————_—_-. ——— xi PREFACE, In organizing the present volume, various authors were invited to prepare ‘chapters. Each author isa world leader in the field of intramolecular and nonlinear fashion. The contributors tothis inaugural volume have ll future contributors to Advances in Classical Trajec- William L. Hase Series Editor *D.L, Bunker, J. Chem. Phys., 1962, 37,393; 1964, 40, 1946. CLASSICAL TRAJECTORY STUDIES OF INTRAMOLECULAR DYNAMICS: LOCAL MODE DYNAMICS, ROTATION- VIBRATION INTERACTION AND THE STRUCTURE OF MULTIDIMENSIONAL PHASE SPACE Gregory S. Ezra2 GREGORY 8. EZRA V. Local Frequency Analysis and Multidimensional Phase Space Dynamics... 0. eee eee eee eeeeeees 31 VL. Future Directions 36 ‘Acknowledgments 36 References 37 the use of fea! frequency analysis for multimode sysems. 1. INTRODUCTION phase space structure? ee a eee ‘motion? on typical molecular potential energy surfaces and the consequences for Specrcscony and dyanics has esived mich seo a has the rol of Trajectory Studies of Intramolecular Dynamics 3 consequences of couplings between vibration- rotational degrees of freedom have been the ‘of new experimental methods alow the {es ofpolyatomicmoleculesto be probed challenges to theory. An important class the overtone excitation of XH bonds (where ion provides bth» webespecrecric pl 0 degree of freedom systems, at least) using the fun- 3 Bensimon and Kadanoft$ and Channon and ‘that exhibit lasicaly chaotic bet (as opposed to integrable systems, for which the classcalfquantum correspondence4 GREGORY 5. EZRA ‘This Chapter is organized as follows. In Section I we deseribe ovr classical 4 Panicular em- of hydrogen peroxide using a realistic potential surface” are discussed. In ‘Addition, we discuss the validity of classical trajectory methods for studying local toned; this approach forms anatural generalization ‘of MacKay, Meiss and Percival Some conclud- ing remarks are given in Section VI lI, DYNAMICS OF OVERTONE STATES ‘A. CH Overtone Dynamics in Benzene: A Model Study XH stretching vibrations difficult due tothe very large muraberofstates involved (although see Trajectory Studies of nramolecular Dynamics 5 ow-resolution overtone absorption spectra can be classical mechanics (see below). © Ha) a f dt expliorlea0), a) ‘An approximate low-resolution spectrum may be obtained by substituting et) = Pale @ ‘quantum calculations using @ limited tr-model basis.” Las and Hase"? have ne linewidths for benzene and perdeuterobenzene using the ‘precise form of this type of potential coupling has been determined i faleulations on various molecules, and the results expressed in terms of a6 GREGORY 5. EZRA switching function (7). The bending force constant asa function of bond length r ea?) = Hine). Oy ‘The sensitivity of overtone dynamics to the nature ofthe switching function is @ ‘question ofsomeimportance. The quasiclasscal trajectory calculations of Lu, Hase and Wolf showed that the CH overtone decay in a model for planar benzene ‘depends noticeably on the presence or absence of potential coupling:® increasing the strength ofthe stretch-bend potential coupling was found to decrease the decay rate for those states exhibiting exponential decay of probability (cf Fig. 1) and to move the onset netial decay (cf. Fig. 1b) to lower values of ca, the smumber of quanta in the CH bond ‘To obtain futh {into the influence of stretch-bend potential coupling, ‘on overtone dynamics, a two-mode model system was studied" The Hamiltonian ‘was chosen to represent the in-plane streich/bend motion of the HCs fragment of, benzene (CH stretch coupled to CCH wag), andis given in detail in Ref. 41. Here, ‘we note only thatthe Hamiltonian includes both kinetic (G-mnatrix) and potential ‘coupling, the latter only through the switching function: Seat rere ) S() = expl-olr~ rah 7> 70 (at) ‘The switching function (Eqs. 4) was employed by Lu et al in thei trajectory calculations on planar benzene, and is a simple yet reasonably accurate repre- ‘entation of the bending force constant ‘correspond (with zero-point energy deposited the bend) was foundto display the ‘with changes in the coupling parameter as the full planar benzene Hamil ‘and values forthe decay rates were obtained in reasonable agreement wi crease slightly the rate of probability decay for those values of nex exhibiting cxiorenta dey, while increasing the coupling to type 3 eliminates exponential . Te ality of our modelo reproduce reds found in ullscle ‘overtone decay. “The phase space structure of two degree of freedom system such as the discussion can be studied using the Poincare surface of section oO) g 8 8 g 8 ‘yo00 ors cas 004s 0080 0075 0090 0105 01 Me g g g g g %o00 0015 0.030 0.045 0.060 0075. TE (6s)8 GREGORY 5S. EZRA (©, ps)every timer= rp, p> 0 for several trajectories ata particular constant energy. 