Real-Time Laser Femtochemistry: Viewing the Transition From

SPECIAL REPORT

Real-Tiime Laser Femtochemfstrv Viewing the transition from reagents to products

Ahmed H. Zewall, California Institute of Technology, and products. Such a description is the goal of molecular Rlchard B. Bernsteln, University of California, Los Angeles reaction dvnamics. Some 3dyears old now, molecular reaction dynamics One of the most fundamental problems in chemistry is has matured into a major field of chemistry. It has had understanding how chemical reactions occur; that is, impact on photochemistry and laser chemistry; laser how reagents make their journey to products. Tradi- mass spectrometry and ultrasensitive detection; iso- tionally, chemists start by studying the thermodynam- tope separation; energy transfer and relaxation in gases ics of a reaction, then its rate, and finally postulate its and solutions; disequilibrium and transport phenome- mechanism. na; atmospheric, combustion, and plasma chemistry; The century old and well-known Arrhenius rate dynamics of gas-solid interactions and heterogeneous equation catalysis; cluster formation and cluster chemistry; and k(T) = A~-~"~~ even dynamic photobiology. The Pimentel Report, "Opportunities in Chemistry," systemizes a large body of experimental data, express- stresses chemical reaction dynamics as one of the most ing a reaction's rate in terms of the activation energy, important current research areas in chemistry. As fur- E,, and the pre-exponential A-factor. For gas-phase bi- ther evidence of the significance of the field, the 1986 molecular reactions, the equation conveys the essence Nobel Prize in Chemistry was awarded to Dudley R. of the reactive encounter: Molecules collide (expressed Herschbach, Yuan T. Lee, and John C. Polanyi for their by the A-factor, which is related to the rate constant for incisive experimental and interpretive work on the collisions), but they react only if collisions are suffi- dynamics of elementary chemical reactions. ciently energetic (the exponential term). Major advances in the experimental study of molecu- The rate constant, k(T), however, does not provide a lar reaction dynamics have come from the application detailed molecular picture of the reaction. This is be- of molecular beam and laser techniaues. In the sim- cause k(T) is an average of the microscopic, reagent- plest molecular beam experiments, a beam of reagent state to product-state rate coefficients over all possible molecules, say A, is directed toward coreagent mole- encounters. These might include different relative ve- cules B (in the form of a target gas or another beam) and locities, mutual orientations, vibrational and rotational the reactive scattering thatproduces product molecules phases, and impact parameters. We needed a way to C and D is observed. The relative kinetic energy of the describe, with some quantitative measure, the process reagents can be changed in these "single collision" itself of chemical reaction: how reagent molecules ap- experiments by varying the velocity of A with respect proach, collide, exchange energy, sometimes break to B. bonds and make new ones, and finally separate into For laser-molecular beam experiments, a laser excites

24 November 7, 1988 C&EN of reaction cross-sections and their angular depen- Photofragmentation of a triatomic dence as a function of collision energy are compared molecule occurs in femtoseconds with theoretical (usually computational) predictions, based on ab initio or semiempirical calculations of potential energy. At the very least, such studies reveal the strength and range of intermolecular forces. Much more has been learned from detailed collision experiments in which the vibrational and rotational states of both the reactant and product molecules are known. Such "state-to-state" experiments provide information on the role of precollision states in determining the states of the product molecules. The problem is that a major feature of reaction dynamics is still unseen experimentally, namely the transition state between reagents:and products. This state often lasts only picoseconds or less, and experiments on a longer time scale provide data that are effectively time-integrated over the entire course of the collision. As emphasized recently by Ian W. M. Smith of the University of Birmingham, U.K., a great deal is known about the "before" and "after" stages of the reaction but it is difficult to observe the "during" phase. Now chemical dynamics is becoming more ambi- tious: It is trying to describekransition-state chemistry. Until recently, this has been approachable only by theoretical studies. The goal now is to make direct, real- Photofragmentation of a triatomic molecule such as ICN is time observations of the process of reaction itself, even shown schematically here. The lower surface represents a though the duration of the event may be a matter of strip of the ground state potential energy surface as a only femtoseconds (1 fs is 10-l5 second). function only of R, the separation between the atom and the Femtochemistry, or chemistry on the femtosecond diatomic fragment. The absorption of a photon (energy hv) time scale, can be defined as the field of chemical dy- excites the triatomic molecule from its ground state to its namics concerned with the very act of chemical trans- upper, repulsive potential energy surface at the start of the formation, the process of breaking one chemical bond reaction. After a few hundred femtoseconds, the photofrag- and making another. On this time scale the molecular ments are essentially free of one another. Experiments can dynamics are "frozen out," and the complete evolution measure the real-time progress of the reaction as the of the chemical event is observed. This time scale is triatomic reactant makes the transition to its diatomic and perhaps the ultimate one as far as chemistry is con- atomic products. Since the recoil velocity of fragments is cerned, but it is a mistake to infer that new studies will tyvicaily 1 km per second, in 100 fs the distance spanned is end at this time resolution. 1 A. [Adapted from M. Rosker, M. Dantus, and A. H. Zewaii, Femtochemistry requires ultrafast laser techniques Science, 241, 1200 (1988).] to initiate and record snapshots of chemical reactions with femtosecond time resolution. It is also advanta- geous to work with molecules in supersonic beams or expanded jets; the expansion extensively cools the re- one of the reagent molecules and thus influences the actant molecules, which simplifies their internal ener- reaction probability or initiates a unimolecular process gy distribution and makes it easier to do state-to-state by energy deposition in a molecule, say ABC. In this so- experiments. However, since collisions play no role on called "half-collision" process, the fragmentation of the femtosecond time scale, experiments also can be the excited ABC molecule into AB and C is the dynami- carried out in bulk gases. cal process studied. In such experiments, the observa- The objective of this article is to highlight recent tions are of the angular, velocity, and quantum state developments in femtochemistry and to make the con- population distributions of the products. nection between these real-time ultrafast techniques The quantum state distribution of product molecules and earlier methods devoted to studies of reaction dy- can be obtained using infrared or visible chemilumi- namics under collisionless conditions. Reaction d)- nescence, laser-induced fluorescence, or resonance-en- namics in collisional environments, including con- hanced multiphoton ionization. Product angular and densed phases, is beyond our scope. velocity distributions are best measured by rotatable Real-time femtochemistry of elementary reactions mass spectrometer detectors with time-of-flight analy- allows reaction dynamics to be viewed in a new light. sis of products. The task is to relate these observables to Because the experiments can utilize laser pulses as the potential energy governing the interaction be- short as 6 fs (as of the writing of this article!), the actual tween molecular or atomic fragments. Measurements progress of the passage through the transition states of

November 7, 1988 C&EN 21 Special Report a reaction can be viewed directly. For photoinitiated In this and all subsequent classical trajectory simula- unimolecular and bimolecular reactions, one can ob- tions of elementary chemical reactions, it has been nec- tain snapshots from which the potential relevant to essary to use numerical integrations of the classical fragment separation and product formation cqn be de- equations of motion with subfemtosecond step sizes in duced with a distance resolution of about 0.1 A. order to accommodate the high-frequency molecular We are now seeing a revival of flash photolysis, be- vibrations occurring in the reactants. The hydrogen gun in 1949 by Ronald G. W. Norrish and George Por- atoms in Hz move with a vibrational period of about 7.6 ter at Cambridge University, in the U.K., but the in- fs; thus, if a single oscillation cycle is to be approximat- creased nine-orders-of-magnitude improvement in ed numerically with more than 10 steps, each step time resolution in femtochemistry allows one to ob- needs to be smaller than 1 fs. In a typical thermal serve the transition states and the fundamental dynam- energy atom-molecule collision, a 1 fs time interval ics of chemical bond rupture and bond formation, with corresponds to an atomic displacement of A, a step subangstrom resolution. size corresponding to less than 1%of a bond length. Most of our ideas about the transition state thus far Ferntochemistry and transition states come from such calculations, rather than from experi- The critical stage of any reaction is the saddle point ments. These calculations form the basis for femtose- separating reagents from products-the crucial transi- cond transition state chemistry as now envisaged. tion state configuration. We will use the term "transi- What is different is that now experimental methods tion states" to encompass all neighboring nuclear con- permit observing the transition state and clocking, in figurations important to the transition from reagents to real time, the formation and decay of the activated products, including the configuration that defines the complex. The goal is to observe and monitor the dy- transition state in theories of chemical reactions. namics of bond rupture and bond formation and to Theoretical femtochemistry has been with us since measure and characterize the transient intermediate en the 1930s, following the introduction of the concept of route to separation into products. the transition state by Michael Polanyi in the U.K. and The "entr6e" into the transient comes via photons. Henry Eyring in the U.S. In a classic work of 1936, The tool for this research is femtosecond transition- Joseph 0. Hirschfelder and Eyring, then at Princeton state spectroscopy of reactions, introduced by the Cali- University, calculated point by point the first classical fornia Institute of Technology group of Ahmed H. mechanical trajectory of a free hydrogen atom ap- Zewail. The combination of ultrashort (picosecond/ proaching the H2 molecule to form an "activated com- femtosecond) pulsed laser interferometric and molecu- plex" or transition state. lar beam techniques, during the past eight years, has made this possible. Pulses as short as 6 fs can now be generated, following the pioneering design of dye la- Classical mechanical calculations trace sers by the AT&T Bell Laboratories group of Charles V. dynamics of a chemical reaction Shank, and used in these experiments.

