VoGne 3i,number 3 ., -CHEMICAL +HYSICS LETTERS 15 March 1975 -,, .-.

: . .

‘.

.- ,_:. ELECTliON ATTACRMENT TO SULPHUR MXAFLUOIUDE:

FQRMiiiON OF STABLE SF; AT LOW PRESSURi’ ..

_

.. _ wtih-ael S. FOSTER* and J.L! EEAUCHAMP* Arthur AmasNoyesLoboratory o~ff~emical Pf~ysics~, California Institute oj’ Technology, .Pkadena, Col~ornia911.25, USA

Received 2Ci August 1974 .:- Reti& manuscript Eccived 18 November 1974

:

EkCt~on attachment to SF6 hss been Stidicd at low prcs;urc in the 8s phk by ion cyclotron resomnce (ICR) spec- ‘tioscopy. Formation of stable SF; tinder cbllision-free conditions is rationdzed on the basis of radiative relaxaticn by (SF;!*. The utility of ICR for inve$gatin~ electron attaclment at extr&mely low pressures is noted.

_:

I.‘In&oduction

Tfie details of electron attachment by sulphur hex&%&de have been the abject of considerable

-.sttenti?n, &muIated not only by intrinsic interest in : electron a&&m&t processes but also 5y the exten- sive use of SF; as ai energy calibration standard for .Iqw electron kergjes [1,2], a detector of scattered ekctrons in inelastic excitatidn processes [2-43, and- efficient eiectron scavenger in radiolysis experiments [5-73, tid an interesting chemical reagent in halide- ion transfer reactions [4,8]. The reaetion of SF, with electrons is’described by the f6llowing scheme:

Lie (SF;)* ion, formed kth rate constant kf, con- this internal excitaticn ,at !east equal to the electron affiity of the molecuIe, and may autodetach an elec-. . . :

‘. Resent address: Chevron Research Company, Richmond, -lifOmia 94802, USA.

‘ti Dieyfus Teacher&olar, 1971-1976. : $ ContriiutionNo. 4944. ,’ ‘.,’ I....‘.:

:.. ._- : 482 1 ..,.., .-I’: , ‘, :‘.-: .., -. .:. ... _, .’ ,;. . . _’

. .

&on (k,), dk:ompose to SF,&), or be stabilized %l-kr by coksion with a species M (k,[M]), or by a radiative process (k,). The autodetachment process in [SF;)* has been especially well studied, both ex- perimentally [g--1 I], and theoretically [12J, and “autodetachment lifetimes” ranging between 10 PS &d 2 ms ha\re been measured. We wish to’repoti here an ion cycIo:ron resonance (ICR) spectroscopy [ 131 investigatior: of electron attachment to SF,, demon- strating a unimoIecuIar (presumably radiative) mech- anism which leads, at low pressure iti the gas phase, to a stable SF; ion. The facility for trapping ions un- der cdnditions where the time between collisions is 100 ms or more tiakes,ICP uniquely suitable for studying su+ processes. ,.

‘The negAve,-ion-mass spectrum of SF6 , taken in ,the normal drift mode of operation at an energy of 70 eV, shook only SF;, and SF; in appreciable abundance, in.a ratio oFapproxin$ely 20 : 1. In the LCR exjleriment,.it has been shown [I 1] that only electrons scattered from the beam and trapped

1 in the poteniia1 well of the source region [ 111 will be attached by SFe. Consistent with ttiis, we-f;-nd

that th$ SF; .@teiuity increases k 9e nominal energy : ‘.

Volume 71, number 3 CHEhlICAL PHYSICS LETTERS

I

15 March 1975

difference in these numbers is discussed in detdl

below.

Fig. 1. Variation with time of SF, and SF; single resonance iOIl iIltellSitiCS fOUOWiIlg a 4 ms, 70 ev electron barn pulse in pure SF6 at 1.5 X lo-’ torr.

