Impulsive Excitation of Cr(CO)6+ during Surface ... - ACS Publications

Oct 1, 1994 - ... M. T. Rodgers , Emilio Martínez-Núñez , Srirangam V. Addepalli and William L. .... Samuel B. Wainhaus, Eric A. Gislason, and Luke...
0 downloads 0 Views 936KB Size
J. Phys. Chem. 1994, 98, 10913-10919

10913

Impulsive Excitation of Cr(C0)6+ during Surface-Induced Dissociation at Organic Monolayers John A. Burroughs, Samuel B. Wainhaus, and Luke Hanley* Department of Chemistry, d c 1I I , University of Illinois at Chicago, Chicago, Illinois 60607-7061 Received: June 15, 1994; In Final Form: August 19, 1994@

20 to 120 eV Cr(C0)6+ ions are scattered off heptafluorobutyrate and hexanethiolate monolayers adsorbed on Ag( 1 1l), and the fragment ions formed by surface-induced dissociation (SID) are detected. The relative fragment ion intensities are monitored to produce SID breakdown curves, which are subsequently compared with threshold photoelectron photoion coincidence (TPEPICO) data to estimate the internal energy of Cr(C0)6+ following the surface collision. The kinetic to internal energy transfer is more efficient for the fluorocarbon than the hydrocarbon monolayer. The scattered ion velocities are centered near 3000, 4000, and 5000 m/s for initial Cr(C0)6+ velocities of 5100, 7800, and 9800 m/s, respectively (30, 70, and 110 eV). A three-step mechanism for SID is proposed in which ions initially undergo impulsive excitation by collision with the surface, inelastically reflect off the surface, and finally dissociate unimolecularly. The experimental kinetic to internal energy efficiencies are fit to an impulsive excitation model which quantitatively duplicates the dependence on both the initial kinetic energy and the effective mass of the adsorbate, at the lower collision energies.

I. Introduction When a polyatomic ion collides with a solid surface, it can dissociatively scatter, nondissociatively scatter, stick to the surface, or implant into the bulk. Dissociative scattering or surface-induced dissociation has been observed experimentally for ions of small c l u ~ t e r s , ~organic - ~ molecules,8-12 and biom~lecules.'~-'~ Surface-induced dissociation (SID) is sometimes used to probe the structures of polyatomic ions in analytical mass spe~trometry.~-' 1,13-15 These and related phenomena have been explored theoretically for ionic and neutral cluster-surface c o l l i s i ~ n s . ~ JPolyatomic ~-~~ ion-surface collisions also play an important role in the growth and etching of materials both directly and through their presence in plasma environments.21s22Polyatomic ion projectiles have been used in secondary ion mass spectrometry to enhance secondary ion yields with respect to atomic ion projectile^.^^^^^ Despite this considerable level of interest, many of the basic phenomena which accompany polyatomic ion-surface collisions remain poorly understood. A paucity of mechanistic studies is especially evident for the chemically bound polyatomic ions of interest to analytical mass spectrometry (Le., organics and biomolecules). Of fundamental importance to polyatomic ion-surface collisions is the kinetic to internal energy transfer which leads to the dissociation of the projectile ion when it rebounds off the surface. The chemical and structural diversity of ion-surface pairs which undergo SID at hyperthermal (10-1000 eV) collision energies imply that they may share a common mechanism of kinetic to internal energy transfer. Several mechanistic pathways for this energy transfer have been proposed, including impulsive excitation, vibrational excitation, or deformation;1,5,25 rotational excitation; and charge transferinduced electronic excitation. 1-4~12 However, only in the cases of diatomic and cluster ions have experiments identified the excitation mechanism.1-6s26 Unambiguous data in the case of chemically bound polyatomics larger than three atoms are particularly lacking. Charge transfer-induced electronic excita@

Abstract published in Advance ACS Abstracts, October 1, 1994.

0022-365419412098-10913$04.5010

tion seems an unlikely candidate for the dominant mechanism since SID of a wide variety of ions with different ionization potentials occurs at electronically disparate surfaces, including clean metal^,'-^,^* graphite,5x6metal oxides,lo semicond~ctors,4~~~ and organic layers.9~28-30Nonetheless, electronic excitation leading to dissociation is clearly important at least in the case of diatomic ionic scattering off clean metal and semiconductor surface^.^-^ Rotational excitation would seem to depend upon the geometry of the incident ion' and is probably inefficient for large ions. Vibrational or impulsive excitation of polyatomic ions is generally indicated by the similarity of the to s time scale of the near-surface interaction of a hyperthermal polyatomic ion to the vibrational period of its low-frequency deformational modes ( ,given in eV, for 70 eV Cr(C0)6+ scattered off the hydrocarbon surface.

