J. CHEM. SOC. FARADAY TRANS., 1994, 90(2), 239-247 Collisional Activation of Large Ions Energy Losses and an Impulsive Collision Theory of Energy Transfer Caroline D. Bradley, Jonathan M. Curtis and Peter J. Derrick* Institute of Mass Spectrometry and Department of Chemistry, University of Warwick, Coventry, UK CV4 7AL Margaret M. Sheil Department of Chemistry, University of Wollongong, Wollongong,NSW,Australia The translational energy losses, A€,experienced by keV beams of singly charged ions of high masses (> 1000 u) in collisions with either an inert-gas atom or hydrogen molecule depend on the mass of the target gas employed. There is no evidence of a dependence on the ionisation energy of the target gas. A€s are similar with He and D, targets, and both are larger than those with Ar.This behaviour is found with organic ions composed of light atoms and with inorganic cluster ions composed of heavy atoms. The measured Us are consistent with internal energy uptake, Q, occurring via direct momentum transfer in an impulsive collision. It is concluded that the Q taken up by an ion is dependent upon the masses of its constituent atoms. In collision with He, a light-atom ion of a given molecular mass takes up more Q than does a heavy-atom ion of similar molecular mass under the same experimental conditions, but the A€ can be similar in the two cases. The development over the past 20 years of techniques for the ion energy and under comparable conditions, the trans-formation of beams of molecule ions and cluster ions' has lational energy losses for ions of different masses showed a opened up possibilities for investigations of the dynamics of direct relationship to incident-ion mass, i.e.were larger for collisions involving very large molecules.2 Collisions involv- heavier ions. Taken as a whole, these findings point over-ing large ions hold intrinsic interest as regards energy transfer whelmingly to a momentum mechanism, rather than elec- and disposal, and understanding such events is directly rele- tronic excitation with minimal momentum transfer. vant to the analytical technique of tandem mass spectrometry Bricker and Russell6 examined more closely the variation which is assuming increasing importance for the determi- in energy losses when using different inert gases as targets, nation of the molecular structure of biomolecules and syn- and reported a linear relationship between the ionisation thetic polymer^.^ In the context of tandem mass energy of the target and the energy loss.On this basis, they spectrometry, collision between incident ions in a beam and argued in favour of excitation of the target gas as a major gas molecules serves to induce dissociation of the ions, so-process during collision. They studied the formation of a called collision-activated dissociation (CAD).3 Using He as fragment ion (m/z 614.2) from protonated chlorophyll-a, m/z the collision gas, there is evidence to suggest that the efi- 893.5. The linear relationship was not universally valid, ciency of the CAD becomes low, if the mass of an incident because with other ions D, behaved as a target gas more like ion mi+ is greater than ca.1500 u given an incident-ion trans- He than Ar (D2has the same mass as He and approximately lational energy of 8 eV (or The original explanation the same ionisation energy as Ar)." The better relationship put foward was that the large ions did not fragment within seems to be between the mass of the target and the energy the time-frame (lo-' s) of the experiment. The long lifetimes loss. That the formation of m/z 614.2 from protonated were attributed to the large numbers of internal degrees of chlorophyll-a is not an exceptional case in any regard has freedom over which excitation energy can be distributed. On been confirmed unequivocally in a careful series of experi- the other hand, it has been suggested that the inefficiency of ments.12 CAD of large ions is more a consequence of weak excitation That the relationship between ionisation energy and energy in the collision.6 loss for the inert gases may be no more than a consequence For many years, the accepted view of CAD of small of ionisation energy of an inert gas being related to mass does organic ions (mass -= 100 u) was that the ion was elec-not in itself disprove a proposal that energy uptake is small tronically excited in the collision, that the translational in collisions of multiatomic ions with light target gases.In energy lost by the ion in the collision was negligible and that which case, the observed dissociation of large ions following there was no direct momentum tran~fer.