Femtosecond Time-Resolved Photoelectron Spectroscopybromine.cchem.berkeley.edu/grppub/fpes28.pdf · many concepts in molecular electronics.9 Femtosecond time-resolved methods involve

Femtosecond Time-Resolved Photoelectron Spectroscopy

Albert Stolow*

Steacie Institute for Molecular Sciences, National Research Council of Canada, 100 Sussex Drive, Ottawa, Ontario, K1A 0R6 Canada

Arthur E. Bragg and Daniel M. Neumark*

Department of Chemistry, University of California, Berkeley, California 94720

Received August 6, 2003

Contents1. Introduction 17192. General Aspects of Femtosecond TRPES 1721

2.1. Neutral and Anionic Systems 17222.2. Photoelectron Angular Distributions 17242.3. Intensity Effects 1725

3. Development of Time-Resolved PhotoelectronSpectroscopy: 1985−1998

1726

3.1. Time-Resolved Photoelectron SpectroscopyExperiments

1726

3.2. Time-Resolved Photoelectron AngularDistributions

1727

3.3. Theoretical Developments 17274. Experimental Methods 1728

4.1. Photoelectron Spectroscopy of Neutrals andAnions

1728

4.2. Photoelectron−Photoion Coincidence Methods 17294.3. Femtosecond Laser Technology 1730

5. Applications 17305.1. Non-adiabatic Intramolecular Dynamics 17305.2. Excited-State Intramolecular Proton Transfer 17365.3. Photophysics of Model Molecular Switches 17375.4. Superexcited States 17385.5. Time-Resolved VUV/XUV/X-ray Photoelectron

Spectroscopy1739

5.6. Neutral Photodissociation Dynamics 17405.7. Neutral Reaction Dynamics 17425.8. Anion TRPES: I2- Nuclear Wave Packet

Dynamics1743

5.9. Dissociation Dynamics of Polyatomic Anions 17445.10. Photodissociation, Recombination, and

Reaction Dynamics in Anionic Clusters1744

5.11. Anion Vibrational Wave Packet andRelaxation Dynamics

1746

5.12. Charge-Transfer-to-Solvent and SolvatedElectron Dynamics

1746

5.13. Anion Intramolecular Electronic RelaxationDynamics

1747

5.14. Photoelectron Angular Distributions 17496. Conclusions and Future Directions 17537. Acknowledgments 1753

8. References 1754

1. IntroductionThe development of femtosecond time-resolved

methods for the study of gas-phase molecular dy-namics is founded upon the seminal studies of Zewailand co-workers, as recognized in 1999 by the NobelPrize in Chemistry.1 This methodology has beenapplied to chemical reactions ranging in complexityfrom bond-breaking in diatomic molecules to dynam-ics in larger organic and biological molecules, and hasled to breakthroughs in our understanding of fun-damental chemical processes. Photoexcited poly-atomic molecules and anions often exhibit quitecomplex dynamics involving the redistribution of bothcharge and energy.2-6 These processes are the pri-mary steps in the photochemistry of many polyatomicsystems,7 are important in photobiological processessuch as vision and photosynthesis,8 and underliemany concepts in molecular electronics.9

Femtosecond time-resolved methods involve apump-probe configuration in which an ultrafastpump pulse initiates a reaction or, more generally,creates a nonstationary state or wave packet, theevolution of which is monitored as a function of timeby means of a suitable probe pulse. Time-resolved orwave packet methods offer a view complementary tothe usual spectroscopic approach and often yield aphysically intuitive picture. Wave packets can behaveas zeroth-order or even classical-like states and aretherefore very helpful in discerning underlying dy-namics. The information obtained from these experi-ments is very much dependent on the nature of thefinal state chosen in a given probe scheme. Transientabsorption and nonlinear wave mixing are often themethods of choice in condensed-phase experimentsbecause of their generality. In studies of moleculesand clusters in the gas phase, the most popularmethods, laser-induced fluorescence and resonantmultiphoton ionization, usually require the probelaser to be resonant with an electronic transition inthe species being monitored. However, as a chemicalreaction initiated by the pump pulse evolves towardproducts, one expects that both the electronic andvibrational structures of the species under observa-tion will change. Hence, these probe methods can be

* To whom corresondence should be addressed. A.S.: telephone(613) 993-7388, fax (613) 991-3437, E-mail [email protected].: telephone (510) 642-3505, fax (510) 642-3635, [email protected].

1719Chem. Rev. 2004, 104, 1719−1757

10.1021/cr020683w CCC: $48.50 © 2004 American Chemical SocietyPublished on Web 03/12/2004

restricted to observation of the dynamics within asmall region of the reaction coordinate.

