J. Phys. Chem. 1987, 91, 1509-1515
1509
Chemical Reactions on Clusters. 5. Gas-Phase Unimolecular Decomposition of the CH,OH, CH,OD, CD,OH, and CD,OD Ions in Association with Argon and Carbon Dioxide Clusters A. J. Stace School of Molecular Sciences, University of Sussex, Falmer, Brighton, BNl 9QJ. United Kingdom (Received: July 23, 1986; In Final Form: October 15, 1986)
Ion clusters of the type Ar,.X+ and (C02),.X+, where X is one of four isotopic variants of methanol, have been formed by electron impact following the adiabatic expansion of an inert gas/methanol mixture. In the case of argon clusters the product ions Ar,-CH20H+ and Ar,-COH+ (or their isotopic equivalents) are observed. For carbon dioxide clusters the only product ion is (CO2).CH2OH+. A consideration of the available experimental data on the fragmentation pattern of methanol ions, and in particular results from chargetransfer and photoelectron spectroscopy experiments, supports the proposal that cluster-bound methanol ions are excited by charge transfer from the inert gas cluster as opposed to an individual inert gas atom or the ionizing electron. From a knowledge of the internal energy of the methanol ions a lower limit to the vibrational predissociation time has been calculated as 10-’2s.
Introduction The detection of van der Waals molecules and clusters frequently involves their ionization by either photons or electrons. Even the most fragile of clusters, for example, those of argon, will form stable ions as the result of ionization by electrons with energies in excess of 50 eV. Such behavior does not imply that electron impact is a “soft” ionization technique but is more a reflection of those processes which lead to the formation and stabilization of ion clusters, Le. autoionization from high-lying Rydberg states’q2followed by self-trapping of the positive ~harge.33~ An additional feature in a number of recent experiments involving cluster ionization has been the appearance of ions containing fragments of the original cluster constituent^,^'^ i.e. (C02),.CO+ ions from the ionization of carbon dioxide cluster^^*^^ and Ar,. CH30CH2+ions from mixed dimethyl ether-argon clusters.’2 In a recent series of experiment^'*'^,'^ we have been concerned with the unimolecular decomposition of organic ions in association with argon and carbon dioxide clusters. Through the preparation of these mixed reactive clusters we have sought to separate the roles played by the various cluster constituents, i.e. energy loss via inert gas evaporation vs. the unimolecular decomposition of the molecular ions. By changing the inert gas component we might hope to modify the efficiency of the evaporation process. In these experiments we have attempted to address the following questions: (1) How does the molecular ion acquire the internal energy necessary for fragmentation? (2) If the cluster as a whole contains (1) Dehmer, P. M.; Pratt, S. T. J . Chem. Phys. 1982, 76, 843. (2) Walters, E.A.; Blais, N . C. J. Chem. Phys. 1981, 75, 4208. (3) Haberland, H. In 13th International Conference on Electronic and Ionic Collisions, Eichler, J., Hertel, I., Stolterfoht, N . , Eds.; North Holland: Amsterdam, 1984. (4) Stace, A. J. Chem. Phys. Lett. 1985, 113, 355. (5) Ono, Y.; Linn, S.H.; Prest, H. F.; Gress, M. E.;Ng, C. Y. J . Chem. Phys. 1980, 72,4242. (6) Ono, Y.; Linn, S.H.; Prest, H. F.; Gress, M. E.;Ng, C. Y. J . Chem. Phys. 1981, 74, 1125. (7) Erickson, J.; Ng, C. Y. J . Chem. Phys. 1981, 75, 1650. (8) Ono, Y.; Ng, C. Y. J . Am. Chem. SOC.1982, 104, 4752. (9) Stephan, K.; Futrell, J. H.; Peterson, K. I.; Castleman, Jr., A. W.; Mark, T. D. J . Chem. Phys. 1985, 77, 2408. (IO) Stace, A. J. J . Phys. Chem. 1983,87, 2286. (11) Stace, A. J. Chem. Phys. Lett. 1983, 99,470. (12) Stace, A. J. J . Am. Chem. SOC.1984, 106, 4380. (13) Walters, E. W.; Grover, J. R.; Newman, J. K.; White, M. G. Chem. Phys. Left.1984, 1 1 1 , 190. (14) Romanowski, G.; Wanczek, K. P. Int. J . Muss Spectrom. Ion Processes 1984, 62, 277. (15) Stace, A. J. J . Am. Chem. SOC.1985, 107, 755.