.

of the electron berm varies from 1 to 100 volt, higher energies resulting in more inelastically scattered eIec- trons. All experiments described here were performed at a nominal electron beam energy of 70 volt.

Fig. 1 is a trapped-ion spectrum 1131 showing the intensity of SF; and SF? as a function of time after a 4 ms eIectron beam pulse at a pressure of I..5 X 10m7 torr. After the beam pulse, the curves rise rapidly, Ievel off at about 70 ms, and remain constant to 300 ms. As the trapping time is length- ened, ion Ioss begins to occur as the ions, diffuse to the walls of the cell because of collisions with the background gas. At this pressure and at a magnetic field of 15 kG, the half life for. SF; ion loss is about 15s.

At times longer than 20 ms, the rising SF; curve. in fig. l’is accurately described by a simple exponen- tial function,

where [SFc(=)] is the SF; ion intensity at long times, when all the electrons have been attached. The apparent bimqlecular rate constant, k,,, for electron attachment is determined as (1.6 +- 0.2) X 10es cm3 molecule-l s-l, which is independent of pressure below’=5 X 10d7 torr. Literature values for +he attachment rate of thermal electrons to.SFs, deter- mined by vhious methods but always in the presence of a large excess of buffer gas, are (2.4 9 0.6) X lo-’ cm3 moIecule_ 1 s-l [ 141: The order-of-magnitude_

: :.

A detded kinetic analysis of the reaction scheme above indicates that the apparent rate coefficient de- scribing the formation of SF; is a function ofpres- sure. When the species M is rrot SF,, stabihzation of (SF;)* by collision (rate = k, [M]) wiI.I yield a larger value of k,PP than in the absence of collisions (high

.pressure limit = k;). Addition of nitrogen to the SF6 increases kapp in the expected manner, up to G.‘Z X lObE cm3 molecule-l S-L at an N, pressure of 7.3 X 10m5 torr. In pure SFb, however, kar.,F is predicted by the kinetic analysis to decrease with increasing SF, pressure, to a limiting value of k,. In agreement with this, kaap decreases monotonically with added SF, to 2.5 X IO+ cm3 moIecule-i s-r at a pressure of 1.5 X lo-’ torr. The SF6 pressure could not be raised higher than this because the totai rate of for- mation of SF; became comparable to the sampling

time of the experiment (~2 ms). A problem with the simplified reaction scheme

above is that kf, k,, and Fid are functions of the kinet- ic energy of the attached electron [iZ,LS]. Thus: a question arises in this work with regard to the in- ‘fluence of the electron energy distribution on the measured rate of attachment. In the “flat” ICR ceU used in these experiments, the scattered electrons trapped in the source-region potential weil will initial-

ly have a kinetic energy distribution ranging from ca. thermal to 0.9 VT, where VT’is the trapping voltage.

Values of V, used in this work ranged between -L .5

and -3.5 volt, and thus an appreciable fracticn of the electrons will initiai!y be epithermal. The average election kinetic energy will rise with increasing VT, and the rate of thermalization of the electrons could have a pronounced effect on the measured vaIue of kam.

To assess these effects, kzPP was determined for a series of tr2pping voltaSes between -1.78 md

-3.60 volt. The results are shown in fig. 2. I: is seen that after a period of relatively slower SF: forma-

.ticn, all the curves come to the same Iimiting slope. Since the rate c@ant describing the formation of SF;, kapp, is calculated from the hrrW.ng slopes of these curves, the measured rate is Czndependenr of VT and thus of the initial electron energy distcibu- tion. As expected, fig. 2 shows that the electron thermalization time increases as the distribution peaks at progressively higher energies (higher VT).

: 483

V6hx-q~ 31, n-amber 3 _. CHEMICAL PHYSICS LETTERS 15 March 1975

.,’ t A 3.60 \

_:

,, .p I 1 I ‘0.01 ‘# A 0 IO 20 30 40 50 60 70 80

Tie fmsec)

.Fig. 2.. Variatiop of 1 _- SFz(t)/SFi(m) rvith Cm? at 1.3 X IO-? torT and at various trapping voltages, VT (in -2). Trapph voltages are negxtivc for negztivc ions. SFz(-)

is the SF-, ion intensity when all the electrons have been

attached, and is equal to the number of free ekctrons pro. durxd during the beam puke.

