Icarus142, 421–444 (1999)
Article ID icar.1999.6226, available online at http://www.idealibrary.com on
Evidence for Methane Segregation at the Surface of Pluto
S. Doute,1 B. Schmitt,2 and E. Quirico3
Laboratoire de Glaciologie et Geophysique de l’Environement, 54 rue Moliere, BP 96, 38402 St-Martin-d’Heres, FranceE-mail: [email protected]
T. C. Owen4
Institute for Astronomy, University of Hawaii, 2680 Woodlawn Drive Honolulu, Hawaii 96822
D. P. Cruikshank4
NASA Ames Research Center, MS 245-6 Moffett Field, California 94035–1000
C. de Bergh4
Observatoire de Paris, 5 Place Jules Janssen 92195 Meudon, France
T. R. Geballe5
Joint Astronomy Center, 660 North A’ohoku Place, University Park, Hilo, Hawaii 96720
and
T. L. Roush
San Francisco State University and NASA Ames Research Center, MS 245-6 Moffett Field, California 94035–1000
Received February 1, 1999; revised August 23, 1999
In May 1995, a set of spectrophotometric curves of the systemPluto–Charon were recorded with the UKIRT telescope equippedwith the spectrometer CGS4. The spectra cover the near-infraredrange between 1.4 and 2.55 µm with a resolution of approximately700. The existence of solid methane is confirmed by numerous ab-sorption bands, and carbon monoxide and nitrogen ices are iden-tified by their respective signatures at 2.35 and 2.15 µm. We havemodeled the spectrum of May 15 that corresponds to the maximumof Pluto’s visible lightcurve using a radiative transfer algorithmdealing with compact and stratified media. A geographical mixtureof three distinct units is required to explain all the significant struc-tures of the analyzed spectrum. The first unit is a thin, fine-grainedlayer of pure CH4 covering a compact polycrystalline substratum
1 Present address: Institute of Geophysics and Planetary Physics. UCLA, 405Hilgard, Box 951567, Los Angeles, CA 90095.
2 Present address: Laboratoire de Plan´etologie de Grenoble, Bˆat. D dePhysique, B.P. 53, 38041 Grenoble Cedex 9, France.
3 Present address: Institut d’Astrophysique Spatiale, Bat 121, Universit´eParis XI, 91405 Orsay, France.
4 Guest observer, United Kingdom Infrared Telescope, Joint AstronomyCenter, Hawaii, USA.
5 Present address: Gemini Observatory, 670 North A’ohoku Place, UniverPark, Hilo, HI 96720.
of N2–CH4–CO, which are in a molecular mixture (concentrationsof CH4 and CO of the order of 0.5 and 0.1–0.2% respectively). Itcovers about 70% of the observed area and corresponds to volatiledeposits that are sublimating under solar illumination. The secondunit is either (a) a single thick layer of pure large-grained methaneor (b) a unit with large-grained CH4 forming a substratum andthe N2–CH4–CO mixture a superficial layer of fine grains covering20% of the surface. Finally, the third unit is bright and spectrallyneutral and is first modeled as a layer of very fine grains of nearlypure N2. Tholin, suggested to explain the red slope in the visible, isalso found to be spectrally compatible with this unit. It covers theremainder of the surface (about 10–15%). All these results allow abetter understanding of the processes of deposition, metamorphism,sublimation, and transport affecting the different ices detected onPluto during its climatic cycles. c© 1999 Academic Press
Key Words: Pluto, surface; ices; spectrophotometry; charon;radiative transfer.
1. INTRODUCTION
larby
has
42
sity
Pluto, the ninth and the last identified planet of the SoSystem, remains the only one that has not yet been visiteda spacecraft. However, our knowledge about this object
1
0019-1035/99 $30.00
Copyright c© 1999 by Academic PressAll rights of reproduction in any form reserved.
o
wher
eian
ia
e
aix
s
d
s
u
dl
-tot
ssi-
and
ob-the
bypec-ricalis aontheintson,lids
nalibed.rum.ndim-tione dif-ms ofribe
s cor-larec-eirthe
ctraay
the
thelv-
etionon,un-re
.ted
sferctrumthe
422 DOUTE
significantly progressed during the last few years thanks tocent instrumental developments.
For the first time, the treatment of Hubble Space Telesc(HST) images, produced in late June and early July 1994the Faint Object Camera (FOC) ESA in blue light, has allowSternet al. (1997) to generate maps nearly covering the themispheres of Pluto with 160-km resolution. These maps sa huge spatial variability of the visible albedo with a dozdistinctive features, including a bright polar cap. The featuare thought to be related mainly to changes in the chemicalphysical properties over the surface.
In the same way, the improvement of ground-based ninfrared (NIR) and, to a less extent, visible (Grundy and F1996) spectrophotometry has dramatically increased our cbilities to determine surface properties. This new generatioobservations has come essentially from the United KingdInfraRed Telescope (UKIRT), Mauna Kea, Hawaii. The initmeasurements were made in 1992 with the NIR spectromCGS4 (Cooled Grating Spectrometer) and covered the rangto 2.5µm at a spectral resolution of 6.7 nm (Owenet al.1993).They included the light reflected by the spatially unresolvdisks of both Pluto and Charon. A spectroscopic analysis ofdifferent reflectance spectra obtained has provided crucial inmation about the nature and physical state of the materialscover the uppermost millimeters or centimeters of the surfof Pluto. The presence of solid methane was confirmed by echaracteristic absorption bands. In addition, carbon monoand nitrogen ices were identified by their respective signatuat 2.35 and 2.15µm. The analysis of the profile of the 2.15-µmband of nitrogen, used as a “thermometer” in the 1993 obvations, led Trykaet al. (1994) to propose a value of 40± 2 Kfor the temperature of nitrogen ice on Pluto. Carbon dioxian ice that is detected on Triton, seems to be absent on Pas are hydrocarbons like C2H2, C2H4, C2H6, etc., but water icecannot be excluded (Owenet al.1993). For Charon, water ice ithe only compound that was recognized unambiguously byspectral signatures at 1.55 and 2.04µm (Buieet al. 1987), butother components may also exist (e.g., Roush 1994).
This study also showed that, on Pluto, the methane molecare most probably diluted in nitrogen ice but not as diluted asTriton. This was deduced by noting that the position and wiof most of the methane bands fit those of a laboratory samp1% CH4 mixed in N2 better than either those of pure CH4 ice orthose of CH4 even more highly diluted in N2 (Schmittet al.1992,Quirico 1995). The higher mean CH4 concentration (2.6%) deduced by the model for Pluto compared with Triton also pointoward an intermediate state of mixing. However, the laboratdata available at that time were insufficient to define this smore precisely. Subsequent extensive laboratory studies onCH4 and CO (as well as several other molecules) and their vous mixtures with solid nitrogen (Quirico and Schmitt 1997aQuiricoet al.1996, 1999, Schmittet al.1998) allowed a detailedspectroscopic analysis of the initial 1992 UKIRT observation
well as a set of spectra taken in 1993. This led to the suggesET AL.
re-
peby
edoownesand
ar-nkpa-of
oml
eter1.4
edthefor-thatceghtideres
er-
e,luto,
its
lesonth
e of
edry
atepureari-,b,
as
that a segregation of methane with an enriched top layer pobly occurs at the surface of Pluto (Schmittet al.1994, Quirico1995) rather than a horizontal distribution of patches of purediluted CH4 (Cruikshanket al.1997, Grundy and Fink 1996).
In May 1995, an additional series of three Pluto–Charonservations were acquired with the UKIRT–CGS4 system insame spectral range but with a spectral resolution improveda factor of about 2.5 (2.64 nm). In this paper, we present a stroscopic analysis of these observations as well as numemodeling of one of these spectra. The scope of this workfirst attempt to derive the horizontal and vertical distributimodes of the different ices detected, to solve the problem ofice segregation. This study also leads to improved constraon the intrinsic chemical and physical properties (compositigrain sizes etc.) and about the geographical extent of the soidentified so far. In the first part of the paper, the observatiodata, as well as the relevant calibration procedures, are descrThe second part is dedicated to analysis of the Pluto spectFirst, we recall the contributions of the modeling approach, apresent the framework and the numerical tool used for our sulations. Second, after modeling and extracting the contribuof Charon to the measured spectral reflectance, we assign thferent bands observed and analyze the methane bands in terthe physical state in which the methane occurs. We then deschow we tested and selected the best surface representationresponding to the different coexisting modes of the molecumixture N2–CH4–CO and pure methane suggested by the sptroscopic analysis. A synthesis of these results, followed by thinterpretation and discussion, is presented in the final part ofpaper.
2. PRESENTATION AND CALIBRATIONOF THE OBSERVATIONS
In May 1995, three nonspatially resolved reflectance speof the Pluto–Charon system were acquired respectively on M11, 13, and 15. The longitudes of the sub-Earth point onsurface of Pluto were respectively 66◦, 313◦, and 200◦.
