Electronic Excitation in Potentially Reactive Atom-Molecule Collisions BY MALCOLM A. D. FLUENDY, KENNETH P. LAWLEY, JOHN MCCALL, CHARLOTTE SHOLEEN AND DAVID SUTTON Department of Chemistry, University of Edinburgh, Edinburgh EH9 355 Received 1 1 th December, 1978 Inelastic differential scattering cross sections for the system potassium + alkyl halide have been measured in the small angle region for Ex between 20-1000 eV". Electronic excitation of both colli- sion partners is seen together with vibrational excitation of the alkyl halide. Evidence is adduced suggesting that excitation occurs by either of two paths corresponding to the preliminary transfer of an electron in the entrance channel or as the colliding pair recedes. A harpooning model incorporating bond stretching in the negative molecular ion is developed that agrees well with most of the observations.A large number of exit channels are open in the collision system alkali atom + alkyl halide at higher energies. They include: M + RX-+ M + RX(RX7) elastic (inelastic) (0 - + M X + R reaction (ii) -+ M+ + RX- chemi-ionisation (iii) -+ M* + RX M +RX*) electronic excitation - + M + R + X dissociation. (v) The first two processes have been extensively investigated at thermal collision energies 1*2 and are well known examples of the electronic harpooning mechanism, subsequent chemical reaction occurring at thermal energies by ionic combination. The chemi-ionisation channel is less well explored3 but provides direct evidence for non-adiabatic behaviour at the ionic/covalent surface crossing.The importance of an ionic surface in coupling ground and excited electronic states of the atom is con- firmed by collision-induced fluorescence ~tudies.~ In the work described here continuing the programme outlined in a previous Faraday Discus~ion,~ we have eliminated the reaction channel by working at high relative kinetic energies and choosing a heavy halogen atom, iodine. Equally im- portant from the point of view of analysis, by confining scattering observations to very small angles ( 5 5") the K atom trajectories are essentially rectilinear and of con- stant velocity. Nevertheless, because the forward momentum is high, interesting regions of the potential inside the harpooning radius can be probed by these small deflections. Electronic excitation of several eV is readily observed.42 ATOM-MOLECULE COLLISIONS EXPERIMENTAL APPARATUS The apparatus used in this work is shown schematically in fig.1. The beam of fast alkali atoms was produced initially as ions by surface ionization and electrostatic focusing. The ion beam was then pulse modulated, using a velocity compression technique described else- where,6 so that the energy loss resulting from a collision could be recorded by measurement of the flight time of the scattered atom and hence the post-collision states of the atom and molecule inferred. After modulation the ion beam was neutralised in a vapour cell and any remaining ions deflected away. r a d i a t i o n heater source oven Gtjst aff son ~~ ~~ t a r g e t beam reservoir r collimating array 3 cross beam chopper motor vapour c e l l ion dump collision zone plates collimating slits f cross beam filament porous tungsten disc t.0.f pulsing lens collimating array flexible bellows / detector detect o r assembly FIG.1 .-Schematic representation of apparatus. The fast neutral beam then intercepted a slow target beam of molecules formed by effusion from a capillary array in a well defined collision zone. This beam was also modulated (at 47 Hi) and the target flux continuously monitored by a gauge placed directly below the collision zone. Potassium atoms scattered from this region were ionised on a cool W wire and detected uia a scintillator and photomultiplier. The detector could be varied in angle with a precision of &0.002". Atom arriirals located in angle by the detector position were arranged to stop a 50 MHz clock running in synchronism with the pulse modulation so that the flight time could be recorded.The collection of data and the operation of the experiment were controlled by an on-line computer.' The signal collection and experimental control arrangement are shown schemati- cally in fig. 2. Hard copy log and graphical output facilities were provided to allow operator intervention, DATA COLLECTION A N D ANALYSIS Count rates are very low in this experiment (<0.01 counts s-l) and periods of E 12 h were required to collect sufficient counts at the widest angles. Data collection thus took place over periods of about five days. During this time the main beam arrival time profile was checked at intervals under program control and data collection suspended and the operator.) DEC 11/45 I run delay C& ? delay ,flip f lop - '/ 1 i l o B i, -tr7 - ready -T multiplexe d a t a target modulator r filament I monitor stepper angle drive