Traditionallv.- - .-..-.- - ~~ ,, svectroscovv-I --- I, has been the method of choice for determining the structure of isolated molecules and their intramolecular dynamics. With the ad- vent of intense, ultrashort laser pulses, and the ability to establish accurate time delays in the range 10 to 1000 fs by micrometer mirror displacements, with auto- and cross-correlation analysis of pulse shapes, it is now possible to do real-time transient spectroscopy of reactions on the femtosecond time scale and thus real-time femtochemistry. Before discussing femtochemistry in detail, we will summarize the key information obtained from time-integrated experimental studies of Rbc - unimolecular and bimolecular ieactions, which will help illustrate the important features of reaction dy- The first classical mechanical trajectory calculation for a namics and the new areas that can be impacted by chemical reaction, indicated by the curved line with arrows, direct real-time (femtosecond) observations. is plotted on the potential energy surface contour diagram in the skewed coordinate system. The reaction is the collin- Birnolecular reactions ear collision of HZ,with H: HaHb+ Hc-HaHbHc-Ha+ HbHc. Since the early reactive beam scattering experiments Initially the distance between Hb and Hc (Rbc)is very large of the 1960s and 1970s, it has become well established and Rabis the hydrogen molecule bond length. Note that the that, for a large class of fast elementary reactions in- trajectory becomes trapped in "Lake Eyring." Although the volving small molecules, the angular distribution of potential energy surface used for this calculation later products with respect to the molecular collision frame- turned out to be flawed, the computation is of historical work-the so-called center-of-mass system-is strong- significance. [Adapted from J. 0.Hirschfelder, H. Eyring, ly anisotropic. This asymmetry comes about because and B. Topley, J. Chem. Phys., 4, 170 (1936).] the transition state does not last long enough for the activated complex to rotate much, a process that takes,

28 November 7, 1988 C&EN Direct-mode reactions take femtoseconds, complex-mode reactions picoseconds The lifetimes of transition states in bimolecular collisions Ha + HbH,-H,Hb + H, Interatomic distance, A can vary from several femtoseconds to many picoseconds. The upper graph shows a single trajectory calculated for the direct-mode exchange reaction Ha + HbHc-tHaHb + Hc9 assuming a collinear configuration. Plotted on the ordinate scale are the interatomic separations Rab, Rbc, and Rca as a functlon of time from an arbitrary starting condition. Note the oscillations in Rbc before the encounter, then the exchange taking place, forming the new molecule HaHb, 0 10 20 30 40 50 whose oscillation is seen in Rab. There is a time interval of Time, femtoseconds about 10 ferntoseconds during which the system is in a KCI + ~a~r-KB~+ NaCl transition state-a collinear, triatomic HaHbHc. The lower Interatomic distance, A graph is a similar portrayal of a single trajectory calculated for the complex-mode, gas-phase metathesis reaction KC1 10 + NaBr-ttetratomic complex (structure shown)-+KBr + NaCI. The "snarled trajectories" are evidence for a long- lived (here about 5 picosecond) collision complex that 5 eventually falls apart to yield products. [Upper graph adapted from M. Karplus, R. N. Porter, and R. D. Sharma, J. Chem. Phys., 43,3258 (1965),lower from P. Brumer and M. 0 1 I I o 2.5 5.0 7.: Karplus, Far. Disc. Chem. Soc., 55, 80 (1973) and P. Tlme, picoseconds Brumer, Ph.D. thesis (1972).] i

typically, a picosecond. Thus, for many such reactions, To obtain reaction cross sections, one must comput- called direct-mode, we have experimental evidence of er-simulate many trajectories and use Monte Carlo transition states lasting less than a picosecond, based methods of averaging over impact parameter, initial on this crude and average rotational period clock. orientation, and vibrational and rotational phases, as On the other hand, for a whole class of complex- pioneered by Martin Karplus, then at Columbia Uni- mode bimolecular reactions, discovered in 1967 by versity; the late Donald Bunker, University of Califor- Herschbach and coworkers of Harvard University, the nia, Irvine; John C. Polanyi at the University of Toron- product angular distributions are symmetrical [n the to; and others. However, only a modest number of center-of-mass system, implying that the complex well-chosen trajectories are needed to visualize the key sticks together for many rotational periods, typically dynamic features of a reaction, especially the forma- several picoseconds, before falling apart to yield prod- tion and decay of the activated complex. "Motion pic- ucts. In some cases, they can be stabilized further by tures" of such trajectories have been made by combin- subsequent, deactivating collisions, as in unimolecular ing many successive snapshots of atomic locations cal- reactions. culated as a function of time. To characterize faithfully Molecular beam scattering experiments also provide the fast vibrations of molecules requires that the snap- data on angular distributions for elastic scattering, as shots be made at very close time intervals. The time well as velocity and angular distributions (and their scale of interest for direct-mode reactions ranges from energy dependence) for inelastic scattering. From this about 50 fs to several picoseconds. It is much longer, of information, many features of the anisotropic intermo- course, for complex-mode reactions. lecular, ground-state, potential surface can be deduced. This great progress in obtaining and simulating The accuracy of the derived potential is ascertained by product angular distributions and energy-dependent back-calculation of the experimental scattering cross reaction cross sections enables one to advance some sections using classical, semiclassical, or (preferably) concepts regarding energy disposal and consumption, quantal scattering computations. Classical mechanics, but these time-integrated observables still do not offer thouah it has some well-known limitations. is verv a direct view of transition states themselves. good"for visualizing the course of the reaction. F& elastic and inelastic scattering, quantal scattering computation has supplanted classical trajectory methods, Photofragmentation reactions but for reactive collisions (even simple atom-molecule In a full collision, two molecules approach one an- exchange reactions), quantal calculations are still diffi- other, may or may not exchange energy, and may or cult and costly, so quasiclassical methods are widely may not break apart as new product molecules. The used. In recent years, however, great progress has been process can be divided into two parts, the incoming made in implementation of accurate quantum scatter- and the outgoing half-collisions. Photofragmentation ing computations on elementary reactions, such as is a classic half-collision process in which a stable mole- H + HZ. cule is photoexcited to an unstable electronic state in a