~ Electronic excita- collisional activation would be a consequence of multiple col- tion was assumed to be followed by internal conversion, and lisions.That is to say, an ion would undergo a number of the ion fragmented by vibrational predissociation.' The collisions and accumulate suficient internal energy for disso- observation that CAD of large ions (the term 'multiatomic' ciation. There is evidence from Fourier-transform ion cyclo- will be employed for the purpose of distinction from polyato- tron resonance spectrometry, to support a mechanism of this mic ions) was typically accompanied by translational energy type in the case of the organic ion valinomycin and low- losses of the incident ions as large as 10-lo2 eV was a clear energy collisions, where the extent of fragmentation was indication of momentum transfer in the c~llision.~~~~' increased on raising the time available for collision and hence The cross-sections for CAD of multiatomic ions were small, and the number of collision^.'^ The alternative view, namely that were consistent with energy uptake by the ion oia direct fragmentations of large ions can be induced by single (keV) momentum The translational energy losses were collisions, has, however, been given support by experiments large for He collision gas, and smaller for heavier target gases at different collision-gas pressures.14*' 'In these experiments, such as Ar.' These trends were observed for biological, the measured energy losses AE associated with certain frag- polymer and inorganic cluster ions.A further significant ment ions were found to be independent of pressure, suggest- trend observed for organic ions was that at a given incident- ing that a single collision induced fragmentation. Assuming that direct momentum transfer occurs, two extreme theoretical treatments can be distinguished. In the one case, the ion can be regarded as a single entity so that the collision becomes quasi-diatomic. Simply conserving energy and momentum then provides the relationship (1) among AE, Q and scattering angle 8 for a single collision. + 2(mi/mg)cos0[1 -(AE/Ei)]1/2 Ei is the translational energy of the incident ion prior to colli- sion and mi and m, the masses of the incident ion and target gas.The earliest AE, for multiatomic peptide ions were related to Q on this basis, making the assumption that scatter was negligible4.59'6 (i.e. 8 = 0). More recently this quasi- diatomic treatment of what is inelastic scattering (Q # 0) has been described as the 'limiting elastic model'.12 The other extreme is where the interaction of the target with the ion is considered to involve only one atom in the ion. The inter- action of this one atom with the other atoms in the ion is considered as a separate subsequent event. On this basis, the following relationships have been derived. ' The constants E and p are given by expressions (3) and (4).'* ma is the mass of the atom in the ion considered to be involved with an atomic target m,.These equations, which will be referred to as the 'impulsive c'ollision transfer (ICT)' model, have been used in considering collision of cluster ion^.'^-^^ (3) (4) 1 In the case of a small target such as He, which of the two extremes is a more plausible model hinges upon the relative strengths of, on the one hand, the interactions between the target and the atom, and, on the other hand, the interactions between atoms within the ion. With an ion composed mainly of hydrogen, carbon, nitrogen and oxygen, and considering a keV collision in which the incident ion loses 10-lo2 eV of kinetic energy, the interactions within the ion are likely to be weaker than the seemingly very strong interaction in the col- lision.This being the case, the impulsive collision in which one atom in the ion is considered to be involved in the colli- sion might be expected to be the more realistic of the two possibilities considered. Molecular dynarnic~~.'~ calculations have supported the impulsive collision model. In the present paper, we report AEs measured for keV collisions of valinomycin molecule ions, containing predomi- nantly hydrogen, carbon, nitrogen and oxygen, and for cluster ions of caesium iodide. AEs have been determined from all sufficiently intense fragment ions in all cases. Our previous studie~~.~,' have, for the most part, presented mean energy losses, i.e. some sort of average over a set of fragment ions from a given incident.Other earlier studies have focussed attention on just one selected fragment ion for a particular incident ion.6.' Comparison between two ions of approximately the same mass, where one consists of a small number of heavy atoms and the other of a large number of light atoms, ought to provide an indication as to which of the two extreme models provides the better description. Valinomycin has a relative molecular mass, M,, of 1110.6 and 498 internal degrees of J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 freedom, whereas (CsI),Cs has an M, of 1172.1 and 21 inter-nal degrees of freedom. Thus in collision, the relative velo- cities and centre-of-mass energies for molecule ions of valinomycin and [(CsI),Cs]+ will be similar, given the same incident-ion