Reactions of Alkali Metal Atoms with Carbon Tetrachloride Rainbow-like Coupling of Product Angle and Energy Distributions BY STEPHEN J. RILEY,? PETER E. SISKA~ AND DUDLEY R. HERSCHBACH Department of Chemistry, Harvard University, Cambridge, Massachusetts 021 38, U.S.A. Received 9th April, 1979 Velocity distributions of alkali metal chlorides reactively scattered from crossed thermal beams of K, Rb or Cs and CC14 have been measured over the range 100-lo00 m s-' at laboratory angles from 10 to 100" with respect to the parent alkali beam. The differential cross sections for reactive scat- tering in the centre-o_f-mass system show strong coupling between the peak position of the product angular distribution 6' and the final relative translation91 energy E', and vary markedly with the identity of the alkali metal atom.For a given alkali metal, 6' shifts to smaller angles as E' increases, and as K -+ Rb -+ Cs the entire pattern shifts toward the forward hemisphere. These properties suggest an analogy to the rainbow effect familiar in elastic scattering. The product distributions can be simu- lated by a simple dynamical model. The most important features are the reaction probability as a function of initial impact parameter, the repulsive force causing dissociation of the unstable CC4- intermediate formed by transfer of the alkali metal valence electron, and the T --f V, R energy transfer induced by release of this repulsion during formation of the product bond. The product velocities obtained 25 years ago for Cs + CC14 by Bull and Moon agree with our data within z 10 %.Molecular beam chemistry in the " early alkali age " drew encouragement from the remarkable pioneering experiment of Bull and M0on.l They bombarded a stream of Cs vapour at right angles by a pulsed, accelerated CCl, beam produced by " swat- ting " with a paddle attached to a high speed rotor. Signal pulses from scattered Cs or CsCl were recorded by a surface ionization detector. Although there was no direct means to distinguish between Cs and CsCI, the observed pulses were attributed to reactively scattered CsC1, on the basis of time-of-flight analysis and blank runs with the CCl, replaced by Hg vapour. These experiments were not emulated, however, and indeed were wrongly discounted in contemporary reviews, evidently because it was expected that elastic scattering must always outweigh reactive scattering.We have previously shown2 that the contrary holds for Cs + CC1, in the pertinent angular region and hence Bull and Moon should be vindicated. On this Silver Jubilee occa- sion, we report a quantitative reactive scattering study which reaffirms the results of Bull and Moon and also reveals a strong correlation between the product scattering angle and translational recoil energy. A correlation between the preferred direction of recoil of the products and the magnitude of the reaction cross-section was one of the earliest themes to emerge from molecular beam experiments and trajectory studies of reaction dynamics. For CHJ, reaction is limited to small impact parameters and most of the alkali halide scatters backwards (" rebounds ") with respect to the incident alkali metal beam.For t Present address : Department of Chemistry, Yale University, New Haven, Connecticut 06520, $ Present address : Department of Chemistry, University of Pittsburgh, Pittsburgh, Pennsyl- U.S.A. vania 15213, U.S.A.28 REACTIONS OF ALKALI METAL ATOMS WITH CCl, Br,, reaction at large impact parameters is domiflant and most of the product goes forward (" stripping "). For CC14, the reaction cross-section is of intermediate size and the product peaks sideways, giving a conical angular distribution about the direction of the initial relative velocity ~ e c t o r . ~ Velocity analysis experiments likewise show a nice contrast. The product translational energy is large for the CH31 case4 and small for the Br, case;' the angular distributions do not change much with the