J. Phys. Chem. 1995, 99, 15488-15496
15488
Elaboration of an Impulsive Model for Collision-Induced Dissociation: Application to CS2+ Ar S+ CS Ar
+
-
+
+
Ronald E. Tosh,* Ani1 K. Shukla, and Jean H. Futrell* Department of Chemistry and Biochemistry, University of Delaware, Newark, Delaware 19716 Received: March 15, 1995; In Final Form: May 16, 1995@
The crossed-beam method provides uniquely valuable information for deducing details of energy transfer, reaction mechanisms, and broad features of energy disposal in chemical reactions. This crossed-beam study of the dynamics of elastic scattering of CS2+ and the lowest energy dissociation channel CS2+(X2rI,) S+(4S,) C S ( X P ) on collision with argon at 17.8 eV collision energy provided the impetus for developing a “knockout” model for collision-induced dissociation. The occurrence of a ridge of intensity of S+ originating within and extending outside the elastic scattering circle is quantitatively rationalized by this model. Symmetry of the scattering ridge about a point on the relative velocity vector and the radius of the ridge which connects maxima in relative intensity ranging from zero scattering angle to 44” (LAB)are quantitatively connected by the model to the bond dissociation energy of CS2+. The model requires that the potential curve describing the interaction of S+ and CS be essentially “flat” when the momentum exchange collision with Ar occurs. Consequently, we infer that transfer of kinetic energy into internal energy of CS2+ equal to its bond dissociation energy occurs in the approach trajectory. It is suggested that this involves the electronic excitation CS2+(X’rI,) CS2+(B2c).
-
+
-
I. Introduction The pioneering series of papers Zdenek Herman published with the late Richard Wolfgang dramatically changed our thinking about mechanisms of ion-molecule reactions.’-4 These first studies of the reaction dynamics of proton-transfer reactions using the crossed molecular beam chemical accelerator EVA5 showed unequivocally that the mechanism was direct and did not involve formation of a randomized complex, even at very low collision energies. They elaborated a model for this class of direct reactions4which they described as “polarizationreflection” to distinguish this from Henglein’s “spectatorstripping” mode16v7 to account for the fact that the sharply forward peaked distribution of the protonated product was shifted to higher velocity than predicted by the spectator stripping model. This results from the polarization force of attraction on the reactants exceeding the polarization force of attraction between the products, resulting in a net acceleration clearly evident in the reaction contours. A similar impulsive reaction model advanced by Chang and Lights reproduced the essential features of angular scattering reported experimentally. The same kind*of classical model has been shown to describe the reaction dynamics of other proton-transfer ion-molecule
reaction^.^
+
Several beam studies of atom-diatomic collisions, A BC, have provided compelling evidence for impulsive scattering in which A strikes only one of the atoms, B, in the molecule (C is then a spectator). In such collisions, it is possible for the final velocity of A to lie on a sphere whose center is the velocity of the A/B center of mass (CM). Observations of such scattering are reviewed by Brenot and Durup-Ferguson.’O Another example of this type of process is presented in the single-beam study by Martin and co-workers,Ii which provides evidence of impulsive scattering that results in electronic excitation of the diatomic particle. Among the earlier measurements of BC* recoil velocities that are consistent with this mechanism were those reported by Mahan and co-workersi2 and by Fayeton et al.I3 ~~
@
Abstract published in Aduunce ACS Abstrucrs, October 1, 1995.
Extensions of the model to include interactions with the third body C are briefly reviewed in the article by Brenot and DurupFerguson, and an extensive list of references to such work may be found in the recent theoretical study by Song et al.I4 Some of this work concerns reactive collisions between A and BC, and in this connection we cite the recent interesting study by Loesch and Moller’5in which impulsive scattering concepts are invoked to explain recoil velocity correlations in the products that result when the orientation of reactant BC is controlled. Molecular beam methods have provided equally valuable reaction dynamics information for characterizing fundamental mechanisms of energy transfer for collision-induced dissociation (CID) reactions of polyatomic ions. The single-beam study by Mahan and co-workersI6 of the dissociation of diatomic and triatomic cations of oxygen and nitrogen on collision with He at relative kinetic energies 1-3 times their bonding energy concluded that a stripping model best described their results. For methane and propane cations, Herman et al.I7 demonstrated that a quite different mechanism, a two-step model in which the collision activates the ion and it subsequently dissociates into products, provides a better description of the dynamics of CID of more complex molecular ions. The current status of our understanding of CID reaction dynamics and the central role of the two-step model in interpreting energy transfer and dissociation dynamics have been reviewed recently.’* Recent studies in our laboratory of the collisional activation and dissociation of the carbon disulfide cation have demonstrated a significant departure from predictions based on the two-step model. A particularly significant observation is scattering of S+ outside the elastic scattering circle with distinctive symmetry. This has led us to carry out the present research and develop a classical kinematic model directly inspired by the pioneering work in ion-molecule reaction dynamics just described. The model involves collisional scattering of a sulfur ion by an argon atom with the CS neutral acting as a spectator and analyzes the dynamics of the three product particles to deduce the energetics consequences of interpreting experimental results using this model. The model
OO22-3654/95/2099-15488$09.00/0 0 1995 American Chemical Society