I-

--

I

solid = H. hollow = F

V = m/z 136, Cr(C0):

.....

v

L

-

0 = m/z 52, Cr' 0 = m/z 80, Cr(C0)' v = m/z 108, cr(c0);

100

.b E 60 e

~

20

120

Figure 2. SID breakdown curve of the relative intensity of Cr(CO),+ (x = 0-6) fragment ions versus initial Cr(CO)s+ kinetic energy, for scattering off the hydrocarbon surface.

4-

70

1

.u

LLI

.e-

9.14

-

e

H

'

E 50-

60

-

4-

9,lO

CI

L

.e-

6.86

6.21

*y

...... ....

9...*.,0/

.-E

..

E

.w

40

0 E

d

20

0 20

40

60 80 100 ion Kinetic Energy (eV)

120

Figure 3. SID breakdown curve of the relative intensity of Cr(CO),+ ( x = 0-6) fragment ions versus initial Cr(C0)6+kinetic energy, for

2000

4000

6000

8000

Velocity (m/s)

Figure 5. Scattered ion velocity distributions from 30 eV Cr(C0)6'

scattering off the fluorocarbon surface.

collisions with the hydrocarbon (solid points) and the fluorocarbon surfaces (open points).

observations are consistent with a previous study using peffluoro carboxylic acids adsorbed on silver which concluded that the carbon backbone was directed at -40" from the surface normal with the carboxylate head group symmetrically bound to the silver as a bridging ligand.43 We conclude that our fluorocarbon monolayers are bound in the same manner, resulting in a C,F,terminated surface. We examined the preparation of hexanethiolate monolayers on Ag( 111) from hexanethiol vapor in a previous study,'''' and so only the salient results are summarized here. We found that the S-H bond is cleaved when hexanethiol is adsorbed on Ag(11l),forming a saturated monolayer of the thiolate. The carbon backbone of the hexanethiolate was found to be oriented near the surface normal, albeit with a higher degree of disorder and a different saturation coverage than monolayers prepared from solution. Thus, our hydrocarbon monolayers are terminated with C,H, groups. B. Fragmentation of Cr(CO)s+. The breakdown curve for Cr(C0)6+ colliding with the hydrocarbon monolayer is shown in Figure 2. Relative intensities of Cr(CO),+ (x = 0-6) species resulting from these collisions are measured as a function of incident ion kinetic energy. The general profile of the SID breakdown curve shows similar behavior as the Cr(C0)6+ TPEPICO data.39 Figure 3 illustrates the Cr(C0)6+ SID breakdown curve when the fluorocarbon monolayer is used as the collision target. This profile also resembles the TPEPICO breakdown curve. There are subtle distinctions between the two SID breakdown curves which indicate a difference in the

kinetic to internal energy transfer characteristics, as will be discussed below. The angular dependence of energy deposition is examined by rotating the collision target about its axis and detecting the scattered ions with the mass spectrometer, for a fixed incident to detector angle of 90'. Figure 4 shows how varying the angle of incidence from 25" to 65" during 70 eV Cr(CO)s+ impact off the hydrocarbon affects the relative intensities of Cr(CO),+ fragments. As the angle of incidence increases, the Cr+ (mlz 5 2 ) ion signal increases and the Cr(C0)2+ (mlz 80) intensity declines. This behavior denotes an increase in kinetic to internal energy transfer as indicated by the internal energy values noted above each angle in Figure 4 (see below for determination of internal energies). C. Velocity Distributions of Scattered Ions. The kinetic energy of the ions leaving the surface region are measured using the Bessel box on the mass spectrometer and then converted to scattered ion velocities. For Figures 5-7, the data collected from the fluorocarbon surface are displayed as open points whereas solid points are used for data from the hydrocarbon surface. The curves drawn through these points result from second-order fast Fourier transform smoothing of the data to remove statistical noise and enhance the structure observed in the distributions. Figure 5 exhibits the velocity distributions of Cr(CO),+ ions exiting the surface following 30 eV collisions. Cr+, Cr(CO)+, and Cr(CO)Z+ are detected at this collision energy, but counting statistics for the Cr+ ions are not sufficient to allow presentation

10916 J. Phys. Chem., Vol. 98, No. 42, 1994

Burroughs et al.