3 Figure 2a coresponds to type 3 coupling and energy equal othe zeotvorder a 15 910 GREGORY 5. EZRA Figure 2. (contiwed} ‘opento question. The experiments of Page et." on overtone spectaofje-conled ‘quasiclassical CH overtone linewidh for planar benzene with the fraction of 2e10- otmal modes.” By educing the zero-point Denzene by a factor fy = 0.04, a linewidth a satisfactory theoretical justification. The role of zero-point energy in classical simulations of linewidths deserves a thorough ‘Trajectory Studies of haramolecular Dynamics " reappraisal in the light of recent suggestions for imposing constraints on the energy/mode in classical trajectories.” Further aspects of the clssical/quantum correspondence for linewidths and unimolecular decoy rats are discussed below. B. Overtone Induced Dissociation of HOOH peroxide molecule (HOOH) in the gas phase however incomplete on thet ‘and, both the rate and productstat distribution inthe overtone induced disocis, tion of HOOH ean be accounted for using a statistical adiabatic channe theory”* puted by Uzer i channel calcuations”™ ‘components, suggestive of the existence of phase space bottlenecks, although2 GREGORY S. EZRA ‘other explanations are possible (Se alo the recent classical trajectory caleula- tions by McDonald etal). Extension of these measurements 1 the Vor = 6 case ‘would obviously be desirable, ‘Thefirst classical trajectory calultion ofthe overtone induced decomposition of the vou = 6 state of nonotating hydrogen peroxide were carried out by Uzer et a1 and by Sumpter and Thompson." In altionto giving the calculated dissocia tion lifetimes noted above, these studies ilustate the importance of the HOO bending modes in the shor-time transfer of energy out ofthe OH overtone and also reaction coordinate, gains energy only slowly until a sudden “impulsive” of energy to the OO bond occurs followed by immediate dissociation A sig- “The potential energy surface used by Uzer et al incorporates spectroscopic and thermodynamic data, and models the OH and OO bond sireches as Morse oillatrs, and the HOO bends as harmonic oscillators. (Torsional motion was between the zeroth-order modes is introduced in two ways: off-diagonal terms in the F-matrix; and switching functions describing the attenuation of the HOO harmonic bend force constant by stretching ofthe adjacent HO or OO bond. As noted in Refs. 42 and 82, tis potential surface is unsatisfactory in several respects. ‘The switching functions used in Ref. 82 actually describe stretch-bend potential coupling in a CCH fragment: switching functions appropriate for the OOH fragment have since become available“ In adition the off-diagonal F-matrix Shanges in the assumed potential surface. n the light ofthe results on CH -decay in benzene discussed above and of the pathways for energy flow xed by Uzer et al, we expect that the form of the stretch-bend potential ‘be important in determining dissociation lifetimes, and this expecta- decay of HOOH using potential surfaces in which the shortcomings ofthe surface Trajectory Studies of Intramolecular Dynamics 3 ‘of HOO bend force constant with (b) Function describing HCH *fevel Figure 3._ Switching functions describ (00 bond length. (a) MP2/6-3 attenuation In CHa. () ab initio results at of Uzer etal. have been corrected “The elation between the timescale for classical ‘unimolecular decay and that for intramolecular vibrational energy redistribution ‘has also been examined, The potential surfaces used are ab intio/empirical func tions of the form used by Uzer et al. but modified so thatthe deficiencies noted ‘above have been remedied. First, the potentials behave in a physically reasonable fashion a roo —* ©. Off-dingonal harmonic force Field potential coupling terms ste fit to a sum of nonbonded Lennard-Jones interactions, which decay £02840 38 700 = ®. In addition, the OH bond force constants relax to thei values for free (OH, and the torsional potential (included in our calculations) i represented as @ dissociates. The role of the switching ‘AD ino caleulaons (using the Gaussian package) have been made to determi switching | functions S(roo) and S(roui) describing the potential energy tween the HOO bead and the OO and OH bond seth coordinates (ching functions S(roo) are plottedin Figure 3. The frst, 3, is determine o oust together with second- the sb initio results for fective bend force constant to the val equilibrium geometry,“4 GREGORY S. EZRA ‘hile the sold line labeled (a isthe fitting function. The switching function is ft remarkably well by a simple BEBO! form: ‘Sir00) = exp(-1.468 Aro). © ‘The second function 3b describes the HCH bend attenuation for methane, and as used in the trajectory calculations of Uzer et al.