encoder modulation r e lay drivers t i a 0.0.c. i modulation I display rp 4-1 logging teletype teletype f l i p fi scaler 4 I 4 I ] clock c" divider44 ATOM-MOLECULE COLLISIONS alerted if any significant changes in the beam fluxes or other operating conditions took place. The angular scan was made automatically to a predetermined sequence, angle changes being initiated automatically when a set precision had been reached. The data reported in this paper were accumulated over a period of about eighteen months during which time the equipment was removed from one building to another and a number of small changes made.Partly as a result the time location of the primary beam pulse varied by as much as 30 ns between different experiments. The data were therefore adjusted in time so as to be relative to the unscattered beam arrival as measured in each experiment. Any accompanying variation in the pulse width was corrected by a process of deconvolution and reconvolution to a standard pulse width, the stability of these operations being checked by trials with synthesised noisy data. Inconsistencies of this type between different experimental runs rather than counting statistics account for most of the noise seen in the results. After these adjustments had been made in the laboratory frame the data were transformed into the c.m.frame using the most probable laboratory velocities. RESULTS These results are most compactly presented as c.m. contour maps showing the variation in the product of the scattered intensity and the square of the scattering angle, Z(x)x2, as a function of the variable z (t = collision energy x scattering angle, Ex) and the post-collision velocity. The contour map in fig. 3 shows such a plot for K + Ar and illustrates the energy b 4 1 2 scattering angle 1 deg FIG. 3.-K + Ar scattering at 108 eV c.m. collision energy. The thick lines indicate energy losses of 0.0 and 1.6 eV (centre of mass frame).M. A . D . FLUENDY, K . P . LAWLEY, J . MCCALL, C . SHOLEENANDD. SUTTON 45 resolution since in this E region the scattering is at least 98 % elastic.The island of intensity at slow exit velocities is due to the K4' isotope present at ~6 % abundance. The other contour maps in fig. 4-6 show similar plots for methyl and propyl iodide at various initial collision energies. In comparison with the K + Ar data considerable inelasticity, particularly at the wider angles, is immediately apparent and can be seen to onset at specific Ex. + .- u 0 0, w - I 0 5 scattering angle I deg 2 FIG. 4.-K + CH,I scattering at 164 eV collision energy. The thick lines indicate energy losses of 0.0, 1.6 and 3.47 eV (centre of mass frame). Cuts through the surface show the intensity of scattered K atoms as a function of the energy lost by them in collision are perhaps more suggestive. A number of ex- amples computed by averaging together several sets of independent observations in a narrow range of angle are shown in fig.7 and 8. The solid curves on these figures show the results of a deconvolution procedure using the 0" profile as a reference profile. The peaks are sharpened by this process but can already be distinguished in the unprocessed data ; moreover, independent angular scans yield peaks which move smoothly with angle as in fig. 9-1 1 . The enhanced scattering profiles prepared in this way are combined to yield similarly sharpened contour maps as shown in fig. 12 and 13. Time of flight data of this type are of limited value in molecular systems because it is not possible to associate a given velocity change in the K atom with a specific exit channel, owing to the number of closely spaced energy states.Thus, in the K + RI system the K ionisation continuum starts at 4.34 eV and there is a near con-46 ATOM-MOLECULE COLLISIONS tinuum of vibration-rotation states in each electronic level of RI. Table 1 summarises the relevant information for K and CH31 (C3H,I is similar). In view of this continuum of vibronic levels, it is remarkable that discrete energy losses are observed at least up to 10 eV at the largest angles of scattering. r. u 0 0, > + .- d 0 5 3 d 1; :j -5 7 62 0 2 scattering angle / deg and 3.47 eV (centre of mass frame). FIG. 5.