Impulsive Model for Collision-Induced Dissociation resembles in some respects the knockout model described by Mahan and co-workersI6 and rejected by them on energetic grounds. It will be shown that this model presents a very satisfying interpretation of the title reaction at a collision energy roughly 3.5 times the S+--CS bond energy. Moreover, the present work furnishes perhaps the first observation of a system in which the recoil sphere of a fragment ion is concentric with that of the target neutral. As we shall argue below, this permits a complete specification of the final product velocities. The CID of the carbon disulfide cation is interesting from several perspectives. Pninslow and Armentr~ut'~ have measured cross sections for formation of S+ and CS+ in CID with Xe at and a few electronvolts above the thermochemical thresholds for forming these products and have demonstrated that relatively accurate bond energies may be deduced from this kind of singlebeam experiment. They report the formation of S+(4S,) occurs at the thermochemical threshold to break the S+-CS bond even though it is spin-forbidden. The cross section is less than 1 A2 at 16 eV (linear extrapolation suggests that the total cross section should be about 0.8 A2 for the experiments reported here if we assume Ar and Xe have similar behavior as collision partners), very much less than the geometric cross section of 68 A2 for repulsive c ~ l l i s i o n s . 'The ~ rather small cross section for this reaction was logically attributed to the fact that it is a spin-forbidden transition. Sohlberg and Chen20have interpreted the Prinslow and Armentrout results using an RRKM model to describe the dissociation of the excited CS2+ cation at threshold and over a range of about 4 eV. However, highly nonstatistical behavior sets in above 7.8 eV collision energy. Several intriguing features2',22in the CID of this cation prompted us to undertake an investigation of its reaction dynamics using the crossed-beam method. We recently reported23the general dynamics of CS2+ CID with Ne and Ar at several collision energies and have shown that the two product channels exhibit very different reaction dynamics and highly nonstatistical dissociation behavior. This earlier paper should be consulted for a general summary of CID dynamics for this system. The results and conclusions reported here are an extension of that study in which we have focused our attention on understanding the apparently superelastic scattering of S+ which occurs at a collision energy of 17.8 eV and accounts for about 30% of the observed reaction dynamics.
11. Experimental Methods The crossed-beam instrument used for the present study has been described in detail e l s e ~ h e r e ? and ~ . ~only ~ salient features are described here. The CS2+ primary ion beam was generated by injecting 750 eV energy electrons into a high-pressure source where CS2 pressure was of the order of 1 Torr. CS2' ions trapped inside the plasma undergo several hundred collisions with neutral CS2 and electrons, resulting in relaxation to predominantly ground (electronic and vibrational) state ions. The ion source was maintained near the desired final ion kinetic energy and ions exiting the source were accelerated to 750 eV for mass analysis by a magnetic sector. The mass-selected ion beam was decelerated to the final collision energy by an exponential deceleration lens and intersected with a supersonic beam of neutral argon. The argon beam was generated by supersonic expansion of neat argon through a 100 pm diameter capillary nozzle and passing the central core through a 1 mm diameter aperture skimmer into a series of two differentially pumped collimating chambers before reaching the collision center. The product ions were energy analyzed by a 90" cylindrical energy analyzer, mass analyzed by a quadrupole mass filter and finally detected by an electron
J. Phys. Chem., Vol. 99, No. 42, I995 15489 multiplier configured for pulse counting. The neutral beam was chopped mechanically for signal averaging. The laboratory angular and energy distributions of product ions were measured at a series of angles from 0" upward as long as a sufficient number of signal counts were present to give reasonable energy distributions. The observed spread of the primary ion beam at 51.5 eV laboratory energy was 1.35 eV (fwhm) with an angular spread of nearly 1.5" (fwhm). The line width of the energy analyzer was about 2.7% of the transmitted energy. In the present experiments, product ions were accelerated by an additional 3 eV for transmission and analysis through the energy analyzer. The experimental energy distributions determined as a function of laboratory scattering angles were converted into velocity distributions and transformed into Cartesian probabilities. The necessary transformation relationships and procedure for data analysis specific to CID processes have been described in detail in several earlier publication^.^^*^^
In. Results and Discussion A. CS2+ Elastic Scattering. Prinslow and Armentr~ut'~ have measured absolute cross sections for all major reaction channels for CS2+ impacting on Xe over the collision energy range from below 2 to 16 eV. Since the summation of all reactive channels at 16 eV is less than 4% of the geometric cross section, it may be inferred that nondissociative scattering is the principal outcome of colliding this cation with the rare gases. This is supported in the present research, in which we have investigated both the predominant, nonreactive channel and S+ formation. Figures 1 and 2 present our experimental data for nondissociative scattering of the carbon disulfide cation colliding with argon at a collision energy of 17.8 eV (51.5 eV laboratory energy). The data were taken at 1" intervals over the range 0"-33", at which point the detector angle becomes tangent to the elastic scattering circle (ESC) indicated in the figures; spectra taken at angles less than 4" are not included because of spillover of the much more intense primary CS2+ in that region. A mesh plot of the scattered parent-ion intensity is shown in Figure 1. Also shown is the elastic scattering circle with the center of mass (CM) for this collision process as origin. This figure shows that the scattered signal consists of essentially one feature which is well-described as elastically scattered CS2+. Moreover, the scattered intensity is mostly confined to the "forward" half-space and decreases monotonically with increasing scattering angle, approaching 0 at = 33". Figure 2a shows a semilog mesh plot of the same data. It is apparent that elastic scattering of CS2+ is strongly forward with any backscattering being several orders of magnitude smaller than the forward-scattered intensity. Figure 2b is the corresponding contour plot, showing contours of equal intensity, with an overlay of the ESC and circles corresponding to nondissociative inelastic scattering for conversion of translational energy into electronic excitation to CS2+(A211,) and (B2&+) (see Table 1). Translational spectroscopy studies28of this system at 5 kV laboratory energy have demonstrated that both channels are moderately populated using the rare gases as targets, including Ar. A specific objective of our study was to determine whether inelastic scattering to generate these excited states is significant and establish the relevant kinematics. Since it is known that CS2+(C2Z:) lies just above the thermochemical threshold for Sf(4SU)formation and predissociates to this limit, we anticipated that the kinematics for electronic excitation of the primary ion beam might be strongly correlated phenomenologically with this reaction. Figure 3 indicates schematically the procedure which was followed to search for evidence for inelastic scattering over the
15490 J. Phys. Chem., Vol. 99,No. 42, 1995
Tosh et al.