® The third function, 3c, was obtained using an SCF calculation with a smaller 6-31G** basis? and was used in the trajectory calculations reported in Ref. 42. rences between the various switching functions. The (presumably most accurate) function of curve 3a shows very strong stretch-bend, ‘coupling in the region of small bond displacements (-0.5 < Ar e, the absorption spectrum consists of a set of discrete peals, each of which is homogeneously ‘broadened with a width determined by the unimolecular decay rate Tp (Fig. 62). ‘Assoon as Tp becomes ig. 6). Provided the potential used is accurate and the application of classical mechanics warranted, our trajectory results f (6 ae consistent with the dissociation is sufficiently slow that it is possible to resolve transitions to ‘molecular eigenstates, and to ‘extract homogeneous contributions tothe linewidihs inthis case Tp» 0.006 em, whereas I # 0.6 cm) ‘Onemajor question regarding the above interpretation of te observed lineshape for HOOH concems the density of molecular vibrational eigenstates. An accurate calculation of the density of vibrational states of HOOH at the relevant energy rough est harmonic state ‘vibrational eigenstates with significant OH stretch charTrajectory Studies of tnramolecular Dynamics 8 intenaiy 08. “100 -50 oo 50 100 inthe experiment of Lnoet al), How many ofthese stetes are appreciably coupled to the OH stretch mode remains to be determined hy study of accurate dynamical models, "We now turn to the question ofthe validity of classical mechanics, both for calculating the unimolecular decay rate and the absorption profile, A general “discussion of the accuracy of classical trajectories for studying unimol processes and IVR has been given by Hase,’ while a seview of quantum nonclassical) effects in bimolecular reactions has been given by Schatz. classical/quantum correspondence for nonintegrable systems is an active area of study which is discussed inthe light of recent advances in noolinear dynamics in Ref. 39. Ti order to assess the i accuracy of « unimolecular decay rate calculated for a given potential surface using classical mechanics, we note that classical greater than of Tess than the ‘One reason the classical rate cam be greater than the quantum classical zero-point energy as discussed above in connect decay. Classically, zero-point energy placedin a molecule att ajectory ‘may transfer into motion along the reaction coordinate, thereby lowering the sm value. 50 100 dissociation rates can Figure 6. Absorption lineshapes ca WT; =40,To=20. Ti = 04, units arbitrary. lle’s mode la T= 4.0, To = 0.4 1.0 throughout (page 18). Intensity i “lo cera Figure 6. (continued) range of 0.8 cm" (J and K quantum numbers forthe initial and final state are known.20 GREGORY. EZRA classical threshold relative to the quantum value 87! 10 tum valve.®7! Methods for constraining the every pet mde o be grater tan te zero-point value toughnut slecony calculation have recently been; oat caleaaionhaere proposed,” but have not yet been extensively tested, ear triatomics.™ In the language of nonlinear dynamics, barriers broken scparatrices™) through which the clas- which are stable for all time in classical dissociate by a nonclassical mechanism. flow via dynamical tunnelling has also been discussed by for example by Kay ina detailed classical/quantum > Trajectory Studies of Intramolecular Dynamics a respect to the classical This discrepancy between classical and quantum mechanics has been ascribed to quantum trapping by cantor!or scars of periodic orbits," or to Anderson-type localization of quasienergy states™ (see also Ref. 114, Gray has shown forthe PFO model for He. dissociation that selection of trajectory inital conditions from the Wigner distribution comesponding to the initial (= 0) quantum state ves beter agreement with the quantum dissocis- tion rate than the usual quasilassical initial conditions with zero-point energy." ‘Unimolecular decay lifetimes for HOOH, von = 6, have been calculated using Wigner sampling