1

fi

2000

2000

4000

6000

8000

Velocity (m/s)

Figure 6. Scattered ion velocity distributions from 70 eV Cr(C0)6+ collisions with the hydrocarbon (solid points) and fluorocarbon surfaces (open points). Initial Velocity '9800 m/s

2000

solid = ti hollow = F

4000 6000 Velocity (m/s)

8000

Figure 7. Scattered ion velocity distributions from 110 eV Cr(C0)6+ collisions with the hydrocarbon (solid points) and fluorocarbon surfaces

(open points). of this minor channel from the hydrocarbon surface. The average velocity for any given Cr(CO),+ species is -3000 d s , much lower than the initial velocity of -5100 d s . The velocity of the ions leaving the hydrocarbon monolayer is, on average, greater than that of ions leaving the fluorocarbon monolayer. Figure 6 illustrates the scattered ion velocities which result from 70 eV Cr(C0)6+ collisions with the two surfaces. The initial velocity of the projectile ions is -7800 d s , and the velocity distribution for each fragment ion is centered near 4000 d s . There is little difference here in the velocities of ions between the hydrocarbon and fluorocarbon surfaces. Additionally, peaks in the velocity distributions are detected at 70 eV collision energy which are not evident at 30 eV. When the incident ion energy is raised to 110 eV, only Cr+ and Cr(CO)+ ions are detected. Figure 7 shows the velocity distributions of these fragment ions. The peaks which were first noticeable at 70 eV kinetic energy are much more pronounced at 110 eV. Initial ion velocities at 110 eV are -9800 d s . Again, all detected scattered ions have similar velocities, centered close to 5000 d s . Note that the velocity distributions (fwhm) widen as the projectile ion kinetic energy is increased. The effect of changing the incident angle on the velocity distributions of a representative fragment ion is displayed in Figure 8: the Cr(CO)+ ion intensity is plotted against its velocity at 25", 45", and 65" for 70 eV Cr(C0)6+ SID off the hydrocarbon surface. As the incident angle is increased, the

m/z = 80, Cr(C0) -t

fi;n;Ezity

4000 6000 Velocity (m/s)

8000

Figure 8. Scattered ion velocities of Cr(CO)+ ions from 70 eV Cr(CO)6+ collisions with the hydrocarbon surface at 25", 45", and 65" incident angles.

ions coming off the surface have less kinetic energy. Again, we see no less than two peaks in the velocity profiles. The general trend seems to be that ions are concentrated in the lowenergy peak at high angles of incidence whereas they are distributed into a high kinetic energy peak at lower angles. D. Calculation of Average Internal Energy. The average internal energy, < Q> ,is estimated by fitting our SID breakdown curves to similar breakdown curves from threshold photoelectron photoion coincidence (TPEPICO) data reported in the literat ~ r e The . ~ ~TPEPICO breakdown curves in the literature39are reported as ionizing photon energies versus fragment ion abundances. These data have been digitized by scanning the published figure into a computer file. The photon energies of the TPEPICO data are then converted to internal energies by subtracting out 8.24 eV for the ionization energy of Cr(C0)639 and 0.35 eV for the internal energy. The 0.35 eV of internal energy originates from the electron impact ionization process and has been estimated by assuming similar Franck-Condon factors as with photoionization, as determined from the reported half-width half-maximum of the Cr(CO)s+ photoionization peak.39 The corrected TPEPICO data are fit by a simplex algorithm to our breakdown curves (Figures 2 and 3) to give the final values. This fitting procedure assumes that the internal and kinetic energies of the CO and Cr(CO),+ fragments are the same for TPEPICO and SID. Taking into account the day-to-day fluctuations, our approximation of initial internal energy, the TPEPICO data digitization, and the fitting process, we expect that the estimated values are accurate to within -0.5 eV, as shown in Figure 9 by the error bar. Figure 9 shows the estimated < Q > values as a function of ion kinetic energy, where the hydrocarbon data are plotted as open squares and the fluorocarbon data as solid squares. The lines drawn near the data points correspond to our impulsive excitation model predictions, as will be explained below. Previous studies using Cr(C0)6+ ions collided off self-assembled monolayer surfaces found that the average internal energy depositions from 25 eV collisions were roughly 2.8 and 4.7 eV off stainless steel and fluorinated surfaces, r e s p e c t i ~ e l y .Our ~~ values for < Q > at 20 eV collision energy are 3.6 eV from the hydrocarbon monolayer and 5.0 eV from the fluorocarbon. Our energy transfer efficiencies are -18% for the hydrocarbon and -25% for the fluorocarbon, in comparison with the literature values of 12% and 19% off similar surfaces.29 This discrepancy most likely arises from the different methods used to estimate the internal energies: we use TPEPICO data whereas the literature values were estimated from an algorithm using the appearance potentials of the Cr(CO),+ fragments.

