L-TR-S-0286L -89

Ion and Electron Interactions atThermal and Suprathermal Energies

David SmithNigel 0. AdamsCharles R. Herd

In

University of BirminghamSchool of Physics and Space Research

(Birmingham B15 2TT, UNITED KINGDOM

I 30 September 1989 O I& DTIC

Scientific Report No. 2 JAN! 1 1990

APPROVED FOR PUBLIC RELEASE; DISTRIBUTION UNLIMITED

GEOPHYSICS LABORATORYAIR FORCE SYSTEMS COMMANDUNITED STATES AIR FORCEHANSCOM AIR FORCE BASE, MASSACHUSETTS 01731-5000

90 01 16 120

"This technical report has been reviewed and is approved for publication"

" '- I - / " .

JOHN F. PAULSON JOHN E. RASMUSSENContract Manager Branch Chief

FOR THE COMMANDER

ROBERT A. SKRIVANEKDivision Director

This report has been reviewed by the ESD Public Affairs Office (PA) and isreleasable to the National Technical Information Service (NTIS).

Qualified requestors may obtain additional copies from the DefenseTechnical Information Center. All others should apply to the NationalTechnical Information Service.

If your address has changed, or if you wish to be removed from the mailinglist, or if the addressee is no longer employed by your organization,please notify GL/IMA, Hanscom AFB, MA 01731. This will assist us inmaintaining a current mailing list.

Do not return copies of this report unless contractual obligations ornotices on a specific document requires that it be returned.

Unclassified

SeCURITY CLASSIFICATION OF THIS PAGE

REPORT DOCUMENTATION PAGEla. REPORT SECURITY CLASSIFICATION lb. RESTRICTIVE MARKINGSUnclassified2a. SECURITY CLASSIFICATION AUTHORITY 3. DISTRIBUTION/ AVAILABILITY OF REPORT

Approved for public release;2b. DECLASSIFICATION / DOWNGRADING SCHEDULE Distribution unlimited

4. PERFORMING ORGANIZATION REFORT NUMBER(S) S. MONITORING ORGANIZATION REPORT NUMBER(S)

GL-TR-89-0266

6a. NAME OF PERFORMING ORGANIZATION 6b. OFFICE SYMBOL 7a NAME OF MONITORING ORGANIZATION

University of Birmingham (If applicable) European Office of Aerospace Research and

I_ Development6c. ADDRESS (City, State, and ZIP Code) 7b ADDRESS (City, State, and ZIP Code)

School of Physics & Space Research 223/231 Old Marylebone Road

Birmingham B15 2TT, UNITED KINGDOM London NWI 5TH, UNITED KINGDOM

8a. NAME OF FUNDING /SPONSORING (8b. OFFICE SYMBOL 9 PROCUREMENT INSTRUMENT IDENTIFICATION NUMBER

ORGANIZATION (If applicable) AFOSR-87-0355

Geophysics Laboratory LID

8c. ADDRESS (City, State, and ZIP Code) 10 SOURCE OF FUNDING NUMBERSHanscom AFB PROGRAM PROJECT ITASK WORK UNIT

Massachusetts 01731-5000 ELEMENT NO NO. NO. ACCESSION NO62101F 2303 GI AJ

I1 TITLE (Include Security Classification)

Ion and Electron Interactions at Thermal and Supratbermal Energies

12. PERSONAL AUTHOR(S)David Smith, Nigel G. Adams, Charles R. Herd

13a. TYPE OF REPORT 13b. TIME COVERED 14 DATE OFREPOqT (Yeir Month, Day) 15. PAGE COUNTScientific #2 1 FROM TO 1989 Septemoer 0 _ 122

16 SUPPLEMENTARY NOTATION

17. COSATI CODES I 18 SUBJECT TERMS (Continue on reverse if nece.sr and identifyby ocknrumber)

FIELD GROUP SUB-GROUP. Electron attachment ,Flowing atte g ow Langmuir

FIELGROUP__ sU_-GROUP- IDissociative recombination probe /

Ioninolecule reactions ".Selected ion flow tube.

19. ABSTRACT (Cotitinue on reverse if necessary and identify by block number)The flowing afterglow Langmuir probe (FALP) apparatus has been used to study (i) dissociative

electron attachment reactions of several haloethanes; (ii) the neutral products of the

dissociative recombination reactions of several ground state polyatomic ions (using the FALP

together with LIF and VUV spectroscopic techniques). The selected ion flow tube (SIFT) appar-

atus has been used to study (iii) the reactions of ions in the series PH +(n=O to 4) and thereactions of Kr+, Xe+, Kr2 + and Xe+ with several molecular gases. Details of these studies

are given in seven appendices to the report.

20. DISTRIBUTION/AVAILABILITY OF ABSTRACT 21. ABSTRACT SECURITY CLASSIFICATION

OUNCLASSIFIED/UNLIMITED E0 SAME AS RPT 0 DTIC USERS Unclassified

22a. NAME OF RESPONSIBLE INDIVIDUAL 22b. TELEPHONE (Include AeaCode),22c OFFICE SYMBOLJohn Paulson ,. GL/LID

DD FORM 1473. 84 MAR 83 APR edition may be used until exhausted. SECURITY CLASSIFICATION OF THIS PAGEAll other editions are obsolete. Unclassified

PREFACE

Page No.

I. INTRODUCTION 1

II. SUMMARY OF RESULTS 2

III. CONCLUSIONS 5

APPENDIX 1 IONIC REACTIONS IN ATMOSPHERIC, INTERSTELLAR 6

AND LABORATORY PLASMAS.

APPENDIX 2 FALP STUDIES OF ELECTRON ATTACHMENT REACTIONS 27

OF C 6 F 5C1, C6F 5Br AND C 6F-I.

APPENDIX 3 A STUDY OF DISSOCIATIVE ELECTRON ATTACHMENT 40

TO CF 3 SO 3 CH 3 AND CF 3 SO 3 C 2 H 5 TRIFLATE ESTERS

USING THE FALP APPARATUS.

APPENDIX 4 STUDIES OF DISSOCIATIVE ELECTRON ATTACHMENT TO 47

SOME HALOETHANES USING THE FALP APPARATUS:

COMPARISON WITH DATA OBTAINED USING NON-THERMAL

TECHNIQUES.

APPENDIX 5 DEVELOPMENT OF THE FLOWING AFTERGLOW/LANGMUIR 62

PROBE TECHNIQUE FOR STUDYING THE NEUTRAL PRODUCTS

OF DISSOCIATIVE RECOMBINATION USING SPECTROSCOPIC

TECHNIQUES : OH PRODUCTION IN THE HCO 2+ + e

REACTION.

APPENDIX 6 OH PRODUCTION IN THE DISSOCIATIVE RECOMBINATION 74

OF H 30, HCO 2 AND N2 OH+ : COMPARISON WITH

THEORY AkTI INTERSTELLAR IMPLICATIONS.

APPENDIX 7 A SELECTL. ')N FLOW TUBE STUDY OF THE REACTIONS 96

OF THE PHn + IONS (n - 0 to 4) WITH SEVERAL

MOLECULAR GASES AT 300K.

APPENDIX 8 REACTIONS OF Kr +, Kr 2 , Xe + and Xe 2 + IONS WITH 104

SEVERAL MOLECULAR GASES AT 300K.

(iii)

PREFACE

This work is part of a larger programme of ionic

reaction studies at thermal energies conducted by the senior

authors of this report. The overall programme includes studies

of ion/neutral reactions, electron attachment, electron/ion

recombination, ion/ion recombination, and other plasma reaction

processes. The work is largely intended as a contribution to

the physics and chemistry of natural plasmas, such as the

ionosphere and the interstellar medium, and of laboratory plasma

media, such as gas laser systems, etchant plasmas and combustion

plasmas. A great deal of relevant data has been obtained

principally by exploiting the variable-temperature selected ion

flow tube (VT-SIFT), the variable-temperature selected ion

flow drift tube (VT-SIFDT) and the variable-temperature flowing

afterglow/Langmuir probe (VT-FALP) techniques which were

developed in our laboratory. Part of the overall programme is

supported by a grant from the Science and Engineering Research

Council. ,

(v)

I INTRODUCTION

Our major commitment under the terms of the current grant is

to study various gas-phase ionic reaction processes at thermal

and near-thermal energies, with special reference to processes

that may occur in the terrestrial atmosphere and in combustion

and discharge plasmas. Such processes include ion/neutral

reactions, electron attahment, electron/ion recombination and

ion/ion recombination (mutual neutralization). All these

processes can be studied in our laboratory following our

development and construction of the very versatile selected ion

flow drift tube (SIFDT) and flowing afterglow/Langmuir probe

(FALP) techniques. Using these techniques, rate coefficients

and ion products of reactions can be determined in many cases

over the wide temperature range 80-600K and at reactant

Ion/reactant neutral centre-of-mass energies up to about one

electron volt in some cases. Numerous reactions have been

studied during the past few years and many have been presented in

previous reports and research publications. The relevance of

much of our experimental work to the terrestrial atmosphere,

interstellar gas clouds and laboratory etchant plasmas is

explained in the brief review paper which is presented as

Appendix I to this report.

During the period covered by the current grant, we have added

laser and vacuum ultraviolet spectroscopic techniques to the

already versatile FALP apparatus and this has greatly extended

the processes that we can study. In this report, we refer to

1

the further studieg we have made of electron attachment.

electron/lon recombinetion, specifically the neutral products of

some such reactions, and some ion/molecule reactions. Details

of much of this work are given in the reprints and preprints of

research papers which are included as Appendices 2-8 of this

report. A brief summary of the most important results is given

in the next section.

II SUMMARY OF RESULTS

(a) Electron Attachment

In the previous Interim Scientific Report No.1, we mentioned

the results of our study of the attachment reactions of C6F5Cl,

C6FsBr and C6F I. The detailed results of this study are given

in this report as Appendix 2. Also mentioned briefly in the

previous report was our study of the attachment reactions of

methyl and ethyl triflate and the details of this study are

presented in Appendix 3.

More recently, we have completed a study of the dissociative

attachment reactions of four haloethanes, CH 3CCl3, CH2CICHC1 2 ,

CF 3CC1 3 and CF 2ClCFC1 2 , which was carried out under truly

therualised conditions at 298, 385 and 470K using our flowing

afterglow/Languuir probe (FALP) apparatus. In all four

reactions the only product ion was ClV; in the majority of the

reactions the attachment rate coefficient, 8 , increases rapidly

with increasing temperature. The values of S obtained have been

compared with those derived from non-thermal swarm and

photoionization techniques in which the 8 are observed to

decrease with increasing mean electron energy, K, at a fixed gas

temperature of 298K. The different senses of the variations

2

of B with temperature and with B are discussed, and are

rationalised in terms of the thermal and non-thermal nature of

the different experiments, in Appendix 4 to this report.

Our most recent attachment study has involved some

bromomethanes and bromoethanes. The surprising result of this

study is that dissociative attachment reactions between these

molecules results in the production of not only Br- ions as

expected but also a significant fraction of Br ions. The

details of these observations will be discussed in a research

paper which is now in the draft stage.

(b) Electron/Ion Dissociative Recombination

We have previously used the FALP apparatus to determine the

dissociative recombination reaction rate coefficients,a, for a

wide variety of polyatomic ions and the data have been presented

in several research papers and in previous scientific reports.

During the last year, we have begun a programme of measurements

to determine the neutral products of the dissociative

recombination reactions of some polyatomic ions. This has been

made possible by the application of laser induced fluorescence

(LIF) and vacuum ultraviolet (VUV) spectroscopic techniques to

recombining FALP plasmas consisting of electrons and one ionic

species only. In the first measurements, we have determined the

fraction of OH molecules (by LIF) produced in the recombination+ 4

with electrons of the ions HCO 2 , N20H , 02H and H36 . The

intensity of the LIF from OH radicals was calibrated against the

OH number density by first creating a swarm of H atoms in the

flow tube in the absence of ionization, determining the number

density of H atoms in the plasma (by utilizing VUV absorption)

and then converting all of the H atoms to OH radicals using the

3

reaction H+M02 *OH+NO. The details of the experimental approach

to this interesting study are given in Appendix 5 to this report.

The fractions of OH generated following dissociative

recombination of the above-mentloned ions are reported in the

preprint given as Appendix 6 to this report.

These data on the neutral molecular products of dissociative

recombination of ground state polyatomic ions are the first ever

obtained. Work is proceeding to determine the fractions of H

and 0 atoms produced in the above recombination reaqztions.

Preliminary data have been obtained and these will be

consolidated and then presented in research publications and

future reports.

(c) Ion/Neutral Reactions

We have continued to exploit our versatile selected ion flow

tube (SIFT) apparatus to study a range of ion/molecule reactions.

Specifically, we have carried out a detailed study at 300K of

the reactions of the phosphorus-bearing ions PHn+(n=0 to 4) with

many molecular gases. Many reaction processes are observed,

but especially interesting are the phosphorus atom insertion

reactions which produce ions such as HnPN , HnPO +, HnPS and

HnCMP . A considerable amount of kinetic data and information

has been obtained from this study. For example, upper limits to

the heats of formation of HPO, H2 PO , PCH 2 + and HCP have been

obtained. The details are given in the research paper included

as Appendix 7 to this report.

We have also carried out a detailed study at 300K of the

reactions of the spin states (2p 2 Pj) of Kr' and Xe* and the2_ T

reactions of Kr 2 + and Xe2 using our SIFT apparatus. Again, a

4

great deal of information has been obtained from this substantial

study, including the dissociation energies of the molecular ions

Kr 2+ and Xe2+ . Full details are presented in the research

paper included as Appendix 8 to this report.

III CONCLUSIONS

Our studies of electron attachment reactions, ion/neutral

reactions and especially the products of dissociative

recombination reactions have continued apace during the last

year. Some seven research papers (included as appendices)

summarise the results obtained. Our determination of the

neutral products of the dissociative recombination reactions of

ground state polyatomic ions are the first ever obtained. The

LIF and VUV spectroscopic techniques that we have developed,

together with optical spectroscopy, will be exploited further in

the coming years to determine the neutral products of a greater

range of dissociative recombination reactions. Optical

emissions are to be expected from certain positive ion/electron

recombination reactions and also from some positive ion/negative

ion neutralization reactions. A concerted effort will be made

to look for such emissions using the FALP in the near future.

Our programme of electron attachment studies will also be

continued using the FALP., Using the SIFT, we will soon

initiate a detailed survey of the reactions of ions observed from+&

silane (SiHn reactions), germane (GeHn reactions), and diborane

(BH2 and B2Hn reactions).

5

APPENDIX 1

IONIC REACTIONS IN ATMOSPHERIC, INTERSTELLAR

AND LABORATORY PLASMAS

N.G. ADAMS AND D. SMITH

Contemp.Phys., 29, 559 (1988)

6

CONTFMP. PHYS.. 1988. VOL. 29, No. 6, 559-578

Ionic reactions in atmospheric, interstellarand laboratory plasmas

Nigel G. Adams and David Smith, Department of Space Research,University of Birmingham, Birmingham BI5 2TT England

ABSTRAC'T. The ionic reactions that can occur in ionized gases, and especially those in naturally-occurrinelow-temperature gaseous plasmas. are discussed These reactions include several types of positive ion andnegative ion reactions with molecules. tin-electron dssociative recombmation reactions. electronattachment reactions and ton-ion mutual neutralization reactions. The laboratory techniques used to stud*and to determine the rate coefficents for these reactions are brieily decribed. Three case studies arepresented which illustrate how the various types of tonic reactions modify the ionic and neutral compositionof ion ed media. These case studies describe the reactions which (1) lead to the observed ions in the terrestnalatmosphere; (2) produce the neutral molecules and molecular ions that ar detected in interstellar gas clouds:and (3)can occur in the laboratory discharge plasmas that ire used to etch the surfaces of semcrnduclors inhe preparation of microprocessorn.

I. IntroductionMixtures of many gases are quite stable and unreactive at normal temperatures. Forexample, the terrestrial atmosphere near sea level is fortunately very stable since norcaction occurs bctwccn N z and 02 and so toxic gases such as NO2 are not produced insignificant amounts. Also the problem of atmospheric pollution would be much worseif common atmospheric pollutants were reactive with atmospheric gases. However,when gases are exposed to a source of radiation, intense electric fields, and such like, sothat the molecules are dissociated to atoms or especially when the neutral speciesbecome ionized, their reactivity increases enormously. Thus, when ionization is createdin a gas mixture, complex ionic reaction processes are initiated which may greatlymodify the molecular species within the ionized gas. For each positive, singly-chargedatom or molecule produced in the ionization process, a free electron must also beproduced so that the ionized gas remains electrically neutral overall. Then the mediummay be termed a gaseous plasma, although this term has to be applied carefully sinceother cnteria must be obeyed for ionized gases to be properly termed gaseous plasmas(McDaniel 1964). In this article, we will be concerned largely with weakly ionized gasesat relatively low temperatures, many of which can be correcdy termed low-temperaturegaseous plasmas. These include the terrestrial ionosphere (200 K < T 4 2000 K), someinterstellar gas clouds (10 K < T < 200 K) and laboratory etchant plasmas. In someplasmas, the temperatures of the component particles are not eq'al (for example, theelectron temperature in the F-region of the terrestrial ionosphere exceeds the ion andneutral temperatures by about a factor of three; see Smith and Adams (1980)) and it isimp,)rtant to appreciate this fact when attempting to understand the complex reactionprocesses that occur. The electrons that are generated in the initial ionization processmay remain free for a period or become 'attached' to neutral species to form negativeions. Free electrons and negative ions can react with positive ions, thus neutralizing theplasma and acting to return the medium to the neutral gaseous state. It is these'recombination' reactions together with the numerous ion-neutral reactions whichcombine to produce the rich ion physics and chemistry of low-temperature plasmaswhich we describe in this article.

7

56 N. G. Adams and D. Smith

When gaseous plasmas are only weakly ionized, the collision frequencies betweenions and neutrals will greatly exceed those between charged particles (positive ions.negative ions and electrons) and so ion-neutral reactions and electron-neutral(attachment) reactions will generally precede recombination reactions. Hence a greatvariety of ionic species can be synthesised in the plasma before recombination reactionscan c-, "jr to produce new neutral species.

Why do ion-neutral reactions and recombination reactions proceed so much morerapidly than neutral-neutral reactions? Why do ion-neutral reactions proceed morerapidly as the temperature of the plasma is reduced contrary to the behaviour of mostneutral-neutral reactions? The simple answer lies in the nature of the forces that comeinto play during the collisional process. The force between two neutral molecules actsover a relatively short range. This limits the collision frequency and thus the so-calledcollisional rate coefficient, which is an indicator of the maximum efficiency of thereact'on. Moreover, most neutral-neutral reactions are inhibited by activation-energybarriers, which can be thought of as potential barriers which the reactants mustovercome to achieve new bonding arrangements. The presence of such barriers impliesthat neutral-neutral reactions should proceed more rapidly as the temperature of thereactants is increased and this is often observed to be the case. So, neutral-neutralreactions will generally proceed only slowly in low-temperature plasmas but there arewell known exceptions; notably some reactions involving neutral radical species(Howard and Smith 1983).

The forces between charged (ions) and neutral species are clearly different. When anion approaches a non-polar molecule the electric field associated with the ion distortsthe charge distribution within the molecule, thus inducing an electric dipole. Thisresults in an attract ve force between the ion and the molecule described by aninteraction potential. V(r proportional to r '. where r is the ion-molecule separation.This force is relatively long-range compared to that between neutral molecules andthus ensures that the corresponding capture cross-sections exceed those for neutral-neutral interactions by about an order of magnitude at room temperature. Anexpression for the collisional rate coefficient. k,, can be readily derived, which shows itto be dependent only on the polarizability of the molecule, x, and the reduced mass ofthe reactants, A (k,=2xq/u)"., where q is the ionic charge, and is typically10 -cm -'l and. significantly. is indcpendcnt of temperature T (although the capturecross-section. which is inversely proportional to the relative velocity of the reactants.varies as T- ",), This simple expression for k, has been vindicated by numerousexperimental measurements of the rate coefficients, k, for ion-molecule reactions (forthe experimental methods see the next section). If the reactant molecule possesses apermanent dipole moment then Kjr) includes an additional term. This results in anincrease in k, by a factor dependent on the dipole moment of the molecule., u, and,significantly, renders k, dependent on temperature:

k, = 2xqiu " 2[z "' + Cao(2, xrk r) ,

where C is a constant (Su and Bowers 1979). Thus for a typical polar molecule (e.g. H 20

or NH A) k, for the reaction with a given ion exceeds that for the reaction of the same ionwith a non-polar molecule of similar mass and polarizability by about a factor of two atroom ten ,crature. However, k. values for reactions between ions and polar moleculesexcced those for reactions between ions and non-polar molecules by a factor of abouttcn at the low temperaturts of interstellar clouds. It is interesting to note that if any

8

Ionic reactions in plasmas 561

intrinsic energy barrier to molecular rearrangement exists in the ion-moleculeinteraction, then these energy barriers can be overcome by the energy gained by thereactants in the attractive field. This energy cannot, however. be used to promoteendothermic reactions since as the ion-neutral products of the reaction separate thereactive system must 're-absorb' this energy.

When the two reacting species are both charged, as is the case for the dissociativerecombination reaction of a molecular positive ion with an electron and for the mutualneutralization reaction of a positive ion with a negative ion, the attractive force isCoulombic and long-range with Vr) proportional to r- ' (Bardsley and Biondi 1970,Moseley et al. 1975). It is therefore expected that the capture cross-sections, and hencethe rate coefficients for such reactions, will exceed those for ion-neutral reactions, andthis is observed experimentally. Indeed, dissociative recombination coefficients, .,, andmutual neutralization coefficients, 2,, exceed ion-molecule collisional rate coefficientsby typically two to three orders of magnitude at room temperature. Also, theorypredicts, and experiments confirm, that both i. and a, should increase with decreasingtemperature. This fact, and the fact that many ion-neutral reactions also proceed morerapidly at low temperatures, is important to the proper understanding of the reactionsoccurring in low-temperature plasmas such as some terrestrial atmospheric regionsand interstellar gas clouds, to which we give special attention below.

2. Ionic reaction proceses in gaseous plasmas--- brief overviewFor simplicity, we consider first the ionization of an unreactive mixture of two diatomicmolecular gases, A2 and B. Later, in sections 4-6, we consider real situations as casestudies. Even in such a relatively simple A2/B2 mixture many different reactive speciescan be generated by subjecting the gas mixture to an electrical discharge or irradiatingit with energetic particles or photons. In the primary processes, positive ions and freeelectrons (direct ionization, atomic and molecular radicals (dissociation) and negativeions (electron attachment) may be formed. Negative-ion formation requires thatelectronegative gases (such as O , halogens, chlorofluorocarbonsI be present. Amultitude of excited states of A, and B2 and of the newly created charged and neutral

Table 1. A list of the primary processes that occur when a mixture of two molecular gLes(A, and B2) is subjected to a source of ionization, such as a gas discharge or energeticphotons, and the subsequent reaction procses that can occur between the ions, electronsand neutral species which generate new ionic and neutral species in the gas.

No. Process Description Comments

Pr m r 'Y processes(i) A, -A - A Dissociation The energies o( the

B: - B + B electrons and photonswhich cause dissociation

(i0 A: -A,' +e Direct Ionization and ionization must exceedB: .B. +e the threshold energies for

these processes(liii , -A' - A -e Dissociative

B2 - B *e Ionization

The dissociative attachment and direct attachment processes given in (5 a) and (5 b)below may also be primary processes.

9

562 N. G. Adams and D. Smith

TaMe I. (Ciintnued)

Reactions following ionizationI (a). A + B, - Be + A Charge transfer Which process dominates is

(exchanp) determined by theenergetic of the reactions.

Ilb). -. S + B + A Dissociation charge The rate equation for A"transfer would, for example, be

d(A "I/d, = - k, (A * ](92],I (c. -. ABt + B Ion-atom interchange where [ ] denote

(or atom abstraction) concentrations. WhenB2]>>[A*]. as is common.

[A'] decays exponentially.2. A + B-AB +e Associative detachment Important in plasmas with

negative ions and large

concentrations of atomicradicals (such as H or 0).

3. At +B,+M-A 182 +M Ion-molecule In atmospheric air, Massociation or clustering would surely be either N,(M is the third body or 0. This type ofwhich removes energy reaction is very likely inreleased in forming the high-pressure Iow-A 1 B2 molecule) temperature media.

4. A+e-.-A+A Dissociative An efficient process; veryrecombination fast compared to radiative

recombination, namelyA- +c-A-k+v. Rateequation for [A*] isd[.A, 3/* =i - - .[A' ][c].When (A;] C e], thesolution is:(A.1 "' =(l:A ' i-a,..

This condition canbe realized in somelaboratory plasmas.

5(a). A+e-A +A Dissociative electron For the dissociativeattachment proces. the electron

afinity of A must exceedthe bond dissociationenergy of A,. The directprocess can occur if A, hasa finite electron affinity.The rate equation for5 aL for example, isdae):dr = - OA,]Ce].

5(bl. A, +e+M--A" +M Direct electron 'A hen (A.]>>(e]. theattachment Jolution is exponential.

6lal. A ', -- A + B, Mutual neutralization The ternary process 6(b)c:ould 'esilt in the

6ihl. A' - B_ - M- Ternjrv ionic formation of ABA- B. , +M recombination molecules. The rate

equations are similar tothat for the dissociativerecombination process (4).

10

Ionic reacon. in plasma.s 563

species may be present and this constitutes a very reactive mixture. Many scqucntialand parallel reactions can occur between the positive and negative ions and neutralswhich generate different, often polyatomic, ions. Also, recombination (neutralization)reactions between positive ions and electrons and between positive ions and negativeions can occur which act to return the ionized gas to the neutral state (generating otherneutral species in the gas). The various reaction processes are summarised in table i.Limitations are placed on the reactions by the energetics: reactions cannot occur inlow-temperature plasmas if they are significantly endothermic. Nevertheless, evenunder the extremely low temperatures of some interstellar gas clouds, many ionicreactions occur which greatly modify the composition of the clouds. Specific examplesof the generalized reaction processes listed in table I are presented in sections 4-6.

It is clear that a great deal of laboratory data is required before a quantitativeunderstanding of the overall reaction kinetics and ion chemistry of low-temperatureplasmas can be obtained. In response to this, several important experimentaltechniques have been developed in recent years including those described below.

3. Laboratory techniques for studying ionic reactions at thermal energiesThere are two basic methods for the study of ionic reactions: beam methods which yieldcross-sections from which rate coefficients must be obtained indirectly; and swarmmethods which yield rate coefficients directly.

Beam techniques involve the generation of well collimated beams of the reactivespecies-ions, electrons or neutrals-which cross at some well defined angle under verylow ambient pressure conditions. At the region of interaction the species may react, andfrom the reduction of intensity of one (or both) of the beams, the cross-section for thereaction is determined as a function of the centre-of-mass interaction energy E,.. Ionbeams of a given mass-to-charge ratio, m/q, are usually prepared using mass filters(magnetic sectors instruments are the most common), and the ionic products of thereactions are usually detected using mass spectrometers. Pulse-modulation techniquesare often used to determine more accurately the often very small fractional attenuationof the beams and to discriminate against reactions with background gas. By varying theinteraction angle and determining the distribution of the products of the reactions as afunction of the scattering angle (differential scattering methods), a great deal ofinformation about the dynamics and mechanisms of the reactions as well as thedifferential scattering cross-sections. a, can be obtained (McDaniel et al. 1970). Thedisadvantage of conventional beam methods for the provision of kinetic data ofrelevance to real low-temperature plasmas is that the experiments cannot be performedto sufficient accuracy at sufficiently low E,, (that is, at E,. equivalent to roomtemperature and below) to give rate ccefficients k of acceptable accuracy. (Note that kis related to a by the relation

k=f f(Ec.)O E.wjdE.,

where f(E,.) is the Maxwell-Boltzmann thermal energy distribution function. Thussince to determine formation and loss rates in real media we need k as a function of thetemperature. T, we have to measure a as a function of Ec. and then take a thermalaverage.) To alleviate this disadvantage, merged beam experiments have beendeveloped to study binary ion-electron and ion-ion neutralization reactions (see

11

564 N. G. Adam.s and 0. Sma,

table I) in which the reactive species are formed in beams which are merged to overlapcollinearly (Moseieye al. 1975. Broujllard and McGowan 1983). Thus, by adjustingthe laboratory velocities of the two beams to be similar, the E.,, can in principle be madevery small, that is, similar to those E,, appropriate to low temperatures. These mergedbeam methods have had some success, although uncertainty still surrounds the dataobtained at low E,., principally because of the extreme difficulty of ensuring that thebeams are precisely collinear and are not interacting at a small angle or diverging. Somedoubt also exists concerning the internal energy states of the reactant ions generated forsuch single collision experiment& a very important point when it is considered that thereactivity of positive ions with electrons and with negative ions can be very dependenton the internal energy of the ions. It is for these reasons that kinetic data obtained fromswarm experiments are strongly favoured for modelling the ion chemistry of theterrestrial atmosphere (see section 4) and interstellar clouds (section 51.

Swarm methods involve an ensemble of reactive particles, ideally possessing .a

Maxwellian distribution of velocities which interact with another ensemble of particleswhich ideally also possess a Maxwellian velocity distribution at the same temperature.The experiment then reduces to the determination of the rate of loss of one or both ofthe reactants. and then a rate coefficient for the reaction can be determined directly at afixed temperature (or as a function of temperature in many experiments). Using thesemethods, kinetic data have been obtained at temperatures as high as 10' K and below10K (see Adams and Smith (1983) for example).

Perhaps the most successful techniques for the study of ion-neutral reactions in thethermal energy regime are the ion cyclotron resonance (ICR) technique and the fast-flow tube techniques. In the ICR technique (Mclver 1978), a swarm of ions of a givenm/q is trapped in a cell under low-pressure conditions ( < 10 ' torr) by a combination ofmutually perpendicular static electric and magnetic fields. The trapped ions aredetected and their mlq established by matching the frequency of a probing radiofrequency (rf.) field impressed on to the cell to the cyclotron frequency of the ions.Reactive gases are introduced into the cell at (known) low pressures and the change inthe r.f. signal strength (resonance absorption) due to the trapped reactant ions ismonitored as a function of time at a fixed pressure of the reactant gas. Hence the ratecoeficient for the ion-neutral reaction is determined directly at a particular interactionenergy which is usually equated to the temperature o(th cell walls. Ionic products'ofthe reaction are identified by sweeping the rl. probing field and recognising resonanceabsorption at the different frequencies appropriate to the mlq of the product ions.Valuable data have been obtained for ion-neutral reactions using ICR methods, whichhave been used both to determine the mechanisms of reactions and to establish relativethermochemical quantities such as proton affinities. They have also been valuable inthe elucidation of interstellar ion chemistry. The major criticism of such data lies in theuncertainty in the energies of the interacting ions (which it has been suggested may be inexcess of that appropriate to the cell temperature by about 0) 1-2eV). The internalenergy of some ionic species in ICR cells is also uncertain since these ions are produceddirectly in the low-pressure (CR cell and therefore may nor be thermalized to the cellwall temperature.

The most reliable kinetic data for modelling low-temperature ionic reactions areobtained by using higher-pressure, collision-dominated media included in which arethe so-called afterglow techniques, in particular the flowing afterglow technique, andthe similar (but subtly different) selected-ion flow-tube technumc Isee Adams and Smith(1987, 1988) for example). The first of the afterglow plasma tchniques to be developed

(2_

Ionic reutlions in ph snia~is 565

was the stationary afterglow in which ionization is created in a pure gas or a gasmixture by a short ionizing pulse (both r.f. and dc. voltage pulses have been used). Thereactions of the ions are then studied in the afterglow plasma following the cessation ofthe pulse when the ion (and electron) energies have relaxed to those appropriate to thegas temperature. Ion identification and the temporal decay of the ion density areusually monitored by observing the ion currents diffusing to the sampling orifice of amass spectrometer located in the walls of the 'essel containing the afterglow plasma(McDaniel and Mason 1973). Such experiments are usually performed at a relativelyhigh pressure ofan inert (unreactive) gas such as helium (to inhibit diffusive loss of ionsand electrons from the plasma) containing small admixtures of reactive gases. The timeconstant for the loss of a given reactive ion is measured at several pressures of thereactant neutral gas and the truly thermal rate cocfficicnt for the ion-neutral reaction isthen readily obtained. A few ion-neutral reactions of importance in the terrestrialionosphere were studied using this technique more than twenty years ago, some over asignificant temperature range (200-6 0 K). The flowing afterglow and selected-ionflow tube techniques have been used much more widely for the study of ion neutralreactions (see below).

The stationary afterglow has been used to greater effect to determine positive ion-electron dissociative recombination coefficients, 2, (process (4) in table 1). For thesemeasurements, the afterglow plasma is created in a small (dimensios - I cm)microwave resonator and the decrease in the electron number density with timeresulting from the recombination process is obtained by monitoring the temporalchange in the resonant frequency of the cavity. If only a single positive ion species ispresent in the afterglow (this objective is not always easy to achieve) as determined by awall-sampling mass spectrometer and if the pressure in the cavity is sufficiently high toinhibit diffusive loss of ionization then, since the ion and electron densities are equal.the 2. value is readily determined (Bardsley and Biondi 1970. Prior to the inception ofthe flowing afterglow/Langmuir probe technique (see below, the majority of theavailable thermal energy z, data were obtained using the stationary afterglowtechnique and this established the order of magnitude of a, for dissociativerecombination reactions.

With the advent of fast-flow tube techniques. initially the flowing afterglow in 1964followed by the selected-ion flow tube about a decade later, the study of ion-neutralreactions was transformed, because the types of reactions that could be studied and therate of provision of reliable kinetic data increased dramatically. The enormousadvantage of the flowing afterglow over the stationary afterglow technique is directlyattributable to the transformation of the time domain of the latter to the steady-statedistance domain (via gas flow) of the former (Ferguson et al. 1969). In the flowingafterglow, ionization is created in the upstream region of a fast-flowing inert carrier gas(usually pure helium at a pressure of about I torr) by some form of electrical dischargeor ion source. Plasma ions and electrons are convected away from the ionization sourceand thus an afterglow plasma is created in the downstream region of the low tube(typical dimensions of which are 110 cm long by 8 cm diameter) in which the ion andelectron temperatures are equal to the carrier-gas temperature. The ions arrive at asmall orifice where they are sampled, analysed and detected by a differentially-pumpedquadrupole mass spectrometer detection system. In a pure helium carrier gas the onlyions detected are He' and He:. Reactaot gases can then be added to the thermalizedafterglow and the masb spectrometer detector registers a reduction in the He' and He,ion signals and the appearance of product ions.

