOt kHR nR][kHOt kOt nt] (A5c)

JOHNSON ET AL846

Comparing with Y` in the simple precursormodel in Eq 4b then s R [(kHOt nH)f] R [sd nkHOt][kHR nR] In addition the product [sO2sp] R sd9 sd9 [kOOt n][kOt nt] That is the dis-sociation cross sections are modified by reactionrates

Quite remarkably it is seen in Eq A5c that thesteady-state yield Y` can depend inversely onthe density of traps nt in the temperature andflux regimes for which the approximations arerelevant That is if there are competing destruc-tion process for trapped O then at high trap den-sity a mobile O must find a trapped O before itbecomes trapped or before mobile H destroys theOt In this way the increasing O2 yield with in-creasing T can be related to the change in the trap-ping density nt with the formation temperatureof the ice sample As the temperature increasesand the density of traps in the bulk decreases thesurface and grain interfaces become the only trap-ping sites

ACKNOWLEDGMENTS

REJ acknowledges support by a Gledden Fel-lowship in Chemistry School of Biomedical andChemical Sciences at The University of WesternAustralia and by grants from the NSF Astron-omy Program and NASArsquos Planetary Geologyand Geophysics Program PDC acknowledgessupport by an Australian Postgraduate Award

ABBREVIATIONS

HST Hubble Space Telescope MJQ MatichndashJohnsonndashQuickenden NIMS Near Infrared Map-ping Spectrometer UV ultraviolet

REFERENCES

Ayotte P Smith RS Stevenson KP Dohnarsquolek ZKimmel GA and Kay BD (2001) Effect of porosityon the adsorption desorption trapping and release ofvolatile gases by amorphous solid water J Geophys Res106(E12) 33387ndash33392

Bahr DA Famaacute M and Baragiola RA (2001) Radiol-ysis of water ice in the outer solar system sputteringand trapping of radiation products J Geophys Res 10633285ndash33290

Baragiola RA (2003) Microporous amorphous water icefilms and astronomical implications In Water in Con-

fining Geometries edited by JP Devlin and B Buch El-sevier New York in press

Baragiola RA and Bahr DA (1998) Laboratory studiesof the optical properties and stability of oxygen onGanymede J Geophys Res 103 25865ndash25872

Baragiola RA Atteberry CL and Bahr DA (1999) Ori-gin of solid oxygen on Ganymede Reply to commentby RE Johnson on ldquoLaboratory studies of the opticalproperties and stability of oxygen on Ganymederdquo JGeophys Res E104 14183ndash14187

Baragiola RA Vidal RA Svendsen W Schou J ShiM Bahr DA and Atteberrry CL (2003) Sputteringof water ice Nucl Instrum Methods Phys Res B 209294ndash303

Bar-Nun A Herman G Laufer D and Rappaport ML(1985) Trapping and release of gases by water ice andimplications for icy bodies Icarus 63 317ndash332

Baverstock KF and Burns WG (1976) Primary pro-duction of oxygen from irradiated water as an expla-nation for decreased radiobiological oxygen enhance-ment at high LET Nature 260 316ndash318

Benit J and Brown WL (1990) Electronic sputtering ofoxygen and water molecules from thin films of waterice bombarded by MeV Ne1 ions Nucl Instrum Meth-ods Phys Rev B 46 448ndash451

Benit J Bibring J-P della-Negra S LeBeyec YMendenhall M Rocard F and Standing K (1987)Erosion of ices by ion irradiation Nucl Instrum Meth-ods Phys Rev B 1920 838ndash842

Brown WL Augustyniak WM Lanzerotti LJ John-son RE and Evatt R (1980) Linear and nonlinearprocesses in the erosion of H2O ice by fast light ionsPhys Rev Lett 43 1632ndash1635

Brown WL Augustyniak W M Simmons E Mar-cantonio KJ Lanzerotti LJ Johnson RE BoringJW Reimann CT Foti G and Pirronello V (1982)Erosion and molecular formation in condensed gasfilms by electronic energy loss of fast ions Nucl In-strum Methods B 1 307ndash314

Brown WL Augustyniak WM Marcantonio KJ Sim-mons EN Boring JW Johnson RE and ReimannCT (1984) Electronic sputtering of low-temperaturemolecular solids Nucl Instrum Methods B 1 307

Buratti B (1995) Photometry and surface structure of theicy Galilean satellites J Geophys Res 100(E9) 19061ndash19066

Buratti B and Veverka J (1983) Voyager photometry ofEuropa Icarus 55 93ndash110

Calvin WM and Spencer JR (1997) Latitudinal distri-bution of O2 on Ganymede observations with the Hub-ble Space Telescope Icarus 130 505ndash516

Calvin WM Clark RN Brown RH and Spencer JA(1995) Spectra of the icy Galilean satellites from 02 to5 mm a compilation new observations and a recentsummary J Geophys Res 100 19041ndash19048

Calvin WM Johnson RE and Spencer JA (1996) O2

on Ganymede spectral characteristics and plasma for-mation mechanisms Geophys Res Lett 23 673ndash676

Calvin WM Aninich VG and Brown RH (2004) Oxy-gen temperature and phase effects Icarus (in press)

OXIDANTS AT EUROPA 847

amined in this paper additional experiments areneeded to determine the chemical state of irra-diated ice at low fluences Of immediate inter-est is the identification of stable molecular pre-cursors required for the formation of O2 Basedon the available laboratory data individual per-oxide molecules trapped in the ice do not appearto be the principal precursors but trapped O(Matich et al 1993) remains a viable candidateAlthough the data suggest that the local hydro-gen-bonding network is important in the pro-duction of O2 the effect of the formation tem-perature and radiation damage on the structureneed to be measured It is also important to de-scribe quantitatively the relationship betweenH2 loss and the formation of O2 and H2O2 overa range of temperatures and fluences Finallyone would like to determine the steady-statedensities of biochemically interesting moleculesformed by radiolysis in an ice containing sulfurand carbon at Europarsquos surface temperaturesand then determine their stability with increas-ing temperature and pressure Such experimentsstill need to be carried out Here we have pre-sented a model that brings together the variouslaboratory data and now can be used to considerthe formation trapping and transport of relevantoxidants at Europa

APPENDIX

Introduction

In this Appendix we examine rate equationsthat describe various aspects of the radiolysis orphotolysis of ice These equations need to be un-derstood if one is to be able to calculate the for-mation trapping and transport of oxidants Theemphasis is on the formation of molecular oxy-gen We integrate these equations over the escapedepth of O2 to give the simplified rate equationsdiscussed in the text in which number densitiesare replaced by column densities (eg Eqs 1 and2) In this way we can examine the relationshipbetween the rate constants and the effective pre-cursor formation and destruction cross sectionsextracted from data We first consider a precur-sor such as H2O2 that is formed from two exci-tation events This bears on the observed linear-ity of the oxygen yield versus fluence We thenconsider the equations for trapped versus a tran-

siently mobile species We use a reaction betweena mobile and trapped O as an example althoughO 1 OH can also give oxygen Finally we ex-amine the possible dependence of the O2 forma-tion process on the density of traps (defectsvoids and grain boundaries) as a function of tem-perature The results are discussed in the text

For clarity all possible radiolytic species arenot included in the discussion below Thereforepotentially important products such as HO2 andthe charged species (eg H3O1 the trapped elec-tron etc) are missing In each section we attemptto make a point related to production of O2 andother oxidants by using a subset of the productsrather than attempt a complete description withmany unknown rate constants

Fast versus slow processes

To describe the O2 production rate from a pre-cursor such as H2O2 we first use a simple but in-structive set of rate equations to describe the for-mation of H2O2 We then relate these to thesimplified rate equation for the precursor in thetext Eq 1 and show how the postulated precur-sor formation cross section sp is related to theH2O dissociation cross section sd by the reactionrates

Using the reactions

H2O 1 radiation R H 1 OH (sd)

H 1 H R H2 (k2H)

H 1 OH R H2O (kHOH)

OH 1 OH R H2O2 (k2OH)

the reaction pathways are described by

H2

Hqk2H

1 OHH2O 1 radiation mdashRsd H

OHH2O2

OH Rk2OH

kHOH

HH2O

OXIDANTS AT EUROPA 843

The corresponding rate equations in an irradiatedvolume are

dnOHdt 5 sd f n 2 2 k2OH nOH nOH

2 kHOH nH nOH (A1a)

dnHdt 5 sd f n 2 2 k2H nH nH

2 kHOH nH nOH (A2a)

dnH2O2dt 5 k2OH nOH nOH

2 destruction processes (A3a)

Here sd is the dissociation cross section f is theradiation flux and n is the molecular numberdensity of ice with nOH and nH the density of dis-sociation products OH and H respectively Therate constants k2OH k2H and kHOH describe thereactions above In this model the H2 formed isassumed to escape for temperatures above about20K We do not distinguish between trappedspecies and mobile species although in fact thatdistinction is important as discussed in the nextsection That is any OH produced at low tem-perature by one ion would likely be trapped priorto interacting with an OH produced by a subse-quent ion At higher temperatures this can breakdown as discussed in the text We first relate theabove set of equations to Eq 1 in the text and thenconsider precursor destruction processes

Equation 1 in the text is obtained by solvingthese equations together and then by integratingover a depth Dx Here Dx is the smaller of the pen-etration depth of the particle and the O2 escape(percolation) depth Assuming the molecular den-sity of ice n is constant with depth Dx n 5 N inEq 1 Comparing Eq 1 with Eq A3a the precur-sor formation rate sp f N is equivalent to [k2OH

nOH nOH] Dx Assuming that the processes in EqsA1a and A2a are fast then the steady state is read-ily obtained In addition if k2H k2OH there ispreferential loss of H2 at short times as suggestedby the laboratory data If steady state is reachedthe production of H2 is equal to that of H2O2 us-ing only the equations above Further one obtains

[k2OH nOH nOH] Dx R [fp sd] f N (A4)

with

fp 5 05[1 1 05 kHOH(k2H k2OH)05] (A5a)

Therefore the precursor formation cross sectionsp in Eq 1 is a fraction fp of the total dissocia-

tion cross sd Since laboratory data suggest theH2O2 fraction in an irradiated ice is small thenkHOH (k2OH k2H)05 which is likely givingfp R [(k2OH k2H)05kHOH]

