J. CHEM. SOC. FARADAY TRANS., 1994, 90(7), 935-939 Appearance Energies of Small Cluster Ions and their Fragments Brett R. Cameron, Craig G. Aitken and Peter W. Harland" Chemistry Department , University of Canterb ury , Chris tch urch ,Ne w Zealand Appearance energies of the cluster ions (CO,); (2 d n d 4), (N2,0); (2 < n < 4) and (NH,),,NH: (0 <n < 7), and the cluster ion fragments (N20 * 0)' and (N20. NO)' have been determined by electron impact ionization of neutral clusters formed in a supersonic molecular beam. Results obtained for (CO,); (2 <n d 4), (N20),+ (2 < n < 4), (N,O NO)+ and (NH,),NH,+ (0 < n d 2) are in general agreement with previously reported appear- a ance energies for these species, while the appearance energies of (N20* 0)' and (NH,),NH; (3 < n < 7) have been measured for the first time.Binding energies deduced from appearance energy measurements for the (CO,);, (CO,); and (N,O); cluster ions are observed to be in accord with results obtained using ion-molecule equilibrium methods. Possible mechanisms for the formation of the cluster fragment ions (N20-O)+ and (N20-NO)+ are discussed. Atomic and molecular clusters have been the subject of extensive experimental and theoretical research activity for more than 30 years. The fundamental aim of this research effort has been to gain an improved understanding of the evolution from the atomic or molecular properties of a system to its bulk phase properties with increasing cluster size. For example, atomic and molecular clusters represent small, isolated systems which may be used to test and further our understanding of amorphous solids, catalysis, liquid structure and solvation effects.Despite the intensive research activity involved, however, the characterization of cluster species is still in its early stages. Considerably more particle- specific information will be required in order to develop a complete understanding of the mechanisms of cluster forma- tion and elucidate the factors governing the structure and sta- bility of such species. The use of supersonic molecular beams to generate clusters which are rotationally and vibrationally cooled provides a suitable environment for obtaining this information. The essentially collision-free environment of the molecular beam provides the opportunity to investigate the clusters formed in the supersonic expansion in the absence of any further aggregation and problems relating to particle specificity may be largely resolved through the application of ionization and mass-filtering techniques. In this paper we report an investigation of the appearance potentials for the cluster ions (CO,); (2 < n G 4), (N,O),f (2 < n < 4) and (NH3),NH,' (0 < n < 7), and the cluster ion fragments (N20 * 0)' and (N,O. NO)'.The chemistry of CO, cluster ions is of considerable interest in ionospheric studies of the predominantly CO, atmospheres of Mars and Venus,' while ammonia clusters are of interest with regard to the energetics of gas-phase proton solvation.Accurate appearance energies are required for the determination of binding energies and enthalpy changes associated with various steps in cluster formation. A knowledge of cluster-ion appearance energies may also be used to obtain information on rearrangement processes and internal cluster ion-molecule reactions which may follow from electron impact or photo- ionization of neutral clusters. Appearance energies for (CO,); (2 < n < 4), (N,O)J (2 < n G 8), (N,O.NO)+ and (NH3),NH,' (0 < n < 2) have been previously while those determined for (N,O -0)' and (NH3),NH,' (3 < n < 7) represent new results. Experimental The gas mixture under study was expanded from a high-pressure stagnation reservoir through a commercial electro- magnetic pulsed valve (General Valve Corporation, model 9-181) into the first of two differentially pumped vaccum chambers.The valve was modified by the inclusion of a small stagnation volume between the 0.8 mm orifice in the valve and a 50 pm shaped orifice in the exit plate. This gave higher cluster densities than obtained from a valve fitted with a 50 pm orifice. The central core of the pulsed supersonic expan- sion was sampled by a 1.0 mm skimmer (Beam Dynamics) located cu. 300 nozzle diameters (15 mm) from the nozzle exit. The skimmed supersonic beam was allowed to enter the ion source of a Vacuum Generators SXP300 quadrupole mass filter located 10 cm downstream from the skimmer assembly in the second differentially pumped chamber.The electron energy distribution was estimated to be cu. 