with the potential suriace discussed above: lifetimes are = ‘with 28.3 ps (= 0) and 19.9 ps (with extensive comparison of quasilasical samy energy and Wigner sampling of intial conditions for HOO! (@=5,6, has beenmade by Getino Sumpter and Santamaria ‘model potential surfaces ‘consider the conditions under which the quasiclasical pr entirely consistent wi homogeneously broadened Ie2 GREGORY 5. EZRA UL. CLASSICAL TRAJECTORY STUDIES OF ROTATION-VIBRATION DYNAMICS ‘The interaction of molecular rotation with interna vibrational motion ian impor- tant, but as yt, a elatvely poorly undersood factor in determining rates of 1VR (Ref. 19 and references therein) and unimolecular decay.® Large-amplitude :* ‘ucla vibrations can generat substantial vibrational angular momentum, and.can ‘also lead to large changes in moments of inertia over a vibrational period; both these effects couple rotations with vibrations. The relative importance of rotation jn inducing energy flow between vibrational modes versus its role as an active ‘degree of freedom able to exchange enesgy with vibration i an interesting open question.#® Rotation-vibration (RV) are important for understanding ‘Although there ha recently been much progress in the development of quantum mechanical methods for calculation of RV states for realistic Hamiltonians at nonzero total angular momentum,!® classical and semi-classical approaches study of high angular momentum () states (j~ 50-100; recall nonrigid molecules,“ 4-dimensional representation of 1, @ comprehensible repre- smple, the rotational energy ‘or rotation in three dimensions, be followed, e., three Euler ‘There are however two space: the magnitude ofthe he total angular momentum ization of rigid asymm in their work on semict (Gee also Refs. 46 and 126). used by Frederick of the rigid bender Trajectory Studies of Inramolecular Dynamics 2 ‘A. Rigid Bender Model ‘coupling inthis model. Nevertheless, a rich variety of dynamical behaviors found. Asthere are only two degrees of freedom, a“rotational sos” can be usedto examine the dynamics. In this way we have explored the RV phase space as a function of| o of the rotational angular momentum onto the is taken to bisect the BAB bond angle of the molecule. q and p are the Here, Jr, Jy and Jz are smolecule-fixed x,y, and 2 axes. while the x-axis is perpendicular smass-weighted bending coordinate and momentum, respectively. A(q),B(q) and C are twice the inverse moments of inertia, and A(0) > (0) > C. As the bending motion preserves C2, symmetry, the inertia tensor is always diagonal. The simplest posible dependence (line) ofthe moments of inertia onthe bending coordinate ‘was assumed in our calculations, and the bending potential V(q) was taken to be cof the form Vig) = Vania! + tat ® ‘Both harmonic (5 = 0) and anharmonic potentials were studied. A set of parameters appropriate for the molecule HzO is given in Ref. 44. The magnitude of the parameter a determines the strength of the interaction between rotation and the Jan PB) sin(x) (oa Syn AP Bost) (9) Jenk, @) ‘where (K, 0 are the canonical variables for rotation, and jis the magnitude of the ‘otal angular momentum. With the choice of2-axis asthe axis of quantization, kis imately @ conserved quantity for k= When the angular momentum vector ligned along the space fixed z-axis, the conjugate angle x, becomes equal to Euler angle y (minus). Su ‘of Eqs. 9 into the Hamiltonian (Eq,4 GREGORY 5. EZRA ‘yields the final form of the rigid bender Hamiltonian in terms of the four variables oap): Hyp Alg\k + (F ~ BB (q)c0sy + Cain’) + Yop? + Veg). (10) (in general, Eqs. 9 are substituted into the classical RV Hamiltonian fortheN-mode ,- problem of interest to give a system with N+ 1 degrees of freedom.) A rotational ‘0s is defined by pioting (b,x) for g = 0, p> 0, for trajectories of the fully coupled ied by Fredet their study of RV interaction in triatomics.