Trapped versus mobile species

To examine the fluence dependence discussedin the text we note that trapped OH and freshlyproduced transiently mobile OH can be treatedas different species (eg Matich et al 1993)Therefore assuming no trapping of H the OHrate equation above would be replaced by twoequations one for the trapped species nOHt andthe other for the freshly produced mobile speciesnOH Assuming the reaction of two mobile OHunits is unlikely and that peroxide is formed fromthe interaction of a trapped OH with a subse-quently produced OH an expanded set of equa-tions is obtained

OH 1 trap R OHt (kOHt)

OH 1 OHt R H2O2 (kOHOHt)

H 1 OHt R H2O (kHOHt)

then the reaction pathways are described by

qH2

H2O 1 radiation mdashRsdH

qk2H

1 OHH

trap OH

OH RkOHt

OHt kOHOHtH2O2

kHOHkHOHt

H H2O HH2O

The resulting set of equations is

dnOHdt 5 sd f n 2 kHOH nH nOH

2 kOHt nOH nt 2 kOHOHt nOH nOHt (A1b)

dnOHtdt 5 kOHt nOH nt 2 kOHOHt nOH nOHt

2 kHOHt nH nOHt (A1b9)

dnHdt 5 sd f n 2 2 k2H nH nH 2

kHOH nH nOH 2 kHOHt nH nOHt (A2b)

JOHNSON ET AL844

dnH2O2dt 5 kOHOHt nOH nOHt

2 destruction processes (A3b)

where nt is the trap density in the irradiated iceAt low T we can assume nt is large so thatdntdt 0 simplifying the set of equations If theprecursor of interest is trapped O as discussed inthe text then a similar set of equations can be con-structed for O and Ot as in the example to followOf course a full description over a broad rangeof temperatures would require the considerationof the both trapped and mobile H O and OHthe rate of change of trap density and the vari-ous destruction processes

The above equations should be solved simul-taneously but for the incident particle fluxes typ-ically used in these studies the diffusion of mo-bile species to traps is relatively fast This is trueeven at low temperatures if there is a significantdensity of trapping sites This is usually the casefor samples formed at low T Therefore in a lowtemperature solid at low dose rates there is not asignificant steady-state background of diffusingOH There are trapped or freshly produced OHunits which rapidly react Averaging over thebombardment rate this can be approximated bywriting dnOHdt 0 and then separating nOH outin Eq A1b Assuming also that H diffuses rapidlyand either reacts or forms H2 we also writednHdt 0 on the average Therefore betweenimpacts only species trapped at defects pores orgrain surfaces are assumed to be present as sug-gested by the lack of a dependence on the inci-dent beam flux In this limit the rate of change oftrapped OH roughly simplifies to

dnOHtdt 5 2 k2H nH nH 2 2 kOHOHt nOH nOHt

It is seen that the production of peroxide dependson the loss of H by formation of H2 as discussedand at steady state the production rate for per-oxide equals the rate of formation of H2 in theseequations

Treating the mobile and trapped species sepa-rately we can further simplify the rate equationsand obtain the precursor model used in the textThat is integrating over the relevant depth Dxthe densities are replaced by column densitiesThe rate constants ki are then related to effectivecross sections si In this way cross sections aregiven as an interaction length for each freshlyproduced mobile radical [ki nj] R [si f] The

trapped OH column density NOHt and peroxideproduction rate can be estimated from ldquolinearrdquorate equations

dNOHtdt lt [sd f] N 2 2 [s2OH f] NOHt

2 [sHOH f] NOHt (A6a)

dNH2O2dt lt [s2OH f] NOHt 2 s NH2O2 (A6b)

Here sd is again the dissociation cross sectionand the destruction processes in Eq A3b are con-tained in s The second and third terms on theright in Eq A6a are the loss of OH due to the pro-duction of H2O2 and H2O In the integration ofEq A1b9 [2 kOH OHt nOH] becomes [2 s2OH f] and[kHOHt nOH] becomes [sHOH f] Therefore thedensity and the interaction length of mobile H arecontained in sHOH If H2O2 is indeed the pre-cursor then in Eq 2 in the text [sp f N] is be-comes [s2OH f NOHt] from Eq A6b Accountingfor precursor destruction which can give backtwo OH and treating H2 can both be included inthe integrated model via the si values

The simplified set of equations above allows usto examine the relative time (fluence) scales forprecursor formation (in this case H2O2) SolvingEqs A6a and A6b the column density of OH andthe concentration of precursors versus fluence cpcan be obtained Using the column of H2O2 pro-duced as the precursor column Np in Eq 2 thenO2 yield can be obtained The results can be com-pared with the results from Eqs1 and 2 in thetext The average precursor (H2O2) concentrationover the column depth of water molecules N is

cp 5 fp (sds)[1 2 exp(2s F)] 2 (ssr)

[1 2 exp(2sr F)] (A7)

with

fp 5 2s2OH(sr 2 s) (A5b)

In these equations sr 5 (2s2OH 1 sHOH) andssr 1

It is seen that the precursor concentration withtime cp depends on two rates the reaction ratesthat determine sr and the destruction rates in sAs in Eq 4a the yield can be written as

Y 5 sO2 cp N (A8)

If sr and s are comparable then the yield at small

OXIDANTS AT EUROPA 845

fluence is quadratic in F However if sr sthen the linear dependence on fluence is obtainedwith a small shift in the fluence as shown in Eq4c in the text Further the steady-state yield isnow

Y` (T) 5 sO2 [fp sd] Ns (A9)

That is sp R [fp sd] as in Eq A4 and the tem-perature dependence for the incident electrondata must be contained in [sO2 fp] or in the ratio[sO2 s2OH]sHOH]

For comparison with their precursor modelSieger et al (1998) examined the possibility of ob-taining the O2 precursor by considering reactionsfor a density of diffusing dissociation products asin Eqs A1 and A2 rather than a trapped productinteracting with a freshly formed dissociationproduct Based on the above the lowest fluence atwhich O2 is first detected experimentally sets alimit on the time scale for mobilendashmobile reactions

Effect of trap density

Equations A1b A1b9 A2b and A3b can be ap-proximately solved assuming that the dnOHdtdnHdt and dntdt are very small on the averageas discussed above In order to examine the roleof the trap density nt which can depend on thetemperature and method of formation and on theradiation history of the ice we use the slightlysimpler precursor model In this model the O pro-duced directly by dissociation traps forming aprecursor O2 is then formed when a freshly pro-duced O or other species reacts with a trappedoxygen atom Ot

H2O 1 radiation R H2 1 O (sd)

O 1 trap R Ot (kOt)

O 1 Ot R O2 (kOOt)

The reaction pathways for the oxygen species are

H2O 1 radiation mdashRsd9

trap OO2

kOtOt kOOt

H2 1 OOHt

kOR H(kHO t)

ROR

We assume for simplicity both H2 and O2 escapefrom depth Dx after formation We also allow mo-bile H to react with Ot (kHOt) removing trappedO and both the mobile H and O to be removedby other reactants R (eg OH or contaminants)This leads to a simplified set of equations not di-rectly involving OH or peroxide

dnOdt 5 sd9 f n 2 kOt nO nt 2 kOR nO nR

2 kOOt nO nOt (A1c)

dnOtdt 5 kOt nt nO 2 kOOt nO nOt

2 kHOt nH nOt (A1c9)

dnHdt 5 sd f n 2 kHOt nH nOt

2 kHR nH nR (A2c)

dnO2dt 5 kOOt nO nOt (A3c)

These equations of course act in parallel with theequations above for OH and mobile OH may re-act with trapped O to form O2 Here we focusonly on mobile O

Since the diffusion of H and O are fast and thedensity of traps is large the nH and nt averagedover the bombardment rate can be taken veryroughly as time independent Further if we as-sume the number of trapped O nOt is small com-pared with the number of sites at which H canreact then nH [sd f nkHR nR] For a low ra-diation flux f nH is small on the average as as-sumed Since the O2 yield YO2 is equal to[(dnO2dt) Dxf] with Dx the escape depth thenthe steady-state yield Y` can be obtained ana-lytically

[Y`][sd9 N] 5 1 1 (q 2 1) 2 [(q 2 1)2

1 8 q d]054 d (A10)

q 5 [kOt (nt 1 nR)] (kHOt nH)(sd9 f n) kOOt

5(sdsd9) [kOt (nt 1 nR)] kHOt[kOOt kHR nR]

d 5 05 [1 1 nt(nt 1 nR)]

We note that for nt large q can be large Assum-ing that q 1 and nt nR then Eq A10 be-comes

Y` R [sd9 N]q 5 (sd9 N) (sd9sd)

[kOOt kHR nR][kHOt kOt nt] (A5c)

JOHNSON ET AL846

Comparing with Y` in the simple precursormodel in Eq 4b then s R [(kHOt nH)f] R [sd nkHOt][kHR nR] In addition the product [sO2sp] R sd9 sd9 [kOOt n][kOt nt] That is the dis-sociation cross sections are modified by reactionrates

Quite remarkably it is seen in Eq A5c that thesteady-state yield Y` can depend inversely onthe density of traps nt in the temperature andflux regimes for which the approximations arerelevant That is if there are competing destruc-tion process for trapped O then at high trap den-sity a mobile O must find a trapped O before itbecomes trapped or before mobile H destroys theOt In this way the increasing O2 yield with in-creasing T can be related to the change in the trap-ping density nt with the formation temperatureof the ice sample As the temperature increasesand the density of traps in the bulk decreases thesurface and grain interfaces become the only trap-ping sites

ACKNOWLEDGMENTS

REJ acknowledges support by a Gledden Fel-lowship in Chemistry School of Biomedical andChemical Sciences at The University of WesternAustralia and by grants from the NSF Astron-omy Program and NASArsquos Planetary Geologyand Geophysics Program PDC acknowledgessupport by an Australian Postgraduate Award

ABBREVIATIONS

HST Hubble Space Telescope MJQ MatichndashJohnsonndashQuickenden NIMS Near Infrared Map-ping Spectrometer UV ultraviolet

REFERENCES

Ayotte P Smith RS Stevenson KP Dohnarsquolek ZKimmel GA and Kay BD (2001) Effect of porosityon the adsorption desorption trapping and release ofvolatile gases by amorphous solid water J Geophys Res106(E12) 33387ndash33392