The CGS4 spectrometer measures the flux received byUKIRT telescope with a spectral resolution of 2.64 nm (resoing powerλ/dλof approximately 750). TheH range 1.4–2.1µmand theK range 1.9–2.55µm were recorded separately. Thmeasurement of the lines of an argon lamp and the observaof the atmospheric bands of the OH radical, both in emissiensure a good spectral wavelength calibration with a meancertainty of±1 nm. The details of the calibration procedure amentioned in Quiricoet al. (1999). The following paragraphdescribes the way the radiometric calibration was achieved
The radiative power that arrives on the detectors is integrafor 1 h on average. The resulting curveS(λ), measured in dig-ital units, is deeply marked by absorptions due to the tranthrough the terrestrial atmosphere. As a consequence, a speS✹(λ) of a solar-type star is recorded in the same region of
tionsky with only a short time shift. Care was taken to observe this
T
a
oeea
y
n
onth
no
on-ysig-
lear,eay ofures
13,r
iderved.rre-
e. 2)
CH4 SEGREGA
star, BS5384, at an airmass close to the midpoint of the airmrange for which Pluto was observed, leading by a simple rS(λ)/S✹(λ) to the correction of the solar spectral features athe disturbing atmospheric bands (principally those of CO2 andH2O). Only the strongest lines of telluric H2O in the range 1.8–1.95µm cannot be removed accurately, since the atmosphdistribution of H2O varies over short time intervals. In the samtime the radiometric calibration of the data is performed csidering the star BS5384 as a blackbody, the temperaturwhich is 5900 K. In that case, the infrared photometric msurements of Leitherer and Wolf (1984) allow us to calculits absolute spectral fluxFB(λ; T = 5900 K) as received by thetelescope. Dividing the Pluto–Charon spectrumS(λ) by the mea-sured spectrumS✹(λ) of BS5384 and multiplying the result bthe blackbody fluxFB(λ; T = 5900 K) yields for the flux of theplanet and its satellite
F(λ) = S(λ)/S✹(λ) ∗ FB(λ; T = 5900 K). (1)
Finally, the astronomical spectrum is converted into a geomrical albedoAg according to the expression
Ag(λ) = D2d2F(λ)
R2Fs(λ), (2)
whereD andd respectively correspond to the heliocentric a
geocentric distances of Pluto–Charon,R is their mean quadraticFIG. 1. Calibration operation applied to the 05/11/95 observation of Pluto. TheH region (dot–dash line) is adjusted to theK region by a multiplication factor
(Buieet al.1997). In addition, anH spectrum recorded on May
to form a new coherent spectrum (solid line). This spectrum is compared(dotted and dashed lines).
ION ON PLUTO 423
asstiond
ericen-
ofa-te
et-
d
radius,F(λ) is the corrected and calibrated flux as receivedEarth, andFs(λ) is the solar flux at 1 AU, as tabulated by Smiand Gottlieb (1974).
After applying this calibration procedure, it appears thatphotometric coherence is achieved between theH andK spec-tral regions as judged from the overlapping range 1.9–2.1µm. IntheK region the measurements of May 11, 13, and 15 give csistent albedo spectra while theH region displays unexpectedlhigh and variable values. On May 11, clouds attenuated thenal from the calibration star BS5384 while itsH -band spectrumwas being measured. On May 13 and 15, the skies were cbut we observed the star atH through a partially opened dombecause of concerns that the starlight would saturate the arrthe CGS4 spectrometer. Following the data reduction proceddescribed above, we just multiplied theH -band spectra from thethree nights by scaling factors (0.75, 0.47, and 0.375 for 11,and 15 May, respectively) to match theK -band spectra in theiregions of overlap. We note that the scaledH -band fluxes areconsistent with those measured in earlier years when a wslit was used and a better photometrical calibration achieFor example, a spectrum measured on April 16, 1994, cosponding to the visible lightcurve minimum of Pluto at 96◦ lon-gitude, is very consistent with most of theH andK spectrum ofMay 11, 1995, at 66◦ longitude (Fig. 1). On the lightcurve, thcorresponding visible albedos differ by a bit less than 5% (Fig
with better-calibrated spectra from 05/27/92 and 04/16/94 at similar rotational phases
)
424 DOUTE ET AL.
FIG. 2. Visible geometric albedo lightcurve of the Pluto–Charon system as calculated by Buieet al. (1997). The locations of Pluto (sub-Earth longitude
during the three 1995 UKIRT–CGS4 spectrophotometric observations, the 1992 and 1994 observations used to check our calibration operation, as well as at thetime of the eclipse observation of Charon in 1987, are indicated.do
nrn
t
l
na
o
att
s tone-ionlesson-uc-er,eralnceIf itthenn ofsim-calsorp-sitytainthewelltiesface
27, 1992, at 51◦ longitude, i.e., only 15◦ away from the 05/11/95observation, also fits the scaled 1995H spectrum within a fewpercent. Nevertheless, part of the two surrounding spectranot match exactly with the 1995 measurement, essentiallyside the strong methane bands. The noticeable difference inservation conditions (−15◦ and+30◦ longitude for the 1992 and1994 spectra relative to the 1995 one) or a possible evolutiothe surface of Pluto may explain the spectral differences in teof average surface composition and/or physical state. Wethat the variations in the “continuum” around 1.5µm (Fig. 1)are correlated with the visible albedo (Fig. 2), thus favoringfirst explanation.
In conclusion, we believe that our initial photometric cabration is correct for all three spectra in theK region and afterrescaling for theH -region spectra. Knowing that the 1992 a1994 reference spectra at lower resolution display an estimuncertainty of±15% and considering the difference with thspectrum of May 11, we think that our calibration operatiyields all threeH andK May 1995 flux spectra with a conservative error of±20% in albedo. This uncertainty applies over moof the spectral range, as multiplying factor, or a broad specslope. In the following we use the calibrated and concaten1.4- to 2.55-µm flux spectra, except for May 13 restrictedthe H region due to some spurious instrumental effects in
K -region data.oesut-ob-
ofmsote
he
i-
dted
en-sttraltedohe
3. ANALYSIS OF THE OBSERVATION
3.1. Framework of the Modeling Approach
In the visible and near-infrared ranges, the flux that comeEarth from the Pluto–Charon system, or from any other platary body in general, totally or partly arises from the reflectand absorption of the solar light that penetrates more ordeeply into the surface. The intimate interaction with the cstituent materials implies that their physicochemical and strtural properties potentially mark the spectral signal. Moreovas light attenuation into the materials often changes by sevorders of magnitude at different wavelengths, every reflectalevel of a spectrum probes a particular range of depths.is modulated enough, the general shape of the spectrumbecomes very sensitive to the occurrence of a stratificatiothe surface properties. For all these reasons, beyond aple identification of the chemical nature and of the physistate of the components, as expressed by the precise abtion band positions (see Section 3.3), modeling of the intenand the shape of the bands is the only efficient way to obadditional information (see Section 3.4). These are mainlychemical composition of the compounds and the texture, asas the granularity of the materials they form. These propercan be obtained as a function of depth in the case of sur
stratification.T
an
ie
tr
ina
chlith
cesph-g
re-e athere.fulent
lutoec-t a
c
on-at
dλ= 0.1µm (Fig. 3). For that, we have to choose a suitable, but
CH4 SEGREGA
Things can be a bit more complicated, however. Most oftime different kinds of uniform surface units, which form somthing like a more or less regular geographical mosaic that isspatially resolved by the observation, may contribute to the sspectral signal. Then, they linearly add their respective cobutions according to their surface proportion and the viewgeometry. This kind of addition is called a “geographical mture.” Geographic mixing is always implied for disk-integratspectral measurements such as those we consider here, belateral inhomogeneities always occur on planets.
For the modeling of the spectrum of the Pluto–Charon stem, we use an efficient radiative transfer algorithm calculathe spectral bidirectional reflectance of an icy or mineral stified surface (Dout´e and Schmitt 1998). Each layer of suchsurface can present a granular or a compact texture. The athe algorithm is to determine, for any kind of illumination aviewing conditions, the link between the physical, chemical,structural properties of the underlying medium and its obserspectral signature. A potential stratification is described byvertical superposition of any number of plane-parallel homoneous layers.
Practically, the algorithm is integrated into a software thatglobally simulate and then analyze by interactive inversionperspectral data according to a wide variety of observationasurface situations (Dout´e 1998). Indeed this system can deal w
7
one
spectrophotometric data (as is the case here with Pluto–Charon)FIG. 3. Low-resolution modeling (solid line) of the spectral geometric albedo of Charon calculated by Roushet al. (1996) (dashed line), from the 04/23/8
simplified, surface representation of Charon. The simplest
mutual event observations (Buieet al.1987). In this model 60% of the surfacecomponent.
ION ON PLUTO 425
thee-notmetri-ingx-dcause
ys-ingat-a
m ofdnd
vedthege-
any-
and
and with hyperspectral images, both corresponding to surfathat can present any kind of chemical composition, geograical, and vertical distribution of the materials. The modelinis performed by integration of the radiative spectral powerflected by the different surface units that generally constitutplanetary surface. This radiative transfer algorithm ensurescalculation of each contribution of such a geographical mixtuAll the above-mentioned characteristics lead to a very powersystem to reach our main objective: to test and select differsurface representations of Pluto.
3.2. Modeling and Extraction of the Contributionof Charon
We first need to separate the respective contributions of Pand of its satellite. The problem consists of extracting a sptrum of Charon from our observations which were acquired afar higher resolution (dλ= 2.64 nm) than any available Charonspectrum. Roushet al. (1996) derived the spectral geometrialbedo of Charon from very poorly resolved (dλ= 0.1µm) spec-trophotometric data acquired between 1.5 and 2.5µm during aspecial mutual event: the total eclipse of Charon by PlutoApril 23, 1987 (Buieet al.1987). We first modeled this geometric albedo curve with high-resolution simulations deresolved
is covered by H2O ice (grain size 50µm) and the remainder by a spectrally neutral
n
n0
a
i
t
-
ioi
l
o13,wa-we
thether
byits
t thede-
thes-
ra-lar,
nd
426 DOUTE
we found is a uniform hemisphere covered by a geographmixture of two kinds of surface units: (1) a unitUw, which is agranular optically semiinfinite layer of water ice, and (2) a uUn, which is spectrally neutral and of unknown chemical natuFor the first unit, intended to reproduce the absorption bawe use new optical constants of crystalline water ice at 4(Grundy and Schmitt 1998). The second unit permits the 2-µmflux of the strong H2O bands to rise to the correct level. This wimpossible to realize with a hemisphere entirely covered wH2O grains. The principal parameters inferred by the invers(surface proportion of the water icePsurf
H2O= 60%, mean grainsizedH2O= 50µm) as well as the chosen surface representamay not be very significant, nor well constrained, but the desolved spectrum they generate by direct modeling is withinerror bars of the geometric albedo of Charon (see Fig. 3).