product translational energy or with the identity of the alkali metal atom.For the CC14 case, we find a contrary trend: the preferred direction of the product shifts for- wards rapidly as the translational recoil energy increases and also as K + Rb -+ Cs. Similar product angle-energy coupling is often found in nuclear physics, where it is usually associated with an angular momentum restriction6 or with the effect of a Coulomb barrier.7 Also, a theoretical study of the H + H2 reaction' predicts that the product distribution shifts from backwards-peaked to a forward-directed cone as the collision energy (about equal to the product recoil energy in this case) is increased to values 2-5 times the barrier height.In all these examples the form of the product angle-energy correlation resembles the " rainbows " of elastic scattering. EXPERIMENTAL CONDITIONS The apparatus and experimental procedures are described elsewhere. The reactant beams, which intersect at 90°, emerge from ovens mounted on a platform that is rotated to scan the scattering angle. A two-filament surface ionization detector is used to distinguish reactively scattered MCl from the nonreactively scattered M atoms. The velocity distribu- tions are measured with a small, slotted-disc analyser.'' The resolution is 20 % (f.w.h.m.) and velocities from 20 to 1000 m s-' can be conveniently measured. Parent beam para- meters are given in table 1 . To facilitate kinematic calculations, the observed number density distributions are fitted to a functional form, P(u) = (u/6)" exp ((n/m)[l - ( ~ / 0 ) ~ ] } .TABLE 1 .-VELOCITY DISTRIBUTIONS IN PARENT BEAMSO beam T,/K T,/K a/m s-' 6/m s-' n, ml nz mz cs Rb K CCI4 623 553 281 376 4.5 3.2 10.4 1 .o 623 543 348 456 11.7 0.7 10.9 0.9 653 613 527 712 9.8 0.8 5.3 1.5 328 - 190 235 2.2 3.2 3.1 1.8 Temperatures Tu and TL refer to the upper (source) and lower (supply) chambers of the alkali metal beam oven. The quantity a = (2kTu/M)+ is the most probable velocity for a Maxwell-Boltzmann velocity distribution. Other parameters are defined in eqn (1); subscript 1 refers to u < 6, subscript 2 to v > 8. The D parameter represents the peak velocity and the exponents n and m are obtained from a least-squares fit to the data points.(For a Maxwell-Boltzmann beam, D = o! and n = rn = 2.) The CC14 beam velocity distribution was measured in an auxiliary experiment employing the negative surface ionization technique." The usual surface ionization filaments were replaced with an activated, thoriated tungsten filament of low work function, and the filament bias was reversed so that negative ions could be collected. Beam modulation and phase-sensitive detection were employed to suppress contributions from background electron emission. Under these conditions, a CCI4 beam yields C1- ions with z 1.5 % efficiency.S . J . RILEY, P . R . SISKA AND D . R . HERSCHBACH 29 RESULTS AND KINEMATIC ANALYSIS Fig. 1 shows velocity distributions of reactive scattering at several laboratory (LAB) angles.The relative intensities at the various angles were determined by normalizing the integrated flux (area under the velocity scan) at each angle to the total flux as measured in the previous angular distribution experiment^.^ The peak veloci- ties shift rapidly with laboratory angle, much more so than for any other alkali metal 0 500 1000 0 500 1000 lab velocity / m s-l 0 5 00 1000 FIG. 1 .