-K + CHJ scattering at 81 eV collision. The thick lines indicate energy losses of 0.0, 1.6 DISCUSSION At scattering angles x 5 5", the momentum transfer perpendicular to the incident velocity is small (x = final transverse momentum/incident forward momentum).The various small-angle approximations are valid and the momentum changes in the forward direction cancel on the incoming and outgoing halves of the trajectory. Under these conditions the maximum energy transferable to vibration/rotation of the molecular partner is where X = I or R, the end struck, and a " forceless " oscillator has been assumed. The maximum energy thus transferred at the angles of observation will be <0.5 eV, far smaller than most of the observed energy loss channels. The extensive vibronic energy transfer that is observed can only occur if the potential energy surfaces areM . A . D . F L U E N D Y , K . P . LAWLEY, J . MCCALL, C . SHOLEEN A N D D . SUTTON 47 profoundly modified in the course of the collision.The source of this alteration is clearly the crossing onto the ionic state. The harpoon model, equivalent to adiabatic behaviour at the ionic/covalent crossing, is well established as the mechanism for chemical reaction in many alkali metal systems at thermal energies. At the collision energies of the present experiments the reactive channel is essentially closed because the fast K+ ion cannot accelerate the c .- u 0 aJ > L I 61 0 3 scattering angle / deg and 3.47 eV (centre of mass frame). FIG. 6.-K + C3H71 scattering at 166 eV collision. The thick lines indicate energy losses of 0.0, 1.6 I- ion sufficiently rapidly to capture it before leaving the ionic surface. The residual electronic excitation is like the grin on the face of the Cheshire Cat, the aftermath of a much more profound electronic rearrangement.We develop a model to account for the broad features of the observed scattering in two stages. As a first approximation, the collision is assumed to be isotropic and sudden with respect to the R-I motion, i.e., the R-I bond is clamped at its equilibrium value throughout the collision. The behaviour of the various diabatic potential sur- faces can then be displayed solely as a function of the K-I coordinate, fig. 14(a). The vertical electron affinity of the alkyl iodides is sufficiently small (-0.9 eV) for the ionic state to intersect all the K* channels (including the ionised continuum) and thus to provide a route for populating these states. Excited electronic states of RI, except the A state, lie above the dissociation limit of K+RI- (5 eV) and must then be populated48 ATOM-MOLECULE COLLISIONS by a different mechanism. We speculate that an excited charge transfer state is involved but the mechanism will not be discussed here.The only important adjustable parameter in these potentials is the short range repulsion behaviour of the ionic state and the coupling matrix elements at the various crossings. Whatever the values of the parameters, some simple consequences arise because any exit channel can be reached via two paths, according to whether or not the electron is transferred on the first passage of the ionic/covalent crossing. 0 energy loss /eV FIG. 7.-Energy loss profiles observed at various scattering angles for K + CH31 at 81 eV collision energy.The dashed curves are observed values and the solid lines their deconvolution. (a) 61, (b) 122 and (c) 203 eV". The predictions of this model (using the potentials shown, the Landau-Zener approximation and the classical small-angle formulae to evaluate the cross-sections) are compared with experiment at 164 eV in fig. 15, the energy loss data being par- titioned in accord with the asymptotic energy losses assuming only electronic excitation. The model is partially successful, especially in predicting the narrow angle thresh- olds of K* and CHJ* ( A ) state onsets. If the route to these states involved a cross- ing on the respulsive wall of the potential, the angular threshold would appear at much larger angles and the intervention of a strongly attractive surface is unambiguous.The model is less satisfactory in predicting the change in angular onsets of the vari- ous channels with incident energy. These thresholds are seen to occur at lower E values in the 81 eV data, whereas the basic model necessarily predicts constant Ex values (assuming straight line trajectories). More important differences are seen inM. A . D . FLUENDY, K . P . LAWLEY, J . MCCALL, C . SHOLEEN A N D D . SUTTON 49 111 I: I 6 O energy loss /eV I FIG. 8.-Energy loss profiles observed at various scattering angles for K + CH31 at 164 eV collision energy. The dashed curves are observed values and the solid line their deconvolution. (a) 75, (6) 450 and (c) 900 eV". the energy loss spectrum where the model only permits energy losses corresponding to the electronic states of the separated species.The observations (e.g., fig. 7 and 8) show a much larger number of discrete energy loss processes, some of them (those < 1.6 eV) not being assignable at all to electronic excitation. The most serious assumption of the basic model lies in the neglect of the internal motion of the target molecule. s, changes in the C-I bond distance can occur which greatly alter the vertical electron affinity and hence the posi- tion of the ionic/covalent crossing. Such effects have been discussed by other workers8 in connection with chemical reaction and chemi-ionisation. The initial crossing at R1 yields CHJ- in a strongly repulsive state [fig. 