energy in each case (vide infra).The global con- straints of conservation of energy and momentum will be approximately the same in the two cases. The dynamics of energy transfer are predicted to be approximately the same if the quasi-diatomic model applies, whereas the ICT model predicts that Q should differ significantly. Whether or not the Qs are similar should be evidenced by the degrees of fragmen- tation. According to Rice-Ramsperger-Kassel-Marcus theory, the species with the smaller number of atoms should exhibit a stronger dependence of rate of fragmentation upon internal energy,23 if critical energies are similar. Experimental All measurements have been made using an unusually large research mass spectrometer (see Fig.l), which has been described previ~usly.~~ Incident ions were selected using a magnetic sector (nominal radius of ion optical axis 800 mm), the resolution of which was sufficient to separate the pure I2C[M + K]+ valinomycin ion from the species containing one 13Catom. The collision cell (Fig. 1) was 10 mm in length, and the surrounding area was differentially pumped. The ion gauges in the vicinity of the collision cell were calibrated for different gases ; conductances and pumping speeds associated with the components of the differential pumping system were determined in separate experiments. The translational ener-gies of undissociated incident ions and of fragment ions were measured using a cylindrical electric sector (radius of ion optical axis lo00 mm).The acceptance angle of the electric sector was &lo(in the plane of deflection) in terms of the direction of travel of ions out of the collision cell. Electric- sector potentials were varied under computer control from a few V to their maximum values (&340 V for ion energy 10.5 keV) in steps of 120 mV. Electric-sector potentials were calibrated against a digital voltmeter of known accuracy. Metastable peaks (i.e.peaks due to fragment ions from spon- taneous unimolecular decomposition in the field-free region between the sectors) of test compounds appeared at their expected positions (to better than 0.1 V). All spectra were recorded using ion counting and signal averaging. The particle-multiplier detector and counting electronics 'floated' at negative potentials of up to 40 kV, in order that ions would be accelerated into the multiplier and detected effi- ciently. Signals were transferred from the counting electronics to the computer via optical fibres.Field desorption (FD) was employed as the ionisation method for valinomycin. The FD emitters were 25 pm tung- sten wires covered in carbon microneedles and had been acti- vated in an atomosphere of ben~onitrile.~' Samples were loaded by repeatedly dipping the emitter into a concentrated sample solution. The emitter was positioned 2 mm from a grounded counter electrode and a positive potential of 10.5 kV applied. A small heating current was passed through the emitter wire to promote desorption.Caesum iodide cluster ions were formed by bombardment of a sample-coated target with 8 keV xenon atoms [fast atom bom bardmen t (FAB)]. In all collision experiments, the incident-ion energies were 10.5 keV. The collision conditions were such that the incident-ion beam was reduced to 3540% of its original intensity when the target gas was admitted into the collision cell. The AEs were obtained from the measured electric-sector potentials corresponding to peak maxima of the fragment ions. Let the electric-sector potential required for transmis- J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 24 1 second field-free region rst field-free region I ?-protective cage II I I I L----I '----J Fig. 1 Schematic of the ion-beam spectrometer sion of the incident ion in the absence of gas be q.If the incident ion mi+ were to decompose to a fragment ion mf+ with no loss of translational energy, the fragment ion would be transmitted at an electric-sector potential V, [eqn. (4)]. V, = (mf/mi)b (4) Suppose that the incident ion mi+ loses translational energy, AE, as a result of collision and subsequently decomposes to the particular fragment ion m,,. This is the so-called 'two-step model'.' A smaller electric-sector potential, called V, , will be required to transmit the fragment ion mf+ formed from an incident ion which has lost translational energy AE. Eqn. (5)shows the relationship of AE to V,. AE = (mJm,)(EJK)(Q-V,) (5) Ei is the translational energy of mi+ prior to collision.If the incident ion mi+ loses energy AE in the collision but does not decompose, the energy-deficient mi+ ion will appear at a lower electric-sector potential, 5.Eqn. (6) shows the relation-ship of AE to 5. AE = [(V, -5)/q]Ei (6) In general, the shift in electric-sector potential (V, -Vd is dif-ferent for different fragment ions, and under a given set of conditions is a reproducible characteristic of any particular fragment. The precision