The present review article focuses upon gas-phasetime-resolved photoelectron spectroscopy, a probemethodology that has been demonstrated to be ableto follow dynamics along the entire reaction coordi-nate. In these experiments, the probe laser generatesfree electrons through photoionization or photode-tachment, and the electron kinetic energy (and/orangular) distribution is measured as a function oftime. Time-resolved photoelectron spectroscopy (TRP-

ES) was first developed in the early 1980s andapplied to electronic dynamics on semiconductorsurfaces.10 In this review of the literature from 1999until the present, we focus only on applications offemtosecond TRPES to isolated molecules and clus-ters, both neutral and negatively charged, in the gasphase. The first reviews of TRPES were publishedin 2000-2001,11,12 followed soon after by others.13-17

TRPES experiments have been performed on neu-tral molecules for several years using nanosecond(ns), picosecond (ps), and femtosecond (fs) lasers togenerate the pump and probe pulses. The type ofdynamics one can follow depends on the temporalresolution of the laser system. TRPES experimentswith ns or ps resolution have been used to probelifetimes and radiationless decay pathways of excitedelectronic states. With fs resolution, experiments ofthis type can be done on very short-lived electronicstates. In addition, one can follow a host of vibra-tional dynamics including dissociation, vibrationalrelaxation, and coherent wave packet motion. Theresults of ns or ps TRPES experiments are oftencomplementary to information obtained with othertechniques such as laser-induced fluorescence anddispersed emission. Femtosecond TRPES experi-ments are differential versions of experiments inwhich mass-selected ion yields are measured as afunction of pump-probe delay. For example, inphotodissociation problems, ion-yield experiments arevery useful for identifying any transient species andmonitoring the production of products,18 whereasTRPES provides an additional level of detail concern-ing the energy content of and charge redistributionin these species as a function of time.

Albert Stolow was born in Ottawa. He began studies in biochemistry atQueen’s University, Kingston, but turned to chemistry and physics afterhis second year. He did his Ph.D. studies in physical chemistry with JohnC. Polanyi at the University of Toronto. Albert was an NSERC postdoctoralfellow with Yuan T. Lee at the University of California, Berkeley, wherehe became interested in the dynamics of polyatomic molecules. In 1992,he joined the Femtosecond Science Program of the Steacie Institute forMolecular Sciences (National Research Council of Canada) to begin aresearch program in femtosecond dynamics based upon time-resolvedphotoelectron spectroscopy. He is currently a Senior Research Officer atthe NRC. His research interests include the development and applicationsof femtosecond time-resolved spectroscopies for the study of non-adiabaticmolecular dynamics in both gas and condensed phases, the coherentquantum control of molecular processes using nonperturbative laser fields,and the behavior of polyatomic molecules and clusters in strongnonresonant laser fields.

Arthur Bragg was born in 1977 in Rochester, Michigan. He studiedchemistry and physics at Albion College, Albion, Michigan, as a DowChemical and Barry M. Goldwater Scholar, graduating with a B.A. andhonors in 1999. Art subsequently received a National Science Foundationpredoctoral fellowship and began graduate studies at the University ofCalifornia, Berkeley. He currently studies the dynamics of anions usingtime-resolved photoelectron imaging in the group of Professor Daniel M.Neumark, with particular interest in the intramolecular electronic relaxationand alignment dynamics of photoexcited molecular and cluster anions.

Daniel Neumark received his B.A. in chemistry and physics and his M.A.in chemistry from Harvard University in 1977, where he did undergraduateresearch with Dudley Herschbach. He spent a year at Cambridge Universityworking with David Buckingham, and then attended graduate school atBerkeley, where he joined the research group of Yuan Lee. He receivedhis Ph.D. in physical chemistry in 1984, and then worked as a postdoctoralresearcher with Carl Lineberger at the University of Colorado, Boulder,from 1984 to 1986. In 1986, he returned to Berkeley as a faculty memberin the Chemistry Department. His research interests focus on severalareas in chemical dynamics and spectroscopy, including (i) studies ofreaction dynamics using transition-state spectroscopy, time-resolvedphotoelectron spectroscopy, and state-resolved photodissociation experi-ments on stable molecules and reactive free radicals, (ii) size-dependentspectroscopy and dynamics of clusters ranging from semiconductor clustersto He nanodroplets, and (iii) the effects of clustering and solvation onfundamental chemical processes. These issues are addressed through acombination of negative ion photodetachment and neutral beam experi-ments.

1720 Chemical Reviews, 2004, Vol. 104, No. 4 Stolow et al.

TRPES experiments have also been performed onmass-selected negative ions. While the more complexsources and lower number densities in negative ionexperiments present challenges that are absent in theneutral experiments, detachment energies are gener-ally significantly lower than ionization energies inneutral species and are typically overcome with easilygenerated probe laser wavelengths. Whereas in stud-ies of radiationless transitions most neutral TRPESexperiments “lose track” of dynamics once internalconversion to the ground state occurs, this limitationdoes not hinder negative ion experiments. In addi-tion, studies of clusters are straightforward in nega-tive ion experiments because the ions can be mass-selected prior to their interaction with the laserpulses; analogous neutral studies require collectingphotoions in coincidence with photoelectrons so thatthe identity of the ionized species can be ascertained.