14(b)], assuming a vertical transition.As the C-I bond stretches on the ionic surface, the ionic/covalent crossing moves to larger R values (fig. 16) and on the return of the electron a large amount of energy can be dumped in the Me-I vibration. The extent of such energy transfer clearly depends on the time spent on the ionic surface and ranges from zero if the motion at R1 is diabatic (electron not transferred) to actual dissociation of the Me-I bond if the MeI- surface is sufficiently repulsive. Since there are in general two classical paths leading to a particular angle of deflection (if b < &), corresponding to diabatic or adiabatic motion at R1, each electronic exit channel should be accompanied by two distinct peaks in the time of arrival spectrum. During the collision lifetime, typically50 ATOM-MOLECULE COLLISIONS Our second model, then, is to permit relaxation of the C-I bond in the ionic state by introducing a term Vio"(RR-I) = A exp [-a(& - RkO?)] (2) into the total potential energy.The I-K interaction remains coulombic and there is no K-R interaction. One result emerges immediately from this model. If the para- meters in Vion(&) are taken to be those of the isolated ion,' far too much vibrational excitation is predicted even in the ground electronic exit state. In fact, we would have a runaway situation with extensive bond dissociation (and probably chemi-ionisation). In practice (fig. 9-1 1) the vibrational energy gain in both the ionic K and K* channels is quite small ( z 1 eV) and almost constant with Ex after the threshold.The CHJ- ion is thus perturbed by the passing K+ ion and we can very crudely incorporate this effect in the model by making a adjustable. However, even this degree of freedom is not sufficient for the data to be fitted; if the vibrational energy gain in the K* (ionic) channel is fitted, too little energy loss occurs in the ground state channel. In qualitative terms, the initial acceleration of the methyl group after 2 x F - :I 2 - 0, - * 0 0 0 v u v " " 0°C 0 1: L 1 I I I I I I I I 0 500 1000 t IeV deg FIG. 9.-Plot showing the location of the peaks observed in the energy loss measurements as a function of the reduced scattering angle, z. CH31 + K 164 eV collision energy. 0 FIG. 10.-Plot showing the l m i o n of peaks observed in the energy loss measurements. CH31 + K at 81 eV collision energy.M .A , D. FLUENDY, K . P. LAWLEY, J . MCCALL, c . SHOLEEN AND D. SUTTON 51 I -- 6 300 t l e V d e g 0 FIG. 11 .-Plot showing the location of peaks observed in the energy loss measurements. C3H71 + K at 166 eV collision energy. I I I I 1 1 2 3 4 5 scattering angle I deg FIG. 12.-Contour plot showing CH31 + K scattering as a function of energy loss and c.m. scattering angle at a collision energy of 164 eV and after enhancement by deconvolution. Thick lines are drawn at energy losses of 0.0, 0.86, 1.6 and 2.8 eV.52 ATOM-MOLECULE COLLISIONS v) 0 x d al 1 060 1.720 scattering angle /deg FIG. 13.-Contour plot showing CHJ -+ K scattering as a function of energy loss and scattering angle at a collision energy of 81 eV and after enhancement by deconvolution.Thick lines are drawn at energy losses of 0.0, 0.72, 1.4, 2.0 and 3.2 eV. electron transfer seems to be rapid, but the repulsion soon drops almost to zero. The functional form of eqn (2) must be wrong and the dependence of Vion on RMeK should be introduced. This could be interpreted as due to the repulsion of the departing Me group by the K+ ion, or the change in bond order of MeI- due to partial back transfer of the electron to K+. Nevertheless, relaxation of the Me-I bond on the ionic surface is a key step in the collision process. Besides leading to extensive vibrational excitation, the deflection of trajectories sampling the ionic surface will depend on the extent of R-I relaxation during the collision. The angular thresholds for all electronic processes fed by the ionic surface will not thus scale with Ex, and will also depend upon the reduced mass of RI.These effects can be seen in fig. 17, where the energy losses calculated from this TABLE I.-EXCITED STATES OF K AND CH3I K energy /eV CHSI energy/eV 42s, 0.0 X(%) 0.0 42P3,; 1.62 A* 3.47* 52s+ 2.61 B, C(E) 6.10, 6.16 3’D;,4 2.67 D, ( E ) 6.77 5’P+,; 3.06 E, 7.30 I.P. 4.34 Rydberg states Rydberg states F, G 9.4, 9.8 LIP. 9.54 * Onset of continuous adsorption; peak at 4.5 eV.M . A . D. FLUENDY, K . P . LAWLEY, J . MCCALL, C . SHOLEEN AND D. SUTTON 53 R I i FIG. 14.-(a) Isotropic diabatic potential model for K + RI interaction. (b) CHd and CHJ* potentials. Dashed curves show the perturbation used to obtain the approximate fit described. FIG.15.