J. Phys. Chem., Vol. 98, No. 42, 1994 10917

Impulsive Excitation of Cr(C0)6+

I

0

0:.

I/

ov 0

M, = 2a M , = a5 M,= 169 Hydrocarbon Surface Fluorocarbon Surface

1

'

20

40 60 80 100 Ion Kinetic Energy (eV)

120

Figure 9. Estimated average intemal energy, , as a function of incident ion kinetic energy for Cr(CO)6+scattered off the hydrocarbon (open squares) and fluorocarbon surfaces (solid squares). Predicted values from the impulsive collision model are shown by solid (Ms= 85) and dashed (Ms= 169) lines. The most accurate bond energies of chromium carbonyl ions have been measured using collision-induced d i s s ~ c i a t i o nand ,~~ one might ask why we have chosen to utilize instead the previous TPEPICO data. In a TPEPICO experiment, a photon of known energy is absorbed by the molecule and an electron is ejected with low energy. If the molecule is large, like Cr(CO)6+, we expect complete randomization before dissociation occurs. We expect similar behavior in the SID experiments, provided the assumption of unimolecular dissociation remains valid. Furthermore, even though photoionization deposits narrower internal energy distributions into the ion than does SID, this fitting method should still estimate the correct average internal energy. The distribution of internal energies is broader in collision-induced dissociation, and the breakdown curves do not resemble those of TPEPICO and SID, which are actually quite similar. While Cr-CO bond energies may be more accurately calculated from the CID data, we are interested here in determining the absolute Cr(C0)6+ internal energies. Such internal energy calculations are best done in the fashion described above rather than by creating an algorithm to reprocess the bond energies back into the internal energies.*

IV. Discussion A. Summary of Results. Surface-induced dissociation (SID) breakdown curves for Cr(CO)a+ scattered off fluorocarbon and hydrocarbon monolayers are measured from 20 to 120 eV and found to closely resemble the TPEPICO breakdown curve. By fitting our data to the TPEPICO results, we calculate the kinetic to internal energy transfer during SID. Consistent with previous studies, we find that the kinetic to intemal energy transfer is greater for fluorocarbon than for hydrocarbon monolayer^.^^ This energy transfer is also greater for ions reflected toward the surface normal than toward the surface tangent, for a fixed scattering angle of 90". The scattered ion velocities of the fragment ions are centered around 3000,4000, and 5000 m / s for initial Cr(CO)6+ velocities of 5100,7800, and 9800 m/s, respectively (30,70, and 110 eV). The scattered ion velocities are approximately independent of the mass of the fragments, although the distribution profiles do show structure which is dependent upon the fragment mass, the incident velocity, and the incident angle. B. Three-Step Mechanism of Surface-Induced Dissociation. We propose here a three-step mechanism for SID in which polyatomic ions initially undergo impulsive excitation by collision with the surface, then inelastically reflect off the

surface, and finally undergo unimolecular dissociation in the gas phase. We show that the impulsive step in this model quantitatively duplicates our experimental values for the kinetic to internal energy transfer at the lower collision energies. Finally, we use the inelastic reflection and unimolecular dissociation steps to qualitatively explain the other features of our data. Molecular dynamics and trajectory calculations of clusters indicate that the excitatiodscattering event is a concerted process in which impulsive excitation and reflection occur nearly s i m u l t a n e ~ u s l y .However, ~ ~ ~ ~ ~ ~the ~ ~ concept ~ ~ ~ of sequential multiple collisions is applied in hyperthermal atomic ion and neutral scattering off surface^^^-^^ both to allow analytical calculation of kinetic energy losses and to present a conceptually simple picture of what are actually complex kinetic events. Similarly, it has been argued that noble gas cluster scattering off surfaces can be described as a sequence of atom-surface and atom-atom binary collision^.^^^^^ By utilizing a similar approach, we strive to present a predictive and intuitive model for SID of chemically bound polyatomic ions. C. Impulsive Excitation at the Surface. An impulsive excitation model has been developed which predicts that the energy transfer during the collision of a polyatomic ion and a target gas atom depends upon their masses and that of the excited atom imbedded in the ion.32333We apply this model to describe the average intemal energy, , imparted to a polyatomic ion during a surface collision as