-Velocity distributions of reactively scattered alkali metal chloride at indicated laboratory angles, for reactions of (a) Cs, (b) Rb and (c) K with CC14. The parent alkali metal beam is directed at 0”, the CCI4 beam at 90”. Open circles are experimental points, solid curves are smoothed data. Dotted curves show distributions back-calculated from the nominal centre-of-mass cross-section of eqn (3), dashed curves from the least-squares fitted polynomial expansion of eqn (4).The ordinate scale is such as to assign 10 units to the largest experimental peak for each reaction (60” for K, 50” for Rb, 30” for Cs).30 REACTIONS OF ALKALI METAL ATOMS WITH cc1, system so far studied. A distinct bimodal structure is also present, again varying strongly with angle. The low velocity components correspond approximately to the LAB velocity of the centre-of-mass in each case. The observed intensity at a particular LAB velocity and angle, ILAB(v, 6), is re- lated to the differential reaction cross-section in the c.m. system, Icm(u, 0), by an aver- age over the parent beam distributions. This is given by ILAB(V, 8 ) = dmdvl jomdv2ni(vi)nl(vz) vJIcm(u, 8>, (2) where v1 and v2 are the velocities of the reactant beams, n, and n, the number densities, V = ( u l + uZ)* the initial relative velocity, and J = v2/u2 is the Jacobian factor for the c.m.-+ LAB flux transformation. In previous work Icm(u, 8) has been extracted by approximate iterative methods which usually assume the velocity and angle de- pendence to be separable. Such methods proved utterly unsuccessful for the CC14 reactions. This led to the development of two much more efficient procedures, neither of which involves assuming separability of the velocity and angle dependence. The first method merely carries out the LAB -+ c.m. transformation directly by selecting a single representative velocity in each of the parent beams in order to define the transformation uniquely.Then, aside from a normalization factor, The resulting " nominal " c.m. cross-section depends to some extent on the velocities used for the parent beams. However, trial calculations using synthetic data con- firm that, for values of v1 and v2 near the most probable velocities of the parent beams, the nominal result provides a good approximation to the true cam. cross-section. A " best fit " nominal cross-section can be determined by varying v1 and v2 and back- calculating ILAB(v, 0 ) for comparison. Table 2 gives " best fit '' parameters. TABLE 2.-vELOCITY PARAMETERS FOR KINEMATIC ANALYSIS' system u1 u2 UO cs + cc14 450 260 250 Rb + CCl, 500 200 400 K + CC14 800 187 800 Units are m s-l; ul(alkali metal) and v4(CC12) are velocities of the parent beams used in the nominal kinematic transformation of eqn (3); uo is the width parameter for the least-squares fitted polynomial expansion of eqn (4).The second method employs the c.m. --+ LAB transformation with Icm(u, 6) repre- sented by a convenient functional form, Icm(u, 6) = exp [ - ( u / u , ) ~ ] 2 ai juku$. (4) ij Here u, = u cos 6, u, = u sin 6 are the Cartesian coordinates of the c.m. velocity vector. The gaussian factor causes the cross-section to vanish as u +- 00, at a rate governed by the width parameter, uo. The coefficients a i j in the polynomial factor are determined by a least-squares fitting procedure, which compares the transformed and velocity-averaged intensity computed from eqn (2) and (4) with the observed LAB data.Computational details and an analysis of the method are presented else- where,12 including criteria for choosing the size of the coefficient matrix, aij. A 6 x 6S . J . RILEY, P . E . SISKA A N D D. R . HERSCHBACH 31 matrix was used here. The only free parameter in this procedure is uo, and it may be varied to optimize the fit. The result is not strongly dependent on the choice of uo. Table 2 includes the values adopted, which are z 10 % lower than the peak velocity of the nominal c.m. cross-section. Fig. 1 shows the LAB distributions backcalculated from both the nominal and least squares c.m. cross-sections. As expected, the nominal cross-section tends to be too broad and incapable of reproducing fine structure, since it incorporates the velocity averaging inherent in the experiment.Yet the quality of the fit obtained with the nominal method is much better than any achieved using the previous iterative approxi- mations. The least squares method reproduces the data more closely; the standard error of fit is 2.7 % for Cs, 5.1 % for Rb and 3.4 % for the K