-Isotropic sudden model; comparison with observations for CHJ + K at 164 eV. model are displayed against the corresponding scattering angle. In the Ex region around 150 eVo, particularly at 81 eV collision energy, a rainbow feature can be seen where two branches for the ionic ground state scattering coalesce. The invariance of vibrational excitation with angle of scattering is a remarkable feature of the plots and again points to a relatively small shift of the ionic/covalent seam with changing transit time over the surface. Finally, the differential cross sections for the channels identified are displayed, together with the model predictions, in fig. 18 and 19. The observed very narrow angle elastic scattering is normalised to the model.54 ATOM-MOLECULE COLLISIONS FIG.1 &--High energy trajectories on an ionic/covalent surface. Two trajectories are shown, corre- sponding to different initial kinetic energies (EA < EB). The crossing point on the outward path (R,) is very sensitive to E. In case A, Rz is so large that dissociation or ionisation would result. The K trajectory in real space is inset. I I 3- > : 2 - cn lA 0 - , * ----- - - - __ M . . . . . . . . . . . . . . . . . . M - * - - - * . . - ..... ..... .- * _----- -- . . . . . . . . . . . 0 500 1000 - ,elastic I elasticN T l e V d e g FIG. 17.-Comparison of observed (0) and bond stretching model predictions (M) for the energy loss as a function of reduced scattering angle. The subscript I indicates motion on the neutral surface. Model and experiment are in accord in predicting an increasing energy loss as the mass of R decreases and as the collision lifetime increases.Dashed curve, 81 eV Mel; solid curve, 164 eV MeI; dotted curve, 166 eV PrI.M . A . D . FLUENDY, K . P . LAWLEY, J . MCCALL, C . SHOLEENANDD. SUTTON 55 SO00 m u - D N ;? 5 - - - b ' t / e V deg - .- "'4 t P T - , -I A A 0 500 1000 t / e V deg 2750r FIG. 18.-Differential cross-sections for K + CH3I at 164 eV. (0, 0 ) ground state N, I; (A, A) K* (4p) N, I ; (0, 1) CH31* ( A ) N, I. Lines are the model fit (-) N and (- - - -) I. 0 3 000 T t eV deg FIG. 19.-As for fig. 18, but at 81 eV. T I e V deg CONCLUSIONS Our conclusions as to the processes involved may be summarised with reference to fig. 14(a) as follows: (a) Each electronic exit channel is accompanied by two vibrational channels, one with small or zero internal energy change, the other with substantial vibrational excitation.These two channels correspond to, respectively, diabatic or adiabatic (harpooning) behaviour at the first crossing R1. Both channels are important in the experimental energy range. (b) The negative ion state involved is a repulsive state, but with rather different characteristics from the isolated RI- ion. In particular, the amount of bond stretch- ing is less than expected (at least in the configuration probed in the bound state exit channels) and points to some containment of the alkyl group. ( c ) The differential cross section summed over all discrete exit channels (ground plus excited states) is approximately constant over the x range from 0.5 to 5" (LAB).This strongly suggests that continuum processes (bond dissociation and ionisation), unless they onset at very small angles of deflection, play a negligible role at impact parameters 5 0 L j .56 ATOM-MOLECULE COLLISIONS ( d ) The excitation of the A state of CH31 is observed to have two energy loss con- tributions and angular thresholds, but these have less intensity than in the K* channel. An ionic surface again probably intervenes because of the small angular thresholds. But even without bond stretching, the ground state K+CH31- surface would lead to a crossing at GZ 50 A on the outward branch, at which point the coupling matrix element between the two states would be essentially zero. Some electronic excitation of the negative ion may be involved, i.e., harpooning to a different empty orbital. (e) Discrete energy losses >5 eV are observed and these must correspond to electronic excitation of the alkyl iodide. Since these energy levels are above the energy of the separated K+RI- ion pair, the ground state ionic surface cannot be involved in their coupling. J. L. Kinsey, Molecular Beam Reactions (M.T.P. Int. Rev. Sci., Physical Chemistry, 1972) ser. 1, vol. 9. M. E. Gersh and R. B. Bernstein, J . Chem. Phys., 1972, 56, 6131. A. P. M. Baede, Charge Transfer between Neutrals at Hyperthermal Energies, Adv. Chem. Phys., (Wiley, Chichester, 1975), vol. 30. V. Kempter, Electronic Excitation in Collisions between Neutrals, Ado. Chem. Phys., (Wiley, Chichester, 1975), vol. 30. M. A. D. Fluendy, K. P. Lawley, J. M. McCall and C. Sholeen, Faraday Disc. Chem. SUC., 1977, 62, 149. J. M. McCall and M. A. D. Fluendy, J . Phys. E, 1978,11,631. ed. R. A. Rosner, B. K. Penney and P. N. Clout (Advance, London, 1975). W. E. Wentworth, R. George and H. Keith, J. Chem. Phys., 1969,51, 1791. ’ M. A. D. Fluendy, J. H. Kerr, J. M. McCall and D. Munro, On-line Computing in the Laboratory, * J. A. Aten, G. A. H. Lanting and J. Los, Chem. Phys., 1977,19,241.