0.85 eV full width at half maximum (FWHM). Output pulses from the channel- tron electron multiplier were passed through a high-Q 2 MHz notch filter to eliminate rf pick-up from the quadrupole driver circuitry and amplified with a fast preamplifier fol- lowed by an amplifier and pulse amplitude discriminator combination. The signal-to-noise ratio of the beam signal was optimised using a simple gating arrangement. The TTL output pulses from the pulse-counting preamplifier were split and fed into two and-gates. Using a pulse generator and a pulse delay unit, two 5 V gates of identical width were independently delayed with respect to the nozzle trigger pulse in order to correspond with different regions of the signal pulse envelope.The first window was positioned over the ion arrival time distribution resulting from the pulsed supersonic beam in order to sample signal plus background, while the second was positioned somewhat later in time, sampling only back- ground signal. Output pulses from the two and-gates were counted through a counter-timer. Computer control of elec-tron energy and mass selection was implemented using custom-built 12-bit digit al-to-analogue converters incorpor- ated into the mass spectrometer control unit. The pulsed nozzle was generally operated at a frequency of 10 Hz, with an open time of not more than 2 ms. Background pressures of and Torr were maintained in the expansion chamber and the mass spectrometer chamber during normal operation of the nozzle.The temperature of the nozzle was monitored using a thermocouple attached to the body of the valve. The potential difference across the thermocouple was calibrated and amplified using a simple fixed-gain circuit, the output of which was supplied to one of 16 14-bit analogue-to-digital conversion channels monitored. The reservoir pressure was monitored using an MKS Baratron (loo00 Torr) connected to an MKS type 286 con-troller. Experiments were carried out automatically by scan- ning from low to high and from high to low electron energy for increments of 0.04 or 0.08 eV. Reproducibility from week to week was excellent and the ionization efficiency curves were automatically analysed using a linear least-squares pro- cedure to locate and tabulate the threshold and any breaks in the curves.All of the ionization efficiency curves reported were caljbrated against argon and the molecular ion recorded simultaneously. Results Neutral CO,, N20 and NH, clusters were produced by expanding gas mixtures containing 100 Torr of argon and 500 Torr of CO,, N,O or NH, made up to a total pressure of ca. 4OOO Torr with helium at a reservoir temperature of 295 K. These mixtures were found to produce supersonic beams of sufficiently high cluster content for reliable determi- nation of appearance energies for the cluster ions (CO,); (2 < n < 4), N20i (2 < n d 4) and (NH,),NHZ (0 d n d 7).For the N,O mixture it was also possible to determine the appearance potentials for the cluster ion fragments (N,O. 0)' and (N,O. NO)'. While monomer and cluster speed distributions were not measured, we would expect the parallel translational temperature of the cluster beams to be close to the value of CQ. 6 K measured previously for a pure helium beam under the same source condition^.^ The extent of rotational cooling which occurs during the supersonic expansion of these gas mixtures is unclear and will depend on the efficiency of rotational to translational energy transfer in collisions between the molecules and clusters and the rare-gas atoms. We tentatively suggest that the terminal rotational temperatures of the molecules and clusters will be less than 40 K in all cases. No clusters containing helium or argon atoms were observed for any of the gas mixtures examined.Ion counts were measured at up to 100 points with a typical counting period of 5 s at each electron energy repeat- ed some 15 to 20 times to obtain an average count with an acceptable standard error. Average ion counts were also recorded at an electron energy of 70 eV at the beginning and end of each run. Depending upon the number of ions exam- ined, a run could take up to 3 or 4 h to complete and it was therefore necessary to consider any potential sources of long- term experimental instability that might adversely affect the accuracy of the measurements. In particular, it is known that the cluster content of supersonic molecular beams is highly sensitive to variations in source pressure and .temperature," and these variables were carefully monitored throughout