November 7, 1988 C&EN 27 Special Report

Crossed-beam experiments reveal reaction dynamics The state-of-the-art of both theoretical and experimental transition state that lasts for a very short time. Very recently, chemical reaction dynamics can be seen in recent work Valentini has observed sharp structure in the energy depen- based on crossed-beam methods on two elementary reac- dence of the reaction cross section for specific vibrational tions. The best-known example is the hydrogen exchange states of the products. Such dynamical resonances have reaction, H + H2 -+ H2 + H (right). The potential energy been confirmed theoretically by Truhlar and Donald J. Kouri surface for this reaction and its various isotopic counterparts of the University of Houston. was computed, point by point, by Per E. M. Siegbahn of the University of Stockholm, Sweden, and Bowen Liu of the IBM research laboratory in San Jose, Calif., using ab initio quantum chemical procedures believed to give energies with an accuracy of 0.1 kcal per mole. These data were later fitted to analytical functions by Donald G. Truhlar and Charles J. Horowitz at the University of Minnesota to make them easier to use. This "best-known" potential energy surface is desig- nated LSTH, and extensive quanta1 computational advances using it have been made by the groups of Aron Kuppermann at California Institute of Technology, Robert E. Wyatt at the University of Texas, Austin, and John C. Light at the Universi- ty of Chicago. The lower panel shows calculations for the expected angular distribution of the product molecule HD from the H + D2 reaction, using quasiclassical trajectory calculations on the LSTH surface. Such calculations by Truhlar and Normand C. Blais of Los Alamos National Laboratory, and by Howard R. Mayne at the University of New Hampshire predict very well the experimental detailed differential reactive scattering cross sections for this same reaction obtained experimentally by J. Peter Toennies and coworkers at Max Planck Institute for Fluid Dynamics, Gottingen, West Germany, shown in the upper panel. In an important 1986 paper, they showed that the HD angular distributions are strongly anisotropic (in the scattering angle O), implying a short-lived transition state. Quasiclas- sical trajectory calculations of vibrational and rotational state distributions of the HD product are also in reasonable accord Experimental reactive-scattering contour diagram in the center with laser experiments by the groups of Richard N. Zare at Of mass represenfaflon for the D -I-H2 crossed-beam reaction. The HD product flux-veloclt~1s seen to be elected preferentially Stanford University and James J. Valentini at the University of sideways wlth respect to the relative velocity (of about 3 km per California, Irvine. second). [e is the product-scattering angle in the center of mass The reactive cross sections shown here are for a collision system, is the in the of mass system, and is energy Of .5 eV. These experimental resultsare Only the velocity In the laboratory system.] The results are well repre mimicked by the quasiclassical trajectory calculations, but senfed by theory (lower panel). [Adapted from R. Gbtting, H. R. by a simple classical impulsive model consistent with a Mayne, and J. P. Toennies, J. Chem. P~YS.,85, 6396 ( ig86)] time that is short compared with any significant nucle- of product recoil, thereby providing knowledge of the ar motion. Following the excitation, the two fragments process of fragment separation. The new and impor- separate and accelerate to a terminal relative velocity tant information on the dynamics of a variety of photo- governed by the available energy for translation. dissociation reactions obtained from these experiments Since the classic work on photodissociation in 1934 permits deduction of the time-integrated population by the group of Alexander N. Terenin at the University distribution and angular distribution of the photofrag- of Leningrad, there have been many important techno- ment product states. logical advances. State-to-state dynamics of photodis- Such angular distribution experiments also provide sociation reactions have now been studied with the a method for inferring the dissociation lifetime of the help of lasers and techniques such as photofragment laser-excited molecule ABC. Basically, if the molecule time-of-flight spectroscopy. The pioneering theoreti- dissociates before it has time to rotate significantly, cal and experimental work of Richard N. Zare of Stan- then the original alignment of the excited molecule ford University, Richard Bersohn at Columbia Univer- induced by the laser is preserved and the fragments sity, and Kent R. Wilson at the University of California, will be ejected in certain well-defined directions. On San Diego, made it possible to study the correlation the other hand, if the dissociation is very much slower between laser polarization and the direction and speed than the average rotational period, the photoproducts

28 November 7. 1988 C&EN Collision energy = 1.82 kcal per mole Experimental product flux-velocity contour map in the center of mass system for DF from the F + D2 reaction at two different colllslon energles. The clrcles denote veloclty limits for the speclfled DF vlbratlonal states. For the lower collision energy case, DF Is predominantly backward scattered, that is, antipar- allel to the lncldent F veloclty, for all observed vibratlonal states. For the hlgher colllslon energy case, the DF Is strongly forward scattered (attributed to resonance) for the v = 4 state. [Adapted from D. M. Neumark, A. M. Wodtke, G. N. Robinson, C. C. Hay- den, K. Shobatake, R. K. Sparks, T. P. Schafer, and Y. T. Lee, J. Chem. Phys., 82, 3067 ( 1985)]

upper panel shows a typical flux-velocity contour map plotted in the center-of-mass system, indicating backward scattering of the DF product in vibrational states 1, 2, 3, and 4. But at a different collision energy (lower panel) there is a drastic change in the angular distribution, which now shows a forward peak for the v = 4 state. This is a clue that quasibound states are formed in the encounter, which decay preferential- Collision energy = 3.32 kcal per mole ly into product molecules in this vibrational energy state. (Analogous experiments with F + H2 show such an anomaly at another energy, with a forward HF peak for v = 3.) Various semiempirical and ab initio potential energy surfaces have been put forward by James T. Muckerman of Brookhaven National Laboratory, John C. Polanyi at the Uni- versity of Toronto, and by Henry F. Schaefer Ill at UC Berke- ley, which can account for many of the gross dynamical features of the reaction and even predict certain resonances in the scattering cross sections. However, as yet no calculation does full justice to the experimental, vibrationally specific product angular distributions measured over a range of collision energies for the three isotopic reactions of F with H2, HD, and D2. Very recently, William H. Miller and coworkers at UC Berkeley have obtained accurate quanta1 scattering computational results, which perhaps can be extended to deal with the entire body of experimental observations. Be- The second important bimolecular reaction that has been cause of the mainly direct characteristicsof the reaction, it is studied is F + H2-) HF + H and its isotopic variants. A much clearly one in which the lifetime of the transition state is in larger experimental chemical dynamical database is avail- the femtosecond rather than the picosecond regime, except able for this reaction, the most detailed of which comes from for the long-lived resonances. As Lee points out, one of the the crossed molecular beam studies of Yuan T. Lee and co- as yet incomplete tasks of the molecular beam method is the workers at the University of California, Berkeley (above). The "direct experimental probing of potential energy surfaces." will have an isotropic angular distribution. The disso- have been obtained by the Caltech group for several ciation lifetime can be estimated from the degree of elementary, laser-initiated, unimolecular reactions, anisotropy of the observed photofragment angular dis- and these temporal results have been compared with tribution. The technique is similar to the use of the those from spatial (orientation) experiments. deviation from symmetry in product angular distribu- In certain polyatomic molecules, such as the methyl tions in crossed-beam experiments to estimate the life- halides, a quasidiatomic approximation can account for time of the transition state. Many recent advances have their continuous absorption spectra. However, to inter- been made in this area, as illustrated by experiments pret the photofragment distribution data, it is neces- involving polarization analysis of photofragment fluo- sary to take into account the internal excitation of the rescence from the groups of Paul L. Houston at Cornell fragments. Various simple classical models for half- University, Stephen R. Leone at the University of Colo- collisions have been useful in accounting for experi- rado, John P. Simons at the University of Nottingham, mental data on fragment energy distributions, but clas- U.K., and Richard N. Dixon at the University of Bristol, sical trajectory simulations provide more insight, just U.K. as they do for full collisions. The most popular ap- In the case of half-collisions, subpicosecond and proach for the description of dissociation trajectories femtosecond measurements of the dissociation lifetime has been that of simple impulsive forces between frag-

November 7, 1988 C&EN 29 Special Report

Photofragment spectroscopy experiments reveal reaction dynamics

The photofragment spectroscopy of tally and theoretically for many mole- tion of both the carbon monoxide and three- and four-atom molecules pre- cules. hydrogen products. By energy conser- sented here shows how dynamical in- Experiments have revealed the en- vation, the remaining rather large frac- formation is extracted and is typical of tire rotational and vibrational distribution of the available energy is known to many studies, made in the U.K., West energy Germany, Switzerland, and the U.S. The group of C. Bradley Moore at the -H University of California, Berkeley, has provided a large body of experimental data on the photodissociation of formaldehyde that has helped to establish how the excited formaldehyde mole- - cule breaks apart to form molecular hydrogen and carbon monoxide (and also atomic hydrogen plus the formyl radical). The path for this photodissociation is shown at right. The initial, excited state of formaldehyde is well characterized by spectroscopic studies. In- ternal conversion causes coupling between the first excited singlet state of the molecule (S,) and the quasi- degenerate rovibrational levels of the ground state (So""). Consequently, the - initial state of the reaction cannot be considered to be a discrete state, but - rather a lumpy continuum of states. Molecules in these states reach the reaction coordinate by a process ,c= 0 The energy path along the potential energy surface called intramolecular vibrational-ener- describing the formaldehyde unimolecular decomposition gy redistribution, and such processes reaction: H2CO+H2 + CO. [Adapted from C. 13. Moore have been characterized experimen- and J. C. Welsshaar, Ann. Rev. Phys. Chem., 34, 525 ( 1983)]

ments. One imagines instantaneously breaking the tive mainly to the nature of the potential surface in the bond and then makes classical mechanical predictions region of product separation. On the other hand, the of what would follow, using conservation of energy rates of these reactions represent the flux of excited and momenta. For example, if a torque is produced, molecules traversing the transition state and therefore then one would expect rotational excitation of the frag- relate to the nature of the surface near the critical tran- ments. sition state configuration. Real-time picosecond mea- Even better, of course. would be exact auantal calcu- surements of state-to-state rates have now been made lations on ab ikitio potential surfaces. As ii also the case and can be used to test theories of unimolecular reac- for full collisions, these calculations are just a little tions and to help elucidate the nature of the potential beyond the present state of the art. An extra difficulty energy surface near the transition state. with photofragmentation reactions is the need to cai- Experimental state-to-state (microcanonical) rate culate not just the ground-state potential energy sur- constants for the reaction NCNO -+ NC + NO have face, but also that of the excited state and any interac- been measured at Caltech as a function of energy above tions with other, possibly nearby or crossing, surfaces. threshold for bond rupture. For this same reaction, Even in the simplest case, where there is only one Curt Wittig and Hanna Reisler of the University of excited state, there are problems in the implementation Southern California have made a thorough study of the of time-independent quantum mechanical procedures. NC and NO product state distributions. The microcan- However, experimental advances by many research onical rate constants show an unexpected nonmono- groups in the U.S., West Germany, France, Switzer- tonic structure in the energy dependence, and deviate land, and the U.K. have aided the application of theory, significantly from theoretical calculations based on the particularly to experiments on the photodissociation phase space theory that was used successfully to calcu- reactions of simple molecules, such as HOH, HOOH, late product state distributions. Similar studies of the