13

566 N. G. Adams and D. Smith

Although the flowing afterglow is an extremely versatile apparatus in that a widerange of ionic species can be generated in the plasma by sequential gas additions alongthe flow tube, often more than one ion species is generated following the addition of agiven gas (for example, when CH. is added to a helium afterglow then CH *, CH2,CH3 and CH4 are all produced and these ions also react rapidly with CH) and thiscan greatly complicate the identification of the ion products of the reactions and theelucidation of reaction mechanisms. This problem has been overcome by thedevelopment of the selected-ion flow tube (SIFT) technique (Smith and Adams 1979,1987) in which ions are generated in an ion source external to the flow tube (see figure 1)and after mass selection (according to m/q) by a quadrupole mass filter, a single ion typeis injected through an orifice into the relatively high-pressure carrier gas in which theions thermalize in collisions with carrier gas atoms as they are convected along the flowtube. By introducing reactant gases into the ion swarm (which is not a plasma!) reactionrate coefficients can be determined and product ions identified in the same way as inflowing afterglow experiments. Note that in the SIFT, ion product distributions canalso be determined. The secret of the operation of the SIFT is in the introduction of thecarrier gas via a venturi inlet which surrounds the ion-injection orifice (see figure 1).This inhibits back-diffusion of carrier gas through the ion-injection aperture, whichallows a relatively large aperture to be used without raising the pressure in the injectionmass filter to prohibitively high values. This simple device has enormously increasedthe versatility of fast-flow tube techniques. It is now possible to study reactions ofvirtually any ionic species that can be prepared in an ion source with a wide range ofreactant gases and vapours. Hence the rate coefficients for several thousands ofreactions have been measured using SIFT apparatuses in several laboratories aroundthe world (Albritton 1978). a significant fraction of which have been obtained over thetemperature range 80-600K (currently some twenty SIFT apparatuses are beingexploited). A notable success of such experiments is the contribution they have made tothe understanding of the reactions occumng in interstellar gas clouds (see section 5)

Another valuable experimental technique which has been developed recently tostudy ion-neutral reactions is the CRESU technique (*Cinetique de Reaction enEcoulement Supersonique Uniforme', that is Reaction Kinetics in Uniform SupersonicFlow') by which reactions can I- studied at temperatures down to 8 K. This is achievedby producing ionization upstream in the very low-temperature isentropic core formedby the expansion of gas via a de Laval nozzle (Rowe et al. 1984). Rate coefficients arethen determined in a similar manner to the fast-flow tube techniques. A similarexperiment in which ions are created in a free jet expansion is providing kinetic data onion-neutral reactions at temperatures below I K (Mazely and Smith 1988)! Both thesevery low-temperature experiments offer much for the understanding of the reactionswhich are thought to occur in cold interstellar clouds.

In contrast to the above, the development of the selected-ion flow-drift tube hasallowed the ion energy to be elevated above that appropriate to the carer-gastemperature by impressing an electric field along the axis of a selected-ton flow tube (seefig. I). This allows reactions to be studied at suprathermal energies approaching a fewelectron volts in the centre-of-mass frame and bridges the energy gap between the trulythermal data obtained from flowing afterglow and SIFT experiments and the dataobtained using the beam techniques referred to earlier. The data obtained fromselected-ion flow-drift tube experiments is also contnbuting to the understanding ofion-neutral reactions occurring in shocked regions of interstellar gas.

i 4

Ionic reat tions in plasmas 567

to~U C 0 - C

40 C,,

~C.

US z &~ --

j C

C.. C -a K ~

E q- E v-~

~~2 8

I - I s

; 2 2E - -~:CY~C 0I7

-8 w0 -.I i V

C w

o--

-or

2 C

C u E

a -

-C 'Dt

15

56( N. G. Adums and D. Smith

Fast-flow tube techniques have recently been applied to the study of plasmaprocesses other than ion -neutral reactions, notably dissociative recombination,electron attachment and ion-ion mutual neutralization (see table I). To achieve this.the flowing afterglow/Langmuir probe apparatus was conceived ldescribed inBrouillard and McGowan (1983)). Essentially the flowing afterglow/Langmuir probeapparatus is a conventional flowing afterglow to which has been added the Langmuirprobe diagnostic technique with which the plasma electron and ion number densitiescan be measured at all positions along the axis of the flow tube. Knowledge of ionneutral reactions is required in order to create plasmas comprising ions of the desiredtype. Under conditions of high pressure and high ionization density the ions are lost bydissociative recombination, whence the measurement of the gradient of the electron orion density along the flow tube provides a measurement of the recombinationcoefficient. 2. for the reaction. Similarly, the addition of electron attaching gases to theafterglow initiates attachment reactions which result in a loss of electrons and this isreadily recorded by the Langmuir probe. The attachment coefficients Plare also readilyderived (at low ionization density to inhibit recombination) as are rate coefficients formutual neutralization reactions. These flowing afterglow/Langmuir probe studies haveprovided much of the available data on mutual neutralization reactions of atmosphericimportance (see section 4) and on dissociative recombination reactions of importancein interstellar clouds (see section 5).

Some o(the important advances made in the understanding of the fundamentals ofionic reaction processes are described in some recent review papers (Bowers 1979.Adams and Smith 1983). In the following sections, the basic elements of the ionicreactions occurring in the terrestrial atmosphere, in interstellar gas clouds andlaboratory etchant plasmas will be outlined as illustrations of the utility of the kineticdata relating to thermal-energy ionic reactions.

4. Ionic reactions in the terrestrial atmospereWere it not for the presence of ionizing radiations. the atmosphere of the Earth wouldbe chemically very stable since, at normal temperatures, the dominant gaseousconstituents are unreactive. In practice, the action of solar radiation, galactic cosmicrays and radioactive emanations from rocks ensures that the atmosphere is partiallyionized at all altitudes and thus that a wide variety of ionic reactions occur.

Owing to gravity, the density of the atmosphere decreases roughly exponentiallywith altitude. Thus, the incoming intense solar radiation first interacts with a relativelylow-density atmosphere of N, and 02 enriched in the lighter gases hydrogen andhelium owing to gravitational separation. The destructive short wavelength (extremeultraviolet and soft X-ray) components of the solar radiation efficiently dissociate andionize the molecules creating the upper ionosphere (the F-region) in which the primaryions are largely the single-charged species H *, He *, N ' and 0* (balanced by an equalnumber of electrons thus creating a quasi-neutral electron-ion plasma medium). Theshorter wavelength radiations do not penetrate to the lower altitudes and there the lessenergetic species 0'. O" and N.Z" are produced as the primary ions (in the so-calledE-region of the ionosphere). At lower altitudes still lin the so-called D-region. see figure2) ionization is mainly caused by the solar Lyman-a and Lyman-#l line radiations whichare able to penetrate into this region of the atmosphere through 'absorption windows'in the 02 spectrum and thus to selectively ionize NO molecules (I.) and metastable 020A.3,) molecules (L,) thus creating NO' and 0: ions in the D-region (Wayne 1985). Ataltitudes below - 50km. solar radiations do not ionize the atmosphere but do

10

Iri( reuions in plastmas 6

INO

N0100

so EleFtreg-on

70 ~ ~ N J-0. /N//

1 ~ ~ ~ 164 -region

6-(higher- 7order

/ clusters)

Chartcd particle number denity (cm -)Ftgure I. The ionized regions of the Earth's atmosphere. The F. E and D regions of the

ionospnere are designated iccording to the 'ledges' observed in the electron density.Typical altitudinal profiles o(the various positive ion number densities in the positive ton-electron plasma are also shown. The negative ion types in the positive ion-negative tonplasma of the lower 0-region are known but the detailed attitudinal profiles of density arenot well characterised and so only the approximate total negative ion number densuy. - _dushed line) is tndic~atcd The proliles of the electron density. N~.and the total positive Iondensity. N ... ire also included It is assumed that quasi- neutralitV existS throughout the.itmnusphec. that is N N . in the elewarn-fon plasmi and N -zN. in the ion-toolplasma (from Smith ind Adams 1980.

17

570 N. G. Adams and D. Smith

dissociate molecular oxygen producing oxygen atoms which react to form the ozone(0) layer in the stratosphere. Other molecules from natural sources (such as CH4 andNHj) and from man-made sources (such as chlorofluorocarbons, CFCs) are alsodissociated and the atomic (e.g. Cl) and molecular (e.g. CH 3) fragments initiate a rich,even damaging and dangerous neutral chemistry (the CFCs are strongly implicated inthe destruction of the ozone layer and in causing the 'ozone hole' over the antarctic;McElroy (197)). Ionization in the lower atmosphere (the lower stratosphere andtroposphere) is created by penetrating galactic cosmic rays and radioactive emanationsfrom rocks, These energetic radiations ionize the medium non-selectively and thus themajor ions produced at low altitudes are N* , N;, 0* and 0* (see figure 2 forapproximate ion number densities in the various atmospheric regions). Negative ionsare also efficiently formed in the lower atmospheric regions, by the attachmentprocesses given in table I.

What are the actual ion types observed in the various atmospheric regions? The firstin situ measurements of atmospheric ions were carried out almost thirty years ago in theE- and F-regions of the ionosphere using rocket-borne mass spectrometers. 0 , N *and N ' ions were detected as expected, but surprisingly NO* ions were seen to be amajor component (see fig. 2). NO* ions result from the reactions of 0 * ions with N2and N * ions with 02, for example

0' +N2-.NO +N. (!)

In this reaction, ions of lower recombination energy, RE (which for NO' is 9-25 eV) areformed from ions of higher RE (for O +, 136eV). This is an important general principlein ionic reactions and leads to so-called terminating ions (it is clear that 0 * ions cannotbe formed from NO ions at low interaction energies). Note also that the N * and Oions are formed in the ionosphere by photo-ionization of N and 0 atoms and by thereactions of He * with N2 molecules and O molecules. As summarised in figure 3, it canbe seen that the ion chemistry of these ionospheric regions involves the conversion ofatomic ions to molecular ions. This is important in controlling the degree of ionizationof the plasma because the ncutralization of an atomic positive ion by an electron is aslow (inefficient) process whereas, in generaL dissociative recombination of molecularions is very efficient. Thus the production of and recombination of N;, 0; and NOlimits the build-up of ionization in the upper ionosphere. The good agreement obtainedbetween model predictions and in situ observations has demonstrated that thereactions occurring in these regions are properly understood.

At low altitudes where the gas pressure is much larger, and especially in thetroposphere where a number of reactive trace impurities exist and where electronattachment proceeds by converting free electrons to negative ions, a greater variety ofionic reactions occurs. In the D-region, the primary positive ions are NO * and O* andthe primary negative ions are 0- and 02, and these undergo clustering reactions(see process (3), table I), exemplified by the important reaction

NO +N* +N2 -. NO* N2 +N, (2)

in which the weakly-bonded species NO .N, is formed. Such a species can undergo so-called switching reactions with minority neutral species such as H2O.

N N2 + H20-.NO'.H20 + N, 43)

forming the more strongly bonded cluster ion NO .H20. Further reactions of his ionwith water molecules occur (see Thomas (1974)). The primary negative ions 0- and 0,are converted to the more stable (less reactive) species NO; and C0- in reactions with

18

Ionic reactions in plasmas 57)

'NI0

H (H 0,

(,. I

00

0.\

Figure 3. A limited scheme describing the ion chemistry of the upper ionosphere. The ionicspecies are arranged from top to bottom of the diagram in order of reducingrecombnaton energy. The major primary ions produced by the action o(solar photons onthe ambient gas are H'. He'. N and 0 * in the upper part of the region and 0*, N and0. in the lower part of the region. The major terminating ions throughout the region areNO* and 01 (see textk and these ions and also N2 ions can be destroyed by dissociativerecombination forming neutral atoms as shown.

NO and CO, and then clustering reactions occur. In the D-region. neutralization ofcharged species results from both dissociative recombination of the cluster positive ionswith electrons and also mutual neutralization reactions between the ambient positiveions and negative ions.

In the lower stratosphere and troposphere, the ambient positive ions and negativeions are complex cluster ions of the type H "(H2O),X. and HSOT.Y, respectively,where X are pollutant molecules having basic properties, such as CH 3CN and NH 3,and Y are acidic molecules, including HzSO and NHO3, and m and i are smallintegers. The balance between production of ionization and its loss by neutralizationreactions results in a steady-state number density of positive and negative ions of about2 x 103 cm - )in air at sea level. It is interesting to note that an excess of negative ions(over positive ions) is considered by some to give individuals a feeling of well being!

S. Ionic resctious in interstellar clouds The formation of interstellar moleculesDiatomic molecules isuch as CH and CN) were first detected in diffuse interstellarclouds some fifty years ago by their characteristic absorption spectra in the visibleregion. More recently, other simple molecules (such as CO and OH) have been detectedin these clouds via their characteristic absorption spectra in the vacuum ultravioletregion by using spectrometers borne on satellites, a necessary procedure to avoidabsorption by the atmosphere. With the advent of radio.- and millimetre-waveastronomy, many complex molecular species have been detected in the denser andcolder clouds of gas and dust in the galaxy (of which the Orion Molecular Cloud is thebest known example) IRydbeck and Hjalmarson 1985). The gas in these clouds is very

19

572 N. G. Adams and D. Smith

largely H2. with CO being the next most abundant molecule with a relative numberdensity some 10' times less than that of H2. Many polyatomic species have beendetected (see table 2 and Smith (1987)). The most complex species detected to date is thelinear HCoCN. although there is no doubt that even more complex molecular speciesexist in dense interstellar clouds.

The question arises as to how these molecules are formed in these harsh regions ofspace in which the temperatures are as low as 10 K (in dense clouds) and gas numberdensities are typically only 103 -10"cm - '. Gas-phase reactions between neutralmolecules cannot be important because, as mentioned in section I, such reactions aregenerally prohibitively slow at low temperatures. The condensation of neutral atomson to the very cold surfaces of the micrometre-sized dust grains which pervade denseclouds (the Horschead Nebula in Orion is the best known manifestation of a dust andgas cloud) can result in the production of molecules by heterogeneous catalysis, butthere is no satisfactory explanation of how these molecules could be released into thegas phase. In the early 1970s, it was proposed that the gas-phase reactions of positiveions with neutrals, many of which proceed more rapidly as the temperature isdecreased, result in the synthesis of polyatomic molecular ions which then undergodissociative recombination with electrons producing the observed neutral interstellarmolecules (Herbst and Klemperer 1973). In recent years, detailed models have beenconstructed of the gas-phase ion chemistry of interstellar clouds, which predict therelative abundances of many interstellar molecular species which are in reasonableagreement with astronomical observations. Thus it is now generally believed that thesynthesis of most interstellar molecAles occurs via ionic reactions in the gas phase.Detailed discussions are given by Dalgarno and Black (1976) and Smith and Adams1 1Q81). Here we briefly describe some of the basic reactions.

Diffuse clouds, as the name implies, are partially transparent to visible radiationsand ultraviolet radiations of longer wavelengths than the threshold wavelength forionization of H atoms (radiations with wavelengths shorter than this are stronglyabsorbed). Since the ionization energy of carbon atoms (I 1-26eV) is less than that of Hatoms (13.59 eV) then C * ions and electrons can be generated by photo-ionization of Catoms by stellar vacuum ultraviolet radiation. Galactic cosmic rays can also penetratediffuse clouds and, of course, are non-selective in the species they ionize. Since H, H 2and He are the major constituents of these clouds then H *, H and He * are generatedin the cosmic ray ionization process. Dense clouds of gas and dust are opaque tovacuum ultraviolet radiation and the only major source of ionization is penetratinggalactic cosmic rays, which in these clouds composed of principally H2 and He alsogenerate H *. H' and He * as the primary ions. Two very rapid reactions then occur inboth diffuse and dense clouds:

H2 +H, -H+ + H, (4)

He' +CO-,C +O+ He. (5)

Both reactions occur on every collision between the reactants as numerous laboratoryexperiments have shown. Thus both C * and H * ions are the important precursor ionsto most other ions formed in interstellar clouds.

As can be seen in table 2. several hydrocarbon molecular species have been detectedin interstellar clouds, but not the simplest hydrocarbons CH,. CH, and CH,. Thesespecies and also H., N2 and 0, must be present but, unfortunately, because of their

20

Ionic reactions in plas mas 573

Tabe 2. Molecules observed in interstellar clouds, and in circumstellar envelopes (indicaied byan asterisk). Molecules of a given atomicity are ordered from left to nght in increasingmolecular weight. Note that several singly-charged molecular-ion species have beendetected (CH *,SO". HCO". N2 H ".HCS-. HO .HCN" and HCOI. Bothlinear(I-Iand cyclic (c-1 isomers ofCH have been detected. The cyclic molecules c-CH, c-C H andc-SiC arc eamcntially iriangular in structurc.

Number of atoms Chemical species

2 H, CH. CH -. OH. C2, CN, CO. NO. HCL CS, SiO, PN, SN,. 50, S0, SiS.3 HO. CH, HCN, [INC. HCO. HCO', N2H . HNO, H2S, HCS', Siq,

CS. OCS, SO,4 NH3,. HO '. C2H. HCN, H2CO. I-C3H, c-CjH. HNCO. HCO, HCS,

CCN. CO. HNCS, CS.5 CH:, CH2NH, Sil:, c-C,14f. CHICO, NH2CN. HCOH. CH, HC2CN.6 CH CHJOti. CHCN. ClINC. NiCHO. CHJSII. CH7 CH3NH,, CH 3 CH. CH 3CHO. C2H:N, CH. HCs N .

8 CH3C2CN, CHJCO2 H.9 CHOH, CHOCH , , C2H,CN, CH3 C.H. HCCN.

to CHCOCH 3, CH 3 CCN.II HCCN.1213 HC,,CN.

symmetry (no permanent dipole moment) they do not radiate via rotational transitionsin their ground vibronic states. Such molecules have been detected by infraredabsorption spectroscopy in the warm, denser atmospheres of evolved stars (i.e. incircumstellar shels). Methane (CH,) and smaller one-carbon hydrocarbons can besynthesised by the following reaction sequence:

C H I H .H2 HH; -- CH- CH 3 1CH;---CH;. (6)

CH CH, CH 2 CH1 ,CH.

CH+ can also be formed directly from H 2 and C' via the process of radiativeassociation C+ +H2-.CH " + hv. Rzactions of the hydrocarbon ions in sequence (6)wit% neutral hydrocarbons can thcn form larger species. e.g.

CH, eCH- "CH--CH2 , C2 H3. CH 4,. (7)

Further reactions then lead to the observed polyatomic hydrocarbons (see table 2).The most abundant interstellar molecular species containing nitrogen and

hydrogen is ammonia, NH 3 . It can be formed by the following reaction sequencestarting from N* ions:

NH2. NHH H2 H eN+ NH*HN NHN -NH--NH* NH 3 . (8)

The NH 3 can then react with C* thus:

NH 3 , eC-- HCN , H,CN-.CN. HCN (9)

21

574 .V. G. Adams and D. Smith

producing both HCN * and HZCN which can then dissociatively recombine to formthe cyano compounds CN and HCN that are observed in interstellar clouds. Althoughthe recombination coefficients, , for many of the ions involved in these schemes havebecn determined (mostly using the FALP apparatus described in section 3), theidentities of the neutral products are not yet certain. Other CN-bearing molecules (suchas the cyanopolyynes) can be formed by reactions of HCN with hydrocarbon ions. Thereactions of hydrocarbon ions with NH, form the amino group of interstellarmolecules, including CH2NH and CH3NHI:

CH;N-H ,CH NH+. CH3NH. +- CH2NH. CH3NH 2. (10)

Interstellar molecules are of great utility in astrophysics. In particular, thecharacteristic radiation emissions from CO molecules are used extensively to detectinterstellar matter in the Milky Way and indeed in other galaxies. Such emissions arealso used as unique probes of the interiors of giant molecular clouds (such as the Orion,ind Sagittarius clouds) in which star formation is occurring as is indicated by the" clumping' of the gas due to gravitational collapse. Also the frequency resolution ofradio telescopes is such that they can easily distinguish between emissi-.is due to CO inall combinations of the I"C, I"C, 160, '"0 and "0 isotopic variants (Winnewisseret al. 1979). Via such observations, it has been observed that the "C: "C ratio ininterstellar CO is typically 1 :40 and thus exceeds the ratio in solar terrestrial materialwhich is 1 :89. This was initially thought by astrophysicists to be due to the greaterproduction of "C (by the process of nuclear burning in the interiors of stars) in otherregions of the galaxy compared-to production in the pir-solar nebula from which theSun and the planets were formed. However, it has been shown by SIFT experimentsthat ionic reactions of the type involving isotope fractionation are responsible forenriching the stable heavy isotopes of elements into interstellar molecules. Thus thereaction:

1C3 + +"CO _- -. - +f CO +0.003eV (11)

Iractionates' the I "C it, iO by virtue of the smaller vibrational zero-point-energy of" CO compared to "CO. Since this energy difference is small this phenomenon is onlyefficient at I v temperatures such as those pertaining to dense interstellar clouds.Isotope exchange reactions ,. p np .L ly efficient in fractionating deuterium, D, intomany interstellar molecules; indeed, the reaction:

D* + Hiz-H + + HD+O4eV (12)

ensures that most of the D in dense interstellar clouds is locked into HD molecules.Many similar ionic reactions result in the fractionation of deuterium into many otherinterstellar molecules (Smith and Adams 1984 b).

Also worthy of note in this section is another example of how interstellar molecules(together with a knowledge of ionic reactions) can provide the astrophysicist withvaluable information. Since ionization plays a part in inhibiting the rate of gas cloudcollapse to form stars, it is important to know the degree of ionization, 4e), in the densemolecular clouds where protostars form. This is an erudite topic but, in essence, theions and electrons in the clouds are coupled hydromagnetically to the ambientmagnetic field and the ions are coupled to the neutral gas by viscous forces Le. bycollisions. Thus. since the motions (in-fall) of the ions and electrons are inhibited by themagnetic field so are the neutral molecules and in this way gravitational collapse is

22

Ionic reactions in plasma.s 575

inhibited. An estimate of the degree of ionization x(ce = (e]/[H ,]) can be obtained byconsidering the rate of formation and rate of loss of HCO * ions which are observed inall dense molecular clouds. ric) turns out to be - 10- ' for most dense interstellarclouds.

6. Ionk reactions in laboratory discharge plasmasAs was mentioned at the start of section I, when an electrical discharge is passedthrough a gas mixture a wide variety of reactions can occur between the new speciesgenerated (namely electrons, ions and radicals) and these new species can also reactwith the ambient gas. These reactions may generate radiation at specific wavelengths(line radiations) and in broad bands (continuum radiations Radiation can also begenerated most efficiently by the direct excitation of the ambient gas by electrons whichhave been accelerated to sufficiently high energies in the discharge. Thus electricaldischarges have been used for decades as sources of line radiations (for example, low-pressure mercury and sodium lamps) and broad-band radiations (for example, high-pressure mercury and xenon lamps). More recent applications of electrical dischargesthrough gases are as flash lamps for the optical pumping of lasers, in the directgeneration of laser radiations (intracavity excitation) and in the etching ofsemiconductor surfaces (commonly silicon) for the preparation of microprocessors. Inthe early stages of the development of'these techniques a 'trial and error' approach wasusually taken to optimize the desired light intensity, etching rate, etc. Thus the totalpressure, partial pressures of the component gases in mixtures, discharge power, and soon were vaned. largely in ignorance of the processes which were occurring.

With advances in the understanding of atomic and molecular physics, of ionic andradical reactions in gases, and in the techniques to study these individual processes,more attention has been given to the processes occurmng in laboratory dischargedevices. This has, of course, been greatly stimulated by the commercial, industrial andsocial importance of devices such as microprocessors and lasers.

Plasma etching of'the surface of silicon wafers usualy involves exposing the surfaceto an electrncal discharge through a gas mixture containing a fluorine- or chlorine.bearing molecule. Halogens (F2 or C12) can be used but these (particulady F2 ) aredifficult and dangerous gases to deal with and so fluoro- or chloro-carbons are usuallypreferred. The important point is that the plasma ions sensitize the surface and then theneutral radical species generated in the discharge, particularly F and CI atoms but alsomolecular radicals such as CF, actually etch the surface (removing Si as vol;,tilehalides). Thus the production of reactive radicals in the discharge must be madeefficient (Smith and Adams 1984 a). Various mixtures are commonly used, for example,O,iCCI, and O,,CF,. Radicals can be produced directly in the discharge bydissociative electron collisions with the halogen-beanng molecules (for example.CCI, + e -CCII + Cl + el but efectron collisions also result in dissociative attachmentwhich generates a negative ion and a neutral radical, for example

CCI +e-.CI" 4CCI,, (13)

Positive ions formed in the discharge can undergo reactions with negative ions thus:

O" +Cl--0, +Cl 114)

23

576 N. G. Adums and D. Smith

again generating radicals. Dissociative recombination can also produce radicals, forcxamplc

Cc1; +C-CCI. +C1. (15)

Radical neutrals can also be destroyed in reactions with ions such as via the process ofassociative detachment (see table I): for example

CI" +CI-.CI, e 1161

and also in reactions with other neutral radical species (such neutral-neutral reactionsare undoubtedly important but are beyond the scope of this review). Reactions (13) to(16) are included with other reactions in figure 4 which indicates some of the manyreactions that can occur following a discharge through an 0 2/CCI , mixture

It is evident that a detailed knowledge of the basic reaction processes which canoccur in these discharge plasmas assists in the optimization of discharge conditions,namely the choice of appropriate gas mixtures, partial pressures of the components.current through the discharge, and so on, to obtain the desired etch rate. The alreadylarge and growing amount of data available on gas.phase reactions of ions, electrons

A E0-.o01-.- o-.-_.oo

8 F

0,. 0.-*- C 1 .~ l

Ca -. procm 1C1

O't ;h.0, C1. O. -1. CIZ.

Figure 4. Some reactions which can occur in an etchant plasma created by passing an cetcicaldischarge through a mixture of 0 and CCI. These reactions include positive-ion, negative-ion mutual neutralizaton (Box A +- Box 81. positive-fonwelectron dissociativerecombination 1ox B +Box C). negative ton reactions with 0, and CCI4 Box A - Box Eand positive-ton reactions with 0, and CCI, (Box 8 -Box F. The so-called terminatingspecies indicated include somc of the pnmar cpcic s created in thc discharge and someformed in subsequcnt reactions

24

Ionic reucnors in plasma s 577

and neutral species, obtained by techniques such as those described in section 3. is beingused to model the reactions occurring in etchant plasmas and gradually developingsemiconductor etching into something better than a black art'.

Throughout this brief review we have concentrated on the reactions of chargedspecies in gaseous plasmas but have occasionally alluded to the importance of thereactions of neutral radicals. Mention must also be made, in concluding, that excitedneutral species, in particular electronically excited neutrals which can possess manyelectron volts of energy, can also be involved in the complex reaction schemes whichcollectively describe the physical and chemical evolution o" gaseous plasmas. Excitedneutrals can collide with other neutrals transferring their excitation energy and in somecases causing dissociation and ionization. For example, a discharge through helium ora gas mixture conlaining helium undoubtedly results in the production of heliummetasable atoms Iboth He(2'S) and He(23S) are metastable and long-lived sinceradiative relaxation to the ground state is forbidden). The electronic excitation energyof He(2 3S) is 19-8 eV and, in collision with most atoms or molecules, Penning ionizationoccurs, for example

He(21S) + N 2 -. N * +c + Hc( 1So- (17)

This process can therefore represent a significant source of ionization in a gas. Also, if amixture of helium and neon is discharged then He(2S)(thc electronic excitation energyis 20-61 eV) can excitc the Nc atoms to a specific radiating state:

Hc(2 IS) + Nc IS)-Ne(3s2 ) + Hee'S) (18)

L. Ne(2p,) + 632-8 am radiation.

This causes a population inversion between two states of Ne atoms and it is thisphenomenon that has been exploited in the helium/neon laser. Other lasers also rely onthis phenomenon. as discus.ed in several reviews (see Nighan (1982), for example).

7. Concludiq remarksIn this review we have indicated how the creation of ionization in a gas mixture initiatesa variety of ionic reactions that can rapidly change the chemical composition of themedium. Laboratory experiments have been described which can be used to determinethe rate coefficients for individual ionic reaction processes. Following this, we havediscussed the ways in which data relating to many ionic reactions are being used tosuccessfully explain how the ions observed in the terrestrial atmosphere are formed,how the observed interstellar molecules are synthesised and how the reactive radicalspecies are produced in laboratory etchant plasmas. The study of ionic reactions ingases remains an exciting and interesting interdisciplinary research topic whichdemands a knowledge of atomic and molecular physic, spectroscopy, massspectrometry and gas-phase chemistry. It is an ideal discipline for training researchscientists both in the fundamentals of theory and experiment, and thus for providingversatile scientists to meet the requirements of modern industrial research anddcvclopmcnt.

25

378 Ionic ractions in plasmas

AcksowlgmeolsOur work in the many aspects of ionic physics referred zo in this paper has beengenerously funded over many years by the Science and Engineering Research Counciland the United States Air Force.

Refereu-esAiMms N. G.. And SiTt, D.. 1983. R'ations of Small Transient Species, edited by A. Fonuijn and

M. A. A. Clyne (London: Academic Pressl, pp. 311-385- 1987, Science Progress, 71, 91;1988, Ttchniques for the Study of Gas-Phase Ion-.olecule Reactions, edited byJ. M. Farrar and W. H. Saunders Jr (New York: Wiley) pp. 165-220.

ALBIiTTON. D. L, 1978, Atomic Data. nucl. Data Tables, 22I.DARDSLEY. J. N., and BioNDI, M. A., 1970, Adv. atomic molec. Phys., 6 I.Bowams, M. T. (editor), 1979, Gas Phase ton Chemistry, Vois I and 2 (New York: Academic

Press).BROUILLAito, F., and MCGOwAN. J. W. (editors), 1983, Physics of Ion-Ion and Electron-lon

Collisions (New York: Plenum Press).DALGARNO, A, and BLACK, J. H, 1976. Rep. Prog. Phys., I, 271.FEGGaoN, E E., FmsENu-L, F. C.. and SC4MELTEKOPF, A. L., 1969, Adv. atomic molec. PhYs.. S,

I.HERST, E., and KLI-rWERER. W. 1973. Astrophys. J.. 185. 505.HOWARD, M. J., and SMITH, I. W M., 1983. Prog. Reaction Kinetics, It, 55.MAZELY. T. L., and SMITh, M. A, 1988, Chem. Phys. Lett., 144, 563.McDANIEi- E. W., 1964, Collision Phenomena in Ionized Gases (New York: Wiley).MCDANIEL, E W., CERMAK, V, DALGARNO. A, FERGuSON, E. E., and FRIEDMAN, L, 1970,

Ion-Molecule Reactions (New York: Wiley), pp. 107-15.McDANIEL, EL W, and MASoN, E A.. 1973, The Mobility and Diffuion of ions in Oases (New

York: Wiley). p. 99.McELROY. M. B. 1987, Recent Studies of Atomic and Molecular Processes, edited by

A. El Kingston (New York: Plenum), pp. 63-78.MclvE, It. T. JiL 1978, Rev. scient. Instrum., 49, 11i.MosEi.n, J. T, OuoN, R. E_ and PrrtsoN, J. R, 1975, Case Studies in Atomic Physics, 5, .NiGmA, W. L (editor), 1982, Applied Atomic Collision Physics, Vol. I11, Gas Lasers (New York:

Academic Pressl.Row . B. R., Dt.PEmYRAT.G., MAIQE"Tr.J. B.. and GAucmEL. P_ 1984,J.chem. Phys.l , 4915.RyomecL 0. El H. and HJALmAvSON. A, 1985, Molecular Astrophysics edited by G. H. F.

Dwercksen, W. F Huebncr and P. W Langhoff (Dordrecht Reidel), pp. 45-175.SmIT. D., 1917, Phil. Trans. R. Soc., Lout. A, 323, 269.SmiTH, D, and ADAM. N. G.. 1979, Gas Phase Ion Chemistry, Vol. I.edited by M. T. owen New

York: Academic Press, pp. 1-44; 1980 Topics in Current Chemistry, 39, Plasmn ChemistryI, I; 1981, int. Rev. phys. Chem.. I, 271; 1984a. Pure app. Chem., S 175; 1984b, IonicProcesses in the Gas Phase, edited by M. A. Almoster Ferreira (Dordrecht: Reideilpp. 41-46; 1917, Ado. atomic molec. Phys. 24, I.

Su, T.. and Bowxas, M. T, 1979. Gas Phase Ion Chemistry, Vol. I, edited by M. T. Bowers iNewYork: Academic Pressl, pp. 83-118.

TOmio s. L., 1974, Radio Sci, 9, 121.WAYNE. R. P, 1985, Chemistry of Atmospheres (Oxford: Clarendon Press).WINNEwiss, G., CHURCHWELL E.. and WALMSLEY. C. M, 1979. Modem 4spec s ofMicrowa;e

Spectroscopy. edited by G. W Chantry (New York: Academic Press) pp. 313-501.

26

APPENDIX 2

FALP STUDIES OF ELECTRON ATTACHMENT REACTIONS OF

C 6 F5 C1, C6 F5 Br AND C6 F I

C.R. HERD, N.G. ADAMS AND D. SMITH

Int.J.Mass Spectrom. Ion Proc., 87, 331 (1989)

27

International Journal of Mass Spectrometry and Ion Processes. 87 (1919) 331- 342 331Elsevier Science Publishers 8.V.. Amsterdam - Printed in The Netherlands

FALP STUDIES OF ELECTRON ATTACHMENT REACTIONSOF C.F.Ct C.F.Br AND C6F9l

CHARLES R. HERD. NIGEL G. ADAMS and DAVID SMITH

Department of Space Research. University of Birmingham. Birmingham BJ3 277 fGi. Britain)

(First received 5 July 1988; in final form 12 August 1988)

A BSTRACT

Electron attachment reactions of thermal electrons with CFCl. CFBr and CFslmolecules have been studied at 300 and 450 K using the flowing afterglow/Langmuir probe(FALP) technique. The rate coefficients at both temperatures are large and increase withincreasing temperature. Different ionic products are obtained in the reactions; C, F,CI - onlyfor CFC1. predominantly CF, for CF1. and predominantly CFBr- at 300 K forCF, Br with the percentage of Sr- increasing with increasing temperature and constitutingca. 401 of the total products at 450 K. Comparison of these "high premsire FALP data withprevious -low pressure" data obtained from an ton beam time-of-flight experiment, hasenabled the autodetachment lifetime of the exited intermediate ion (CFI-)* to bebracketed between 10 - ' and 10 - 6 s. Also an upper limit to the C-Br bond dissociationenergy in CF FBr of ca. 84 kcal mol- has been deteruned from a consideration of theinetics of the C,F,6Br reacUon. In a parallel study, the reactions of CF 1 ions with severalmolecular gases have been studied using a SIFT apparatus and these results are reported inan Appendix.