In a typical instrument for a gas-phase experiment,a skimmed neutral molecular beam or mass-selectednegative ion beam interacts with pump and probelaser pulses in the interaction region of a photoelec-tron (PE) analyzer. Since these experiments involvepulsed lasers, time-of-flight electron energy analyzershave been used in most time-resolved PE experi-ments to date, although both 2D and 3D imagingmethods are beginning to contribute. It is particularlyuseful to use high collection efficiency analyzers toincrease the data acquisition rate; hence, parabolicreflectors19 or, more commonly, “magnetic bottle”analyzers20-22 with a collection efficiency of 50% orbetter have been incorporated into many of theinstruments currently in use. The recent develop-ment of high-power femtosecond lasers with repeti-tion rates of 1 kHz or higher has also greatlyfacilitated time-resolved PE experiments with fem-tosecond temporal resolution. The combination ofphotoelectron spectroscopy and femtosecond lasers isparticularly appealing because the energy resolutionof typical electron energy analyzers (10-50 meV) iscomparable to the energy resolution of a typicalfemtosecond pump-probe experiment (∼25 meV for100 fs pulses).

TRPES is particularly well-suited to the study ofultrafast non-adiabatic processes because photoelec-tron spectroscopy is sensitive to both electronicconfigurations (molecular orbitals) and vibrationaldynamics. An elementary but useful picture is thatemission of an independent outer electron occurswithout simultaneous electronic reorganization of the“core” (be it cation or neutral)sthis is called the“molecular orbital” or Koopmans’ picture.23 Thesesimple correlation rules indicate the continuum stateexpected to be formed upon single photon, singleactive electron ionization or detachment of a givenmolecular orbital. The probabilities of partial ioniza-tion into specific continuum electronic states candiffer drastically with respect to the molecular orbitalnature of the probed electronic state. If a givenprobed electronic configuration correlatessupon re-moval of a single active outer electronsto the groundelectronic configuration of the continuum, then thephotoionization probability is generally higher thanif it does not. This suggests the possibility of using

the electronic structure of the continuum as a probeof non-adiabatically evolving electronic configurationsin the neutral excited state, as was first proposedtheoretically by W. Domcke in 1991.24,25

2. General Aspects of Femtosecond TRPESFrom a physical point of view, time-resolved ex-

periments involve the creation and detection of wavepackets, defined to be coherent superpositions ofmolecular eigenstates. Due to the differing energyphase factors of the eigenstates, this superpositionis nonstationary. Extensive theoretical treatment ofthe propagation and probing of nonstationary statesin a TRPES experiment may be found elsewhere;26-32

only a descriptive summary of the theory follows.The three aspects of a femtosecond pump-probe

wave packet experiment are (i) the preparation orpump, (ii) the dynamical evolution, and (iii) theprobing of the nonstationary superposition state. Theamplitudes and initial phases of the set of preparedexcited eigenstates are determined by the amplitude(i.e., the “spectrum”) and phase (i.e., the “temporalshape”) of the pump laser field and the transitionprobabilities between the ground state |g⟩ and theexcited state of interest. Once the pump laser pulseis over, the wave packet ø(∆t), given by eq 1, evolvesfreely according to relative energy phase factors inthe superposition. The an complex coefficients contain

both the amplitudes and initial phases of the exact(non-Born-Oppenheimer) molecular eigenstates |Ψn⟩,which are prepared by the pump laser. The En arethe excited-state eigenenergies, given here in wave-numbers. The probe laser field interacts with thewave packet typically after the pump pulse is over,projecting it onto a specific final state |Ψf⟩ at sometime delay ∆t. This final state acts like a “moviescreen” onto which the wave packet dynamics isprojected. Knowing something about this “moviescreen” and whether the projection is “faithful” willobviously be important. The time dependence of thedifferential signal, Si(∆t), for projection onto a singlefinal state |Ψf⟩ can thus be written:

where

The complex coefficients bn contain the an from eq 1as well as the probe transition dipole moment andgeneralized vibronic overlap factors to the final state|Ψf⟩. The most detailed information in this final stateis the resolved differential signal Si(∆t). It arises froma coherent sum over all two-photon transition am-

|ø(∆t)⟩ ) ∑n

an |Ψn > e-i2πcEn∆t (1)

Si(∆t) ) |⟨Ψf| µb(rb)‚EBPROBE |ø(∆t)⟩|2 )

|∑n

bn e-i2πcEn∆t|2 (2)

bn ) an⟨Ψf| µb(rb)‚EBPROBE(∆t) |Ψn⟩

Si(∆t) ) ∑n

∑men

|bn||bm| cos{(En - Em)2πc∆t + Φnm}

Femtosecond Time-Resolved Photoelectron Spectroscopy Chemical Reviews, 2004, Vol. 104, No. 4 1721