*” Long-time correlations are found in the apparently chaotic RV ‘tajectories, and are presumably associated with the bottleneck defined by the ‘broken separatrix dividing A-type (high-) from C-type (low-E) regions of phase space. ‘Quantum rotational sos forthe rigid bender problem have been calculated by ‘Martens using rotational coherent states, and a remarkable correspondence found between the classical and quantum phase space structure. B. Rigid Twister Model (Our interest in rotational eects in overtone induced dissociation of hydrogen ‘motions are identified, where extensive rotaton-torsion energy flow is associated with the latte. Of particular interest isthe observation that changes inthe cis and vibcation-rotation coupling in HOOK. ‘We note also that Sumpter and Thompson®™ have recently studied the effects ‘of molecular rotation on the intramolecular vibrational dynamics of hydrogen peroxide. Addition of rotational energy equivalent to 1500K was found o enhance Trajectory Studies of Intamolecular Dynamics 2 k wo 70 90 0085 SSC* S20 xr gure 7. Rotational kx surface of section forthe rigid bender model: j= 35, € 2015 au, a= 100, the rate of energy flow out ofthe initially excited OH overtone and increase the tate of dissociation approximately twofold. C. Full Dimensional H20 ‘Our work onthe rigid bender model has been extended to the full rotation vibra- ee below) fottion induced cots couplings sof paral interest. ‘The classical RV dynamics of HzO has been studied including all degrees of face expressed in terms of geometrically defined internal ton to molecule-fixed Eckart coordinates‘ is made to% GREGORY 5. £784 Trajectory Stukes of intramolecular Dynamics 2 (Our studies ofthe full RV dynamics in 1:0 reveal interesting ptters rotation uegtot ofthe son energy from one bond tothe of a SLAWNMMMA fess Gain etsy ena aa “Ye o ro te te 10, the resonance melee) low in HO. Sequential passage from the local to normal back tothe local mode regime can be seen. ton-vibration interactions in molecules exhibiting local mode HO. tractable using Fourier methods.2% GREGORY 5. EZRA 5 ge. fences eeaeeaeesae NV 5 ooo 1s500 Wop alto boo ewe Energy 3 » ae fw jooo 19000 1eoo aod atsna _eatao Energy Figure, The ciference AE between primitive semilasia reat values and quantum variaionat eigenvalues versus quantum ener Y sate of 20. (a Hoy Bunker potential. Lai-Hagsrom potential. Al energies are IV. THE STRUCTURE OF MULTIDIMENSIONAL PHASE SPACE ‘As mentioned in the Introduction, following the fundamental work of MacKay et al. Bensimon and Kadanoff * and 'otranspotin arewprseving maps, much een research on asia models or ‘unimolecular and bimolecular processes in two degree of freedom (dof) systems has focussed on the importance of dynamical bottlenecks in nonsatstical be- havior-3°** Two types of partial bericrare important in 2-dof systems: cantor. Which are remnants of invariant tori that have broken up under the influence of 5. and separatrices™ which form the boundaries of and separatrices has been incor- porated into modified s ‘Waals complexes?™2 _Phase space structures associated with tapping and long (Cintinsicnon-RRKM” behavior") in 2-dof may usefully be cha Trajectory Studies of tramolecular Dynamics 2 of relations between frequencies of motion. The least permeable cantor, which {form minimum flux intramolecular bartiesso IVR, ae empirically found tobe the ‘remnans of invariant tori withthe mos rational golden") frequency ratios. I is intuitively reasonable thatthe most robust tori (the fst to break up under perturbation) are those farthest out of resonance. Trapping of trajectories nzide a Tesonance zone is associated witha locking ofthe two fundamental frequencies into a particular rational ratio.” “The nature of bottlenecks responsible for long-time trapping in N'= 3 mode systems and the quantitative description of phase space transport in multimode systems remain oustanding problems. The structure of phase space for N= 3 dot is qualitatively different from the two-mode case, due tothe fact that invatiant [N-tori do not partition the 2N-1 dimensional energy surface nto disjoint regions for N > 2 Relaxation pathways and the origins of lang time correlations a tindeftood inthe multidimensional ease. The current level of understanding of 2- dof has been obtained using techniques, such a the Poincare surfee of section? that are not realy extended to higher dimensional systems. New methods and ‘concepts are therefore required. ‘In two-dimensiona area-preserving maps (which may be Poincre return maps for 2-dof Hamiltonian systems), partial barriers ae formed by taking the union of segments ofthe stale andunstable manifolds of periodic oti ‘tion of resonance boundaries given by MacKay et al. Part constructed by joining up the gaps in eantor (or petiodic thereto). A derumerablese of par into disjoint regions, and a description: phase volume across the region boundai eal in this approach to transpor isthe tumstil™ or lobe, which rt from onesie of the partial barrier othe othe. the flux per iteration crossing the given partial orbits 'A promising generalization of the turnstile based