each run.Owing to the small flow of gas through the 50 pm nozzle, the reservoir pressure was observed to drop by not more than 2 or 3% over a 4 h period of continuous operation and no change in nozzle temperature was detected. The semi-log plot has been used in this study to determine the cluster-ion appearance potentials with an estimated accuracy of kO.1 eV in all cases. This error limit includes both statistical and systematic errors. Although nominally less accurate than ideal photoionization measure- ments, reproducibility is excellent and certainly good enough to allow critical comparisons to be made between our results and those of previous experimental and theoretical investiga- tions.The appearance potential of Ar' used as the primary electron energy scale calibrant for all of the appearance potential measurements was taken to be 15.76 +_ 0.01 eV.13 C02 Clusters Illustrative examples of ionization efficiency curves measured for (CO,);, (CO,); and (CO,): are shown in Fig. 1, where every second point has been omitted for clarity. Calibration of the electron energy scale was achieved through a concur- J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 160 140 7v) l2OI100 v)c.5 80-8-$ 60-.-0,-v) 40 -t I *O t L 1 I 1 ,I 11.0 11.5 12.0 12.5 13.0 13.5 14.0 14.5 15. uncorrected electron energy/eV Fig.1 Measured ionization efficiency curves for (0)(CO,);, (0) (CO,): and (A) (CO,):. Every second experimental point has been omitted for clarity. rent measurement of the ionization energy for Ar'. The cor- rected values of the (CO,);, (CO,); and (CO,); appearance energies and the calculated binding energies are shown in Table 1 with literature values for comparison. N20Clusters Ionization eficiency curves for (N20)i (2 < n < 4) and the cluster ion fragments (N200)' and (N,O * NO)' are illus- * trated in Fig. 2 and 3. The appearance energies determined for these species are summarized in Table 2. The value of 12.3 eV determined for the appearance energy of (N20);is in Table 1 Appearance energies (Eapp)and binding energies (E,,) for (CO,): (2 G n G 4) EdeV for (C0,):-CO, ion E.ppleV this work literature (CO,): 13.1 f0.1 0.73 0.675a*b 0.564"*'(W,= 12.8 k 0.1 0.36 0.32b (CO,), 12.6 k 0.1 0.26 0.22d These values have been corrected to 0 K by Linn and Ng4 and are therefore lower than the values stated by the authors.Ref. 14. Ref. 1. Ref. 15. Table 2 Appearance energies (Eapp) and binding energies (EJ for N,O clusters EdeV ion E,JeV this work literature (N,O * O)+ 14.6 f 0.1 (N,O * NO)+ 14.3 f 0.1 17.0 f0.2 (N2O): 12.3 & 0.1 0.6 1 0.56" 0.57'*' w201: 12.1 f 0.1 0.22 (N,O),f 12.0 f0.1 0.12 a Ref. 4. Ref. 18. Corresponds to a temperature of 481 K and may not be directly comparable. J. CHEM.SOC. FARADAY TRANS., 1994, VOL. 90 C 0300, 0L c 250 3 C I s s s-_--_-0 -_ _-44f -10 11 12 13 14 15 16 17 18 uncorrected electron energy/eV Fig. 2 Measured ionization efficiency curves for (0)(N,O):, (0)(N,P); and (A) (N20)f.Every second experimental point has been omitted for clarity. excellent agreement with that of 12.35 f0.02 eV reported by Linn and Ng4 and the value of 12.394 & 0.015 eV reported by Kamke et d6It can be seen from Fig. 4 that the shape of the ionization eficiency curve measured for the (N,O * NO)+ cluster ion fragment is significantly different to the shape of the curves obtained for any of the other ions illustrated in Fig. 1-5. Apart from the ionization threshold at 14.3 eV, there is a sharp change of slope at ca.17.0 eV, suggesting the presence of a second threshold for the formation of this ion. The shape of this curve was found to be totally reproducible. The (N,O*NO)+ion was also observed by Linn and Ng,4 who estimated an appearance energy of 14.01 eV, in reason- able accord with the lowest-energy threshold of 14.3 eV determined in the present study. A second threshold at ca. 17.2 eV is also apparent on the (N,O *NO)+photoionization efficiency curve recorded by Linn and Ng. Although they attempted no interpretation of this, it lends support to our observation of a reproducible, higher-energy threshold. The appearance energy for the cluster fragmentation product 50 -. 3. . 0 -z--O. 0 c--.C C0 3 0. 1 , -12 13 14 15 16 17 18 19 20 uncorrected electron energy/eV Fig. 3 Measured ionization efficiency curves for (0)(N,O.O)+ and (0)(N,O.NO)+, with Ar' (0)as reference. Every second experimental point has been omitted for clarity. 