plitudes consistent with the pump and probe laserbandwidths and therefore implicitly contains inter-ferences between degenerate two-photon transitions.It can be seen that the signal as a function of ∆tcontains modulations at frequencies (En - Em), theset of level spacings in the superposition. This is therelationship between the wave packet dynamics andthe observed pump-probe signal. It is the interfer-ence between individual two-photon transitions aris-ing from an initial state, through different excitedeigenstates, and terminating in the same single finalstate, which leads to these modulations. It is, inessence, a multilevel quantum beat. The powerspectrum of this time domain signal is comprised offrequencies and amplitudes (i.e., modulation depths)which give information about the set of level spacingsin the problem and their respective overlaps with aspecific, chosen final state |Ψf⟩. Different final states|Ψf⟩ will generally have differing overlaps with thewave packet leading to differing amplitudes in thepower spectrum (due to differing modulations depthsin the time domain signal). By carefully choosingdifferent final states, it is possible for the experimen-talist to emphasize particular aspects of the dynam-ics. In general, there is often a set of final states |Ψf⟩which may fall within the probe laser bandwidth. Wemust differentiate, therefore, between integral anddifferential detection techniques. With integral detec-tion techniques (e.g., total fluorescence, total ionyield), the experimentally measured total signal,S(∆t), is proportional to the total population in theset of all energetically allowed final states, ∑Si(∆t),created at the end of the two-pulse sequence. Infor-mation is lost in carrying out the sum ∑Si(∆t) sincethe individual final states |Ψf⟩ may each have differ-ent overlaps with the wave packet. Therefore, dif-ferential techniques such as dispersed fluorescence,translational energy spectroscopy, or photoelectronspectroscopy, which can disperse the observed signalwith respect to final state |Ψf⟩, will be important.Clearly, the choice of the final state is of greatimportance, as it determines the experimental tech-nique and significantly determines the informationcontent of an experiment. In many studies of non-adiabatic processes in polyatomic molecules, theinitially prepared “bright” state is often coupled to alower-lying “dark” state. The fact that this latter stateis “dark” means that observation of its dynamics ispotentially obscured, depending greatly upon thenature of the final state.

The molecular ionization (or detachment) continu-um as the final state in polyatomic wave packet ex-periments has both practical and conceptual advan-tages:24,25,33,34 (a) Charged particle detection is extreme-ly sensitive. (b) Detection of the ion or preselectionof the anion provides mass information on the carrierof the spectrum. (c) Ionization/detachment is alwaysan “allowed” process, with relaxed selection rules dueto the range of symmetries of the outgoing electron.Any molecular state can be ionized/detachedsthereare no “dark” states. (d) The final state in an ioni-zation/detachment process is often a stable cation/neutral species providing a well-characterized tem-plate on which to project the evolving wave packet.

(e) Highly detailed, multiplexed information can beobtained by differentially analyzing the outgoingphotoelectron as to its kinetic energy and angulardistribution. (f) Higher order (multiphoton) processes,which can be difficult to avoid in femtosecond experi-ments, are readily revealed. (g) Photoelectron-pho-toion coincidence and anion photodetachment mea-surements allow for studies of cluster solvationeffects as a function of cluster size and for studies ofvector correlations in photodissociation dynamics.

2.1. Neutral and Anionic SystemsAs discussed in the Introduction and illustrated by

the work of Baumert35 and Neumark,36,37 TRPES hasthe multiplexed ability to follow detailed vibrationalwave packet dynamics in excited molecular states,in some cases all the way to the products. In the workby Baumert and co-workers on Na2, two-photonexcitation was used to create a wave packet on theexcited 2 1Πg state of Na2, and the PE spectrum wasmeasured after delayed ionization with a probe pulse,as schematically depicted in Figure 1a. The photo-electron spectrum changes with the phase of thewave packet because of the difference between thepotentials for the 2 1Πg state and the X 2Σg

+ state ofNa2

+, with higher (lower) electron kinetic energy atthe inner (outer) turning point. As a result, a plot ofthe photoelectron spectra vs time (Figure 1b) showsthe complete excited-state wave packet dynamics.The negative ion TRPES experiments on I2

- byNeumark and co-workers probed the time scale fordirect dissociation to I(2P3/2) + I- on the repulsive A′2Σg,1/2

+ state, schematically drawn in Figure 15a; thePE spectrum evolves from a broad transient featureassociated with dissociating I2

- to very sharp featuresassociated with the atomic I- product in about 300fs, as shown in Figure 2.

Hayden38 first demonstrated that dynamics due tonon-adiabatic processes can also be monitored byTRPES, using internal conversion in hexatriene asan example. Non-adiabatic processes lead to changesin electronic symmetry, and, as proposed by Dom-cke,24 the simple Koopmans’ picture suggests that onemight be able to directly monitor the evolving excited-state electronic configurations during non-adiabaticprocesses while simultaneously following the couplednuclear dynamics via the vibrational structure withineach photoelectron band. In Figure 3, we show apicture of excited-state polyatomic wave packet dy-namics probed via TRPES, shown here for the caseof neutral molecule dynamics. (A similar picture canbe envisaged for the case of anion excited-statedynamics.) A zeroth-order bright state, R, is coher-ently prepared with a femtosecond pump pulse.According to the Koopmans’ picture, it should ionizeinto the R+ continuum, the electronic state of thecation obtained upon removal of the outermost va-lence electron (here chosen to be the ground electronicstate of the ion). This process produces a photoelec-tron band ε1.