Using the above values for Charon, we calculate the cresponding high-resolution (dλ= 2.64 nm) spectrum subsequently transformed into flux using Eq. (2). This flux contribtion is then removed from the Pluto–Charon 1995 observatto derive a Pluto-only spectrum. We finally convert the resultspectrum of Pluto to geometric albedo thanks again to Eq.(Fig. 4).
As the eclipse observation corresponds to a Pluto centralgitude of 180◦, the removal of Charon spectral contributionstrictly valid only for the May 15, 1995, observation (200◦ Plutolongitude). Near-infrared spectral variations with Charon lo
ulated by
at 1.69µm that we attribute to pure solid CH4 is a special case
gitude have been observed and attributed to changes in waterFIG. 4. Geometric albedo of the Pluto–Charon system from the 15 May observation (dotted line), of Charon (dashed line) at high resolution calc
discussed in more detail below.
fitting the low-resolution observations (see Fig. 3), and of Pluto alone (solid
ET AL.
ical
itre.ds,K
sithon
ionre-the
or-
u-ns
ng(2)
on-is
n-
ice absorption (Boshet al. 1992, Buie and Shriver 1994), sthe removal of the eclipse observation from the May 11 or1995 observations may lead to incorrect subtraction of theter ice signature of Charon. For that reason, in this paperrestrict our modeling effort to the May 15 observation. Onother hand, the following spectroscopic analysis of the two oPluto spectra (at 66◦, May 11, and 313◦, May 13), based mainlyon CH4 band positions, should not be significantly alteredthe first-order removal of Charon water ice absorption using180◦ synthetic spectrum.
3.3. Spectroscopic Analysis
After some discussion of band assignments we presenspectral analysis applied to the higher-resolution spectrascribed in this paper.
3.3.1. Band Assignment
Our high-resolution spectra of Pluto exhibit 20 bands inrange 1.4–2.5µm. About half of them were observed and asigned by Owenet al.(1993) to N2, CH4, and CO. We completethis identification and indicate, for each band, the specific vibtion mode of the corresponding molecule (Fig. 5). In particuthe observation of the second overtone of solid CO at 1.579µmconfirms the identification of this molecule on Pluto. The ba
line) after extraction of the Charon contribution to the total flux.
CH4 SEGREGATION ON PLUTO 427
to tha
FIG. 5. Identification of the absorption bands and illustration of the choice of the significant characteristic spectral features on the spectrum of Plut are good criteria for the estimation of the modeling quality: (a)H region, (b)K region.di
o
t
a
u
n
n
e
e
c
needndidenggheak
13w-
dat
e-y
ceuld
.easeithalsopref-tiont ofandoc-e ofg oftions
Natice,uld
er-ing
any
n
genionHan-
428 DOUTE
Compared with Triton, there is no positive sign of the thrstrong CO2 ice bands around 2µm, thus excluding the presencof this molecule in significant amount at the surface of PluThe problem of water ice is more complicated because its brabsorption bands combined with the broad CH4 bands make itsdirect identification very difficult. As an illustration, just looking at the spectrum of the Pluto–Charon system displayeFig. 4, it is not evident that it contains the signature of wateralthough this component actually contributes to at least 13%the projected surface! This point is further addressed belowthe detailed modeling of the spectrum (see Section 3.4), butfor the observation simulated in this paper (5/15/95).
Only one band that is present in all three of the May 19observations, but not observable in the former lower-resoluobservations, cannot be attributed to one of the three molecidentified previously on Pluto. It peaks around 1.75µm, and isslightly shifted from one observation to the other in the ran1.749–1.752µm. This band has also been observed in the sptrum of Triton where it has been suggested that ethane (C2H6)or perhaps propane (C3H8) diluted in solid nitrogen may bea potential candidate (Quiricoet al. 1999). Comparisons withthe available laboratory data on various molecules (QuiricoSchmitt 1997a, Quiricoet al.1999) show that pure C2H2, C2H4,C2H6, C3H8, HCN, HC3N, CH3OH, SO2, NO, NO2, and NH3
ices cannot explain this band but that C2H6 and possibly C3H8
diluted in solid nitrogen display a strong band shifted only0.003–0.006µm from the observed feature. However, the limited relevance of the currently available data on both molec(measured at 1% concentration inα-N2 at 21 K) does not explainthe remaining shift, thus still precluding a firm identificatioLaboratory data on these molecules at more relevant temptures and concentrations are clearly needed.
3.3.2. Physical State of the CH4 Molecules
Using the laboratory data of Quirico and Schmitt (1997we can compare the methane bands in the Pluto spectrathose of pure solid methane around 40 K and those of C4
diluted in solidβ-N2 around this same temperature (Fig. 6The first striking point is that the weak, but real, band presin all three observations around 1.69µm is active only in puresolid methane. Indeed, this band has never been observed isample of CH4 diluted in N2 (neither inα- or β-N2 phases) atany temperature. This band is clearly a witness of the presof pure methane at the surface of Pluto.
The second point is that the peaks of the strongest bandthe K region (at 2.20, 2.32, 2.37, and 2.43µm), particularlyfor the 15 May spectrum, are situated at positions intermate between those of pure and diluted methane, contrary tosame absorptions in the spectrum of Triton that perfectly fitpositions of diluted CH4 (Fig. 6b). These bands are also muwider, partly due to saturation of the bands (at 2.32 and 2.37µm)but possibly also due to the simultaneous contribution of b
forms of methane. If we now look at the medium bands in theHET AL.
eeeto.oad
-in
ce,ofbynly
95ionules
geec-
nd
by-les
.era-
a),withH).ent
any
nce
s in
di-the
theh
oth
region, they more closely fit the wavelength of diluted methafor the 11 May observation, but they are still slightly shifttoward low frequency compared with Triton in the 15 May aparticularly in the 13 May spectra (Fig. 6a). In addition the sband near 1.78µm is less pronounced than on Triton, indicatia possible contribution of pure methane, partly filling the troubetween the side band and the main band. Finally the two wbands around 1.48 and 1.95µm follow similar trends, the 11and 15 May bands being at intermediate positions and theMay bands almost fitting the positions of pure methane. Hoever, the 1.69-µm band of pure CH4 seems also to be shiftein the same direction compared with its theoretical position40 K. This shift corresponds to the limit of accuracy of our wavlength calibration. A different CH4 temperature on Pluto maalso contribute to this shift.
Although all this information points toward the coexistenof pure and diluted methane at the surface of Pluto we shotake the shift information in theH region with some cautionNevertheless, for the 15 May observation the apparent decrof the diluted character of methane when looking at bands wincreasing strengths, probing less deeper into the surface,suggests that at least part of the pure methane should beerentially accumulated near the surface. A different situaseems to occur for the 11 May spectrum: a larger amounpure methane possibly showing up at depth (weak bands)diluted methane dominating the top layers. The effectivecurence of such vertical stratifications, and of any other typCH4 segregation, should be confirmed by a detailed modelinthe whole spectrum for various possible surface representa(see Section 3.4).
It should be noted that the phase diagram of the binary2–CH4 mixture (Prokhvatilov and Yantsevich 1983) tells us thabsolutely pure methane ice cannot coexist with nitrogenbut that a small fraction of nitrogen (3–4% around 40 K) shobe diluted in it. The positions of the CH4 bands in such a systemare slightly shifted from the positions for pure CH4 ice, as shownby laboratory spectra (Quiricoet al.1996, Schmittet al.1998).However, this shift is smaller than the resolution of the obsvations and does not affect our interpretation. In the followwe continue to refer to this CH4-rich mixture containing a fewpercent nitrogen as “pure methane.”
These intermediate positions of the CH4 bands provide impor-tant information about the physical state of CH4 at the surfaceof Pluto. On the other hand, they preclude the extraction ofinformation or constraint on the temperature of the N2–CH4–COmixture and on the crystalline phase of solid N2 based on theprecise position of the CH4 bands, as was possible for Tritowhere only diluted CH4 is present (Quiricoet al.1999). In par-ticular, the most sensitive band relative to the phase of nitroat 2.32µm is strongly perturbed by the suspected contributof pure CH4 ice. The detailed modeling of the shape of the C4
bands may help constrain the surface temperature, but it isticipated that the blending between the two CH4 contributions
will dominate the spectrum. The CO bands being weak and quiteCH SEGREGATION ON PLUTO 429
alsoration
4
FIG. 6. Comparison of the 1995 albedo spectra (a:H region, b:K region) of Pluto with laboratory transmission spectra of pure CH4 ice at 40 K (bandpositions indicated by vertical dotted lines) and of CH4 diluted inβ-N2 ice at 40 K (positions: vertical dot–dash lines). The spectrum of Triton (09/07/95) isshown for comparison (from Quiricoet al.1999). All spectra are shifted vertically for clarity. The horizontal error bar corresponds to the wavelength calib
uncertainty (±1 nm).t
t
pu
r
e
c
th
re
y-
-
re
pa-vi-
c”
”a-
beripts
of
2-ionuching
430 DOUTE
insensitive to temperature and phase, the only remainingperature tracer seems to be the weak nitrogen band at 2.15µm.As has been noted before, its broad shape is clearly indicativthe predominance of theβ phase, i.e., a temperature larger th35.6 K. But only a thorough modeling of the shape of the weN2 band together with the underlying CH4 band may providesome additional constraint on the temperature of the nitrogrich deposits.