4500 'I , I,, I 3°1 3 OJ 0-4000 0: 0 3500 31 0 0 -3 15001 10001 L 5001 c , /,I 1 7 8 9 10 11 12 13 14 15 16 17 uncorrected electron energy/eV Fig. 4 Measured ionization efficiency curves for (0)NHf, (0) (NH,)NH:, (A) (NH,),NHf and (0)(NH,),NHf. Every second experimental point has been omitted for clarity. (N,O -0)' has not been previously reported. In view of the agreement observed between the appearance energies of (N,O)l and (N,O.NO)+ determined in the present study and those obtained by Linn and Ng4 using photoionization, we might expect the appearance energies of 12.1, 12.0 and 14.6 eV obtained for (N,O)l, (N,O)Z and (N,O.O)+, respectively, to be equally reliable.The appearance energies for (N,O): and (N,O),' reported in the photoionization study of Kamke et aL6 are 12.29 & 0.02 and 12.26 f0.04 eV, respectively, or ca. 0.2 eV higher than the electron impact threshold reported here and listed in Table 2. Although, in principle, photoionization should yield more accurate thresholds with lower uncer-tainty, experimental photoionization data do not often measure up to these expectations. The appearance energies o01 12400 01 0-3-2200 0 2000c 0 0 1800tc C 11 7* 16001 S 140Ob 0 12001 _-..-0 --.g.1000~ CI, .ii 8001 t 600 C 7 8 9 10 11 12 13 14 15 16 uncorrected electron energy/eV Fig. 5 Measured ionization efficiency curves for (0)(NH,),NHf, (0)(NH,),NHf and (A) (NH,),NHf. Every second experimental point has been omitted for clarity. The ionization efficiency curve for (NH,),NHf would be superimposed on that for (NH,),NHf and has been omitted from the figure. The higher signal level for (NH,),NHf over its neighbours, (NH,),NHf (Fig. 5) and (NH,),NHf, reflects the higher stability of this cluster. for the (N,O); (1 < n < 8) cluster ions are reported by Kamke et d6with experimental uncertainties from f0.015 eV for n = 2 to kO.04 eV for n = 8.However, inspection of the experimental data shown in Fig. 1 of ref. 6 for the (N,O)T ions shows little correspondence between the reported values and the thresholds anticipated from the data. The threshold regions are smeared, noisy and the shape of the curves varies considerably from n = 1 to n = 8. The reported thresholds and uncertainties are the result of an empirical multi- parameter fitting procedure, which cannot gurantee a unique solution. So, despite photoionization thresholds quoted to two or three decimal places with uncertainties in the meV range, some consideration must be given to the data treat- ment. Electron impact ionization thresholds do return reli- able values within the stated uncertainty, although it must be acknowledged that recoil energy and internal excitation in the ionization process are folded into the absolute values measured by either technique.Using the appearance energy value of 12.3 eV measured for (N,O)l with the ionization energy of 12.886 f0.002 eV for N2016 and the estimated intermolecular binding energy of 0.02 eV for the neutral dimer,17 the bond dissociation energy of (N20); has been calculated to be 0.61 eV, in good agree- ment with the value of 0.56 eV reported by Linn and Ng4 and the value of 0.57 eV determined by Illies" using ion- molecule methods. Note, however, that this latter value relates to a measurement of AH5 for the association reaction of N20 and N20' at 481 K and therefore may not be directly comparable. The (N20)2+ binding energy calculated using the photoionization data reported by Kamke et d6 would be 0.512 eV, which seems a little low.Assuming the same binding energy of 0.02 eV for (N,O), and (N20)4, we calculate bond dissociation energies of 0.22 and 0.12 eV for (N,O); N,O and (N20)3+N,O, respectively, compared with values of 0.124 and 0.05 eV calculated using the photo- ionization thresholds reported by Kamke et aL6 The appearance energy of 14.6 eV determined for the (N20.0)' cluster ion fragment is observed to be 2.3 eV greater than that of (N,O)l, close to the difference of 2.4 eV between the appearance energy of N20+ from N,O (12.886 ev) and the appearance energy of 0' from N20 (15.29 eV).I6 This observation may be rationalized by writing the structure of (N,O 0)' as 0' * N20 and to postulate that the forma- tion of this species involves the ionization and fragmentation of one of the N20 monomer units in (N,O), without any significant perturbation of the accompanying cluster mol- ecule.Note that the same argument would also apply to the species (CO* CO,)' and (NH, * NH,)+ observed by Stephan et An alternative mechanism proposed by them for the formation of these cluster fragment ions involved an internal ion-molecule reaction. If the ion within the (N,O), cluster is initially formed in