Bahr DA Famaacute M and Baragiola RA (2001) Radiol-ysis of water ice in the outer solar system sputteringand trapping of radiation products J Geophys Res 10633285ndash33290

Baragiola RA (2003) Microporous amorphous water icefilms and astronomical implications In Water in Con-

fining Geometries edited by JP Devlin and B Buch El-sevier New York in press

Baragiola RA and Bahr DA (1998) Laboratory studiesof the optical properties and stability of oxygen onGanymede J Geophys Res 103 25865ndash25872

Baragiola RA Atteberry CL and Bahr DA (1999) Ori-gin of solid oxygen on Ganymede Reply to commentby RE Johnson on ldquoLaboratory studies of the opticalproperties and stability of oxygen on Ganymederdquo JGeophys Res E104 14183ndash14187

Baragiola RA Vidal RA Svendsen W Schou J ShiM Bahr DA and Atteberrry CL (2003) Sputteringof water ice Nucl Instrum Methods Phys Res B 209294ndash303

Bar-Nun A Herman G Laufer D and Rappaport ML(1985) Trapping and release of gases by water ice andimplications for icy bodies Icarus 63 317ndash332

Baverstock KF and Burns WG (1976) Primary pro-duction of oxygen from irradiated water as an expla-nation for decreased radiobiological oxygen enhance-ment at high LET Nature 260 316ndash318

Benit J and Brown WL (1990) Electronic sputtering ofoxygen and water molecules from thin films of waterice bombarded by MeV Ne1 ions Nucl Instrum Meth-ods Phys Rev B 46 448ndash451

Benit J Bibring J-P della-Negra S LeBeyec YMendenhall M Rocard F and Standing K (1987)Erosion of ices by ion irradiation Nucl Instrum Meth-ods Phys Rev B 1920 838ndash842

Brown WL Augustyniak WM Lanzerotti LJ John-son RE and Evatt R (1980) Linear and nonlinearprocesses in the erosion of H2O ice by fast light ionsPhys Rev Lett 43 1632ndash1635

Brown WL Augustyniak W M Simmons E Mar-cantonio KJ Lanzerotti LJ Johnson RE BoringJW Reimann CT Foti G and Pirronello V (1982)Erosion and molecular formation in condensed gasfilms by electronic energy loss of fast ions Nucl In-strum Methods B 1 307ndash314

Brown WL Augustyniak WM Marcantonio KJ Sim-mons EN Boring JW Johnson RE and ReimannCT (1984) Electronic sputtering of low-temperaturemolecular solids Nucl Instrum Methods B 1 307

Buratti B (1995) Photometry and surface structure of theicy Galilean satellites J Geophys Res 100(E9) 19061ndash19066

Buratti B and Veverka J (1983) Voyager photometry ofEuropa Icarus 55 93ndash110

Calvin WM and Spencer JR (1997) Latitudinal distri-bution of O2 on Ganymede observations with the Hub-ble Space Telescope Icarus 130 505ndash516

Calvin WM Clark RN Brown RH and Spencer JA(1995) Spectra of the icy Galilean satellites from 02 to5 mm a compilation new observations and a recentsummary J Geophys Res 100 19041ndash19048

Calvin WM Johnson RE and Spencer JA (1996) O2

on Ganymede spectral characteristics and plasma for-mation mechanisms Geophys Res Lett 23 673ndash676

Calvin WM Aninich VG and Brown RH (2004) Oxy-gen temperature and phase effects Icarus (in press)

OXIDANTS AT EUROPA 847

The corresponding rate equations in an irradiatedvolume are

dnOHdt 5 sd f n 2 2 k2OH nOH nOH

2 kHOH nH nOH (A1a)

dnHdt 5 sd f n 2 2 k2H nH nH

2 kHOH nH nOH (A2a)

dnH2O2dt 5 k2OH nOH nOH

2 destruction processes (A3a)

Here sd is the dissociation cross section f is theradiation flux and n is the molecular numberdensity of ice with nOH and nH the density of dis-sociation products OH and H respectively Therate constants k2OH k2H and kHOH describe thereactions above In this model the H2 formed isassumed to escape for temperatures above about20K We do not distinguish between trappedspecies and mobile species although in fact thatdistinction is important as discussed in the nextsection That is any OH produced at low tem-perature by one ion would likely be trapped priorto interacting with an OH produced by a subse-quent ion At higher temperatures this can breakdown as discussed in the text We first relate theabove set of equations to Eq 1 in the text and thenconsider precursor destruction processes

Equation 1 in the text is obtained by solvingthese equations together and then by integratingover a depth Dx Here Dx is the smaller of the pen-etration depth of the particle and the O2 escape(percolation) depth Assuming the molecular den-sity of ice n is constant with depth Dx n 5 N inEq 1 Comparing Eq 1 with Eq A3a the precur-sor formation rate sp f N is equivalent to [k2OH

nOH nOH] Dx Assuming that the processes in EqsA1a and A2a are fast then the steady state is read-ily obtained In addition if k2H k2OH there ispreferential loss of H2 at short times as suggestedby the laboratory data If steady state is reachedthe production of H2 is equal to that of H2O2 us-ing only the equations above Further one obtains

[k2OH nOH nOH] Dx R [fp sd] f N (A4)

with

fp 5 05[1 1 05 kHOH(k2H k2OH)05] (A5a)

Therefore the precursor formation cross sectionsp in Eq 1 is a fraction fp of the total dissocia-

tion cross sd Since laboratory data suggest theH2O2 fraction in an irradiated ice is small thenkHOH (k2OH k2H)05 which is likely givingfp R [(k2OH k2H)05kHOH]

Trapped versus mobile species

To examine the fluence dependence discussedin the text we note that trapped OH and freshlyproduced transiently mobile OH can be treatedas different species (eg Matich et al 1993)Therefore assuming no trapping of H the OHrate equation above would be replaced by twoequations one for the trapped species nOHt andthe other for the freshly produced mobile speciesnOH Assuming the reaction of two mobile OHunits is unlikely and that peroxide is formed fromthe interaction of a trapped OH with a subse-quently produced OH an expanded set of equa-tions is obtained

OH 1 trap R OHt (kOHt)

OH 1 OHt R H2O2 (kOHOHt)

H 1 OHt R H2O (kHOHt)

then the reaction pathways are described by

qH2

H2O 1 radiation mdashRsdH

qk2H

1 OHH

trap OH

OH RkOHt

OHt kOHOHtH2O2

kHOHkHOHt

H H2O HH2O

The resulting set of equations is

dnOHdt 5 sd f n 2 kHOH nH nOH

2 kOHt nOH nt 2 kOHOHt nOH nOHt (A1b)

dnOHtdt 5 kOHt nOH nt 2 kOHOHt nOH nOHt

2 kHOHt nH nOHt (A1b9)

dnHdt 5 sd f n 2 2 k2H nH nH 2

kHOH nH nOH 2 kHOHt nH nOHt (A2b)

JOHNSON ET AL844

dnH2O2dt 5 kOHOHt nOH nOHt

2 destruction processes (A3b)

where nt is the trap density in the irradiated iceAt low T we can assume nt is large so thatdntdt 0 simplifying the set of equations If theprecursor of interest is trapped O as discussed inthe text then a similar set of equations can be con-structed for O and Ot as in the example to followOf course a full description over a broad rangeof temperatures would require the considerationof the both trapped and mobile H O and OHthe rate of change of trap density and the vari-ous destruction processes

The above equations should be solved simul-taneously but for the incident particle fluxes typ-ically used in these studies the diffusion of mo-bile species to traps is relatively fast This is trueeven at low temperatures if there is a significantdensity of trapping sites This is usually the casefor samples formed at low T Therefore in a lowtemperature solid at low dose rates there is not asignificant steady-state background of diffusingOH There are trapped or freshly produced OHunits which rapidly react Averaging over thebombardment rate this can be approximated bywriting dnOHdt 0 and then separating nOH outin Eq A1b Assuming also that H diffuses rapidlyand either reacts or forms H2 we also writednHdt 0 on the average Therefore betweenimpacts only species trapped at defects pores orgrain surfaces are assumed to be present as sug-gested by the lack of a dependence on the inci-dent beam flux In this limit the rate of change oftrapped OH roughly simplifies to

dnOHtdt 5 2 k2H nH nH 2 2 kOHOHt nOH nOHt

It is seen that the production of peroxide dependson the loss of H by formation of H2 as discussedand at steady state the production rate for per-oxide equals the rate of formation of H2 in theseequations

Treating the mobile and trapped species sepa-rately we can further simplify the rate equationsand obtain the precursor model used in the textThat is integrating over the relevant depth Dxthe densities are replaced by column densitiesThe rate constants ki are then related to effectivecross sections si In this way cross sections aregiven as an interaction length for each freshlyproduced mobile radical [ki nj] R [si f] The

trapped OH column density NOHt and peroxideproduction rate can be estimated from ldquolinearrdquorate equations

dNOHtdt lt [sd f] N 2 2 [s2OH f] NOHt

2 [sHOH f] NOHt (A6a)

dNH2O2dt lt [s2OH f] NOHt 2 s NH2O2 (A6b)

Here sd is again the dissociation cross sectionand the destruction processes in Eq A3b are con-tained in s The second and third terms on theright in Eq A6a are the loss of OH due to the pro-duction of H2O2 and H2O In the integration ofEq A1b9 [2 kOH OHt nOH] becomes [2 s2OH f] and[kHOHt nOH] becomes [sHOH f] Therefore thedensity and the interaction length of mobile H arecontained in sHOH If H2O2 is indeed the pre-cursor then in Eq 2 in the text [sp f N] is be-comes [s2OH f NOHt] from Eq A6b Accountingfor precursor destruction which can give backtwo OH and treating H2 can both be included inthe integrated model via the si values

The simplified set of equations above allows usto examine the relative time (fluence) scales forprecursor formation (in this case H2O2) SolvingEqs A6a and A6b the column density of OH andthe concentration of precursors versus fluence cpcan be obtained Using the column of H2O2 pro-duced as the precursor column Np in Eq 2 thenO2 yield can be obtained The results can be com-pared with the results from Eqs1 and 2 in thetext The average precursor (H2O2) concentrationover the column depth of water molecules N is

cp 5 fp (sds)[1 2 exp(2s F)] 2 (ssr)