HONO, NCNO, ICN, CH3I, and CHJONO. unimolecular reaction HOOH -+ 2 OH also have been Product state distributions are expected to be sensi- carried out by F. Fleming Crim of the University of

30 November 7, 1988 C&EN go into the relative translation of ener- Bersohn at Columbia University, David terized with the help of theoretical gy of the product molecules, as verified A. Micha at the University of Florida, developments by Reinhard Schinke of by molecular beam translational spec- Keiji Morokuma at the lnstitute for Mo- Max Planck lnstitute and Richard N. troscopy experiments. The nature of lecular Science in Japan, and others Dixon at the University of Bristol in En- the transition states in this reaction has also contributed toward the under- gland. A great deal now is known about been deduced by Henry F. Schaefer Ill standing of photofragmentation. the way this molecule, which is bent in and coworkers, then at UC Berkeley, The energy path for the photodisso- its excited state, dissociates. from computations of relevant portions ciation of a second polyatomic mole- The upper excited potential energy of the potential energy surface. Dy- cub, water, shown below, is much surface for water is repulsive. The pho- namical calculations utilizing this sur- different. This photofragmentation has toinduced reaction, at 157 nm, can be face by William H. Miller and cowork- been thoroughly studied experimental- described as ers, also at Berkeley, reproduce the ly by Peter Andresen of Max Planck H,O H20* -+ H OH main features of the experimental lnstitute for Fluid Dynamics in Gotting- % + energy disposal patterns. These and en, West Germany and John P. Simons The OH photofragment molecules other studies show how state-to-state at the University of Nottingham, U.K. are vibrationally hot but rotationally population analysis can help establish The angular distribution, population, only "warm." A considerable fraction a microscopic mechanism for the and rotational alignment of the frag- of the available energy in the reaction photofragmentation. Theoretical work ments produced after excitation to the goes into translational recoil between by Karl F. Freed at the University of absorption continua have been charac- the H and OH fragments. When the Chicago, Moshe Shapiro at Weizmann water molecule bond breaks, there are lnstitute of Science in Israel, Richard two possibilities for the rotation of OH: either in the same plane as the oxygen p-orbital (n-state) or perpendicular to it (n' state). Experiments show that the dissociation leads to preferential population of n+. This OH alignment leads to the conclusion that the photofragmentation process is essentially coplanar, as depicted in the figure. This in- 4) verted population for the lambda dou- Plctorlal representation of the coplanar dissociation ee44 blet explains the origin of interstellar ##* of bent excited water. The unpaired electron of the water molecule orbltal (shown as vertical) correlates with the OH (also OH maser emission formed via ultravi- vertical) orbital. [Adapted from P. Andresen, G. S. Ondrey, 6. Tltze, olet irradiation of water molecules in H and E. W. Rothe, J. Chern. Phys., 80, 2548 ( 1984)] the galactic environment.

Wisconsin (product state distribution) and the Caltech ments nor the product state distribution data provide a group (real-time rates). direct view of the transition states. The phase space theory, the adiabatic channel model, and the RRKM theory for describing unimolecular re- Uni- versus bimolecular reactions actions can now be critically tested on the microcanoni- Unimolecular and bimolecular reactions have many cal, molecular level. A particularly good example is dynamic aspects in common, and the connection be- recent work on the photofragmentation of ketene by tween these two classes of reactions is important to the group of C. Bradley Moore at the University of femtochemistry. A few examples illustrate this. Hersh- California, Berkeley, who examined product state dis- bach showed how the product angular distribution in tribution near threshold energy, and the group at Cal- the reaction C12 + H -+ HC1 to C1 reveals the depen- tech who determined the state-to-state rates for the dence of the reaction probability on the relative orien- same reaction. They find that although phase space tation of colliding partners in the transition state. He theory agrees with observations near threshold, it must also showed the close analogy between the dynamics be modified to account fully for the dynamics at all of this type of direct-mode, atom-molecule reaction energies. This problem is currently being addressed and the corresponding photon-molecule dissociation theoretically by the groups of Rudolph A. Marcus of reaction: The HC1 product recoil velocity distribution Caltech, William H. Miller at UC Berkeley, Jiirgen Troe from the H + C12 reaction is much like that of the at the University of Gottingen, West Germany, and product chlorine atoms in photodissociation. Because others. Progress will be achieved when the topography the photofragment angular distribution mimics the of the potential energy surface involved is better orientation of the photo-excited HC1 molecule (or that known from ab initio quantum mechanical computa- of the transition state HClC1) before it rotates, the dis- tions and when the quantum dynamics can be solved. sociation is described as prompt; that is, it occurs before Although these studies have provided many details the molecule (or transition state) has time to rotate. of the dynamics, neither the state-to-state rate measure- This example indicates the connection between an

November 7, 1988 C&EN 31

Product angular distributions give clues to transition state lifetimes

H + CI,-HCI + CI hv + CI,-CI + CI collision, most of the product HCI molecules recoil to the left; the red cone encloses the region of maximum HCI intensity. Similarly, the CI product recoils to the right into the region enclosed by the blue cone. Superimposed is a contour map indicating the angular and recoil velocity distri- H butions for the HCI product. (Higher numbers indicate great- er concentration of product molecules). For the photodissociation of C12, the cones and contour maps show a very similar distribution for the two CI atom products. These patterns are quite different from that for the reac- Potential electronvolts tion of Hg + 12+Hgl + I, shown in the lower left figure. Here Hg + I, + [IHgl] +Hgl + I Hgl product recoils both forward and backward with respect to the center-of-mass (at the origin of the contour map). This forward-backward symmetry is characteristic of a long-lived collision complex that undergoes many rotations before it breaks apart. For the upper reactions, by contrast, the collision complex lasts such a short time that it does not have time to rotate before it breaks apart. The lower right figure is the experimentally derived potential energy curve - for the Hg 4- 12 reaction. Note the deep well corresponding lHgl to the binding energy of the lHgl collision complex, another Studies of product distribution in unimolecular and bimolec- indication that the complex will be long lived. [Upper fig- ular reactions reveal reaction dynamics. The upper two ures adapted from D. R. Herschbach, Far. Disc. Chem. figures compare the product flux-velocity distributions for a Soc., 55,241 (1973). Lower from M. M. Oprysko, F. J. Aoiz, bimolecular reaction, H + C12+HCI + CI (left) and the M. A. McMahan and R. B. Bernstein, J. Chem. Phys., 78, unimolecular photofragmentation reaction, hv + CI2-+CI + 3816 (1983) and T. M. Mayer, J. T. Muckerman, B. E. CI (right). For the bimolecular reaction, the H reactant Wilcomb, and R. B. Bernstein, J. Chem. Phys., 67, 3522 approaches from the left, the C12 from the right. After (1977).]