3.4. Test and Selection of the Best Surface Representations
3.4.1. Presentation
The previous spectroscopic analysis suggests the coexisof two phases of methane on the surface of Pluto: the pure sform and methane diluted in solid nitrogen. Three simpletential modes of coexistence can be imagined: (I) a gran(intimate) mixture, (II) a vertical stratification, and (III) a geographical mixture. An intimate mixture of grains respectiveformed by pure methane and “CH4-contaminated” nitrogen isthe first possibility (I). A second possibility (II) would be a vetical segregation of pure CH4 and of a N2–CH4–CO molecularmixture. The third possibility (III) is that more or less extendhorizontal patches of pure and diluted CH4 coexist on the sur-face of Pluto. Nevertheless, as shown in Section 3.3.2, thbasic surface representations will have to be combined inferent ways to provide the best solutions for the problem ofspecies distribution. Due to the geographically integrated naof the available spectra, it is not yet possible to derive a prespatial distribution of the different physicochemical units we cexpect at the surface of Pluto. Consequently, we propose a cidealization of the three basic representations of differentiawe have described, each covering the entire visible hemispof Pluto:
I. First basic representation Ri (intimate): A uniform depositcomp
utem
osed of one single thick layer:TABLE IThe Different Parameters of the Basic Surface Representations Ri, Rs, and Rg
most of the flux at 2.32 and 2.37µm (this implies a minimu
ET AL.
em-
e ofanak
en-
enceolido-lar-ly
-
d
esedif-thetureiseanrudeionere
—a unit Ui formed by a homogenous population of puCH4 and N2–CH4–CO grains intimately mixed (Gc and Gm,respectively).
II. Second basic representation Rs (stratified): a stratified anduniform deposit Us composed of two superimposed distinct laers:
—a superficial layer Lc of finite depth, formed by a homogeneous population of pure CH4 grains,
—an optically semiinfinite substratum Lm, formed by ahomogenous population of N2–CH4–CO (molecular mixture)grains.
III. Third basic representation Rg (geographic): a uniformgeographical “mosaic” of two distinct units:
—a unit Uc formed by an homogenous population of puCH4 grains,
—a unit Um formed by an homogeneous population of N2–CH4–CO (molecular mixture) grains.
These three configurations imply three different lists of freerameters for the models, which are listed in Table I. In the preous and all following notation of the “units,” U, the subscripts “refers to pure CH4, “m” to the N2–CH4–CO molecular mixture,“i” to the intimate (granular) mixtures of “c” and “m” grains, “sto the superposition of “c” over “m”, “r” to the reverse stratifiction, and “n” to the neutral or fine-grained N2 unit. All surfacerepresentations (except the basic representation Rg≡Rcm) that isa geographical combination of two or more of these units willdescribed by the concatenation of the corresponding subsc(e.g., Risn≡Ui +Us+Un).
The first difficulty that arises when modeling the spectrumPluto with either basic surface representation Ri, Rs, or Rg, isthe obvious impossibility of fitting the albedo level for the 2.3and 2.37-µm CH4 bands. Indeed, due to the required saturatof these two bands, the models always lead to albedos mlower than observed. There are two possibilities of increasthe albedo: (1) add a new geographical unit that will contrib
aa
aeri
u
n
op
g
cu
d
afio
oo
d toag-sionws
ym-
ling
s ofaceedsnal-
f then-
thede-
at
mitthem),f thetical
useare
an
the
-all
datacon-
yredrstrdtheonve-
ysonseedH
CH4 SEGREGAT
level of brightness for the constituent materials); (2) mix at lepart of the surface grains with some other particles that hto be bright enough (high single scattering albedo) to increthe reflectance around 2.32 and 2.37µm in particular. In bothcases, the additional necessary component(s) should havemost neutral spectral reflectance over the entire range, sincdo not observe any significant continuum slope in the spectand only a single weak unidentified localized signature (posslinked with units containing CH4, if the ethane identification isconfirmed). Among the molecules that have been identifiedthe spectrum of Pluto, only nitrogen, even a with small amounCH4 and CO, can present this neutrality, if it is present as grathat are sufficiently fine. This property comes from the weaness and the uniqueness of the N2 absorption band at 2.15µm.Since a contrary H2O ice, suspected at some places on Pl(Cruikshanket al.1997), presents in the range 1.4–2.5µm sev-eral strong and broad absorption bands, this compound caplay the same role, even with submicron-sized grains. Comporganic mixtures, such as tholins, have been suspected tpresent on Pluto as they may explain the red slope of the stra in the visible range (Grundy and Fink 1996). Although ncompletely neutral, tholins formed from CH4–N2 gas mixtures(“Titan tholins”) could be quite bright in the near-infrared ranfor very small grains [imaginary index of refractionk< 10−3
between 1.0 and 2.4µm (Khareet al. 1984)]. However, ournear-infrared data provide no identification, nor any constraon its presence. We thus do not consider tholins in our modethis point but, in a second step, we will simply test its constency as a quasi-neutral unit Un with our best surface representations (see Section 3.4.2-d). Consequently, we choose to adfirst a new unit of fine-grained nitrogen ice to the geographimosaic with the other compounds. Forming a granular mixtof N2-rich grains within the unit Um would have been anothechoice but it is thermodynamically very improbable to haveintimate mixture of two very different grain sizes of the samvolatile material. In addition, because of the nonlinear behavof a granular mixture, it is necessary to introduce an extremlarge quantity of fine-grained nitrogen to produce a substanrise in the reflectivity at 2.32 or 2.37µm where methane ishighly absorbing, whatever its state of dilution (pure or dilute
Consequently, we add to all our future representations baon Ri, Rs, and Rg a spectrally neutral geographical unit Un (finegrains of nitrogen) that is described by the following parametesurface proportionPsurf
n , mean grain sizern, scattering phaseparametergn.
One can note that several scattering parametersg appear inthe previous list and in Table I. They express the angular disbution of the scattered power by a volume element of a layerare related to the Henyey–Greenstein phase function as deand illustrated by Hapke (1993). They cannot really be cstrained by a disk directionally integrated measurement asthe case for our spectrum of Pluto. Without any additional infmation, we decided to set all these parameters to zero (isotr
scattering grains). This particular choice may have a notiION ON PLUTO 431
stve
ase
n al-we
umbly
int ofinsk-
to
notlex
beec-
ot
e
intl atis--dalre
raneiorelytial
).sed
rs:
tri-ndnedn-
it isr-pic
able effect on the values of the parameters closely relateg, i.e., mainly the grain sizes, but not on their orders of mnitude. For the other parameters, the results of the invershould be relatively insensitive to our choice. Table I also shothat a finite layer is parameterized by its optical depthτ , ratherthan by its real thicknessl . Indeed only the former quantithas a direct role in the radiative transfer calculation. It cobines two free parameters: the real thicknessl and the densityof the layer, which both cannot be constrained by the modealone.
To calculate the scattering and the absorption propertiethe materials that constitute the different layers of the surfunits, the radiative transfer model we use for the analysis netheir optical indices versus wavelength. The spectroscopic aysis has given some information about the physical state oidentified components: N2, CH4, and CO (see Section 3.3). Cosequently, we select a set of optical indices, measured inlaboratory and compatible with these constraints. Below, wescribe these data, ice by ice:
Pure CH4 ice: We use a set of measurements performed40 K (unpublished, Grundyet al., in preparation) and similar tothe data at 24 and 30 K (Quirico and Schmitt 1997a, Schet al. 1998). The effects of temperature affect principally tposition and the width of the bands (as well as the continuubut remain weak in these data, especially at the resolution oPluto observations. Thus, the appropriateness of the 40 K opconstants is ensured if the actual surface temperature of CH4 iceis not too different from 40 K. This may be the case becatemperatures slightly higher (up to 54 K) than the averageexpected for this kind of deposit due to its lower volatility thN2 ice (Stansberryet al.1996).
Pure N2 ice: The laboratory data we use were obtained inβ phase at 36.5 K, slightly above theα–β phase transition, byGrundyet al. (1993) and Quirico (1995).
Molecular mixture N2–CH4–CO: It is not possible to experimentally obtain the optical index spectra corresponding tothe potential values for the concentrationsCCH4 andCCO in solidnitrogen. As a consequence, we numerically generated theneeded from a sequence of experimental measurementscerning N2–CH4 and N2–CO mixtures at 36.5 K obtained bQuirico and Schmitt (1997a,b). For that purpose, we considethree real or virtually “pure” reference components. The fione is identified with pure N2 ice, and the second and the thiones with methane and carbon monoxide initially diluted innitrogen matrix, but artificially normalized to unit concentrati(see Quiricoet al.1999). Then we assume that, for each walength, the scalar product of the complex indices of N2, of thenormalized diluted CH4, and of the normalized diluted CO bthe vector [(1-CCH4 -CCO), CCH4, CCO] gives the optical constantof the mixture. The normalization and scalar product operatido not change the spectral effects linked to the dilution. Indthe slight high-frequency shifts of the absorption bands of C4
ce-diluted inβ-N2 compared with pure CH4 (4 to 6 nm in theH and
i
tyse
d
t
et
titto
i
e
g
n
r
t
eric
icalneely,hat-um(R
e
st areverthe
l-oththey
ti-es ofounthean-
nsita-theachf theters,
o-
on-
Onthe
sen-
432 DOUTE
K regions) strongly determine the modeling of the astronomspectrum of Pluto.
The inversion mode we use in an interactive trial-and-errorquence of direct modeling that progressively improves the fithe spectrophotometric curve of Pluto. For that purpose we trreproduce the main and more significant spectral characteriof the spectrum. These spectral characteristics are the shapintensity, and the position of the principal absorption bandsmethane, nitrogen, and carbon monoxide, as well as the presor the absence of very distinctive weak structures, all relateCH4 in its two phases. Aside from the bands of N2 at 2.15µmand CO at 2.35µm, the methane features can be arrangedphysical reasons into four independent groups (see Fig. 5):
Group 1: the four strong bands of theK region at 2.2, 2.32,2.37, and 2.42µm. This group also contains the slight bulgaround 2.28µm standing on the high-frequency wing of th2.32-µm band. This feature appears to be a significant strucbecause it can be reproduced by the modeling.
Group 2: the three medium bands of theH region at 1.67,1.72, and 1.79µm with, on the short-wavelength wing of thfirst and last ones, a small side band with a position thatypical of methane diluted in nitrogen.
Group 3: the weak band at 1.85µm (H region).Group 4: the weak absorption of pure methane that appe
at 1.69µm on the hill separating the bands at 1.67 and 1.72µm.