We now consider any non-adiabatic coupling pro-cess which transforms the zeroth-order bright stateR into a lower lying zeroth-order “dark” state, â, asinduced by promoting vibrational modes of appropri-


ate symmetry. Again, according to the Koopmans’picture, the â state should ionize into the â+ ioniza-tion continuum (here assumed to be an electronicallyexcited state of the ion), producing a photoelectronband ε2. Therefore, for a sufficiently energetic probephoton (i.e., with both ionization channels open), weexpect a switching of the electronic photoionizationchannel from ε1 to ε2 during the non-adiabatic pro-cess. This simple picture suggests that one candirectly monitor the evolving excited-state electronicconfigurations (i.e., the electronic population dynam-

ics) during non-adiabatic processes while simulta-neously following the coupled nuclear dynamics viathe vibrational structure within each photoelectronband. In other words, the electronic structure of themolecular ionization (or detachment) continuum actsas a “template” for the disentangling of electronicfrom vibrational dynamics in the excited state, as wasfirst experimentally demonstrated for fs non-adia-batic processes in 1998-1999.39,40

The two limiting cases for Koopmans’-type correla-tions in such experiments, as initially proposed byDomcke,24,25 have been demonstrated experimen-tally.41,42 The first case, Type I, is when the neutralexcited states R and â clearly correlate to different

Figure 1. Femtosecond PE spectroscopy of Na2. (a)Potential energy curves of Na2/Na2

+ and experimentalexcitation schema: one- and two-photon excitation at 618nm creates vibrational wave packets at the inner turningpoint of the A 1Σu

+ and 2 1Πg states, respectively, withdynamics probed via nuclear-induced changes in the ion-ization potential. (b) Time-resolved photoelectron spectrashowing coherent wave packet motion on the excited Na2states. Reprinted with permission from ref 35. Copyright1996 American Physical Society.

Figure 2. Experimental fs photoelectron spectra of I2-.

Spectra at various delays (-30 to 320 fs); A, B, and Ccorrespond to the X 1Σg

+, A′ 3Π2u, and A 3Π1u states of I2measured in a one-photon UV background spectrum, whiledashed lines indicate the results of time-resolved wavepacket simulations. Reprinted with permission from ref 37.Copyright 1999 American Institute of Physics.

Figure 3. A TRPES scheme for disentangling electronicfrom vibrational dynamics in excited polyatomic molecules.A zeroth-order electronic state R is prepared by a fs pumppulse. Via a non-adiabatic process it converts to a vibra-tionally hot lower-lying electronic state, â. The Koopmans’-type ionization correlations suggest that these two stateswill ionize into different electronic continua: R f R+ +e-(ε1) and â f â+ + e-(ε2). When the wave packet haszeroth-order R character, any vibrational dynamics in theR state will be reflected in the structure of the ε1 photo-electron band. After the non-adiabatic process, the wavepacket has â zeroth-order electronic character; any vibra-tional dynamics in the â state will be reflected in the ε2band. This allows for the simultaneous monitoring of bothelectronic and vibrational excited-state dynamics. Re-printed with permission from ref 91. Copyright 2000Elsevier.


ion electronic continua, as suggested by Figure 3.Even if there are large geometry changes uponinternal conversion and/or ionization, producing vi-brational progressions, the electronic correlationsshould favor a disentangling of the vibrational dy-namics from the electronic population dynamics. Asan example, Stolow and co-workers studied ultrafastinternal conversion in a linear polyene (decatetraene,DT). The DT optically bright S2 state internallyconverts to S1 on ultrafast time scales. As discussedin more detail in a following section, the S2 stateelectronically correlates with the D0 cation groundstate, and S1 correlates with the D1 cation firstexcited state. In Figure 4 (top), we show the energylevel scheme relevant to this experiment. A femto-second pump pulse at 287 nm prepared the excitedS2 state at its origin. This then evolved into avibrationally hot S1 state via internal conversion. Thecoupled states were probed using single-photon 235nm ionization. As suggested by Figure 3, the evolving

electronic and vibrational character of the wavepacket alters the photoionization electronic channel,leading to large shifts in the time-resolved photo-electron spectrum. The experimental photoelectronkinetic energy spectra, shown in Figure 4 (bottom),indeed show a dramatic shift as a function of time.This shift is the direct (as opposed to inferred)signature of the changing electronic-state characterinduced by non-adiabatic coupling. This exampledemonstrates the selectivity of the molecular ioniza-tion (or detachment) continuum for specific excited-state configurations and shows that electronic popu-lation dynamics can be disentangled from vibrationaldynamics during ultrafast non-adiabatic processes.40,42

The other limiting case, Type II, is when the neutralexcited states R and â correlate equally strongly tothe same ion electronic continua. This is expected tohinder the disentangling of electronic from vibra-tional dynamics42 and is discussed in more detail ina following section.