ular-formed KClNa collision complex. Independent John C. Polanyi and his collaborators have used the evidence from the study of cluster beams of mixed emission in the course of a unimolecular reaction alkali halides by the group of Ernst Schumacher at the Institute for Inorganic and Physical Chemistry in UV NaI + N~I#*+ Na* + I Berne, Switzerland, has shown that a relatively stable KClNa complex can be formed in a jet-cooled beam. and of a bimolecular reaction These are good candidates for future femtochemistry experiments to establish connections between unimo- F + Na, -+ [N~N~F]~'-. Na* + NaF lecular and bimolecular reactions on the same potential energy surface. to characterize their transition-state regions. Very recently, femtochemistry observations also have been Time-integrated spectra of the transition region made at Caltech on the first of these reactions. For the In recent years, there has been great progress in bimolecular reaction, no lasers need be used since the using time-integrated spectroscopic techniques to sodium product Na* emits light. The transition state, probe reactions during the reactive process. Absorp- species [NaNaF]+* has a lifetime of 1 picosecond or tion, emission, scattering, and ion spectroscopy meth- less, while the lifetime of Na* is about low8second. ods have all been utilized. The spectra of transient Therefore, the light emitted from the transition state is species intermediate between reagents and products about lop4 the amount emitted from the Na product. are expected, of course, to be different from those of the Spectral data show wing emission that extends over stable products or reagents. For one thing, transition several hundred angstroms from the sodium D line. energies and spectral linewidths for these intermediate Thus, the relative intensity of the two emissions at a species will be shifted and broadened because of per- given wavelength is less than the expected by still turbations caused by the proximity of product frag- another factor of 100 or so because of the spectral ments, as shown for many systems including those breadth of the wings. Polanyi's group has deduced studied by Alan Gallagher and coworkers, then at the some potential models that can account for these obser- University of Colorado, the group of William C. Stwal- vations. In a 1987 study, Polanyi examined the hot ley at the University of Iowa, and others. atom reaction of D -t Hg, and found evidence for a

November 7, 1988 C&EN 33 Special Report

used have pulse durations of several nanoseconds or Emission spectroscopy helps are in a continuous-wave mode. The collision complex, characterize transition state however, lasts on the order of picoseconds. Thus, light Log intensity, arbitrary units absorption during its lifetime will be extremely small. Such experiments have been carried out on the reaction - 1

The laser in this case was tuned to wavelengths where - 3 neither reagents nor products absorb, so thk signal is attributed to the transient complex. As Brooks pointed out recently, "The problem facing the experimentalist [when the yield of emission or absorption is very small] - 5 is to design a system where the effect of photon absorp-

Wave packet propagation describes the - 7 process of photofragmentation Potential energy

- 9 5

Wavelength, nm Shown here is the emission spectrum for Nal" in the process of falling apart. The off-scale peak at 589 nm is the D- line resonance fluorescence of product sodium. The broad spectrum near this peak, called wing emission, comes from several sources, including scattered light and the Lorent- zian linewidth of the D-line. The observed "extra" intensity on the blue side is attributed to fluorescence from the excited state intermediate, Nalt in the process of failing apart. The calculated wing shape is based on a theoretical model and an approximate potential curve for the excited state of Nal. [Adapted from H.-J. Foth, J. C. Polanyi, and H. H. Telle, J. Phys. Chem., 86, 5027 (1982).]

DH2# transient species. Such experiments are clearly relevant to the important Hg family of reactions. In 1984, James L. Kinsey and coworkers, then at Mas- sachusetts Institute of Technology, showed that measurements of near-resonant Raman scattering spectra from laser-excited methyl iodide or ozone molecules in the very process of dissociation can be used to characterize the dynamics along the reaction coordinate. The concept involved is best visualized with the help of Internuclear separation, RAB wave packet theory, developed by Eric J. Heller, then at UCLA. Basically, if a laser prepares a packet of excited Simplified potential energy curves describe the photodlsso- molecules on an upper potential energy surface, this ciatlon of a hypothetical molecule, ABC, into A 4- BC. The packet will in time spread out and move to larger inter- lower curve shows the ground state and the upper one the nuclear distances, leading to photodissociation. The excited, dissociative state. In the description provided by Raman-shifted spectral wavelengths will depend on wave packet theory, the laser photon transfers the ground the ground-state characteristics, but the intensities of state wave function to the exclted state "packet" at the these spectral transitions will reflect the evolution of tlme shown as to. During the process of falllng apart, an the wave packet on the upper surface. From these spec- exclted ABC molecule can fluoresce to various vibrational tra, Kinsey and coworkers deduced the nature of the states of the ground state. This emission spectrum, shown excited-state surface. in the lower left corner, consists of a series of broadened Brooks and Curl have invoked a different method to lines spaced according to the ground state vibrational lev- study the transient intermediate in bimolecular reac- els. [Adapted from D. Imre, J. L. Klnsey, A. Sinha, and J. tions. As with the two previous examples, the lasers Krenos, J. Phys. Chem., 88,3956 (1984).]

34 November 7, 1988 C&EN

Special Report

transit. The resulting spectra show the CN fragment in system built at Caltech is based on the pioneering de- the process of separation from the iodine atom in the sign provided in 1981 by Shank and colleagues at reaction AT&T Bell Laboratories.-They have shown that 90-fs pulse widths at 100-MHz repetition rate can be ob- ICN* -+ [I....cN]' * -+ I + CN tained from a colliding-pulse-mode locked ring dye laser. The same technique has recently been applied at With the help of pulse compression and group veloc- Caltech to reactions of alkali halides, where resonance ity compensation techniques, a number of groups have trapping of the departing atoms was observed generated shorter femtosecond pulses; the shortest is 6

-+ fs by the AT&T group. These pulses can be amplified, NaX* [N~--X]'* -+ Na + X and pulse broadening caused by group velocity disper- (X = I, Br) sion in the amplifier can be eliminated by a special It is interesting that this first experimental observation optical arrangement. Thus, femtosecond pulses are of trapping resonances in dissociation reactions, which generated at different wavelengths, opening up a helped establish the "femto age," dealt with systems number of possibilities for studies on this time scale. from the "alkali age" of crossed-molecular-beam stud- The characterization of the pulses is made by standard ies. auto- and cross-correlation techniques. For subpicose- The idea is as follows: The pump pulse excites a cond and picosecond experiments,-two synchronously target molecule, say ABC to a dissociative state. The pumped dye lasers (with pulse compression capability) probe pulse, delayed by a variable time, detects the are used in similar arrangements. photofragment product, say AB, as it is being formed, For experiments on rotationally and vibrationally in the process of separation from C. The probe laser is cold molecules, a standard molecular beam source is first tuned to a wavelength corresponding to a known used (employing a pulsed, seeded supersonic beam) excitation resonance of the stable AB species, either with electron impact and laser ionization, time-of- yielding laser-induced fluorescence or bringing about flight mass spectrometer, and laser-induced fluores- multiphoton ionization. In either case, the excitation cence capabilities. As shown by Zare, laser-induced allows detection of AB. The resulting photon or pho- fluorescence is a very sensitive method of detection, toion signal is recorded as one point on a curve of and the work of Donald H. Levy, Lennard Wharton, intensity (measuring concentration) as a function of and Richard E. Smalley at the University of Chicago time. The delay is altered systematically until an entire showed that cooling of molecules simplifies the spectra curve for this wavelength is obtained. There is an in- greatly. Laser ionization time-of-flight mass spectros- duction period corresponding to the time required for copy, as introduced by Bernstein, then at Columbia the AB species to separate effectively from the force University, and Edward Schlag at the Technical Uni- field of C and to attain, asymptotically, its normal iden- versity in Munich, helps identify the fragment of inter- tity as AB in its final rotational and vibrational energy est. Thus, mass and quantum state resolution is possi- states. ble. The laser beamsintersect the molecular beam (or The delay curve is repeated, detuning the probe laser the gas) in a very small interaction zone. The time at by small successive increments to reach the fragment absorption wavelength in-the transition region. At each wavelength, a buildup and decay of the transition state is recorded. A typical curve goes through the maximum at short delay times and then decays asymptotically to a constant signal level, dependent upon the detuning in- crement. This entire set of experiments can be repeated at different pump pulse wavelengths, thus changing the available energy for product recoil. The resulting data constitute a surface from which can be deduced the potential energy for the formation of AB from the excited state of ABC. Two technologies are involved in much of femtochemistry (and the earlier, sister picochemistry) studies under collisionless conditions: molecular beams and the generation and characterization of ultrashort laser pulses. The laser Part of Caltech's femtosecond apparatus, including its laser components, is mounted on

. 36 November 7, 1988 C&EN Femtosecond lasers combine with molecular beams for femtochemistry experiments Time-of-flight mass spectrometer

interferometer

Pulsed nozzle Caltech apparatus for femtochemistry and picochemistry experiments includes, at left, colliding-pulse-mode (CPM) locked ring laser, YAG-pumped laser amplifier system, pulse compression system, and an arrangement for pump and probe pulse generation. At right is the molecular beam system, including the time-of-flight mass spectrometer, interfaced to the laser system. The delay of the probe pulse is set by the Michelson interferometer arrangement. Shown also are the different nonlinear techniques-second harmonic generation (SHG), stimulated Raman scattering (SRS), and infrared mixing-used in general in these pump and probe femtochemistry experiments. The polarization of the pump and probe laser beams can be adjusted separately. For some experiments a flowing gas system is used instead of the molecular beam. [Adapted from M. Rosker, M. Dantus, and A. H. Zewail, J. Chem. Phys., 89, 61 13 (1988)l.