0022-3654/87/209 1- 1509$01.50/0
sufficient energy to break a covalent bond, what prevents the more weakly bound inert gas component from shattering? (3) Can the reaction tell us anything about the position of the molecular ion with respect to the inert gas component? For the purposes of discussion we shall refer to COz as an inert gas in the sense that it does not fragment when associated with an organic ion. In respect to the above questions, our experiments so far1*12J5 have led us to draw the following conclusions: (1) in large clusters excitation of the molecular ion appears to proceed via a chargetransfer mechanism; (2) the molecular ion decomposes because its lifetime is very much shorter than the time necessary for energy randomization to the inert gas component; (3) the molecular ion appears to “sit on” rather than “within” the ion cluster; (4) individual ion clusters appear to exhibit behavior which could be associated with their undergoing a phase transition. Until now evidence for the charge-transfer conclusion has come primarily from a consideration of reaction energetics; no fragmentation processes have so far been observed for which to > IP (cluster) - IP(X), where to is the critical energy of reaction and X is the organic molecule. However, two very recent photoionization studies of van der Waals molecules by Ding et a1.16 and Kamke et al.17 have provided strong additional support for the above proposal. In particular, Kamke et al.I7 have shown that photoexcitation of the inert gas atom in the complex C6HSCN.Ar leads to ionization of the benzonitrile molecule. The term “intramolecular Penning ionization” has been proposed as a classification of such processes.” The presence of reaction fragments could be taken as good evidence of nonstatistical behavior in ion clusters. Consider, for example, the behavior of acetoneargon clusters, Ar,.(CD3)2CO+.1S Following electron impact ionization these ions lose CD3 from the acetone moiety to give Ar,CD3CO+. The relative intensity of the product ion for n = 35 is not very much smaller than that observed for n = 1 or 2;15 however, in going from n = 1 to 35 an extra 102 vibrational degrees of freedom have been added to the ion! If the argon component were an efficient energy sink, as in the case of covalently bonded atoms, then we would expect to observe a rapid decline in product ion intensity as a function of increasing argon cluster size. Instead the continued appearance of fragment ions, irrespective of cluster size or constitution,15 suggests that energy transfer from a vibrationally excited parent ion, such as (CD3)2CO+,is slow in comparison to the reaction
0 1987 American Chemical Society
1510 The Journal of Physical Chemistry, Vol. 91, No. 6, 1987
Stace
time scale. Our reason for assuming that the inert gas component does not participate in energy randomization can be illustrated by using the RRK expression for the probability that given an internal energy E then an energy 1 to will be found in any one of the vibrational coordinates of a clusterI8
P(t0,E) = (1 - t 0 p ) N - l
(1)