where Ei is the initial kinetic energy of the ion, Ma is the mass of the atodfhnctional group in the ion excited during the collision, Ms is the effective mass of the surface, and Mionis the mass of the entire ion. The effective mass of the surface used here replaces that of the collision gas atom in the original CID model. One important consequence of this model is that increases linearly with the incident ion energy. This is evident in Figure 9 where the solid and dashed lines are predictions from this impulsive model. In each case, the Ma value is set at 28 Da for the mass of a CO group on Cr(C0)6+. An M, value of 85 Da corresponding to CH3(CH2)5 is used to generate the solid line whereas an Ms of 169 Da for CF3(CF2)2 is used to generate the dashed line. When so applied, eq 1 fits our experimental data well except at the higher initial kinetic energies (see below). Attempts to fit our data to an excited atom mass corresponding to that of 0, 2C0, and 3CO either do not fit as well or require unrealistic effective surface masses. Data for the inelastic scattering of atoms and molecules off liquid fluorocarbon and hydrocarbon surfaces have used similar effective surface masses of 109 and 75 Da, respectively, when single collisions were assumed.47 These calculations indicate that the efficiency of the kinetic to internal energy transfer increases with the effective mass of the surface, as previously s u g g e ~ t e d . ~ These % * ~ results imply that Cr(C0)6+ dissociates because one CO molecule collides elastically with the surface and then rebounds off the Cr atom, followed by rapid energy randomization and then dissociation. The C-0 stretch is probably not effectively excited in the surface collision because, with a stretching frequency of 2000 cm-l, it is too stiff. In contrast, the relatively floppy Cr-C stretches and bends are easily excited since they mostly occur near 400 cm-1.37 A similar argument can justify the large effective surface masses used to fit our data: only low-frequency bending modes of the adsorbate are easily deformed by the surface collision, whereas the C-H, C-F, and C-C stretches are too high in frequency. Alternatively, the large effective

10918 J. Phys. Chem., Vol. 98, No. 42, 1994 surface mass may be correlated with an enhancement of the surface impact area which results from the 45" incident angle: the requirement that only one CO is excited could be maintained under these circumstances by the need for near head-on collisions to effectively excite a C r C O stretch. The impulsive excitation model described here is completely analogous to the "reverse hard cube" behavior initially identified by trajectory calculations for ArJPt( 111) s ~ a t t e r i n g . ' ~Reverse *~ hard cube behavior is characterized both by a decrease in the fragment ion velocity as the incident angle increases (see Figure 8) and a peak in the fragment ion intensity near specular (data not shown) for a fixed scattering angle of 90". This behavior has been explained p r e v i o u ~ l yin~ the ~ ~ ~following ~ manner: C r C O collisions inside the Cr(C0)6+ following the initial impulsive excitation tend to scatter fragments toward the surface tangent with higher velocity and those toward the surface normal with lower velocity. This is exactly the behavior observed in Figure 8, although we cannot make an exhaustive comparison with the trajectory calculations since we cannot independently rotate our detector angle. Similar {collision energy/atom}/{bond energy} ratios for our experiments and the Ar trajectory calculations help explain the appearance of reverse hard cube behavior in these otherwise dissimilar systems. Yet more evidence for the impulsive model comes from the data presented in Figure 4, where the experimental value increases by 3.3 eV as the incident angle is increased from 25" to 65". For the same incident angular change, the kinetic energy of the scattered Cr(C0)6+ decreases by a roughly equal amount, 3.1 eV (calculated from Figure 8 by assuming the scattered Cr(CO)6+ has the same velocity change as the detected Cr(CO)+ ion). This transfer of excitation from internal to kinetic energy clearly supports impulsive excitation. Although the impulsive model fits well with the estimated values for up to -70 eV collisions, the experimental observations deviate from those predicted at higher kinetic energies (Figure 9). Two possible explanations for this deviation are offered: either the kinetic energy of ejected CO molecules released from the excited Cr(C0)6+ differs for SID and TPEPICO or the unimolecular dissociation assumption breaks down because the cluster is dissociating at the surface. Another point of variation is the presence of an abrupt step in the estimated between 7 and 9 eV (Figure 9). This step is artificial and is caused by the difficulty in obtaining an accurate simplex fit of the SID data to the TPEPICO data since the latter display a relatively complex fragmentation pattern at this point. Data at kinetic energies of 55 and 75 eV for the fluorocarbon and hydrocarbon monolayers, respectively, support this assertion. D. Inelastic Reflection Off the Surface. The impulsive collision model does not accurately predict the observed kinetic energy losses or account for the 90" reflection off the target: it incorrectly predicts both much smaller energy kinetic losses which are strongly dependent upon the surface mass and much smaller scattered angles (calculations not shown). We therefore invoke a second mechanistic step, inelastic reflection off the surface, which causes the projectile ion to be further slowed and deflected through 90" without additional internal excitation. Attempts to model our kinetic energy losses by the hard cube have been unsuccessful. instead, the kinetic energy losses can qualitatively described by the reverse hard cube behavior described previously; l7vZ6 the scattered ion energies depend significantly upon the atom-atom collisions which occur in the scattered polyatomic ion following the surface collision. An extreme example of this behavior is observed in the case of c 6 0 + scattering off graphite and diamond: the scattered ion kinetic energies are effectively invariant over the 50-450 eV