achieved in the determination of AEs depends upon the mass, m,, of the fragment ion measured, relative to the mass, mi, of its parent ion. Provided a peak was smooth and symmetrical, its position, V,, in terms of electric-sector potential could be determined to k0.016 V.This uncertainty arose very largely in determining the cen-troid of a peak. The consequent uncertainties in AE range from f0.5 eV at a mass ratio mf/mi of 0.9 to f5 eV at 0.1. Taking the results for [(CsI),Cs] + with He, the uncertainties arising in this way are 21.0 0.5 eV [(CsI),Cs]', 49.0 0.7 eV [(CsI),Cs]+, 66.7 f0.8 eV [(CsI),Cs]+, 53.9 & 1.0 eV [(CsI),Cs]+, 61.6 f1.4 eV [(CsI),Cs]+ and 57.9 & 2.4 eV [(CsI)Cs]+. The values of the AEs determined (Tables 1-3) are dependent upon experimental conditions, in particular collision conditions and ionization conditions in the source. The values in the tables were, as far as possible, based on measurements made under identical conditions, so that com-parisons can be made among values in the tables for the pur-poses of elucidating the effects of constituent-atom mass upon AE.Results AEs obtained on the basis of expression (5) from measure-ment of the fragment ions of the [M + K]' m/z 1149.7 of valinomycin are shown in Table 1. The molecular structure of valinomycin is shown in Fig. 2. The mass assignments for the fragment ions are based on the four-sector fast atom bom-bardment mass spectrum of the [M + K]+ of valinomycin, which is shown in Fig. 3. Energy spectra [these spectra will be referred to as mass-analysed ion kinetic energy (MIKE) spectra] are shown in Fig. 4. AEs obtained from fragment ions of [(CsI),Cs]+ m/z 1172.1 and also of [(CsI),Cs]' m/z 1951.7 are shown in Tables 2 and 3, respectively.The MIKE spectra of [(CsI),Cs]+ and [(CsI),Cs]' are shown in Fig. 5 and 6, respectively. The contributions to the fragment-ion peaks from metastable decompositions were found to be neg-ligible in the valinomycin case and do not affect the energy loss values. Metastable decompositions contributed to a small number (<20%)of the cluster-ion fragment peaks. With both the [M + K] valinomycin ion and the caesium+ iodide cluster ions, AEs tend to be similar for a given frag-ment ion with either He or D, (Tables 1-3). These AEs tend to be greater than the corresponding vaues for Ar (Tables 1-3). The Qs (Tables 1-3) have been calculated from the cor-responding AEs through eqn. (2)-(4) in the cases of He and Ar. These equations cannot be applied directly, if the incident ion undergoes more than one collision prior to fragmentation.Further, if the incident ion collides and breaks down to a fragment ion and that fragment ion collides and decomposes to give the observed fragment ion, the energy shift from the observed fragment ion will yield an exaggerated AE.26 The large AEs for m/z 132.9 from the caesium iodide cluster are possibly influenced by this latter effect. A factor affecting the interpretation of observed energy losses is the extent of scat-tering as a result of collision. An ion is able to sustain a larger AE while being deviated through a small angle when J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 Table 1 AE and Q (according to ICT theory) (in eV) for valinomycin [M + K]+ m/z 1149.7 colliding with He, D, and Ar fragment ion m,+/u AE(He) Q(W UD,) AE(Ar) QtAr) loss of CH, 1133.7 40.3 14.8 40.9 6.7 5.7 loss of H,CO 11 19.7 58.5 21.6 43.5 19.0 16.2 loss of C3H, 1106.7 38.2 14.1 38.5 18.0 15.4 loss of CH, and C3H7 1091.6 64.8 23.9 67.9 15.4 13.2 [(HVLV),VHV + K]+ -2H 1075.7 39.7 14.6 44.8 27.3 23.3 [(HVLV),VHV + K]+ -CH, + H 1063.7 63.4 23.4 67.2 19.5 16.7 [(HVLV),VLV + K]+ -2H 1047.7 59.2 21.8 76.9 11.4 9.7 [(HVLV),VLV + K]+ -CH3 -H 1033.6 57.5 21.2 33.9 [(HVLV),HV + K]+ 978.6 63.4 23.4 68.3 35.0 29.9 [(HVLV),HV + K]+ -CH3 + H 964.6 47.5 17.5 69.5 19.4 16.6 [(HVLV),VL + K]+ + 2H 952.6 65.7 24.0 72.2 33.6 28.7 [(HVLV),VL + K]' -CH3 + H 936.6 66.2 24.4 71.3 27.3 23.3 [(HVLV),V + K]+ -CH3 + 2H 865.6 82.9 30.6 67.4 25.8 22.0 [(HVLV),L + K]' + H 852.0 104.5 38.6 80.3 45.7 39.0 [(HVLV),L + K]+ -CH3 -C,H7 + 2H 795.5 55.4 20.4 62.3 [(HVLV), + K]+ -CH3 + H 765.5 56.2 20.7 68.6 [(HVLV), + K]+ -CO + H 753.5 76.3 28.2 86.3 [(HVLV)VHV + K]+ -CH3 + 2H 694.5 56.8 21.0 88.4 [(HVLV)HVL + K]+ -CH3 + H 666.4 68.2 25.2 [(HVLV)HVL + K]+ -CO + H 653.5 72.4 26.7 [(HVLV)HV + K]+ -CH3 + H 594.4 65.4 24.1 ~ H : hydroxyisovaleric acid, L: lactic acid, V: valine.Table 2 AE and Q (according to ICT theory) (in eV) for [(CsI),Cs] m/z 1172.1 colliding with He, D, and Ar + C(CSU3CSI+ 912.34 19.2 0.1 20.8 2.3 0.5 C(CsI),Cs!+ 652.53 55.8 0.4 48.9 3.7 0.8 C(CSI)CSl 392.72 69.2 0.5 55.2 4.8 1.oDl+ 132.91 189.5 1.3 35.6 10.4 2.2 Table 3 AE and Q (according to the ICT theory) (in eV) for [(CsI),Cs]+ m/z 1951.7 colliding with He, D, and Ar [(csI),csI+ 1691.76 21.0 0.6 20.8 0.3 0.1 ~(CSI),CSl+ 143 1.95 49.0 1.4 43.1 6.5 1.5 C(C~~)'aCSl+ 1172.15 66.7 1.9 49.0 7.5 1.7 [(CSI) 3CSI + 9 12.34 53.9 1.5 45.5 13.4 3.0 C(CsI),Cs!