INTRODUCTION

Electron attachment reactions have been studied using a variety oftechniques which fall into two distinct categories: (i) low pressure tech-niques in which electrons interact with molecules. M, forming an excitednegative ion (M - ) * which does not suffer collisions with ambient moleculesduring its lifetime, and (ii) high pressure techniques in which collisions ofN - ) * can occur which can modify the internal energy of (%M - ) *. The frstcategory includes ion beam (tame-of-flight. TOF) 11J, ICR 121 and photoiorui-zation cell 131 techniques. and the second category includes drift tube [41,Cavallen [51 and FALP [6.71 techniques. Attachment reactions are classifiedaccording to the observed products; when the negative ion products are theparent molecular negative tons (e.g., SF,- from SF and CF from CF 4 )then the process is termed non-dissociative attachment, but when the

0168-1176/89/$03 50 0 1989 Elsevier Science Publishers 8.V

28

332

product ions are fragments of the molecules (e.g., Cl- from CC 4. Br_ fromCH3Br) then the process is called dissociative attachment. In some electron/molecule reactions both dissociative and non-dissociative product channelsare evident (e.g., SFr and SF7 from SF, at temperatures ; 400 K, and Br-and C,6FBr- from C6FBr. see below).

The capture of an electron by a molecule may result in the rapid andspontaneous dissociation to the fragment negative ion and a neutral atom ormolecule (e.g., CCI + e - CI- + CC13). The rate coefficient (or cross sec-tion) for the process then depends on the efficiency of the electron capturestep which is usually dependent on the interaction temperature and energyof the reactants but not on the pressure of the ambient gas, since thedissociative attachment will usually occur on a timescale much shorter thanthe collision time, T,. of (M-) with an ambient atom or molecule (except atvery large ambient gas pressures). However, when (M - ) * has an appreciablelifetime against autodetachment, %, such that r. > r,. then superelasticcollisions with ambient molecules can reduce the internal energy of (M-)oand prevent autodetachment thus leading to the stable parent negative ion.M -. or modify the product ion distribution (i.e., as between parent ions orfragment ions, see below). For such reactions it is to be expected thatdifferences will be evident between the rate coefficients and product iondistributions determined using low pressure and high pressure methods andthis is a major point to be drawn by comparing the results of the collision-dominated FALP experiments reported in thus paper with those obtainedfrom earlier low pressure experiments. It is worthy of note here that when -ris long (but necessarily t -r,) then the excited ion (M-)* may be stabilisedby the emission of an infrared photon. Several such (M-)* have beenidentified in ICR experiments [21 and the radiative lifetimes, i-r, have beendetermined.

During the last few years, we have developed the flowing afterglow/Langmuir probe (FALP) technique to study electron attachment reactionsand to determine attachment rate coefficients, fl, and product ion ratios16.71. Thus the 0 values have been determined for a number of dissociativeand non-dissociative attachment reactions over a temperature range of about200-600 K in helium carrier gas at a pressure of akut 1 torr. In all thereactions studied, except that of hexafluorobenzene. C6 F6. the 0 value foreach reaction was either sensibly independent of temperature over theavailable range. or increased with temperature and then "activation energies"were determined for the reactions. The f value for the reaction of electronswith C6F (in which only the parent ion C.F6- was formed) very surprisinglywas observed to decrease dramatically with increasing temperature [8). Thisanomalous behaviour was subsequently verified using different techniques[9.101. However, no totally convincing explanation for this behaviour has yet

29

333

been given (11]. In pursuit of a better understanding of this unusualphenomenon. we chose to study electron attachment to the monosubstitutedperfluorobenzenes C. FsCI, C6F Br, and C6FsI using the FALP apparatus.

Naff et al. (11 studied these reactions some years ago using their elegantand productive TOF mass spectrometer method. For these C F5X molecules(X = Ci, Br, I), and for ill-defined electron energies < 0.1 eV in their ionsource, they observed that X - and C6 F - ions were formed in all three cases.The ratios of the maximum current levels of X- to CF,- were 100: 1, 3: 1and 1: 3 for X = Cl. Br and I respectively. The parent ions C.FsC- andCF 5Br- were also observed and the autodetachment lifetimes. -,, weredetermined to be 17.6 ps and 20 ps respectively for the excited ions(C6F5CI-)" and (CF 5 Br-) ° . C6F 5 - was not observed, indicating that ',for any (C. Fs I -) formed was < I ps which was the smallest r. measurablein the TOF experiment. Conditions in the FALP plasma are, of course, quitedifferent than in the TOF experiment. In the FALP plasma the ion collisiontime with carrier gas (helium) atoms, 7, is about 10- " s which is muchsmaller than the %o given above for the (C6 FsC-)" and (C6 FsBr-)" ions.Also. since the temperatures of the reactant electrons and molecules (andhence their interaction energies) are precisely defined in the FALP. thendifferences in the product ion distributions of X. C6 F - and C6 FX-between the FALP and TOF experiments are to be expected.

EXPERIMENTAL

The FALP technique has been described in detail previously [6.71 and itsapplication to the study of electron attachment has been thoroughly de-scribed in other papers [6.7,12). Briefly. a thermalised afterglow plasma iscreated in helium carrier gas at about I torr pressure flowing along a flowtube (which is about 100-cm long and 8 cm in diameter) with a massspectrometer sampling system located at the downstream end to identify thepositive ion and/or negative ion types which, with free electrons, comprisethe plasma. Controlled amounts of electron attaching gases are introducedinto the thermalised flowing plasma and the modification of the gradient inthe electron number density, n,. along the axis of the flow tube (:-direction)due to the attachment process is determined using a Langmuir probe thatcan be positioned anywhere along the axis of the flow tube. Attachmentcoefficients. 0. are derived from the analysis of the n, versus : dataobtained at fixed flow rates of the attaching gas into the afterglow [61. Theattachment reactions which are the subject of this paper are very rapid at thetwo temperatures for which measurements were made (300 K and 450 K. seeTable 1) and so only very small number densities (- 10 '0 cm -3 ) of theattaching gases were required to establish workable gradients of n. along

30

334

TABLE I

Rate coefficients and product ton distributions for electron attachment to C.FCI. CFBrand C6F, I

Reaction Temperature - 300 K Temperature - 450 K

0 (cm3 s" I) Ion products , (cm3 s - I) Ion products

C6 FCI+ e 8.4x 10- C6 F,CI- (100) 1.3x 10- ' CF 3CI- (100)CF 5Br+e 8.3x 10 C, FBr -(Z97) 2.4x 107 CFBr- (-60)

Br- ( :< 3) Br- ( - 40)CF,I+e 3.1 xO-' CF ( a 95) 1.2x10-' CFs (-100)

CFsl- ( s 5) 1- (trace)

Thus, dilute (about 1%), accurately known mixes of the C6 F5 X gases withhelium were made up to facilitate the accurate determination of the flowrates of C6 F5 X into the afterglow plasma. Since the number densities of theattaching molecules are so small. ion/molecule reactions cannot occur toany significant extent along the axis of the flow tube and so the product ionsof the attachment reactions, as determined by the downstream mass spec-trometer, are not confused. The values of P determined in this study areconsidered to be accurate to ± 25%. although relative values of $ for thedifferent reactions are more accurate.

RESULTS AND DISCUSSION

As was previously mentioned, earlier FALP studies showed that ,8 for theattachment of electrons to C6 F. decreased dramatically with increasingtemperature and that only the parent ion CFJ was observed as a product ofthe reaction. Naff et al. [l] have measured the r, for (C6 F-)* to be 12 its.Longer autodetachment lifetimes of > 30 ps [13) and ca. 50 jus [14 havebeen measured for (C.F6-)* formed by electron transfer to C6 F6 fromRydberg atoms. It is apparent that these r values are much greater than the-; values in the FALP plasma so all the (C.F -)* formed in the FALPplasma could be collisionally stabilised. Since there are no exothermicdissociative channels in the C6 F6 attachment reaction then the observationthat C.F - is the only product ion is understandable. This same argumentapplies to the C6 F C1 reaction and therefore only the parent negative ionwas observed to be formed in the FALP plasma. However, for both theC6 F, Br and C6 F5 I reactions, dissociative channels are observed in the FALPexperiments; notable is the attachment reaction of C6 FI in which the majorproduct ion was CJ F-. (In a parallel project, we have studied variousion/molecule reactions of C6 F - using a selected ion flow tube (SIFT)apparatus and the results are presented in the Appendix.)

31

335

As expected, some differences are apparent between the FALP data andthe TOF data of Naff et al. (1]. Consideration of the FALP and TOF dataprovides information concerning the thermochemistry of these attachmentreactions and the lifetimes of the excited (C FsX-)* ions. The reactions ofC4 F,I C6 FCI and C FBr are discussed separately below in the order stated.

Electron attachment to C6 FI

For this attachment reaction the major product was C6Fs- (> 95%) atboth 300 and 450 K. The parent ion C.F5[- was a minor product it 300 Kand only a trace of I - was observed at 450 K (see Table 1). A considerationof the thermochemistry for the two dissociative product channels gives

C6F51 + e - I- + CF5 + (4.3 ± 1.0) kcal mol - ' (la)

-. C6F + I - (3.0 ± 1.4) kcal mol' (1b)

The reaction ergicities were calculated using the accurately known electronaffinity of I atoms. EA(1) = (70.482 ± 0.023) kcal mol - [151, and the lesscertain values of EA(C 6F5 ) (63.2 ± 1.0) kcal mol - 1[16,171 and the bonddissociation energy D(CF-I) = (66.2 ± 1.0) kcal mol' 1181.

It is interest;r 6 that the major product ion is C.F - since this productchannel arently endothermic whereas the exothermic I - product ismnnir.i AZmila behaviour has been observed by Compton and Reinhardt117) for the collisional ionisation of C6 F6 by fast alkali atoms, i.e.. the moreendothermic production of C F,- appeared at a lower centre-of-mass energythan that for the less endothermic production of F-. The reason for this typeof behaviour is not yet understood; however, consideration of the availabil-ity of acceptor states for the free electron favours electron attachment to themulti-state polyatomic C4 F, radical rather than the singlet ground state -.

The derived endothermicity of channel (l b) is not inconsistent with thepresent data if the following argument can be accepted. The # values for thereaction at both temperatures are appreciable fractions of the upper limit(collisional) value, ft, which we estimate to be about 2-3 x 10-' cm 3 s-1(using the theory of Kots [191). The P value at 300 K is measured to be3.1 x 10-Icm' s - I (see Table 1), and if this value is less than the ft. valueby virtue of the reaction endothermicity or an activation energy barrier. .E.

then assuming a simple Boltzmann relationship (i.e., 0i =,8. exp -

1E/RT) %E is calculated to be about 1.2 kcal mol-. Also. if the factor-of-four larger value of the measured 0 at 450 K (see Table 1) is alsoattributed to the effect of AE. then combining the P values at the twotemperatures and again assuming the simple Boltzmann law, the % E value iscalculated to be about 2.4 kcal mol-'. It should be said that from these

32

336

measurements it is not possible to distinguish between an activation energybarrier or an endothermicity. However it is notable that If A E is assumed tobe due to an endothermicity then it is entirely consistent with AE calculatedabove for CtF - production (reaction lb). On this basis, the D(C 1 Fs-i) andthe EA(C6 F,) adopted in the calculation of AE are seemingly quite accurate(to within about 1 kcal mol-1 as indicated above).

As mentioned above, the parent ion C6 FsI - was discernible in the production spectrum at 300 K in the FALP experiment whereas in the TOFexperiments of Naff et al. [1], in which negative ions with lifetimes < 1 Pscould not be detected, parent ions were not detected. Hence the implicationis that (CFl-) * has r < I Ass but greater than the r, value in the FALPplasma which is about 0.1 Ais. This therefore brackets the T. of (C 8Fsi-)*between 10- s and 10-" s. The lower limit to r. may be somewhat greaterthan that quoted since helium, in general, inefficiently collisionally stabilizesexcited ions and so several collisions may be needed to stabilize the(C6FsI-)*. In view of this, it seems probable that r, for (C.F 51-)* at 300 Kis close to 10- s. Also. because C6 F51- is absent from the product ionspectrum at 450 K a significantly lower r. for (CFl-)* at this highertemperature is indicated.

Electron attachment to C, F Cl

The parent negative ion C6 FC- was the only significant product ionobserved in the FALP experiment at both temperatures, whereas bothC6 FC- and CI- were produced in the ion source of the TOF experiment(1]. CI- and CF - production are both quite endothermic

C6F 5C! + e -. CF s + Cl- -(8.8 ± 2.2) kcal mol-' (2a)

- C 6 F - + Cl - (29.0 ± 2.4) kcal mol- I (2b)

and thus the truly thermal FALP experiment is consistent with theseenergetics. The reaction ergicities were calculated using EA(CI) = (83.336 ±0.092) kcal mol - I [151, D(C 6Fs-C) -=(92.2 ± 2.2) kcal mol - [201 and theEA(C.F.) as given in sub-section 1. The long i, of 17.6 ps for the (C*FsC-)*[11 together with the large ft values measured in the FALP experiment(which are appreciable fractions of ft.. at both temperatures, see Table 1).implies that thermal electrons are efficiently captured by the C6 F5CI mole-cules and that the (C.F.CI-)' ions are effectively stabilised in coflisionswith He atoms. The increase in 0 with temperature (in this case about afactor of 2 between 300 K and 450 K) is not uncommon for non-dissociativeattachment reactions and could be a manifestation of a small activationenergy barrier to electron capture (equated to %E) which as before. is

33

337

estimated from the kinetics to be ca. 0.8 kcal mol -'. It is unlikely to be dueto a variation of %" with temperature (since -a %;) or a variation of thecollisional stabilisation efficiency of (C6 F5CI -)* by He atoms over such arelatively small temperature range. It is worthy of note that the sense of theP dependence on temperature is opposite to that for the P for non-dissocia-tive attachment to CF 6 (the C6F6 case has been discussed in detail inprevious papers [8-11 1). A final note worth making here is that the observa-tion of Cl- as a major product in the TOF experiment implies that a fractionof the electrons in the TOF ion source must be suprathermal by at least 0.3eV (rather than the upper limit energy of 0.1 eV indicated by Naff et al. [11)if the energetics given with Eq. 2a are correct. Alternatively, the literaturevalue for D(C 6F,-CI) may be too large by a few kcal moi

Electron attachment to C4F 5Br

The major product ions of this reaction were C6FBr- (> 97% at 300 Kand - 60% at 450 K) and Br- (< 3% at 300 K and - 40% at 450 K) withthe percentage of Br- and the overall attachment rate coefficient, /.increasing with increasing temperature (see Table 1). The energetics for Br-and CF - production in this reaction are uncertain because an accuratevalue of D(CF 5-Br) is not available. A value of D(C6 F5 -Br) = 105.9 kcalmol - ' has previously been reported 121) but this must be greatly in errorsince using this value, an endothermicity of 28.4 kcal mol-1 for Br-production is indicated and this cannot be remotely correct since Br-production is facile at 450 K in the FALP plasma. However an estimate ofD(C6 Fs-Br) a 80 kcal mot - ' can be made by considering the trend in theanalogous CHX compounds (where X = F. Cl. Br. and 1) [151 and thisvalue along with the accurately known EA(Br) of 77.530 + 0.092 kcal mo-'has been used to 'alculate the ergicities of the reactions

C,FBr + e - Br- + C6 F5 - - 3 kcal mol-' (est) (3a)

-" CsF + Br - - 17 kcal mo- (est) (3b)

Clearly the production of CF - is very endothermic and indeed is notobserved in significant quantities in the FALP.

The appearance of C6 FsBr- as the major product at 300 K in the FALPexperiment is consistent with the long lifetime of the (C,FBr-)" iondetermined by Naff et al. II], implying that this excited ion can be stabilizedagainst autodetachment in collision with He atoms (- ,. 7-,). Notwithstand-ing the fact that (CF 5Br-) ions can undergo collisions with He atomsduring their lifetime, which may modify the nascent state distribution of theexcited ions. we treat the product channels as separate reactions and, as

34

338

before, assume a Boltzmann law to derive a A E value for each reaction.Thus A E for Br- production is calculated to be ca. 6 kcal mol- '. As beforeIhis may represent a barrier to dissociative electron attachment or theendothermicity of the reaction and again these are indistinguishable. How-ever. large barriers ( Z 3 kcal mol - 1) to dissociative electron attachment arenot uncommon 16,71 (note that the barrier for CF,- production from C6 F1is not greater than 2 kcal mol-). However, if AE for Br- productionderived from the experiment represents an endothermicity of the reaction,then D(C6F-Br) = A E + EA(Br) =_ 84 kcal mol-1. In view of the uncertain-ties as to whether AE represents an endothermicity or an activation energybarrier, this value should be viewed as an upper limit to D(CF 5-Br). Thisupper limit is not inconsistent with the estimate of 80 kcal mol obtainedfrom the analogous CHBr bond dissociation energy.

It is also interesting to note that an energy barrier of ca. 1.0 kcal mol-,can be deduced for C6FBr- production which is similar to the barrier forCFCi- production mentioned previously. So it appears that the mecha-nisms for parent ion production for CjFCI and CoFBr are similar (C6FIforms very little parent ion at 300 K).

CONCLUDING REMARKS

These studies of electron attachment to CbF 5CI. CFBr and C6 F ,. haveclearly shown how different product ions can result from electron attach-ment studied in a low pressure (beam) experiment in which the electrons aresuprathermal in the ion source and in a "high" pressure experiment in whichelectrons are truly thermal. The production of (C6 F I-)* in the FALPexperiment but not in the TOF experiment has been used to bracket thepreviously unknown lifetime to autodetachment. 'r. of (C6F1-)" in therange 10- s < , (C6F5 -) < 10' s. It has also been shown that thermalelectron attachment to C6FBr yields both C6 F5Br- and Br-. from which anupper limit to D(CF 5-Br) of ca. 84 kcal mol - 1 has been established.

A most important conclusion to be drawn from these studies is that therate coefficients and product ions resulting from attachment reactions ofelectrons with a given molecular species can be dependent on the environ-ment in which the reaction occurs. Even where the electron/molecule (M)interaction energies or temperatures are the same in two systems but wherethe ambient pressures are very different (such that the -r value for collisionsbetween the nascent parent negative ion (M -) * and ambient gas atoms ormolecules vary greatly relative to the autodetachment lifetime of (M-)*)then different results are to be expected. Thus. for example. it is dangerousto assume that results obtained in very low pressure time-of-flight experi-ments and in low pressure experiments, including FALP results obtained at

35

339

pressures of typically I torr, are appropriate to electron capture detectorswhich generally operate at ambient pressures close to atmospheric pressure.Recently, we observed that the CO 3 radicals generated from the dissociativeattachment reactions ol CC 4 with thermal electrons in the FALP plasmarapidly undergo dissociative attachment with thermal electrons 1121. How-ever, we have been informed by Grimsrud [221 that he sees no evidence forsuch a phenomenon in his high pressure electron capture detectors. Thisdifference may be a manifestation of the production of vibrationally-excitedCC! 3 radicals in the CCI4 + e dissociative attachment reactions which arenot effectively quenched to the ground vibrational state before interactingwith an electron in the lower pressure FALP but are quenched at the higherpressure of the capture detector. This would imply a barrier to dissociativeattachment to CC' 3 radicals which is effectively overcome by vibrationalexcitation. More work is necessary to clarify this.

ACKNOWLEIXEMENT

We are grateful to the United States Air Force for financial support ofthis work.

REFERENCES

I W.T. Naff. RLN. Comptoo and C.D. Cooper. 1. Chem. Phys., 54 (1971) 212.2 IL. Woodin, MS. Faster and J.L. Beauchamp. J. Chem. Phys.. 72 (1980) 4223.3 A. Chutlias and SH. Alajajian, Phys. Rev., 131 (1985) 2835.4 LM. Chani, A.V. Phelps and M.A. Beond Phys. Rev. Lewt., 2 (1959) 344.S R-W. Conuptoa and G.N. Haddad. Aust. J. Phys., 36 (1963) 15.6 E Alg., N.G. Adams and D. Smith. J. Phys. B, 16 (1964) 1433.7 D. Smith, N.G. Adams and E_ AIse , J. Phys. , 17 (1914) 461.S N.G. Adama. D. Smith. E. Alge and J. Burdon. Chem. Phys. Let., 116 (1985) 460.9 S.M. Spyrou and L.G. Cbnstophorou. J. Chem. Phys., 82 (1915) 1048.

10 N. Hr dande-Gi. W.E. WnrworIb. T. Lmeru and EC.M. Chem J. Chromatopr., 312(1964) 31.

11 LG. Christaphoro. J. Chem. Phys., 83 (1985) 6543.12 N.G. Adams, D. Smith and C.R. Hcrd. lot. J. Mass. Spectroa. lo Processes. 84 (1988)

231.13 I. Dimxd and R. Boner. J. Chem. Ptiys.. 74 (19111) 2355.14 R.W. Maraww, C.W. Walter. KA. Smith and F B. Dunning J. Chem. Phys SS (1983)

2853.15 T.M. Miller, to R.C. Weast (Ed.). CRC Handbook of Chcrustry and Phyc 66th eda..

ClC Prsa. B= Raton, FL 1915. E-62.16 F.M. Page and G.C. Goode. to Negative Ions and the Magnetron. Wiley Intersci m

New Yorh, 1969.17 R.N. Compton and P.W. Reinhardt, Chem. Phys. Lett.. 91 (1982) 268.18 MJ. Kjreeh. SJ.W. Price and W.F. Yared, tnt. J. Chem. Kinet., 6 (1974) 257.

36

340

19 CE. Kiots, Chem. Phys. Lett., 38 (1976) 61.20 K.Y. Choo, O.M. Golden and S.W. Benso Int. J. Chem. Kinet., 7 (1975) 713.21 SJ.W. Price and H.i. Saptano. Can. J. Chem.. 52 (1974) 4109.22 E Grinmsrud pnvate communication, June 19U.

APPENDIX A

Ion / molecule reactions of C6 F,-

Little is known about the reactivity of _F - ions, so we chose toinvestigate this in order to gain a more complete understanding of thechemistry of this ionic species. Reactions of C6 F - with CH3 Ci, CHBr,CH3 I, HF, HCl. HBr and C6 FCl were studied at 300 K using a selected ionflow tube (SIFT) apparatus which has previously been described in detail[All. The carrier gas was pure helium at a pressure of about 0.5 torr.Copious amounts of C6 F - were produced by electron impact on C.Fl in ahigh pressure ion source and injected into the carrier gas. Reactant gas wasadded downstream and the binary rate coefficients, k2 , were determined in

the usual manner (All. The k, values and the ionic products observed arelisted in Table Al.

The k2 values for the reactions with CH3C! and HF (reactions (a) and (d)in Table A1) were immeasurably small, being < 10- 13 cm3 s - . This is notsurprising for reaction (d) since it is calculated to be endothermic by about 6kcal mol'. The reactions with CHBr and CHI (reactions (b) and (c))seemingly occur via a binary nucleophilic substitution mechanism (A2,A31(an Ss 2 displacement mechanism [A4D giving Br- and I- as the productions, respectively. These reactions have small reaction efficiencies (k2/k., 0 ,where kA, is the collisional limiting rate coefficient cakulated using theaverage dipole orientation theory (ADO) [AS; see Table AI). Many organicion/molecule reactions proceed with k2 well below their collisional limitingvalues, this being qualitatively ascribed to the effect of a double-wellpotential surface [A2,A6]. However, it is interesting to note that in reactions(a) to (c) the nucleophile, qFf, and the substrate, CH 3, are obviously thesame and thus the leaving group ability of the halogen ion should be animportant factor affecting the efficiency of the reactions. It is known thatthe leaving group abilities of the halogen ions in the gas phase are the sameas in the liquid phase, i.e., I- > Br- > C - and, as can be seen from TableAl. the efficiencies of reactions (a)-(c) do indeed increase with the leavinggroup abilities of the halogen ions. Note also that the reaction efficiencyincreases with the exothermicity of the reactions.

The reaction with HO and HBr (reactions (e) and (f) in Table AI) givingC!- and Br- as product ions are proton transfer reactions and proceed with

37

341

TABLE Al

Rate coefficients k, . at 300 K for the reactions of C1 F;- with the molecules indicated asmeasured using a selected ion flow tube (SIFT). Also listed are the collisional linuting valuesof the rate coefficients kAo '. the reaction efficiences (k /kAoco) and the reaction ergicities&H, (calculated from the bond dissociation energies and electron affinities given in ref. 15)

Reaction k, (ciu s-') kA.o (cn s-') k2/kAo o 4H,(kca mol")

(a)CF,- +CH3C: <101 3 1.40lXO' 7.0xl0 - -41no reaction

(b) CF - + CH 3Br 4.4x10- 2 1.14x10 - 3.9x10 - -49-. CF, CH, + Br-

(c)CFs" +CH 3 I 6.5x10 - t' 1.04x10- ' 6.3x10 2 -57- CFCH, + I -

(d) C5 F- +HF:noreaction <to- l.79x10- ' 56x10 - 's +6(e)C, F +HCI 6.0x10- '0 1.05x10- 9 5.7xlt - -41

- CFH + Cl -(f) CF;- +HBr 2.0x10 - 0 7.44x10 - 10 2.7x10 - ' -41

- CFH + Br -(g)CF" +CqFCI 1.9x 10-' 0 - - -45

- (C1F,), +C-

The polarisabilities and dipole moments of the neutral molecules which were required tocalculate kAco were obtained from ref. 15 and the dipole locking constants were determinedfrom data in refs. A9 and AIO.

k 2 values which are appreciable fractions of, although significantly smallerthan, their collisional limiting values. Proton transfer to negative ionsgenerally occurs at or near the collisional limiting value for localised anions(e.g., 0-, OH- and C2H-) [A7). However there are many examples ofrelatively slow proton transfer for delocalised anions, e.g., CH.O- and(C5 H5)CH2 A6]. C,---s - is a delocalised aion and thus it is not surprisingthat the k, values for reactions (e) and (n) are significantly smaller thantheir collisional limiting values.

The reaction with C F5CI (reaction (g)) proceeds with a k2 value of1.9 x 10-"' cml s-' which again is significantly smaller than the collisionalvalue and as before the halogen negative ion is the product. (Since thepolarizability of C6FC! is unknown, k,0 could only be estimated as5.2 X 10- ° cm' s - ' (using an estimated polarisability of 5 X 10- 24 cm3)giving a reaction efficiency of 0.3). This reaction probably proceeds via anaddition elimination mechanism [A81, with the strong electron withdrawingF atoms stabiising the resulting carbanion which then eliminates CI- toproduce (C6F,) 2. Dissociative charge transfer from C6 F - to C6FCI to giveCl- and two C6F, radicals is not possible since this process is ca. 55 kcal

38

342

mol-I endothermic. The relative inefficiency of this reaction may be attri-buted to a barrier on an, as yet, undetermined potential surface.

REFERENCES

Al (a) D. Snith and NG. Adams. in M.T Bowers (Ed.). Gas Phase Ion Chemistry. Vol. I.Academic Press, New York, 1979. p. 1. (b) D. Smith and N.G. Adams. Adv. At. Mol.Phys.. 24 (1988) 1.

A2 W.N. Olmstead and I.I. Brauman, J. Am. Chem. Soc., 99 (1977) 4219.A3 M. Tanaka. G.I. Mackay, J.D. Payzant and D K. Bohme. Can. J. Chem.. 54 (1976) 1643.A4 T.H. Lowry and K.S. Richardson. Mechanism and Theory in Organic Chemistry,

Harper and Row, New York. 1976.A5 T. Su and M.T. Bowers, in M.T. Bowers (Ed.), Gas Phase Ion Chenustry, Vol. 1,

Academic Press. New York. 1979, p. 83.A6 W.E. Famneth and J.1. Brauman, J. Am. Chem. Soc., 98 (1976) 789).A7 D.K. Bohme. in P. Ausloos (Ed.), Interactions Between Ions and Molecules. Plenum

Press, New York 1975, pp. 489-504.A8 I.M. Hams and C.S. Wamser. Fundamentals of Organic Reaction Mechanisms. Wiley.

New York. 1976.A9 T. Su and M.T. Bowers. hit. J. Mass Spectrom. Ion Phys.. 12 (1973) 347.

AIO T. Su and M.T. Bowers. Int. J. Mass Spectrom. Ion Phys.. 17 (1975) 211.

39

APPENDIX 3

A STUDY OF DISSSOCIATIVE ELECTRON ATTACHMENT

TO CF 3 SO3 CH3 AND CF3 S0 3 C2 H5 TRIFLATE ESTERS USING

THE FALP APPARATUS

C.R. HERD, D. SMITH AND N.G. ADAMS

Int.J.Mass Spectrom.Ion Proc., 91, 177 (1989)

40

International Journal of Mass Spectrometry and Ion Processes. 91 (1989) 177- 182 177Elsevier Science Publishers B.V., Amsterdam - Printed in The Netherlands

A STUDY OF DISSOCIATIVE ELECTRON ATTACHMENTTO CF3SO 3CH 3 AND CF3SO3C 2H s TRIFLATE ESTERSUSING THE FALP APPARATUS

CHARLES R. HERD, DAVID SMITH and NIGEL G. ADAMS

School of Physics and Space Research, University of Birmingham.Birmingham BI5 2TT G. Briain)

(Received 16 December 1988)

ABSTRACT

The rate coefficients. 8. for dissociative electron attachment to the inflate estersCF, SOCH, and CF3SOC2H,. have been detcrmned at 300. 385 and 475 K using the FALPapparatus. The f for both reactions are small ( - iO- " cnm3 s- ') at 300 K. in contrast to theA for trilic acid. CFJSOH. which is very large ( - 10-' cm' s- 1) at 300 K. In a parallelstudy, the reactions of F-, CI-. Br - and I- with the two esters have been studied at 300 Kusing the SIFT technique. These ion/molecule reactions proceed at or near their collisionalrates, again producing the tnflate anion CFS03-. and reveal that the dissociative electronattachment reactions of these esters are exothermuc.

INTRODUCTION

The extreme acidity in the liquid phase of the (Bronsted-type) superacidCF3SO 3H, commonly called triflic acid, has its equivalence in the veryefficient gas-phase dissociative attachment it undergoes when reacting withfree electrons in a plasma

CF3SO3H + e -. CFS03- + H (1)This may be viewed as the acid donating a proton to an electron. (the base)generating the triflate anion CFSO- [1]. This anion is extremely stable(unreactive) and this implies a large detachment energy or equivalently thatthe radical CF3SO 3 has a very large electron affinity. Indeed. from thestudies of the ion chemistry of tnflic acid. a gas phase electron affinity of 5.5eV for CF3SO3 has been estimated 121 which exceeds that of most otherspecies [31. The combination of a very large electron capture rate coefficientand the generation of an extraordinarily stable negative ion makes triflicacid a very attractive scavenger of electrons from ionized gases and plasmas.

41

178

However, the problem is that it is a very corrosive material. Attractivealternatives as electron scavengers may be the methyl and ethyl esters oftriflic acid, CF3SO3 CH 3 and CFSO3C2 Hs, which are quite stable. easier tohandle and less corrosive. The major question is whether they undergodissociative attachment with electrons at normal temperatures. We havetherefore studied these attachment reactions and the ion chemistry of thesetriflate esters to determine their suitability as efficient electron scavengers inthe gas phase.

EXPERIMENTAL

The electron attachment studies were carried out using the flowingafterglow/Langmuir probe (FALP) apparatus which has been described indetail previously [4]. Measured flow rates of the vapours of the esters wereintroduced into a thermaized flowing afterglow plasma and the increase inthe rate of reduction of the electron density along the axis of the flowingplasma was determined using a movable Langmuir probe. The analysis ofthe data to determine attachment coefficients. Pi, is straightforward and isweU documented (4-61. The product ions of the reactions were identifiedusing a downstream mass spectrometer sampling system. The #-values forboth esters were measured at three temperatures 300, 385 and 475 K.

In order to better understand the energetics of reactions (2) and (3) below,studies were also made of the gas phase reactions of the halogen atomicnegative ions F-. C!°, Br- and I- with the CF3SO 3CH 3 and CF3SO 3C2H5 at300 K. These studies were carried out using a selected ion flow tube (SIFT)apparatus which has also been described in detail previously (7] and which isnow a commonly-used technique for studying ion/molecule reactions atthermal energies [8].

RESULTS AND DISCUSSION

Since the value of 8 at 300 K for the triflic acid reaction (1) is very large(1 x 10-7 cm 2 s-') (1] being close to the predicted upper limit value forattachment reactions (ft. - 3 x 10 -'cm2 s- 1) (9] it was somewhat surpris-

ing to find that the $-values for the esters were some three orders-of-magni-tude smaller at the same temperature at - 1 -

'° cm 3 s -I (see Table 1).

However, although slow, these reactions did proceed via dissociative attach-ment

CF3SO3CH 3 + e - CFSO3- CH 3 (2)

CF3SO 3C2 Hs + e CF3SO + C, H5 (3)

42

179

TABLE I

Rate coefficients for electron attachment to CFJSO3CH 3 and CF 3SO3C 2 H,

Reaction Electron attachment coefficient X 10 I- 0 (cm, s - I)

300 K 385 K 475 K

CF3SO3CH3 + e 1.8 10 23CF3SO 3CzH s +c 2.7 6.7 6.5

generating the stable triflate anion in both cases. The ,8-value for reaction (2)increased dramatically with temperature, at 475 K being some 40 timesgreater than at 300 K. An additional value of Pi at 385 K allowed a 3-pointArrhenius plot to be constructed (Fig. 1) and the slope of the line indicatesan activation energy barrier or an endothermicity for reaction (2) of about170 meV (we show below that reactions (2) and (3) are exothermic).However, even if this rapid rise in ,P with temperature was maintained athigher temperatures, the ,#-value would not reach ,.. (i.e. become com-parable to triflic acid as an electron scavenger) before the ester thermallydecomposed. (The decomposition of these esters is discussed in ref. 10; itoccurs only at much higher temperatures and pressures than are used in thepresent experiments.) The values of fi for reaction (3) are also given in Fig. Iwhere it can be seen that Pi appears to be reaching a maximum at the highertemperatures. so no meaningful value for an activation energy or thereaction endothermicitv can be obtained.