Burroughs et al. collision energy range, being completely dominated by the internal energy released from the compressed cluster when it springs from the s u r f a ~ e . ~ - ~ ~ ~ ~ One can still refer to the second step in the SID process as inelastic scattering since all atoms in the projectile must reflect at 90" off either the surface atoms or other atoms in the projectile, in the case of the upper half of the projectile which never directly contacts the surface. During this inelastic scattering event, all the atoms in the projectile lose roughly equal amounts of kinetic energy to the surface. A clearer picture of this step awaits trajectory or molecular dynamics calculations on this system. E. Unimolecular Dissociation away from the Surface. The last step in our mechanism describing SID is unimolecular dissociation far from the surface. Given the 10-13-10-14 s time scale that the ion and surface are in intimate contact, the -100 ps time scale of dissociation for Cr(C0)6+ (estimated for gasphase collision-induced d i s s ~ c i a t i o n )is~ far ~ too long for the process to occur at the surface. Instead, the ion must fragment during its flight from the surface to the detector entrance (calculated to be 4-20 ps, within 1 order of magnitude of the literature dissociation time). The primary evidence supporting unimolecular dissociation is that at each collision energy, the different mass fragment ions have roughly the same average velocity. If the Cr(C0)6+ ions dissociated on the surface, the velocities would be expected to differ according to the mass of the fragment ions. F. Structure in Scattered Ion Velocity Distributions. Perhaps the most intriguing data collected here are the scattered ion velocity distributions. Peaks appear in these distributions which are most pronounced for higher initial velocities. We suggest that these peaks result from the ejection of one or more CO molecules from the Cr(C0)6+ ions. Since we are only detecting ions along one axis, the Cr(CO),+ species have an equal probability of ejecting CO in the forward or reverse direction with respect to the mass spectrometer. Those ions ejecting COS in other directions serve to broaden out the distribution or, in more severe cases, will not be detected since the ions will be forced off axis by momentum conservation. Support for this argument comes from the fact that these peaks become enhanced as the incident ion kinetic energy is increased. One expects that as the projectile ion energy is raised, the scattered Cr(C0)6+ ions will have more internal energy to dissipate, causing the CO species to be thrown off with more energy. Also, since there will be some distribution in the kinetic energy of each ejected CO, the width of the distributions should become wider as the energy is increased, as observed. An alternative explanation for the appearance of peaks in the velocity distributions involves the geometry of approach of the incident ion to the surface. Results from trajectory calculations for (260' scattered off diamond surfaces indicate that the initial position and orientation of the ion dominates the outcome of the collision process.20 This may also be true for Cr(C0)6+ SID off organic monolayers. The Cr(C0)6+ ions may strike the surface with any of three different orientations: a direct CO hit, two CO groups aligned towards the surface, or across the face of the octahedron with three CO groups. However, this argument seems to be in contradiction with the incident angle-induced shift in velocity distribution from one peak to another which is observed in Figure 8. Furthermore, alternate collision geometries would be expected to yield the same peak structure for different fragment ions, which is not observed here. Yet another alternative explanation for the velocity peak structure is that the highest velocity peak derives from single collisions with the surface, the next peak from double collisions,