[1 2 exp(2sr F)] (A7)

with

fp 5 2s2OH(sr 2 s) (A5b)

In these equations sr 5 (2s2OH 1 sHOH) andssr 1

It is seen that the precursor concentration withtime cp depends on two rates the reaction ratesthat determine sr and the destruction rates in sAs in Eq 4a the yield can be written as

Y 5 sO2 cp N (A8)

If sr and s are comparable then the yield at small

OXIDANTS AT EUROPA 845

fluence is quadratic in F However if sr sthen the linear dependence on fluence is obtainedwith a small shift in the fluence as shown in Eq4c in the text Further the steady-state yield isnow

Y` (T) 5 sO2 [fp sd] Ns (A9)

That is sp R [fp sd] as in Eq A4 and the tem-perature dependence for the incident electrondata must be contained in [sO2 fp] or in the ratio[sO2 s2OH]sHOH]

For comparison with their precursor modelSieger et al (1998) examined the possibility of ob-taining the O2 precursor by considering reactionsfor a density of diffusing dissociation products asin Eqs A1 and A2 rather than a trapped productinteracting with a freshly formed dissociationproduct Based on the above the lowest fluence atwhich O2 is first detected experimentally sets alimit on the time scale for mobilendashmobile reactions

Effect of trap density

Equations A1b A1b9 A2b and A3b can be ap-proximately solved assuming that the dnOHdtdnHdt and dntdt are very small on the averageas discussed above In order to examine the roleof the trap density nt which can depend on thetemperature and method of formation and on theradiation history of the ice we use the slightlysimpler precursor model In this model the O pro-duced directly by dissociation traps forming aprecursor O2 is then formed when a freshly pro-duced O or other species reacts with a trappedoxygen atom Ot

H2O 1 radiation R H2 1 O (sd)

O 1 trap R Ot (kOt)

O 1 Ot R O2 (kOOt)

The reaction pathways for the oxygen species are

H2O 1 radiation mdashRsd9

trap OO2

kOtOt kOOt

H2 1 OOHt

kOR H(kHO t)

ROR

We assume for simplicity both H2 and O2 escapefrom depth Dx after formation We also allow mo-bile H to react with Ot (kHOt) removing trappedO and both the mobile H and O to be removedby other reactants R (eg OH or contaminants)This leads to a simplified set of equations not di-rectly involving OH or peroxide

dnOdt 5 sd9 f n 2 kOt nO nt 2 kOR nO nR

2 kOOt nO nOt (A1c)

dnOtdt 5 kOt nt nO 2 kOOt nO nOt

2 kHOt nH nOt (A1c9)

dnHdt 5 sd f n 2 kHOt nH nOt

2 kHR nH nR (A2c)

dnO2dt 5 kOOt nO nOt (A3c)

These equations of course act in parallel with theequations above for OH and mobile OH may re-act with trapped O to form O2 Here we focusonly on mobile O

Since the diffusion of H and O are fast and thedensity of traps is large the nH and nt averagedover the bombardment rate can be taken veryroughly as time independent Further if we as-sume the number of trapped O nOt is small com-pared with the number of sites at which H canreact then nH [sd f nkHR nR] For a low ra-diation flux f nH is small on the average as as-sumed Since the O2 yield YO2 is equal to[(dnO2dt) Dxf] with Dx the escape depth thenthe steady-state yield Y` can be obtained ana-lytically

[Y`][sd9 N] 5 1 1 (q 2 1) 2 [(q 2 1)2

1 8 q d]054 d (A10)

q 5 [kOt (nt 1 nR)] (kHOt nH)(sd9 f n) kOOt

5(sdsd9) [kOt (nt 1 nR)] kHOt[kOOt kHR nR]

d 5 05 [1 1 nt(nt 1 nR)]

We note that for nt large q can be large Assum-ing that q 1 and nt nR then Eq A10 be-comes

Y` R [sd9 N]q 5 (sd9 N) (sd9sd)

[kOOt kHR nR][kHOt kOt nt] (A5c)

JOHNSON ET AL846

Comparing with Y` in the simple precursormodel in Eq 4b then s R [(kHOt nH)f] R [sd nkHOt][kHR nR] In addition the product [sO2sp] R sd9 sd9 [kOOt n][kOt nt] That is the dis-sociation cross sections are modified by reactionrates

Quite remarkably it is seen in Eq A5c that thesteady-state yield Y` can depend inversely onthe density of traps nt in the temperature andflux regimes for which the approximations arerelevant That is if there are competing destruc-tion process for trapped O then at high trap den-sity a mobile O must find a trapped O before itbecomes trapped or before mobile H destroys theOt In this way the increasing O2 yield with in-creasing T can be related to the change in the trap-ping density nt with the formation temperatureof the ice sample As the temperature increasesand the density of traps in the bulk decreases thesurface and grain interfaces become the only trap-ping sites

ACKNOWLEDGMENTS

REJ acknowledges support by a Gledden Fel-lowship in Chemistry School of Biomedical andChemical Sciences at The University of WesternAustralia and by grants from the NSF Astron-omy Program and NASArsquos Planetary Geologyand Geophysics Program PDC acknowledgessupport by an Australian Postgraduate Award

ABBREVIATIONS

HST Hubble Space Telescope MJQ MatichndashJohnsonndashQuickenden NIMS Near Infrared Map-ping Spectrometer UV ultraviolet

REFERENCES

Ayotte P Smith RS Stevenson KP Dohnarsquolek ZKimmel GA and Kay BD (2001) Effect of porosityon the adsorption desorption trapping and release ofvolatile gases by amorphous solid water J Geophys Res106(E12) 33387ndash33392

Bahr DA Famaacute M and Baragiola RA (2001) Radiol-ysis of water ice in the outer solar system sputteringand trapping of radiation products J Geophys Res 10633285ndash33290

Baragiola RA (2003) Microporous amorphous water icefilms and astronomical implications In Water in Con-

fining Geometries edited by JP Devlin and B Buch El-sevier New York in press

Baragiola RA and Bahr DA (1998) Laboratory studiesof the optical properties and stability of oxygen onGanymede J Geophys Res 103 25865ndash25872

Baragiola RA Atteberry CL and Bahr DA (1999) Ori-gin of solid oxygen on Ganymede Reply to commentby RE Johnson on ldquoLaboratory studies of the opticalproperties and stability of oxygen on Ganymederdquo JGeophys Res E104 14183ndash14187

Baragiola RA Vidal RA Svendsen W Schou J ShiM Bahr DA and Atteberrry CL (2003) Sputteringof water ice Nucl Instrum Methods Phys Res B 209294ndash303

Bar-Nun A Herman G Laufer D and Rappaport ML(1985) Trapping and release of gases by water ice andimplications for icy bodies Icarus 63 317ndash332

Baverstock KF and Burns WG (1976) Primary pro-duction of oxygen from irradiated water as an expla-nation for decreased radiobiological oxygen enhance-ment at high LET Nature 260 316ndash318

Benit J and Brown WL (1990) Electronic sputtering ofoxygen and water molecules from thin films of waterice bombarded by MeV Ne1 ions Nucl Instrum Meth-ods Phys Rev B 46 448ndash451

Benit J Bibring J-P della-Negra S LeBeyec YMendenhall M Rocard F and Standing K (1987)Erosion of ices by ion irradiation Nucl Instrum Meth-ods Phys Rev B 1920 838ndash842

Brown WL Augustyniak WM Lanzerotti LJ John-son RE and Evatt R (1980) Linear and nonlinearprocesses in the erosion of H2O ice by fast light ionsPhys Rev Lett 43 1632ndash1635

Brown WL Augustyniak W M Simmons E Mar-cantonio KJ Lanzerotti LJ Johnson RE BoringJW Reimann CT Foti G and Pirronello V (1982)Erosion and molecular formation in condensed gasfilms by electronic energy loss of fast ions Nucl In-strum Methods B 1 307ndash314

Brown WL Augustyniak WM Marcantonio KJ Sim-mons EN Boring JW Johnson RE and ReimannCT (1984) Electronic sputtering of low-temperaturemolecular solids Nucl Instrum Methods B 1 307

Buratti B (1995) Photometry and surface structure of theicy Galilean satellites J Geophys Res 100(E9) 19061ndash19066

Buratti B and Veverka J (1983) Voyager photometry ofEuropa Icarus 55 93ndash110

Calvin WM and Spencer JR (1997) Latitudinal distri-bution of O2 on Ganymede observations with the Hub-ble Space Telescope Icarus 130 505ndash516

Calvin WM Clark RN Brown RH and Spencer JA(1995) Spectra of the icy Galilean satellites from 02 to5 mm a compilation new observations and a recentsummary J Geophys Res 100 19041ndash19048

Calvin WM Johnson RE and Spencer JA (1996) O2

on Ganymede spectral characteristics and plasma for-mation mechanisms Geophys Res Lett 23 673ndash676

Calvin WM Aninich VG and Brown RH (2004) Oxy-gen temperature and phase effects Icarus (in press)

OXIDANTS AT EUROPA 847

dnH2O2dt 5 kOHOHt nOH nOHt

2 destruction processes (A3b)

where nt is the trap density in the irradiated iceAt low T we can assume nt is large so thatdntdt 0 simplifying the set of equations If theprecursor of interest is trapped O as discussed inthe text then a similar set of equations can be con-structed for O and Ot as in the example to followOf course a full description over a broad rangeof temperatures would require the considerationof the both trapped and mobile H O and OHthe rate of change of trap density and the vari-ous destruction processes

The above equations should be solved simul-taneously but for the incident particle fluxes typ-ically used in these studies the diffusion of mo-bile species to traps is relatively fast This is trueeven at low temperatures if there is a significantdensity of trapping sites This is usually the casefor samples formed at low T Therefore in a lowtemperature solid at low dose rates there is not asignificant steady-state background of diffusingOH There are trapped or freshly produced OHunits which rapidly react Averaging over thebombardment rate this can be approximated bywriting dnOHdt 0 and then separating nOH outin Eq A1b Assuming also that H diffuses rapidlyand either reacts or forms H2 we also writednHdt 0 on the average Therefore betweenimpacts only species trapped at defects pores orgrain surfaces are assumed to be present as sug-gested by the lack of a dependence on the inci-dent beam flux In this limit the rate of change oftrapped OH roughly simplifies to

dnOHtdt 5 2 k2H nH nH 2 2 kOHOHt nOH nOHt

It is seen that the production of peroxide dependson the loss of H by formation of H2 as discussedand at steady state the production rate for per-oxide equals the rate of formation of H2 in theseequations