data. Also, whereas the nominal method deteriorates at the " edges " of the data (at high velocities and at 10 and looo), the least-squares method provides smooth extrapolations beyond those regions. Fig. 2 shows c.m. differential cross-sections, plotted as angular distributions for various c.m. velocities. The similarity between the nominal and least-squares results again indicates the value of the quick and essentially trivial nominal procedure.The most striking aspect of these distributions is the strong coupling between c.m. velocity and angle. As the product exit velocity increases, the peak of the angular distribution shifts rapidly to smaller scattering angles. The width of the angular distribution also decreases as the exit velocity increases. Another notable feature is the rapid drop in intensity on the small angle side of each peak, in contrast to a more gradual fall-off towards large angles. Fig. 3 shows the flux distribution of final relative translational energy of the products, E' (for MC1 us. CCl,) at various angles. Aside from normalization, this is given by P(E', 8) = Icm(u, O ) / U . ( 5 ) The abscissa scale is in terms of the fraction, ftrans = E'/Etot, of the total available energy, E,,, = E + Eint + ADo.Here E and Ei,, are representative initial relative translational and internal energies of the reactants. ADo is the reaction exoergicity obtained from the MCI and Cl-CC1, bond dissociation ene~gies.'~.'~ Values for the K/Rb/Cs cases are: E = 6.3/6.7/8.4; Eint = 2.5 (diatomic approximation); ADo = 155/134/138 kJ mol-'. The corresponding fraction of energy appearing as vibrational and rotational excitation of the products is given byfint = 1 - ftrans. The strong coupling between the product translational energy and scattering angle is again evident in these distributions. The " mountain ridge " or locus of largest intensity is indicated by a dashed curve. The deviation of these loci from a circle is another measure of the coupling. Also evident is the marked shift to higher exit velocities and wider scattering angles as Cs -+ Rb -+ K.The contour maps show evidence of some low velocity structure at small angles. This structure corresponds to the bimodal form of the LAB velocity distributions where it is enhanced by the u2/u2 Jacobian in the c.m. -+ LAB transformation of eqn (2). The low velocity structure is likewise enhanced in the energy distributions by the u-' factor in eqn (9, as indicated by the dashed portions of the curves in fig. 3. However, in the contour maps it is a relatively minor feature and will not be considered here. The low velocity region in c.m. space is the region most severely " washed out " by velocity averaging, and a reliable quantitative determination of such structure cannot be obtained without resorting to velocity selection of the parent beams.Fig. 4 displays contour maps of the c.m. cross-section.32 REACTIONS OF ALKALI METAL ATOMS WITH cc1, 1 . 0. c.m. scattering angle, 8 /O FIG. 2.-Angular distributions of reactively scattered alkali metal chloride for various velocities u in the centre-of-mass (c.m.) system, for reactions of (a) Cs, (6) Rb and (c) K with CCI.,. By convention, 0" refers to the initial c.m. direction of the reactant alkali metal atom velocity, 180" to the CCI4 c.m. velocity. Dotted curves show results from nominal kinematic transformation, dashed curves those obtained from the least-squares procedure.S . J . RILEY, P . E . SISKA AND D . R . HERSCHBACH 33 f int 1.0 0.9 0.8 0.7 0.6 0.5 1.0 0.5 0 rr H ; 1.0 L - Y- Y Q z U .- In c al T) 3 x 0.5 4 )I 0 1.0 0 .5 C 0 0.1 0.2 0 . 3 0.4 0.5 fraction of product energy in translation, ftrans FIG. 3-Energy distributions at various c.m. scattering angles for alkali metal chloride product, derived from the least-squares fitted polynomial cross-sections, for reactions of (a) Cs, (6) Rb and (c) K with CC14. Upper abscissa scale gives fraction of available energy appearing in internal excitation, lower scale that in relative translation of products.34 REACTIONS OF ALKALI METAL ATOMS WITH cc1, 90" 0" 90' FIG. 4.