T (K)S" 50 00 300 25010 21

E1(F, $03 CN 3 -- l"0v

Fig . I Plots of the attachment c:oefficients. ,.vs. reciprcal temperature. r"- .., rutts-t.yqeplotsl for the disccat,.'e attachment reactions of electrons with CFS03CH) (0) andCFSOIC, H, w',. The slope a( (he line for CFSOICH!3 (0) provides in activation energy forthe reaction of 170 meV as shown.

43

IIso

TABLE 2

Rate coefficients for the reactions of CFISOCH 3 and CF3SO 3C2H, with F. C1- Br. and

k' (cms) -/2

CF 3SO3CH, 3 F 2.1 x t0' 0.242Cl - 1.7 x I0' 0.185

+ Br - 9.6 x 10 -'0 0.136+1- 8.9x10 - 0 0.118

CFiSOC: H, + F- 3.2x lO-' 0.241+Cl- 2.1 x 10-' 0.184Br 1.5x 10- ' 0.135

+1- 7.8x 10- 0.116

The product ion is CF 3SO - in all cases and it can be seen that k2 decreases as DC- 7 2

(where )t is the reduced mass) as expected if the reactions proceed at the collision rate.

In order to determine whether reactions (2) and (3) were exothermic orendothermic we studied the reactions of the esters with the halide ions

X- + CFSO3R - CF 3SO3 + RX (4)

where X is F, Cl, Br or 1, and R is CH3 or C2 H5 . All of these reactionsproceed very efficiently, essentially on every collision. although the preciseefficiencies cannot be determined because the polanizabilities and the dipolemoments of the esters, which are needed for calculation of the collisionalrate coefficient, are not available, The measured rate coefficients, k2 , aregiven in Table 2 where it can be seen that the k z-values decrease roughly inaccordance with ;A-1/2 (where #L is the reduced mass of the reactants) aspredicted by theories for determining collisional rate coefficients [111. Thisvariation with reduced mass is itself an indicator that the reactions proceedat or near the collisional rate. That the reactions of the halide ions with bothesters occur so rapidly indicates that they are exothermic. If a thermody-namic cycle is constructed the enthalpy change, .H, for the reactions isgiven by

AH = D(CF3SO3-R) + EA(X) - D(R-X) - EA(CF3SO 3) (5)

(note the formation of the neutral product R-X is assumed and this must beso or the reactions would be very endothermuc). Because -% H < 0. an upperlimit to D(CF 3SO 3-R) can be determined and it is found that the smallestvalue for this upper limit is obtained for X = I. An upper limit of - 5 eVwas deduced using the following values: EA(CF 3SO3) = 5.5 eV [2, EA(I) "3.0591 eV. and D(CH 3-1) - 2.4 eV (D(CH,-l) is similar and is - 2.3 eV)131. Thus the dissociative electron attachment reactions (2) and (3) areexothermic by at least 0.5 eV, i.e.. EA(CF 3SO3) - D(CF3SO 3-R), and there-

44

181

fore the slope of the Arrhenius plot in Fig. I yields an activation energy andnot an endothermicity for reaction (2).

It is worthy of note that the reactions between the halide ions and thetriflate esters occur essentially on every collision generating the CF3SO3 ionwhereas significant energy barriers exist to the formation of this anion viadissociative electron attachment. Barriers to binary ion/ molecule reactionsare the exception rather than the rule and when they are exothermic theyusually (but not always) proceed rapidly [121. However, there are manyexamples of activation energy barriers to dissociative electron attachment14.51 and these are usually rationalized in terms of the need for vibrationalenergy in the reactant molecule and./or electron energy in order to access arepulsive potential curve along which the product negative ion and fragmentneutral can separate [13).

Note that these results again indicate that it is usually incorrect to makethe assumption that dissociative attachment will proceed efficiently even ifexothermic channels are available.

Finally, as we stated in the Introduction, the detachment energy ofCF 3SO is large and it is worthy of mention here that we have obtainedindependent information that this is indeed the case. Using a tunable lasersystem, we have been unable to photodetach electrons from CF3SO; withphotons of wavelength 226 nm = 5.3 eV (photodetachment from I - has beenachieved with 226 nm photons), suggesting (but not proving) that theEA(CF3SO 3) exceeds 5.3 eV, in accordance with the ion-chemical data [11.

ACKNOWLEDGEMENT

The authors are grateful to the United States Air Force for financialsupport of this work.

REFERENCES

I N.G. Adams. D. Smith. A.A. Viggiano. J.F. Paulson and MJ. Henchman. I Chem. Phys.,84 (1986) 6728.

2 JF. Paulson. A.A. Viggiano, M. Henchman and F. Dale. Proc. Symposium on Atomnicand Surface Physics, Obernraun. Austra. 9-15 Feb. 1986. p. 7.

3 T.M. Miller. tn R.C. Weast (Ed.), CRC Handbook of Chemistry and Physics. 66th edi..CRC Press. Boca Raton. FL, p. E-62.

4 D Smirth. N.G. Adams and E. Age. J. Phys- B. 17 (1984) 461.5 E. Alge. N.G. Adams and D Smith. 1. Phys. B. 17 (1984) 3827.6 N.G. Adams. D. Smith and C.L Herd, int. 1. Mass Spectrom. ton Processes. 84 (1988)

243.7 N.G. Adams and D. Smith. Int. J. Mass Spectrom. Ion Phys.. 21 (1916) 49.8 D Smith and N.G. Adams. Adv At Mol. Phys.. 24 (1987) 1.9 C E. Klots. Chem. Phys. Lett.. 38 (1976) 61.

45

182

10 T. Gramstad and R.N. Haszeldine. J. Chem. Soc.. (1957) 4069.11 T. Su and MT. Bowers. in M.T. Bowers (Ed.). Gas Phase Ion Chenustry. Vol. 1,

Academic Press, New York. 1979, p. 83.12 N.G. Adams and D. Smith. in A. Fontijn and M.A.A. Clyne (eds.). Reactions of Small

Transient Species. Academic Press, New York. 1983, p. 311.13 L.G. Christophorou. D.L. McCorkle and A.A. Christodoulide. in L.G. Christophorou

(Ed.). Electron-Molecule Interactions and their Applications, Vol. 1. Academic Press,1984, Chap. 6.

46

APPENDIX 4

STUDIES OF DISSOCIATIVE ELECTRON ATTACHMENT

TO SOME HALOETHANES USING THE FALP APPARATUS:

COMPARISON WITH DATA OBTAINED USING

NON-THERMAL TECHNIQUES

D. SMITH, C.R. HERD AND N.G. ADAMS

Int.J.Mass Spectrom.Ion Proc. (1989), in press

47

STUDIES OF DISSOCIATIVE ELECTRON ATTACHMENT TO

SOME HALOETHANES USING THE FALP APPARATUS:

COMPARISONS WITH DATA OBTAINED USING NON-THERMAL TECHNIQUES

D. Smith, C.R. Herd and N.G. Adams

School of Physics and Space Research

University of Birmingham

P.O. Box 363

Birmingham, B15 2TT

England

48

ABSTRACt

Measurements are reported of the rate coefficients, 0. for the

dissociative attachment reactions with electrons of the four haloethanes

CH3CCI 3 , CH2CICHCI 2 , CF3CCI 3 and CF2CICFCI 2 , obtained under truly thermal

equilibrium conditions at 298, 385 and 470K using the flowing

afterglow/Langmuir probe (FALP) technique. In all four reactions the

only product ion observed was CV'. These values of P are compared with

the equivalent $ obtained by the non-thermal swarm a[,d krypton

photoionization techniques for which 6 values were deduced as a function

of mean electron/molecule interaction energy, E, at a fixed reactant

molecule temperature of -298K. The different senses of the variations of

0 with temperature and with E are discussed and are rationalized in terms

of the thermal and non-thermal nature of the different experiments and

not as 'discrepancies' as described previously.

49

INThODUCTION

During the past few years, we have developed our flowing

afterglov/Langmuir probe (FALP) technique to determine electron

attachment rate coefficients, 0, under truly thermal conditions for a

number of molecular species over the approximate range of temperature

from 200 to 600K (i.e. mean interaction energies, -25 to -77 meV ( m

(3/2)kT) [1]). Several examples are seen of both direct attachment, in

which negative ions KX" of the parent molecules KX are formed, and

dissociative attachment, in which fragment ions X" and fragment radicals

Hf are formed (2,3]. An obvious feature of the collected data is that

when dissociative attachment is inefficient at room temperature, i.e.,

when the measured P is much less than Pmax, the upper-limit value for the

process [4,5,6]. then an approximately exponential increase of 0 with

increasing temperature, T, is observed and then 'activation energies',

Ea, can be derived for exothermic dissociative attachment (2,3]. The

increase in P with T is often very marked; for example, the A for

dissociative electron attachment to CH3Br (producing Br' and CH3)

increases by about a factor of 100 between 300K and 500K, which is

equivalent to an Ea of some 300 meV (3]. It is just as obvious from

published data obtained from swarm experiments (7] and by the krypton

photoionization technique [8] in the low energy (thermal) regime, that

when the gas temperature is held constant and the mean electron energy t

is increased then the 0 for the same dissociative attachment reaction

often decreases, sometimes quite dramatically. Thus, increasing the

internal temperature (energy) of the molecule tends to increases the 6

while increasing the electron/molecule interaction energy E (and thus.

equivalently, the translational temperature of the reactants) in general

tends to decrease the 0, the latter being In accordance with theoretical50

predictions (see ref.9 and references therein). In the truly therualisad

plasma medium of the FALP experiment, where both the internal energy and

the interaction energy are changed, the 0 is therefore a convolution of

these two opposing effects.

In this paper, we report the results of a FALY study of the P for

four haloethanes CH3CCl 3 , CH2CICHCI 2 , CF3CCl 3 and CF2CICFCI2 which have

previously been studied using the krypton photolonization technique

and/or the (electron) swarm techninque for which the opposite trends in

the true temperature and electron energy variations of the 8 are again

very evident.

EXPERIMCENTAL

The FALP apparatus and the method for determining 6 have been

discussed in detail previously [2,3]. Basically, a flowing thermalized

plasma is created in helium carrier gas (at a pressure of - 1 torr) by

using a microwave discharge located in the upstream region of a flow tube

which is -100 cm long and 8 cm in diameter. The carrier gas/plasma is

convected along the flow tube by means of a large Roots pump creating a

thermalised afterglow plasma in the downstream region. A mass

spectrometer sampling system is located downstream for identification of

the positive and negative ions in the afterglow plasma as well as to

identify the negative ions of the electron attachment reactions.

Reactant (attaching) gases are introduced into the flow tube in

controlled amounts and the resulting increase in the gradient of the

electron density, ne, along the flow tube (z direction) is determined

by a Languuir probe which can be positioned at any point on the flow tube

axis. The straightforward analysis [2,3] of the ne versus z data

provides a value for the attachment rate coefficient, . Because the51

for CF3CC1 3 was quite large (-2x10. 7 cu 3 s "I at 298K) a dilute (-I%),

accurately known mix of CF3CCI3 with helium was made to facilitate the

accurate measurement of the very small flow rates which were required of

this reactant gas into the thermalised afterglow plasma. Measurements of

0 were made at temperatures of 298, 385 and 470K.

RESULTS

The P values obtained in this study are given in Table 1 together

with those previously determined using the swarm method at a fixed buffer

gas temperature, T, of 298K and at a mean electron energy E(-39 meV)

equivalent to this temperature [7]. It can be seen that there is

reasonably good agreement between the P obtained from the FALP and from

the swarm experiments under comparable conditions (i.e. at T -298K,

equivalent to -39meV), particularly when sensible uncertainties in both

experiments are considered (the present $ values are considered to be

accurate to ± 20). The only product ion observed in all cases in the

FALP was CIJ.

It is interesting to note the trends in 8 between the different

isomeric forms of these ethanes as well as the effect on P of

substitution of F atoms for H atoms. It can be seen from Table 1 that

for a given temperature, in going from CX2 CICXCI 2 to CX3 CCI 3 (where X-H

or F) that there is an increase in 0 of more than one order of magnitude.

Also, in going from the H atom substituted ethanes to the F atom

substituted ethanes there is an increase in $ of slightly more than one

order of magnitude (i.e. in going from CH3 CCI 3 to CF3 CCI 3 or CH2 ClCHCI 2

to CF2CICFCl 2 ). The exact reasons for these two phenomena is not yet

unoerstocd fully, but these trends have been observed previously by

McCorkle et al, [7]. They suggest that the increase in 8 in going from

CX2 CICXCI 2 to CX3 CCl 3 results from a smaller bond dissociation energy of52

the C-Cl bond in CX3CCI 3 as compared to CX2ClCXC1 2 , which lovers the

asymptotic limit of the negative ion potential energy curve and in effect

increases the probability of separating to give the product negative ion

and radical.

As mentioned above, the only product ion observed in all cases was

Cl', but it has previously been reported by Alajajian and Chutjian [10]

that in the reaction of CF3CCl 3 with near thermal electrons another

product ion besides CV" was formed (they assumed parent ion formation

although this ion was not observed). However, for a 0 to 250 amu mass

scan of the quadrupole mass filter on the FALP, the parent ion CF3 CCI 3 "

was not observed, indicating that if it is formed it is not stabilized

against autodetactment in the FALP plasma.

The major point to which we wish to draw attention is the comparison

between the changes which occur in the f with increasing T, as determined

in the FALP experiment, and with increasing E, as determined in the swarm

experiment. The data for both experiments are presented in Fig. I

(filled symbols are FALP data; open symbols refer to swarm data). The

rapid increase in the 0 with increasing T for CH3 CCl 3 and CF2 CICFCI 2 is

not mirrored for a changing t in the swarm experiment, but rather only a

very small, if any, increase in 0 occurs for t within the thermal range.

For CF3CCl 3, the FALP data indicate that P does not increase with T; but

it is clear that the 0 at 298K is close to the upper- limit value for

this process, 6max, and so 0 cannot increase by much as T is increased.

Note, though, that the 0 decreases with increasing E for this reaction.

For the much slower reaction of CH2 CICHCl 2 the FALl data again indicates

a rapid increase of 0 with T but the 0 increases only slowly with

increasing E in the swarm experiment.

53

Before comenting further on the FALP data vis-a-vis the avars data,

it is pertinent to consider the data obtained for the CH3 CC1 3 reaction in

comparison to that obtained using the krypton photoionization method [10]

and these are presented in Fig. 2. Also included in this figure are the

data obtained previously from both the FALP and the krypton

photoionization experiments for the dissociation attachment reactions of

CH2 Br2 (3,10]. It is clear from these data that again the 0 for these

reactions increase with increasing T and decrease slowly with increasing

E in the thermal energy regime. Thus a pattern emerges for dissociative

attachment reactions in the thermal energy (temperature) regime as

studied by the truly thermal FALP experiment and the non-thermal swarm

and krypton photoionization methods. How can the obvious differences be

explained ?

DISCUSSION

The maximum electron capture cross section by a molecule, a. is

predicted theoretically to vary as E-1 (according to s-wave capture

theory (6]) and this has been substantiated by many experimental

measurements of the cross sections for rapid dissociative attachment (9].

Thus it is to be expected that the P for fast dissociative attachment

reactions will also reduce with increasing t (since A = a v- and V_ - jl/2

then 0 - E-1/2 ). Also, if a 0 measured at 298K in the truly thermal

FALP experiment or in the low energy limit of the swarm experiment is

found to be large and close to 0ax, then it is reasonable to assume that

no energy barrier exists to inhibit dissociative attachment. If.

however, a P is small (<<Omax) at 298K and the reaction is exothermic,

then, as previously mentioned, an energy barrier is assumed to be

reponsible and an 'activation energy', Ea, can be derived (2,31. The

origin of this energy barrier has been discussed by Wentworth (11 who54

suggests that it results from an unfavourable crossing of the bound M-X

potential curve with a repulsive curve of the M-X" system. Vibrational

excitation of the M-X bond, which can result from an increase in

temperature, accesses the repulsive curve allowing more facile separation

to K and X- to occur. Thus 0 will increase with increasing T and this

will be dramatic when Ea is large.

Using these ideas the differences between the T and E dependences of

can be qualitatively explained. When P is large and close to #max

(i.e. when there is no energy barrier and the repulsive curve passes

through the minimum of the M-X potential well) then increasing T cannot

produce an increasing P but an increasing t can result in a decreasing 8

(since the electron capture rate decreases with increasing E). However

when 0 is small (<<max, i.e. an energy barrier exists due to

unfavourable curve crossings) then 0 will invariably increase with T in

the FALP experiment but how will it vary with E ? Inspection of the data

in Figs. 1 and 2 indicates that when P increases rapidly with T (i.e.

when Ea is large) then 0 does not decrease rapidly with E and can even

increase somewhat as is the case for CH2ClCHCl2 .

We suggest that (i) this may be because in collision of the

attaching electron with the molecule in the swarm experiments, partial

internal (vibrational) excitation of the reactant molecule (or

equivalently of the transient molecular negative ion) results [12] which

tends to enhance # by more readily accessing the repulsive M-X" curve and

therefore mitigates the inhibiting effect an increasing E has on the

capture rate. Alternatively, (ii) an increase in a with E may also

result from an unfavourable curve crossing which then requires an

increase in electron energy for a favourable Franck- Condon transition to

occur from low lying vibrational states to the repulsive M-X" curve [9].55

Independent of whether explanation (1) or (ii) is appropriate, this

phenomenon will be more evident when Ea is large. When the Ea is

relatively small (and P only increases slowly with T) then the 0 would be

expected to decrease, albeit slowly, with increasing E . This reasoning

satisfactorily explains the different trends in the variation of p with T

and with E. So, rather than these differences being 'discrepancies' as

has been suggested by others [8], differences are actually to be

expected. It should be noted again that increasing the temperature of

the FALP plasma, of course, results in both an increased rovibrational

temperature of the reactant molecules and an increase in the

electron/molecule interaction energy. Thus the influence of T on P is a

convolution of these two opposing effects. Following these ideas,

inferences may be drawn as to the E dependence of P when the T dependence

is known and vice versa. Further investigations are continuing in our

laboratory into this phenomenon.

Finally, it should be noted that these conclusions do not relate to

direct attachment (referred to in the Introduction) but only to

dissociative attachment. The direct attachment process appears to be

much more complex and capture resonances appear to be involved [9]. The

0 for this process may even decrease rapidly from very high values close

to 82ax as is the case for SF6 " formation with increasing E [9] and for

C6F6 " formation with increasing T [131.

ACKNOWLEDGEMENT

We would like to thank the United States Air Force for financial

support of this work.

56

RKFEJICKS

1. P.W. Atkins, Physical Chemistry, Third Edition, W.H. Freeman and

Company, New York, 1986, p.8 .

2. D. Smith, N.G. Adams and E. Alge, J. Phys. B: At. Hol. Phys., 17

(1984) 461-472.

3. E. Alge, N.G. Adams and D. Smith, J. Phys. B: At. Mol. Phys., 17

(1984) 3827-3833.

4. C.E. Klots, Chem. Phys. Lett., 38 (1976) 61.

5. J.M. Warman and M.C. Sauer Jr., Intern. J. Radiac. Phys. Chem., 3

(1971) 273.

6. A.A. Christodoulides, L.G. Christophorou, R.Y. Pai and C.M. Tung,

J. Chem. Phys., 70 (1979) 1156.

7. D.M. McCorkle, I. Szamrej, and L.G. Christophorou, J. Chem. Phys.,

77 (1982) 5542.

8. S.H. Alajajian and A. Chutjian, J. Phys. B: At. Hol. Phys., 20,

(1987) 5567.

9. L.G. Christoporou, D.L. McCorkle and A.A. Christodoulides in

Electron-Molecule Interactions and their Applications. Vol.1., Academic

Press, 1984, Chapter 6, pp. 556-557.

10. S.H. Alajajian and A. Chutjian, J. Phys. B: At. Mol. Phys., 20

(1987) 5567.

1i. W.E Wentworth, R. George and H.Keith, J. Chem. Phys., 51 (1969)

1791.

12. H.S.W. Massey, Negative Ions, Third Edition, Cambridge University

Press, London, 1976, p. 169.

13. D. Smith, N.G. Adams, E. Alge and J. Burdon, Chem. Phys. Lett., 116

(1985) 460.

14. H. Shimamori and Y. Nakatani, Chem. Phys. Leet., 150 (1988) 109.57

TABLE 1. Rate coefficients, 0, determined on the FALP apparatus for

electron attachment to CH3CC1 3 , CH2ClCHC12 , CF3CCl 3 and CF2ClCFC12 at the

temperature indicated. The product ion in all cases was Cl. Also

given are the data obtained in a swarm experiment [7] at a gas

temperature of -298K and a mean electron energy of 39meV.

FALP SWARM

p, cm3s l , c23 s-l

298K 385K 470K

Reaction (39meV) (50meV) (63meV) 39meV

CH3CCI 3 + e l.5xlO "8 3.9xi0 "8 9.3x0 "8 1.5x10 8

CH2CICHC12 + e 3.1xlO "10 9.5x10 10 5.0xlO "9 1.8x10 10

CF3CC1 3 + e 2.4x10 7 2.4x10 7 --- 2.8x10 7

CF2ClCFC12 + e l.1x10 8 2.0x0 8 5.5x10 8 1.1x10- 8

58

Figure 1

The rate coefficients, , as a function of k obtained from the

present FALP studies (filled symbols) and from previous swarm studies (7]

(open symbols) for the dissociative attachment reactions of electrons

with CH3CCl3 (e,o), CH2ClCHCl2(A, 6), (73CC1 3(W,0) and CF2ClCFCI 2 (*,O).

FALP data were obtained under truly thermal equilibrium conditions at

three temperatures (298, 385, and 470K). The swarm data were obtained at

a fixed temperature of the attaching gas molecules of -298K and at

elevated mean electron energies, E as plotted. To facilitate comparison

of the FALP and the swarm data, the temperatures of the FALP measurements

are converted to mean energies according to E - (3/2) kT. The upper

solid line is the theoretical maximum value of P, i.e. #max as a

function of i obtained from s-wave capture theory [141.

Figure 2

The rate coefficients, 0, as a function of E obtained from FALP

studies (filled symbols) and those obtained using the krypton

photoionization technique (open symbols) for the dissociation attachment

reactions of CH3CCI 3 (, o) and CH2Br2 *,I with electrons. The CH2Br2

data were obtained from previous studies (FALP, ref (31, Kr

photoionization method, ref [10]). The FALP data were obtained under

truly thermal equilibrium conditions at three temperatures (298, 385 and

470K). The Kr photolonization data were obtained at a fixed temperature

of the attaching gas molecules of -298K and at elevated mean electron

energies. E as plotted. To facilitate comparison of the FALP and Kr

photoionization data the temperatures of the FALP measurements are

converted to mean energies according to E - (3/2) kT. The upper solid

line is the theoretical maximum value of , i.e. #max, as a function of

obtained from s-wave capture aeory [14].

59

rig. 1

10

.7110

6100

Fig. 2

-610

-7

E

10-

20 50 80 110 140 170 200E(meV)

61

APPENDIX 5

DEVELOPMENT OF THE FLOWING AFTERGL'4/LANGMUIR

PROBE TECHNIQUE FOR STUDYING THE NEUTRAL PRODUCTS OF

DISSOCIATIVE RECOMBINATION USING SPECTROSCOPIC TECHNIQUES:

OH PRODUCTION IN THE HCO 2 + e REACTION

N.G. ADAMS, C.R. HERD AND D. SMITH

J.Chem.Phys., 91, 963 (1989)

62

Development of the flowing afterglow/Langmulr probe technique forstudying the neutral products of dissociative recombination usingspectroscopic techniques: OH production In the HCO' +e reaction

Nigel G. Adams, Charles R. Herd, ar,d David SmithSchool of Physics and Spai Resrarch Univtniy of Birmmnhom. P 0. Box 363. BiminghomB15 27. England

(Received 2 March 1989; accepted 31 March 1989)

The flowing afterglow/Langnuir probe (FALP) technique has been extended to enable the

neutral products of electron-ion dissociative recombination in thermalized afterglows to be

identified by spectroscopic methods. Absolute number densities of H atoms in the afterglow

have been determined using vacuum ultraviolet (VUV) absorption at the L. wavelength. Byexploiting the reaction H + NO 2 - OH + NO, all of the H atoms can be incorporated into OH

molecules and thus observation of the intensity of laser induced fluorescence (LIF) If,obtained by exciting the (1.0) band of OH(A 17.-X Zl). allows a calibration to be obtained of

1. against the known number density of OH X *'I (v = 0) in the afterglow. Following this

procedure, a recombining HCO2 /electron afterglow was probed for production of ground

state OH X In (v' = 0) using LIF and it was established that OH(v" = 0) resulted from 17%

of the recombining ground state HCO,* ions. It was also established that a further 17% of therecombinations resulted in OH(v" >0), i.e., that, in total, (34 ± 6)% of the HCO' ions

recombine to produce OH X 2 1 radicals, either directly or via the electronically excited 4 '1

state. Details of the calibration procedure for H and OH number densities, of the ion chemistry

involved in the production of the HCO: afterglow plasmas and of the checks carried out to

establish that the fluorescence observed was from OH produced in the recombination reaction

are presented. Dunng these experiments, the rate coefficient at 300 K for the H + NO2

reaction was determined to be 1.3 x 10 -'o cm' s- ' from observations of the H-atom decay as afunction of NO2 number density in the afterglow (in good agreement with previous

determinations). Also the rate coefficient for the quenching reaction of OH( v > 0) with NOto produce OH(v = 0) was determined to be 6x 10" "cml s-'

The U.S. Government Is authorized to reproduce and sell this report.Permission for further reproduction by others must be obtained fromthe Copyright owner.

L INTROOUCTION Green and Herbst" have used a statistical phase space model

Elecuon-ion dissociative reconbination is an impor- to predict the products of the dissociative recombination of

tan ionization loss process in natural plasmas such as inter- HICN *, HJO *, CH, and NH.- with electrons. Bates "

steflar gas clouds" and planetary ioosperes.' Over the has used a more intuitive approach to predict the recombina-

years, the rate coefficients a. for the dissociative recombina- tion products of many ions; be argues that the neutral prod-

boo of many molecular ion species have been measured ucts which involve the least rearrangement wdl be favored.

mainly at room temperature using stabonary' and Bowing The approach of Bates has recently been extended by includ-

afterglow' techniques and deduced from cross-section mea- ing quantum chemical calculations of the electron density

surements made using the merged beam technique.' Subse- distributions' and by considering the likelihood of favorable

quently several reviews have appeared in the literature curve crossings of the bound ionic potenual curve with re-

which describe the theoretical u experimental aspects Of pulsive curves to products.2 '

ths process Ard pve tables of a, values."' However, very However, for specific recombination reactions these

little information is available concermng the identity of the theoretical approaches predict very different products for

neutral products of dissociative recombination reactions and the same reaction. """ , Some of these theoretical findings

that available relates only to excited ions." Atomic products have been discussed in a recent review. 2 Because of the pau-

have been dentihed for the recombination of02, t2 NO',3 city and uncertainty of the available information on the

CO ,,,.,s and H 20',"1 but the ions were internally excited products of dissociative recombination and because of the

to unknown degrees; only in the case of N2' (v = 1) was the desperate need for such data to include in models of molecu-

ltte o( excitation known " lar synthesis in interstellar clouds.1 we have extended the

Some theoretical effort also has been directed to predict- flowing afterglow/1angmuir probe (FALP) technique'

ing the products of dissociative recombination reactions. (which we have previously used to determine dissociative

Detailed calculations have been made of the ion and neutral recombination coeficients" 1) to permit the quantitative de-

potential energy curves implicated in the dissociative recom- termination of the neutral products of the dissociative re-

bination of O2 ,s H,- ,9 and HCO' , ' while Herbst' and combination of some ground state ions. To this end, laser

j -,hem ;-iys 91 2). 5 .ur '989 w02 . 06os,89 40963-IIS02 '0 89Alnr aw i irs5 t of C S 96363

induced fluorescence (LIP) and vacuum ultraviolet (VUV) kim to prodece OH im a vadM of ma vbd d"fsspectroscopic diagnostic techniques have been applied to re- However, OHl 1%v 0).i expea lo be repf* qimBW&d 10combining plasmas created in the PAL? apparatus. This is the v- - 0 vibrationaul level by further resea Wkh N%,the first study in which the molecular products of dissocia- since this is facilitated by the deep potential wed is thetive recombination have been identified and it is the exten- (HNO 3) interediate complex."Ile rt coefficient k forsion of the FALP technique and its application to the study quenching of OH( y = 1) by No, at 298 K is largeof the recombination of HCO,' ions with electrons that are I(k(298) = 4.$X 10-11 cm' s'1).3

the subject of this paper. 11. EXPERIMENTALOnly three product channels are energetically possible The FALP technique. without spectroscopic diagnoa-

in the HCO2* recomnbination reaction viz. tics. has been described previously in some detai in the liter-

HCO2* +.-H + C02, &H = - 185 kcal mol, (I A) ature. 7.3 Therefore we discuss the basic FAL? apparatus

OH + O. H 10 kal ol lb) only briefly and concentrate more fully on the spectroscopi-OH+ C. & - ~) calmoP. (b) aspects of the present experiments. A diagram of the aPPar-

-0 + H + CO, &H -58 kcal moP-'.(c) tus is given in Fig. 1. Ionization is created in a microwave

The recombination coefficient a, for the overall process has discharge through flowing He carrier gas which con vectu the

recently been measured to be 3.4x i0-'cm's-'at 300 K.18 ion/electron plasma along a stainless steel flow tube (I cm

In the present study, the fractions ot the reactions leading to diameter; I mn length). The ion types in the thermalizad

OH in both the ground and excited vibrational states of the afterglow plasma so tanned are identified in the downstream

electronic ground state have been determined from absolute region by a mass spectrometer/channeltron detection sys-

intensity measurements using LIF. T'he fluorescence intensi- tern. At a He pressure of 1.6 Torr (typical of the pressures

ty trom laser excitation at rovibronic transitions in used in these studies), the He 'ions formed in the discharge

OH[JA 'Y(W = I) -X -n1(V = 0) 1 was related to the abso- are rapidly converted to He? viz.

lute OH X ' V 0) number density using the calibration He* + He + He-He,* + He (3)

reaction (nate that He2* can also be produced in the discharge by theH + NO, - OH + NO, &H -29. 5 kcal mo '. (2) Hornbeck-Malnar process which involves colisions be-

This reaction has been has been widely studied [ the rate tween highly excited He* atoms and He. i-e-,

coefficent at 300 K.kW00) = 1.3 x I0' 0 cm's '12 and is HeO + He - Hez* + e). Helium mnetastable atoms, which

YUV Lighit Ndl : AG Laser]

I oublflq/Nzaifl Dye Laser

De t t tcto OtivIt,FIGI A sceti Diaria h oigat wLnmrpveFL)aorw.n~wn~dn dtdcdfwrcm LF pU~(dcomLns DrifceIS gO anCNadV inir Jket DV top~i shrge (o e bona i6M 5.N

d0Afftra oa moensi h I w~a Vpd bu~qat rapu.adtewnosi h U ~ a rm a riu I t ond.Aosbw h sieLngu r.cadheowsucn a ptO etIdt su h OSp n

Str~aoeoe ,-nlt Tubeafr ~he hn rmrmprtDiffusin Root

PJJp PumpyS \Law Seas 2 5 ily'8

ar Mo produced io the discharge and which. if alowed to, tio mom (i.e., port 14) from a maximum d - I x 10'would resultin adistributed sourceof ionization in the after- cm"' at which recombination is the dominm. ionizaoglow due to Penning reactions, are completely removed from loss process, down to densities for which recombination lon

the afterglow by adding Ar (at a partial pressure of - 3 is insignificant, so that the plasma ion composition and themTorr) just downstream of the discharge (at inlet port 12; ion-molecule chemistry can be determined using the down-see Fig. 1) viz.; stream mass spectrometer.

He-(2 'S, 2 'S) + Ar -Ar* + He + e. (4) As previously mentioned, in these studies of the neutrL1 4 products of dissociative recombination, the OH product of

Aralso reactswith He,* [k(200) = 2.0x 10 -'cm' s-') the recombination of HCO,* reaction (I)1 hasbeenmoni-v Hz* tored using LIF. The LIF technique has been widely used to

He" + Ar-.Ar" + 2He. (5) monitor radical species and in prticular the OH radical has

Thus an Az *Ie afterglow plasma is created. The presence of been thoroughly studied.3"' However, to our knowledge,At in the flow tube was also observed to reduce the amount the LIF technique has not been used previously to determineof short wavelength radiation which was transmitted from absolute densities of OH radicals produced in ionic reactionsthe microwave discharge to the downstream spectroscopic and therefore this requires some discussion. The experimen-diagnostics (this is especially important for the absorption tal arrangement is shown in Fig. 2. A laser beam at a wave.measurements at the hydrogen L. wavelength, see Sec. length of- 281 nm is produced by frequency doubling theIII A). The addition of H2 (at inlet port 13) to the Ar+/e output of a pulsed dye laser (Spectron SL400B operating

plasma then rapidly converts it to an H," / e plasma via the with a Rhodamine 6G dye) which is pumped by the doubled

reactions output of a Nd:YAG laser (Spectron SL803). This laserA~r + H, - ArH * + H beam is passed across the diameter of the flow tube 17.5 cm

downstream of the CO 2 addition point (see Fig. I). and

- H,* + Ar, (6) when spacially filtered provides an energy of typically I mJ

ArH'(Hl* ) + H 2 - H; + Ar(H). (7) in a4 mm diameter laser beam of 10 ns pulse duration. Thefluorescence from the OH in the plasma is detected at right

This afterglow plasma is a unique polyatomic molecular angles to the laser beam (see Fig. 2). Collection of the flu-ion/electron plasma because H," ions in their ground vibra- orescence is via a quartz lens (focal length 3.8 cm) situatedtional state (v = 0) do not dissociatively recombine with in a side arm of the flow tube which images the fluorescmceelectrons at a significant rate." ' H,- (v) can dissociatively production region (i.e., part of the volume occupied by therecombine" but, in the presence of sufficient H, it is rapidly laser beam) onto a slit and thence through an interferencequenched via the reaction ' filter to a photomultiplier (EMI 9813QB) which is cooled to

H3" (v) + H,- H," + H2. (g) about - 30 'C to minimise noise counts. After amplifica-tion, the output pulses are accumulated by a gated photon

Thus the ionization density in the H,"/c (and Ar"/e) plas- counter (Stanford Research Systems, model 400) and thenma decays only by the process of ambipolar diffusion which transferred via an IEEE interface to a microcomputer whichis quite slow at the pressures used in these expenments. is used for data analysis.