Treating the mobile and trapped species sepa-rately we can further simplify the rate equationsand obtain the precursor model used in the textThat is integrating over the relevant depth Dxthe densities are replaced by column densitiesThe rate constants ki are then related to effectivecross sections si In this way cross sections aregiven as an interaction length for each freshlyproduced mobile radical [ki nj] R [si f] The

trapped OH column density NOHt and peroxideproduction rate can be estimated from ldquolinearrdquorate equations

dNOHtdt lt [sd f] N 2 2 [s2OH f] NOHt

2 [sHOH f] NOHt (A6a)

dNH2O2dt lt [s2OH f] NOHt 2 s NH2O2 (A6b)

Here sd is again the dissociation cross sectionand the destruction processes in Eq A3b are con-tained in s The second and third terms on theright in Eq A6a are the loss of OH due to the pro-duction of H2O2 and H2O In the integration ofEq A1b9 [2 kOH OHt nOH] becomes [2 s2OH f] and[kHOHt nOH] becomes [sHOH f] Therefore thedensity and the interaction length of mobile H arecontained in sHOH If H2O2 is indeed the pre-cursor then in Eq 2 in the text [sp f N] is be-comes [s2OH f NOHt] from Eq A6b Accountingfor precursor destruction which can give backtwo OH and treating H2 can both be included inthe integrated model via the si values

The simplified set of equations above allows usto examine the relative time (fluence) scales forprecursor formation (in this case H2O2) SolvingEqs A6a and A6b the column density of OH andthe concentration of precursors versus fluence cpcan be obtained Using the column of H2O2 pro-duced as the precursor column Np in Eq 2 thenO2 yield can be obtained The results can be com-pared with the results from Eqs1 and 2 in thetext The average precursor (H2O2) concentrationover the column depth of water molecules N is

cp 5 fp (sds)[1 2 exp(2s F)] 2 (ssr)

[1 2 exp(2sr F)] (A7)

with

fp 5 2s2OH(sr 2 s) (A5b)

In these equations sr 5 (2s2OH 1 sHOH) andssr 1

It is seen that the precursor concentration withtime cp depends on two rates the reaction ratesthat determine sr and the destruction rates in sAs in Eq 4a the yield can be written as

Y 5 sO2 cp N (A8)

If sr and s are comparable then the yield at small

OXIDANTS AT EUROPA 845

fluence is quadratic in F However if sr sthen the linear dependence on fluence is obtainedwith a small shift in the fluence as shown in Eq4c in the text Further the steady-state yield isnow

Y` (T) 5 sO2 [fp sd] Ns (A9)

That is sp R [fp sd] as in Eq A4 and the tem-perature dependence for the incident electrondata must be contained in [sO2 fp] or in the ratio[sO2 s2OH]sHOH]

For comparison with their precursor modelSieger et al (1998) examined the possibility of ob-taining the O2 precursor by considering reactionsfor a density of diffusing dissociation products asin Eqs A1 and A2 rather than a trapped productinteracting with a freshly formed dissociationproduct Based on the above the lowest fluence atwhich O2 is first detected experimentally sets alimit on the time scale for mobilendashmobile reactions

Effect of trap density

Equations A1b A1b9 A2b and A3b can be ap-proximately solved assuming that the dnOHdtdnHdt and dntdt are very small on the averageas discussed above In order to examine the roleof the trap density nt which can depend on thetemperature and method of formation and on theradiation history of the ice we use the slightlysimpler precursor model In this model the O pro-duced directly by dissociation traps forming aprecursor O2 is then formed when a freshly pro-duced O or other species reacts with a trappedoxygen atom Ot

H2O 1 radiation R H2 1 O (sd)

O 1 trap R Ot (kOt)

O 1 Ot R O2 (kOOt)

The reaction pathways for the oxygen species are

H2O 1 radiation mdashRsd9

trap OO2

kOtOt kOOt

H2 1 OOHt

kOR H(kHO t)

ROR

We assume for simplicity both H2 and O2 escapefrom depth Dx after formation We also allow mo-bile H to react with Ot (kHOt) removing trappedO and both the mobile H and O to be removedby other reactants R (eg OH or contaminants)This leads to a simplified set of equations not di-rectly involving OH or peroxide

dnOdt 5 sd9 f n 2 kOt nO nt 2 kOR nO nR

2 kOOt nO nOt (A1c)

dnOtdt 5 kOt nt nO 2 kOOt nO nOt

2 kHOt nH nOt (A1c9)

dnHdt 5 sd f n 2 kHOt nH nOt

2 kHR nH nR (A2c)

dnO2dt 5 kOOt nO nOt (A3c)

These equations of course act in parallel with theequations above for OH and mobile OH may re-act with trapped O to form O2 Here we focusonly on mobile O

Since the diffusion of H and O are fast and thedensity of traps is large the nH and nt averagedover the bombardment rate can be taken veryroughly as time independent Further if we as-sume the number of trapped O nOt is small com-pared with the number of sites at which H canreact then nH [sd f nkHR nR] For a low ra-diation flux f nH is small on the average as as-sumed Since the O2 yield YO2 is equal to[(dnO2dt) Dxf] with Dx the escape depth thenthe steady-state yield Y` can be obtained ana-lytically

[Y`][sd9 N] 5 1 1 (q 2 1) 2 [(q 2 1)2

1 8 q d]054 d (A10)

q 5 [kOt (nt 1 nR)] (kHOt nH)(sd9 f n) kOOt

5(sdsd9) [kOt (nt 1 nR)] kHOt[kOOt kHR nR]

d 5 05 [1 1 nt(nt 1 nR)]

We note that for nt large q can be large Assum-ing that q 1 and nt nR then Eq A10 be-comes

Y` R [sd9 N]q 5 (sd9 N) (sd9sd)

[kOOt kHR nR][kHOt kOt nt] (A5c)

JOHNSON ET AL846

Comparing with Y` in the simple precursormodel in Eq 4b then s R [(kHOt nH)f] R [sd nkHOt][kHR nR] In addition the product [sO2sp] R sd9 sd9 [kOOt n][kOt nt] That is the dis-sociation cross sections are modified by reactionrates

Quite remarkably it is seen in Eq A5c that thesteady-state yield Y` can depend inversely onthe density of traps nt in the temperature andflux regimes for which the approximations arerelevant That is if there are competing destruc-tion process for trapped O then at high trap den-sity a mobile O must find a trapped O before itbecomes trapped or before mobile H destroys theOt In this way the increasing O2 yield with in-creasing T can be related to the change in the trap-ping density nt with the formation temperatureof the ice sample As the temperature increasesand the density of traps in the bulk decreases thesurface and grain interfaces become the only trap-ping sites

ACKNOWLEDGMENTS

REJ acknowledges support by a Gledden Fel-lowship in Chemistry School of Biomedical andChemical Sciences at The University of WesternAustralia and by grants from the NSF Astron-omy Program and NASArsquos Planetary Geologyand Geophysics Program PDC acknowledgessupport by an Australian Postgraduate Award

ABBREVIATIONS

HST Hubble Space Telescope MJQ MatichndashJohnsonndashQuickenden NIMS Near Infrared Map-ping Spectrometer UV ultraviolet

REFERENCES

Ayotte P Smith RS Stevenson KP Dohnarsquolek ZKimmel GA and Kay BD (2001) Effect of porosityon the adsorption desorption trapping and release ofvolatile gases by amorphous solid water J Geophys Res106(E12) 33387ndash33392

Bahr DA Famaacute M and Baragiola RA (2001) Radiol-ysis of water ice in the outer solar system sputteringand trapping of radiation products J Geophys Res 10633285ndash33290

Baragiola RA (2003) Microporous amorphous water icefilms and astronomical implications In Water in Con-

fining Geometries edited by JP Devlin and B Buch El-sevier New York in press

Baragiola RA and Bahr DA (1998) Laboratory studiesof the optical properties and stability of oxygen onGanymede J Geophys Res 103 25865ndash25872

Baragiola RA Atteberry CL and Bahr DA (1999) Ori-gin of solid oxygen on Ganymede Reply to commentby RE Johnson on ldquoLaboratory studies of the opticalproperties and stability of oxygen on Ganymederdquo JGeophys Res E104 14183ndash14187

Baragiola RA Vidal RA Svendsen W Schou J ShiM Bahr DA and Atteberrry CL (2003) Sputteringof water ice Nucl Instrum Methods Phys Res B 209294ndash303

Bar-Nun A Herman G Laufer D and Rappaport ML(1985) Trapping and release of gases by water ice andimplications for icy bodies Icarus 63 317ndash332

Baverstock KF and Burns WG (1976) Primary pro-duction of oxygen from irradiated water as an expla-nation for decreased radiobiological oxygen enhance-ment at high LET Nature 260 316ndash318

Benit J and Brown WL (1990) Electronic sputtering ofoxygen and water molecules from thin films of waterice bombarded by MeV Ne1 ions Nucl Instrum Meth-ods Phys Rev B 46 448ndash451

Benit J Bibring J-P della-Negra S LeBeyec YMendenhall M Rocard F and Standing K (1987)Erosion of ices by ion irradiation Nucl Instrum Meth-ods Phys Rev B 1920 838ndash842

Brown WL Augustyniak WM Lanzerotti LJ John-son RE and Evatt R (1980) Linear and nonlinearprocesses in the erosion of H2O ice by fast light ionsPhys Rev Lett 43 1632ndash1635

Brown WL Augustyniak W M Simmons E Mar-cantonio KJ Lanzerotti LJ Johnson RE BoringJW Reimann CT Foti G and Pirronello V (1982)Erosion and molecular formation in condensed gasfilms by electronic energy loss of fast ions Nucl In-strum Methods B 1 307ndash314