where N is the number of vibrational degrees of freedom. Taking E = 4 eV (see later) and using Ar35-(CD3)2CO+ as an example, there are three possibilities to consider: (1) to = 0.04 eV (appropriate for the removal of an argon atom) and N = 129, this gives P(c0,E) = 0.27; (2) eo = 1.0 eV (appropriate for the breaking of a covalent bond) and N = 129, this gives P(co,E) = (3) to= 1.0 eV and N = 24 (the acetone ion alone), this gives P(to,E) = 10". It can be seen from this simple calculation that if energy randomization into the inert gas component were taking place, then it would provide the ion cluster with a far more facile decomposition route than the one it actually chooses to take. With regard to energy randomization, the situation in clusters is going to be significantly different from that considered appropriate for molecules or ions composed of covalently bonded atoms.I8 The energy loss mechanism in clusters held together by weak van der Waals bonds will consist of evaporation induced by vibrational predissociation. Hence, any vibrational energy lost from a reaction site into an inert gas cluster will most probably never return to the polyatomic ion. However, those same statistical factors which govern the unimolecular decomposition of an ion18 could also, through the introduction of a time lag, play a part in determining the efficiency of vibrational predissociation. As a result the unimolecular decomposition of a cluster-bound polyatomic ion could compete effectively with vibrational predissociation even when the internal energy of the ion is quite close to the critical energy of reaction.I5 Although predissociation often only requires the presence of a single quantum of vibrational energy,I9 the absence of a strong coupling mechanism between the excited mode and the predissociative mode can mean that the time scale for such an event is very often much longer than a vibrational period.20g21 In a polyatomic the statistical partitioning of vibrational energy could increase the predissociation time scale still further.15 In this paper we present the results of a detailed study of the unimolecular decomposition of methanol ions in association with argon and carbon dioxide clusters. The reactions of isolated gas-phase methanol ions have been the subject of numerous publications,224 and we have drawn extensively upon this wealth
(18) Forst, W. Theory of Unimolecular Reactions; Academic Press: New York, 1973. (19) Gentry, W. R. In Resonances in Electron-Molecule Scattering, van der Waals Complexes and Reactive Chemical Dynamics, Truhlar, D. G., Ed.; American Chemical Society: Washington, DC, 1985; ACS Symp. Ser. No. 263. (20) Levy, D. H. Adu. Chem. Phys. 1983, 47, 363. (21) Beswick, J. A.; Jortner, J. A h . Chem. Phys. 1983, 47, 363. (22) Friedman, L.; Long, F. A,; Wolfsberg, M. J . Chem. Phys. 1957, 27, 613. (23) Wilmenius, P.; Lindholm, E. Ark. Fys. 1962, 21, 97. (24) Sjogren, H . Ark. Fys. 1967, 33, 597. (25) Beynon, J. H.; Fontaine, A. E.; Lester, G. R. Int. J . Mass Spectrom. Ion Phys. 1968, I , 1. (26) Refeay, K. M. A.; Chupka, W. A. J . Chem. Phys. 1968, 48, 5205. (27) Gordon, S. M.; Krige, G. J. Isotope, 18th Annual ASTM Conference; San Francisco, 1970. (28) Brehm, B.; Fuchs, V. Kebarle, P. Int. J . Mass Spectrom. Ion Phys. 1971, 6, 279. (29) Warneck, P. 2. Naturforsch. A 1971, 26a, 2047. (30) Lifshitz, C.; Tiernan, T. 0. J . Chem. Phys. 1972, 57, 1515. (31) Corval, M.; Masclett, P. Org. Mass Spectrom. 1972, 6, 51 1. (32) Cooks, R. G.; Hendricks, L.; Beynon, J. H. Org. Mass Spectrom. 1975, 10, 625. (33) Jonsson, B.-0.; Lind, J. J . Chem. SOC, Faraday Trans 2 1976, 72, 906. (34) Berkowitz, J. J . Chem. Phys. 1978, 69, 3044. (35) Sharp, R. C.; Cutshall, E. A.; Muschlitz, Jr., EE. Int. J . Mass Spectrom. Ion Phys. 1978, 27, 5 5 .