+ 652.53 61.6 1.7 44.3 24.1 5.4 C(CSI)CSl 392.72 57.9 1.6 40.9 31.2 7.0 CCSI + 132.91 65.8 14.7 Val Lac Va I HYV VaI Lac -CH-NH-C-CH-o -C -CH -NH -C-CH -? 0-CH-C-NH-CH-C-0-CH-C-NH-CH-C-0-CH-C-NH-CH-6 I I1 I I1 I II I II I I1 I 11CH 0 CH 0 CH, 0 CH 0 O Oti&/ 'CH, H,C /\ CH, H,C/ 'CH, H,C CH, H,C CH, Hyv Val Lac Val HYV Val Fig.2 Structure of the cyclic depsipeptide valinomycin (M, = 11 10.6). Hyv = hydroxyisovaleric acid, Lac = lactic acid. J. CHEM. SOC.FARADAY TRANS., 1994, VOL. 90 100-n s W >-4-.-v) al .-E 50-0, .--U 0-Fig. 3 Four-sector tandem mass spectrum of valinomycin [M + K]' ions formed by fast atom bombardment less massive target gases are employed.'2*'7 When the range of observable scattering angles is fixed, as in our experiments, the range of observable AE will be somewhat smaller in the case of Ar than with He or D2.Under the experimental con- ditions used, incident ions of valinomycin, [(CsI),Cs] and+ [(CsI),Cs]+ would not have been lost in the ion optical plane as a result of a single collision with either He or D2. The variations in AE from one fragment ion to another for a given target (Tables 1-3) reflect the dynamics of the com- peting fragmentation channels. The collision of a multiatomic ion with the target will give rise to a distribution of internal energy states of the ion.Different states and total energies apparent m/z , . . , I, . . , . , .-.-qw.. , . , . , , . , , . . . , , 200 300 ESA potentialp n apparent m/z -.-'-----' '".I ' ' ' ' '.' ' ' I ' '.' ' ' ' 200 300 ESA potentialp favour different fragmentation pathways, because the ener- getics of different fragmentation pathways differ from each other. Consequently, it is to be expected that the Qs associ-ated with formation of individual fragment ions and hence the AEs manifested will differ from one fragment ion to another. The AE observed represents the sum of the recoil energy of the target-gas atom and the internal energy uptake Q of the ion.The recoil energy of a light target such as He (or D2)can be greater than that of the heavier target Ar. The Qs could be similar in all three cases, even though the observed AEs are greater for the lighter targets. For the valinomycin ion, the fragmentation patterns observed in the spectra (Fig. 4) are broadly similar for all three target gases. There are ---'I "' 'I ' ' ', I ' ' p600 800 1000differences in the qualities of the spectra due to a lower frag- mentation efficiency observed for Ar, i.e. the number of frag- ment ions measured compared to the total number of apparent m/z -,----7-------r-----, . , Y., 1 . . . . . , . , . . . . , 200 300 incident ions entering the collision cells was 5% as compared with 10% for He and D,.The lower fragmentation efi- ciencies with Ar may be due solely to scattering but there could be other contributing factors such as charge exchange. For [(CsI),Cs] +,the intensities of the lighter fragment ions relative to [(CsI),Cs]+ (Fig. 5) are higher in the case of Ar than with He or D,. The difference is more pronounced for [(CsI),Cs] (Fig. 6). Again overall fragmentation efficiencies + are greater with He than with Ar, which may be attributed to scattering effects. The ICT theory has been used to calculate Qs associated with different fragment ions, using expressions (2)-(4) and the measured AEs (Tables 1-3). The calculations have been per- formed for the atomic targets.Use of the expressions requires a decision as to the mass of the interacting atom within the ion which is involved in the collision. In the case of valinomy- cin, the values of Q in Table 1 have been calculated with a hypothetical atom of mass 6.803 u (average of masses of ESA potentialp Fig. 4 Field desorption MIKE spectra of valinomycin [M + KJ+ ions with (a) helium, (b)deuterium and (c) argon as target gases atoms in valinomycin). For [(CsI),Cs] and [(CsI),Cs) +,+ again the averages of masses of the atoms were considered. The calculations show the Qs for an ion composed of pre- dominantly light atoms (valinomycin) to be large in most cases (tens of eV) and to be similar overall whether the target be He or Ar. The means of the Qs in Table 1 are 23 eV and 20 eV for He and Ar, respectively. The Qs for the 'heavy atom' cluster ions (Tables 2 and 3) are much smaller, although the AEs are of similar magnitude to the valinomy- cin case. With one or two exceptions, the calculated Qs for fragment ions from valinomycin exceed 10 eV and the major- ity exceed 20 eV (Table 1).With one exception, the Qs from caesium iodide clusters are < 10 eV and the majority are only J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 n -1000 300 ESA potentialp 500 1000 apparent m/z 300 ESA potentialp (c1 ,-500 1000 apparent m/z r-~--r-vr.