Brown WL Augustyniak WM Marcantonio KJ Sim-mons EN Boring JW Johnson RE and ReimannCT (1984) Electronic sputtering of low-temperaturemolecular solids Nucl Instrum Methods B 1 307

Buratti B (1995) Photometry and surface structure of theicy Galilean satellites J Geophys Res 100(E9) 19061ndash19066

Buratti B and Veverka J (1983) Voyager photometry ofEuropa Icarus 55 93ndash110

Calvin WM and Spencer JR (1997) Latitudinal distri-bution of O2 on Ganymede observations with the Hub-ble Space Telescope Icarus 130 505ndash516

Calvin WM Clark RN Brown RH and Spencer JA(1995) Spectra of the icy Galilean satellites from 02 to5 mm a compilation new observations and a recentsummary J Geophys Res 100 19041ndash19048

Calvin WM Johnson RE and Spencer JA (1996) O2

on Ganymede spectral characteristics and plasma for-mation mechanisms Geophys Res Lett 23 673ndash676

Calvin WM Aninich VG and Brown RH (2004) Oxy-gen temperature and phase effects Icarus (in press)

OXIDANTS AT EUROPA 847

fluence is quadratic in F However if sr sthen the linear dependence on fluence is obtainedwith a small shift in the fluence as shown in Eq4c in the text Further the steady-state yield isnow

Y` (T) 5 sO2 [fp sd] Ns (A9)

That is sp R [fp sd] as in Eq A4 and the tem-perature dependence for the incident electrondata must be contained in [sO2 fp] or in the ratio[sO2 s2OH]sHOH]

For comparison with their precursor modelSieger et al (1998) examined the possibility of ob-taining the O2 precursor by considering reactionsfor a density of diffusing dissociation products asin Eqs A1 and A2 rather than a trapped productinteracting with a freshly formed dissociationproduct Based on the above the lowest fluence atwhich O2 is first detected experimentally sets alimit on the time scale for mobilendashmobile reactions

Effect of trap density

Equations A1b A1b9 A2b and A3b can be ap-proximately solved assuming that the dnOHdtdnHdt and dntdt are very small on the averageas discussed above In order to examine the roleof the trap density nt which can depend on thetemperature and method of formation and on theradiation history of the ice we use the slightlysimpler precursor model In this model the O pro-duced directly by dissociation traps forming aprecursor O2 is then formed when a freshly pro-duced O or other species reacts with a trappedoxygen atom Ot

H2O 1 radiation R H2 1 O (sd)

O 1 trap R Ot (kOt)

O 1 Ot R O2 (kOOt)

The reaction pathways for the oxygen species are

H2O 1 radiation mdashRsd9

trap OO2

kOtOt kOOt

H2 1 OOHt

kOR H(kHO t)

ROR

We assume for simplicity both H2 and O2 escapefrom depth Dx after formation We also allow mo-bile H to react with Ot (kHOt) removing trappedO and both the mobile H and O to be removedby other reactants R (eg OH or contaminants)This leads to a simplified set of equations not di-rectly involving OH or peroxide

dnOdt 5 sd9 f n 2 kOt nO nt 2 kOR nO nR

2 kOOt nO nOt (A1c)

dnOtdt 5 kOt nt nO 2 kOOt nO nOt

2 kHOt nH nOt (A1c9)

dnHdt 5 sd f n 2 kHOt nH nOt

2 kHR nH nR (A2c)

dnO2dt 5 kOOt nO nOt (A3c)

These equations of course act in parallel with theequations above for OH and mobile OH may re-act with trapped O to form O2 Here we focusonly on mobile O

Since the diffusion of H and O are fast and thedensity of traps is large the nH and nt averagedover the bombardment rate can be taken veryroughly as time independent Further if we as-sume the number of trapped O nOt is small com-pared with the number of sites at which H canreact then nH [sd f nkHR nR] For a low ra-diation flux f nH is small on the average as as-sumed Since the O2 yield YO2 is equal to[(dnO2dt) Dxf] with Dx the escape depth thenthe steady-state yield Y` can be obtained ana-lytically

[Y`][sd9 N] 5 1 1 (q 2 1) 2 [(q 2 1)2

1 8 q d]054 d (A10)

q 5 [kOt (nt 1 nR)] (kHOt nH)(sd9 f n) kOOt

5(sdsd9) [kOt (nt 1 nR)] kHOt[kOOt kHR nR]

d 5 05 [1 1 nt(nt 1 nR)]

We note that for nt large q can be large Assum-ing that q 1 and nt nR then Eq A10 be-comes

Y` R [sd9 N]q 5 (sd9 N) (sd9sd)

[kOOt kHR nR][kHOt kOt nt] (A5c)

JOHNSON ET AL846

Comparing with Y` in the simple precursormodel in Eq 4b then s R [(kHOt nH)f] R [sd nkHOt][kHR nR] In addition the product [sO2sp] R sd9 sd9 [kOOt n][kOt nt] That is the dis-sociation cross sections are modified by reactionrates

Quite remarkably it is seen in Eq A5c that thesteady-state yield Y` can depend inversely onthe density of traps nt in the temperature andflux regimes for which the approximations arerelevant That is if there are competing destruc-tion process for trapped O then at high trap den-sity a mobile O must find a trapped O before itbecomes trapped or before mobile H destroys theOt In this way the increasing O2 yield with in-creasing T can be related to the change in the trap-ping density nt with the formation temperatureof the ice sample As the temperature increasesand the density of traps in the bulk decreases thesurface and grain interfaces become the only trap-ping sites

ACKNOWLEDGMENTS

REJ acknowledges support by a Gledden Fel-lowship in Chemistry School of Biomedical andChemical Sciences at The University of WesternAustralia and by grants from the NSF Astron-omy Program and NASArsquos Planetary Geologyand Geophysics Program PDC acknowledgessupport by an Australian Postgraduate Award

ABBREVIATIONS

HST Hubble Space Telescope MJQ MatichndashJohnsonndashQuickenden NIMS Near Infrared Map-ping Spectrometer UV ultraviolet

REFERENCES

Ayotte P Smith RS Stevenson KP Dohnarsquolek ZKimmel GA and Kay BD (2001) Effect of porosityon the adsorption desorption trapping and release ofvolatile gases by amorphous solid water J Geophys Res106(E12) 33387ndash33392

Bahr DA Famaacute M and Baragiola RA (2001) Radiol-ysis of water ice in the outer solar system sputteringand trapping of radiation products J Geophys Res 10633285ndash33290

Baragiola RA (2003) Microporous amorphous water icefilms and astronomical implications In Water in Con-

fining Geometries edited by JP Devlin and B Buch El-sevier New York in press

Baragiola RA and Bahr DA (1998) Laboratory studiesof the optical properties and stability of oxygen onGanymede J Geophys Res 103 25865ndash25872

Baragiola RA Atteberry CL and Bahr DA (1999) Ori-gin of solid oxygen on Ganymede Reply to commentby RE Johnson on ldquoLaboratory studies of the opticalproperties and stability of oxygen on Ganymederdquo JGeophys Res E104 14183ndash14187

Baragiola RA Vidal RA Svendsen W Schou J ShiM Bahr DA and Atteberrry CL (2003) Sputteringof water ice Nucl Instrum Methods Phys Res B 209294ndash303

Bar-Nun A Herman G Laufer D and Rappaport ML(1985) Trapping and release of gases by water ice andimplications for icy bodies Icarus 63 317ndash332

Baverstock KF and Burns WG (1976) Primary pro-duction of oxygen from irradiated water as an expla-nation for decreased radiobiological oxygen enhance-ment at high LET Nature 260 316ndash318

Benit J and Brown WL (1990) Electronic sputtering ofoxygen and water molecules from thin films of waterice bombarded by MeV Ne1 ions Nucl Instrum Meth-ods Phys Rev B 46 448ndash451

Benit J Bibring J-P della-Negra S LeBeyec YMendenhall M Rocard F and Standing K (1987)Erosion of ices by ion irradiation Nucl Instrum Meth-ods Phys Rev B 1920 838ndash842

Brown WL Augustyniak WM Lanzerotti LJ John-son RE and Evatt R (1980) Linear and nonlinearprocesses in the erosion of H2O ice by fast light ionsPhys Rev Lett 43 1632ndash1635

Brown WL Augustyniak W M Simmons E Mar-cantonio KJ Lanzerotti LJ Johnson RE BoringJW Reimann CT Foti G and Pirronello V (1982)Erosion and molecular formation in condensed gasfilms by electronic energy loss of fast ions Nucl In-strum Methods B 1 307ndash314

Brown WL Augustyniak WM Marcantonio KJ Sim-mons EN Boring JW Johnson RE and ReimannCT (1984) Electronic sputtering of low-temperaturemolecular solids Nucl Instrum Methods B 1 307

Buratti B (1995) Photometry and surface structure of theicy Galilean satellites J Geophys Res 100(E9) 19061ndash19066

Buratti B and Veverka J (1983) Voyager photometry ofEuropa Icarus 55 93ndash110

Calvin WM and Spencer JR (1997) Latitudinal distri-bution of O2 on Ganymede observations with the Hub-ble Space Telescope Icarus 130 505ndash516

Calvin WM Clark RN Brown RH and Spencer JA(1995) Spectra of the icy Galilean satellites from 02 to5 mm a compilation new observations and a recentsummary J Geophys Res 100 19041ndash19048

Calvin WM Johnson RE and Spencer JA (1996) O2

on Ganymede spectral characteristics and plasma for-mation mechanisms Geophys Res Lett 23 673ndash676

Calvin WM Aninich VG and Brown RH (2004) Oxy-gen temperature and phase effects Icarus (in press)

OXIDANTS AT EUROPA 847

Comparing with Y` in the simple precursormodel in Eq 4b then s R [(kHOt nH)f] R [sd nkHOt][kHR nR] In addition the product [sO2sp] R sd9 sd9 [kOOt n][kOt nt] That is the dis-sociation cross sections are modified by reactionrates

Quite remarkably it is seen in Eq A5c that thesteady-state yield Y` can depend inversely onthe density of traps nt in the temperature andflux regimes for which the approximations arerelevant That is if there are competing destruc-tion process for trapped O then at high trap den-sity a mobile O must find a trapped O before itbecomes trapped or before mobile H destroys theOt In this way the increasing O2 yield with in-creasing T can be related to the change in the trap-ping density nt with the formation temperatureof the ice sample As the temperature increasesand the density of traps in the bulk decreases thesurface and grain interfaces become the only trap-ping sites