The reason .tiat ail the curves have the same limiting slope is that the +&on relaxation process is evident- Ii muc.h faster than the attachment process, meas- .ured by. k,, . Sulphur hexafluoride is probably par- .ticularly efficient at relaxing electronS via autode- tactient of- traiisiefii (SF;)*; leaving a vibrationatly excited SF6 ,ncutral [9]. :

The eonsttit SF; ion intensity between.70 and 360 ms i;l fig. 1 suggests that one of the two stab& zaticz _qechanisms in the scheme above iS operative on (SF;)* in this experiment. if (SF;)* were de- .‘.aching an electrdn and being reformed con+nualiy, .tie Fadtial,loss of electrons from the cell and the de- Fompositiqn to 8FL should be manifest as,a pro> ‘n&mced decline in SF: ,ion intensity with time,

which is not observed. The collision frequency’of SF; with its parent neutral can be estimated from the ion-inductd dipole collision rate given by [ 161

ic =-2ne(cu/p))“2 = 7.00 X lo-lo cm3 molecule-l s-l’,

where e is the I:lectronic charge, u the pola&ability of SF,(= 6.56 X.1O-24 cm3 [l?]), and p the reduced mass of the collision pair. At a pressure of 1.5 X lop7 torr, the mean time between collisions is thus 200 ms. This is substan:ially lonber than the rise time of the SF: curve in fig. 1 and implies that collisional stab&

-7ation of (SFi)* to SF6 is ukniportant under these catiditions. Thr; possibility of an electron transfei reaction from (SF;)* to SF,, which might occur at a faster rate than the collision frequency, was eliminated by ejecting 32SF; from the cell and noting that no change occurred in the intensity of 34SF; as a func- tion of trappinE time. In agreement with the results of kifshiti et al. [ 181, the electron transfer reaction has a rate constant of <1O-*2 cm3 molecule-’ s-l. As the only remaining possibility, a radiative process on a time scale of several milliseconds (probably in- volvin~ vibrational modes) is proposed as the mecha- nism leading.to stable SF; in the low pressure trapped- ion experiment.

Unequivocal evidence for the stability of the SF:

ion in fig. 1 wa.s obthed from an electron scaveng-

ing experiment in which’CC1, was added to the SF,. Carbon tetrachloride undergoes efficient dissociativti attachment (u s 100 A2 for thermal electrons [19]) to give Cl- in an irreversible process. Thus, if SF; autodetaches in the presence of Ccl,, a trapped-ion : experiment should show a ,decrease in’SF6 and an in- ,crease in Cl- as a function of time. Fig. .3 presents the data for a 3.0: 1 mixture of SF6 and Ccl4 at a total pressure o! 1.8 X 10s7 torr. After about 60 ms for attachment #of all the electrons, both SF; and Cl- ion intensities are constant with time. That is, the SF; ion is oaf undergoing autodetachment. On 3 shorter time scale, however, Cl- and SF6 are in- deed coupled. Continuous ejection of SF; from the All (ejection time = 400 p) reduces the Cl- inten- sity’by 6S%, i.e., under the conditions of fig. 3, at

least 65% of the Cl- ion is derived from electrons originally autodl:tached from (SF;).

From tie data of fig. 3, it is possjble to calculate relative appven.: rates of electron-att&hment by SF6

:

. .