AG.CHOH,I \
\
"I
-
Figure 1. Typical examples of the recorded mass spectra. Also present ArWl+ Ar) peaks. are isotope O6Ar and 38Ar)and metastable (Ar,'
+
of information in order to provide a quantitative interpretation of our observations. In particular, it will be seen that recent gas-phase photoionization data16,'7,34*36-38 have been of considerable assistance in determining the mechanism responsible for exciting the reactant molecular ions while they are in association with the inert gas clusters. With knowledge of the internal energy of the reactant ion, it has been possible to place a lower limit on the time scale for vibrational predissociation. Experimental Section
Neutral clusters were generated by the adiabatic expansion of a gas mixture through a 100-km pulsed nozzle operating at approximately 20 Hz. Following collimation through a 0.5" skimmer positioned 2 cm from the nozzle, the modulated cluster beam was ionized by electron impact and mass analyzed on a modified AEI MS 12 mass spectrometer. The ion signal was monitored via a lock-in amplifier (Brookdeal 9503) which took its reference from the unit responsible for driving the pulsed nozzle. Most of the mass spectra were recorded with a methanol concentration of the order of 100 ppm and a nozzle stagnation pressure of 46 psi. The most suitable inert gas/methanol ratio was determined through trial and error for each different sized inert gas cluster. A very small amount of methanol was introduced into the nozzle and gradually diluted with inert gas. During the dilution process the clusters of interest were often seen to be accompanied by clusters of the type Ar,.CH30H2+, which was taken as an indication that a methanol dimer had clustered with the argon prior to ionization. A further reduction in the methanol concentration was always effective in eliminating the dimer contribution. As with previous experiments in this ~ e r i e s , ' & ' ~the . ' ~ presence of impurities and possible mass coincidences with isotope peaks placed limitations on the accuracy with which measurements could be made. The presence of trace quantities of O2 and N2 caused the most difficulties because Ar,-02+ has the same nominal mass as A r , C H 3 0 H + and Ar,CD20+, and the mass of Ar,.N2+ coincides with that of Ar,CO+. Only by repeatedly degassing each liquid sample could the interference from the presence of air be minimized. Ion peaks resulting from the mass combination 36Ar-Ar,+overlapped with those from Ar,.CD,OD+; by recording the intensities of the former in the absence of CD,OD a correction was made to each of the Ar,.CD,OD+ intensities. A further source of interference came from hydrogen atom transfer, possibly due to the presence of water, on the surfaces of the inlet system and the pulsed nozzle. In particular, the labile oxygen-bonded deuterium atoms on CH,OD and CD,OD were very susceptible to exchange. The presence of this process could be detected by examining the mass spectra for pure methanol (36) Niway, Y.; Tsuchiya, T. Ado. Mass Specrrom. 1980, 8A, 56. (37) Momigny, J.; Wankenne, H.; Krier, C. Int. J . Mass Spectrom. Ion Phys. 19, 35, 151. (38) Nishimura, T.; Niwa, Y.; Tsuchiya, T.; Nozoye, H . J . Chem. Phys. 1980, 72, 2222. (39) Mashni, M.; Hess, P. Chem. Phys. Let?. 1981, 7 7 , 541. (40) Galloy, C.; Lecomte, C.; Lorquet, J. C. J . Chem. Phys. 1982, 77, 4522.
The Journal of Physical Chemistry, Vol. 91, No. 6, 1987
Chemical Reactions on Clusters clusters. For example, in the case of CH,OD, the effect of exchange was to produce mixed ion clusters of the type [(CH,OH),.(CH,OD),]D(H)+. In order to minimize the influence exchange may have on our results, the apparatus was first "seasoned" with DzO, then baked, and further "seasoned" with samples of the deuteriated alcohols. In spite of these precuations it was not possible to eliminate the exchange component. Because of the experimental difficulties summarized above, we consider the results obtained for Ar,.CH30H+ and Ar,.CD30H+ to be more reliable than those obtained for Ar,.CH30D+ and Ar,.CD30D+. In the case of the C0,-methanol clusters, (COz),+ clusters themselves fragment to give as one of the products (C02),-02+ where m < n.14 By measuring the intensities of the latter in the absence of methanol a correction was made to the (CO2),.CH30H+ intensities. Figure 1 shows examples of the recorded mass spectra for CH30H and C D 3 0 H clustered with argon. Most of the experimental observations were of a quality similar to those given in the figure. Because the ions of interest all had relatively low intensities, the mass spectrometer was operated with both the source and collector slits set to their maximum values. The subsequent loss of resolution meant that it was not possible to make accurate measurements on processes which involved the loss of 1 amu from ions with masses beyond 900 amu. Hence all the results presented here are for Ar,.X+ and (COJn.X+ with n 5 21. During a typical experiment, the pressure in the nozzle expansion chamber was approximately lo4 Torr, in the collimation chamber it was lo6 Torr, and in the ion source of the mass spectrometer the presure remained below 10" Torr. The latter low value allows us to disregard the possibility that ion-molecule reactions are responsible for the effects we observe. In the field-free region between the ion source and the magnet the pressure was of the order of Torr. We believe this value to be low enough as to minimize the possibility of introducing a collision-induced component in to the observed reactions.