ACKNOWLEDGMENTS

REJ acknowledges support by a Gledden Fel-lowship in Chemistry School of Biomedical andChemical Sciences at The University of WesternAustralia and by grants from the NSF Astron-omy Program and NASArsquos Planetary Geologyand Geophysics Program PDC acknowledgessupport by an Australian Postgraduate Award

ABBREVIATIONS

HST Hubble Space Telescope MJQ MatichndashJohnsonndashQuickenden NIMS Near Infrared Map-ping Spectrometer UV ultraviolet

REFERENCES

Ayotte P Smith RS Stevenson KP Dohnarsquolek ZKimmel GA and Kay BD (2001) Effect of porosityon the adsorption desorption trapping and release ofvolatile gases by amorphous solid water J Geophys Res106(E12) 33387ndash33392

Bahr DA Famaacute M and Baragiola RA (2001) Radiol-ysis of water ice in the outer solar system sputteringand trapping of radiation products J Geophys Res 10633285ndash33290

Baragiola RA (2003) Microporous amorphous water icefilms and astronomical implications In Water in Con-

fining Geometries edited by JP Devlin and B Buch El-sevier New York in press

Baragiola RA and Bahr DA (1998) Laboratory studiesof the optical properties and stability of oxygen onGanymede J Geophys Res 103 25865ndash25872

Baragiola RA Atteberry CL and Bahr DA (1999) Ori-gin of solid oxygen on Ganymede Reply to commentby RE Johnson on ldquoLaboratory studies of the opticalproperties and stability of oxygen on Ganymederdquo JGeophys Res E104 14183ndash14187

Baragiola RA Vidal RA Svendsen W Schou J ShiM Bahr DA and Atteberrry CL (2003) Sputteringof water ice Nucl Instrum Methods Phys Res B 209294ndash303

Bar-Nun A Herman G Laufer D and Rappaport ML(1985) Trapping and release of gases by water ice andimplications for icy bodies Icarus 63 317ndash332

Baverstock KF and Burns WG (1976) Primary pro-duction of oxygen from irradiated water as an expla-nation for decreased radiobiological oxygen enhance-ment at high LET Nature 260 316ndash318

Benit J and Brown WL (1990) Electronic sputtering ofoxygen and water molecules from thin films of waterice bombarded by MeV Ne1 ions Nucl Instrum Meth-ods Phys Rev B 46 448ndash451

Benit J Bibring J-P della-Negra S LeBeyec YMendenhall M Rocard F and Standing K (1987)Erosion of ices by ion irradiation Nucl Instrum Meth-ods Phys Rev B 1920 838ndash842

Brown WL Augustyniak WM Lanzerotti LJ John-son RE and Evatt R (1980) Linear and nonlinearprocesses in the erosion of H2O ice by fast light ionsPhys Rev Lett 43 1632ndash1635

Brown WL Augustyniak W M Simmons E Mar-cantonio KJ Lanzerotti LJ Johnson RE BoringJW Reimann CT Foti G and Pirronello V (1982)Erosion and molecular formation in condensed gasfilms by electronic energy loss of fast ions Nucl In-strum Methods B 1 307ndash314

Brown WL Augustyniak WM Marcantonio KJ Sim-mons EN Boring JW Johnson RE and ReimannCT (1984) Electronic sputtering of low-temperaturemolecular solids Nucl Instrum Methods B 1 307

Buratti B (1995) Photometry and surface structure of theicy Galilean satellites J Geophys Res 100(E9) 19061ndash19066

Buratti B and Veverka J (1983) Voyager photometry ofEuropa Icarus 55 93ndash110

Calvin WM and Spencer JR (1997) Latitudinal distri-bution of O2 on Ganymede observations with the Hub-ble Space Telescope Icarus 130 505ndash516

Calvin WM Clark RN Brown RH and Spencer JA(1995) Spectra of the icy Galilean satellites from 02 to5 mm a compilation new observations and a recentsummary J Geophys Res 100 19041ndash19048

Calvin WM Johnson RE and Spencer JA (1996) O2

on Ganymede spectral characteristics and plasma for-mation mechanisms Geophys Res Lett 23 673ndash676

Calvin WM Aninich VG and Brown RH (2004) Oxy-gen temperature and phase effects Icarus (in press)

OXIDANTS AT EUROPA 847

Carlson RW Anderson MS Johnson RE SmytheWD Hendrix AR Barth CA Soderblom LAHansen GB McCord TB Dalton JB Clark RNShirley JH Ocampo AC and Matson DL (1999a)Hydrogen peroxide on the surface of Europa Science283 2062ndash2064

Carlson RW Johnson RE and Anderson MS (1999b)Sulfuric acid on Europa and the radiolytic sulfur cycleScience 286 97ndash99

Carlson RW Anderson MS Johnson RE SchulmanMB and Yavrouian AH (2002) Sulfuric acid pro-duction on Europa the radiolysis of sulfur in water iceIcarus 157 456ndash463

Carlson RW Anderson MS and Johnson RE (2004)Radiolytic carbon compounds on Callisto Geophys ResLett (in press)

Carr MH Belton MJS Chapman CR Davies MEGeissler P Greenberg R McEwen AS Tufts BRGreeley R Sullivan R Head JW Pappalardo RTKlaasen KP Johnson TV Kaufan J Senske DMoore J Neukum G Schubert G Burns JAThomas P and Veverka J (1998) Evidence for a sub-surface ocean on Europa Nature 391 363ndash365

Chyba CF (2000) Energy for microbial life on EuropaNature 403 381ndash382

Chyba CF and Hand KP (2001) Planetary sciencemdashlifewithout photosynthesis Science 292 2026ndash2027

Cooper JF Johnson RE Mauk BH Garrett HB andGehrels N (2001) Energetic ion and electron irradia-tion of the icy Galilean satellites Icarus 149 133ndash159

Cooper PD Johnson RE and Quickenden TI (2003a)A review of the possible optical absorption features ofoxygen molecules in the icy surfaces of outer solar sys-tem bodies Planet Space Sci 51 183ndash192

Cooper PD Johnson RE and Quickenden TI (2003b)Hydrogen peroxide dimers and the production of O2in icy satellite surfaces Icarus 166 444ndash446

Delitsky ML and Lane AL (1998) Ice chemistry on theGalilean satellites J Geophys Res 103 31391ndash31403

Domingue DL and Hapke BW (1992) Disk-resolvedphotometric analysis of European terrains Icarus 9970ndash81

Domingue DL and Hendrix A (2004) Temporal vari-ability in the surface chemistry of the icy Galilean satel-lites Icarus (in press)

Domingue DL Hapke BW Lockwood GW andThompson DT (1991) Europarsquos phase curve implica-tions for surface structure Icarus 90 30ndash42

Engdahl A and Nelanders B (2002) The dimmer in a tootight matrix cage Phys Chem Chem Phys 4 2140ndash2143

Fanale FP Granahan JC McDord TB Hansen GHibbitts CA Carlson R Matson D Ocampo AKamp L Smythe W Leader F Mehlman R Gree-ley R Sullivan R Geissler P Barth C Hendrix AClark B Helfenstein P Veverka J Belton MJSBecker K Becker T and the Galileo NIMS SSI UVSInstrument Teams (1999) Galileorsquos multiinstrumentspectral view of Europarsquos surface composition Icarus139 179ndash188

Geissler PE Greenberg R Hoppa G McEwen A

Tufts R Phillips C Clark B Ockert-Bell M Helfen-stein P Burns J Veverka J Sullivan R Greeley RPappalardo RT Head JW Belton MSJ and DenkT (1998) Evolution of lineaments on Europa clues fromGalileo multispectral imaging observations Icarus 135107ndash126

Gerakines PA Schutte WA and Ehrenfreund P(1996) Ultraviolet processing of interstellar ice analogs1 Pure ices Astron Astrophys 312 289ndash305

Ghormley JA and Stewart AC (1956) Effects of g-radiation on ice J Am Chem Soc 78 2934ndash2939

Gomis O Satorre MA Leto G and Strazzulla G(2004) Hydrogen peroxide formation by ion implanta-tion in water ice and its relevance to the Galilean PlanetSpace Sci (in press)

Grundy WM Spencer JR and Buie MW (2001) Ther-mal inertia of Europarsquos H2O ice surface componentBull Am Astron Soc DPS 33 4706

Gurman V Batyuk V and Sergeev G (1967) Photoly-sis of frozen dilute solutions of hydrogen peroxide inwater Kinet Katal 8 527ndash531

Hall DT Strobel DF Feldman PD McGrath MAand Weaver HA (1995) Detection of an oxygen at-mosphere on Jupiterrsquos moon Europa Nature 373 677ndash679

Hall DT Feldman PD McGrath MA and StrobelDF (1998) The far-ultraviolet oxygen airglow of Eu-ropa and Ganymede Astrophys J 499 475ndash481

Hansen GB and McCord TB (1999) Amorphous andcrystalline ice on the Galilean satellites a balance be-tween thermal and radiolytic processes [abstract 1630]In Lunar and Planetary Science Conference XXXI [book onCD-ROM] Lunar and Planetary Institute Houston

Hansen GB and McCord TB (2000) The distribution ofamorphous and crystalline ice on Ganymede Bull AmAstron Soc DPS 32 3903H

Hart EJ and Platzman RL (1961) Radiation chemistry[abstract 1630120] In Physical Mechanisms in RadiationBiology 1 edited by M Erreva and MA Forssberg Aca-demic Press New York p 99

Hendrix AR Barth CA and Hord CW (1999a)Ganymedersquos ozone-like absorber observations by theGalileo ultraviolet spectrometer J Geophys Res 104(E6)14169ndash14178

Hendrix AR Barth CA Stewart AIF Hord C andLane AL (1999b) Hydrogen peroxide on the icyGalilean satellites [abstract 2043] In Lunar and PlanetaryScience Conference XXX [book on CD-ROM] Lunar andPlanetary Institute Houston