approach to transport in sasional systems has recently been given by Wiggins." In this approach, ers for N-dof systems are constructed from segments ofthe stable and invariant manifolds through one iteration ofthe ‘We ate currently applying this theory to phase space transport in 4d symplectic maps, which are discrete analogues of three-mode continuous time Hamiltonian20 GREGORY 5. EZRA “3200-16 Irancey/ncoy) 43 00 60 120 180 240 300 360 420 400 S40 500 Time (e105) Unimolecular decay of ensemble under a 4-d symplectic mapping that vibrational n of Heels" The solic systems. Both coupe standard maps and map mosetng he pedsocnon der Waals complenes Xz...Y are being investigated’ 100 iterations of the map (about 15 ps; one timestep is 6000 au). We are currently Trajectory Studies of Inramolecular Dynamics 3 ‘investigating the phase-space features that determine ‘map. One very important point to emerge from inthe case of “nice” intersections may turnstiles be defined a all,°° and this fact may well pose a severe obstacle to extension of tumnstile based theories of transport roach to the study of multidimen- analysis” Although more empiri is method has proved to be a useful ransfr in tee-mode systems, sional dynamics, the method of ‘eal than the rigorous approach, diagnostic approach for underst V. LOCAL FREQUENCY ANALYSIS AND. MULTIDIMENSIONAL PHASE SPACE DYNAMICS anspor or trapping) may be defined by between frequencies of motion. LFA therefore focuses attention direct important aspect of the dynamics. (The significance of pairwise frequent in multimode system: although the motion at 20,000 Jong time scales (~ 50 ps), quant time averages of zeroth order aci ‘constant over many vibrational periods (- 1 ps). Trapping of trajectories in2 GREGORY 5. EZRA 45 é e é (e099 32 3 ae ean OTe 4 a H 3 Cred Te Zao) ro PN gy ST a9 es! Opens Figure 11. Frequency ratio space (or planar OCS." Resonance lines (lid) and Pairwise noble conditions (dashed) are indicated. short (~ 1 ps) spectral estimators are jectories!*” or shorttime studied by following the changes ofthe local frequencies with time, Trajectory Studies of nvamolecular Dynamics T = 26.907 od amplitude (arb.) 4 ps segment 2 conance zone. The indicated peaks in the power spectra are frequency ratio atains a given value. By simpl ‘given condition on the local frequencies is sats34 GREGORY S. EZRA Trajectory Studies of intramolecular Dynamics 35 Action Space — ‘crossing of irrational surface along transverse 1 Energy Surface the local frequencies, without having o construct such surfaces. It Figure 14. Schematic view of phase space structure for three degree of freedom possible to examine candidates for dynamical botlenecks in systems, such as planar OCS.** slong resonance channels is observed in planar OCS.” We have also found ing of trajectories inthe vicinity ofsesonance junctions.” thai, in sional energy sional phase stam ‘The dense network of resonance lines in the frequency-ratio plane forms the so-called Arnold web2!42350 At the intersection of dimension one ess known in the small perturbation limit as Arnold diffusion.?445° Rapid transport36 GREGORY $. EZRA Itis not yet known which conditions F(a) are associated with dynamically ‘mechanisms for transport a resulting view of the phase space structure in 3-dof is shown schem: discussion of the dynamics of highly excited acetylene.™* VI. FUTURE DIRECTIONS fundamental questions semain to be addressed in future work. bbe preferable (see the contribution by Skodje and Davis to this Volume). ACKNOWLEDGMENTS owotkers and colleagues for enjoyable and fruitful collabora- Trajectory Studies of inzamolecular Dynamics 37 REFERENCES. Bunker, Dy Hass, WLI. Chom Pls. 1973, 9, 462138 GREGORY 5. EZRA . Gre Wis Lovenes Soe CB Chen Pow th fe 00 | “| = | \ Trajectory Studies of lnramolecular Dynamics 33 67. Cds, WB; Shins, RB.J. Chem. Phys. 1984, 81,6224 st, W.P- Chem, Phy Let 18S, 17, 600; Chem, Phy. 1986, Thompson, DLJ. Chem. Py. 1985, 2, 4557; 1987, 6, 805; Chem. 2. 107. Gy, SK; Rice, S.A Farad Dist Chem, See 185, 8,207.GREGORY S. EZRA ic THE ROLE OF MODE-MODE ENERGY TRANSFER IN UNIMOLECULAR REACTIONS 144, Dave, MJ. Chom Pt Let 1984, 10,491; Davis, MI. Wagner, A.ACS Symp. 1984, 263 1S. Brigham, £0, The Fax Fourier Trnsor: Prete Hall: Balewood Cis, NI, 1974 146. Maple, SL. Digital Spectra Analyse with Applcoins, Prete Hal: Englewood Cis, NI 1987 Roy, Stamper, LG. Pelt, G-A; Gray, S Roi D.W. Pls Rep. 1991205, 109. 154, Holme, 1A; Levine, RD. Chem Pl. 1989, 131,109 John S. Hutchinson