Figure 1. Mesh plot of the nondissociative scattering of CS2+(X211g)on collision with Ar at 17.8 eV collision energy. Velocity axes labeled by “CSzf” and “Ar” indicate the reactant velocities in the superposed Newton diagram with elastic scattering circle (ESC). Velocities are in d s .
Figure 2. Semilogarithmic mesh plot and contour diagram of same data as in Figure 1. Velocities are in d s . (a, top) Mesh plot with ESC and Newton diagram superposed. (b, bottom) Cartesian velocity contour diagram showing relative intensities referenced to elastic scattering circle (ESC) and circles of constant velocity corresponding to excitation CS2+(Xzllg) CSz+(A21T,) and CSz+(B2&+), respectively. Relative intensities for three of the contours are indicated.
-
laboratory angles from 4” to 33”. At each angle investigated, the energy profile was evaluated and compared with the
distribution measured at 0” for the primary ion beam. This distribution is well-described as a Gaussian peak with a mean
Impulsive Model for Collision-Induced Dissociation TABLE 1: Ionization Energies of CS2+ electronic state
ionization energy (eV) 10.06 12.69 14.4 16.19 17.06
a Taken from: Turner, D. W.; Baker, C.; Baker, A. D.; Brundle, C. R. Molecular Photoelectron Spectroscopy; Wiley-Interscience: New
York, 1973.
,A
"CS;
deg.
1
Figure 3. Angular scans of CS2+ intensity at selected angles of 7", 20°, and 32". Laboratory scattering angles referenced to Newton diagram for CS2+ scattering by Ar. Perpendicular lines dropped from maxima are referenced to the E X . Distributions of roughly equal height are obtained by multiplying the scans at 20" and 32" by factors of 5 and 500, respectively. See discussion in text.
energy of 51.5 eV and a fwhm breadth of 1.35 eV. Actual experimental scans projected on the Figure 3 diagram at three selected angles and differing in relative intensity by about a factor of 1000 show the quality of the data and illustrate how our data were analyzed. They are reasonably described as Gaussian peaks whose widths gradually increase with laboratory angle. This is readily understood as sampling an increasing range of CM scattering angles as the sampling angle approaches the line tangent to the ESC. There is also a monotonic increase in experimental error as the tangent line is approached, where the count rate has declined dramatically relative to the lowest angle sampled. Within our experimental error each angular scan is centered on the ESC over the entire range investigated, and there is no discemible evidence in the contour plot for electronic excitation. Stated more explicitly, excitation into the A2UUstate of CS2+ cannot be ruled out entirely because the measured contours partly overlap the A circle; excitation into the state would appear to be ruled out. The overlap of the A circle is small, however, and appears to be a consequence of the instrumental line width of the energy analyzer rather than energy conversion into electronic excitation. Inspection of the semilog surface plot of Figure 2a reveals a very small (5.1% of the maximum intensity) amount of signal appearing in the lowenergy tail of scans near the forward direction. While this might be evidence for electronic excitation, close examination of this region, as well as the low-energy tails of the energy distributions at higher angles, did not reveal any distinct features that could be unambiguously ascribed to excitation into any of the states listed in Table 1. The measured width of the parent-ion beam also does not permit us to rule out the occurrence of vibrational excitation within the X state, which can support several electronvolts of vibrational energy; however, the symmetry of the contours about the ESC at all angles suggests that it, too, is not very likely. The lowest dissociation limit for CS2+ lies about 4.8eV above the ground state,29 so that vibrational excitation within the X
B2C
J. Phys. Chem., Vol. 99, No. 42, 1995 15491 state would produce events in the velocity space extending from the ESC to well inside the B circle. However, the centroid of the scattered distribution remains centered on the ESC at all angles, and its shape and width are readily rationalized in terms of the energy distribution of the primary ion beam. Further, since no superelastic scattering is observed, we may infer that the reactant ion beam contains a negligible population of excited states, a conclusion also supported by lifetime arguments.30 With the exception noted above, there appears to be no tail of CS2' intensity penetrating into the interior of the ESC at any angle. While it was our expectation that we would observe excitation of CS2+ with increasing scattering angle (corresponding to decreasing impact parameter), our experiments provide no definitive evidence for nondissociative inelastic scattering of CS2+ in 17.8 eV collisions with argon.3' We reach the remarkable conclusion that the probability for electronic a n d or vibrational transitions in CS2+ undergoing collisions with Ar at 17.8 eV CM collision energy appears to be negligible by comparison to elastic scattering. Moreover, elastic scattering is largely restricted to the forward hemisphere, falling essentially to zero at CM scattering angles above 90". B. CID To Produce S+. Mesh and contour plots of the S+ velocity distributions at the same collision energy are shown in Figure 4a,b, respectively. These plots were generated from energy distributions recorded at lo-2" intervals for 6lab between 0" and 22" and 3"-4" intervals for 61ab between 22" and 44". Unlike the parent ion results discussed above, these plots extend to 