floating optical table that measures 20foot X 5foot

November 7, 1988 C&EN 37 Special Report which the initial laser pulse arrives at this zone marks ever, the potential energy surfaces are more complex the beginning of the experiment and establishes the and the data must be inverted to obtain the shape. We zero of time; the probe pulse follows. have recently developed such an inversion procedure. The old concept of Michelson interferometry is used Progress in femtochemistry does not require new to clock the experiments. Time is determined by con- developments in laser technology. For transform-lim- trolling the distance traveled by the two pulses (3 pm ited (Gaussian) pulses at the wavelength of the ICN distance is equivalent to 10 fs), starting when the two experiments, the spread of the wavelength-intensity pulses overlap. For each experiment at a chosen delay distribution of the pump laser is 3.5 nm for a 40-fs time, typically in the range of -100 to +I000 fs, the pulse; for the probe laser, 4.4 nm. If spectroscopic reso- detected ionization or fluorescence signal is integrated lution is required, no advantage is to be gained from for a sufficient length of time to yield a measure of total further shortening the laser pulse. In fact, because of intensity. The experiments are repeated at different the uncertainty principle, it is of little value from a delays and the so-called "transient" (a curve of intensi- chemical viewpoint to use pulses much shorter than ty as a function of time) is constructed. those now available. For example, at X = 308 nm, 7 fs In these experiments, probe pulse wavelengths are corresponds to 20 nm; thus the energy of the excitation generated by nonlinear methods, using doubling and is E = 4.02 eV, with an uncertainty of f 0.13 eV. This is a mixing crystals, or by producing a continuum of differ- total energy width of 0.26 eV, or 6 kcal per mole. It is ent wavelengths by focusing the femtosecond pulse on soon going to be necessary to compromise between water in a jet or a cell. The same can be done for the temporal and energy resolution. From a chemical dy- pump laser, and the desired wavelengths are selected namics point of view, since intermolecular and intra- by a monochromator or by using interference filters. molecular atomic velocities are usually of the order of The entire experiment is repeated at each new wave- magnitude of 1 km per second, a time interval of 1 fs length of interest to construct the spectra of the reac- corresponds to an atomic displacement of about 0.01 A. tion at a given available energy and at different prob- This is certainly "slow motion." Even for the fastest ing wavelengths. The Caltech machines utilize 40 fs reaction of all, H + Hz, the necessary time resolution is pulses from the colliding pulse-mode laser, 60 to 100 fs only about 10 fs. pulses from the amplified colliding pulse-mode laser system, and subpicosecond or 2- to 5-picosecond pulses Ferntochemistry of bimolecular reactions from the synchronously pumped dye laser systems. These examples illustrate the potential to conduct real-time femtochemistry observations of the dynamics Ferntochemistry of unimolecular reactions of a wide variety of photon-induced unimolecular pro- As mentioned above, the first of these femtosecond cesses, both dissociations and isomerizations. But the studies dealt with unimolecular reactions. The experi- field of chemical dynamics is broader than this. The ments were done on the elementary reaction of ICN + central theme of chemistry is the transformation of a I + CN. The pump pulse was at 307 nm and the probe set of reactant molecules into product molecules. The was set at a wavelength of 388.5 nm (the absorption overall mechanism of reactions usually involves a peak for free CN fragments) or detuned by as much as number of elementary steps, many of which are bimo- 10 nm to detect perturbed CN at these wavelengths, via lecular in nature. In the H2+ F2 chemical laser reaction, its laser-induced fluorescence. The signal is propor- for example, the key elementary step is the fast abstrac- tional to the number of perturbed CNs. As anticipated, tion of a hydrogen atom from the hydrogen molecule: when tuning to the perturbed CN fragment absorp- F + H2 + HF + H. In this bimolecular reaction, an tion, the transients exhibit a buildup and a decay char- H-H bond is transformed into an H-F bond via a three- acteristic of the short-lived (about 10-l3 second) transi- center transition state in such a way that the nascent tion states. On-resonance absorption of the free CN HF product is vibrationally and rotationally excited. fragment gives the time for dissociation of the ICN into Molecular beam studies of reactive scattering have the fragments, 205 f 30 fs. characterized a great many such elementary bimolecu- The time for bond breaking depends on the charac- lar reactions in terms of postcollision attributes, such as teristics of the potential energy surface, and the ob- energy distribution in the products and linear and an- served transients provide a way to view the dynamics gular momentum distribution. These attributes repre- on these potential energy surfaces. Basically, the fem- sent asymptotic properties that have been time-inte- tosecond probe pulse "sees" the potential at different grated over the course of the collision. It is highly intramolecular separations. desirable to be able to observe the collision itself in real For a simple, classical bond-breaking process, assum- time. ing pure exponential repulsive forces, the change of An important new issue arises for bimolecular reac- the potential with time and then the corresponding tions that is not a concern in the unimolecular case. For change with intermolecular separation is obtained di- a unimolecular reaction, the reaction begins when the rectly from the data. Such classical modeling by Ber- laser pulse initiates excitation and subsequent dissocia- sohn and Zewail, as well as a quantum-wave-packet tion. Thus, the interval between the firing of this laser treatment of the dissociation by Dan G. Imre at the pulse and the subsequent probe laser pulse represents University of Washington, can reproduce the main fea- the actual amount of time that the reaction has been tures of the experimental observations and provide taking place. But for a bimolecular reaction, whether in potential-energy-surface parameters. In general, how- bulk or in crossed beams, there is no comparable way to

38 November 7, 1988 C&EN 1 Van der Waals precursor molecule can be used to study bimolecular reaction Laser photolyzes HI Hot H atom attacks OCO Collision complex forms And breaks into products 4+ - 0 7

A T~me + A Pump laser Probe laser

directly. The van der Waals complex holds the reactants in close proximity to each other, thereby eliminating the time ordinarily required for them to find each other and collide. The pump laser initiates the reaction by photoiyzing the HI component of the precursor molecule. The probe laser, tuned to a resonance wavelength of the free OH, detects product formation. The laser-induced fluorescence signal of the OH shows a delay of several picoseconds due to the formation and decay of the HOCO complex. The inset at left By using a van der Waals precursor molecule, IH*-OCO, the shows the energetics of the overall reaction. [Adapted from lifetlme of the collislon complex for the bimolecular reac- N. F. Scherer, L. R. Khundkar, R. B. Bernstein, and A. H. tion H + C02+[HOCO] #+HO + CO can be measured Zewail, J. Chem. Phys., 87, 1451 (1987)]

establish the beginning point for the reaction. Even if Reisler, and coworkers at USC, and by Benoit Soep and one attempted to initiate a reaction by using a short coworkers at the University of Paris in France. These laser pulse to form radicals that can then react (such as, groups obtained the first studies of product state distri- for instance, in the reactions R + R'H --+ RH + R', or R' butions from reactions within the complexes. + R + RR'), the radicals would have to first find each In the picosecond experiment done in a collabora- other and collide before reaction can possibly occur. tion between Bernstein and Zewail's group, the birth In gas-phase systems, the average time between col- of OH from the reaction of atomic hydrogen with lisions is inversely proportional to the pressure; at sub- carbon dioxide was observed atmospheric pressures, these times are typically in the H OCO -* [HoCO]~-+ OH CO nanosecond to microsecond range, many orders of + + magnitude slower than the femtoseconds to picosec- The precursor molecule, IH.-OCO, was formed in a onds needed to traverse the transition state. Even for free-jet expansion of a mixture of HI and C02 in an crossed molecular beam experiments under the most excess of helium carrier gas. As pointed out by Wittig, favorable conditions, there is an uncertainty of many such van der Waals molecules have favorable geometry nanoseconds as to when the reactants will collide with that limits the range of impact parameters and angles one another. So there appears to be no hope of deter- of attack of the H on the OCO. The geometries of many mining the time of formation and decay of transient such van der Waals complexes have been determined collision complexes with lifetimes of only femtose- accurately by William Klemperer and coworkers at conds to picoseconds. Harvard University, using molecular beam electric res- However, a special trick enables the starting time of onance techniques. the reaction, that is the zero of time, to be established An ultraviolet-laser picosecond pulse initiates the within an uncertainty governed only by the duration experiment by photodissociating the HI, ejecting a of the laser pulse, for a whole class of bimolecular translationally hot H atom in the general direction of reactions. This now has opened up the possibility of the nearest 0 atom in the C02. A probe laser pulse studying real-time dynamics of bimolecular reactions. tuned to a wavelength suitable for detection of OH is The first of these experiments has been conducted at delayed relative to the photolysis pulse by specified Caltech in the picosecond, rather than femtosecond, time increments from zero to several tens of picosec- time domain, but the same principle applies for femto- onds. The OH appears after a few picoseconds. This chemistry. The method uses a beam of a van der Waals delay time is a direct measure of the lifetime of the "precursor molecule" containing the potential re- HOCO collision complex, formed from the H + C02 agents in close proximity, as prescribed by Wittig, reaction, at the given relative initial translational ener-