The detail of the iterative procedure to reach a good fit for Pluspectra is more complicated and intuitive than for Triton (Quiret al. 1999) where each compound clearly dominates parthespectrum. Generally, we first adjust the surface proporand grain size of the neutral unit to obtain the good albedthe maximum absorption of the saturated CH4 bands (2.32 and2.37µm). Second, the most intuitive part, we play around wthe parameters defining the relative distribution of pure andluted CH4 (representation dependent) and with the grain sizpure CH4 to best fit the strength of the other methane banthe detail of their shape (shoulders, wings, etc.), as well asmaximum albedos between them (especially around 1.68, 12.23, 2.35, and 2.41µm). We do not take too much into account the continuum around 1.52, 2.0, and 2.47µm and theweaker CH4 bands to constrain our fit. This is due to the laruncertainty on the very low values of our optical constant s[for absorption coefficients< 0.002 cm−1; see Fig. 1 of Quiricoet al. (1999)] and on the absolute albedo calibration. We thadjust the N2 grain size, with corresponding compensationthe diluted CH4 content, to best fit the 2.15-µm nitrogen bandstrength and shape. Finally, the diluted CO abundance is tuto fit the 2.35-µm band. Several iterations (and parameter setivity tests) between these steps are generally necessary to oa good fit. This fit is not unique but is one of the best compmises in terms of maximum deviation from the observation. Tassessment of a “good fit” is visual but, as a matter of faccorresponds to albedo differences of less than 0.005 over mo
the spectrum, with significant irreducible deviations kept beloET AL.
cal
se-ofto
tics, theof
enceto
for
eeure
is
ars
o’scoof
ionat
thdi-of
ds,the.74,-
eets
enof
nedsi-
btaino-he, itst of
0.02 (see Figs. 12, 13), except in the continuum and atmosphdisturbed ranges (<1.62, 1.8–2.1 and>2.44µm) for the reasonsexplained above.
3.4.2. Analysis
The first task to achieve to precise the horizontal and vertdistribution modes of the different detected ices is to confirm oof the main results of the spectroscopic analysis above, namthat there is a methane segregation at the surface of Pluto, wever its nature. For that, we try to simulate the Pluto spectrwith even simpler representations than the three basic onesi,Rs, and Rg), i.e., considering only one form of CH4, either pureor diluted.
The first Rmn, is a geographical mosaic of two units: onunit Um formed by a molecular mixture of N2–CH4–CO andone unit Un (N2). As is illustrated in Fig. 7, the simulationdo not reproduce the positions of the methane bands thaintermediate between the pure and the diluted forms. Moreoit appears impossible to simultaneously fit, even roughly,intensity and the width of the N2 and all the CH4 bands. Finallythe characteristic signature of pure CH4 at 1.69µm is absent.In a second simple representation, Rcn, the unit Um is replacedby a unit Uc composed of pure CH4 grains. Even more criticamisfits are obtained (shifts, shapes, much too strong 1.69µmband, etc.) (Fig. 7). These models prove more directly that bpure and diluted methane coexist at the surface of Pluto, butdid not tell which type of differentiation occurs.
To discriminate among the different basic types of differenation, we performedthe second task dedicated to test a serimore or less complex surface representations taking into accthese differentiations. They are divided into three families. Tfirst and the third families are respectively composed of the grular mixture Ri and the geographical distribution Rg, completedwith the neutral unit Un as described before. The representatioof the second family contain Un and at least one stratified un(generally Us). A given representation tries to reduce the limittions of the previous one. In this paper, we choose to focus onbest fits of the Pluto spectrum we have obtained so far, with erepresentation. In each case, we give a synthetic scheme orepresentation, the list of the associated modeling parameand their best-fit values.
a. “Geographical representation Rin with a granular mix-ture.” According to the phase diagram of the binary N2–CH4
molecular mixture (Prokhvatilov and Yantsevich 1983), the cexistence of almost pure CH4 grains (with small amount of nitro-gen) and nitrogen-rich grains saturated by CH4 can occur undersome temperature and mean composition conditions. If the cditions are met, the concentration of CH4 in the N2 matrix isconstrained to a value of approximately 2% around 40 K.the other hand the CO concentration is not constrained byN2–CO phase diagram.
Taking these constraints into account, our simplest repre
wtation Rin includes a unit Un and an intimate (granular) mixtureCH4 SEGREGATION ON PLUTO 433
FIG. 7. Two “best” simulations for a simple geographical mosaic of a neutral unit Un (fine-grained N2) and a molecular N2–CH4–CO mixture (Rmn) or pure
s
n
a
ol
e
its:
nditinest
athee butined
f
re
, therivedtheetic
CH4 (Rcn).
of pure CH4 ice grains and N2–CH4–CO grains with the CH4concentration fixed to 2% (unit Ui). We took the optical constantof pure CH4 for the former grains, as only very small spectrchanges are expected and measured for methane ice with 1nitrogen diluted in it (Quiricoet al.1996, Schmittet al.1998).For the 2% CH4–N2(–CO) grains we used the values of diluteCH4 (stated limit of validity, see Section 3.4.1.).
The best fit of the observation by this representation is shoin Fig. 8. Although a satisfactory fit of most of the CH4 bandsis achieved, there are at least three misfits that cannot be sowith this representation. First the strength of the N2 band is morethan an order of magnitude too weak. Increasing its strengthincreasing its grain size strongly and irremediably saturatesmethane bands. The high (2%) CH4 concentration in nitrogenis surely the principal reason. Second, the low-frequency wiof the strongest CH4 bands, especially at 2.32 and 2.37µm,are not well reproduced, indicating some lack of pure methclose to the surface. On the other hand, this representationtoo much pure CH4 at some depth, as witnessed by the tstrong 1.69-µm band. All these results, obtained with a verticahomogeneous distribution of the different types of grains, poto the existence of a stratification favoring the presence of pmethane close to the surface. This leads to the next typsurface representations.
b. “Geographical representations with stratification.”Am-
ong this category, the selected representations of the surfacal–2%
d
wn
lved
bythe
gs
nehaso
lyinture
of
Pluto were inspired by two preliminary trials. The first one Rsn
consists of a uniform geographical mosaic of two distinct un(1) a unit Us associated with the basic vertical distribution Rs (alayer of pure CH4 grains over a N2–CH4–CO mixture), and (2) aneutral unit Un composed of fine grains of nitrogen. The secotrial, Rrm, is similar to the first but the two layers of the unUs are inverted, giving Ur. Both simulations reproduce the mafeatures of the spectrum quite well but do not show all the findetails at the same time (the bulge around 2.28µm, the sidebands in theH region, and the absorption of pure methane1.69µm). Nevertheless they tell us that if a stratification of tdeposits really occurs, pure methane should lie at the surfacalso at some depth. This conclusion leads to the first retarepresentation:
i. Representation Rscn: A uniform geographical mosaic othree distinct units:
—a unit like Us of pure CH4 grains stratified over a N2–CH4–CO molecular mixture,
—a unit Uc formed by a homogeneous population of puCH4 grains,
—a unit Un of fine spectrally neutral grains of nitrogen.
Table II contains a synthetic scheme of this representationlist of the associated parameters, as well as the values defrom the best modeling of the spectrum. Figure 9 illustratesquality of the fit obtained between the observed and the synth
e ofspectra (see also Fig. 12).434 DOUTE ET AL.
FIG. 8. “Best” simulation for a simple geographical mosaic (Rin) of a neutral unitUn (fine-grained N2) and an intimate (granular) mixture of pure CH4 grainsand N –CH –CO molecular mixture grains (U).
t
r
ym-
bal
2 4 i
In the K region (Fig. 12b), the spectral position, the widand the shape of the methane bands at 2.20 and 2.32µm and, toa lesser extent, at 2.37 and 2.43µm are quite well reproducedIn particular, we can note that the slight bulge around 2.28µmis well fit by the synthetic spectrum. The absorption of nit
gen at 2 rly .15µm and of carbon monoxide at 2.35µm correctlyTABLE IIThree-Unit Surface Representation Rscn and Its Best-Fit Parameter Values
strong. The small side bands around 1.65 and 1.78µm clea
h,
.
o-
marks their presence by two small spikes. However, the asmetry that affects the wings of CO band is reversed. In theHregion (Fig. 12a), the simulated spectrum reproduces the gloshape of the intermediate bands of CH4, with the exception of theintensities of the 1.67- and 1.72-µm bands which are slightly too
5
Rsc
CH4 SEGREGATION ON PLUTO 43
to II.
p
t-u
t-
,in
ti
e
-
,u
a
eatedt of
terstions
.
tumter
re-ave
y aaneh-
p-
e
FIG. 9. Surface representation Rscn: the best fit of this representation
stand out and coincide rather well with the observation. Onother hand, we can note that the small absorption spike ofmethane at 1.69µm is present but a bit too weak. The shape aparticularly the width and intensity of the 1.85-µm band are nowell fitted, but between 1.82 and 1.95µm imperfect cancellation of telluric absorptions probably occurred in the spectrof Pluto.Finally the continuum around 1.5, 2, and 2.5µm is notwell reproduced, although the four weak CH4 bands at abou1.45, 1.48, 1.93, and 1.97µm are satisfactorily simulated. However, we know that in these ranges of very weak absorptionoptical constants of CH4 diluted in nitrogen are very uncertaand noisy (see Quiricoet al.1999).
ii. Representation Rsrn: For the second selected represention, we consider a uniform geographical mosaic of three distunits:
—a unit Us (CH4 stratified over a N2–CH4–CO molecularmixture),
—a second unit Ur similar to the previous one but with thlayers Lc and Lm inverted (N2–CH4–CO over pure CH4),
—a third unit Un of fine spectrally neutral grains of nitrogen.
Table III displays a synthetic scheme of the Rsrn representationthe list of the model parameters, as well as their best fit valFigure 10 shows the best fit obtained between the observedsimulated spectra (see also Fig. 12).