Hori A and Hondoh T (2002) Theoretical study on thediffusion of gases in ice Ih by the molecular orbitalmethod [abstract A-1046] In Abstract Conference on thePhysics and Chemistry of Ice National Research Councilof Canada Montreal

Ip W-H (1996) Europarsquos oxygen exosphere and its mag-netosphere interaction Icarus 120 317ndash325

Johnson RE (1989) Sputtering of a planetary regolithIcarus 78 206ndash210

Johnson RE (1990) Energetic Charged-Particle Interactionswith Atmospheres and Surface Springer-Verlag Berlin

JOHNSON ET AL848

Johnson RE (1998) Sputtering and desorption from icysurfaces In Solar System Ices edited by B Schmitt andC deBergh Kluwer Dordrecht The Netherlands pp303ndash334

Johnson RE (1999) Comment on ldquoLaboratory Studies ofthe Optical Properties and Stability of Oxygen onGanymederdquo by RA Baragiola and DA Bahr J Geo-phys Res 104(E6) 14179ndash14182

Johnson RE (2001) Surface chemistry in the Jovian mag-netosphere radiation environment In Advances Series inPhysical Chemistry Vol 11 Chemical Dynamics in ExtremeEnvionments edited by R Dessler World Scientific Sin-gapore pp 390ndash419

Johnson RE and Jesser WA (1997) O2O3 microat-mospheres in the surface of Ganymede Astrophys J Lett480 L79ndashL82

Johnson RE and Quickenden TI (1997) Photolysis andradiolysis of water ice on outer solar system bodies JGeophys Res 102 10985ndash10996

Johnson RE Lanzerotti LJ and Brown WL (1982)Planetary application of ion-induced erosion of con-densed-gas frosts Nucl Instrum Methods 198 147ndash158

Johnson RE Nelson M McCord TB and Gradie J(1988) Analysis of voyages images of Europa plasmabombardment Icarus 75 423ndash436

Johnson RE Carlson RW Cooper JF Paranicas CMoore MH and Wong M (2003) Radiation effects onthe surfaces of the Galilean satellites In Jupiter Satel-lites Atmosphere and Magnetosphere edited by F Bage-nal Cambridge University Press Cambridge UK inpress

Jurac S Johnson RE and Richardson JD (2001) Sat-urnrsquos E ring and production of the neutral torus Icarus149 384ndash396

Kargel JS Kaye JZ Head JW III Marion GMSassen R Crowley JK Ballesteros OP Grant SAand Hogenboom DL (2000) Europarsquos crust and oceanorigin composition and prospects for life Icarus 148226ndash265

Khriachtchev L Pettersson M Jolkkonen S PehkonenS and Rasanen M (2000) Photochemistry of hydrogenperoxide in Kr and Xe matrixes J Chem Phys 1122187ndash2194

Khusnatdinov NN and Petrenko VF (1992) Photolu-minescence of ice Ih in a spectral region 180ndash300nm InPhysics and Chemistry of Ice edited by N Maeno and THondoh Hokkaido University Press Sapporo pp 163ndash169

Kimmel GA and Orlando TM (1995) Low-energy(5ndash120 eV) electron stimulated dissociation of amor-phous D2O ice D(2S) O(3P220) and O(1D) yields andvelocity distributions Phys Rev Lett 75 2606ndash2609

Kimmel GA Orlando TM Vizina C and Sanche L(1994) Low energy electron-stimulated production ofmolecular hydrogen from amorphous water ice JChem Phys 101 3282ndash3286

Landau A Allis EJ and Welsh HL (1962) The ab-sorption spectrum of solid oxygen in the wavelengthregion from 12000 Aring to 3300 Aring Spectrochimica Acta 181ndash19

Langford VS McKinley AJ and Quickenden TI(2000) Identification of H2O-HO in argon matrices JAm Chem Soc 122 12859ndash12863

Leblanc F Johnson RE and Brown WL (2002) Eu-roparsquos sodium atmosphere an ocean source Icarus 159132ndash144

Lefort M (1955) Chimie des radiations des solutions aque-ses In Actions Chimiques et Biologiues des Radiations Vol1 edited by M Haissinsky Masson Paris pp 203ndash247

Leto G and Baratta GA (2003) Lyman alpha photon in-duced amorphization of Ic water ice at 16 K Effects andquantitative comparison with ion irradiation AstronAstrophys 397 7ndash13

Livingston FE Smith JA and George SM (2002) Gen-eral trends for bulk diffusion in ice and surface diffu-sion on ice J Phys Chem A 106 6309ndash6318

Madey TE Johnson RE and Orlando TM (2002) Far-out surface science radiation-induced surface processesin the solar system Surf Sci 500 838ndash858

Matich AJ Bakker MG Lennon D Quickenden TIand Freeman CG (1993) O2 luminescence from UV-excited H2O and D2O ices J Phys Chem 97 10539ndash10553

Mauk BH Gary SA Kane M Keath EP KrimigisSM and Armstrong TP (1996) Hot plasma parame-ters of Jupterrsquos inner magnetosphere Geophys Rev 1017685ndash7696

McCord TB Hansen GB Matson DL Johnson TVCrowley JK Fanale FP Carlson RW SmytheWD Martin PD Hibbitts CA Granahan JCOcampo A and the NIMS Team (1999) Hydrated saltminerals on Europarsquos surface from the Galileo NIMSinvestigation J Geophys Res 104 11827ndash11851

McEwen AS (1986) Exogenic and endogenic albedo andcolor patterns on Europa J Geophys Res 91 8077ndash8097

McGrath MA Feldman PD Strobel DF RetherfordK Wolven B and Moos HW (2000) HSTSTIS Ul-traviolet imaging of Europa Bull Am Astron Soc DPS32 3409

Moore MH and Hudson RL (2000) IR detection ofH2O2 at 80 K in ion-irradiated laboratory ices relevantto Europa Icarus 145 282ndash288

Noll KS Johnson RE Lane AL Dominigue DL andWeaver HA (1996) Detection of ozone on GanymedeScience 273 341ndash343

Noll KS Rousch TL Cruikshank DP Johnson REand Pendleton YJ (1997) Detection of ozone on Sat-urnrsquos satellites Rhea and Dione Nature 388 45ndash47

Orlando TM and Sieger MT (2003) The role of elec-tron-stimulated production of O2 from water ice in theradiation processing of outer solar system surfacesSurf Sci 528 1ndash7

Orlando TM Sieger MT and Simpson WC (1998)Laboratory studies of the production of O2 on icy satel-lites by electronic excitation [abstract 1394] In Lunar andPlanetary Science Conference XXX [book on CD-ROM]Lunar and Planetary Institute Houston

Paranicas C Carlson RW and Johnson RE (2001)Electron bombardment of Europa Geophys Res Lett 28673ndash676

OXIDANTS AT EUROPA 849

Paranicas C Ratliff JM Mauk BH Cohen C andJohnson RE (2002) The ion environment near Europaand its role in surface energetics Geophys Res Lett 2914127

Prockter LM and Pappalardo RT (2000) Folds on Eu-ropa implications for crustal cycling and accommoda-tion of extension Science 289 941ndash943

Reimann CT Boring JW Johnson RE Garrett JWand Farmer KR (1984) Ion-induced molecular ejectionfrom D2O ice Surf Sci 147 227ndash240

Rowland B Fisher M and Devlin JP (1991) Probingicy surfaces with the dangling-OH-mode absorptionlarge ice clusters and microporous amorphous ice JChem Phys 95 1378ndash1384

Ruiz J and Tejero R (2000) Heat flows through the icelithosphere of Europa J Geophys Res 105(E12) 29283

Rye RR Madey TE Houston JE and Holloway PH(1978) Chemical-state effects in Auger electron spec-troscopy J Chem Phys 69 1504ndash1512

Selby B (2003) A kinetic study of UV-excited ice lumi-nescence [PhD Thesis] University of Western Aus-tralia Perth

Shematovich VI and Johnson RE (2001) Near-surfaceoxygen atmosphere at Europa Adv Space Res 27 1881ndash1888

Sieger MT Simpson WC and Orlando TM (1998)Production of O2 on icy satellites by electronic excita-tion of low-temperature water ice Nature 394 554

Smith RS and Kay BD (1997) Adsorption desorptionand crystallization kinetics in nanoscale water films Re-cent Res Dev Phys Chem 1 209ndash219

Spencer JR and Calvin WM (2002) Condensed O2 onEuropa and Callisto Astron J 124 3400ndash3403

Spencer JR and Klesman A (2001) New observations ofmolecular oxygen on Europa and Ganymede Bull AmAstron Soc DPS 33 4704

Spencer JR Calvin W and Person J (1995) CCD spec-

tra of the Galilean satellites molecular oxygen onGanymede J Geophys Res 100 19049ndash19056

Strazzulla G (1998) Chemistry of ice induced by ener-getic charged particles In Solar System Ices edited byB Schmitt and C deBergh Kluwer Dordrecht TheNetherlands pp 281ndash302

Taub IA and Eiben K (1968) Transient solvated elec-tron hydroxyl and hydroperoxy radicals in pulse-irradiated crystalline ice J Chem Phys 49 2499ndash2513

Vaghjiani GL Turnipseed AA Warren RF and Rav-ishankara AR (1992) Photodissociation of H2O2 at 193and 222 nm products and quantum yields J ChemPhys 96 5878ndash5886

Watanabe N and Kouchi A (2002) Measurements ofconversion rates of CO to CO2 in ultraviolet-inducedreaction of D2O(H2O)CO amorphous ice Astrophys J567 651ndash655

Watanabe N Horii T and Kouchi A (2000) Measure-ments of D2 yields from amorphous D2O ice by ultra-violet irradiation at 12K Astrophys J 541 772ndash778

Westley MS Bargiola RA Johnson RE and BarattaGA (1995) Ultraviolet photodesorption from water icePlanet Space Sci 43 1311ndash1315

Williams TL Adams NG Babcock LM Herd CRand Geoghegan M (1996) Production and loss of thewater-related species H3O1 H2O and OH in dense in-terstellar clouds Monthly Notices R Soc 282 413ndash420

Address reprint requests toDr RE Johnson

Engineering PhysicsThornton Hall B103

University of VirginiaCharlottesville VA 22904

E-mail rejvirginiaedu

JOHNSON ET AL850