Since H,* has a small proton affinity IPA(H,) = 101.3 Throughout these studies, the laser was used to excitekcalmo-'] it can transfer aproton to manyother species th A"-.(V = 1) to Xl'r(v" =0) transition of OK (atwhich can be added to the flow tube (at inlet port 14) thuscreating a plasma, which can recombine." e.g..

H,' + CO, - HCO,' + H 1, &H 2 - 7.7 kcal moP'(9 ChoomuliarPhtouiip(9) Coolerfpt e

This approach to the production of plasmas consisting of Rotatablparti, ila protonated species has been used previously in the ,,,_ Interar.redetermination of the recombination coefficients for many Slit Fiter

protonated specs." The exothermtity of reaction (9) is _r -z

such that the product HCO," can be produced in vibration- L Windowsally (and perhaps electronically) excited states, however Ftow 0iar't

any excitation of this ion will be rapidly quenched by the Tb

proton transfer reaction:

(HC , )', +C0 2 -HCO, + t,-0,)a. I\ Laser"

The electron (and thus the ion) densities on the aus of 8101 Beamthe afterglow plasma can be measured readily using a Lang. Ou mp Sapphi r

muir probe which can be moved along the whole length of Windows

the flow tube."' Also, by adjusting the position o(the micro-wave discharge on the glass tube (see Fig I) to change the Vacuum Jacket

diffusive loss of ionization along that tube. the ionization FIG 2.. cheimrac .ip-a ohe LIF MuxmxT=C2 .SC.fJn lrux T Yst- %' ed

density can be vaned it the upstream .,nd of the recombina- m' nig Angles 'Z Ufh 1xi, " )w iL- b

65I Pi l"S, I I I I o 2I juty '98I

(counts)

100

281 262 263 284Laser Wavelength (nml

FIG. 3. A laser excitation spectrum OOH X ll( v - 0) rdicals produced ia the diaocittive remoaibtnation of HC 2 with electrons. This spectrum wMobtained by s.anning the Lhw wavtlengph through the rotaonal structure o( the vibronic transition A I(V, = )- x Il(v" = 0) ( -281-284un) anddetecting the fluorescenAe ntensity , (onunLper S0laweshots) m theA 1Z( = i) -x In(v" - 1) band (- 312 um). Note the inal background conlevel and the corresponidingly large signal-to-noise levels on the most intense lines- The two identified rovibromc tines, A, (I) aM P, (5). we those which wereused in the present studies (we the test).

- 281 nm) and the nonresonance fluorescence was observed III. CALIBRATION PROCEDURE: LIF INTENSITY VS OHin the 4 Y.(v' = l)-X 2 l(v" = I) transition using a 10 NUMBER DENSITYnm bandpass interference filter with a central wavelength of In order to determine the fraction of the HCO" ions312 nrn. This filter prevents scattered laser light from reach- that recombine to produce OH X ifn (v = 0) it is necessarying the photomultiplier and results in practically zero back- to relate the fluorescence intensity in a given rovibronic limground noise counts. Contributions to the signal from back- to the OH number density in the flow tube- To attempt toground sources were further minimized by gating the photon obtain this relationship by calculation is fraught with diffi-counter to count only for a period o( 3,us following a delay of culty since, in addition to having to depend on Einstein cod- 50 ns from each laser pulse (the tngger pulse for the pho- ficients, Franck-Condon factors and HoIn-London factors,ton counter was provided by a photodiode on to which a it also depends on various inaccurately known instrumentalsmall fraction of the laser beam was deflected) thus allowing factors. Thus, it is more accurate to calibrate by using a reac-photon counting only during the time period in which the tion such as reaction (2). H + N03 -OH + NO, to relatemajority of the OH ,4 2 states spontaneously radiatively observed OH fluorescence intensity to a known OH numberdecay. 4 - ) To enable the photon counter to count rapidly density which can be obtained by first determining the H-during the gated period without significant pulse pde-up, a atom number density.fast photomultiplier (3 ns FWHM) and electronics with> 100 MHz bandwidth were used.

An example of part of the excitation spectrum of A. Determination of the H-atom number densJatOH X '1 (W = 0) [which was produced in the HCO " + e It is straightforward to obtain the H-atom columwn den-recombination, reaction (Ib) ] obtaned by scanning the I&- sity across the flow tube using VUV absorption spectrmcopyser over the wavelength range from - 281 to -24 nm is at the H-atom resonace line (L. at 121.57 am)."' Thus,shown in Fig. 3. All of the rotational lines are identified as knowing the radial profile of H atoms (see Sec. IV C) thenoriginating from the OH(.4-X) transition'" and the signal to the H-atom number density can be determined. The tech-noise is seen to be excellent (- 100:i on the most intense tuque used, as llustrated in Fig. I, is fairly standard' and iline). Following Guyer et al., ' and assuming that the laser as follows. L. radiation is produced in a microwave dis-transition is saturated (as is expected tor the laser beam in- charge source (with a 50 W power dissipation in these ex-tensity cmpioyed in this study), then the rclative intensities peruments) through nominally pure He at a pressure ofof the lines can be shown to be consistent with a rotational -0.05 Toff in a I cm diameter Pyrex tube terminated with alevel population distribution of OH X :1(v" = 0) which is MgF window. Since the oscillator strength of the transitionBoltzmann at a temperature of 3 10 K i.e., essentially at the f is so large, such lines are often self-reversed. However, de-temperature of the experiment ( - 300 KY. Thus the OH tailed exprimen,,a studies" of the profiles ofL, lines fromrotational modes, as expected,' are thermalized in multiple discharges similar to that used in the present work havecollisions with the He carrier gas atoms prior to laser e.xcita- shown that significant self-reversal only occurs withtion. This is not the case for the vibrational modes and so >0.01% H2 mixed with He at a pressureof - 2 Torr. This adetailed information as potentially available concerning the a much greater H, partial pressure than that in the presentOH product vibrational distnbution. source (which is < 10 ' TofT) when "pure helium" is used.

66

In addition, the present soune is thoroug ly degased by thermal Doppler broadeniag (as aa been shown to be thebeating under vacuum and with a continuous Bow o(beium case for microwave discharge sources o(the type ied in thispassingthroughit toremovewater andany volatilehydro- work") then a = (T,/T, )' where T, and T. are thecarbons from which H atoms can be generated. Also the source and afterglow temperatures, respectively. Thus themicrowave discharge is operated as close as possible to the deduced column density is not a sensitive function of theMSFj window to minimise ihe column density of H atoms source teinperature. For this type of source T, has been pre-that the L. radiation must pass through, thus minimizing viously measured to range from 305 to 900 K depending onany self-reversal. Note that it was not possible to check di- the microwave power level." These data suggest a T, of 650rectly for self-reversal in the L. line from the source because K for the power level of 50 W used in the present experiment.of the limited dispersion of the VUV monochromator used in That this T, was an appropriate value was confirmed bythese experiments (see below). However, for the O-atom adding a trace of N2 to the lamp and observing the rotation-resonance line triplet, which was also emitted by the source, ally resolved emission from the B 21 ;the individual components could be resolved and their inten- (v=0)-X'1*(v" = I) transition in N*. Analysis ofsities were in the ratios 'P2: 'P,. 'Po = 5:3:1 which are the the relative intensities of the rotational lines" giv,' a valueratios of the statistical weights thus showing that no self- for the temperature of 650 K. It is known that the B 2*reversal occurred for this line. We are therefore confident state of N * is populated by Penning reactions with He meta.that there is no significant self-reversal for the L. line. stable atoms'0 (reaction with He* produces N1 in the

The L. radiation is directed across the diameter of the C 21 . state) and that the N rotational populations reflectflow tube at a distance 35.5 cm downstream of port 14 (see those of the N2 precursor. Thus since the rotational popula-Fig. I) and onto the entrance slit of a I m vacuum mono- tion of the N, will be equilibrated by collisions at the kineticchromator equipped with a 1200 lines mm -' grating to dis- temperature of the He and H atoms, then the kinetic tern-pene other interfering radiations originating in the L. perature of the H atoms is expected to be - 650K. Note thatsource and in the discharge which produces the flowing any error in this temperature will contribute to some degreeafterglow plasma. It is then detected at the exit slit of the to the error in the H-atom density determination (typically,monochromator by a channeltron multiplier which, because an error in T, of 100 K results in an error in N of - 6% ).of its large work function, is only sensitive to VUJV radiation Taking all known sources of error into account, we estimate(wavelength < 150 nm) and thus has a low background that the deduced H-atom densities are accurate to withincount rate. The He it line at 121.52 n, which is close in 15%.wavelength to L.L is a possible source of interference in thesestudies. Fortunately, however, spectral cans at maximum 8. The calibration remaction: H + NO 2 - OH + NOresolution over the L. region showed that there was no sig- For this calibration the K-atom density in the flow tubenificant contnrbution due to this line from the L. source.

The fractional absorption .4 of the L. radiation which was maxunised by passing H, through the microwave dis-

occurs in its passage through the fBow tube is related to the charge and by adding N, at inlet port 12 (see Fig. I ). By

H-atom column density by the expression (see, for example, passing H, through the discharge the H2 was partially disso-

Mitchell and Zemansky) ciated to produce H atoms with the accompanying produc-

tion of an H)" e plasma. The introduction of N, to this plas-

A i -I, = f;1, exp( -k.L)dv (Ill ma resulted in a rapid loss of H)' and the conversion to an

!, 1 l. dv NH'/e plasma (following proton transfer from H, to N2)which at high electron densities produces additional H

Sp ( I + ap)" (12) atoms via the recombination reactionP )N2H * + e -N, +H. (5

where the frequency dependent absorption coefficient k. inthe case of Doppler broadening (see below) is given by This reaction also resulted in a large reduction in the dec-

S 2(v -Y)(in2: iron density in the region of the afterglow just upstream of

k. = expl-2 -v 0 )(-n) 1 , (13) inlet port l4from lxlO'°to Sx l0 cm' NO. was thenIIv added at inlet port 14 and the OH LIF fluorescence and H-

where atom absorption w'ere monitored turther downstream at dis-

I (In n2) eV tances If of 17 5 cm and 1, of 3 5 cm, respectively (see Fig.o = -. " / \ . (14) 1) .Since the electron density and the H, density were so

* =v 2mtc small, the calibration via reaction (2) was not signifcintly

1, and 1, are the incident and tr'asrt.-td intfnsiti1, i. he influenced ( -0.5%) by production of OH in the reaclionfrequency dependent emission line profile, p is an integer, V0is the frequency at the line centre, a is the ratio (Av,/v) H + NO 2- KNO" - - H2where Av, is the full width at half-maximum (FWHM) of LNO + OHthe L. line emitted by the source, and Avo is the FWHM ofthe absorption line, L is path length through the absorbing HNO ± H2O (16)medium, N is number density of H atomsf is the oscillator even though the rate coefficient for this reaction is large."strength for the transition and m is the electron mass. If both Figure 4 shows semilogarithmic plots of the H-atom 3b-the emission and absorption line profiles are determined by sorption. I, - 1,, [see Eq. ( 11) and the OH LIF intensity

67J cl- n Ptys ,/ 91. %40 2. 15 'Uty '969

producedl in thie HCO; recombinatin reacton was simaflh................... than for the H + NO, reaction.

IV. RESULTS ON FROM THE KCO; +bO' URECOMBINATION

A. OH(v* si:) production

In order to ensure that the observed LIF fluorescenceIi-ti intensity If, following lase excitation of the

tal 0.1A 21W( - z)- 2n wv =0o) transition in OH. was origi-kowf (coiwrs) notint from OH produced in the dissociative recombination

of HCO,* with electrons [ reaction (I ) ], two experimentalchecks were made. These checks proved conclusively that

10 _101 the observed fluorescence originated from the HC02 - re-combination. First, the fluorescence was monitored as afunction of the flow of CO2 introduced at inlet port 14 and

r thus as a function of the CO2 number density in the fiow0 12 I S~ 7S 1x1~'tube, [C02 ].This varies the amount of HCO* produced by

141140 2 ILN02) (CA inoic c~i" proton transfer from H,* Ireaction (9) 1 and consequentlythe amount of OHWv 0) which can be produced in the

FIG 4 Diereaein the absoqxion I. - I. (cns par 50 s) oiL. radiaticin by recombinauion reaction (I). A sam ple of the data obtained isH atocm (inan&i) as a function of the product 1. 1 NO. I dw tohe calbra given in Fig. 5 where it can be seen that!,( increases smoothlyboo ruiown H NO, -OH +. NO I resctn (2)] (NOI a the numbei with [C02 1. reaching a maximum at large [CO21. in the

denit o(NO mo~,l~ ntoduedint de aerow t i~t14and. ~ manner expected for production via reaction (9) followedreactm length beween this port and the positn at which dic H atom an by reaction ( I) The form of this increase is not simply relat -ontored (see Fg n. Also liven isthe meated uscrase in am LIF inteayofdesisae

OH It (open ciecle) (fluoresce hoo ons e 5W W,, hml) r, tk ed only to I[CO 2 1 since the HC02* and electrondestsarP, (5) line in the A 1 W~ - l) - zfln(v, 0) ecmuticin spwrnm (wee Fig also involved in the reaction sequence. The formnofthis curve3) with walan If iNO, I where I, a tk reactto 1ength betweeni pn 14 and has been modeled takcing into account the production ofdhe Imet bem. Boths I. ad If ame inctuded stice thes raclioi lengili fo H HCO * in reaction (9) and its loss by recombination to pro-7%c aned OHrk twmv2 obW~ne by~ uitiig(5 td OH L floeonritni duct OH [reaction (I) I and is consistent with the known

f teasmptotic vwatrp~ 1, NO. I (as Vvmu by the dahe he) Thes rate coefficits for these reactions"bwuW n that for the L. abtlKi pw va~sis of the rate ooeffiet for tthe The scnd experimental check was to monitor If as aH + '4O Illwswa which an in good apgecnent with the lterature value (wee function of the initial HCO, density (i.e., before recombi-the tAt). nation had occurred). For these measurements, a [CO2 I

was established in the flow tube which was sufficient to pro-

for the P, (5) line against I. 1NO, I anid If[ NO., 1, respective- duct the maximum I,( (see Fig. 5). For this situation.ly. where [ NO,) is the NO2 number density in het flow tube(determined from the NO, flow rate which was measured inthe conventional manner for fiow tube experiments) and I.and 1,, are the appropriate reaction lengths. T1he linear de-crease in the H-atomn absorption with [NO,) gives a rate 1ocoefficient forreaction (2)o(l.3 ± 03) X 0' cm's-'.in excellent agreement with the literature value at 300 K of1 3X 10 - '0ci's'as obtained from several studies.o The frate coefficient for this reaction can also be deduced from the (counts)increase in the OH LWF intensity by subtracting the mea-sured intensities at various [ NO, Ifrom the constant value atlaige [ NO,2 I (the dashed line in Fig. 4). This agan providesa linear decrease from which the same rate coefficient isagain obtained. Thus the K-atom column density and the 10 2OH LIF intensity for the P,(5) lne are related via reaction 2 3'u1300

(2) and a calibration can be obtained for [ OH]I using [ H [ CO I1 I motec C'deduced from L. absorption measurements. In addition, the FGS nrei h I iniyfro N1 poo xrt e 5 nrelative intensities of the P, (5) line and the more intense OM) 5it tnenig he ime dneny fran OI a taiceo rotspt th0 RuR,( 1) line weremeasured at alower [OH] inordertoobtain Sitl kneithe.4 mangC. iub- ny .o ~iCaott sectumg omr t& 3)a calibration of the OH LWF intensity for the R,(1) line Th =cmrmftsfom H(Y' - 01produtson imtk HCOW dmcutbviwhich was unaffected by pulse pile-up in the detection sys- rnenbmoon with electra-n1 rwum (1) 1. the WCii being produced bytern (set Sec. 11). This extends to lower values the range of pruiton trar r-ftr H, toCO Iratam (9)J1 For huiberdiscssim ei[OH]I which can be measured and is useful since the [ OH)I t-

68

HCO" is rapidly produced by reaction (9) and any excited RIO*HCO " is rapidly quenched by reaction (10) before recom- 5

bination can occur. Thus the initial HCO," density wasequal to the H," density just upstream of the port throughwhich the CO, was added and therefore equal to the electron ',

density n, (0) at that axial position. The electron density canof course be determined very easily using the Langmuir N 3probe. The n, (0) was then varied as described in Sec. 11. If

The way in which I, is expected to vary with m, (and .

hence [OH)) requires some discussion. When recombin- (Cm couns 2tion of HCO2" with electrons is the dominant loss mecha- 2nism. the variation of the HCO number density I HCO" Iwith distance z along the flow tube is given by I

d [ HCO: !UP d - a.[HCO2, ]n,(z), (17)dz

where u, is the plasma flow velocity (this is readily deter- 00 1 2 3 . 5 6 7 6,,10'mined, see Ref. 53) and a, is the ion-electron recombination n (0) (CM,)coefficient. For [HCO,' 1, = n,(z), as in the present situa-tion, this has the solution FIG 6 A mpk plot of " , (0), I, alinst ow, (0) for the deuocmuve ruomb-

aio of HC" with electrons [racimn ( I n,(0) the elctron densit

HO. = (,z + Jut ulpsream of the pon t "which CO, is aed L the H1 / afterlow

HCO, ' plasma. and I/ s the LJF ntensity from OH (photon counts per 500 imer

shots) fothe R,()line ntheA -(= i)-Xv= lv" = 0) exciLation sp c-where the subscripts 0 and z identify the initial number den- trum (see Fig 3) at a disance to( 17 5 cm downstream from the CO: addibon

sity and that at position z, respectively. As for Eq. (17), the point. Thi plm ishnear as expected from Eq. (20) From calibration damta

rate of production of OH can be represented as such as ipven i Fig 4.1, cm be related to the OH(v' O) number density and

d OH then the frambon of the HCO - ions that recombire to produce OH(v 0)

- - -fa, [ HCO: ],n,(z), (19) be ro t

dz coffier can be o esa from the inte'pt~ Of the tine on the ordinate Ism

wheref"is the fraction of the recombinating HCO1 ions that Eq 2

result in the production of OH Substttution for [HCO" 1,from Eq. (18) and integration yields [for OHW = ) quenching by NO, the

=HC02 1 _ HC 0 + (20] rate coefficient at 298 K, i.e., k(298) = 3.8x 10 -" cm'

(OH 1, f aji s - ' " I but unlike NO, [ reaction (2) 1, it reacts only very

Thus since [ HCO." 1, = n, (0) and, [OH 1, is related to If, slowly with H-atoms.' Thus it is ideal for use in detectingthen a plot of n, (0)-2/, vs n, (0) should be linear. Such a OH(v > 0). viz.pkot is given in Fig. 6. Note that it is linear as predicted and ifthe relationship between If and [OH), is known (see Sec. OH(v >0) + NO-OH(v = 0) + NO. (21)

III A). thenfand a, can be determined from the slope and The variation of If with [NO is shown in Fig. 7 for theintercept, respectively. Since it is the relationship between I, situation where a large [CO2] had been added to ensure thatand [OH (v' = 0)] that is actually determined, thenfrepre- an HCO,'/e plasma was rapidly produced. It can be seensents the fraction of recombinations that produce from the upper curve (filled circles) that as [ NO I increasesOH(w = 0). The above experimental checks confirm that 1, increases showing that additional OH(v = 0) is beingthe fluorescence I/and thus the OH(v = 0) is originating produced from the quenching of OH(v" > 0) generated infrom the HCO:" recombination. HCOi- recombimation The increase in OH(v = 0) due to

the quenching of OH( v > 0) can be seen more clearly from1. Otvv > 0) prOductiOn the lower curve (open circles) which is obtained by subtract-

Examination of the energetics quoted in reaction (I) ing the constant contribution to If due to the OH(v = 0)shows that, in this recombination. it is also possible to pro- produced directly in the recombination reaction. Compan-

duce highly vibrationally excited OH [it is also energetically son of the asymptotic value of 1, from this curve at largepossible to produce OH in the electronically excited A 'I [NO] with the value of 1, at zero [NO] gives a ratio ofstate, see Eq. (22d), however. tbis would have rapidly radia- OH( v >0) to OH( v = 0) produced in the HCO2 reac-tively relaxed to the ground state before the position at tion of - I [this ratio will be unaffected by any nascent rota-which the OH was monitored .To determine the fraction of tional excitation in the OH(v" = 0) product of reactionOH (v > 0) produced in the recombination. NO was added (20) since this will be rapidly quenched by the He carrierto the flow tube at a position where all of the recombination gas]. Note that an overall quenching rate coefficient, i.e..had occurred but upstream of the point at which I. was mon- that for reaction (21 ), can be deduced from the lowest curveitored. NO is known to vibrationally quench (open circles) in Fig 7 By subtracting the values of1/ on

69i C, n PlysVol 91,4c 2. 15 July '989

wene conducted to estabb tha thm we so qumdchefects due to the pruence of te vaui ohr pm t eflow tube. OH(v >0) was produced via the HCO * +erecombination in the upstream region of the dow tube aidAr, HI, and CO, were added separately in large concentra-

10) tions at port 14. Addition of these gases at number densitiesof -2x 10" cm - (which is equal to or greater than thenumber densities of these gases used in the recombination

if experiments) produced no significant increase in the€ountsl -OH(v" = 0) fluorescence and thus it is concluded that

quenching o(OH(v >0) (produced in the present experi-ments) by these gases at the number densities used was notsignificant. The fact that the OH(v = 0) fluorescence didnot increase when CO was added also established that the H

1__atoms present in the flow tube reacted only insignificandy

0 1 2 3 1 5 6 7 8 9 u10 with CO2 to produce OH which is consistent with previous

[NO) (motec cm') findinp."

FIG. 7. Variation indh NO number densty in th afterglow ( NO I of the LIF C. Difluolvo and reactive losses of H and OHmtuawty fro OH If (ph on ount pe 20 law *h % fUlia cuJle ) foe the

.(I) ine m the A'1(V = ) -xfl(n =0)o ewuit (- It isa now only necea to relte f 1 [OH(v = 0)]c oFi.l ). OH X 2(W >0) an x iv' 0) one pde dpa,.sm 0 obtain absolute values for the fractions of OH(v >0) andNO aldiuci point from the di scuiuve reombnation o HCO" The in- OH(v" = 0) which ire produced in reaction (1). The dis-rme i If with (NOI i due to quenching o(OH " > 0) by NO to pntduce cussion of the calibration procedure given in Sec. III A re-

OH(v" =0) 1tre (2)). Thus thechanlembetwe wro[ JNOI lates 1. for the OH(v' = 0) produced in the H + NO 2 reac-the AMPWM vatue as large (NO . oe to the valueO If atZe I NO]. tion to the initial H-atom column density. However, in orderrtret die mou OH( v"> 0) neC to OH(" - 0) hich Pro to relate I to (OH(v' = 0) 1 it is necessary to know the H-duced e the HC" mctnatinm rema. Tbe u-mm m OH(V" >0) cab uma @Kx. (open acks) by triaCth d tr of atom radial density profile across the fow tube so that [H]Of(fv -0) from /, The ecfave m r t te qulich.ain ai can be deduced. From an empiical expression, the diffusioncm b daWumd rm the %ope of the phi (v=&) ,xh , auiawl coefficient D for H atoms in He at a pressure of 1.6 Torr andby iibuactang the values representmed by the CaM ciavie flt the armitom at temperature o( 300 K used in these experiients can bevaluh a k Ie (NO. a matel by thme dahmi him calculated tobe -(1300 ± 200) cm3 -'. lus, under the

pressure coodiions ofthe present experments, H atoms canreadily diffuse to the walls of the Bow tube. In an earier

this curve from the asymptotc value at large [NO], the de- study ofthe reaction of H atoms with ions. Howard nelj.cry curve (triangles) can be constructed which provides a concluded that H atoms are no lost sagnifcandy on therate oefficiento (6 ± 1.5) X 10" "cm' s' for the rewuon walls of stamlesessl Bow tubes of simila dimensions toa 300 K. It is interesting to note that this value is similar to that used in the present study." This has been confirmed bytL. value o( (3.8 :t 0.6) X 0- cms-' Y obtaine d Row ae1. " wo also showed in aow ube tt there is soat 300 K.' and within the nse of values (1.5 X 10" to significant radial variation of H-atom density. So. in the7.5 x 10- cm1 s-) 2 previously obuined, for the quench- present studies, it is to be expected that the H-atom density isin o(OH(v = I) by NO (a rate coefficient for the quench- sensily independent o( radial position and based on this,mg o(OH(v = 2) by NO o( 1.2 10'- * cm's' has also [H nd thus OH(v = 0) 1 can be eas ly deduced from thebeen obtained" . H-atom column density and [OH(v" = 0)1 can be related

Ifthe ratio OH (v > 0)/OH (v' = 0) is tobe equated to to!I.thAt resulting from the HCO(" + e recombinatoo rewtion. At this point, it is worth considering whether sigiicanttb it isnecessary toestabish that OH(W > 0) is not sigpf- radial diffusiou and wall recombinatio o of OH occursic.ndy radiatively or colisionally quenched in its passage between its production and detecton. This general situatonving the Bow tube. Radiauve lifetimes o( OH(v >0) for has been analyzed by Kaufman" who showed that, in thev' = Ito 15 (for reJaxat-oo to all lower v levels) have been steady state for k only by wall recombinatio. the frac-calculated by LAngboffer al." and range from 93 to $ ma. tional difference between the number density onrespectively. Tberefore, there is expected to be very little [OHI.,and that atthe walls, [OH , (in termsoftheaver-radiative relaxation ofOH(v" > 0) in the - 4 ms time scale age number density f O-H I) isofthe present experiments. Also, it is known that collisional J011 - (0111 3kquenching o(OH(v" = 1) by He and At is very slow; the = ,3.!...(22)mtecoeffciens for these .iactions at 298Kare x 0" [10 -D

and < I X 10" "cm' s -, respectively' and at the pressures where k. is the wall recombination coefficient. r is the Bowo( He and Ar used in the present experiments are equivaient tube radius, and D is the diffusion coefficient for OH radicalsto reaction time constants of 350 ms and 9.4 s. respectively, at the pressure of the experiment. The value of D for OHThus quenching by He and Ar is insignificant. Experiments radicals in helium can be estimated from the D of oxygen

70j nCn Pty%.Vl .N 2. 1 l^ 'S"y

atoms in helium" to be - (436 ± 150) cm, s-I at a helium ad (Sal .d-)pressure o( 1.6 Tor. The value o(k, has been found to vary - H('S,,,) + CO (I'1," - 15.1 (2.a)widely, rmnging from 8 to 50s-I (and perhaps even greater) -H('S,,, + CO(A D0,(A.) - 53.6 (22b)-OH(X 'f) +O('Z") -I150I (22c)for tubes of diameter 2.5 cm depending on whether the sur- -OH(A In) + co( II) -636 (22d)

tace is nominally clean Pyrex or is Pyrex coated with H3 PO. -o('P1) + H'S,,,) + CO(X ") - 57.9suitably aged. '4*i Assuming the largest value of k., theworst case situation, correcting it for the difference between obtained from Refs. 63-65. Production of CH X 2In and

previous and present flow tube diameterss and using an ap. %X'18,i- is close to thenroneutral (AH - + 0.4 kcalpropriate value for D in Eq. (22) gives an [OH]. mol-'),however, this channel requires a great deal o( rear-

S65% [OH I., a small radial gradient. Further, from Kauf- rangement of the (HCO2)* after neutralization and beforeMan5 its dissociation and is therefore unlikely. The experimental

(k / 4k. D\. ( data obtained so far show that - 35% ofthe recombinations(j Uos ( finally result in the production of OH in the electronic

where %* is the bulk velocity of the flowing He gas. which ground state. However, at least some of this product could

predicts only a 13% reduction in [OH] between its produc- result tram the initial production of OH A 11 which would

tion at port 14 and its detection 175 cm downstream. In any then radiatively decay to the ground state with a lifetime

case, the calibration procedure corrects for such a reduction - 700 ns'"' Indeed recently, we have detected weak optical

and thus it should not influence the determination of [OH] emissions from OH A m1- X 2fn transitions in a recombin-

from the recombination reaction. ing HCO1 le plasma (unpubfished results). Therefore a

Finally, it is necessary to consider whether there are any fraction of the detected OH X 2fl (v" = 0 and v" > 0) ori-

reactive loss processes for OH. OH can react with the H, 0, ginated from nascent OH A "I. However, because of the

and OH present in the flow tube, viz, weakness of the emissions, our judgement is that an appre-ciable fraction of the detected OH X 2 ( = o and v > 0)

OH + H 1-HO + M are nascent species. That such a large fraction of the detectedOH X 21" are in the lowest vibrational level for such an ex-

- H, + 0, (24) othermic process [channel 22(c) ] can be explained, at least

OH + 0-0. + H, (25) qualitatively, since the 0-H bond length in HCO2 (as de-w , termined by structural calculations" to be 0.9979 A) is very

OH + OH -HO + M, (26) similar to that of OH in the X12f state (0.9697 A).' Thus,upon recominbation of HCO to produce OHX If, there is

however, the small concentratons of these radicals com- no significant change in the OH equilibrium bond length,pounded with the small rate coefficients "' renders these loss resulting in unimal vibrational excitation of the nascentprocesses insignificant. OH product. The remaining excess energy in the recombina-

tion reaction would then appear largely as kinetic energy of0. Absolute OH(v =0) and 0H(v* > 0) products of tho the separating products. It should be noted that it is also14CO" recombination reaction possible to produce CO2 in the 4 1B2 state [ reaction (22b) I

The percentage of OH(v" = 0) produced in the HC0 2* and radiation from the decay of this state should also berecombination [ reaction ( I ) I has been deduced using the searched for by emission spectroscopy.calibration procedure [ reaction (2) and Sec. III B 1, the data To our knowledge, no theoretical effort has been direct-

in Fig. S together with the measured value of n. (0), and the ed towards predicting the products of the reocnbination ofslope of Fig. 6 [see Eq. (20) 1. In this manner, it was deter- HC01". However, the theoretical ideas of Bates"'s can bemined that OH(v" = 0) is a 17% product and thus from the used to make some predictions. The structure ofHCO* has

OH(v > O)/OH(v" = 0) ratnoo - I deduced in Sec. III B been calculated to be HOCO with the positive charge local-that OH(v' > 0) is also about a 17% product giving a totalOHXfin of (34 +6)%. tized on the 0 atom between the H and the C "'& For this

Once the contnbution of OH(v" = 0) to the product stuati, the ideas of Bates suggest that H + CO. and

distribution has been determined. then the data in Fig.6 to- OH + CO would be the likely product channels. Further,

gether with Eq. (20) can be used to determine the recombi- Bates' has suggested that the likelihood of a favorable curvenation coefficient a.. for the reaction. Thus a, has been de- crossing to products will be greatest when both products aretermined to be (28 + 0.8) x 10-' cm' s at 300 K in radical and molecular. Such an effect would be expected to

reasonable agreement with the value of (3.4 ± 0.85) x 10-' enhance the probability of dissociation to produce

cm' s -I previously determined in the usual FALP expen- OH + CO relative to H + CO. However, it should be noted

ments (from the axial electron density gradient in the flow that if the OH is produced in the A 1I * state with v > 2 then

tube) at the same temperature." rapid predissoiation to 0 and H can occur." Although

V. DISCUSSION these qualitative predictions are consistent with the presentexpernmental observations, it is obviously important for

The energetically allowed product channels for the dis- quantitative theoretical predictions to be made which in-sociative recombination of HCO," %ith electrons are clude the effects of curve crossings.

71j C? wn Poys, Vo 91 NO 2.'. .uty '90

V. cONCLUSM~ 'ID RhY. 0. hg. La, nd L J. litm J. Ci rys. 4& 4137(1977).

A technique has been decribed which can be used to N A Ouice k d . . Z.i, I O sa s 3n 7, 429 (1 73.

quntitatively idenify the neutral products of lectron-ion "f vAl, 3I. k. sio . . C. Ga , I. L Q 7&fec. md N. Mo"193

disociative recombination when these are amendable to de- Chs. PbL Lam 134 317 (1966).tection by laser induced fluorescence. In parnicular, recom- U. . Rove F. Vallee, ). L Q reffc, I. C Garnet, and M. Morka

Chea. PhyL U. 845 (1917).bination reactions o(protonated species can be studied. Data "). L Querf l. It. .Rowe. M. Morla, J. C Gaoret, &d F. VaJ, Plm .have been presented for the HCO * recombination reaction et. Spam Stc 53. 263 (1983).which show thai OH in its ground electronic state (in both '8 L Gubemum,. Rd. 0 p. 167.• "H. H. Micbh and R. H. Hobs Aaarophys. ILest.261.27 (1934).its v' = 0 and vibratkxoaly excited states) is a substantial N. H. rMhel A. U H bbs ReeamkntIwg. T2L7 9).

product being - 35% of the product distribution (note that Expir, .and AP#k,1*o edited by J. B. A. MiWbd sad . L Outb-this percentage may include a contribution due to the A 'I * a. (World Scintiic, Sitippo IrM), p6 61.state which rapidly radiatively decays to the ground state). "E. Herbs. AKwOpy J. M 506 (1978).S. Grmi.d .H otiyJ ma, 121 c1979).These data have implications to the chemistry of CO . in D2S. Gm. - Hers 7. 49 (Jn9,).

interstellar gas clouds and this, together with data concern- "D p. R.S A.tlIph1. J. Ltt 3L L45 (1916).ing the products of ocher recombination reactions is the sub- 'D . mes. .V dem, AM/ norrnea,,4c n.loknwr, hasuesf an-ject of a separate paper" In addition, in the HCO,* recoMa- bA. AE. Kjpto .(Plenum. Londcm. 1987). p. I

D.It Baem Astrophys. J. (m pess)bination it is energetically possible for OH and CO2 to be "D. I. gates &-d E. Hebst in Rd. 11. p. 41produced in their first electronically excited states. The 'N. G. Ad&= and D Smith. ChmL Phy Lm 14411(193).quantitative detection of radiation from these excited species NM. I. Howard ad I W M Smith. Prol Rmct Kine. 12,5 (1913).using emission spectroscopy would provide information on O Kleierman and 1. W M. Snmh. I Cwmw Soc Farad Tram 2 13,229

(1987).energy disposal in the reaction and such observations re "1. W .M. S'udi (private wximunwauon).planned. Also it is energetically possible for H and 0 atoms 'I. W M. Smith. and M. D Wdhasm J Chen. Soc. Faraday Tram. 281. 149to be produced. Efforts are underway in our laboratory to (1985),improve the sensitivity of atomic absorption and atomic flu- "D Smth and N. G Adam in Re )a p. 501.