~ro~ume 3 1 ntimber 3 > CHEMICAL PHYSICS LETTERS 15 hfaich 1975

Fig. 3. Variation with time of SF; and Cl- ion abundances

following a 5 ms, 70 eV electron beam pulse in a 3.0 : 1 mixture of SF6 and CCL at a total pressure of 1.8 X 10m7 tom. Single resonance ion intensities have been mass corrected

to given ion abundances.

and CClb. From the known ratio of neutrals (3 : 1) and the measured ratio of SF{ to Cl- (0.28), CC14 is found to be 11 times more efficient in,attaching electrons to give Cl- than is SF6 to give SF6 under low-pressure conditions (no collisions). Thus the rate

coefficient for thermal electron attachment in Ccl, is calculated to be (I .g + 0.2) X lo-? cm3 molecule-l s -l. This is somewhat lower than the value of 2.8 X lo-’ cm3 fiolecule-l s-l reported by Christophorou and co-workers [ 15,191. Alternatively, from their determination that SF6 and Ccl4 have the same rate of electron attachment under high-pressure conditions [ 15,191, one concludes that (SF;)* in the present experiment is 10 times more likely to auto- detach an electron than to undergo radiative relaxa- tion; i.e., k,/k, = 10. We have measured the rate of electron attachment in pure CCid by a procedure identical to that described here for SF6. However, Ccl4 is re!atively poor at thermalizing electrons and the ratedetermining step in Cl- production is relaxa- tion of the electron energy distribution, not the ac- tual attiuh-nent reaction. Addition of CO2, which is

The ICR technique appears to be singularly well- suited for investigations of electron attachment at low pressure, in the absence of collisions, where radia- tive processes may be significant. ExSension of the methodology described here to a variety of dissocia- tive and non-dissociative attachment processes is in progress.

References

[I] C.F. Brian, Intern. J. Mass Spectrom. ion Phys. 3 (1969) 197.

[2] R.N. Compton, R.H. Huebner, P.W. Reinhardt ad L.G. Cluistophorou, J. Chem. Phys. 4E (L968) 9OL.

[3] M.-J. HubinFranskin and J.E. CoUin, Intern. I. him Spectiom. Ion Phys. 4 (1970) 151; 5 (19701 163, 255.

[4] J.A. Stockdale, D.R. Nelson, F.J. Davis and R.N. Cgmpton, J. Chem. Phys. 56 (1972) 3336.

[5] J.M. Warman, K.-D. Asmus and RH. SchuIer, Advan. Chem. Ser. 82 (1968) 25.

[IS] R.C. Rumfeldt. Can. i. Chem. 49 (1971) 1262. [7] G.R.A. Johnson and J.L. Redpath, 5. Phys. Chem. 72

(1968) 76.5. [a] J.C. Haar&z and D.H. McDartiel, I. Am. Chem. Sot. 9.5

(1973) 8562, and referencesthcrein. [9] R.N. Compton, LG. wtophorou, G.S. Hurst and

P.W. Reirhard+.J. Chem. Phy% 4.5 (1966) 4634.

reasonably effective at the,rm&zing electrons, in- [lo] P.W. Harland and J.C,J. Thyme, J. Phys. Chen. 75

creases k,,,(CI~_) d ramati&lly. At the highest CO2 (1971) 3517.

pressure employed, 3 X 10m6 torr, k had increased !ll] R.W. Odom, D-L. Smith and JH. Futrell, Chem. Phys

to 1.4 X low7 cm3 molecule-? s -I .“%e results ob- titters 24 (1974) 227, and references therein

,.(i2] C.E. Klots, J. Chem. Plqx. 46 (1967} 1197.

485

tained with Ccl4 are very similar to those derived by -Warman and Sauer using an entirely different tech- riique [20].

3. Conclusions

The distinguishing features of the present work, 2s

opposed to previous investigations of SF6, are (1) the experimental time scale is several orders of magnitude longer and (2j ‘he pr,essure is sufficiently 10~ that

collisional processes are not significant. Formation : of stable SF, at low pressure is reasonaMy accountgd for by invoking a radiative de-excitation mechanism. Such a mechanism would not be attendant to previous experiments, which involved either much shorter time scales 0: much higher pressures. The results indicate that caution is advisable in interpreting experiments which purport to measure autodetachment lifetimes.

:m.

,_ \.--,, __. . ..

,. :