2.2. Photoelectron Angular DistributionsThe other component of the molecular ionization

or detachment continuum is that of the free electron.Within the molecular frame, the symmetries of theoutgoing electron partial waves are likewise relatedto the symmetry of the electronic state undergoingionization. This can be viewed most simply (eq 3) asbeing due to the requirement that the product of thesymmetry species of the prepared excited state Γex,the dipole operator Γµ, the ion state Γ+, and the freeelectron wave function Γe must equal or contain thetotally symmetric irreducible representation ΓTS ofthe molecular symmetry group, in order that thetransition be allowed.

If, owing to a non-adiabatic process, the symmetryspecies of the initial zeroth-order state Γex changes,then the symmetry species of the outgoing electronΓe- must also change in order that the product remain(or contain) the totally symmetric species ΓTS. It isthe symmetry species of the outgoing electron Γe- thatrelates to the photoelectron angular distribution.Hence, measurement of time-resolved photoelectronangular distributions (PADs) also provides a probeof electronically non-adiabatic processes. We note,however, that a change in symmetry is not alwaysrequired to produce a change in the PAD. Twoexcited-state wave functions of identical electronicsymmetry may have quite different ionization matrixelements, leading to differences in the PADs. Thephotoemission is, in general, sensitive to orientationof the molecular frame with respect to the photoelec-tron recoil vector (however, unfavorable ionizationdynamics may obscure this effect). In favorable cases,a laboratory-frame alignment of the molecular framewill lead to anisotropies in the PAD. A coherentsuperposition of rotational states leads to a time-evolving molecular axis distribution in the laboratoryframe. In this situation, the PAD may evolve due tothe evolution of the rotational superposition state andcan directly reflect the axis distribution. The use of

Figure 4. (Top) Energy level scheme for TRPES of all-trans-decatetraene (DT), a molecule exhibiting Type IKoopmans’ correlations. The pump laser prepares theoptically bright state S2. Due to ultrafast internal conver-sion, this state converts to the lower-lying state S1 with∼0.7 eV of vibrational energy. The expected complementaryType I Koopmans’ correlations are shown: S2 f D0 + e-(ε1)and S1 f D1 + e-(ε2). (Bottom) TRPES spectra of DTpumped at 287 nm and probed at 235 nm. There is a rapidshift (∼400 fs) from ε1, an energetic band at 2.5 eV due tophotoionization of S2 into the D0 cation ground electronicstate, to ε2, a broad, structured band at lower energies dueto photoionization of vibrationally hot S1 into D1, the firstexcited electronic state of the cation. The structure in thelow-energy band reflects the vibrational dynamics in S1.These results illustrate the disentangling of the vibrationalfrom the electronic population dynamics. Reprinted withpermission from Nature (http://www.nature.com), ref 40.Copyright 1999 Nature Publishing Group.

Γex X Γµ X Γ+ X Γe- ⊇ ΓTS (3)


time-resolved PADs in excited-state molecular dy-namics has been recently reviewed, and we refer thereader to these reviews for detailed discussions.13,14,17

The measurement of time-resolved PADs is expectedto be particularly valuable when dynamical processesare not discernible from a photoelectron kineticenergy analysis alone. In general, PAD measure-ments are more involved and their calculation morechallenging in polyatomic systems than for the caseof photoelectron energy distributions. A considerationof practical issues such as the nature of the labora-tory-frame alignment (i.e., how the pump pulsecreates an anisotropic distribution of excited-statemolecular axes in the laboratory frame) and thesensitivity of the ionization dynamics to the orienta-tion of the molecular frame is important. Unfavorablelaboratory-frame alignment can potentially obscurethe underlying photoionization dynamics and reducethe efficacy of the PAD as an experimental probe.

2.3. Intensity EffectsAmplified femtosecond laser pulses are inherently

intense. The neglect of intensity effects can lead tomisinterpretation of results. For example, even mod-erate (f/40) focusing of a 100 fs pulse (at λ ) 300 nm)yields an intensity of 1012 W/cm2 with a pulse energyof only 1 µJ! It can be seen that amplified femtosec-ond pulse experiments will be very difficult to carryout in the perturbation theory (Golden Rule) limit.As the strong laser field physics of molecules is amplydiscussed elsewhere,43 we summarize here only brieflysome potential manifestations of nonperturbativeelectric fields that can appear in almost any femto-second experiment using amplified pulses. A char-acteristic phenomenon observed in photoelectronspectroscopy via strong field ionization is above-threshold ionization (ATI), the absorption of ad-ditional photons by an electron already above theionization potential (IP) that leads to a series of peaksin the photoelectron spectrum spaced by the photonenergy.44

Other strong field phenomena may also directlyaffect femtosecond photoelectron spectra. As dis-cussed below, strong Rabi oscillations can lead toaligned states and to ground-state wave packets.Another important related consequence is the dy-namic (or AC) Stark effectsthe shifting and modifi-cation of potential energy surfaces during the strongelectric field of the laser pulse. As discussed below,this leads to an alteration of the molecular dynamicalevolution and to sweeping field-induced resonancesthat may introduce phase shifts in the probe lasersignal. From a very practical point of view, it is oftendifficult to use one-photon preparation of an excited-state wave packet unless the absorption cross sectionfor this step is large, as high-intensity femtosecondpulses generally favor nonlinear over linear pro-cesses. Since the ratio of one- to two-photon crosssections is fixed (for a given molecule) and the ratioof the one- to two-photon transition probabilitiesscales inversely with intensity, the only possibilityis to reduce the laser intensity (W/cm2) until the one-photon process dominates. The signal levels at thispoint, however, may be very small, and very sensitive

detection techniques such as single particle countingare usually required.