There are very few differences between this simulation
n, whose qualities seem to be globally equivalent. We onthe spectrum of Pluto. The best-fit parameter values are reported in Table
theurend
m
our
a-nct
es.and
nd
note, in theH region, the improvement of the intensity for thintermediate methane bands and the unfortunately correldegradation of the intensities of the two side bands. The fithe hill around 1.75µm also deteriorates a little bit.
It should be mentioned that, among the different paramethat are used in the current representation, the concentraC2
CH4andC2
CO as well as the mean grain sizerm2 of the molecularmixture superficial layer of the unit Ur are not well constrainedIndeed, this layer has a limited influence on the weak N2 band,which is essentially produced and controlled by the substraof the unit Us. We can be much more confident of the paramevalues obtained for the last-mentioned unit Us.
We do not proceed further with the stratified surface repsentations and keep the two best-fitting simulations we hobtained so far.
c. “Geographical representation Rmcn without stratification.”To test if a stratification really has to be invoked, we now studsimple geographical distribution of pure methane and methdiluted in N2 (with CO). We thus consider a uniform geograpical mosaic of three distinct units:
—a unit Um formed by a homogeneous population of N2–CH4–CO (molecular mixture) grains,
—a unit Uc of volatile deposits made of a homogeneous poulation of pure CH4 grains,
—a third unit Un of fine spectrally neutral grains of pur
lynitrogen.436
are con
FIG
DOUTE ET AL.
TABLE IIIThree-Unit Surface Representation Rsrn and Its Best-Fit Parameter Values
te also
vel
In fact, Rmcn correspond to the basic representation Rg plus theneutral unit Un. The synthetic scheme of the representation,list of the associated parameters, as well as the best-fit va
tained in Table IV. Figure 11 illustrates the quality of theof precision displayed by the previous modeling Rscn and Rsrn
. 10. Surface representation Rsrn: the best fit of this representation to
helues
fit reached between the observed and synthetic spectra (seFig. 13).
The global quality of this simulation comes close to the le
the spectrum of Pluto. The best-fit parameter values are reported in Table III.
437
ture at 2a
FIG. 1
CH4 SEGREGATION ON PLUTO
TABLE IVThree-Unit Surface Representation Rmcn and Its Best-Fit Parameter Values
t
thed
all2,
but without reaching it completely. If we exclusively concentraon the noticeable differences between Rsrn and Rmcn, we can saythat the latter fits the troughs on each side of the 2.42-µm bandof methane better but not the strong saturated absorptions atand 2.37µm, which are too broad (Fig. 13b). In addition, thband at 2.32µm does not correctly display the characterisbulge around 2.28µm, which is a significant feature. The signa
.2µm is also a bit too strong and wide. If we now pa 4
ttention to the nitrogen and the carbon monoxide bands, we stillrelative intensities).
1. Surface representation Rmcn: the best fit of this representation t
te
2.32eic-y
give preference to the Rsrn simulation. Indeed, the Rmcn simula-tion carves the former and lowers the latter a bit too much. Onother hand, the spectrum Rmcn better fits the observation towarthe continuum below 2.14µm than the representation Rsrn. Inthe H region (Fig. 13a), the agreement of the Rmcn simulationappears to be significantly less satisfactory if one refers tothe significant CH4 spectral features described by the groups3, and 4 in Section 3.4.1 (too narrow CHbands with less good
o the spectrum of Pluto. The best-fit parameter values are reported in Table IV.
438 DOUTE ET AL.
FIG. 12. Comparison between the simulations corresponding to the representations Rscn and Rsrn (see text): (a)H region, (b)K region.
CH4 SEGREGATION ON PLUTO 439
FIG. 13. Comparison between the simulations corresponding to the representations Rsrn and Rmcn (see text): (a)H region, (b)K region.
E
i
y
rr
bt
mt
v
o
a
f
a
a
r ofard
chrededss ofed
theveryof
t thed byndneov-
ilee asil-
wasthedel,
6),ort
s lo-d thenethe
rlytion
that
ye,e
ss ofh
face
440 DOUTE
We also tested a slightly more complicated representatRimn, by replacing the pure CH4 unit Uc of Rmcn with an intimatemixture Ui. This did not significantly improve the fit, especiallat the (too) strong bump around 2.28µm.
d. Representations Rscn and Rmcn with tholins. Tests of thecompatibility of tholins as an alternative material for the neutunit Un have been performed for the simplest stratified repsentation, Rscn, and for the geographical one, Rmcn. Interpolatedoptical constants from “Titan tholins” have been used (Khaet al. 1984). The main result is that very similar fits are otained with the same remaining localized misfits. However,presence of tholins implies very small grains (∼1–3µm) and aslightly increased surface area for this unit (about 16%) dueits lower albedo around 2.3–2.4µm. Only limited changes in theother parameters (mainly a decrease in the N2 grain size of themolecular mixture) are necessary to compensate the reductioarea of the other units. Materials having optical properties silar to those of these particular tholins seem then to be compawith the near-infrared spectrum of Pluto.
3.4.3. Synthesis
The series of simulations we have performed clearly allowto select as the best representations Rscnand Rsrn, which are char-acterized by the presence of one or two stratified units. Howesituations without any stratification, composed of a geograpcal mixture of three distinct units (pure CH4, molecular mixtureN2–CH4–CO, and fine grains of nitrogen) cannot be excludeven if they lead to less satisfactory fits of the observed spectrFor both kinds of simulations we point out that:
• A pure methane/diluted methane segregation appears tnecessary, whether it is vertical and horizontal or purely hozontal, but not granular. Note that “pure” methane can contup to 3% nitrogen around 35–40 K.• The areas with at least the superficial layer composed
pure methane always represent an important fraction ofwhole surface of the visible hemisphere of Pluto (88, 69, a46% for Rscn, Rsrn, and Rmcn, respectively). They are characterized by a granular texture of relatively small grains (a fehundred micrometers).• The concentrations of methane and carbon monoxide
luted in nitrogen are about five times higher on Pluto thanTriton (Quiricoet al.1999), despite the similarities between thtwo bodies.• The huge mean grain size (of the order of several centim
ters) that is obtained for the molecular mixture N2–CH4–COwhen it forms an optically semiinfinite substratum probablynot real. But it indicates that these deposits present veryoptical interfaces per unit volume and therefore probably haa compact polycrystalline texture with large evolved crystrather than an accumulation of independent particles. This kof situation was considered by Eluszkiewiczet al. (1998).• A spectrally neutral bright area (or one with a very sm
spectral slope and/or very weak features) is needed over ab
T AL.
on,
ale-
re-
he
to
n ofi-
ible
us
er,hi-
edum.
beri-in
ofthend-w
di-one
e-
isewvelsind
ll
12–16% of the surface. Pure nitrogen with fine grains (orde10 µm) was first considered here as the most straightforwcandidate, but this nitrogen can in fact contain as much CH4 andCO as the molecular mixture unit does. Simulations with sua mixture for the Un unit show that only very limited parametechanges for the three main surface representations are neto recover very similar fits. Alternatively, tholin-like materialare also found to be compatible with the spectral constraintthis unit. Any other bright neutral candidate in the near-infrarrange can be considered for this unit.• Water ice, the most widespread compound throughout
outer Solar System, does not seem to be necessary to obtainsatisfactory fits of the spectrum of Pluto near the maximumits visible lightcurve (around 200◦ longitude).
4. INTERPRETATION AND DISCUSSION
The existence of seasonal transport of volatile molecules asurface of Pluto was suggested 15 years ago and first modeleCruikshank and Silvaggio (1980), Trafton and Stern (1983), aStern and Trafton (1984). However at that time, only methawas identified and then considered in the models. The discery of solid N2 and CO (Owenet al.1993) led to considerationof nitrogen (with a trace amount of CO) as the main volatmolecule driving the interhemispheric transport and methanan “impurity” having a special behavior due to its lower volatity (Trafton 1990, Benchkoura 1996, Stansberryet al. 1996).As a result, a possible segregation of nitrogen and methanealready predicted by Lunine and Stevenson (1985) but forsurface of Triton. Using a globally averaged seasonal moTrafton (1990) predicted that a layer of pure CH4 only a fewmolecules thick should overlay the N2–CH4 solid solution (“de-tailed balancing layer”). On the other hand Benchkoura (199using a time-dependent latitudinal model of volatile transpincluding differential sublimation of the N2–CH4–CO mixture,obtained a geographical segregation of pure methane patchecated at the boundary between the substrate (water ice?) ansublimating N2-rich polar cap. These moderately hot methaareas cover a few to almost 40% of Pluto’s surface duringseasonal cycle.
Finally, starting from the spectroscopic analysis of the eaUKIRT–CGS4 spectra suggesting a possible vertical segregaof CH4 at the surface of Pluto (Schmittet al.1994, Quirico 1995),one of us performed laboratory experiments demonstratingslow sublimation of N2 ice with low CH4 concentration (0.2%)leads to the progressive formation of a fine-grained CH4 layer ontop of the sublimating N2-rich molecular mixture (see Stansberret al. 1996). Extrapolating this experiment to Pluto’s surfacStansberryet al. (1996) considered that the continuation of thsegregation process may result in an ever-increasing thicknethe layer of CH4 grains, finally leading to thick lag deposits withigher temperatures than the volatile N2-rich molecular mixture.Modeling the vertical exchanges of methane between the sur
outand the atmosphere, they concluded that only a few percent ofCH4 SEGREGATION ON PLUTO 441
FIG. 14. Possible “scenario” of the sublimation and recondensation processes that affect the ices and the surface state of Pluto, as suggested by our modeling.
h
b
r
(m
e
e
db
actrts
sea-re-adeedre-
ta-
d theod-rusttthe
hat
.ec-Hepth,of oldingover
chfineringnof
the surface covered by nearly pure and warm (∼53 K) methanepatches is sufficient to produce the high methane atmospabundance (1%) required to explain its elevated temperatuabout 100 K (Yelle and Lunine 1989, Lellouch 1994, Stroet al.1996).