November 7, 1988 C&EN 39 Special Report gy. The lifetime changes with translational energy. A ized radiation and the outgoing velocity vector of a new series of experiments using femtosecond pulses photofragment. For a number of molecules, including will follow the formation and decay of the collision CH31, these angular distribution experiments imply complex in the femtosecond region, and be compared dissociation lifetimes of the excited states of less than a with predictions, such as the classical trajectory calcu- picosecond. For CH31, this promptness has been lations by the group of George C. Schatz at Northwest- checked by direct real-time measurements, which gave ern University, based on the ab initio calculated poten- a value of 0.4 picosecond or less. tial energy surfaces made by Larry Harding at Argonne A new spatial technique, introduced by the Bern- National Laboratory. stein group at UCLA, makes use of a beam of oriented It is probable that this technique will be applicable to symmetric top molecules prepared by an electrostatic a wide variety of bimolecular reactions, including not hexapole. Orientation of the molecules means not only only those with transition states that last for picosec- that one can distinguish "heads" from "tails," but that onds but also direct-mode reactions with transition the experimenter can cause them to be arranged state lifetimes in the femtosecond range. As with uni- "heads-up" or "heads-down" at will. Polarized laser- molecular reactions, off-resonance detection should al- induced photofragmentation of the oriented molecules low detailed measurements of the buildup and decay of within an electric field is carried out, and the ratio of the transition state with the zero of time for the bimo- the intensity of upward to downward-ejected photo- lecular reaction known to femtosecond resolution. fragments is measured. Strong up/down asymmetries are observed for several iodo-alkanes, confirming de- Femtochemistry and oriented molecules finitively the promptness of the photodissociation. Connections can be made between real-time experi- Photofragmentation of oriented molecule beams ments on the dissociation of the photoexcited states clearly yields independent spatial information, com- and experiments dealing with the spatial distributions plementary to the traditional angular distribution mea- of the photoproducts. In the traditional spatial experi- surements, on the promptness of the decay of the pho- ments, the angular distribution of the photofragments toexcited state. Moreover, the ability to select the rota- is measured by photolyzing randomly oriented tional quantum states of the molecules opens up the ground-state molecules with linearly polarized light. possibility of studying the influence of rotation on the The intensity of the ejected photofragments varies lifetime of the excited state. New theoretical advances with the angle between the electric vector of the polar- will be required, since the role of the orientational

Pump-probe experiments show photofragmentation of CH,I takes less than a picosecond Potential energy

Concept of a picosecond laser experiment to determine the bond breaking time for methyl iodide is shown schematically at left. The pump laser excites the CH31 molecule from the ground state to a higher energy repulsive state. Iodine product is detected by a delayed probe pulse using three- photon ionization. Varying the delay time between the pump and probe lasers produces the results shown below. Analysis gives an upper limit of 0.4 ps for the time required for separation of the photofragments. [Adapted from J. L. Knee, L. R. Khundkar, and A. H. Zewail, J. Chem. Phys., 83, 1996 (1985); L. R. Khundkar and A. H. Zewail, Chem. Phys. Lett., 142, 426 (1987).]

Iodine ion signal

I, CH, separation distance Time, picoseconds

40 November 7, 1988 C&EN Photofragmentation of oriented molecules confirms prompt dissociation

Orientation field plate ------_-----\ ------Beam of oriented

___.t

Time, nanoseconds

direction and the CH3 downward, the excited (repulsive) + state must be so short-lived that the molecules do not lose their orientation. Typical experimental results are shown Vertically above. The first arriving peak represents iodine atoms that polarized were ejected in the upward direction from the oriented laser beam -0 beam; the second peak comes from the downward-directed I, which takes a longer time to arrive at the detector. The degree of orientation can be estimated by comparing the "Snapshot" of polarized laser fragmentation experiment on intensity of the two peaks. This experiment confirms that oriented CH31 molecules shows, above, the molecules the lifetime of the transition state is less than a picosecond. passingthrough the laser beam with the I end up. Since the I [The experiment is that of S. R. Gandhi, T. J. Curtiss, and R. photoproduct is found to be ejected primarily in the upward B. Bernstein, Phys. Rev. Lett., 59, 2951 (1987).] quantum number in photoexcitation and subsequent and Donald G. Truhlar, now at the University of photofragmentation of the excited state has not yet Minnesota. been established. Interesting quantum interference Manifestation of these resonances in the real-time and coherence effects may be observable, as discussed probing of fragment separation would be a slowdown by Stuart A. Rice, Shaul Mukamel, Robin Hochstrasser, in the appearance of free fragments and possibly the and Zewail in the U.S., Moshe Shapiro in Israel, and appearance of oscillations reflecting the vibrational Paul W. Brumer in Canada. resonance frequency of the wave packet of the dissoci- For bimolecular reactions, crossed-beam reactive ating fragments. The simplest case would be a diatomic scattering experiments with oriented reagent mole- salt molecule, MX (from the alkali age), for which the cules confirm the shortness of the life of the collision initially excited state, MX*, is covalent. This state inter- complex in direct mode reactions. In the future, femto- acts with the ground electronic state, which is ionic chemistry may be extended to study the stereochemical (M+X-). Then the dynamics of bond breaking may in- dynamics of these reactions in real time. volve the phenomenon termed "potential curve crossing" as a result of this covalent-ionic interaction. Femtochemistry and resonances In a classic series of papers dating from the 1960s, R. The femtochemistrv discussed thus far has dealt with Stephen Berry and coworkers at the University of Chi-

elementary chemical reactions, such as ICN* -+ cago provided the foundation for the description of the [I...CN]#* -+ I + CN, for which bond breaking takes alkali halide potentials and the avoided crossing which place on dissociative potential surfaces. The molecules defines the adiabatic (trapped MX*) and diabatic (MX* are in "transition" for only a few hundred femtose- --+ M + X) behavior of the excited salt molecule. En conds or so, as evidenced by the rise and decay ob- route to products, the [M...X]#* transition state mole- served in the ICN experiments and confirmed by the- cules "decide" between the covalent and ionic poten- ory. If, however, in the process of falling apart, the tials. Either the wave packet of the MX* will be trapped system encounters a well in the potential surface, or if on the adiabatic potential curve without crossing or it there is more than one degree of freedom involved, the will dissociate by following the diabatic curve. The two system can be "trapped" and thus exhibit behavior limits have entirely different temporal behavior and, if indicative of quasi-bound states, or resonances. Reac- there is trapping, the frequency and amplitude of the tive scattering resonances were predicted based on the- oscillations will provide details of the nature of the ory in the early 1970s by Raphael D. Levine of Hebrew surfaces and the strength of the coupling. Roger Grice University of Jerusalem, Aron Kuppermann at Caltech, at the University of Manchester in the U.K. and

November 7, 1988 C&EN 41 Special Report

Femtochemistry reveals fragment trapping in alkali halide photodissociation

tempt toecrossto the ionic curve at an internuclear distance of 6.9 A. Molecules that attempt this crossing become trapped and oscillate back and forth in the upper potential energy well. Such trapping can be seen experimentally on the femtosecond timescale, above. The red curve shows Wave packet descrlption of the photofragmentationof sodi- the Na-l bond resonating in trapped molecules. The blue um iodide with two electronic degrees of freedom is shown curve is that of the sodium when separating from the iodine schematically above left. Light excites a packet of ground along the covalent curve. The curves were observed with state Nal molecules into an initial excited state wave pack- different degrees of detuning. [Adapted from M. Rosker, T. et, shown as the upper bell-shaped curve in the diagram. S. Rose, and A. H. Zewail, Chem. Phys. Lett., 146, 175 From here the molecules may follow the purple diabatic (1988) and T. S. Rose, M. Rosker, and A. H. Zewail, J. potential curve and dissociate covalently, or they may at- Chem. Phys., 88, 6672 (1988).]