0" because there is no background content of S+ present in the primary ion to obscure this information (except for a small contribution from background CID of CS2+ which is removed by signal averaging). Analogous to elastic scattering, reaction to form S+ also results from forward scattering. Not unexpectedly, the velocity vector distribution for this endothermic reaction is much broader than the elastic scattering channel discussed above. As shown quantitatively in Figure 4, the distribution is peaked at about 8" with respect to the lab frame. It exhibits an extensive low-energy tail and a ridge proceeding out to higher scattering angles. An overlay of the parent-ion ESC is included for reference and indicates that most of the activity is confined to the interior of the circle, consistent with our expectations for endothermic reactions. The location of this peak is unremarkable and could be interpreted as an impulsive scattering event which converts somewhat more translational energy into intemal energy than is required to drive a dissociation which is endothermic by 4.8 eV. In an earlier p ~ b l i c a t i o nwe ~ ~ discussed this reaction channel assuming the conventional two-step activatioddissociation model for polyatomic ion CID applies to this system. In the present study of CS2+ CID, carried out with higher precision energy and angular resolution over a broader range of scattering angles than our earlier study,23additional features are evident in Figure 4. In particular, a considerable fraction (about 30%) of the total S+ signal lies at laboratory angles clearly outside of the ESC, which by the conventional model would be interpreted as resulting from superelastic processes. However, the search for nondissociative inelastic scattering of CS2+ described above, supported by other e ~ i d e n c e , ~ rules "~~ out any significant excited-state population in our reactant ion beam, a statement further strengthened by our use of a highpressure ion source in which the nascent population is relaxed by hundreds of collisions before extraction. Moreover, the contour plot (Figure 4b) shows that this contribution is part of a ridge that includes the peak of the distribution well inside the ESC, which represents the most probable scattering angle and
15492 J. Phys. Chem., Vol. 99, No. 42, I995
Tosh et al.
LAB Oli* Figure 4. Mesh plot and contour diagram for the reactive scattering of CS2+ with Ar at 17.8 eV collision energy. The Newton diagram showing the velocity vectors of the reactants and the ESC are superposed on the experimental data. Velocities are in mls. (a, top) Linear scale three dimensional plot shown with arbitrary vertical scale. (b, bottom) Cartesian velocity contour diagram showing relative intensities as a function of velocity and scattering angle. The solid circle is the ESC reference circle used for deducing energy transfer, and the dashed line is an arbitrary circle drawn with the CM of Ar/S collisions as origin.
energy for S+ production. The continuity of the Figure 4 distribution suggests that both the peak and the ridge extending outside the ESC may be generated by a common mechanism. To obtain a more precise specification of the ridge than is possible with the computer-averaged velocity contour diagram, energy spectra at each angle were carefully examined to determine the range of laboratory velocity that brackets the top of the peak. Figure 5 illustrates this procedure, which is in all respects analogous to the analysis of Figure 3 in the preceding section. Figure 5 includes a circle drawn through the locus of points defined by the peak maxima. The ESC from Figure 4 is also included as a reference circle. It is evident from this figure that the most probable velocity falls well inside the ESC at low scattering angles, approaches the ESC with increasing angle, intersects it at e]&,e 25", and lies outside the ESC at higher angles. According to the conventional two-step model, this would signify that deposition of energy into the parent ion decreases
with scattering angle, which to our knowlege has never before been observed phenomenologically in CID of polyatomic ions. Additionally the circle traced by the ridge maxima has as its origin a point on the relative velocity vector that we tentatively identified as the A r W CM, suggesting that ArlS+ momentum is conserved as an additional constraint to conserving ArICS2" momentum. As we shall argue below, this is the signature of a process in which S+ is produced in a prompt dissociation event following what may be essentially described as a collision in which Ar has struck only an S atom in the parent ion. Knockout Kinematics. For reasons that are obvious from Figure 4b, we consider in this section collisions in which one particle, A, strikes a submolecular moiety within a polyatomic particle, B. We first derive the kinematic relationships for this model that follow from kinetic energy and momentum conservation and explore their consequences by means of Newton diagrams. Finally, we consider the dissociation of B and the
Impulsive Model for Collision-Induced Dissociation
J. Phys. Chem., Vol. 99, No. 42, 1995 15493
n
i Figure 5. Angular scans at selected angles of 6", 20°, and 40" showing relative intensities of S+ as a function of velocity at each angle. Perpendiculars dropped from the maximum are referenced to the dashed circle shown in Figure 4, with the CM for S'/Ar scattering located on the relative velocity vector as origin. Distributions of roughly equal height are obtained by multiplying the scans at 20" and 40" by factors of 2 and 13, respectively.