'4 D. K. Bol.me. N 0. Adam M. Mose nwi. D R Duwikm and E. E. Frgu-orescence spectroscopy at the atomic resonance fine wave- Wft J. C%. Phys. 5S5094 (1970).lengths (e.g.,/. ) to enable the contributions ofO and H to "N. G Ada nuD Snth. and E gAt. J Che PhyL It 1771 (194).the product distibution to be quantitatively determined. "J. K. Kim L P Theord. and W T Huntrew Jr, tntJ. Mass Spectrom ta

When this is achieved, the total product distribution of ths " h , M 2 (1974).

reaction will be specilied. This then will give guidance to (1975).theorists in the calculation of the product distributions for 'N. A. A. Clym and W S N* i Aewaca I,niedia a At Ga Pthis type of reaction. eSted by D, W Ser (Aca . New York. 19). M 1-5 .

*A. C Eckbeth P A. BKzyk. and J F Vetbick. Pv Enrgy CombuMSCL 1. 253 (1979).

'V M. A , tum.G Ei and S R. Leom G Ga i /OR Chen vrA

ACKNOWLEOGUMENTS Imu aod 141. edited by K T Bow n (Acadom New York, 1914). VdI. .W 1-39.

We are gratefui to the USAF and the SERC for financial "A. C- Ekbmk Law Dqr ,stwu r Co m,ma. Te,,,,,ow and Spernmsupport of this work. We are also prateful to Lhe Univeruty of (Abacti. Tmbndge Wek 191).

t a ph w K. t- Gennm.I Chem. Pbyi 6, 5252 (1975).zimugham for the award of a VlU 'nt with wbxh it Was -W L DmO i ,d J. L Kiney. J uent. Spwrn- Ra, T-rru n. 233possible to purchase the specoscopic equpmenL We wish to (1979). w H Smi, 3& G. Ehnerpmli and N. H. h s 1 Ceu.m PhyLthank Profemor I.W.M. Smith for usdl discussions concern- 6L 2791 (19741.mg neutral-neutral reactons. Thanks are also due to Mark "G H. Dke and H. M. Cnofute. J Qua". Spixuwr R dia Transer 2.

97(1962)Geoghegan for his assistance with the data analyss. D. It. G . L Huvmd, d S, R. L w J. Chm. Phys 79. 1259 (193).'A. C. G. Mitchell and !,[ W zanky. Rwiancr Rdauam and Excuad.4hou (CAridge Una iry. New York. 1971).

-M. Dudac. 0 Poait. B. K Raw J L Quea.. md Mada. .PhyL D 146 " (183)

'A. D[Wproo sad I Itl Black. Re. Prg Pby. 3,573 (1976). -A. LF-hatiG . %mitr. and D R_ Wood, J Chem. Phlyl 55607(19").:) Souath ad N. 0. Adaim ln Rev Phyt Chbm. 1. 271 (1901). 'G Keoztrg, Wokerr S-piiwm and Wolerlr SDkni f Sp itr *'4-'T I. Mdiar. 0 J DeFriet A. D McLean. ind E. Hers, k.azr As- atomr.Mokcilue (Van %ostrand. Pr, ce on 1950).uaphyt 1M4 250 (],1). 'A. L Schmdukopf, E E. Farno, and F C Febsod2 Chenm Phy. Ut

*G C. Raid. Adv. AL Mol.FPayi. 6, 375 (1976). 29t 1961).'D Suthand N. 0 Adm, Top. Car Chem. 19.I (1910). "1 .,- Bur.J L Dun. J Mc Ewan. %1 %4 Sunon. E Roche. andH IF J MehI r and M. A. Bi.y. i i 11,b .264 (1969) S;haf. J Chem Phy,3 .606 t 190),". Alge ,N G Adams. tnd . a ) Pbys 816. 1433 (1993). "Y Ike= S. Matsuok .4. Takeb mand .Vtgpno, CooP I.dOle-

'D. AmerachR, K Cacak. k. Caudan. T D. Galy. C. J Keyser. . W cule Amit. Raw, Coasanxiat g 1966 ,M!aulai Tokyo, 1987).McGowmn.J B A. NtcheandS. F J. Wdk. J Pbyv BI, 3797(1977). "N G Adam M. J Church. and D Smth. J Phys. D& 1409 (1975)

'I. N bardaey and M. A. hond, Adv At. Mol. Phy. 6.I (1970). -S. . Lamghuff. X, L Jaffe. J Hr L"a, and & Dalp oa. GeogpyL Res. L.*J. B. A. Mitchell and J. W. McOaw*, t Physic aflon-Ion and Ekcima- tO. " (1993)

lom Co/intuu, edited by F Oroudlard and 1. W McGoma (Plesm, "t. W M. Smith. Top Can Chan X. 112 (966)New Yort. 1913),p 279. 'E.. %avimt ndT R. Marrem Adv AL N. Physl, &15 (1970).

"N G Adams and 0 South. n Rate Cweffirwteu ,n .4tAtocnwry, a6ted "C I Ho-astt F C FehwneK. nd MI McFaland. I Chan Pbys 60,5T6by T I Millar and 0 A. Williams (Iuver., Dordrecht. 19811), pi. 173. (19-4).

"2F VuJkJe. B It. Rowe, J L. Queffeke. 1. C C, &mtM M. Morlam, J. "F Kaufman. ?rol React Kie 1, I (1961).Chan Pliys (to be published). -P Bemhat. J Gvphiys. Rca. K4 N3 t9I").

72

J Cm !Pimys, voi i31 No 2. 5 iJuly I 9

Ay,~d Not W Wrt OH 9w0lonbW + OhINf onJ* m u

"3. 0. Asima9M .3.. Margiiaa lad F. LAAAA& J am rY 6k )301 46CAC Notw 4V'C~fm, sud ft 4ft dt wi by L. C. TAM(19IMj. (Oncai Rbe. I=DOra naos.Mu IOU).. P FIX

"Ni. A. A- Oye sad & Downa 3. Chin. Sor. Faraday Trm. X. 70. 23 ftM. J. Fkch. K. F. Schader Ill. ad 33S Dinkley. 3. PhyLm s . 11292(1974). (0915.

"hM. A. A. Opuia and P 4. Hok. J. Chem. Soc. Faraday Trans,. 275. 58X. 'K. P. Huber and 0. Hcrzbcq. MoesA.- Sporvr. ead Mokruws, Sntrc.(1979). i'm',. IV. Cwrman aDiatomar Melinda (Van Noruad RanhoKd New,

". 0. Adosm D. smith. m. rCy. 0 Javabery, N. D. Twiddy, ad. E . York. 3979).Forgusam .Cben. Phys. (uins). '8H. F. Schaefer 111. in loA eooClwteIlam Spct-i,~yndStrruaj

"N. Mi. lostock. K. Drain. 3. W. Suuier. and L T. Heryos. 1. Phys. ad by I. P. Maier (Elsevwa. Amsterdam. 1989) pp. 109130.Cben. Rd. Data- .ESupp. 1 (1977). '*C. It. Herd. N. 0. Adam and D Smith. Astrophy. J. (submitted).

73

APPENDIX 6

OH PRODUCTION IN THE DISSOCIATIVE RECOMBINATION

OF H 3O+, HCO 2 + AND N 2OH+: COMPARISON WITH THEORY

AND INTERSTELLAR IMPLICATIONS

C.R. HERD, N.G. ADAMS AND D. SMITH

Astrophys.J., in press, 20th Jan (1990)

74

OH PRODUCTION IN THE DISSOCIATIVE RECOMBINATION

OF H3 0+ , HCO 2+ AND N2 OH+ : COMPARISON WITH

THEORY AND INTERSTELLAR IMPLICATIONS

Charles R. Herd, Nigel G. Adams and David Smith

School of Physics and Space Research

University of Birmingham

P.O. Box 363

Birmingham, B15 2TT

England

Short Title: OH Production in Recombination

75

ABSTRACT

The percentages of OH radicals produced in the dissociative

recombination reactions of H30+, HCO2+ and N20H

+ with electrons at 300K

have been determined to be (65 ± 10)%, (34 ± 6)% and (31 ± 5)%

respectively. The experiments were conducted using a flowing

afterglow/Langmuir probe (FALP) apparatus modified to include

spectroscopic diagnostics. In the case of H30+, the amount of OH

produced is intermediate between the theoretical predictions of Bates and

of Herbst. The importance of this finding in determining the dominant

oxygen containing species in dense interstellar clouds is discussed in

relation to recent chemical models of these clouds. In the case of

HCO2+, it is concluded that recombination will be a significant overall

loss process for HC02+, and thus C02 , in dense interstellar clouds for

fractional electron abundances 10.

76

I. INTRODUCrION

Quantitative gas-phase ion-chemical models of diffuse and dense

interstellar clouds have been developed which, for many of the species

observed, sensibly predict their abundances (Millar and Nejad 1985;

Herbst and Leung 1986; Millar, Leung and Herbst 1987). Since the

majority of the observed species are neutral, then after initial

production of ionization by photo and/or cosmic ray ionization followed

by sequences of positive ion- neutral reactions which generate polyatomic

ions, a charge neutralization process - usually dissociative electron-ion

recombination - is invoked to produce the observed neutral molecules.

For example, OH and H20 are often assumed to be produced in the

recombination with electrons of the observed interstellar ion, H30+,

which is believed to be produced in the reaction sequence (Wootten et al.

1986)

0 H2 +11 +11 +H+ 0 0+ -- OH+ H2 O 1H20+ H 30+ -. OH, H20 (1)

H

H2 + -- H 3+

For the accurate prediction of the abundances of the observed spezies,

values for the rate coefficients and product distributions of both the

ion- neutral and dissociative recombination reactions are required. Many

laboratory data are available for ion- neutral reactions (e.g. see

Anicich and Huntress 1986; Ikezoe et al. 1987) and these have been

included in detailed models (e.g., see Herbst and Leung 1986) but, much

fewer data are available concerning dissociative recombination -

particularly in relation to the neutral products (Adams and Smith i 88a).

Although laboratory data are available for some ions not implicated in77

Interstellar syntheses (N2+, NO+, C02+ and 420+; see, for example, Rove

et al. 1987, and the review by Adams and Smith 1988a), these data only

relate to the atomic products of the recombination of these ions in

unknown states of excitation (except for the case of N2+ (v' - I),

Queffelec et al 1985).

Sove effort has been made theoretically to predict the products of

recombination. Detailed potential curves have been calculated for 02+

(Guberman 1983) and some curves are also available for HeH+ (Michels

1989), H3+ (Michels and Hobbs 1984) and very recently HCO+ (Kraemer and

Hazi 1988). Herbst (1978) and Green and Herbst (1979) have applied phase

space ideas to predict the reaction products of dissociative

recombination, and Bates (1950, 1986, 1987) has considered which products

would be likely in terms of the valence bonds which might break based on

their bonding/anti-bonding character. However, for the case of H30+, the

Bates and Herbst approaches predict very different results. The

energetically allowed reaction channels are (see Table 1):

H30+ + e - H + H20 (2a)

- OH + H2 (2b)

- 0 + H + H2 (2c)

- OH + 2H (2d)

Bates (1986) previously predicted that H20 production (channel (2a)) is

dominant, the other channels being minor, whereas Herbst calculated that

OH production via channels (2b) (80%) and (2d) (I0%) is dominant. These

theories dep-nd on appropriate crossings of the potential curve(s) for

the neutralized excited polyatomic molecule and the repulsive potential

energy curve(s) for the products. This has very recently been re-

considered by Bates (1989) who concluded that channel (2b), producing OH

and H2 , would be enhanced on the basis that there are likely to be more

favourable cure crossings for t )se products which are molecular and711

radical. Because these theoretical approaches are in the development

phase, and not least because of the difficulty in calculating potential

curves for polyatomic ions, quantitative experimental identification of

the products is essential both to guide theory and to improve models of

interstellar molecular synthesis. Reaction (2) is very critical to these

models since if the fraction of the .ecombinations, f, resulting in OH

production is small, then the dominant oxygen-containing species is H20,

whereas if f is large, then the dominant oxygen containing species is 02

(Millar et al. 1988). The effect of varying f in the models has been

considered by Sternberg, Dalgarno and Lepp (1987) and Millar et al.

(1988).

Our approach to this situation has been to determine f for the

recombination of ground vibronic state H30+ utilizing the flowing

afterglow/Langmuir Probe (FALP) technique modified to include laser-

induced fluorescence (LIF) detection of radical species (Adams, Herd and

Smith 1989). In addition, we have determined the fraction of

recombinations leading to OH for HCQ2+, which is an observed interstellar

ion (Thaddeus, Cuelin and Linke 1981; Minh, Irvine and Ziurys 1988),

and for N2OH+

I. EXPERUCENTAL

The FALP technique incorporates laser induced fluorescence (LIF) for

the detection of molecular radicals and VUV absorption spectroscopy for

the detection of atoms, and has been discussed in detail recently (Adams,

et al. 1989) and will therefore only be mentioned briefly here.

Ionization is created upstream in a flowing He carrier gas (at a typical

pressure of 1.6 Torr) by a microwave discharge; then Ar and H2 are

sequentially addeu dov-stream to produce a non-recombining H3+/e

afterglow plasma by the reactions

29

2He Ar H2 H2

He-+ H e Ar+ ArH+ H3+

Ar H2 (5)

HeO H2+

Further downstream, H20, CO2 or N20 are added and a recombining plasma is

produced by proton transfer from H3+ e.g.

H3+ + H20 - H3O + + H2 (6)

Any internal excitation of the product ion is rapidly quenched in the

presence of excess reactant neutral by near-resonant proton transfer,

e.g.

H30+ * + H20 - H30+ + H20* (7)

At the initial electron densities (-I X 1010 cm "3) in the recombining

plasma, the ions are efficiently neutralised to give products. Any OH

X2H(v"-O) resulting from the dissociative recombination reactions of

H30+, HCO 2+ and N20H+ with electrons is detected further downstream by

laser excitation of the A2Z(v'-l) - X2a(v'-O) rovibronic transition of

OH and detection of fluorescence in the (I,1) band. An absolute

calibration of the fluorescence intensity against OH X2H(v"-O) number

density was obtained and has been discussed in detail previously (Adams

et al. 1989).

In order to ensure that the OH observed resulted from the

recombination reaction, the OH fluorescence intensity, If. was measured

as a function both of the flow of the added reactant gas (except in the

.ase of H20 for which it is difficult to vary the flow for large flows)

and as a function of the electron density, he(O), Just upstream of the

addition point of the gas. A recombining H30+/electron plasma was also

created by 'ding 02 upstream into a pure He carrier gas and adding H2

80

downstream whence the reaction sequence

02 H2 H2 H2

He+ - 0+ H -+ H20

+ - H30

+ (8)

produced H30+. Also, 02+ is produced in the initial reaction in sequence

(8); however this ion is unreactive with H2 and therefore could not

contribute to OH production, In this case, If was then measured as a

function of H2 flow. The dependence of If on both reactant gas flow rate

and ne(O) confirmed that OH is a product in the recombination of H30+,

HC02+ and N2OH+ with electrons. All measurements were made at 300K.

III. RESULTS AND DISCUSSION

a) OR Products of RecovabInatlon: Comparlson v.ith Theory

Information obtained concerning the neutral products of the

dissociative recombination of H30+, HC02+ and N20H

+ is given in Table 2

together with the recombination coefficients, ae' which were determined

previously at 300K (Smith and Adams 1984; Adams and Smith 1988b). The

recombination of HCO+ with electrons was also studied but an OH product

was not observed even though this product channel is somewhat exothermic

(-0.8 eV). This seems reasonable as it would require a bond to be formed

between the H and 0 atoms which are separated by the central C-atom (in

the FALP, utilization of the reaction H3+ + CO produces HCO+ and not the

higher energy isomer HOC+; Dixon, Komornicki and Kraemer, 1984).

However, the absence of OH as a product did show that there were no

significant extraneous sources of OH in the afterglow plasma.

In Table 2, the percentages of the recombinations which produce OH

are separated into those resulting in the ground and the vibrationally

excited states of the electronic (X2U) ground state. These molecules in

the ground electronic state are not necessarily primary products of the

recombination reactions since, in the recombination of all three ionic

species, the electronically excited A2E state of OH is accessible which

can rapidly radiatively relax to the X2U state (German 1975). It is

worthy of note that the vibrationally excited levels of the A2Z state

(v'>2) can rapidly predissociate to 0 and H (Smith, Elmergreen and Brooks

1974). It should be noted that vibrational excitation in the reaction

products is not important in modelling the chemistry of interstellar gas

clouds since vibrationally excited states relax to the vw-O level of the

X2R state in times much shorter than a collision time with even the most

abundant species, H2 (Langhoff et al. 1983). Thus, only the total

fraction of the recombinations which lead to OH is significant,

independent of its nascent state distribution. However, it is

interesting to note that radiative vibrational relaxation within the X2U

state will result in infrared emissions which might be detect Ae in the

directions towards interstellar gas clouds.

It is important to compare the experimentally determined product

distributions with theoretical predictions in order to assess the

validity of the various theoretical approaches to dissociative

recombination. For the case of H30+ recombination, (65±10)t of the

recombinations result in the production of OH. Herbst (1978),

considering only channels leading to the production of the electronic

ground states of OH and H20 (channels 2(b) and 2(d), and channel 2(a)

respectively in Table 1), concluded that -90% of the recombinations would

result in OH production, the remaining -101 producing HZO. If the

only energetically accessible C2,ctronically excited state of OH, the A2Z

state, had been included in the calculation, then OH production would

presumably be even more strongly favoured compared to H20. Thus Herbst's

theory significantly overestimates the amount of OH production. The

converse applies to the earliest theory of Bates (1986, 1987) which

predicted that H20 production would be very much dominant. This was

substantiated by Millar et al. (1988) who extended the Bates theory by

making quantum chemical calculations of the charge distributions in the

recombining ions and predicted that H and H20 (channel 2(a)) are the only

products. This of course significantly underestimates the OH production.

Hewever, these two different theoretical approaches both rely on the

existence of readily accessible repulsive potential curves to the

predicted products. This has very recently been recognised by Bates

(1989) who argues that there would be the greatest likelihood of

favourable curve crossings if the products of the recombination were both

molecular and radical. As far as the H30+ + e recombination is

concerned, this means that the products OH + H2 would be more likely

than previously thought.

Also, the phase space approach of Herbst requires that the potential

surface of the neutralized ion be fully explored before dissociation into

products. If many favourable curve crossings exist then this is unlikely

to happen and this would influence the predicted product distribution. In

view of this discussion, it appears that the availability of accessible

curve crossings should be taken into account in theoretically predicting

product distributions to be used in the models of molecular synthesis in

interstellar clouds.

For the other two recombination reactions that we report he:e (i e.

HCO2+ + e and N2OK

+ + e), no explicit theoretical predictions have been

made. However, since the structures of HCO 2+ and N2OH+ have been

determined theoretically then the original Bates' theory (1986, 1987) can

be used to make some predictions. The lowest energy form of HCO2+ has

been shown theoretically to be H&O (Frisch, Schaefer and Binkley 1985;

83

Schaefer 1988) and thus the product channels H + CO2 (channel 3(a), see

Table 1) and OH + CO (channel 3(b)) would both be predicted, consistent

with the experimental data for the production of OH (a (34±6)t product,

see Table 2). That the channel giving the two molecular products (one

being a radical, Bates 1989) is not dominant, might imply that some

predissociating A2Z state of OH is generated in the recombination

reaction. In fact, in preliminary emission studies of this recombination

reaction, weak A2E 4 X2H emissions of OH have indeed been observed

indicating that some of the OH is produced in the A2Z state.

Rice, Lee and Schaefer (1986) calculated that the lowest energy form

of N2OH+ is NAOH and thus N2 + OH (channel 4(b), see Table 1) would be

expected as a product channel, again in accordance with the experimental

data for the production of OH (a (31±5)% product, see Table 2), although

again the channel leading to the two molecular products, one which is a

radical (N2 and OH) is not dominant. The other channel (4(f)) predicted

on the basis of these theoretical ideas, N + NOH (or N + NO + H) would

require the breaking of a nitrogen-nitrogen triple bond and is therefore

considered to be unlikely. Similar arguments apply to channels 4(d) and

4(e) (see Table I). Obviously, further experiments are necessary to

determine the contribution to the product distribution due to atomic

products (using VUV absorption and fluorescence spectroscopy) and to

electronically excited products (using emission spectroscopy as mentioned

above). Such experiments are in progress.

b) Interstellar ImplIcatlons: 83O+ + e

The experimental data for this reaction indicate that 65% of the

recombinations lead to OH (see Table 2), the remaining percentage being

distributed in an as yet unknown manner between the other energetically

84

accessible product channels (see Table 1). As stated above these data

were obtained at 300K, however the accessible curve crossings are

unlikely to change significantly with decreasing temperature and thus it

is not unreasonable to apply these results to low temperature

interstellar gas cloud chemistry. Following the ideas of Millar et al.

(1988), it is possible to clarify the situation as to the identity of the

dominant oxygen- containing species in dense interstellar clouds. The

implications of different product ratios of the H30+ recombination

reaction to interstellar chemistry have been discussed in detail by

Sternberg et al. (1987) and by Millar et al. (1988) who considered

possible scenarios for various fractions, f, of the H30+ recombinations

which produce OH. For an f - 0.65 in a cloud with an H2 number density

of 104 cm"3 at a temperature of 50K, a depletion factor for carbon and

oxygen of 0.1, a cosmic ray ionization rate of 0l'7s "I and a total metal

fractional abundance of 1.5 x 10-8 . Sternberg et al. (1987) predicted in

the steady state, the fractional abundances (relative to H2) of H30+, OH,

H20, C+ , CH4, C2H2 , and C3H2 (these are listed in Table 3 together with

the fractional abundances actually observed in various interstellar

clouds). It can be seen that the calculated values for H30+ , OH and CH4 ,

within error, are in agreement with direct observations. However, the

fractional abundance predicted for C3H2 (7.3xlOl) is significantly

smaller than observations. It should be noted that Millar et al. (1988)

predict a fractional abundance of C3H2 in agreement with Sternberg et al.

(1987) using a steady state model (both with f-O.l) but additionally

predict much larger fractional abundances (-10- 7) using an early-time

model. Nevertheless it is gratifying to see that, to a large degree, the

direct observations are consistent with predictions based on the product

distribution experimentally measured at 300K for the H30+ recombination

reaction, as reported in the present study.

85

In the earlier models of molecular synthesis (see Prasad and

Huntress 1980; Craedel, Langer and Frerking 1982; Millar and Freeman

1984; Leung, Herbst and Huebner 1984; Herbst and Leung 1986), an f value

of 0.5 (i.e. close to experimental value of 0.65) was fortuitously

adopted. However, it should be noted that it was generally assumed in

these models that the fraction of the recombinations which do not produce

OH lead to the production of H20. There is no experimental evidence yet

to support this assumption and, indeed, since the A2U excited state of OH

is accessible and predissociates for v'22 (Smith et al. 1974), it is

possible that the production of 0 and H atoms (channel (2c)) could occur.

However, because OH is a dominant product from the recombination of

H30+, the dominant oxygen containing species In dense interstellar gas

clouds should be 02 (formed via the reaction of OH + 0) rather than H20

(Millar et al. 1988).

c) Interstellar Implications- 8C02 + + e

Much less effort has been directed towards the interstellar ion

chemistry of HCO2+ compared to that of H30

+ . The ion HC02 + has been

detected in Sgr.B2 (Thaddeus,Cuelin and Linke 1981) with a fractional

abundance of >10-10 (Minh et al. 1988) which is about two orders of

magnitude greater than expected on the basis of molecular synthesis in

cold quiescent interstellar gas clouds (Herbst and Leung 1986). The

abundances of HC02 + and CO2 are closely linked by the proton transfer

reactions:

H3+ + CO2 CHCO2

+ + H2 (9)

N 2H++ CO2 - HCO2+ + N2 (10)

HC0 2+ + CO ---. HCO+ + CO2 (11)

together with the more minor reactions

8o}

HCO2+ + H20 (NH3. etc) - H3O+ (NH4

+, etc) + CO2 (1i)

(Herbst et al. 1977). The abundances of H3+. N2H

+ and CO are such that

reactions (9) to (II) are the dominant production and loss processes for

HCO 2+ and CO2 and thus effectively cycle CO2 between its protonated and

unprotonated forms with the number density ratio, R, - (HC02+]/[C021

being relatively unaffected by other overall production and loss

processes (see reactions (15) to (19)). By considering that HC02+ and

CO2 are in dynamic equilibrium, then

k9[H 3+ ] + klo[N 2H

+]

R k1l[CO ] + k12[H20 (NH3 etc.)] (13)

The rate coefficients, k9 and kl0 have been measured at 300K (Ikezoe et

al. 1987) and it is assumed that these are appropriate to interstellar

cloud temperatures (a reasonable assumption for proton transfer reactions

which proceed at or close to the collisional rate; note that for

reactions (12), since H20 and NH3 have appreciable permanent dipole

moments, then the collisional rate coefficients at interstellar

temperatures will be appreciably greater than at 30OK; Adams, Smith and

Clary 1985). The rate coefficient for reaction (l) has not been

measured but is confidently expected to be equal to the collisional

value. Using these values of rate coefficients together with the

fractional abundances of the species included in equation (13) as

determined in the dense cloud model of Herbst and Leung (1986), then

R - [HCO+j - 10- 5

[C021 (14)

It is now of interest to consider the overall loss processes for these

coupled species, HC0 2+ anc- 32, and to determine how effective

87

dissociative recombination is as a loss process. Loss processes for

HCO2+ and CO2 (Herbst et al. 1977; Prasad and Huntress 1979) are

hCO2+ + 0 - HCO+ + 02 (15)

HC02+ + N H HCO+ + NO (16)

He+ + CO2 C CO+ (80), C02+(l0%), O+(l0%) (17)

C+ + CO2 - CO+ + CO (18)

H+ + CO2 - HCO + + 0 (19)

as well as those dissociative recombination channels of the HCO2+ + e

reaction (reactions (3b) and (3c) in Table 1) which do not produce CO2

and which comprise at least 34% of the recombinations. Consideration of

these loss processes gives:

d[HC02 + ] + d(C0 21

dt dt - -(kl[OI+kl6[N])(HC02+1 (kl7[He+]+kl8(C+]

+kz9 fH+J)IC02) P ,eIe)IHCO2+] (20)

where p is the fraction of the recombinations which generate products

other than CO2 . The relative importance of these loss processes can be

assessed by utilizing the value of R in equation (14), the fractional

abundances of the various species in equation .(19) as e-termined in the

dense cloud model of Herbst and Leung (1986), and the known rate

coefficients for the reactions. k17 and k18 have been determined to be -

1 x 10- 9 cm3s-I and k19 to be 3xlO "9 cm3 3-1 at 300K (Ikezoe et al. 1987)

and since they aie close to their respective collisional rate

coefficients they are not expected to vary significantly with temperature

(since 1,2 does not have a dipole moment). The rate coefficients, k1 5

and k1 6 have not been measured, however, by consideration of similar

exothermic atom reactions (Viggiano et al. 1980), tlhe rate coefficients

are expected to be a significant c.-action (-20-50% at 300K) of their

L8

collisional values. The recombination coefficient, ae, for the HCO 2 + a

reaction has been measured to be 3.4 x 10" 7cm3s' at 300K (Adams and

Smith 1988b), and in common with the a. for other reactions, is expected

to increase to - I x l0-6cm3s-l at interstellar cloud temperatures (Adams

and Smith 1988a). Application of these data to equation (19) shows that

the HC02+ + e recombination reaction will be a competitive loss process

for HC0 2+ and CO2 for values of the product p.x(e) of approximately 10-8,

where x(e) is the fractional abundance of electrons relative to H2. Thus,

since p 0.35, then x(e) - 3x 10.8, which is within the range of x(e) of

10- 7 to 10-9 commonly used in models of interstellar gas clouds (Prasad

and Huntress 1979; Graedel, Langer and Frerking 1982, Leung, Herbst and

Huebner 1984; Herbst and Leung 1986; Millar et al. 1988). Therefore,

dissociative recombination of HC02+ is expected to be an important loss

process for HCO2+ and thus for CO2 in interstellar clouds. However, more

detailed modelling is obviously required to establish the magnitude of

the contribution of dissociative recombination to the loss of HCO2+/C02.

Finally, when the detailed product distributions of other

recombination reactions have been determined, there could be unforeseen

implications to interstellar molecular synthesis.

ACKNOWLEDGMENTS

We are grateful to the USAF and the SERC for financial support of

this work. We also wish to thank Professor Sir David Bates for providing

us with a preprint of his most recent theoretical paper on dissociative

recombination.

89

TABLE 1

Exothermicities (eV) for Possible Channels of the

Dissociative Recombination Reactions Indicated

Reaction (2) Reaction (3) Reaction (4)

Channel HjO+ + e H&O + e NAOH + e

(a) H + H2 0 + 6.4 H + C02 + 8.0 H + N2 0 + 7.6

(b) OH + H2 + 5.8 OH + CO + 6.9 OH + N2 + 10.4

(c) O + H + H2 + 1.5 O + H + CO + 2.5 O + H + N2 + 6

(d) OH + 2H + 1.3 OH + 2N + 0.6

(e) NH + NO+ 6.5

(f) N + H + NO + 2.7

NOTES. The energetics were calculated for the ground rovibronic

states of the species indicated using data obtained from Rosenstock er

al. (1977), Weast (1985) and McIntosh, Adams and Smith (1988).

90

TABLE 2

Percentage Product Distributions and Rate Coefficients, ae,

Measured at 300K for the Dissociative Recombination

Reactions Indicated

H30+ + e H&O + e N§OH + e

OH(v"-O)+H 2(2H) 46% OH(vo-O)+CO 17%,- OH(vw-0)+N 2(2N) 14%65% 34% 31%

OH(v"'>O)+H 2 (2H) 19% / OH(v*>O)+CO 17%34 OH(v*>O)+N 2 (2N) 17 7

other products 35% other products 66% other products 69%

ae (a), cm3s-

1.0x0"6 3.4x10"7 4.2x10"7

(a)Smith and Adams 1984; Adams and Smith 1988b.

91

TABLE 3

Comparison of Model Predictions (Sternberg et al. 1987) and

Observed Fractional Abundances (Relative to H2)

of Several Interstellar Species

Predicted Observed

Species Fractional Fractional Source Reference

Abundance Abundance

H30 + 1.0X10 .9 -10.9 OC-l,SgrB2 Wootten et al. 1986

OH 5.8x10"8 3x10"7 TMC-I Irvine et al. 1985, 1987

6x10 "8 TC-1 Leung et al. 1984

H20 7.8x0 7 .

C+ 6.4xl0-10

CH4 5.0xlO"8 <8xlO "7 OMC-1 Knacke et al. 1985

C2H2 7.7410 "10

C3H2 7.3x0 1 1 2xlO 8 TMC-E Irvine et al. 1987

>5xl0 "9 a Katthevs and Irvine 1985

5x10 0 SgrB2 Thaddeus et al. 1987

lxl0"9 a Irvine et al. 1987

7x0 11 OMC-1 Thaddeus et al. 1985

NOTES. The predicted fractional abundances are those obtained using

a value of 0.65 for the fraction of H30+ + e recombinations leading

to OH in the chemical model of Sternberg et al. (1987).

92

REFERENCES

Adams, N.G. and Smith, D. 1988a, in RACe Coefficients In

Ascrochemlscry, eds. T.J. Millar and D.A. Williams (Kluver Academic

Publishers, Dordrechc) pp.173-192.

Adams, N.G. and Smith, D. 1988b, Chem. Phys. Leer., 144, 11.

Adams, N.G., Smith, D. and Clary, D.C. 1985, Ap. J. (Letters), 296, L31.

Adams, N.G., Herd, C.R. and Smith, D. 1989, J. Chem. Phys., submitted.

Anicich, V.G., and Huntress, W.T. Jr. 1986, Ap.J. Suppl., 62, 553.

Bates, D.R. 1950, Phys.Rev., 78, 492.

Bates, D.R. 1986, Ap. J. (Letters), 308, L45.

Bates, D.R. 1987, in Modern Applications of Atomic and Molecular

Processes ed. A.E. Kingston (Plenum, London) pp 1-27.

Bates, D.R 1989, Ap. J, preprint.

Dixon, D.A., Komornicki, A. and Kraemer, W.P. 1984, J. Chem. Phys., 81,

3603.

Frisch, N.J., Schaefer, H.F.1I1 and Binkley, J.S. 1985, J. Phys. Chem.,

89, 2192.

German, K.R. 1975, J. Chem. Phys., 63, 5252.

Graedel, T.E., Langer, W.D. and Frerking, N.A. 1982, Ap. J. Suppl., 48,

321.

Green, S. and Herbst, E. 1979, Ap. J. 229, 121.

Guberman, S.L, 1983, in Physics of Ion-Ion and Electron-Ion

Collisions, eds. F. Brouillard and J.W. McGowan (Plenum, New York)

pp. 167.

Herbsc, E. 1978, Ap. J., 222, 508.

Herbst, E., and Leung, C.M. 1986 M.N.R.A.S.,222, 689.

Herbst, E., Green, S., Thaddeus, P. and Klemperer,W. 1977, Ap. J., 215,

503.

Ikezoe, Y., Matsuouka, S., Takebe, M., and Viggiano, A. 1987, Gas Phase

93

Ion Molecule Reactions Rate Constants through 1986. (Maruzen Company

Ltd, Japan).

Irvine, V.1., Goldsmith, P.F., and HJalmarson, A. 1987 in Interstellar

Processes, eds. D.J. Hollenback and H.A. Thronson Jr. (Reidel,

Dordrecht) pp. 561-609.