some electronically excited state N20+*, then the (N,O.O)+ ion may be produced by the following sequence of reactions: (N,O), + e-+N,O.N,O+* + 2e-+N,O; + N, (1) The N,O; species produced in this manner would be expected to have an appearance energy greater than that of (N20); by at least the additional energy required for the electronic excitation of the parent ion.Irrespective of which- ever mechanism applies, we can use the measured appearance potential of 14.6 eV for (N,O-O)', the thermochemical threshold of 15.29 eV for the formation of 0' from N,016 and the binding energy of 0.02 eV for (N2O)i7 to estimate a lower bound of 0.71 eV for the bond dissociation energy of (N,O * 0)'. The lowest energy threshold observed for the formation of (N20* N0)'corresponds to an appearance energy of 14.3 eV, J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 while the break at higher electron energy equates to an ion- ization threshold of ca.17.0 eV. Note that the differences of 2.0 and 4.7 eV observed between these thresholds and the appearance energy of 12.3 eV obtained for (N20)2f corre- spond very well to the differences of 2.124 and 4.854 eV between the ionization energy of 12.886 eV for N,O' and the thermochemical thresholds of 14.19 and 17.76 eV for the fragmentation processes N,O + e--+NO+(X'E') + N(4S)+ 2e- (2) and N,O + e-+NO+(X 'E') + N(,P) + 2e- (3) respectively.' Applying the same argument used for (N,O.O)', it seems reasonable to write the structure of (N,O NO)' as NO+ -N,O and to describe the formation of this species as involving the ionization and fragmentation of one of the N,O molecules in the neutral dimer without any significant perturbation of the other.Realistically, the mol- ecules making up a cluster must exert an influence on one another. This might well be expected to include a lowering of the ionization energy with increase in cluster size, as observed in this and other studies. This adds support to the mechanism of cluster ionization described above. Linn and Ng4 observed that the photoionization efficiency curve they measured for (N20* NO)' had essentially the same profile as that of NO' produced from the fragmentation of N20+. Such an observa- tion indicates that the fragmentations of N20+ and (N,O)l to form NO' and (N,O -NO)', respectively, follow similar reaction pathways, further supporting the notion that ioniza- tion of the neutral N,O dimer occurs on a single monomer unit to form N,O'.N,O.The neutral monomer in N20' -N,O then acts simply as a spectator in the fragmen- tation process leading to the formation of (N,O. NO)+. Note that N20' may undergo another fragmentation process leading to the formation of NO'(XIZ') and N(,D). The thermochemical threshold for the formation of NO+ by this process would be 16.57 eV,16 suggesting that another break in the ionization efficiency curve of (N,O.NO)' may be expected between 14.3 and 17.0 eV. We were unable to detect this break, although there was some evidence for such a feature on the photoionization efficiency curve measured by Linn and Ng.4 Our failure to observe this feature may be attributed to the low-energy resolution of the instrument employed for the present study.The fragmentation of N20+-N20 may then be viewed as a set of energy-dependent unimolecular cluster ion dissociation reactions as shown in eqn. (4), analogous to eqn. (2) and (3). -14.6 eV + (N,O)O' + N, + 2e-(N,O)N,O + e--14.3 eV -+ (N,O)NO+ + N(4S) + 2e--16.6 eV + (N,O)NO' + N(,D) + 2e--17.0 eV + (N,O)NO' + N(,P) + 2e-(4) NH,Clusters Ionization efficiency curves for the ammonia clusters are shown in Fig. 4 and 5 and appearance energies are listed in Table 3 for (NH,),NH: (0 < n < 7). Discrepancies between binding energies deduced from ion-molecule equilibria and from cluster-ion appearance energies suggest that electron impact and photoionization fail to yield the true adiabatic ionization energies of these weakly bound species.2- Disso- ciation energies of the (NH,),NHd ions deduced from appearance energy measurements were found to be in poor J.CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 Table 3 Appearance energies (Eapp)for (NH,),NHf (0 < n < 7) ion electron impacta photoionizationb 9.7 9.59 i-0.02 9.2 9.15 f 0.04 9.0 9.03 k 0.04 8.9 - 8.85 - 8.81 - 8.78 - 8.72 - ’’ Reproducibility fO.l eV or better (this work). Ref. 8. accord with those obtained by ion-molecule methods, indica- tive of this failure of electron impact ionization measurements to sample the true adiabatic ionization thresholds for the