Considering the different results presented in this papegether with some considerations from the volatile transport mels described above, we propose a “scenario” describingdifferent transport and segregation phenomena that affecidentified ices and that lead to the surface state as revealethe analysis of the UKIRT–CGS4 near-infrared observationFig. 14). Inversely, the following scenario coupled with sonumerical results from our spectrum modeling will allow usbetter constrain, or discard, some processes included in thferent transport models.
As witnessed by the occurrence of methane segregatioPluto, differential sublimation and/or chemical diffusion seto affect on one side the CH4 molecule and on the other sidN2 and CO within the volatile deposits N2–CH4–CO when sub-jected to solar radiation. At the surface, these processesate a residue of nearly pure methane forming a crust of vable thickness and divided into small grains by the escapthe N2 and the CO gases or by the diffusion process. Athey have been mixed and transported through the atmospthe gases settle as a molecular mixture by direct condenson the coldest areas. These areas, although probably extecover quite a small fraction of the observed hemisphere, becthe telescope essentially receives the light reflected arounsubsolar point, i.e., from the well-illuminated and thus sumating surfaces. Since they are in thermodynamic equilibr
with an atmosphere containing the same molecules, the depericre ofel
to-od-the
t thed byseee
todif-
n onme
cre-ari-e offterhere,ationnded,ause
theli-
ium
evolve more or less rapidly by metamorphism to a comppolycrystalline texture. Finally a new segregation cycle stawhen they undergo the effects of the Sun again as thesonal cycle takes place, temporally modulating the sunlightceived at a given latitude. This general scenario can be mmore specific if we look at the details of the results obtainfor the different units of the three main representations wetained.
First, the additional complexity of the “stratified representions” Rscnand Rsrn compared with the “geographical” one Rmcn
appears to be more consistent, at both the spectroscopic anphysical levels, with the observations and the transport mels. Indeed, over most of the surface, the pure methane coverlying the N2–CH4–CO mixture may be thin enough to lethe radiation reach the compact sublimating deposits. Usingresults of the model of Benchkoura (1996), we calculate tthe sublimation of a mixture with 0.5% CH4 leads to a methanecrust about 100–250µm thick over most of the sublimating areaSuch a layer is optically thin over most of the spectrum. Sond, the contribution of the areas displaying optically thick C4
residues, wherever they are situated at the surface or at dalso appears to be necessary. They attest to the existencesurfaces that have sublimated for a long time (possibly durprevious seasonal cycles), and eventually became coveredby small amounts of fresh deposits of the molecular mixture N2–CH4–CO (Rsrn case). We note that all our models need mubigger grains for these thick crusts than for the methanecrust. This may result from metamorphism processes occurin these warmer CH4 patches after the sublimation and diffusioof the underlying N and CO has stopped. This is another pro
osits2
of their age. Such a situation is particularly clear in the case of the
E
r
l
e
d
t
r
l
i
ra
r
t
n%h
c
,
i
thern
oes
om-uitetheas
rker
tonsromy the
icr of
ternclu-vedatse ininkex-ical
artic-first
om-theer-s ofouldsinghethe
ble:ighlinsanenceichthe
aybeane
dyon-
neu-xed
el-therial.
442 DOUTE
Rsrn representation where the grain sizes differ by one ordemagnitude. When it exists (Rsrn case), the superficial N2–CH4–CO frost layer consists of grains with sizes 20 times smacompared with the grains of the active sublimating N2–CH4–CO areas. This may effectively give evidence of the freshnof this thin deposit relative to the sublimating deposits, probbly condensed several tens of years ago and strongly evolvemetamorphism and densification. The very fine-grained N2-richarea should also have been deposited very recently. These gmay be linked with the micron-sized grains that have beenferred for the thin clouds observed in the atmosphere of Tri(Pollacket al.1990).
The existence of all these differentiation and stratification pcesses on Pluto surely arises from the larger concentrationmethane and carbon monoxide that we obtain for Pluto copared with Triton (Quiricoet al. 1999). Indeed, in the opticalayer the total abundance of methane is close to one-quartethat of nitrogen in the stratified models and reach values silar to the nitrogen abundance in the case of the geographmodel. This abundance is much larger than previously estimaon Pluto [2.6% (Owenet al.1993)] and than the value of 0.1%derived on Triton. Differentiation clearly also displays a molimited extent on Triton. An “upper limit” of 10% surface arecovered by pure methane has been estimated for Triton usingsame model, although there is no spectral evidence of its pence (Quiricoet al.1999). On Pluto, they imply the presence oan important amount of superficial pure methane which is ctainly linked, as modeled by Stansberryet al. (1996), with thedetection of high CH4 concentrations in the atmosphere (∼1%)(e.g., Younget al.1997). Although the thin CH4 crust probablydid not contribute to the upward CH4 mass flux necessary tomaintain the atmospheric concentration in this model, therelarge enough thick CH4 deposits (surface proportion∼20%) inboth stratified representations to comply with these calculatioIndeed only 1 to 3% of surface covered by hot CH4 patches isenough. It should be noted that in our stratified models mosthe methane molecules are in the form of pure ice. Indeed,fraction of diluted CH4 is less than 2% of the total methane abudance in the optical layer. This fraction is even smaller (0.3in our geographical model. This suggests that differentiationefficiently segregated the nitrogen and methane molecules.nondetection of water ice, which is only confirmed at the mment for the longitudes around the maximum of the lightcurvas well as the absence of CO2 ice, is another particularity ofPluto compared with Triton (Cruikshanket al. 1997, Quiricoet al. 1999). As water ice is expected to be present as a rolithosphere by the bulk composition models (see, e.g., St1998), the absence of such bare surface units may meanthe amount of volatile materials, or of their derived productsquite large on the surface. The thick CH4 crust may also hidethe H2O and CO2 ices on Pluto, while the complete sublimatioof the volatiles on Triton uncovers them.
Now we can determine what the best surface representat
we have derived from the analysis of the UKIRT–CGS4 neaT AL.
of
ler
ssa-by
rainsin-on
o-s ofm-
r ofm-icalted
e
thees-fer-
are
ns.
ofthe-)
asTheo-e,
kyernthatis
n
ons
infrared data would imply for the appearance of Pluto invisible if it were imaged by the HST as performed by Steet al. (1997). The methane, whether it is pure or diluted, dnot present any noticeable signature below 0.62µm, with maybean extremely weak residual absorption due to overtone and cbination bands. Nitrogen and carbon monoxide, which are qfeatureless in the near infrared, are totally transparent invisible. Consequently, a 410-nm HST image of Pluto’s diskdescribed by the Rscn, Rsrn, or Rmcn representation would showa quite bright surface on average, with some slightly daregions corresponding to the compact polycrystalline N2–CH4–CO sublimating deposits. Indeed for the latter area, the photravel over a much longer path length before they can exit fthe surface and have then more chance to be absorbed bresidual absorption of CH4. The problem is that the dynamrange of albedo across the planet locally exceeds a facto5 (25% when disk-averaged) on the real HST images (Set al.1997). Such a contrast cannot be explained by the exsive occurrence of highly translucent materials like the evoldeposits of N2–CH4–CO. The CCD spectra of Pluto recordedvarious rotational phases also display a systematic decreaalbedo toward the blue (red slope) (Buie and Fink 1987, Fand Di Santi 1988, Grundy and Fink 1996) that cannot beplained by the ice absorption so far. Nevertheless photochemprocesses can affect these basic molecules. They may be pularly intense within the superficial crusts of pure methane,forming ethane, possibly identified through the 1.75-µm band,next larger organic molecules, and then leading to a very cplex mixture with some compounds strongly absorbing invisible. The results of the HST, CCD, and UKIRT–CGS4 obsvations provide useful constraints on the optical propertiethese unknown compounds. Their absorption coefficient shpresent quite high values in the visible progressively decreaaround 1.0µm from where it should have weak values in tnear infrared, contrasting by a factor of 100 at least withvalue at 410 nm. Different kinds of materials may be suitasome mixtures of hydrocarbons and nitriles displaying a hcarbon-to-hydrogen ratio (Gougeon 1998), like organic tho(Khareet al.1984) and kerogen for example. But minerals calso be considered. For most organic molecules, the existin the near-infrared of many bands of various intensity, whare unobserved, implies that their individual proportions onsurface of Pluto must be less than a certain threshold, mof the order of 0.001%, given the absorption level of methin the near-infrared continuum and theS/N of the spectrum.More complex mixtures, like tholin, as proposed by Grunand Fink (1996), have optical constants consistent with our cstraints. We showed that they may replace nitrogen in thetral unit of our representations, but they may be also miwith methane in the old and thick CH4 crusts. This problemshould be investigated more thoroughly by conjointly moding the visible and the near-infrared spectra to constraincomposition, the location, and the abundance of this mate
r-So, we can note that the search for the dark components onI
e
e
rcn
mc
in
s
th
v
had
e
e
2
u
t
re of
n,
ss,
st
delnear
98.
27,
llite
uto
ared
ent
s,
W.ted
the
azy
rface
e93.
tmo-
ui-
des
dro-
N
n,
CH4 SEGREGAT
Pluto, even if it is a bit constrained, remains an open qution.
5. CONCLUSION
The use of a polyvalent and efficient modeling tool, applito the analysis of high-quality spectrophotometric data, allowus to provide convincing evidence for geographical and vecal segregations of solid methane from the other detected iN2 and CO, on the surface of Pluto. This implies the preseof extended surfaces with superficial pure methane whereintense photochemistry may take place, leading to the darkterial evidenced by the visible HST images. Additionally, sumethane-rich surfaces provide a possible explanation forlarge abundance of CH4 in the atmosphere. All the ices, N2,CH4, and CO, seem to condense, evolve, and sublimatecomplex manner during the seasonal cycle. The kind of hozontal and vertical distributions we have obtained, at leastone hemisphere, allowed us to give a specific description of thprocesses.
This work represents only the first part of an important sries of analyses to be performed in the near future on obvations that correspond to different rotational phases of PluSuch analyses will certainly reveal other interesting informtion, in particular concerning the geographical localizationthe different surface units distinguished by the modeling ofinfrared spectra. As a consequence, it will be easier to studylink between these units and the albedo features of the HSTible images whose chemical and physical nature could be beconstrained.