Irvine, W.1., Schloerb, F.P., Hjalmarson, A., and Herbst, E. 1985, in

Protostars and Protoplanets II (University of Arizona Press, Tucson)

pp. 579-620.

Knacke, R.F., Geballes, T.R., Roll, K.S. and Tokunaga, A.T. 1985,

Ap. J. (Letters), 298, L67.

Kraemer, W.P., and Hazi, A.U. 1988, International Symposium on

Dissociative Recombination, May 29-31, 1988.

Langhoff, S.R., Jaffe, R.L., Lee, J.H., and Dalgarno, A. 1983

Geophys. Res. Lett., 10, 896.

Leung, C.M., Herbst, E., and Huebner, W.F. 1984, Ap. J. Suppl., 56, 231.

Matthevs, H.E., and Irvine, W.M. 1985, Ap. J. (Letters), 298, L61.

McIntosh, B.J., Adams, N.G., and Smith, D. 1988, Chem. Phys. Let., 148,

142.

Michels, H.H. 1989, in Dissociative Recombination: Theory, Experiment

and Applications, eds. J.B.A. Mitchell and S.L. Guberman (World

Scientific Publ. Co., New York).

Michels, H.H., and Hobbs, R.H. 1984, Ap. J. (Letters), 286, L27.

Millar, T.J., and Freeman, A. 1984, M.N.R.A.S., 207, 405.

Millar, T.J., and Nejad, L.A.M. 1985, M.N.R.A.S., 217, 507.

Millar, T.J., Leung, C.M. and Herbst, E. 1987, Astron. Ap., 183, 109.

Millar, T.J., DeFrees, D.J., McLean, A.D., and Herbst, E. , 1988,

Astron. Astrophys. 194, 250.

Minh, Y.C., Irvine, W.M., and Ziurys, L.M. 1988, Ap. J., 334, 175.

Prasad, S.S. and Huntress, W.T. 1979, Ap. J., 228, 123.94

Prasad, S.S. and Huntress, W.T. 1980, Ap. J., 239, 151.

Queffelec, J.L., Rowe, B.R., Morlais, K., Comet, J.C. and Vallee, F.

1985, Planet. Space S¢., 33, 263.

Rice, J.E., Lee, T.J. and Schaefer, H.F.1II 1986, Chem. Phys. Lett.,

130, 333.

Rosenstock, H.K., Draxl, K., Steiner, B.W., and Herron, J.T. 1977,

J. Phys. Chem. Ref. Data, 6, Suppl.l.

Rowe, B.R., Vallee, F., Queffelec, J.L., Comet, J.C. and Morlals, N.

1987, J. Chem. Phys., 88, 845.

Schaefer, H.F. III 1989 in "Ion and Cluster Ion Spectroscopy and

Structure" Ed. J.P. Maler (Elsevier, Amsterdam) pp. 109-130.

Smith, D., and Adams, N.G. 1984, Ap. J. (Letters), 284, L13.

Smith, W.H., Elmergreen, B.G. and Brooks, N.H. 1974, J. Chem. Phys., 61,

2793.

Sternberg, A., Dalgarno, A., and Lepp, S. 1987, Ap. J., 320, 676.

Thaddeus, P., Guelin, K., and Linke, R.A. 1981 Ap. J. (Letters), 246,

L41.

Thaddeus, P., Vrtilek, J.M. and Gottlieb, C.A. 1985, Ap. J. (Letters),

299. 163.

Viggiano, A.A., Howorka, F., Albritton, D.L., Fehsenfeld, F.C., Adams,

N.G., and Smith, D. 1980, Ap. J. 236, 492.

Weast, R.C. (ed.) 1985 CRC Handbook of Chemistry and Physics 66th

Edn. (CRC Press, Boca Raton, Florida) p.

Wootten, A., Boulanger, F., Bogey, M., Combes, F., Encrenaz, P.J.,

Germn, M., and Ziurys, L. 1986, Astr. Ap., 166, L15.

N.G. Adams, C.R. Herd and D. Smith: School of Physics and Space

Research, University of Birmingham, P.O. Box 363, Birmingham B15 2TT,

U.K.95

APPENDIX 7

A SELECTED ION FLOW TUBE STUDY OF THE REACTIONS

OF THE PHn+ IONS (n - 0 TO 4)

WITH SEVERAL MOLECULAR GASES AT 300K

D. SMITH, B.J. McINTOSH AND N.G. ADAMS

J.Chem.Phys., 90, 6213 (1989)

96

A selected Ion flow tube study of the reactions of the PH+ Ions (=0to 4)wfthseveral molecular gases at 300 K

David SmiM, Bruce J. McIntosh, and Nigel G. AdamsSchoo of Plyscs and Space Rearch. Uniwnay of Birmingham P0. Box i63. Birmingham 31J 27.England The U.S. Government Is auth:zrizeo to reproduce and sell this report.

Permission for further reproduction by others must be obtained from

(Received 21 October 1988; accepted 31 January 1989) the COpyright owner.

The reactions of ions in the series PH* (n = 0 to 4) have been studied at 300 K withCH3NH1, PH3, NH3, CH3CCH, H2S. C.H., CHIOH, COS, C2H2, O, H20, CH,, HCN, CO2,CO. and H, using a selected ion flow tube (SIFT) apparatus. The majority are fast binaryreactions, the measured rate coefficients, k, being close to the respective collisional values, andfor reactions involving polyatomic reactant molecules multiple product ions are observed.Most reaction mechanisms are represented in this large number of reactions including chargetransfer, proton transfer, atom abstraction and especially evident is phosphorus atom insertionin which product ions such as H. PN , H. PO , H. PS' and H. C, P are formed. Thereactions of the PH,* ions (n = 0 to 2) with CO2, CO, and H2 proceed by the process ofternary association except for the P reaction with CO, in which PO is formed in a fastbinary reaction. PH,4 reacts only with CHNH and NH 3 by fast proton transfer, a processwhich is always rapid when it is exothermic. Upper limits to the heats of formation of HPO',H2PO *, PC2H7 and HCP have also been deduced from their observation as products of someof the reactions.

I. INTROOUCTION marily concerned with the interstellar implications of the

In recent years. following our development of the select- study but rather with the very interesting fundamental data

ed ion flow tube (SIFT) technique,' we have carried out provided on the kinetics, product ratios, reaction mecha-

detailed studies of the reactions of positive ions in recogniz- nisms, and thermochemistry which this survey of the reac-

able series with a number of molecular species. This work tions of PH: provides

was stimulated by the growing interest in interstellar ionchemistry and the need for kinetic data to substantiate the II. EXPERIMENTALquantitative ion-chemical models that describe the forma- The measurements were carried out in a SIFT apparatustion of the wide variety of molecules that have been detected in helium carrier gas at 300 K. This now standard techniquein interstellar clouds.2, ' Thus we have previously studied the has been described in detail previously, most recently in areactions of ions in the series CH.: (n = 0 to 4)," NH.+ review.' The PH., ions were created in a low pressure clec-(n = 0 to 4), and SH.* (n = 0 to 3), '0 with a variety of tron impact ion source, and after mass filtering, the P+,molecular species many of which are known to exist in inter- PH*, PH,*, PH3*, and PH1 , (the last mentioned beingstellar clouds. Studies of this kind not only provide the criti- produced in the reactions of the primary ions with PH3)cal kinetic data and product ion branching ratios for use in were separately injected into the helium carrier gas at lowthe chemical models but also have revealed, for example, laboratory energies ( < 10 eV) to inhibit collisional dissocia-interesting trends in reactivity with increasing hydrogena- tiou of the ions. Nevertheless a small fraction (up to 10%) oftion of the reactant ions, insights into reaction mechanisms the ions suffered colisional breakup at injection resulting inand thermodynamic data such as proton affinities and heats a small percentage of "unwanted ions" in the flow tube (P 4

of formation of molecules (including radicals) which are from PH* and PH,'; PH' from PH,). It is interesting todifficult to obtain by other means. note that in the breakup of PH2* and PHI* the only signifi-

The discovery of PH 3 in the Jovian atmosphere prompt- cant products were those in which two H atoms (presum-ed Thorne ei al." to study the reactions of ions in the series ably in the form of an H2 molecule) were formed. The slightPH* (n = 0 to 4) with H, CH,, NH,, and PH, and the distortion of the products and product distributions result-reactions of P - and PH" with N., 0,, CO, CO:, H.O, and Ing from the presence of the unwanted ions were accountedHCN using the ion cyclotron resonance (ICR) technique. for in the ion product distributions derived for each reaction.

These studies followed up earlier ICR studies by Holtz Reactant gases were introduced at a controlled fow rate intoet al. " of the reactions of PH, with several ionic species. The the carrier gas/ion swarm and the rate coefficients and thevery recent discovery of PN molecules in the Orion molecu- ion product distributions were determined in the usual man-lar cloud' 3 prompted us to study the reactions of the ions ner.' The reactant vapors H 2O and CHOH were introducedderived from phosphine, PH. (n = 0 to 4), with some 17 as dilute mixtures in helium (2% and 5%, respectively) tomolecular species, most of which are known interstellar spe- inhibit dimenzation and to facilitate accurate flow rate mea-cies, in order to explore the possible routes to the production surements. The HCN was also diluted by helium but to anof PN, and indeed other phosphorus-bearing molecules in uncertain dilution which we ascertained by measuring theinterstellar clouds. Howe-ver, in this paper we are not pn- apparent rate coefficients for the exothermic proton transfer

J Cherm " "(11). 1 une '989 302'.966,89/116213-07$02.10 E 1989 Amencan ins1Wtu Of PhYSICS 6213

97

reactions ofseverad ions (e.g., HCO and N2H*) with the (b

HCN in the mixture, reactions which invariably proceed at -PC:HH (5%)the collisional limiting rate for which the rate coefficient k, in the form ofa semilog plot of the P reactant ion count ratecan be readily calculated using the ADO theory." and the count rates of the product ions PCH - and PC.H."

None of the PH: ions reacted with N2 with a measura- as a function of the flow rate of the reactant gas C.H,. Theble rate coefficient and this offered the opportunity to use N2 slope of the linear P ' decay provides a value for the overallas a relaxant for any vibrational or electronic energy in the rate coefficient for the reaction (I). i.e., k,. In Fig. I (b) thereactant ions, a technique we have used previously to ensure percentages of the product ions are plotted as a function ofthat metastable electronically and vibrationally excited reac- the C2 H2 flow rate to obtain an ion product distribution intant ions in SIFT experiments are relaxed to their ground the normal manner."* (Note that the minority product isstate prior to reaction with the added gases.'" No influence adduct ion PC.H,. which is presumably formed in a ter-was apparent on the product ratios or the rate coefficients for nary collision between P, C2H, and a He atom; we discussselected test reactions, and so we judge that the reactant ions the PCHa" ion later.) During the study of this reaction, andin these experiments are in their groend vibronic states (ap- indeed several other reactions in this series, it was clear thatpropriate to 300 K). the product ion PC2H reacted rapidly with CH, as is ob-

vious in Fig. I (a) from the decrease in the PCH * count rateIlL. RESULTS with increasing CH, flow at the larger flows. The reaction

All of the 16 molecular species included in this study occurring isreact at varying rates with P , PH and PH2', but eight of PC,H - + C:H., -PCH:' + H (2a)the molecular species do not react with PH3*. Since only twoneutrals (NHJ and CHNH2 ) react with PH,, data on the - PCH;. (2b)PH.' reactions are not included in the compilation of ratecoefficients (cm's 'for two-body reactions and cm's-' for Note that an adduct ion (channel (2b) J is formed in this

three-body association reactions) and percentage ion prod- reaction also. The curve of the PC2H ' count rate [PC:H I

uct distributions obtained for the many reactions given in versus CH 2 flow rate (giving rise to a C2H2 number density

Table I. Most reactions are fast binary reactions, the rate (NJ ) can be described by an expression of the form"'coefficients usually being essentially equal to or a significant ak,fraction of k. However, exothermic binary reaction chan- [PC2H ]N - [P Jo (kc _ k,)nels are not available for most molecules in the lower part ofthe table (HCN to H2) and then the much slower (less ef- x [exp( - k, [ N 11) - exp( - k2[N ]) 1,cient) proces of ternary association is observed in severalreactions. ; or these reactions both the effective binary rate wer [P of is th c n rate ofPfpiions zrcoefficient at a helium pressure of 0.48 Torr and the equiva- flow (zero N) of C:H2. k, and k2 are the rate coefficients forlent ternary association rate coefficient are given. The reac- rations (b1t) and (2). a is a coefcient which accounts bothtant molecules are listed in Table I in order of increasing for mass discrimination between P and PCH in theionization energy from CH 3NH 2 (ionization energy is 897 quadrupole detection system and for the nonunity branchingeV as given below the molecule) to H, (ionization energy ratio in reaction (!) and t is the reaction time given by the15.43eV). Also given are the proton affinities above each quotient of the length of the flow tube reaction zone and the

reactant molecule given in units of kcal mol-' as obtained ion swarm flow velocity. Since [P* 1, k, and t are known

from the accepted compilations. "' Similarly. the recombi- then by fitting the [ PC2H I versus YNJ data using an itera-

nation energies are given in eV below each reactant ion and tive nonlinear least-squares procedure," a value for k2 (and

the detachment energy of protons from the ions (which is indeed a) can be obtained. The computer generated fit to the

equivalent to the proton affinity of the PH. - , neutral spe- experimental data points is shown in Fig. I (c) which yields

cies) is given above each ion in kcal mol - ', except, of k2 .9.1 x 10- 0 cm' s-' and a = 0.64. This approach to

course, for P' ions. The proton affinities of P, PH, vnd PH2 rate coefficient determinations is useful when the secondary

were obtained from Ref. I8. Tbe value given for PA (PH,) is ions (inthis case PCH) cannot be formed in anion source

the value obtained experimentally from proton transfer and injected into the SIFT in the usual way or when it ts

studies. " Thus from a glance at Table i. it is clear for which necessary to study the reactivity of a product on of a specific

reactions charge transfer and proton transfer are energeti- ion-molecule reaction.

cally allowed. Also included in Table I are the collisional A similar approach to data reduction can be taken to

rate coefficients for the PH,* reactions (which are essential- determine the rate coefficient for a primary reaction in

ly the same as those for the P , PH , and PH,' reactions) which a primary product or secondary product ion has the

calculated using the ADO theory." same mass/charge ratio as the primary reactant ion Such is

Typical data samples from SIFT experiments have been the cas for the reactions of PH - with CHNH, (secondary

given before in other papers," so here we give only one ex- product ion CHNH 3 , see Table 1) and PH2 with CHOH

ample to illustrate an extension to the usual analysis. In Fig. (secondary product ion CHOH2 * ). In this case the distort-

I (a) the data are presented for the reaction ed (nonlinear) decay plot of the reactant ion count rate isfitted to an appropriate expression" and provides values for

P' + CH,-PCH' + H (95%) (Ia) the primary and secondary (where appropriate) reaction

J. Ch1em Phys. Vol 90. .4o 11. 1 June '989

98

TABLEI R em taand.pe duenbtw othe r P '. PH;. andPf, wtbasies ofmolecuks at )K. The reactant molecules are listed in order oicritang nAto neferlgy and theme

awegass are given below each molecule in eV. above each molecules is givenl its ron affinity in kcal mol '1he proton detachment energies (kcal mo- ') and recombmnation energies (eV) are given above and below thereactant ions. respectively The measured binary rate coefficient in cm' s -' for each reaction is given in theforn 1.3( - 9) (see the P + CH,NH reaction) which means 1.3x 10-'cm's-' Also given are the:colh.smonal rate coefficients k, in cm' s -' as calculated by the Lngevin or ADO theories (Ref 14) as appropriate(k, is essentially the same for all of the ions PH* . = 0 to 4) Ternary association rate coefficients in units ofcm' s - ' obtained at a fxed pressure of 0.43 Ton' are indicated by an astensk. The percentage of each production is given in parentheses aflter the ion. PH,* ions reacted only with CHNHI and NH, (see the text) and sothese data are not included in this table

PA(kcal/mol)1498 1602 1695

k, P PH* PH,' PHJIP(eV) (cm'-) 10486 i0.1s 92 987

214.1 1'J( - 9) 1.3( - 9) 1.5( - 9) 1.7( - 9) 1.9( -9)CHNH, CH:NH;- (68) CH1NH, " (31) CHNH,- (62) CH,NH"8.97 PNH," (32) PNH," (16) CH,NH," (38)

PNH,' (46)

188.6 1 37( -9) 12(-9) 1.3( -9) 1.1(-9) I1( -9)PH, PHi, (26) PH; (18) PH.- (5) PH.9.98 PH"* (62) PH,- (9) P2H"- (171

P:H," (12) P," (12) PH; (78)P:H,' (53)PH," (5)

2040 1,93( -9) Is( -9) 2.1 -9) 2.0( -9) 23( -9)NH, NH,- (52) NH.- (19) NH: (19) NH,10166 PNH," (45) PNH" (23) PNH," (81)

PNH," (53)

182 155( - 9) 1.7( -9) 1.7( - 9) 1.6 -9) 1.6( -9)CHCCH CH -(42) CH," (19) CH," (43) CH; (91110.36 PCH," (52) CH,- (17) PCH (44) PICH" (9)

PCH," (6) PC, " (64) PCH; (7)PCIH." (6)

170.2 144( - 9) 1.41 -9) 1.5( -9) 1.5( -9) 1.0 -9)His HS- (31) HS-(9) HPS- HS"1047 PS" (12) HPS"(64)

HPS" (57) HPS"(27)

162.6 1.24( -9) 1.2( -9) 12( -9) 1.2( -9) 4.7( - 10)C.H. PC-H," PCH (30) PCH;" (12) CH. PH,-10.51 PCH," (70) PCH." (83) Press. adep-

> 4.6( - 26)

1819 1 11( -9) 1.4( -9) 1.9( -9) 2.0( -9) 1.9( -9)CH,OH HPO" HPO" CHOH (65) CHOH"10.85 PCH." (8)

HH1 )0 (27)

151 1.38( -9) 1.1( -9) 1.3( - 9) 9.9( - 10) 4.6( - II)Cos PO (38) HPS" HPS H,PS"11.134 PS"(62)

153 1 12( - 9) 13( - 9) 1,3( - 9) 1.4( - 9) 58( - 10)CH, PC.H-(951 PC.H:" PC.H," PCH,"1141 PCH:" (5)

1009 740( - 10) 56( - 10) 541 - i*n) 87.( - 11) <2( - 13)0. PO PO" PI)12059

165 I 2071 -9) 551 - 10) 12( -9) 4.9( - 10) < 1( - 13)H:O PO" (10) HPO (62) H.PO'(67)12.62 HPO" (90 1 H.PO' (17) HO' (33)

H,O' 21

J Chem PhyS. Vol 90, No 11 I Jun 198999

TAKLE I (.eihmW).

PA( kcal/mol)1498 160.2 1695

k, P PH* PH, PH,'IP(CV) (cm' s') 10.416 10.13 9.82 9.87

132.0 1.15( -9) 9.6( - 10) IM1 -9) 1.1( -9) < 1( - 13)CH. PCH, PCH," (95) PCH."12.62 PCH," (5)

171.4 2.S1( -9) 1.4( - 10) l.7( -9) 1.4( -9) 2.6( - 9)HCN HCN P' H:CN"(95) HXCN" (72) HCN13.59 HCN PH-(5) HCN PH,- (2)

1.2( - 27)0 > 7.3( - 28)1 > 3.4( - 27)

130.9 368( - 10) 4.6( - 10) 86( - 12) 7.5( - 12) < I( - 13)CO. Po" CO, PH - CO, PH,13.769 59( - 28)0 5.l( - 28)0

141.6 3.40( - 10) 3.5( - 13) I 0N - 12) 2.9( - 12) < I( - 13)CO COPH COPH' MPH,14013 2.4( - 29)' 7.01 -29)0 19( - 23)0 <8( - 30)

1013 1.53( - 9) 1.3( --13) 4.3( - 13) IM1 - 12) < I - 13)H, PH,- PH, PH,"15.43 7 5( - 30)0 2.4( - 29)0 6.7( - 29)' <6( - 30)'

rate coefficients and a. The rate coefficients given in Table I for which there are no other exothermic binary channels.for the PH and PH.* reactions given immediately above Most of these reactions are rapid and with often more thanwere obtained using this procedure. one product channel. Clearly, other reaction mechanisms

are dominating proton transfer even when it is energeticallyIV. DISCUSSION AND CONSIDERATIONS OF allowed. In the proton transfer processINDIVIDUAL REACTIONS

The recombination energies of the PH.- ions are rela- PH- (n) + X - P('S 12 ) + XH", (3)tively small and thus charge transfer is energetically allowedin only a few reactions. P can charge transfer only with the the X and XH ' are invariably singlets and so spin inversionfirst five molecules in Table I. PH' with the first three and is required which presumably will tend to inhibit such reac-PH,- and PH,* with only one, CHNH,. Proton transfer is tions, although for the CHNH2 , PH 3, and NH, reactions,energetically favorable in about half of the reactions. Note proton transfer is sufficiently exothermic for the 2D state ofthat even when charge transfer and proton transfer are ener- the P atoms to be accessed and then spin inversion does notgetically allowed they do not always occur. The majority of occur. It is perhaps significant, therefore, that although pro-the reactions in this wide series lead to the production of ions ton transfer does occur in the HCN reaction which is quiteincorporating pl'osphorus bonded to carbon, nitrogen, oxy- exothermic ( > 20 kcal mol '), but for which spin inversiongen or sulfur as appropriate. would be required, the rate coefficient for the reaction is

A. ReactIons of each primari s~ces significantly less than kc (see Table I).It should be noted that there is likely to be an excited 'i

() P " reactions: Even when charge ,ransfer is energeti- state of PH which is only slightly more energetic than thecally allowed (forCHNH.. PH, NH,, CHCCH, H:S) it is 7r ground-state (by analogy with NH- for which there is athe majority channel in only one case (the NH, reaction), r state only - 8o meV above the ground state') and it isalthough the CH:NH:" (with H:) and the C:HI- (with possible that an equilibrium concentration of PH' (4 tr) mayCH) could be the products of dissociative charge transfer in be present in the PH- ion swarm. If this were the case. itthe CH 3NH.. and CHCCH reactions, respectively. In the would not significantly effect the energetics of the PH - reac-large majority of the reactions the P atom is incorporated tions but it could influence reactions because of spin restnc-into the product ions with the release of stable neutral prod- tions. Note that it is not inconceivable that the 9% HS"ucts (e.g., H.,, CO) or atomic radicals (e.g., H, 0). product in the PH -, H.S reaction may be due to the reaction

(i PH .reactions: Although proton transfer is energeti- of PH ' ('n)cally allowed for 10 of the 16 molecules it apparently occi-s (iii) PH reactions: Proton transfer is energetically al-for only five (PH 3, NH,, H.S, H:O, and HCN) and then lowed in nine reacticns but is only observed in fiveonly as a minority channel in all except the HCN reaction, (CHNH:, PH,, NH,, H.O. and HCN) and, again, it is a

J Chem 'y. .Vo, sc 1O '1 1 June'989100

41' only pro~duct (the only exception is the CH CCH reactiohn inwhich an additional minor (9%) channel is obe -ved) at or

1; very close to the collision rate. So there are apparently no10 barriers to proton transfer from PH,* and the production of

the PH, radical and the closed shell (singlet) ions. Note tl.ai2 the observ.-tion that proton transfer occurs to HS but not to

c H20 is consistent with the excepted value for the protonaffinity of PH, of 169.5 kcal mol- '. " In only three other

PC I",reactions, the COS. CIH 2, and CH. reactions, do new___________b,___ chemical bonding arrangements form. In the first of these. S-

o 2 i. 6 a ioo'atom abstraction from the COS results, leaving the stableC2M2 Flow Rate (Torr I %-'I CO neutral and in the second the PCH,1 ion is produced

which we discuss later in Sec. IV B (iii) together with the11 similar ion PC.H, For the reaction between PH,* and

CH. association occurs with a very large rate coefficient, noso variation of the effective binary rate coefficient with pressurei was apparent and so only a lower limit to the ternary associ-

V 6 ation rate coefficient can be given (Table I).- We have noted

40 - in previous studies"2 that when a slightly endlothermaic ( < 5kcal mol -') binary channel is available then association is

20z PM often very rapid and have argued that the occurrence of thisPC ZM? cendothermic step within the excited intermediate complex

0 prolongs the lifetime of the complex and thus facilitates0 2 Is 6 101S collisional association. This so-called 'endlothermic trap-

C H, Flow Rate (Torr iS4 ping" could be occurring here, since the proton transfer

750 between PH,* and C.H, is endlothermic by only about 6 kcal.45 mol'

Unlike P -. PH* ',and PH~,,the radical PH,- ions showlittle propensity to associate with those molecules in which

o no binary channels are available. This is presumably becauseC the outer shell of the central P has a deficiency of only one

, .2 electron and cannot aci as a )one pair acceptor.*x250

B. Reactions of individual reactant molecular species

00 2 1. 6 a 01 Comparisons can be made of the analogous pairs of re-actions, e.g., NH,/PH,, H.S/H,0, COS/CO., etc., and it isC^Flow Rate (Tor r 14,interesting to compare these by inspection of Table 1. How-ever, here we choose to briefly discuss the reactions in the

FIG 1. Sample plots taken from this SIFT study of PH.- reactionts at 30 following groupings although these groupings are artificialK (a) Data obaned fix the reaction o(P - ions viith C.H:, the P primary to a dge:0 H,(t -otiigmlcls ii y

mn(0) aasthe productions PqW (a) and PC.H,' (0) count rates (10 dge i H,(i -otiigmlcls ii ys sampling penod) in semiloganthmic formn are plotted against th CH, drocarbons. (iv) sulphur-containing molecules, (v) oxy-Nlow rate. The rate coefficient for ihe primary reaction is obtained from the gen-containing molecules, and (vi) C0 2, CO, and H.,.slopw of the linear plot (b) The percentages ofthe product ion count rates (j) PH, reactions: Ions involving P-P bonds are readilyare plotted as a funciion of the CH, flow rate and extrapolation to zeroC.H, Rlow provides the ion product disinbution rot the reaction (swe Table formed with the hydrogen deficient ions (P2H:, n<3) be-l) (c) The count rate o( the PC.H' product ion given in (a) is plaitted ing most evident. All of the PH.- (n = 0Oto 3) react furtherseprately asa functiof OC-H, flow rate. The continuous curve through with PH3 (apparently by proton transfer producing PH 4the data points has been fitted according to the procedure outiined in the with very large rate coefficients (as indicated by the fitting oftext anid proviides a value for the rate coefficient for the .,ecodr reactonoIPC,H' with C.H. o9x 10 rm product ion curves as described in Sec. 111. Values of

PAC?..) - 158 kcal mol-' (Ref. 23) and PA(PH 2) = 179kcal mol t ' (Ref. 24) derived from ab initio calculationswould make proton transfer from PH- and P2H to PH3

minority channel in each case except for HCN and the rate exothermic which is what we observe. The reactions of PH.'coefficient is only about half of k, in this case. Again, spin (n = 0 to 4) with PH,3 have been studied also using the ICRinversion is required for proton transfer from the singlet technique"; some differences are evident between these dataPH2' (AO to singlet state molecules producing singlet state and the present SIFT results. Proton transfer channels wereionsand triplet PH ("T.-). not observed in the PH' and PH2- reactions ii the ICR

6iv PH ; reacftons.- Where proton transfer is energeti- experiments. The fCR and SIFT rate coefficients for thescally allowed (in se,.en reactions) it occurs essentially as the reactions are reasoinably consistent.

J1 Gbnr Phys .Vol 90 NO 11. 1June 989101

ga.l %lJ8Wv4E5. Aw"J. Wn4 jIZ..iv IrW, t.unba fla. A go pIclny avaJaaum mi inr acmI~Lm cm run + Lh i r ,as. to pivuiua.cto form P-N bonds is very apparent in these reactions. They PO *, HPO', and H:PO .HPO" and H:P0) are ma)orare all very fast reactions, even those association reactions of product ions in the CH, OH and H:O reactns.HCN which produce HCNP *. HCNPH %. and HCNPH2" (vi) CO., CO. and H. reactions" PH , did not react at .1ions. The ions PNH,* and PNH," are especially evident as measurable rate with any of these three molecules. The reac-product ions (as are the analogous PH 2" and P.H,: ions in tions of PH* and PH2 with CO2 to produce HPO' andthePHreactions),thispresumablybeinganindicatorofthe H2PO , respectively, are exothermic by at least 12 kcalstability of these ions and the strength of the P-N bond. The mol- ' but these channels did not occur, instead these reac-rate coefficients and product distributions obtained for the tions proceeded via relatively slow ion/molecule associ-NH, reactions in the ICR experiment" are similar to the ation. The ternary association rate coefficients increase asSIFT results except that charge transfer between P+ and the polyatomicity of the ion increases for the CO and H:NH, was not observed in the ICR expenment. reactions in accordance with the idea that a greater number

(iii) CH, CCH. C:H,, C.,:,. and CH, reactions. In all of of internal modes are available in which to share the bindingthe reactions with C:H., C:H 2, and CH. incorporation of a P energy thus prolonging the lifetime of the excited complexatom into the product ion resulted with formation of P-C and increasing the probability of collisional stabilization.bonds. This also occurred, but not exclusively, in the In recent experiments, we have shown that these associationCHCCH reactions (see Table1). The ion PCH2*, produced reactions proceed much faster at 80 K in accordance within some of the CH3CCH and CH, reactions, is presumably expectations. This work has some relevance to astrochem-protonated HCP (ab initio calculations indicate protonation istry and this will be discussed in more detail in a subsequentoccurs exclusively onto the C atom"'). PCH 3 and PCH4' paper.(expected to be CHI = PHI*, i.e., P protonated CHI = PH(Ref. 25)J are also efficiently formed in some of these C. Thermochemical limits derived from these studiesCH 3CCH amd CH, reactions. In the reactions with the mol-ecules containing two carbon atoms, the C-C bond is gener- Detailed reaction surveys such as that summarized inally retained with P incorporation. All of these reactions Table I contain a wealth of "hidden" information. The ob-have large rate coefficients, the ions PC2H 2, PC2H 3*, and served production of an ionic species in a reaction in somePC 2H4 being major products. The facile production of cases can place a limit on its heat of formation, its proton

PC2H1 suggests that it has a relatively low heat of forma- affinity, etc., as we have described [see Sec. IV A (iii) I.Thus, we have derived limits for the heats of formation oftion and that it indeed has the predicted cyclic (aromatic) HPO , H2PO . PC2H7, and HCP. These data are given inTable 11 together with the reactions which were exploited to

P derive them.

HC---H

(Ref. 26). The production of the C2H: ions in the V. CONCLUDING REMARKSCH 3CCH reactions presumably results in the elimination of This detailed study of the reactions of PH: ions has notthe stable molecule HCP according to only provided a great deal of kinetic data but also some in-

PH _ , + CH 3CCH- .- 2H + HCP (n = 3 to 5). (4) sight into the nature of these ion-molecule reactions. The

propensity for phosphorus to form bonds with 0, N, C, and SThe ICR results" for the CH. reactions are similar to

the present SIFT results except that a smaller value for the is very evident. Several particular ionic species are obvious asrate coefficient of the PH,* reaction of 5.8X 10_,OcmI s- products of the reactions, notably ions like H. PN,ra otaneffino the H cti. H. PO , H. PS ', and H. C,,, P and from their observationwas obtaned in the aCR study. as products we have been able to derive thermodynamic

facie in these O reactionP d PS b in g articlar, quantities (heats of formation and proton detachment ener-ly evident. The COS reactions proceeded with the elimna- gies) for some. Regrettably, the structures of most of thesetion of CO, thus ions are not known with any certainty, and we can only spec-

PH: + COS-H. PS ± CO (5) TABLE It. Estumatesan dlimi (the hetsofformatoom .H (the ionic

In only the P- reaction was a second channel evident aid neutral speiies indicatied. obtned from cornsdemion of the ,in-tics

( PO' + CS), an observation which we use to place a limit

on the %-H'( PO" ) (e Sec. IV C). Reaction' Deduction() CHOH. 0, and H.O ,eacions. Production of P-O

bondsin these reactionsis facile PO- ions are the exclusive i' .O- HPO' - H %H(HPOs<27kcamolPH"+ HO-H.P0 H -5H,(H:PO'> 1S3 kca ool -

products of the O reactions, although the PH,- reaction, in PH C H--PO' H PHC(HH: ')< , 3kcalimoi -

which H.O is the likely neutral product, has an unusually PH' + CHCCH -C:H, - HCP f?(HCP) < 81kcal tol-6'

small rate coefficient. Since the reaction is exothermic, thispresumably indicates the presence ofan activation bar -.. as 1H' M neutral species from Refs 16 and 27does the absence of any of the three exothermic channels 'JA NAF, Ref 27) etamjes .AHl(HCP) to he 40 '- 15 kca ool

i Cytem PhysVo 90 No 1 11 .une '989

102

Smlc. mcios w Adw w Mw T ST , ol PH. 219 e2g

ulate on these. However, it seems certain that the ion N.G. Adams, D Smith. andi F Paulson. J Chem Pb.7L 2114;1l 9Wi

PC2 H," is probably cyclic and stable. Structural calcula- "D Smih. N G Adams. and W Lrdinger. I Chem Phys ,% . I.

tions for some of the product ions discussed above would be "L. R. Thorne, V G Antcich. and W T Hunirens. (hem Phys L.Ai 91 .

valuable. 162 (1913).If ions such as those met" ned above are formed in "D Holta. J. L. Beauchamp, and J R Eyler. 3 Am ('hem Soc 92. 7WO4

interstellar clouds by similar ion chemistry to that discussed (190)"L. M. Ziurys, Astrophys. J 321. Lil (1987)

here then they could be the precursors or neutral molecules "T Su and M T Bowers. in Gas Phase Ion Chemorsy. edited by M Tsuch as PN which has recently been detected in the Orion Bowers (Academic, New York, 1979). Vol I. p. 13

molecular clouds'" and others such as HCP, PO, PS, and "D Smith, N G Adams. K Giles. and E Herbst. Astron Astrophys. 200.

PCzH which, as yet, have not been detected. We discuss the 191 (igss)."S. G. L--s. I F Liebman. and R D L,-An. I Phys C:hem Ref Data 13.

formation of phosphorus-bearing molecules in interstellar 695 (1984 .

clouds in a subsequent paper. "S G. Lias. J F Liebman. and R. D Levin. Phys C'hem 17. Suppl 8(1988).