C
.-
1
3
2-
AG~.CHOH+
'p
en 1-
30
35
io
i5
50
nozzle pressure/psi
1-I
30
Results and Discussion
1511
35
LO
L5
50
nozzle pressure/psi
Figure 2. Relative product ion intensities plotted as a function of nozzle In the normal mass spectrum of methanol the principal product pressure. For each example the intensity of the product ion ions can be accounted for with the following m e c h a n i ~ m : ~ ~ . ~stagnation ~ on ATl2has been divided by the intensity of the parent ion on Arlz. The error bars represent k1 standard deviation. CH30H+ C H 2 0 H + H CHO' H2 (2)
-
-
+
-+
+ Hz
CHOH'
+
(3)
+
CH3+ OH (4) The thermochemistry and kinetics of each of these reaction steps have been the subject of a number of detailed studies using both photon and electron ionization.22*26~28*29~37 When methanol is clustered with argon the following ions are observed: Ar,. CH30H+,Ar,-CH20H+, and Ar,.CHOH+ for n in the range 1-21, The mass spectra of Ar,.CH30D+ and Ar,.CD30H+ showed the presence of a number of peaks with very low intensities which could possibly have corresponded to the presence of minor products. However, in view of the difficulties discussed in the Experimental Section regarding impurities and hydrogen atom exchange, the origins of these peaks was not pursued. All four isotopic variants showed the presence of an ion which could possibly have been Ar,-CO+. Although the intensity of this peak did not diminish upon further degassing of the liquid samples, interference from Arn.N2+ could not be completely ruled out. Similarly, all four systems produced an ion corresponding to Ar,.CHO+ (Ar,CDO+ in the case of some of the deuteriated samples). In each case, however, the ion had such a low intensity that its origins could not be determined with any certainty. None of the product ions discussed above are observed when pure methanol clusters are ionized by either photons41 or electrons.42 In common with other systems studied,1°-12Js the relative product intensities all showed a marked dependence on the nozzle (41) Cook, K. D.; Jones, G. G.; Taylor, J. W. In?.J . Mass Spectrom. Ion Phys. 1980, 32. 273. (42) Shukla, A. K. Ph.D. Thesis, Southampton University, 1982.
stagnation pressure. Figure 2 shows examples of this behavior resulting from C H 3 0 H and C D 3 0 H clustered with argon. It is reassuring to note that similar behavior has recently been observed for the decomposition of COz ion ~ 1 u s t e r s . lHowever, ~ such behavior on the part of pure carbon dioxide clusters is in contrast with the fact that the fragmentation of neither acetonelS nor methanol ions (this work) in association with COz clusters displays a pressure dependence of the type shown in Figure 2. A further observation we have made is that the degree of fragmentation also depends on the nozzle temperature. Figure 3 shows plots of the total relative product ion intensity as a function of nozzle pressure with the nozzle held at four different temperatures. Both the pressure and temperature dependences are consistent with our original proposal that the inert gas component of the clusters is subject to a phase transition." At a low nozzle pressure (or an elevated nozzle temperature) the inert gas component is liquidlike and its vibrational modes can couple efficiently with the vibrationally excited molecular ion. As a result the cluster rapidly loses energy through the evaporation of argon atoms. At a high nozzle pressure (or reduced nozzle temperature) the inert gas component is more solidlike and can no longer couple effectively with the vibrational modes of the molecular ion. Hence, the ion retains all or more of its internal energy and as a result decomposes. A similar conclusion has also been proposed to explain the behavior of pure COz ion ~ 1 u s t e r s . lOther ~ possible sources of temperature-dependent behavior will be discussed later. Figures 4-7 show the relative intensities of the reaction products for the four isotopic variants plotted as a function of argon cluster size. The product intensities from each of the four molecular ions are approximately equal and the intensities of Ar,,-CH,OH+ (or
1512 The Journal of Physical Chemistry, Vol. 91, No. 6, 1987
Stace .->
3
-is
J"
51
308K
6-
a, C
CD~OH+
.-0
CDOH' a
i i
C
0 c u Y
j
i
+
i i io A i z 1s
14 i 5
Is IS i s i o
$0
$I
n
Figure 6. Same as for'Figure 4, but for Ar,.CD30Ht.