Herschbach have provided a theory for this coupling et dynamics in these prototype systems will be impor- in the Landau-Zener regimes for alkali halides, and tant. Very recently Horia I. Metiu of the University of Mark S. Child at Oxford University has treated the role California, Santa Barbara, Marcus, Zewail, and cowork- of the coupling in determining the lifetime of quasi- ers have shown that this experimental resonance be- bound states. These systems have been studied experi- havior can be reproduced by a quantum calculation. mentally by absorption, emission, and molecular beam These real-time observations of trapping resonance photofragmentation techniques by a number of re- allow the motion of the wave packet to be viewed from searchers. the initial excitation all the way to "infinite separa- Such trapping resonances have been observed in tion," or fragmentation to atoms, making contact with femtochemistry experiments at Caltech on the reac- absorption spectroscopy, atomic beam scattering ex- tions periments, and photofragment translational spectroscopy. Such experiments promise to provide rich dy- NaI* -+ [Na...IIZ* -+ Na + I and namical information bearing on the shape of the poten- NaBr* -+ [Na--BrIZ*-+ Na + Br tial energy surfaces, curve crossings, and interactions For NaI, the oscillations are very strong and the aver- among different degrees of freedom. age period is 1.25 picoseconds, which corresponds to a frequency of 27 cm-I. This leads to the conclusion that Perspectives the packet for NaI is effectively trapped in the adiabatic It has now been 40 years since the introduction to well en route to products, and that the crossing is inef- photochemistry of flash photolysis with a time scale of ficient. The experiments show that the oscillation microseconds. Since then the pulse duration has been damping time for NaI is about 10 picoseconds. Thus, continually improved. In laser femtochemistry, the one crossing on the outward phase per oscillation "shutter speed has been increased by nearly nine or- (about 1 picosecond) has a probability of about 0.1 to ders of magnitude: The present limit of 6-fs time reso; escape from the well. For NaBr, the frequency of oscil- lution corresponds to a spatial window of about 0.06 A lation is similar in magnitude, but severe damping is through which to view the potential energy surface. observed. Thus, the crossing for NaBr is much easier But more important, with the last three orders of than for NaI, consistent with theoretical expectation. magnitude, it has become possible to observe chemis- An interesting question is how the "dephasing" and try as it happens. Femtochemistry is bringing reality to spread of the wave packet influences the dissociation the ephemeral, but all-important, transition states in rate, as determined by the decay of the oscillations. The chemical reactions. This happy marriage of ultrafast experiments show manifestations of this decay and lasers and chemistry promises an exciting future for spreading, and theoretical modeling of the wave pack- this field of real-time molecular reaction dynamics.

42 November 7, 1988 C&EN Suggested Readings Tannor, D. J., Rice, S. A., "Coherent Pulse Sequence Control of Ashfold, M. N. R., Baggott, J. E., editors, "Molecular Photodisso- Product Formation in Chemical Reactions," Adv. Chem. ciation Dynamics," Advances in Gas-Phase Photochemistry Phys., 70, 441 (1988). and Kinetics, Royal Society of Chemistry, U.K., 1987. Truhlar, D. G., "Resonances," ACS Symposium Series 263, Benson, S. W., "The Foundations of Chemical Kinetics," American Chemical Society, Washington, D.C., 1984. Krieger Publishing Co., Malabar, Fla., 1982. Zare, R. N., Dagdigian, P. J., Science, 185, 739 (1974). Bernstein, R. B., Herschbach, D. R., Levine, R. D., editors, Zewail, A. H., Letokhov, V. S., Zare, R. N., Bernstein, R. B., Lee, "Dynamical Stereochemistry," J. Phys. Chem., 91 (Oct. 8, Y. T., Shen, Y. R., Physics Today, 33 (1 1) (November 1980). 1987). "Dynamics of Molecular Photofragmentation," Faraday Disc. Brooks, P. R., "Spectroscopy of Transition Region Species," Chem. Soc., 82 (1986). Chem. Rev., 88, 407 (1988). Eisenthal, K. B., "Studies of Chemical and Physical Processes with Picosecond Lasers," Acc. Chem. Res., 8, 118 (1975). Eyring, H., Lin, S. H., Lin, S. M., "Basic Chemical Kinetics," John Wiley & Sons, New York, 1980. Felker, P. M., Zewail, A. H., "Picosecond Dynamics of IVR," Adv. Chem. Phys., 70, 265 (1988). Fleming, G. R., Siegman, A. E., editors, "Ultrafast Phenomena V," Springer-Verlag, New York, 1986. Herschbach, D. R., "Molecular Dynamics of Elementary Chemi- cal Reactions," Les Prix Nobel in 1986, Elsevier Publishing Co., Amsterdam, 1987. Hochstrasser, R. M., Weisman, R. B., "Picosecond Relaxation of Electronically Excited Molecular States in Condensed Me- dia," in "Radiationless Transitions," Lin, S. H., editor, p. 317, Academic Press, New York, 1980. Jortner, J., Levine, R. D., Rice, S. A., "Photoselective Chemis- try," Advances in Chemical Physics, Vol. XLVII, Pt. 1 & 2, John Wiley & Sons, New York, 1981. Kaufmann, K. J., Rentzepis, P. M., "Picosecond Spectroscopy in Chemistry and Biology," Acc. Chem. Res., 8, 408 (1975). Knee, J. L., Zewail, A. H., "Ultrafast Laser Spectroscopy of Chemical Reactions," Spectroscopy, 3, 44 (1988). Laubereau, A., Kaiser, W., "Vibrational Dynamics of Liquids and Solids Investigated by Picosecond Light Pulses," Rev. Mod. Phys., 50, 607 (1978). Lee, Y. T., "Molecular Beam Studies of Elementarv Chemical Processes," Science, 236, 793 (1987). Ahmed H. Zewail (right) is professor of chemical physics at Letokhov, V. S., "Nonlinear Laser Chemistry," Springer Series California lnstitute of Technology. He received B.S. and M.S. in Chemical Physics 22, Springer-Verlag, New York, 1983. degrees from Alexandria University in Egypt and a Ph.D. Levine, R. D., Bernstein, R. B., "Molecular Reaction Dynamics from the University of Pennsylvania. Following postdoctoral and Chemical Reactivity," Oxford University Press, New work at the University of California, Berkeley, he joined the York, 1987. Caltech faculty in 1976. He is the author of more than 200 Marcus, R. A., "Energy Distributions in Unimolecular Reac- scientific papers, the editor of four books on laser chemistry tions," Ber. Bunsenges. Phys. Chem., 81, 190 (1977). and spectroscopy, and associate editor of The Journal of Pilling, M. J., Smith, I. W. M., editors, "Modern Gas Kinetics," Physical Chemistry. His many awards and honors include Blackwell Scientific Publications, Oxford, 1987. the Camille & Henry Dreyfus Foundation teacher-scholar Polanyi, J. C., "Some Concepts in Reaction Dynamics," Sci- award, the Alexander von Humboldt Award for senior U.S. ence, 236, 680 (1987). scientists (West Germany), the 1985 ACS Eastern New York Shank, C. V., "Investigation of Ultrafast Phenomena in the Section's Buck-Whitney Medal, and a 1987 John Simon Gug- Femtosecond Time Domain," Science, 233, 1276 (1986). genheim Foundation fellowship. Siegman, A. E., "Lasers," University Science Books, Mill Val- Richard B. Bernstein has been a professor of chemistry at ley, Calif., 1986. the University of California, Los Angeles, since 1983. He was Smith, I. W. M., "Direct Probing of Reactions," Nature, 328, 760 previously Higgins Professor at Columbia University. Both (1987). his underaraduate and araduate dearees are from Columbia. " L, He has taught chemistry at a iiunrber of unrz~ersitic.;,iirclrrd- ing Illino~slnstitute of Technology, the Uilizlersity of Miclrr- Reprints of this C&EN special report will be available in black and white at $5.00 per copy. For 10 or more copies, $3.80 pw gun, of Wisconsin, and of Texas. He It; the author c~tj~~z~eral copy. Send requests to: Distribution, Room 210, American books and 300 scientific papers aird it; rdrtor of Chemical Chemical Society, 1155-10th St., N.W., Washin@m, D.C. Physics Letters. Among his nun~erolrsawards art. th~.ACS 20036.On orders of $20 or less, please send check or money Debye and Langmuir awards, the Chemical Ssrolc~~sAward order with request. of the National Academy of Screncrt;, and the 1988 W~~ls/r Award in Chemistry.

November 7, 1988 C&EN 43