ammonia cluster ions. Conclusion Klots and Compton’ suggested that the equilibrium geometry of van der Waals cluster ions produced by electron impact or photoionization may be considerably different from the equilibrium geometry of the neutral precursor. In such situations the Franck-Condon factors near the true adiabatic ionization threshold may be so small that the observation of the adiabatic threshold is precluded.It has been suggested that small Franck-Condon factors near threshold are not a particularly serious problem in the ion- ization of rare-gas cluster species owing to the close spacing of many Rydberg levels throughout the region between the adiabatic and the direct ionization thresholds which may decay via autoioni~ation.~~’~Rydberg states with lifetimes greater than 50 ps and principal quantum numbers 55 < n < 75 have been reported for CO, clusters by Camp- bell and Tittes.” Such long-lived high Rydberg states can be observed only if there are some states for which non-radiative mechanisms of decay, such as autoionization and electronic predissociation, are significantly slower than radiative decay.It has been shown2’ that predissociation rates of molecular Rydberg states are considerably greater for states of low prin- cipal quantum number and while autoionization is the most probable non-radiative decay mechanism, the apparent absence of states with principal quantum number less than 55 observed in the experiment performed by Campbell and Tittes” indicates that predissociation may also be an impor- tant mechanism.It is therefore possible that lower Rydberg states of molecular clusters may, in fact, predissociate instead of decaying to levels of lower energy uia the autoionization process. For this reason, unfavourable Franck-Condon factors may not be completely compensated for in the ioniza- tion of molecular cluster species, making the observation of their true adiabatic ionization potentials unlikely. Some knowledge of Franck-Condon factors for van der Waals clus- ters may therefore be required for the reliable interpretation of cluster-ion appearance potentials. The most probable mechanism for the formation of the cluster fragment ions (N,O * 0)’ and (N20 NO)’ would appear to involve the ionization and fragmentation of one of the N,O molecules in the neutral N,O dimer without any significant perturbation of the second molecule, although alternative mechanisms cannot be discounted.Despite these recognised deficiencies, electron impact ionization efficiency curves can provide sig- nificant mechanistic information, especially where breaks are found and where comparisons with monomer measurements and data collected using other techniques are available. References 1 M. Mautner and F. H. Field, J. Chem. Phys., 1977,66,4527. 2 C. E. Klots and R. N. Compton, J. Chem. Phys., 1978,69,1636. 3 G. G. Jones and J. W. Taylor, J. Chem. Phys., 1978,68,1768. 4 S. H. Linn and C. Y. Ng, J. Chem. Phys., 1981,75,4921. 5 K. Stephan, J. H. Futrell, K. I. Peterson, A. W. Castleman Jr. and T.D. Mark, J. Chem. Phys., 1982,77,2408. 6 B. Kamke, W. Kamke, R. Herrmann and I. V. Hertel, 2. Phys. D, 1989,11,153. 7 K. Stephan, J. H. Futrell, K. I. Peterson, A. W. Castleman Jr., H. E. Wagner, N. Djuric and T. D. Mark, J. Mass Spectrom., 1982, 44,167. 8 S. T. Ceyer, P. W. Tiedemann, B. H. Mahan and Y. T. Lee, J. Chem. Phys., 1979,70,14. 9 B. R. Cameron and P. W. Harland, J. Chem. Soc., Faraday Trans., 1991,87, 1069. 10 Atomic and Molecular Beam Methods, ed. G. Scoles, Oxford University Press, London, 1988. 11 C. A. McDowell, The ionization and Dissociation of Molecules, McGraw-Hill, New York, 1963. 12 R. W. Kiser, Introduction to Mass Spectrometry and its Applica- tions, Prentice-Hall, Englewood Cliffs, NY, 1965. 13 V. H. Dibeler and R. M. Reese, Adv. Mass Spectrom., 1966, 3, 471. 14 R. G. Keese and A. W. Castleman, J. Phys. Chem. Ref: Data, 1986,15,1011. 15 K. Hiraoka, G. Nakajima and S. Shoda, Chem. Phys. Lett., 1988, 146,535. 16 H. M. Rosenstock, K. Draxl, B. W. Steiner and J. T. Herron, J. Phys. Chem. Ref: Data 6, Suppl. 1, 1977,70. 17 H. L. Johnston and K. E. McCloskey, J. Phys. Chem., 1940,44, 1038. 18 A. J. Illies, J. Phys. Chem., 1988,92,2889. 19 C. Y. Ng, D. J. Trevor, P. W. Tiedemann, S. T. Ceyer, P. L. Kronebusch, B. H. Mahan and Y. T. Lee,J. Chem. Phys., 1977, 62,4235. 20 E. E. B. Campbell and A. Tittes, Chem. Phys. Lett., 1990, 165, 289. 21 S. M. Tarr, J. A. Schiavone and R. S. Freund, J. Chem. Phys., 1981,74,2869. Paper 3/05770D; Received 24th September, 1993