recoil of the fragments, which are kinematically different for prompt and slow dissociations. We refer to Figure 6, which shows the disposition of initial and recoil velocities immediately following a collision in which particle A strikes a moiety within the polyatomic particle B, designated B1 (the remainder of B shall be designated B2). Since the recoil of B is determined by the impulse delivered to BI, the recoil vectors for A and BI, as well as A and B, are related by momentum conservation. This has the consequence that the momentum of BZis unchanged by the collision. Hence, under the present restrictions, BZ is a collision spectator, and a measurement of either of the recoil vectors of A or B I (or B if the collision does not result in dissociation) is sufficient to determine all recoil vectors if the reactant vectors are known. In the standard treatment of nonreactive two-body scattering, it is shown that concentric scattering circles define the kinematics for the recoiling particles when translational energy loss is independent of scattering angle or when the collision is elastic.32 In either case the center for both circles is the CM. In the present problem, similar assumptions lead to scattering circles for A (and B I ) and for B that are centered at separate points on the initial relative velocity axis, as shown in Figure 6. Equations for these circles are determined from energy and momentum conservation. For example, when the total energy ET is_wriGen as a function of the final momenta of A, BI and B2 (PA', PI', P i ) and inelasticityQ, substigtion for PI' in terms of momentum of the A/B1 CM PA,^) and P'A yields the following relation:
-
-
'A3 I
ET
-
2(MA + M I )
- - -F22 P) 2 4
If Q is independent-of angle, eq 1 defines a circle in velocity space whose center is VA,I(denoted A/BI in Figure 6) and whose radius, rA, is given in terms of Q and conserved quantities. Similar manipulations show that the corresponding scattering circle for BI is concentric with the A circle but with radius equal to ( M A / M I ) I Awhile , the scattering circle for B has radius (MA/ MB)IAbut is centered on the reactant relative velocity axis at (MI/MB)(VA,~ - Vz) (denoted C B in Figure 6 ) . Energy and momentum conservation conditions also provide the following relationship between energy transferred into B
-
Figure 6. Hypothetical Newton diagram depicting the collision process A B A (BI B2) in which B2 is a spectator. Final velocities, distinguished by primed vectors, lie on scattering circles centered at points on the reactant relative velocity axis (see text for details). The case shown is for Q = 0.
+
+
+
Figure 7. Hypothetical Newton diagram for the same process shown in Figure 6 but identifying the ESC for particle B and chords used for calculating energy transfer into B, AT, and relative energy of B I and B2, AK. The A circle from Figure 6 has been suppressed to simplify the presentation.
and the energy of B I -Bz relative motion:
MIM2 -(V,' 2%
-
- c23* + Q (2)
Equation 2 stipulates that energy transferred out of reactant translation appears as relative motion of BI and BZor disappears into degrees of freedom not represented in the Newton diagram (Le., into the breaking of chemical bonds or into the intemal energy of any of the product particles). Energy transfer, AT, and energy of B I - B ~ relative motion, AK, are represented geometrically by the labeled chords in Figure 7. As is clear from Figure 7, AT and AK vary with scattering angle; however, the difference between them is independent of angle, according to eq 2, since the scattering circles were constructed under the assumption that Q is constant. For the case shown in Figure 7 (and Figure 6), Q = 0, so, in fact, all the energy transferred into particle B must appear in relative motion of BI and Bz. If Q > 0 and is independent of angle, then eq 1 and the preceding discussion imply that the basic symmetries of the recoil loci shown in Figure 7 are unaffected. The radii of the circles are simply shrunk to account for energy transferred into internal modes of A, B1, or B2, as illustrated in Figure 8. This is the situation expected in the case of CID, since it involves the breaking of chemical bonds.
15494 J. Phys. Chem., Vol. 99, No. 42, 1995
Tosh et al.
Figure 8. Scattering circles and chords partitioning the final kinetic energy in a knockout collision where Q > 0. See Figure 7 for comparison and text for details.