U
e
-a 0
+D02Dc4 : CDoD+ A~,.CD~OD+
e
0
e
-
.-x v) C a,
c.
30
c
Lb
35
C
nozzle pressurelpsi
Figure 3. Total (ArI2-CH2OHt+ Ar12CHOHt)product ion intensity plotted as a function of nozzle stagnation pressure with a nozzle held at
different temperatures. For the sake of clarity the error bars have been omitted.
0 u
c
2
;:
2
-0
ea
I
,
,
,
3
4
5
I
+
I I 6
It
12 i s
i4
is k
ti b
te
io A
n A~.CH,OH*
I
4 i i i i i i cbhhhGi5~hCiohzi
n Figure 4. Relative product ion intensity plotted as a function of argon cluster size for the decomposition of Ar,CH30Ht. For each point the intensity of the product ion on Ar, was divided by the intensity of the
parent ion on Ar,. Each measurement was made at a nozzle temperature of 308 K and at a nozzle stagnation pressure of 46 psi. The error bars represent f 1 standard deviation.
A~,.CH,OD+
,o 1a i i i i i i i cb
$1 $2
b
11
is C
$7 18
i o zb n
n Figure 5. Same as for Figure 4, but for Ar,.CH,OD+.
its isotopic equivalent) is for each example always greater than that of Ar,CHOH+. There are no marked intensity fluctuations and certainly none which are common to all four systems. This is in contrast, for example, to the results for acetone and ace-
Figure 7. Same as for Figure 4, but for Ar,.CD,ODt.
tone-d6.15 It was emphasized in the Experimental Section that the CH30H+and CD30H+results were considered to be the most reliable. From a comparison of Figures 4 and 6 it can be seen that beyond n = 12 the product ion intensities from these two ions are very similar. We have stated p r e v i o u ~ l y that, ~ ~ , ~due ~ to cascade effects brought about by inert gas evaporation, the observed product intensities for small argon clusters, n < 10, are unlikely to reflect the characteristic properties of individual ion clusters, i.e. magic number formation, etc. The similarity between C H 3 0 H f and C D 3 0 H + would suggest that there is little or no isotope effect associated with any of the observed fragmentation processes. In turn this would indicate that each reaction proceeds extremely rapidly, i.e. at an internal energy high enough such that the vibrational state densities are insensitive to isotopic substitution. Use of the four isotopic variants has made it possible to confirm that the clustered products are the same as those observed from the decomposition of the respective isolated molecular ions,34i.e. Ar,.CH20H+ rather than Ar,-CH,O+ and Ar,CHOH+ rather than Ar,CH20+. The absence of any ArnCH3+ions from reaction 4 closely parallels our previous o b s e r v a t i ~ n s , ~where ~ ~ ~ * in J ~all cases product ion formation involved the retention of an oxygen atom together with some portion of the hydrocarbon component of the molecular ion. In an analysis of this effectl2.l5it was suggested that the oxygen atom in a product ion provided a natural clustering site for the inert gas atoms, and that it therefore formed an integral part of the structure of an ion cluster (this interpretation did not appear to extend to the product ions on C 0 2 cluster^^^). In each case the hydrocarbon component was assumed to protrude from the main body of the cluster. This model of an ion cluster was adopted for two reasons, first, in order to explain the observation that clusters like Ar,.(C2H5)20t l o and Ar,. (C2Hs)2COt43 appear to be capable of lossing a significant proportion of their hydrocarbon component without seriously disrupting the argon component and, secondly, to account for a number of substantial intensity fluctuations in terms of "magic number" atom combination^.^^^^^ Unlike the reactions of either dimethyl ether12 or acetone15 ions, neither of the methanol decompositions generates a terminal oxygen atom. Therefore, taking into consideration the disruptive influence of the two reactions,
The Journal of Physical Chemistry, Vol. 91, No. 6, 1987 1513
Chemical Reactions on Clusters