-,-, , . . , , . , . , , . , . , . . , . . , . . , -200 300 ESA potentialp Fig. 5 Fast atom bombardment MIKE spectra of [(CsI),Cs]+ ions with: (a)helium, (b)deuterium and (c) argon as target gases ca.1 or 2 eV (Tables 2 and 3). For the clusters, the calculated Qs are also lower on the whole with He than with Ar. The means of the Qs for [(CsI),Cs] + in Table 2 are 0.6 eV and 1.1 eV for He and Ar, respectively; the means of the Q[(CsI),Cs]+ in Table 3 are 1.5 eV and 4.8 eV for He and Ar, respectively. The quasi-diatomic model makes predictions significantly different from those of ICT theory. The quasi-diatomic model considers that, for a given AE, Q depends on the mass of an ion, but not on the mass of its individual constituent atoms. A consequence of this is that for two incident ions of the same mass, where one is composed of heavy atoms and the other of lights atoms, the predicted Qs will be identical for the apparent m/z 19.2 (b)In s W ESA potentialp .--:I ? 1000 1500 i 1 apparent m/z -7.1.... .1.. .I 200 300 ESA potentialp 32.4 (c)1 L1 500 1000 1500 apparent m/z 7 .'...'!..I..--.--....-1 .'I..', I....'.'. t 1 200 300 ESA potentialp Fig. 6 Fast atom bombardment MIKE spectra of [(CsI),Cs]+ ions with: (a) helium, (b)argon and (c) hydrogen as target gases same AE. In Tables 4 and 5, Qs are given for valinomycin and the caesium iodide cluster ions, calculated with 8 = 0. The quasi-diatomic model indicates large internal Qs in most cases for both types of parent ion and further indicates that the Qs are in most cases larger for a given fragment ion with He than with Ar.Discussion The similarity among the fragmentation patterns of the valin- omycin [M + K]' ion found with He, Ar and D, as target gases (Fig. 4) indicates that the internal energies of the parent ions decomposing to the observed fragments are similar to J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 Table 4 AE and Q calculated using the quasi-diatomic model with 8 = 0" for valinomycin [M + K] m/z 1149.7+ fragment ion loss of CH, loss of H,CO loss of C3H7 loss of CH, and C3H, [(HVLV),VHV + K]' -2H [(HVLV),VHV + K]+ -CH3 + H [(HVLV),VLV + K]+ -2H [(HVLV),VLV + K]+ -CH3 -H[(HVLV),HV + K]+ [(HVLV),HV + K]+ -CH3 + H [(HVLV),VL + K]' + 2H [(HVLV),VL + K]+ -CH3 + H [(HVLV),V + K]+ -CH, + 2H [(HVLV),L + K]+ + H [(HVLV),L + K]+ -CH, -C3H7 + 2H [(HVLV), + K]' -CH3 + H [(HVLV), + K] -CO + H [(HVLV)VHV + K]+ -CH3 + 2H [(HVLV)HVL + K]+ -CH3 + H [(HVLV)HVL + K]' -CO + H [(HVLV)HV + K]' -CH, + H m,+b AE(He) QW) WD2) Q(D2) AE(Ar) Q(Ar) 1133.7 40.3 30.0 40.9 29.3 6.7 6.7 1119.7 58.5 35.1 48.5 30.6 19.0 18.7 1106.7 38.2 28.1 38.5 28.1 18.0 17.8 1091.6 64.8 36.3 67.9 35.7 15.4 15.3 1075.7 39.7 28.7 44.8 31.2 27.3 26.8 1063.7 63.4 35.7 67.2 35.7 19.5 19.2 1047.7 59.2 35.1 76.9 35.7 11.4 11.3 1033.6 57.5 34.4 33.9 25.5 978.6 63.4 35.1 68.3 37.0 35.0 34.2 964.6 47.5 3 1.9 69.5 36.3 19.4 19.1 952.6 65.7 35.7 72.2 36.3 33.6 32.8 936.6 66.2 35.7 71.3 35.7 27.3 26.8 865.6 82.9 35.1 67.4 36.3 25.8 25.3 852.0 104.5 29.3 80.3 35.7 45.7 44.3 795.5 55.4 34.4 62.3 35.7 765.5 56.2 34.4 68.6 37.0 753.5 76.3 35.7 86.3 35.1 694.5 56.8 35.1 88.4 35.1 666.4 68.2 35.7 653.5 72.4 35.7 594.4 65.4 36.3 Table 5 AE and Q calculated using the quasi-diatomic model with 8 = 0"derived for fragment ions of [(CsI),Cs] m/z 1951.7+ fragment ion m,+ IU AE(He) Q(W [(CsI),csI + r(csI),csl+ C(CsI),CsI+ C(CSU3CSI 1 C(CSI),cS~ C(CSI)CSl 1172.15 169 1.76 1431.95 912.34 652.53 392.72 66.7 21.0 49.0 53.9 61.6 57.9 14.7 15.9 20.4 20.4 16.6 18.5 CCSI + 132.91 each other.This conclusion is strengthened by the recent dis- covery that fragmentation of cationised valinomycin follow- ing collisional activation can vary dramatically, depending upon the method of formation of the parent The centre-of-mass collision energies for valinomycin [M + K] + colliding with He, Ar and D, are 36, 353 and 36 eV, respec- tively.The AEs vary greatly from one fragment ion to another, but for He and D, most fall in the range 35-60 eV and for Ar many fall below 20 eV. It is pointed out that the AEs falling outside these ranges tend to be for the more minor fragmentations. The strongest fragmentations are loss of C,H;, which has AE of 38.2, 38.5 and 18.0 eV for He, D, and Ar, respectively, and loss of 16 u (presumably CH,), which has AE of 40.3, 40.9 and 6.7 eV for He, D, and Ar, respectively. If both He and Ar deposit the same internal energy, this represents a greater proportion of the centre-of- mass collision energy in the case of He.ICT reproduces these trends satisfactorily. The greater efficiency of internal energy uptake for He is predicted, arising from the better mass-match between the light atoms in the organic ion and He (compared with Ar).'7,28 Considering loss of C,H;, the Q with He is 14.5 eV and with Ar is 15.6 eV according to ICT (Table 1). If for loss of C,H; and He AE is 38.2 eV and Q is 14.5 eV, the recoil energy of He is the difference between, them, i.e. 23.7 eV. The recoil energy for Ar in the case of decomposition