The extension of the foregoing analysis to cases in which B undergoes a prompt dissociation following collision is straightforward. All post collision velocities and masses designated by “B” in Figure 7, for example, describe the CM of the dissociated fragments BI and B2 rather than a real particle. Nevertheless, a complete specification of the kinematics is obtained from a measurement of the recoil of either A or B I . The recoil of the BI/BzCM, which is needed to calculate energy transfer into the B molecule, lies at an intersection of the B circle with the chord connecting the BI and B2 recoil vectors. Q may be calculated as described above (eq 1 or 2) and includes as a minimum the dissociation energy of the B I - B ~bond. If, on the other hand, dissociation is delayed long enough so that no correlation exists between the dissociation axis and elements of the Newton diagram, then a coincidence technique to define the velocity vectors of at least two particles is required to specify the kinematics completely. For the limiting case of completely randomized dissociation processes (or for low kinetic energy release), fragment velocity distributions exhibit the same symmetry as the B circles shown in Figures 7 and 8, and the centroids of these distributions could be used to determine the B circle and AT as described above. In this case, AK may be estimated from the widths of the fragment velocity distributions and Q calculated from eq 2.
IV. Application of Model to S+ Formation Since we began our odyssey with the qualitative observation that S+ follows the kinematic constraints attributed to BI in the case of prompt dissociation (with the possible correlation of scattering angle and dissociation axis), it is appropriate to apply this model quantitatively and examine its consequences. We confess that the hypothetical examples shown in Figures 6-8 have been drawn using the actual mass factors for C, S , and Ar, with CS as spectator. The same model, stipulating a different spectator, could be used to explore the consequences of an impulsive mechanism to produce CS+ (or other moieties of the CS2+ ion) if scattering data suggested some other scattering center. Next we analyze the internal energy change which corresponds to the actual radius deduced for S+ scattering exhibited in Figure 4. Specifically, we use eq 2 to guide the analysis depicted in Figure 9. The solid bars in Figure 9 represent the most probable range of recoil velocity for S+ as deduced from the experimental energy distributions. A circle centered at the Arts+CM and drawn through the bars defines a possible radius for the S+ recoil circle. The corresponding CS2+* circle is determined as described above. This pair of circles is used to calculate a Q for the reaction as follows. We begin by selecting a point on the S+ recoil circle and drawing a line to the tip of the initial CS2+ vector (also the final CS vector). The length of this chord is used to calculate the relative kinetic energy,
Figure 9. Newton diagram of knockout model applied to CID of CS2+ to produce S+, with CS as spectator. Solid bars represent the most probable range of S+ final velocity as determined from energy scans at each angle., Conjugate scattering circles for S+ and CS2+ are used to calculate the Q for the reaction (see text). The length of the bars represents experimental uncertainties in defining the maxima of the angular scans at each scattering angle. The original velocity Newton diagram is depicted as the reference for scattering coordinates. All experimental scans are included in the construction of this figure.
AK,of S+ and CS for this particular scattering event. This chord intersects the CS2+* circle at two points, but only one is the S+/CS CM corresponding to the selected S+ point. A line then
is drawn from the Ar/CS2+ CM to the ESC that passes through the conjugate point on the CS2+* circle. The resulting chord that connects the conjugate CS2+* point to the ESC is a measure of the conversion of reactant translational energy, AT, into internal energy of CS2+ for the same scattering event. Any set of conjugate points on the S+ and CS2+* circles may be used for this evaluation since Q = AT - AK is independent of angle. The process is carried out with the S+ circles of minimum and maximum radius that pass through the bars, to obtain an estimate of the range of Q. Using this procedure, we deduce that the total kinetic energy decreases by 4.5 f 0.8 eV in going from reactants to products, which is remarkably close to the minimum 4.8 eV required to break the S+-CS bond in CS2+. Thus, the scattering data interpreted via the.knockout model indicate that the interaction between the CS2+ cation and Ar atom involves one of the sulfur atoms behaving essentially as a free particle but with energy transfer into CS2+ sufficient to break a strong chemical bond. Following the discussion of Mahan and co-workers,I6 we can suggest that the collision induces a transition to an unbound electronic state of CS2+ just prior to the impact between Ar and S+. Alternatively, we can consider vibrational excitation, in which an asymmetric stretch mode of the CS2+ is the reaction ~ o o r d i n a t e provided ,~~ the kinematic constraints implied by the angular scattering can be satisfied. The essential point is that the energy transferred is sufficient to break the bond and momentum imparted to S+ is not communicated to the CS fragment. This implies that the potential curve at the instant the momentum exchange collision occurs must be essentially “flat” so that S+ is free to interact with Ar alone. Figure 10 is a schematic potential energy surface for C&+ showing on the left hand side the dissociation asymptote to give S+(4Su) as a product. As discussed p r e v i o ~ s l y the ~ ~ most -~~ plausible dissociation path involves an avoided crossing between the B state and the dissociation asymptote which is brought about by spin-orbit coupling. We now suggest that this promotion of CS2+ into the B2&+ electronic state occurs at an early stage of the collision, reducing the force between S+(4Su) and CS(X’Z+) to a very small value, allowing S+ to interact with Ar as a free particle.
J. Phys. Chem., Vol. 99, No. 42, 1995 15495
Impulsive Model for Collision-Induced Dissociation 20 18
16 h
%
Y
Figure 10. Schematic potential energy surface for several electronic states of Adapted from ref 23. The curve-crossing electronic excitation suggested to be responsible for setting the forces between S+(4SU) and CS(XIZ+)to essentially zero is from the B state of CS2+ to the 4Z- repulsive state. See discussion in text.