-Contour map of differential cross-sections for reactively scattered alkali metal chlorides, obtained from least-squares procedure, for reactions of ( a ) Cs, (b) Rb and (c) K with CC&. The intensities are normalized to 100 and contour lines are shown for each 10 units.Tick marks along radial lines appear every 100 m s-l. Open circles and dashed curves indicate the " mountain ridge ", or locus of maximum intensity. Table 3 compares the LAB velocity of CsCl in the Bull and Moon experiments with that computed from our least-squares c.m. cross-section. The velocity of the CC14 beam was taken to be that of the paddle tip with a thermal distribution super- imposed. The Cs beam, also thermal, was given a cosine distribution of directions about 0". The calculated LAB velocity distributions of CsCl scattered at 80, 90 and 100" were averaged to approximate the degree of collimation used by Bull and Moon, and multiplied by two additional factors of velocity to correspond to time-of-flight signal pulses.The agreement is better than 10 %, certainly as good as could be ex- pected in view of the approximations required and uncertainties about precise ex-S . J. RILEY, P. E. SISKA AND D. R . HERSCHBACH 35 TABLE 3.4OMPARISON WITH BULL AND MOON'S EXPERIMENT' rotor tip speed product pulse speed Bull and Moon calc. 1.52 4.7 2.92 4.9 4.33 5.2 6.52 6.7 4.6 5.1 5.6 6.2 Units are lo2 m s-l. perimental conditions. The fact that the agreement remains good at all four rotor speeds is of particular interest, since our calculations assume Icm(u, 0) does not vary significantly with the initial relative translational energy of the reactants. This varies from ~ 4 . 2 to 24 kJ rno1-I in the Bull and Moon experiments, according to the observed speed of the CCI, pulses.RAINBOW MODEL In elastic scattering, rainbow structure usually results from an attractive potential well. However, a repulsive barrier can also produce a rainbow which is similar in all respects except that the rapid fall-off in intensity occurs toward narrow angles rather than wide angles." This is just the behaviour observed for the CC1, reactions. Since elastic rainbows of the usual kind are observed for these systems,I6 any barrier in the reactant trajectory would have to occur rather late in the entrance channel. Indeed, the electron transfer which governs the reactions appears more likely to intro- 0 0 1 2 5 10 20 reduced recoil energy, K FIG. 5.-Rainbow plot for the Morse and Lennard-Jones potentials [ref. (18) and (19)], and for the H + H, reaction [circles, ref.(S)] and deuteron stripping reaction [squares, ref. (7)]. For the H + Hz reaction, the characteristic energy E is taken as 0.4 eV, the minimum saddle-point height for the potential surface used in ref. (8). For the ( d , p ) reaction, the points are shifted an arbitrary amount, corresponding to an E of 25 MeV.36 100 50 \ aD' a; w (31 C tl 0 a t 3 .- E! 20 10 I I I I I I I I REACTIONS OF ALKALI METAL ATOMS WITH Cc& 1 1 1 A 0 0 O O 0 O O 0 0 0 0 0 A A A 0 0 0 0 A A A FIG. 6.-Rainbow plot for reactive scattering of K(O), Rb (0) and Cs (A) atoms from CCI, derived from the least-squares fitted polynomial cross-sections. duce a barrier in the exit ~hanne1.I~ The venerable Evans-Polanyi discussion of curve-crossing suggests the barrier height would vary with the product bond strength and thus would decrease in the order K 21 Rb > Cs.Semiclassical theory shows the rainbow structure is governed by an Airy function form factor, with the rainbow angle 0, located on the " dark " side at a point where I(0) sin 0 has 44 % of its peak inten~ity.'