by loss of C,H; is 2.4 eV. The recoil energy for He is an order of magnitude greater than that for Ar, reflecting the direct momentum transfer supposed by ICT.The critical energies for the decompositions of the valinomy- cin are not accurately known, but most will be several eVs. The compound is cyclic, and many of the observed fragmen- tations necessitate rupture of two, or more, covalent bonds. WD,) QPz) AE(Ar) Q(W 20.8 15.9 0.3 0.4 43.1 21.0 6.5 6.4 48.9 21.0 7.5 7.4 45.5 21.7 13.4 13.2 44.4 22.3 24.1 23.4 40.9 21.0 31.2 30.1 65.8 60.7 Thus, Qs of 10-30 eV are not unreasonable given the com- pound's high number of internal degrees of freedom. The quasi-diatomic model explains the valinomycin results in Table 1 well enough. The calculated Qs are, on the whole, larger for He than for Ar. Considering loss of C,H;, the He figure is 28.1 eV and the Ar figure is 17.8 eV.The calculated Q would fall if non-zero scattering angles were considered, and the dependence of the calculated Q on scattering angle 8 would be stronger in the case of He [expression (l)]. The fragmentation patterns (Fig. 5 and 6) of the caesium iodide cluster ions are not identical. In the case of [(CsI),Cs] , intensities of the lower-mass fragments relative + to that of [(CsI),Cs]+ are higher with Ar than with He. This suggests that Qs are higher with Ar than with He, as loss of CsI to form [(CsI),Cs)+ is the lowest-energy fragmentation pathway and formation of other fragment ions requires higher internal energies. The fragmentation pattern with H, as the target gas (Fig.6) is consistent with this view, in that the relative intensities of the lower-mass fragments are still lower than with He. The centre-of-mass collision energy with H, is half that with He (11 eV as compared with 21 ev), and the mass-match (see below) is less favourable for H, than with He. Thus, it is reasonable to suppose that H,, while perhaps behaving similarly to He, would lead to smaller Qs than would He. The same trend for lower-mass fragments to have lower intensities relative to the CsI-loss fragment [(CsI),Cs]+ with He, compared to Ar, is evident from the spectra of [(CsI),Cs]+ (Fig. 5). Again, the implication is that less Q is taken up in collision with He than with Ar. According to ICT, caesium iodide clusters ought to take up internal energy more efficiently in collision with argon than with helium, because the mass-match of caesium (133 u) is J.CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 Table 6 AE and Q calculated using the quasi-diatomic model with 8 = 0" for fragment ions of [(CsI),Cs] m/z 1172.14+ ~~ ~ II(CsWsl+ 912.34 19.2 16.6 C(CSI)2CS!+ 652.53 55.8 33.8 392.72 69.2 35.7[(CsWICCSl + 132.91 189.5 65.0 better with argon (40 u) than with helium (4 u).17i28The mea- sured fragment ions (Fig. 5 and 6) are associated with AEs generally comparable in size to those observed with valin- omycin [M + K]' in the case of He (Tables 1-3); with Ar the AEs are smaller with the salt clusters than with valinomy- cin. The Qs derived from these AEs by means of ICT are smaller for the cluster ions compared to the organic ion, by something like an order of magnitude for both He and Ar (Tables 1-3).The Qs calculated by ICT for the salt clusters are for the most part less than 1 or 2 eV, which is consistent with the observed fragmentations. The caesium iodide clus- ters are small systems [(CsI),Cs] + and [(CsI),Cs] have 39+ and 21 internal degrees of freedom, respectively, if judged by the standard of valinomycin [Sol degrees of freedom of (M + IS)+]. The binding energies of the clusters are not known as they must depend on cluster structure, but they could be tenths of an eV for loss of CsI are are unlikely to exceed 1.5 eV (for loss of CsI) even in the case of the most stable str~ctures.~~,~~ The cluster ions also undergo sponta- neous unimolecular fragmentation (i.e.in the absence of colli- sion gas). Deposition of a few eV of internal energy into such systems could be expected to induce a significant degree of fragmentation. In the case of [(CsI),Cs]+, ICT indicates that the Qs associated with different fragment ions are larger with Ar than with He for the smallest fragment ions, much the same for Ar and He for [(CsI),Cs]+ and [(CsI),Cs]+ and larger with He in the case of the major fragment [(CsI),Cs]'. These values are not inconsistent with the fragmentation patterns (Fig. 6). In the case of [(CsI),Cs]+, ICT indicates the Qs (derived from measured AE) to be larger with Ar for all frag- ment ions (Table 2), which is consistent with the fragmenta- tion patterns (Fig.5). The quasi-diatomic model does not easily account for the experimental results with the caesium iodide cluster ions. The calculated Qs (Tables 5 and 6) are unacceptably high in most cases. Setting the scattering angle 0 to some finite value (rather than zero) has the effect of reducing the calculated Qs, but