Impulsive Excitation of Cr(C0)6+ and so forth. However, multiple collisions would also be expected to yield the same peak structure for different fragment ions. By these arguments, we conclude that the structure in the velocity distributions results from CO ejections during unimolecular dissociation, although an unassailable conclusion awaits trajectory calculations. G. Considerationof Charge Transfer Effects. Despite the above arguments, it is conceivable that charge transfer effects may play a role in the SID process. Charge exchange between the projectile and the surface have previously been observed for a number of Most incident ions in these experiments are neutralized at the surface, and it is possible that a portion of the scattered ion intensity results from reionization of same. While we have fit our data to a kinematic model, no attempt has been made to exclude a charge transfer mechanism. Nevertheless, the ability of the proposed model to fit the majority of our experimental results indicates that, at least for Cr(C0)6+ scattered off organic surfaces, an impulsive excitation model is primarily responsible for the dissociation event. Any effects of charge transfer appear to be secondary to those of the impulsive excitation. Furthermore, the properties which lead to impulsive excitation will always be present in cases where charge transfer-induced electronic excitation is feasible, indicating that both mechanisms play a significant role in most polyatomic ion-surface pairs.

V. Conclusions From experiments scattering 20- 120 eV Cr(C0)6+ ions off hydrocarbon and fluorocarbon monolayers, we have drawn several conclusions regarding the mechanism of surface-induced dissociation. A three-step model for SID has been proposed which consists of impulsive excitation of the ions via the surface collision, inelastic reflection from the surface, and unimolecular dissociation in the gas phase. The experimentally measured kinetic to internal energy efficiencies have been fit to an impulsive excitation model which quantitatively duplicates the dependence on both the initial kinetic energy and the effective mass of the adsorbate, at the lower collision energies. Inelastic scattering as the second mechanistic step in SID is qualitatively supported by data on scattered ion velocities and angular dependences, which cannot be explained solely by impulsive excitation. Time scale considerations along with the invariance in the average fragment ion velocity with mass indicate that fragmentation is unimolecular, at least for the lower initial kinetic energies studied here. While our three-step model for SID is consistent with all experimental observations in this system, its further verification awaits trajectory calculations and new experiments utilizing other projectile ions and surfaces.

Acknowledgment. We thank Eric Gislason for many thought-provoking discussions and suggestions, without which this paper would not have been possible. We also thank Tianlan Zhang and Douglas Ridge of the University of Delaware for providing the concept and software for fitting our breakdown curves to the TPEPICO data. Finally, we acknowledge the National Science Foundation both for specific support of this project (CHE-9220393) and for the support of L.H. by a Young Investigator Award (1994- 1999). References and Notes (1) van Slooten, U.; Andersson, D. R.; Kleyn, A. W.; Gislason, E. A. Surf:Sci. 1992, 274, 1. (2) Rechtien, J.-H.; Harder, R.; Henmann, G.; Nesbitt, A,; Tellioglu, K.; Snowdon, K. J. S u Sei. ~ 1993, 282, 137. (3) Okada, M.; Murata, Y. Surf:Sei. 1993, 283, 41