~ At high energies, Or becomes inversely proportional to the ratio K = E / E of the collision energy to the well depth or barrier height. Fig. 5 com- pares the &(K) functions computed for a Morse potential16 and a Lennard-Jones (12, 6) potential" with the high energy approximation for the latter case, 0, = 2/K. This shows the angle-energy correlation depends only weakly on the form of the potential. Data points are included for two reactions which are definitely governed by repulsive barriers: a (d, p ) deuteron stripping reaction and the H + Hz reaction.8 The notion of" reactive rainbows " appears to have some merit for these systems.Fig. 6 presents a similar plot for M + CCl,, derived by applying the usual analysis for elastic scattering" to our product distributions. The rainbow model looks dubi- ous, since Or varies more like the inverse square root rather than the inverse first power of the energy. The slopes increase at large E'. This is suggestive because the same trend is noticeable in fig. 5 at low energies; a barrier causes a negative deviation from the K - l line in contrast to the positive deviation caused by a potential well. The curvature thus may indicate that the recoil energies for the CCl, systems are too low Fig.5 and 6 examine the heuristic notion of " reactive rainbows ".S . J . RILEY, P . E . SISKA A N D D . R . HERSCHBACH 37 to show the characteristic K-’ behaviour. If fig. 6 does pertain to rainbow scattering, the relative position of the curves indicates that the barrier height decreases in the order K > Rb > Cs (and the relative heights are roughly in the ratio 5: 3 : 1). This phenomenological rainbow analysis does not deserve much credence. However, it has led to a rather simple model which appears to give a comprehensive account of the reaction dynamics for the CC14 systems as well as the CH41 and Br, systems. DYNAMICAL MODEL There is now a large repertoire of reactive scattering models that are more tractable than full-scale classical trajectory calculations and call for much less information about the potential surface.21 Although we did not find a previous model suitable for the CCl, reactions, the treatment outlined here borrows from several such models; we refer to it as eclectic (to avoid a crowd of acronyms).Our model deals specifically with the electron transfer process, A + BC --f A+ + B-C -+ A+B- + C, and has four main components : the deflection function is taken as the sum of reactant and product portions, 0, + O,, each governed by a two-body central force potential. In the entrance channel the BC bond is fixed at its equilibrium distance for the ground vibrational state; in the exit channel AB is fixed at the r-centroid distance corresponding to its vibrational excitation.The entrance channel potential Vi, is determined from a 2 x 2 secular determinant. The diagonal elements represent covalent (Lennard-Jones) and ionic (Rittner) interactions and the off-diagonal element contains the ionic-covalent coupling term.23 The exit potential V,,, prescribes the decomposition of the transient B - C ion-molecule and is determined from dissociative electron attachment data. 24 (2) As in the DIPR the switch between the entrance trajectory and the exit trajectory is abrupt and the repulsive energy release is evaluated from a Franck- Condon approximation in analogy to photodissociation.26 (3) As in the infinite-order-sudden appro~imation,’~ the collision mechanics is much simplified by fixing the angle between the molecular axis and the radius vector from the collision partner to the molecular centroid.This angle is denoted by q in the entrace channel and 11’ in the exit channel. Thus, only a single representative geometry is specified for the ABC complex at the switchover point. (4) As in the “ half-collision ” model for photodissociation,28 the partitioning of the abruptly released repulsive energy between relative translation E’ of AB and C and internal excitation Elint of AB is estimated from an impulsive approximation akin to the Landau-Teller model for translational-to-vibrational energy With this eclectic model, we find rainbow-like reactive scattering can readily be obtained. The most important factors