the reduction occurs for both the cluster ions and the organic ion. When the magnitudes of Q for the clusters are reasonable, those for the organic ion become unacceptably low. Conclusion AEs measured for the organic valinomycin [M + K]+ ion and for the [(CsI),Cs] ion of similar mass (1 172 u as com- + pared with 1150 u for the organic ion) are of comparable magnitudes in the two cases, both with He and with Ar as the collision gas.AEs measured for the [(CsI),Cs]+ ion are still closer in value to those of the organic ion. AEs are in general greater with He than with Ar, both with the organic ion and the cluster ion. From consideration of the measured fragmen- tation patterns, there is evidence that He and Ar impart similar internal energies Q to the organic ion, but that Ar imparts somewhat more internal energy than He to the cluster ions. These experimental findings are reproduced satisfactorily by ICT. According to ICT calculations, the internal energies 20.8 17.9 2.3 2.3 48.9 31.9 3.6 3.6 55.1 34.4 4.8 4.8 35.6 27.4 10.4 10.3 taken up by the organic ion are similar with He and Ar.The larger AEs with He are accounted for by the predicted larger recoil energies of He. According to ICT calculations, the cluster ions' Qs are smaller by an order of magnitude than those acquired by the organic ion, both with He and Ar. The recoil energies are greater in the case of the cluster ions, hence AEs are as high as with the organic ion. The quasi- diatomic model does not distinguish between an organic ion of mass 1150 u and a caesium iodide cluster ion of similar mass, because only the total mass of the ion is considered in the model. To explain the spectral differences between the organic ion composed of over 150 atoms and the cluster ion of similar total mass composed of nine atoms necessitates that consideration be given to the masses of the constituent atoms in each case.We are pleased to acknowledge the financial support of the Australian Research Council, Kratos Analytical and the SERC. References 1 ion Formation from Organic Solids,ed. A. Hedin, B. U. R. Sun- dqvist and A. Benninghoven, Wiley, Chichester, 1990. 2 S. C. Davis, P. J. Derrick and Ch. Ottinger, Z. Naturforsch. A, 1990,45, 1151. 3 Tandem Mass Spectrometry, ed. F. W. McLafferty, Wiley, New York, 1983. 4 G. M. Neumann and P. J. Derrick, Org. Mass Spectrom., 1984, 19, 145. 5 G. M. Neumann, M. M. Sheil and P. J. Derrick, 2. Naturforsch. A, 1984,39,584. 6 D. L. Bricker and D. H. Russell, J. Am. Chem. SOC., 1986, 108, 6174. 7 MetastabZe ions, ed. R. G. Cooks, J. H. Beynon, R.M. Caprioli and G. R. Lester, Elsevier, Amsterdam, 1973. 8 J. Durup, Recent Developments in Mass Spectrometry, ed. K. Ogato and T. Hayakawa, University Park Press, Baltimore, 1970, p. 92 1. 9 R. G. Gilbert, M. M. Sheil and P. J. Derrick, Org. Mass Spec- trom., 1985,20,43 1. 10 M. M. Sheil, R. G. Gilbert and P. J. Derrick, Advances in Mass Spectrometry, 1985, ed. J. F. J. Todd, Wiley, Chichester, 1986, p. 1161. 11 M. M. Sheil, Ph.D. Thesis, University of New South Wales, Aus- tralia, 1987. 12 A. L. Alexander, P. Thibault and R. K. Boyd, J. Am. Chem. Soc., 1990,112,2484. 13 M. Guilhaus, M. M. Sheil and P. J. Derrick, Org. Mass Spec-trom., 1990, 25, 671. 14 C. D. Bradley and P. J. Derrick, Org. Mass Spectrom., 1991, 26, 395. 15 C.D. Bradley and P. J. Derrick, Org. Mass Specrrom., 1993, 28, 390. 16 M. S. Kim, Org. Mass Spectrom., 1991,26, 565. 17 E. Uggerud and P. J. Derrick, J.Phys. Chem., 1991,95,1430. 18 H. J. Cooper, P. J. Derrick, H. D. B. Jenkins and E. Uggerud, J. Phys. Chem., 1993,97,5443. 19 M. F. Jarrold and J. E. Bower, J. Chem. Phys., 1992, %, 9180. 20 R. C. Mowrey, M. M. Ross and J. H. Callahan, J. Phys. Chem., 1992,%, 4755. 21 M. F. Jarrold and E. C. Honea, J. Am. Chem. SOC., 1992, 114, 459. 22 M. F. Jarrold and E. C. Honea, J. Phys. Chem., 1991,95,9181. J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 247 23 24 W. Forst, Unimolecular Reaction Theory, Academic Press, New York, 1973. (a) D. E. Rogers, Ph.D. Thesis, La Trobe University, Australia, 1980; (b) D. E. Rogers, P. G. Cullis, G. M. Neumann and P. J. Derrick, Ado. Mass Spectrom., 1980, 8, 1729; (c) P. G. Cullis, Ph.D. Thesis, University of New South Wales, Australia, 1988 ; (d) M. G. Darcy, D. E. Rogers and P. J. Derrick, int. J. Mass Spectrom. Ion Phys., 1978,27,335. 27 28 29 30 J. M. Curtis, C. D. Bradley, P. J. Derrick and M. M. Sheil, Org. Mass Spectrom., 1992,21, 502. E. Uggerud and P. J. Derrick, 2. Natuiforsch. A, 1989,44,245. H. J. Hwang, D. K. Sensharma and M. A. El-Sayed, Chem. Phys. Lett., 1989, 160, 243. H. J. Hwang, D. K. Sensharma and M. A. EI-Sayad, Phys. Reo. Lett., 1990,64, 808. 25 26 T-W. D. Chan, A. W. Colburn, D. S. Alderdice and P. J. Derrick, Int. J. Mass Spectrom. Ion Proc., 1992,107,491.M.M. Sheil and P. J. Derrick, Org. Mass Spectrom., 1988, 23, Paper 3/05394F; Received 8th September, 1993 429.