J. Phys. Chem., Vol. 98, No. 42, 1994 10919 (4) Martin, J. S.; Greeley, J. N.; Moms, J. R.; Feranchak, B. T.; Jacobs, D. C. J. Chem. Phys. 1994, 100, 6791. (5) Busmann, H.-G.; Lill, Th.; Reif, B.; Hertel, I. V.; Maguire, H. G. J. Chem. Phys. 1993, 98, 7574. (6) Yeretzian, C.; Hansen, K.; Beck, R. D.; Whetten, R. L. J. Chem. Phys. 1993, 98, 7480. (7) See: Nucl. Instr. Meth. Phys. Res. B 1994, 88. (8) Cooks, R. G.; Ast, T.; Mabud, Md. A. Int. J. Mass Spectrom. Ion Processes 1990, 100, 209 and references therein. (9) Somogyi, A.; Kane, T. E.; Ding, J.-M.; Wysocki, V. H. J. Am. Chem. Soc. 1993, 115, 5275. (10) Dagan, S . ; Amirav, A. J. Am. SOC. Mass Spectrom. 1993, 4, 869. (1 1) Williams, E. R.; Jones, Jr., G. C.; Fang, L.; Zare, R. N.; Garrison, B. J.; Brenner, D. W. J. Am. Chem. Soc. 1992, 114, 3207. (12) Wu, Q.; Hanley, L. J. Phys. Chem. 1993, 97, 2677. (13) Cole, R. B.; LeMeillour, S.; Tabet, J.-C. Anal. Chem. 1992, 64, 365. (14) McCormack, A. L.; Somogyi, A.; Dongre, A. R.; Wysocki, V. H. Anal. Chem. 1993, 65, 2859. (15) Wright, A. D.; Despeyroux, D.; Jennings, K. R.; Evans, S.; Riddoch, A. Org. Mass Spectrom. 1992, 27, 525. (16) Cheng, H.-P.; Landman, U. Science 1993, 260, 1304. (17) Xu, G.-Q.; Bemasek, S. L.; Tully, J. C. J. Chem. Phys. 1988, 88, 3376. (18) Even, U.; Schek, I.; Jortner, J. Chem. Phys. Lett. 1993, 202, 303. (19) Markovic, N.; Pettersson, J. B. C. J. Chem. Phys. 1994,100,3911. (20) Blaudeck, P.; Frauenheim, Th.; Busmann, H.-G.; Lill, T. Phys. Rev. B 1994, 49, 11409. (21) Winters, H. F.; Cobum, J. W. Surf:Sei. Rep. 1992, 14, 161. (22) Hsieh, H.; Averback, R. S.; Sellers, H.; Flynn, C. P. Phys. Rev. B 1992, 45, 4417. (23) Appelhans, A. D.; Delmore, J. E. Anal. Chem. 1989, 61, 1087. (24) Park, M. A., Cox, B. D.; Schweikert, E. A. J. Chem. Phys. 1992, 96, 8171. (25) Miller, S. A,; Riederer, D. E., Jr.; Cooks, R. G . ;Cho, W. R.; Lee, H. W.; Kang, H. J. Phys. Chem. 1994, 98, 245. (26) Xu, G.-Q.; Holland, R. J.; Bemasek, S. L.; Tully, J. C. J. Chem. Phys. 1989, 90, 3831. (27) StJohn, P. M.; Whetten, R. L. Chem. Phys. Lett. 1992, 196, 330. (28) Wu, Q.; Hanley, L. J. Phys. Chem. 1993, 97, 8021. (29) Moms, M. R.; Riederer, D. E., Jr.; Winger, B. E.; Cooks, R. G.; Ast, T.; Chidsey, C. E. D. Int. J. Mass Spectrom. Ion Processes 1992, 122, 181. (30) Pradeep, T.; Miller, S. A.; Cooks, R. G. J. Am. SOC. Mass Spectrom. 1993, 4,769. (31) McLuckey, S. A. J. Am. SOC. Mass Spectrom. 1992, 3, 599 and references therein. (32) Uggemd, E.; Derrick, P. J. J. Phys. Chem. 1991, 95, 1430. (33) Cooper, H. J.; Derrick, P. J.; Jenkins, H. D. B.; Uggerud, E. J. Phys. Chem. 1993, 97, 5443. (34) Thibault, P.; Alexander, A. J.; Boyd, R. K. J. Am. SOC.Mass Spectrom. 1993, 4 , 835. (35) Thibault, P.; Alexander, A. J.; Boyd, R. K.; Tomer, K. B. J. Am. SOC. Mass Spectrom. 1993, 4, 845. (36) Sigmund, P. J. Phys. B: At. Mol. Phys. 1978, 11, L145. (37) Khan, F. A.; Clemmer, D. E.; Schultz, R. H.; Armentrout, P. B. J. Phys. Chem. 1993, 97, 7978 and references therein. (38) Castoro, J. A.; Rucker, P. V.; Wilkins, C. L. J. Am. SOC. Mass Spectrom. 1992, 3, 445. (39) Das, P. R.; Nishimura, T.; Meisels, G. G. J. Phys. Chem. 1985, 89, 2808. (40) Zhang, T.; Ridge, D. P. Proceedings of 41st American Society on Mass Spectrometry Conferences on Mass Spectrom. May 31-June 4, 1993, p177. (41) Allen, J. D., Jr.; Durham, J. D.; Schweitzer, G. K.; Deeds, W. E. J. Electron Spectrosc. Relat. Phenom. 1976, 8, 395. (42) SIMION V. 4.0 obtained from Dahl, D. A,; Delmore, J. E., Idaho National Engineering Laboratory, P.O. Box 1625, Idaho Falls, ID, 83415. (43) Chau, L.-K.; Porter, M. D. Chem. Phys. Lett. 1990, 167, 198. (44) Burroughs, J. A.; Hanley, L. Anal. Chem., in press. (45) Kasi, S. R.; Kang, H.; Sass, C. S.; Rabalais, J. W. Surf: Sci. Rep. 1989, 10, 1 and references therein. (46) Amirav, A.; Cardillo, M. J.; Trevor, P. L.; Lim, C.; Tully, J. C. J. Chem. Phys. 1987, 87, 1796. (47) Saecker, M. E.; Nathanson, G. M. J. Chem. Phys. 1994,100,3999.