- Ar/S+(original)
- ---
Ar/S+(shiCted)
I\ \
vcsg
I
Figure 11. Experimental data from Figure 9 plotted with scattering circles centered on two possible points for the A d s + CM. The solid circle has the Ar/S+ CM calculated from the Newton diagram describing the collision process, while the dashed circle is centered upon a A r / S + CM which is shifted by the appropriate amount to represent "shortening" of the reactant velocity vectors when energy equivalent to the S+- -CS bond energy is transferred into the CS2+ ion.'The original velocity Newton diagram is depicted as the reference for scattering coordinates. All experimental scans are included in the construction of this figure.
This elaboration of the knockout model requires, for energy conservation, that no less than 4.8 eV of translational energy be transferred into electronic excitation of CS2+ in the approach trajectory. This change in the available kinetic energy effectively decreases the velocity vectors of the incoming particles and shifts the Ar/S+ CM toward the reactant CM. Figure 11 is a Newton diagram showing the shift in Ads+ CM required by this constraint. The same solid bars that appear in Figure 9 are drawn to represent the measurements. Separate scattering circles are shown for both the original and shifted Ads+ CM points,
illustrating the angular scattering ridges predicted by the two models. Although experimental errors in our measurements preclude a definitive choice between these two circles, it is evident that there is a better fit to the shifted center deduced from the mechanism, suggesting that electronic excitation precedes the scattering event. Referring now to the velocity contour diagrams in Figure 4, we consider the implications of the assertion that the ridge and the large peak are produced by the same mechanism. Our a priori expectation is that elastic scattering of S+ by Ar should exhibit a peak at 0", just as elastic scattering of CS2+ does (Figures 1 and 2). Experimentally, the large peak is displaced
about 15" away from the forward pole as measured from the A r I S CM. If, nevertheless, the mechanism has the essential character of elastic scattering, we must tum to steric effects in the orientation of the linear CS2+ as the collision occurs to explain the location of the peak maximum at a scattering angle larger than zero. We note that the collision time is about 30 fs, much shorter than the rotational period for thermal energy CS2+, so that the initial orientation is preserved throughout the collision until the prompt dissociation event. Thus, there is a cone of orientation about the forward direction where the struck S+ may experience a second collision with CS. This may be purely repulsive or can be attractive, tending to re-form CS2+ or convert S+ momentum into rotation of CS. This class of trajectories involves the largest impact parameter collisions resulting in impulsive ArIS scattering, effectively diminishing the S + signal in the vicinity of the forward pole. It seems to us most likely that the angular displacement of the maximum is some such steric effect, attributable either to the proximity of CS during the collision or to details of the electronic excitation step as Ar approaches the CS2+ parent ion. Planned additional experiments performed at lower collision energy and with improved resolution may be helpful, as would theoretical analysis of the excitation step. It is further evident in Figure 4 that significant scattered intensity is found off the ridge described by this mechanism, most of it located within the ESC. There are several ways this observation may be rationalized. An obvious one is that a significant fraction of collisional excitation occurs by a mechanism other than the knockout model we have described. One possibility is that momentum exchange for the second mechanism is centered on the conventional CM for the system. This corresponds to the conventional two-step model for CID which has mainly been used to describe CID of polyatomic cations.'* Another possibility is that a range of vibronic excitation occurs, with resulting curvature on the potential surface coupling motion of CS with S + as the momentum-conserving collision occurs. In such a case, the velocity vectors of all three particles are coupled and the constraints depicted in the Newton diagrams of Figures 6-8 no longer apply. Accordingly, the CS fragment is no longer restricted by energy and momentum conservation to the velocity vector of the primary ion and the S+ intensity would tend to fill much of the space within the ESC. Coincidence measurements of all three particles would answer the question but are not technically feasible for us to carry out. Finally, we remark that the constraints on the potential surface that are necessary for the modified version of the knockout model are so unusual that it is unlikely to be an important mechanism for polyatomic cation CID. It may actually be restricted observationally to triatomic ions. If it is applicable to larger ions, it most likely would involve those with special steric properties and higher mass substituents which would emphasize kinematic effects of colliding with a subset of the molecular ion. Large ions are also likely to exhibit delayed dissociation, averaging out some of the correlations. We have found no evidence for such a mechanism, for example, in our studies of several model compounds for organic molecule classes.I8 Finally, the model enables us to rationalize several anomalies in MIKES and other mass spectrometry studies34of CS2+. We plan to explore its utility for elucidating the CID dynamics of other systems.
Acknowledgment. The authors would like to express their gratitude to Professor Douglas P.Ridge for valuable discussions. Support of this work by the National Science Foundation, Grant CHE-9021014, is gratefully acknowledged.
15496 J. Phys. Chem., Vol. 99, No. 42, 1995
References and Notes (1) Heman, Z.; Kerstetter, J.; Rose, T.; Wolfgang, R. J. Chem. Phys.