ACKNOWLEDGMENTS
We thank the staff of UKIRT for their support of these observations. TJoint Astronomy Centre, on behalf of the United Kingdom Particle PhysicsAstronomy Research Council, operates UKIRT. This work has been supportethe French Programme National de Plan´etologie. Sylvain Dout´e acknowledges aMESR grant from the French Minist`ere de la Recherche. The LGGE is associatwith the University Joseph Fourier of Grenoble, France. Finally, we particulathank John Stansberry and Leslie Young as reviewers for their particularly uscomments and suggestions which increase the quality of this paper.
REFERENCES
Benchkoura, A. 1996.Modelisation du transport de volatilsa la surface desplanetes et satellites glaces du Systeme Solaire: Applicationa Triton et Pluton.Thesis, LGGE, University Joseph Fourier, Grenoble.
Bosh, A. S., L. A. Young, J. L. Elliot, H. B. Hammel, and R. L. Baron 199Photometric variability of Charon at 2.2µ. Icarus95, 319–324.
Buie, M. W., and U. Fink 1987. Methane absorption variations in the spectrof Pluto.Icarus70, 483–498.
Buie, M. W., and S. K. Shriver 1994. The distribution of water frost on CharoIcarus108, 225–233.
Buie, M. W., D. P. Cruikshank, L. A. Lebofsky, and E. F. Tedesco 1987. Wafrost on Charon.Nature329, 522–523.
Buie, M. W., D. J. Tholen, and L. H. Wasserman 1997. Separate lightcurvesPluto and Charon.Icarus125, 233–244.
stri-
ON ON PLUTO 443
s-
dedti-es,ceana-
hthe
ari-forese
e-er-to.a-ofe
theis-tter
endby
drlyful
.
m
n.
er
of
Cruikshank, D. P., and P. M. Silvaggio 1980. The surface and atmosphePluto.Icarus41, 96–102.
Cruikshank, D. P., T. L. Roush, J. Moore, M. Sykes, T. C. Owen, R. H. Browand K. A. Tryka 1997. The surfaces of Pluto and Charon. InPluto and Charon(S. A. Stern and D. J. Tholen, Eds.), pp. 221–267. Univ. of Arizona PreTucson.
Doute, S. 1998.Modelisation numerique de la reflectance spectrale des surfaceglacees du Systeme Solaire: Applicationa l’analyse de spectres de Triton ePluton et au traitement d’images hyperspectrales NIMS de Io. Thesis, LGGE,Universite Paris-7.
Doute, S., and B. Schmitt 1998. A multi-layer bidirectional reflectance mofor the analysis of planetary surface hyperspectral images at visible andinfrared wavelengths.J. Geophys. Res. E103, 31,367–31,390.
Eluszkiewicz, J., J. Leliwa-Kopystynski, K. J. Kossacki, and N. Maeno 19Metamorphism of Solar System ices. InSolar System Ices(B. Schmitt,C. de Bergh, and M. Festou, Eds.), Astrophys. Space Sci. Lib., Vol. 2pp. 119–138. Kluwer Academic, Dordrecht.
Fink, U., and M. A. DiSanti 1988. The separate spectra of Pluto and its sateCharon.Astron. J. 95, 229–236.
Gougeon, S. 1998.Contributions experimentalesa l’astrophysique en rayonsX. Thesis, Universit´e Paris-7.
Grundy, W. M., and U. Fink 1996. Synoptic CCD spectrophotometry of Plover the past 15 years.Icarus124, 329–343.
Grundy, W. M., and B. Schmitt 1998. The temperature-dependent near-infrabsorption spectrum of hexagonal H2O ice.J. Geophys. Res. E103, 25,803–25,822.
Grundy, W. M., B. Schmitt, and E. Quirico 1993. The temperature dependspectra of alpha and beta nitrogen ice with application to Triton.Icarus105,254–258.
Hapke, B. 1993. Summary of the irregular-particle model. InTheory of Re-flectance and Emittance Spectroscopy, pp. 144–146. Cambridge Univ. PresCambridge.
Khare, B. N., C. Sagan, E. T. Arakawa, F. Suits, T. A. Callcott, and M.Williams 1984. Optical constants of organic tholins produced in a simulatitanian atmosphere: From soft X-ray to microwave frequencies.Icarus 60,127–137.
Leitherer, C., and B. Wolf 1984. Early-type stars in OB associations ininfrared. I. Extinction law and IR excesses.Astron. Astrophys. 132, 151–162.
Lellouch, E. 1994. The thermal structure of Pluto’s atmosphere: Clear vs. hmodels.Icarus108, 255–264.
Lunine, J. I., and D. J. Stevenson 1985. Physical state of volatiles on the suof Triton. Nature317, 238–240.
Owen, T. C., T. L. Roush, D. P. Cruikshank, J. L. Elliot, L. A. Young, C. dBergh, B. Schmitt, T. R. Geballe, R. H. Brown, and M. J. Bartholomew 19Surface ices and the atmospheric composition of Pluto.Science261, 745–748.
Pollack, J. B., J. M. Schwartz, and K. Rages 1990. Scatterers in Triton’s asphere: Implication for the seasonal volatile cycle.Science250, 440–443.
Prokhvatilov, A. I., and L. D. Yantsevich 1983. X-ray investigation of the eqlibrium phase diagram of CH4–N2 solid mixtures.Sov. J. Low Temp. Phys. 9,94–98.
Quirico, E. 1995.Etudes spectroscopiques proche infrarouges de solimoleculaires: Application a l’etude des surfaces glacees de Triton et Plu-ton.Thesis, LGGE, Universit´e Joseph Fourier, Grenoble.
Quirico, E., and B. Schmitt 1997a. Near infrared spectroscopy of simple hycarbons and carbon oxides diluted in solid N2 and as pure ices: Implicationfor Triton and Pluto.Icarus127, 354–378.
Quirico, E., and B. Schmitt 1997b. A spectroscopic study of CO diluted in2
Ice: Applications for Triton and Pluto.Icarus128, 181–188.
Quirico, E., S. Dout´e, B. Schmitt, C. de Bergh, D. P. Cruikshank, T. C. OweT. R. Geballe, and T. L. Roush 1999. Composition, physical state and di
bution of ices at the surface of Triton.Icarus139, 159–178.C
l
n
w
7,
a-
s
n
k
-
ns
ne
.
444 DOUTE
Quirico, E., B. Schmitt, R. Bini, and P. R. Salvi 1996. Spectroscopy of soices of astrophysical interest: SO2, N2 and N2 : CH4 mixtures.Planet. SpaceSci. 44, 973–986.
Roush, T. L. 1994. Charon: More than water ice?Icarus108, 243–254.
Roush, T., D. P. Cruikshank, J. B. Pollack, E. F. Young, and M. J. Bartholom1996. Near-infrared spectral geometric albedos of Charon and Pluto:straints on Charon’s surface composition.Icarus119, 214–218.
Schmitt, B., S. Dout´e, E. Quirico, A. Benchkoura, C. de Bergh, T. C. Owen, aD. P. Cruikshank 1994. The state and composition of the surface of PLaboratory experiments and numerical modeling.Bull. Am. Astron. Soc. 26,1170.
Schmitt, B., E. Quirico, and E. Lellouch 1992. Near infrared spectra of potesolids at the surface of Titan. InProceedings of Symposium on Titan, ESASpec. Publ. S
¯P-338, pp. 383–388.
Schmitt, B., E. Quirico, F. Trotta, and W. M. Grundy 1998. Optical propertiesices from UV to infrared. InSolar System Ices(B. Schmitt, C. de Bergh, andM. Festou, Eds.), Astrophys. Space Sci. Lib., Vol. 227, pp. 199–240. KluAcademic, Dordrecht.
Smith, E. V. P., and D. M. Gottlieb 1974. Solar flux and its variations.Space Sci.Rev. 16, 771–802.
Stansberry, J. A., J. R. Spencer, B. Schmitt, A. Benchkoura, R. V. Yelle, VRoger, and J. I. Lunine 1996. A model for the overbundance of methan
the atmospheres of Pluto and Triton. Planet.Space Sci. Rev. 44, 1051–1063.ET AL.
me
ewon-
nduto:
tial
of
er
. Y.e in
Stern, S. A. 1998. Pluto and the Kuiper disk. InSolar System Ices(B. Schmitt,C. de Bergh, and M. Festou, Eds.), Astrophys. Space Sci. Lib., Vol. 22pp. 685–695. Kluwer Academic, Dordrecht.
Stern, S. A., and L. Trafton 1984. Constraints on bulk composition, sesonal variation, and global dynamics of Pluto’s atmosphere.Icarus57, 231–240.
Stern, S. A., M. W. Buie, and L. M. Trafton 1997. HST high-resolution imageand maps of Pluto.Astron. J. 113, 827–843.
Strobel, D. F., X. Zhu, M. E. Summers, and M. H. Stevens 1996. Othe vertical thermal structure of Pluto’s atmosphere.Icarus 120, 266–289.
Trafton, L. A. 1990. Two-component volatile atmosphere for Pluto. I. The bulhydrodynamic escape regime.Astrophys. J. 359, 512–523.
Trafton, L., and S. A. Stern 1983. On the global distribution of Pluto’s atmosphere.Astrophys. J. 267, 872–881.
Tryka, K. A., R. H. Brown, D. P. Cruikshank, T. C. Owen, T. R. Geballe, andC. deBergh 1994. Temperature of nitrogen ice on Pluto and its implicatiofor flux measurements.Icarus112, 513–527.
Yelle, R. V., and J. I. Lunine 1989. Evidence for a molecule heavier than methain the atmosphere of Pluto.Nature339, 288–290.
Young, L. A., J. L. Elliot, A. Tokunaga, C. de Bergh, and T. C. Owen 1997
Detection of gaseous methane on Pluto.Icarus127, 258–262.