" Berkowitz. L. A Curtiss. S T Gibson, J. P Greene. G L Hillhouse.ACKNOWLEDGMENTS and- 1 A. Pople I Chem Phys. 54. 375 (1986).

Our thanks are due to the SERC and the USAF for Glosik. A B Rakshit. N. D Twiddy, N. G Adams. and D Smith. JPhys B It, 3365 (1978)

providing financial support of this work. "P R. Bevington, Data Reduction and Error Analysisfor the Physical So-

ences (McGraw-Hill. New York. 1%9), p. 237

'E. Herbst. D. J DeFres and A D McLean. Astrophys. J 321. 898

'D. Smith and N. 0 Adams, Adv. At Mol. Phys. 24, I (1937). (1987).

:E. Herbst and C. M. Ltung, Mon. Not. R Astron. Soc. 222. 619 (1986). E. E Ferguson. D Smith. and N. G. Adams. J. Chem. Phys 81. 742

'D Smith, Philos Trans- R. Soc. London Set. A 323. 269 (1987); 324. 257 (1984).

(1951) 'M T. Nyugen, Chem Phys. Lett 146. 524 1 1988).

'D Smith and N G Adams. int. I. Mass Spectrom. Ion Phys 23, 123 7'M. T Nugen. Chem. Phys Lett 135. 73 (1987)

(1977) :'L. L. Lohr. H B Schiegel. and K Morokuma. J Phys C'hem 31. 1981

'D Smith and N G Adams. Chem. Phys. Lett. 47. 145 (1977). (1984).

"N. G. Adams and D Smith, Chem. Phys Le t. 47. 383 (1977). 'J R Bews and C Glideweil, J Mol Struct. 104. 105 (1983)

'D Smith "rd N G Adams. Chem. Phys. Lett 54, 530 (1978) "'D R Stull and H Prophet. Nail Stand Ref Data Ser Nail Bur Stand

'N G Adams and D Smith. Chem. Phys. Lett 79. 563 (1981). 37 1971).

103

APPENDIX 8

REACTIONS OF Kr+, Kr2 +, Xe+ AND Xe24+ IONS

WITH SEVERAL MOLECULAR GASES AT 300K

K. GILES, N.G. ADAMS AND D. SMITH

J.Phys.B., 22, 873 (1989)

104

J Phys B: At. Mol Opt Phys 22 (19591 913-883 PTinted in the UK

Reactions of Kr , Kr*, Xe and Xe* ions with several moleculargases at 300 K

Kevin Giles, Nigel G Adams and David SmithSchool of Physics and Space Research. University of Birmingham. PO Box 363, BirminghamBI5 2TT, UK

Received 30 .ugust 1988

%bart". The reactions of the atomic tons Kr" and Xe" and the molecular tons Kr: andXe; have been studied with several molecular gases at 300 K using a selected ion nowlubei sF-T). In some of the atomic ion reactions differential reactivity was observed betweenthe spin-orbit states (-P)/ 2 . "P1,,) ofthe ions and, when this was so. the lower energy 'P, Istate of both ions was observed to react more rapidly than the 'P,,. state, in accordancewith the findings of previous work. Whereas charge transfer was the only reaction mechan-sin observed for the atomic ions, both charge transfer and inert gas/reactant molecule

switching were observed in several reactions of the molecular ions, the latter processproducing ions such as KrCH. and Xe('H*. By cotsidenng the kinetics of the near-thermoneutral charge transfer reactions of Kr winth N0 and Xe. with COS. the dissoci-ation energies of the molecular ions have been determined to be V. 15eV for Kr:" and

1l 05 eV for Xe'. both these upper-limit values being in good agreement with previousexperimental and theoretical values.

I. latroductloe

The reactions of the rare gas singly charged atomic ions He', Ne* and Ar have beenstudied extensively at thermal energies (see the compilation of rate coefficients andionic reaction products given by [kezoe et al (1987)). The large recombination energiesof these ions ensure that charge transfer reactions can occur with almost any molecularspecies, and in most cases dissociative charge transfer ts allowed also. The groundelectronic states of both Ne' and Ar" are split due to spin-orbit coupling; the differencein the energies of these states (2P,,2 (lower state) and 'P,, 2 ) is very small for Ne' (i.e.0.097 eV) and differently reacting states of Ne' have not been observed. The differencein the energies of the Ar.( 2P31 ) and A*(PI/) ions is also small (0.178 eV) but in thiscase a difference in the reactivity of the spin-orbit states of Ar" has been observed inexperiments carried out at suprathermal energies utilising TESICO (Tanaka et al 1981),rEpico (Guyon et al 1986) and crossed beam (Liao ef al 1986) techniques. In additionthere exists some evidence from one thermal energy flow tube experment for differentlyreacting forms of At* ions.

The spin-orbit splittings of the ground electronic states of Kr' and Xe * are relativelylarge at 0.666 eV and 1.306 eV respectively and the recombination energies are relativelysmall (Kr"(2P,2)= 13.999 eV, Xe*(Pj 2 ) = 12.130eV), and so differences in the reac-tivity of the spin-orbit states for both ions are more likely. A study of the reactionsof Krt and Xe' with several molecular gases at 300 K in a selected ion flow tube (siFT)

09-53-4075/89/ 0O73 - I IS02.50 e 1989 lOP Publishing Ltd 873

105

874 K Giles, N G Adams and D Smith

some years ago showed that the ) and ! states do indeed react at very different rateswith some gases, and the rate coefficients and the ionic products of the separatespin-orbit states were determined (Adams eta 1980). Charge transfer, as expectedfor rare gas ions, was the only process observed except for the reaction with H? forwhich charge transfer is quite clearly endothermic for both states of Kr" and Xe'.However, reaction did occur between both Kr' states and H2 to produce KrH" ions.The ionisation energies of CH( 12.62 eV) and N:O( 12.89 eV) lie between the recombi-nation energies of the Xe'( 2 P3,,) and Xe'(P,,:) ions and so charge transfer with thelower energy 2 state ion cannot occur; charge transfer was observed between theXe'(P,,2 ) ions and CH. and N2O. The most remarkable result of these experimentswas that when a difference in the reactivity of the spin-orbit states of either Kr' orXc' was observed then invariably the lower-energy 2P3/2 ions reacted more rapidly(the rate coefficients and ionic products of some of these reactions are listed in table1). This work by Adams etal (1980) was followed up by Jones eta (1982) whoexplained the difference in the reactivities of the spin-orbit states of Kr' by consideringthe correlation between reactant and product states, the approach taken to explainwhy the reactivity of the 'P,,2 states of some neutr*4 species, including Br and I atoms(isoelectronic with Kr* and Xe* ions), is often greater than the 2P,/ 2 states (see therecent review by Dagdigian and Campbell (1987)).

The reactivities of the singly charged molecular ions He, Ne. and At; have beenextensively studied and these, like their singly charged atomic counterparts, also arevery reactive (see, for example, the papers by Bohme etal (1970), Collins and Lee(1980) and the papers cited in the compilation by lkezoe etal (1987)). The recombina-tion energies of these ions differ from those of their ground state atomic analogues bythe dissociation energies, Do, of the molecular ions (Do(He;) = 2.45 eV (Khan andJordan 1986), DO(Ne,) = 1.31 ev, Do(Ar.) = 1.30 eV (Michels et al 1978)) and are quitelarge. Charge transfer between these ions and most gases is therefore facile. Therecombination energies of Kr" and Xe; are lower (about 13 eV and I I eV respectively,see 3.2) and so they are expected to react with fewer species, but to our knowledgeno studies of the reactivities of these molecular ions have been carried out.

We report in this paper the results of a study of the reactions of KjrO'P31 2, lPu2),Xe(p 3O 2, 'P1,.), Kr,(A :1":) and Xe'(A ',:) with several molecular gases at 300 Kusing the sFr technique. The atomic ion studies extend those of Adams et al (1980)to involve several more reactant gases. Charge transfer, as expected, commonly occursbut, additionally, the molecular ion reactions reveal that rare gas atom/reactantmolecule exchange (switching) occurs in several reactions. From the results of thisstudy, the recombination energies of the ground states of the Kr,' and Xe" ions andtheir dissociation energies have been determined and arc compared with previousexperimental and theoretical determinations.

2. Experimental details

The stFT technique has been described previously, most recently in a detailed reviewpaper (Smith and Adams 1987) Briefly, ions are generated in an ion source, selectedaccording to their mass-to-charge ratio using a quadrupole mass filter and injectedinto helium carrier gas (at a pressure of typically 0.5 Torr) which is constrained toflow along a stainless steel tube by the action of a large mechanical pump. Reactantgases are introduced into the carrier gas/ion swarm and the reactant ions and the

1C6

Reactions of Kr and Xe tons with molecular gases 875

Tsbl I. A compilation of the rate coefficiens (in cm' s-'. e g. for the Kr'(:P,, 2) reactionwith NH,. the rate coefficient is given as 7 0( -10 which is 7 Ox 10o 'cm' s") and theproduct ton distibutions (percentages given in parentheses after the particular production. unless only a single prolact is .A scr ei (or the react.ons of both spin-ort'it statesof Kr" and Xe* and the reactions of Kr," and Xc. with 13 molecular gases, determinedin a SiFT apparatus at 300 K The recombination energies of the :on species and theionisation energies of the reactant molecules in electron volts are given below each species.Also indicated, in square brackets, are the 300 K collisional rate coefficients for eachreaction for references see tet) For the sake of completeness, the previous data due toAdams rt al( 19801 for the reactions of Krt and Xe" ,,ith NH,. H.S, COS. 02, CH_, N.0.CO. and CO are also included '*here no reaction is observed the rate coefficient isindicated as < I( - 121 cmi s -

Kr' Kr: Xe" Xe:

Reactant "P, :P,

molecule 1400 1467 12 97 12.13 13 44 11,11

N HNH NH; NH; NH;10.17 '0(-101 49(-10) 1.6i-91 83(-01 13(-10) 14(-91

[171(-911 (1.64(-9)] (1.66(-9)) [162(-91]

HIS HS" l3S KrH'20I H.S" H.S" H:S"

1047 KrH"35) S'(45)S' 130) SH'(351

1.0(-91 1,0(-91 1.2( -101 99(-10) 99(-101 1.0(-91[ 19(-91] [1.11 -9)] ( 113 (-9)] (1 01 (-9)]

C:H. CH (45) C:H. C. H "(251 C:H41051 CH; 45) C:H 7)

C.H; 101" 41- 10) "7A t -10) 791-10) 851-10) 9.5(-10) 97(-101

(105 -91] [986(-101] [1.01 (-91] [9 61 (- 10)]

CjH, C,; H(5 C:H ;301 CH 301 C)H;1095 C1:H (15) CH; 1251 CH, (65)

C.H; 160) C)H (51 C 3H; 51C 1 I1 CH1 30)CH (iOM C,H; (101

93(-101 9.3 (-10) &.1(-10) 311-101 8.1(-101 8.6(-101[1.(9 -91] [994 (- 101) (1.02(-91) [9.57 ( - 10))

COS COS' COS' COS' Xe COS* 135)11.13 COS' 115)

4.3(-101 96(-11 3.51-I0) 8.7(-101 -2.0(-Ill 6.51-101(I (Y7 (-91] (9 57 -101] [9 8( -101] [9 12 1 -101]

C.H, - (.H- C.H: XeCH1141 <11-1I1 <1 -121 921-101 501-101 311-11 89 -101

(9591 -101] [9.o i - lOl [9 ISi -101] [11'9 1 - 1]

CH, C:M; 65) C.H; 1351 C:H; 153) XeC:H;11 51 C:H. 1351 C M, 15) .HZ 1 10)

CH. (601 C.H; 135)981-l01 931-10) 9.01-101 921(-101 921-101 6.81-101

[1 05,-91] [9.'9 -10l I 001-911 (9 521-101]

0 o: - o:1206 471-11 <1(-12) 4.3( -101 1.0(-10) 1.1 -101 < 1(-12)

(6 15 -10)] [5.71 ( -10)] [5.3 1 - 01] (5.55 (-10)]

107

876 K Gilej, N 0 Adams and D Smith

Tabile 1. (continued)

Kr* Kr Xe'Xc

Reactant ,12 P/ )1 2P.

molecule 14.00 14.67 12.87 12.13 13.44 11 11

CH4 CH: CH; (90) KiC H : 40) - CH:-12.62 CH: 0 CH: (60)

1.0(-9) 1.0(-9) 7 1(-101 <. I (-121) 90(-101 < 1 -112)(1.03(-9)) (958) -10)] (1-001 -9)) [9.72 1-10)]

HCl HCI' KrHCI* (90) - HCI* XeHCI12.74 l4C 10

3.44-Ill 3.3(-12) 10(-9) < 1 -12) 4.04-11 3 1(-11)(I M3-9)] (.04 (-9)] (1.07 1-9)) [1 01 (-9)1

N'O N.O* KrN 2O* 440) -4 N0-12.89 NzO* (60)

4.0(-10) 1.5 (-111 6.2(-10) < 1 -12) 6.3 (-10) <1 (-121(7 61 f- 10)) (6.93 -10)1 (701(-10)1 (6674 -10)]

Co, co:* - -

13.77 6.34-10) 6.3 (- 10) < 1 (-121 < I (- 12) < 1 (-12) < I I -12)[7.01 (- 10)) [635 k - 10)] (6.57 ( -10)] [6 14(-10)]

CO CO* KrCO, - XeCO,14.01 2.01-10) M5-11) 3.8(10)O <1 (-12) <1H-112) 1.9 (-12)

(7,14 1-10)] [6.6841 - Ml) (6-57(- 10)) [6.511(-10))

product ions of reaction are monitored by a downstreamn quadrupole mass filter/detec-tion System. Pale coefficients and ion product distributions (branching ratios) appropri-ate to the temperature of the reaction (the carrier gas temperature) are derived byrelating the deteced ion count rates to the flow rate of the reactant gas. All theexperiments reported here were carried out at 300 K.

The atomic ions Kr' and Xe' were created from the rare gases in both low-presureand high-pressure electron impact ion sources. In the low pressure source the ionssuffer no collisions with their parent gases, whereas in the high-pressure source theymay undergo many collisions depending on the pressure of the rare gas- The ratiosof the currents of ions in both the 'P 3, and 2P,12 states that emerged fram thelow-pressure ion source were observed to be close to that expected on the basis of thestatistical weights of the states ((2J + 1). i.e. 2:)1 in favour of the 'P, 2 state), whereasgreater fractions of the lower-energy ZP5.2 States emerged rom the high-pressure sourceas the result of quenching of the high-energy state in collisions with the parent gasatoms. The fraction of each state in the carrier gas is evident from their differingreactivity with some reactant gases, an example or this is given in figure I for thereaction of XeW(,P,, P, 2) with C2,H:. A straightforward analysis of the data providesthe fraction of the ions in each state and the separate rate coefficients for the reactions(for further discussion of this see Adams et al11980) and Adams and Smith (1988)).

The molecular ions IKr, and Xe7 necessarily have to be generated in the high-pressure source. They are generated in both binary collisions of Krx and Xe with excitedparent atoms (associative ionisation, e.g., Kr + Kr - K, + e (Samson and Cairns1966)) and in ternary collisions between the atomic ions and ground state atoms (e.g.Kr+ 2Kr -Kr,' +Kr (Smith et al 1972)). No convincing evidence for more than one

108

Reactions of Kr and Xe ions with molecular gases 877

101

Figure 1. A semiloganthmic plot of the count rate of Xc*(P,., P,,2 t) ons versus the flowrate of C2H2 reactant gas obtained in the siry study (open circles) The slope of the fineat cte larger flow rtes provides a value for the rate coefficient for the reaction of Xe'(2 P,,2iwi CH,. The linear plot (full circles) is derived by subtracing the count rates indicatedby the dashed extrapolated line from the Xet(2P,,,. "P,,2 ) count rate at the same Sow rateof CIH. and the slope of the line provides a value for the rate cooefficient for tise reactionof XeY ,P,,, with C,. A bi-exponenitial fitting proceduare using a microcomputer wasused also to determnine the separait rate coefficients with identical results.

reacting species of either K.4' or Xe* was obtained. In several of the reactions of thesemolecular ions, parallel charge transfer and rare gas atom/ reactant molecule switchingwere observed (generating ions such as, for example, KrHCI' from the Krz + HCIreaction). To determine the branching ratio into the charge transfer and switchingchannels it was necessary to account for the several isotopic forms of the product ionscontaining the rare gas atoms. The measured rate coefficients are considered to beaccurate to r20%.

3. Remults

In the previous study (Adams et al 1980) the rate coefficients and product ions for thereactions Of K!" and Xe* with NH,, H2S, COS, 02. CH., N,. C02 and CO weredetermined. In some of these reactions the differing reactivities Of the 2P3 2 and *P112states were very obvious; in others the rate coefficients were indistinguishable at thecollisional limiting (maximum) value of the rate coefficient, k, (calculated using wellknown theory (Su and Bowers 1979)). We have now extended this study of atomicion reactions to include the hydrocarbon reactant gases C.2H,, C:,H,. C.2 H, and C3HSand also HCI, and the reactions of the molecular ions Kr,: and Xc, have been studiedwith these gases and those mentioned above, amounting to 13 gases in all. The ratecoefficients and ion product distributions for both the atomic and molecular ionreactions are given in table 1, together with the previous data for comparison. Thelayout of the table involves some simple logic. The reactant molecules are ordered

109

818 K Giles, N G Adams and D Smith

according to their ionisation energies (given below each molecule in eV) from thesmallest, NH,, to the largest, CO. The accurately known recombination energies forthe 2P3/2 and 2P1, 2 states of each atomic ion and those for the ground state of themolecular ions are also iiven, the latter being more precisely defined by the presentstudies (see 93.2). The organisation of the table places the most exothermic chargetransfer reactions (for ground state products) in the top left-hand comer and the leastexothermic (and some endothermic) charge transfer reactions at the bottom right-handcomer. The thermicities for all the individual reaction channels have been obtainedusing thermochemical data from Rosenstock etal (1977). It has been assumed incalculating the reaction energetics that when two hydrogen atoms are released in thehydrocarbon reactions they appear as an H2 molecule (three H atoms appear as H,and H) thus releasing the H-H bond energy. If this were not the case then all of suchreaction channels observed to proceed would have been very endothermic, and so therate data together with the thermicities is proof that H 2 formation does indeed occurin every case. The thermicity of each reaction is not included in the table to avoidmaking it too complicated. The thermicities for those very interesting reactions thatare close to thermoneutral and which provide data on the recombination energies ofKr; and Xe; are given in 93.2. where individual reactions are discussed.

3. 1. Reactions of Kr * and Xe ' with C2H,, C 2H#, C 2H2, C2H, and HC

The reactions of Kr' and Xe* with the other molecules included in the table havebeen discussed previously (Adams et al 1980) and so they are not discussed here. Inthose studies, the product ions of the reactions of the separate spin-orbit states weredetermined using filter gases to selectively remove either the J or the I state ions fromthe flow tube. This procedure inevitably introduces another reactive ionic species intothe flow tube (e.g. when N20 is used to 'filter out' Xe*(2P,,7) from a mixture of bothspin-orbit state ions, N20' is produced) which can produce other ions in reactionwith the reactant gas and confuse the determination of product ion distributions. Thiswas not too severe a problem for the 'simpler molecules* included in the Adams e al(1980) study, but was too severe for the hydrocarbon molecules included in this study,where, with the exception of CH,, charge transfer and dissociative charge transferwere observed to occur, producing several product ions. So the product ions resultingfrom the separate reactions of each spin-orbit state could not be identified with muchcertainty. To circumvent this problem, two procedures were attempted with limitedsuccess. They were: (i) to add a small amount of filter gas (e.g. N2O) to the high-pressureion source containing the rare gas in order to destroy one state of the atomic ion andthus, hopefully, to introduce only one state into the flow tube (whilst this does reducethe fraction of the more reactive ion it is not sufficiently effective to allow unambiguousdetermination of product ions of me spin-orbit state only); (ii) to obtain the atomicions by injecting the molecular ions (formed in the high pressure ion source) at themaximum lab. energy (50 eV) into the helium camer gas in the flow tube, thus causingdissociation. Both the Kr' and Xe" ions formed in this way were in their lowest-energy

states as shown by measuring the rate coefficients for their reactions with appropriategases (e.g. NO, COS, see table II. Fragmentation to the ) state is to be expected ifthe molecular ions are in their ground states or low-lying excited states (Dehmer andDehmer 1978) and, indeed, this observation is taken as proof that the K.4 and Xe;ions extracted from the high-pressure ion source are predominantly in their groundstates. Unfortunately, only a small fraction (-) of the molecular ions could be

110

Reactions of Kr and Xe ions with molecular gases 879

dissociated in this manner and thus a mixture of atomic and molecular ions was presentin the flow tube to confuse the product ion distributions.

So, no attempt has been made in these studies to associate particular ion productsof the reactions of Kr and Xe* with particular spin-orbit states, except that in somecases it is clear that particular ion products can only be generated by the ! state becauseof energy constraints. For example, the reactions of Xe"(P 2, 2P,, ) with C ,H. proceedto the following products:

Xe'( 2 P12 . "P,) +CH, - C2 H; + H, + Xe (- 1.09, 0.22) (1a)

-. C2H. + Xe (1.62, 2.93). (lb)

The numbers in brackets are the thermicities of the reactions in electron volts (eV) forthe reactions of the 'P,, 2 and :P , , states of the Xe" respectively, the negative numberindicating an endothermic channel. Thus, C H; can only generated from the Xe('P,1 )ions, whereas C.,H* could be generated by both states of the Xe>. All three productsof the Kr* reactions with C2H, can be produced by both spin-orbit states of the Kr.Note that the rate coefficients are not discemably different for the reactions of theindividual spin states of Kr* and Xe with C2H,, CjH# and CzH,. All the reactionsare efficient i.e. the rate coefficients are large fractions of the respective collisional ratecoefficients (given in square brackets below the measured values in table 1) and multipleproducts are observed. For the reaction:

Xe'('P,,2 , 'P.,) + CH.-. CH; + H2+ Xe (.10, 2.41) (2a)

- iso-C 3H; + H + Xe (0.51, 1.82) (2b)

-CH; + Xe (1. 18, 2.49) (20)

the major product (65%) is CIH,. However, production of n-C,H; (i.e. CH 3CH 2CH;)from Xe'('P), 2) (which represents about 66% of the total Xe" signal in the flow tube)is endothermic by 0.20 eV. Thus at least half of the C3H, product ions must originatefrom Xe*( 2 P3,2 ) and are presumably iso-C3H; (i.e. CH)CH'CH 3), the production ofwhich from the Xe'( 2P,,) ions is exothermic by 0.51 eV. Also, due to energeticconstraints, the small CzH product of the Xe" reaction with CzH, must be generatedfrom the Xe'('P.,) ions.

The most unexpected result of these reactions involving hydrocarbon molecules isthe unreactivity of C2H2 with both states of Kr* ions. This result was checked anddouble-checked and is an extremely unusual result for the reactions of energetic ionswith polyatomic molecules, which are normally very efficient. This may be due tounfavourable Franck-Condon factors connecting CH, with the accessible (highlyvibrationally excited) states of C:HT ions. This result is especially surprising sinceK. rapidly charge transfers with C2H, (see table 1) and so also does Xe'(P 3,2 ). Itis perhaps significant that amongst the hydrocarbon molecules included in this study,CH1 is the only one that reacts at different rates with the two states of Xe. Again,as before, it is the ' state which reacts more rapidly whenever differential reactivity isobsened and when both charge transfer channels are exothermic:

Xe'('P3i 2, 'Pt, 2)+C.H 2 - CH, + Xe(0.72, 2.03). (3)

Differential reactivity is also observed for the reactions of both K and Xe" withHCI. Charge transfer between both states of Ksr ions and HCI is exothermic, and

111

880 K Giles, N G Adams and D Smith

again the I state reacts more rapidly (see table 1). However, charge transfer isexothermic only for the reaction of the I state of Xe" with HCI,

XeP(2PJ/Z, :P1 2) + HCI -. HCI" + Xe( -0.61,0.70) (4)

and only this state is observed to react at a significant rate.

3.2. Reactions of Kr' and Xe;

The smaller recombination energies of these molecular ions relative to their atomiccounterparts renders charge transfer endothermic with several of the molecules includedin this study (this is especially so for Xe,). However, when charge transfer is energeti-cally allowed it occurs efficiently, usually producing only the parent ion of the reactantmolecule although there are obvious exceptions to this in which dissociative chargetransfer occurs leading to multiple products (e.g. the reactions with C3H, and C2H.-seetable I). Of much greater interest is the observation that when charge transfer isendothermic or near-thermoneutral, product ions comprising the reactant moleculeand an inert gas atom appear. These atom/reactant molecule exchange or switching"reactions are exemplified by

Xe2" +C2H-' XeCH" + Xe (5)

in which the single product ion XeC 2H: is formed. Charge transfer is endothermicby 0.3 eV in this case. A single switching product is also observed for the reactions ofXe2* with C2H, and HCI which are also endothermic to charge transfer. However, inthe reactions of Kr.4 with CH, and HCI charge transfer is exothermic by 0.25 eV and0. 13 eV respectively and in these reactions the charge transfer and switching productsare both observed (see table 1). In two reactions charge transfer is apparently slightlyendothermic (as determined from the available thermochemical data) yet the chargetransfer products are significant parts of the product distributions (along with theswitching products), these being:

K4r; + N20 NO" + 2K: (60%) (6a)

-KrNO + Kr (40%) (6b)

and the reaction:

Xe. +COS -COS' +2Xe, (15%) (7a)

-. XeCOS" + Xe (85%). (7b)

Note that we assume that the least endoihermic neutral products, i.e. K2 and Xe,.are formed in reactions (6a) and (7a) respectively, and this has implications for thedissociation energies derived below for Kr: and Xe*.

The rate coefficients of the charge transfer reactions 16a) and (7a) obtained as thepercentages of the total rate coefficient in each case are 3.7 x 10- _cm3s '' and 9.8 x10" cm3s- ' respectively. These rate coefficients can be used to define a lower-limitto the recombination energies, R,, of the K.r and Xe, ions on the assumption thatthe measured rate coefficients, ki 300), are lower than the respective collisional values,k,(300), only by virtue of the endothermicities of the reactions. Thus the applicationof the Boltzmann relationship, k(300) =kc(300) exp(-.E/300k.), where k. is theBoltzmann constant, provides an upper-limit value for the endothermicity, AE. Thenthe R. of the reactant ion K, or Xe. here) is smaller than the ionisation energy, I,,

112

Reactions of Kr and Xe ions with molecular gases 881

of the reactant molecule by an amount AE. Thus for reaction (6a), AE takes a maximumvalue of 0.022 eV and the R,(K l) ;P (,(NO) -0.022) = 12.872 eV (using 12.89(4) eVas the accurate value for i,(N:O) (Rosenstock etal 1977)). Using this lower.limitvalue of R.(Kr2*) an upper limit to the dissociation energy, D,(K4;), of the K.r'-K.r

r ion can be obtained. Thus D(K4;) is given by R,(K.r'( 2P,,2))+ Do(K.r2)- R,(K.4),i.e. (13.999+0.016-12.872) =1.143eV or, more correctly, Do(Kr 2')r 1.15eV. Thevalue for Do(Kr 2), and also that for D0(Xe 2) (see below), were obtained from Dehmerand Dehmer (1978). This compares favourably with the value of 1. 15 ± 0.02 eV deducedby Ng et al (1977) from their studies of the photoionisation threshold of Kr 2 neutraldimers (their similar work on Xe 2 dimers is referred to below). Combining these twoexperimental determinations, a most likely value for Do(Kr2) is the value indicatedby the present studies as an upper-limit value, i.e. 1.15 eV. These experimental valuestogether with the available theoretically determined Do, due to Michels et al (1978)and Christiansen et al (1981), are given in table 2.

Using the kinetic data obtained for reaction (7a), SE takes a maximum value of0.064 eV and thus, as before, R,(Xe,) (/,(COS) - 0.064) = 11.110 eV (/,(COS) is takento be 11.17(4)eV for this calculation, i.e. the lower limit of the reported value,11.184±0.01 eV (Rosenstock etal 1977), to enable an upper limit to Do(Xe*) to bedetermined). This value is in agreement with the lower limit of the appearance potentialof Xe2, determined to be Il.10eV by Laporte etat (1983). Then the upper limit toD(Xe,'), the dissociation energy of Xe*-Xe, is given by R.(Xe'( 2P,,))+ Do(Xe,)-R,(Xe2),i.e. (12.130+0.023 - 11.110) = 1.043 eV; so Do(Xej) * 1.05 eV. Thisis accept.able with the value derived by Ng et al (1976) of 1.03 ± 0.01 eV given in table 2. Againas for our value of Do(K'4), our value for DOIXe ) is somewhat smaller than thetheoretical value of Michels etat (1978). So our judgement is that D0(Xe;) is nearlyequal to 1.05 eV.

It is interesting to note that the reaction

Xe., +C5H, -. C'H + 2Xe/Xe 2 (8)

is only exothermic by 0.16 eV yet the measured rate cofficient is close to the collisional

Table 2. The bond dissociation energies of round state Kr. and Xe¢ ions in electronvolts derived from experimental studies and from theory. Those derived from the presentwork (a are considered to be strict upper-limit values for both jons. Other sources are asfollows.

.4 eNial 11977): phoboionisation of Kr, molecules.Pratt and Dehmer 1912): phooonsation of Kr: molecules.

'Abouaf eval (1973): phocodissociaton of Kr. tons.N er a/ (1976): photoionisation of Xe2 molecules.

'Michels e at (1971)'Christiansen tial 11981).

Kr: Xe:

Experiment Theory Experiment Theory

wIt.5 v'I W 1 05" 1,06,1 15t±002' I13 1.03 z 0.01' 1050I 50 2 0.004'1.176 0.0-2

113

882 K Gules, N G Adams and D Smith

value (see table 1) giving some credence to our assumption that for IE = 0, themeasured k can approach k, for many ion-molecule reactions.

4. Conluding remarks

The present studies of the reactions of Kr and Xe* follow the pattern of previousstudies in that where differential reactivity of the 2P,,2 and 2P,/ spin-orbit states ofthese ions is observed, it is the lower-energy 2P3/ 2 State which reacts more rapidly. Asimilar result has previously been obtained for the reactions of the spin-orbit statesof the isoclectronic halogen atoms.

Our comprehensive studies of the reactions of Kr* and Xe2* ions is the first reported.Whereas only charge transfer is observed in the reactions of the atomic ions, bothcharge transfer and inert gas atom/reactant molecule switching is observed in themolecular ion reactions generating ions such as KrCH4 and XeC 2H2. Secondaryreactions occur between the latter ions and the reactant gas generating dimer ions suchas C21-2 . C21-2. These interesting reactions will be discussed in a further publication.

Making reasonable assumptions regarding the kinetics of the near-therinoneutralcharge transfer reactions of 1(4* with N20 and Xe* with COS we have determinedupper-limit values of the dissociation energies of these inert gas molecular ions. Thevalues are in excellent agreement with the values derived from photoionisation studiesof the neutral dimer molecules Kr2 and Xe2 and close to the derived theoretical values.

These studies demonstrate again that studies of ion-molecule reactions at thermalenergies can provide accurate thermochemical data on ions (see the discussion of thisin the paper by Smith et at (1981)). These studies of 1(4* and Xe;* reactions may alsobe of interest to those researching into the processes occurring in rare gas lasers(Laudenslager 1979).

Acknowledgmests

Our thanks are due to SERC and USAF for providing financial support of this work.

References

Abouar Kt Huber 8 A. Cosby P C. Saxon Rt P and Moaseley J T 1913 J. Chmw Phys. (A:406Adams N G and Smith D 1983 TechnuJUEX for the Study of Ion- .'oe.-aae Reacriofu ed J M~ Firrar and

W Saunders Jr (New York: Wiley) p 165Adams N G, Smith D and Alge E 1930 J Phvs. 8 At Woi Phys. 13) 3235Sohme D) V, Adams N4 G. PMosesman ., Dunkin D B and Ferguson E E 1970 1. Chem. PIhys 52 50%4Christiansen P A. Piuer K S. Lee Y S, Yates I H. Ermltr W C and Winter 14 W 1911 J Chgm. PhaIvs IS 5410Collins C 6 and Lee F W 1930 J Cheim Phyj. 72 5381Dagdigian P J and Campbaell M L 19117 Claim. Rev- 37 1Dehmtr P MI and Dehmer J L 1978 1 Claim. Phvs. 69 125Guyon P M. Gover T It and Baer T 1936 Z Py. D 4 39Ikezoe Y, Mtsuoks S. Takebe V! and Viggiano A 1981 Gai Phasje Ion - w.olenale Reaction Rate Constants

Through 1986 1The mlass Spectroscopy Society or Japan, Tokyo. 11'2 Japan)Jones T T C. lirkinshaw K. Jones J D C and rwiddy N D 1982 1. Plays. 0: At. Val. Phavs. 1 2439Khan A and Jordan K( D 1986 Chem. Plays. Lens. 123 363Laporte P, Saile V. Reinanger R,. Asaf U and Steinberser I T 1983 Plays. Rev. A 23 3613

114

Reactions of Kr and Xe ions with molecular gases 883

Laudenslaler J 3 1979 Kinetics of Ion-Mlolectile Reactions ed P Ausloos (New York: Plenum) p405Liao C L, Shao J D, Xu K, Flesch G D. Li Y G and N& C Y 1956 J ChenL Plhys. 15 3874Michels H H. Hobbs R H and Wnght L A 1978 J. Chent Py. 69 5151N& C Y, Trevot D J, Mahan 1 H and Let Y T 1916 J. Chen Phy. (A 4327- 1977 . Cem Piys 66 446Pratt S T and Dehmer P M 1982 Chem. PVys. Lett. 87 533Rosenstock H M. Draxil K, Steiner 8 W and Herron J T 1977 J. Phys. Cem. Ref Data 6 suppi ISamson J A R and Cairns R B 1966 J Opt. Soc An 56 1140Smith D and Adams N G 1917 Adv. At. AoL Phys. 24 ISmith D, Adams N G and Ljndinger W 1991 J. Chen Fiy. 75 3365Smith 0, Dean A G and Plumb I C 1972 J. Pys. 8: At. MAoL Pys. 5 2134Su T and Bowers M T 1979 Gas Pl as Ion Chemit:ry vol 1, ed M T Bowers (New York. Academic) p 83Tanaka K, Durap J, Kato T and Koyano t .91 . Chem. Phys. 74 5561

115