TABLE I: Relative Product Ion Intensities from Methanol Ions for a Range of Excitation Mechanisms mechanism charge transfer' clusterd electron impactC PESb HzO+ Kr+ Ar+ CH4+ hv Ar,+ (CO,),' 70.0 16.0 12.09 ion 1.55" 3.8Sh 4.90h 1.15" 5.1Sh 4.90" 2.90" 70.0h 5.1Sh 1.15" CH30H+ 29 0.4 100 52 39 100 66 96 100 CH20H+ 100 100 65 32 100 100 10 100 100 20 CHOH' 1 9 2 4 52 8 CHO+ 1 5 76 25 56 CHI' 4 100 30 22
IRd hu ?h 60 100 10
36 56
CIC He(23*lS) 9.7 1 37 100 11
66
30
'Reference 30. bReference34. 'Taken from a Kratos MS 80. dReference 39, multiphoton IR laser excitation. 'Reference 35, chemiionization. f n = 12 in each case. gImpact energy in electronvolts. "Maximum energy available to the ion in electronvolts. we would suggest that both the molecular and product ions of molecule. In turn a larger cluster might favor steps 6 and 7, methanol occupy sites on or very close to the surface of the cluster. particularly if the point of ionization is more than the diameter When methanol is clustered with carbon dioxide only two ions of an argon atom away from the organic molecule. It is interesting are observed: ( C 0 2 ) , C H 3 0 H + and (COz),CH20H+, with the to note that hole transport in liquid argon displays a significant latter ion in all cases having an intensity approximately 10%that temperature d e p e n d e n ~ e as ; ~ ~a consequence of this the presof the molecular ion. For this reason we did not consider it sure-dependent behavior we observe in, for example Figure 2, could necessary to present the COz-methanol data in graphical form. arise from the influence cluster temperature has on the excitation Although the product intensities from molecular ions on CO, mechanism. Because both argon and carbon dioxide clusters have clusters are generally lower than those on argon c l ~ s t e r s , 'this ~ ionization potentials which are lower than those of the individual is the first example where there has been a difference in the atom or m o l e c ~ l e , lit- ~has ~ ~not ~ ~been possible to place a precise number of reaction steps.43 value on the quantity IP(c1uster) - IP(molecu1e); also we are not To account for the fact that a molecular ion is able to acquire able to discount the possibility that the ionization potential of the sufficient internal energy to decompose, it has been suggested that clusters may be further modified by the presence of a molecule. some form of charge-transfer mechanism may be in o p e r a t i ~ n . ' ~ ~ ' ~ To assist in the interpretation of our results, use has been made The electron impact process first ionizes, for example, an argon of the extensive range of experimental data available on the atom or group of atoms and the positive charge then transfers to fragmentation of methanol ions.22-39 The variety of different the molecule. The net result is that the internal energy of the excitation mechanisms extend from chemi-ionization using molecular ion increases by the difference IP(c1uster) - IP(mo1metastable helium atoms35to multiphoton IR laser vaporization ecule). Support for this proposal has recently been provided by and ionization of solid methanol.39 Table I presents a compilation a series of photoionization experiments on van der Waals moleof some of the observed product ion intensities together with the c u l e ~ . ' ~In~ terms '~ of the present experiments we might consider intensities measured for Arlz-CH30H' and ( C 0 2 ) l z C H 3 0 H + , the following sequence of events to be taking place inside the ion taken as being typical of the present experiments. A consideration source of the mass spectrometer of the results reveals a number of interesting points; first, the ( C 0 2 ) l z C H 3 0 H +results are not too dissimilar from those obAr,.X + e Ar,*-X e (5) served for very low-energy charge-transfer or electron-impact processes, which suggests an internal energy for the molecular where Ar,*.X denotes