In the electronic continuum, the dynamic Starkeffect manifests itself in terms of the ponderomotivepotential Up, well-known in strong-field atomic phys-ics.44 Classically, the ponderomotive potential of anelectron in a laser field of strength E and frequencyω is given by the time-averaged “wiggling” energy:

A convenient rule of thumb is that the pondero-motive shift is ∼60 meV TW-1 cm-2 at 800 nm(N.B.: 1 TW ) 1 × 1012 W): Up scales linearly withlaser intensity and quadratically with wavelength.Due to the spatio-temporal averaging of intensitiesin the pulse, the ponderomotive effect is observed asa red shift in electron kinetic energy and broadeningof the photoelectron spectrum.45

It is well-known in atomic physics that the dynamicStark effect causes field-induced changes in theenergies of the atomic levels and, hence, leads tofield-induced multiphoton resonances known as“sweeping” or “Freeman” resonances and to finestructure in atomic ATI photoelectron spectra.46 Asthe dynamic Stark effect is general, similar argu-ments apply to molecular systems. Beginning in1996, Baumert and Gerber and their co-workersmade extensive studies on the effect of laser intensityon TRPES.47-51 In particular, using the Na2 systemas a model, they made the first studies of ATIphotoelectron spectra as a function of time, usingintense 40 fs pulses at 618 nm. This study (andfollowing studies52) showed in detail how TRPESmaps the alteration of nuclear dynamics in intenselaser fields, using sweeping resonances as a new toolfor controlling molecular dynamics. Y. Chen and co-workers also reported the observation of intensityeffects on time-resolved PE spectra, using 2 + 1ionization of NO through its C 2Π Rydberg state.53

They observed field-induced Freeman resonancesattributed to the v ) 3 level of the NO A 2Σ+ statebeing shifted about 200 meV into near-resonancewith the NO C (v ) 0) level. A 2003 paper by S. L.Cong and co-workers showed how the dynamic Starkeffect could be used constructively to tune or selectspecific ionization pathways in TRPES of the NOmolecule, with resonances chosen between the F 2∆and E 2Σ+ states.54

Nonperturbative intensities can have a particularlypronounced effect on the measurement of photoelec-tron angular distributions (PADs). The anisotropicnature of the photoexcitation pump process resultsin an alignment of both the angular momentum andmolecular axis distributions of the excited state inthe laboratory frame (defined by the polarizationdirection of the pump laser field). In the Golden Rulelimit, for a single photon pump excitation, a cos2 θdistribution of excited-state molecular axes in thelaboratory-frame results from a parallel transition(sin2 θ from a perpendicular transition). Nonpertur-bative intensities will, due to rotational Rabi cyclingbetween the initial and intermediate states, give rise

Up ) e2E 2

4meω2(4)


to higher than expected alignment and can be usedto align the molecular states in the laboratoryframe.55-57 Seideman has shown that, although thepump laser intensity can have significant effects onPADs, the intensity effects of the probe laser areminimal.58 This is due to the fact that the “lifetime”of a free electron “near” the molecular core is usuallymuch less than the Rabi period. The pump laseralignment of the molecular axis distribution in thelaboratory frame directly affects the observed PADssince the relative contributions of parallel and per-pendicular ionization channels and the interferencesbetween them depend on the degree of alignment. Incases of favorable ionization dynamics, this can beexploited to follow the evolution of molecular axisalignment as a probe of rotation-vibration couplingin both diatomic59 and polyatomic60 molecules.

3. Development of Time-Resolved PhotoelectronSpectroscopy: 1985−1998

3.1. Time-Resolved Photoelectron SpectroscopyExperiments

TRPES experiments performed between 1985 and1998 have been reviewed previously12 and are there-fore considered only briefly in the following discus-sion. The first gas-phase TRPES experiments usingns and ps lasers were demonstrated in the mid-1980s,by Pallix and Colson61 on sym-triazine and by Sekre-ta and Reilly62 on benzene. In both cases, PES wasused to follow excited-state relaxation via intersystemcrossing to a triplet state in real time.

In the early 1990s, experiments by Reilly,63 Knee,64

and their co-workers showed that TRPES with psresolution could be used to follow intramolecularvibrational relaxation (IVR) in excited electronicstates. In IVR, the vibrational character of theinitially excited zeroth-order vibrational state evolves(i.e., dephases) with time, but the molecule remainsin the same electronic state. Reilly’s experimentswere performed on p,n-alkylanilines, while Kneeinvestigated IVR in fluorene, benzene, and aniline‚CH4 clusters.