involved are the range of the reaction proba- bility P,(b) as a function of initial impact parameter; the transition-state geometry ; and the T --f V, R energy transfer induced by the repulsive energy release. Two formulae suffice to describe the role of these factors.The differential cross-section is given by (1) As in the extended optical Here x = 0, + 0, is the classical deflection function; y(rBc) is the initial vibrational wavefunction; I 8E’/2rBcl is a Jacobian factor, the slope of the repulsive potential of the B-C ion at the initial bond distance. All quantities are computed for various38 REACTIONS OF A L K A L I METAL ATOMS WITH CCl, fixed rBC values and the cross-section is then averaged. The energy transfer is given by E&,/Eb = 4 sin2 p cos2 p sin2 (8,’/2)N(p). (7) Here Eh is the sum of the initial relative kinetic energy and the repulsive energy available for disposal in the products (Eb would be E’ in the absence of energy transfer); cos’p = (m,/m,,)(m,/m,,) is a kinematic mass factor, with p the skew angle for axes which diagonalize the kinetic energy.30 The function H(p) is given by Harris;28 it specifies the deviation from adiabaticity in terms of the ratio p of the collision duration to the vibrational period of the product molecule.In the impulsive limit, p = 0 and H = 1. In this limit, eqn (7) is Mahan’s formula for T -+ V transfer in collinear collisions,29 except for the sin2 (0:/2) factor, which in effect inserts an impact parameter. This important factor involves the product contribution to the scattering angle (the zero superscript again indicates a value obtained without including energy transfer).As usual in impulsive models,’* Ei,, may be resolved into vibrational and rotational components using the transition state geometry. Eqn (6) exemplifies the essentially geometrical character 22 of the correlation be- tween the preferred direction of product recoil and the magnitude of the reaction cross- section. If P,(b) is sufficiently long-ranged, the reactants can swing around each other in grazing collisions and the exit repulsion cannot overcome the strong preference for forward scattering that results from the b/sin 0 factor. If P,(b) is shorter-ranged, only closer collisions give reaction and the repulsion then kicks the products apart sideways or backwards. Likewise, for noncollinear transition-state configurations, certain angular regions become inaccessible due to the combined effects of exit repul- sion and angular momentum conservation, Eqn (7) also provides a strong dependence on initial impact parameter.When b is large and the products tend to scatter forwards, the energy transfer delivers more of the exit repulsion into translation whereas the opposite occurs at small impact parameters which favour backward scattering. The magnitude of the effect depends strongly on the mass factors and the adiabaticity parameter. This energy transfer is the major factor governing the rainbow-like coupling of the product angle and energy distributions in the CC14 reactions. Since E’ is thereby enhanced for the more- forward-scattered products which carry large centrifugal angular momentum, the exit parameter b’ shows a maximum as a function of b and hence produces a minimum in the part of BP governed by the exit energy transfer.Although this does not yield axlab = 0, it does give a local minimum in axlab which serves to focus the product intensity in the corresponding angular range. Thus, the eclectic model produces “ energy-transfer rainbows”. Product angle-energy contour maps obtained from the model simulate the experi- mental results quite well for the CCl,, CHJ and Br2 reactions. Fig. 7 shows such maps for the CCl, case. As with any such model, there is considerable leeway in the choice of some parameters, The most reassuring aspects are the comprehensive character of the agreement and trends governed by nonadjustable parameters : the atomic masses.Thus, for the CCl, systems the mass dependence of eqn (7) accounts for the variation seen in the E‘ against 0 coupling with change of alkali metal atoms, including the lithium reaction which shows only weak ~oupling.