the excitation of an electron to a Rydberg ion of 1-2 eV. Our proposal that (CO,), might be the chargestate. The very poor Frank-Condon overlap between the neutral transfer agent would give the methanol ion an internal energy of and ionic states of the argon clusters rules out the possibility of 2.3 eV. Second, although the Ar,CH30H+ results have some direct i~nization.',~ Following step 5 there are two possible routes features in common with those of the high-energy experiments, to ionization: (a) self-trapping of the positive charge followed the latter, unlike the cluster experiments, all show significant by charge transfer, i.e. CHO' and CH3+ions. Finally, the CHOH' ion from the argon Ar,*.X Ar2e-Ar,-z.X e clusters experiments is almost an order of magnitude more intense (6) than the same ion from any other experimental source. Although Ar,.X' (7) Ar,+-Ar,-2-X the argon and carbon dioxide cluster systems were both ionized with 70-eV electrons, it is evident that there are significant difor (b) direct ionization as observed by Kamke et al.I7 ferences between the cluster-bound product ion distributions and those generated by the 70-eV electron impact ionization of the Ar,*.X Ar,.X+ e (8) isolated methanol molecules. The self-trapping mechanism proposed in step 6 has been observed A more quantitative assessment of the energy content of the in photoexcitation experiments on pure and the very methanol ions can be obtained from a consideration of the discrete stable Ar,' dimer is thought to be responsible for the comparatively electronic states in the ion which could be populated as a result slow rate of hole transport in liquid argon.46 of charge-transfer excitation. The photoelectron spectra of Obviously, there is going to be a difference in energy between B e r k ~ w i t z )and ~ Niwa et al.36938show evidence to the effect that an ion X+ generated by steps 6 and 7 and one generated by step specific electronic states are responsible for the formation of 8. The principal contribution to this difference coming from the particular product ions. The ion CHzOH+ originates primarily 1.3-eV binding energy of the argon dimer ion, Arz+. At this stage from the first excited state band which is centered at 13 eV and we do not have sufficient detailed information on the properties has a half-width of approximately 1 eV. The same band also of these systems to comment on the relative efficiencies of the contributes to the formation of CHOH', as does a further band various steps in the mechanism given above. However, as the size centered at 17.5 eV. Although these two ions have approximately of the cluster increases so we might expect the inert gas component equal heats of f o r m a t i ~ n ?there ~ appearance potentials (AP) are to project a much larger ionization cross section than the resident quite different. The AP of C H 2 0 H + is 11.67 eV whereas that of CHOH' at 12.78 eV37 probably reflects the presence of a reverse activation barrier associated with the rearrangement re(43) Bernard, D. M.; Stace, A. J., to be submitted for publication. action which leads to the ion's formation. In the case of CH,' (44) Laporte, P.; Saile, V.; Reininger, R.; Asaf, U.; Steinberger, I. T.Phys.
-
-
+
+
-
-
+
Rev. 1983, A28, 3616. (45) Reininger, R.; Saile, V.; Laporte, P. Phys. Rev. Lett. 1985, 54, 1146. (46) Spear, W. E.; Le Comber, P. G. In Rare Gas Solids, Klein, M. L., Venables, J. A., Ed.; Academic: London, 1977.
(47) Jones, G. G.; Taylor, J . W. J . Chem. Phys. 1978, 68, 1768. (48) Linn, S. H.; Ng, C.Y . J . Chem. Phys. 1981, 75, 4921.
1514 The Journal of Physical Chemistry, Vol. 91, No. 6, I987
Stace
TABLE II: Energetics and Lifetimes for Methanol Ions Excited by Charge Transfer
----
reaction CH20H+ CH,OH+ + H CH;OH+ CHOH+ + H, CH30H+ CH3++ OH CD30D+ CD,OD+ + D CD30D+ CDOD' + D, CD30D+ CD3++ OD
60'
0.83 1.93 2.86
0.87 1.93 2.90
Ar+'