1967,46,2844. (2) Heman, Z.; Kerstetter, J.; Rose, T.; Wolfgang, R. Discuss. Faraday SOC. 1967,44, 123. (3) Hierl, P.; Herman, 2.;Kerstetter, J.; Wolfgang, R. J. Chem. Phys. 1968,48,43 19. (4) Hierl, P.; Herman, 2.;Wolfgang, R. J. Chem. Phys. 1970,53,660. (5) Herman, Z.; Kerstetter, J. D.; Rose, T. L.; Wolfgang, R. Rev. Sci. lnstrum. 1969,40,538. (6) Henglein, A,; Lacmann, K.; Knoll, B. J. Chem. Phys. 1965,43, 1048. (7) Ding, A.; Lacmann, K.; Henglein, A. Eer. Bunsen-Ges. Phys. Chem. 1967,71, 596. (8) Chang, D. T.; Light, J. C. J. Chem. Phys. 1970,52, 5687. (9) Vestal, M. L.; Wahrhaftig, A. L.; Futrell, J. H. J. Phys. Chem. 1976, 80,2892. (10) Brenot, J.; Dump-Ferguson, M. Adv. Chem. Phys. 1992,82, 309 and references therein. (1 1) Martin, S. J.; Heckman, V.; Pollack, E.; Snyder, R. Phys. Rev. A 1987,36,31 13. (12) Cheng, M. H.; Chiang, M. H.; Gislason, E. A,; Mahan, B. H.; Tsao, C. W.; Werner, A. S. J. Chem. Phys. 1970,52, 6150. (13) Fayeton, J.; Pernot, A.; Foumier, P.; Barat, M. J. Phys. 1971,32, 743. (14) Song, J.; Gislason, E. A.; Sizun, M. J. Chem Phys. 1995,102, 4885 and references therein. (15) Loesch, H. J.; Moller, J. J. Phys. Chem. 1993,97, 2158. (16) Cheng, M. H.; Chiang, M.; Gislason, E. A,; Mahan, B. H.; Tsao, C. W.; Werner, A. S . J . Chem. Phys. 1970,52, 5518. (17) Herman, Z.; Futrell, J. H.; Friedrich, B. lnt. J. Mass Spectrom. lon Processes 1984,58, 181. (18) Shukla, A. K.; Futrell, J. H. Mass Spectrom. Rev. 1993,12, 211. (19) Prinslow, D. A.; Armentrout, P. B. J. Chem. Phys. 1991,94, 3563. (20) Sohlberg, K.; Chen, Y. B. J. Chem. Phys. 1994,101, 3831. (21) Laramee. J. A.: Camodv. J. J.: Cooks, R. G. lnt. J. Mass S m t r o m . lon‘ Phys. 1977,43,321. (22) Hirst, D. M.: Jenninas, K. R.; Laramee, J. A.; Shukla, A. K. lnt. J. Mass Spectrom. lon Processes 1985,63, 119.
Tosh et al. (23) Shukla, A. K.; Tosh, R. E.; Chen, Y. B.; Futrell, J. H. lnt. J. Mass Spectrom. Ion Processes, submitted. (24) Blakely, C. R.; Ryan, P. W.; Vestal, M. L.; Futrell, J. H. Rev. Sci. Instrum. 1976,47, 15. (25) Vestal, M.; Blakely, C. R.; Futrell, J. H. Phys. Rev. A 1978, 17, 1321. (26) Futrell, J. H. In Gaseous lon Chemistry and Mass Spectrometry; Futrell, J. H., Ed.; John Wiley: New York, 1986. (27) Shukla, A. K.; Futrell, J. H. In Experimental Mass Spectrometry; Russell, D. H., Ed.; Plenum: New York, 1993. (28) Reid, C. J.; Harris, F. M. lnt. J. Mass Spectrom. lon Processes 1988,85, 151. (29) Momigny, J.; Wankenne, H.; Mathieu, G.; Flamme, J. F.; AlmosterFerreira, M. A. Adv. Mass Spectrom. 1974,6,923. (30) Eland, J. H. D.; Devoret, M.; Leach, S. Chem. Phys. Lett. 1976, 43,97. (31) A reviewer expressed a similar expectation, implying that the absence of backward-scattered CS2+ is consistent with excitation sufficient to dissociate the molecule. In fact, the work presented here and in ref 23 shows that dissociation fragments are found almost exclusively in the forward hemisphere for CSz+/Ar collisions at this energy. However, recently we have undertaken a more thorough search for CS2+ in the backward hemisphere, and our preliminary findings at 8.9 eV CM collision energy show evidence for excitation to the A and B electronic states amounting to a few percent of the elastically scattered signal. The results of this work will be presented in a later publication. (32) Schiff, L. Quantum Mechanics, 3rd ed.; McGraw-Hill: New York, 1968; p 112. (33) The vibrational period of the asymmetric stretch mode for CS2+(Xzll,) is about 28 fs (see ref 20), which is within 10% of the effective collision time for this system at this energy if CS2+ is assumed to be 3 A across. Thus, the conditions for energy transfer from translation into the asymmetric stretch very nearly satisfy the Massey adiabatic criterion for maximum transition probability. Obviously, the absorption of several quanta would be necessary for the system to acquire enough internal energy to approach the 48- asymptote, but in some sense the overall transition might be “resonantly enhanced” because of this. (34) Tosh, R. E.; Shukla, A. K.; Futrell, J. H., to be published. JF9507455