~’ This also accounts for the modest but definite E’ against 8 coupling seen in the Br, reactionsI2 and the lack of such coupling in the CH,I case., For our nominal choice of potential para- meters, we find for the CC14 systems a somewhat bent M/ \C configuration c1S . J . RILEY, P . E . SISKA A N D D . R . HERSCHBACH 39 9 0' ( a ) 120" t 60' 60' ( b ) 90" 120' T / 5 I O0 FIG. 7.-Contour maps for reactively scattered alkali metal chlorides calculated from the eclectic model. Notation and format as in fig. 4. ( a ) Cs + CC14, (b) Rb + CC14, (c) K + CC14.(with q z 20°, q' z l 5 O ) enhances substantially the rainbow-like behaviour of the scattering. Such features suggested by the model invite closer study by means of trajectory calculations. It is fitting that the molecular dynamics of the CCl, reactions seem to involve spinning and swatting, much as in the apparatus of Bull and Moon. Support of this work by the National Science Foundation is gratefully ack- nowledged. T. H. Bull and P. B. Moon, Disc. Faraday Sue., 1954,17, 54. D. R. Herschbach, in Chemical Lasers, Appl. Optics Suppl., 1964, 2, 193. K. R. Wilson and D. R. Herschbach, J. Chem. Phys., 1968,49, 2676. R. B. Bernstein and A. M. Rulis, Faraday Disc. Chem. Soc., 1973, 55, 293. J. H. Birely and D. R. Herschbach, J. Chem. Phys., 1966, 44, 1690.40 REACTIONS OF ALKALI METAL ATOMS WITH Cc14 S.T. Butler, Proc. Roy. SOC. A, 1951, 208, 559. M. Karplus and K. T. Tang, Disc. Faraday SOC., 1967, 44, 56. S. J. Riley and D. R. Herschbach, J . Chem. Phys., 1973, 58, 27. ' L. C. Biederharn, K. Boyer and M. Goldstein, Phys. Rev., 1956,104,383 and work cited therein. lo R. Grice, Ph.D. Thesis (Harvard University, 1967). l1 A. Persky, E. F. Greene, and A. Kuppermann, J. Chem. Phys., 1968,49, 2347. l3 L. Brewer and E. Brackett, Chem. Rev., 1961, 61,425. l4 T. L. Cottrell, The Strengths of Chemical Bonds (Butterworth, London, 1958). l5 K. W. Ford and J. A. Wheeler, Ann. Phys. (N.Y.), 1959,7, 249. l6 R. M. Harris and J. F. Wilson, J. Chem. Phys., 1971, 54, 2088. P. E. Siska, J. Chem. Phys., 1973, 59, 6052 and work cited therein. M. G. Evans and M. Polanyi, Trans. Faraday Soc., 1938,34, 1 1 ; J. C. Polanyi, Chem. Phys. Letters, 1967, 1, 421. R. B. Bernstein, in Atomic Collision Processes, ed. M. R. C. McDowell (North Holland, Amsterdam, 1964), p. 895. l9 E. A. Mason, R. J. Munn and F. J. Smith, J. Chem. Phys., 1966,44,1967. 2o E. F. Greene, G. P. Reck and J. L. J. Rosenfeld, J. Chem. Phys., 1967,46, 3693. 21 J. C. Polanyi and J. L. Schreiber, in Physical Chemistry, An Aduanced Treatise, vol. VIA, Kinetics of Gas Reactions, ed. W. Jost, (Academic Press, New York, 1974), p. 383. 22 D. R. Herschbach, Adv. Chem. Phys., 1966, 10, 319; J. L. Kinsey, G. H. Kwei and D. R. Herschbach, J. Chem. Phys., 1976, 64, 1914. 23 R. Grice and D. R. Herschbach, Mol. Phys., 1974,27,159; R. W. Anderson and D. R. Hersch- bach, J. Chem. Phys., 1975, 62, 2666; S. A. Adelman and D. R. Herschbach, Mol. Phys., 1977, 33, 793. 24 W. E. Wentworth, R. George and H. Keith, J. Chem. Phys., 1969, 51, 1791. See also S. M. Lin, D. J. Mascord and R. Grice, Mol. Phys., 1974, 28,975. 25 P. J. Kuntz, M. H. Mok and J. C. Polanyi, J. Chem. Phys., 1969, 50,4623. 26 D. R. Herschbach, Faraday Disc. Chem. SOC., 1973, 55, 233. 27 R. T. Pack, J. Chem. Phys., 1974, 60, 633; D. Secrest, J . Chem. Phys., 1975, 62, 710; L. W. Hunter, J . Chem. Phys., 1975, 62, 2855. 28 K. E. Holdy, K. C. KIotz, and K. R. Wilson, J. Chem. Phys., 1970,52,4588; R. M. Harris and D. R. Herschbach, J. Chem. Phys., 54, 3652. 29 B. H. Mahan, J. Chem. Phys., 1970,52, 5221. 30 F. T. Smith, J. Chem. Phys., 1959, 31, 1352. 31 D. D. Parrish and R. R. Herm, J. Chem. Phys., 1971,54,2518; C . M. Sholeen and R. R. Herm, J . Chem. Phys., 1976, 65, 5398.