Syage65,66 demonstrated that picosecond TRPEScould be used to follow reaction and solvation dynam-ics within clusters, the first example of a particularlypowerful application of the technique. This experi-ment was applied to excited-state proton transfer inphenol‚(NH3)n clusters. In a seminal study, Weberand co-workers used picosecond TRPES to followspin-orbit coupling (intersystem crossing) from theS1 state in aniline and 2- and 3-aminopyridine,making usesfor the first timesof the very importantrole of electronic symmetry correlations upon ioniza-tion.67 This work presaged the fs studies of non-adiabatic processes that were to follow. C. A. deLange and co-workers used picosecond TRPES tofollow dissociation in the repulsive A band of CH3I,68

and to measure lifetimes of the predissociative B 1E′′and C′ 1A1′ Rydberg states of NH3.69 Fischer, Schultz,and co-workers used picosecond TRPES to measurelifetimes of vibrational levels of the B, C, and Dexcited electronic states of the allyl radical, all ofwhich lie about 5 eV above the ground state.70-72

The first experiments at femtosecond time resolu-tion incorporating electron detection were reportedin the mid-1990s. In 1993, Baumert, Gerber and co-workers,73 followed by Stolow and co-workers in1995,33,34 applied femtosecond time-resolved ZEKEspectroscopy to vibrational wave packet dynamics inNa3 and I2, respectively. Gerber used a femtosecondpump pulse to create a vibrational wave packet inthe B state of Na3 and measured both total ion andZEKE electron yield as a function of pump-probedelay. Stolow’s experiments, applied to the I2 Bstate, were similar in principle but more extensiveand presented an outline of the experimental pros-pects for fs TRPES as applied to excited-state dy-namics in polyatomic molecules, emphasizing the roleof the molecular ionization continuum as the finalstate.

In 1996, Cyr and Hayden38 used femtosecondTRPES in its first application to non-adiabatic dy-namics in polyatomic molecules, investigating thedynamics of internal conversion in 1,3,5-hexatriene.In this work, the S2 state of 1,3,5-hexatriene wasexcited, and a combination of time-resolved ion yieldand PE spectroscopy measurements was used tounravel the ensuing dynamics. The PE spectra forthe cis isomer showed very rapid (∼20 fs) internalconversion from the S2 to the S1 state, followed byIVR within the S1 state on a time scale of 300 fs. Thisexperiment was the first example of how TRPES canbe used to follow ultrafast non-adiabatic dynamicsin polyatomic molecules.

The Stolow and Hayden experiments were soonfollowed by a flurry of activity in other laboratories,including the first negative ion experiments by Neu-mark36 on the photodissociation of I2

- and wavepacket dynamics experiments by Baumert35 on Na2.Baumert also studied intensity effects on the coher-ent control of wave packet dynamics of threshold52

and above-threshold ionization (ATI)48 photoelec-trons, as well as the chirp- and pulse-length depen-dence of molecular photoionization dynamics.74

The first TRPES experiments on metal clusteranions were performed by Gantefor et al.75 in 1998on Au3

-. These negative ion experiments are dis-cussed further in the context of more recent work inthe following sections. Femtosecond TRPES was alsoapplied to studies of dynamics in neutral clusters.In 1997, Soep and Syage studied the dynamics of thedouble proton-transfer reaction in the 7-azaindoledimer.76 By comparing PE spectra obtained by (1 + 1)ionization with 0.8 and 5.0 ps laser pulses, they foundsignificantly higher PE yield with shorter pulseionization, indicating a subpicosecond excited-statelifetime.

Stolow and Blanchet77 applied fs TRPES to thephotodissociation dynamics of polyatomic molecules,using the nitric oxide dimer (NO)2 as an example.Nanosecond studies of (NO)2 photodissociation dy-namics at 193 nm revealed that two product channelsare open:78 (NO)2* f NO(A 2Σ+ v,J) + NO(X 2Π v′,J′)and (NO)2* f NO(B 2Π v,J) + NO(X 2Π v′′,J′′). Fem-tosecond pump-probe measurement of the decaying(NO)2

+ parent ion signal as a function of delay timeyielded a time constant of 322 ( 12 fs, while fs


TRPES spectra showed a prominent sharp peak at0.52 eV due to the appearance of the NO(A 2Σ+,v)photodissociation product with a considerably slowerappearance time of 0.7 ps. This difference in timescales between excited parent signal disappearanceand product signal appearance was attributed to atwo-step, likely non-adiabatic mechanism, ratherthan a direct photodissociation mechanism. Thiswork showed that relying on parent ion yield mea-surements as a function of time in order to determinedissociation mechanisms and time scales may bepotentially misleading, especially if non-adiabaticprocesses are involved.

In 1997 Radloff, Hertel, Jouvet, and co-workersintroduced an important extension of TRPES. Theyapplied the photoelectron-photoion coincidence (PEP-ICO) detection technique to fs pump-probe spectros-copy,79,80 allowing the dynamics of size-selected neu-tral clusters to be studied via TRPES. As anillustrative example, shown in Figure 5, these au-thors studied internal conversion dynamics in ben-zene and benzene dimers. They found that the decayof the S2 to S1 states was very rapid for both species(about 50 fs), whereas the S1 f S0 IC decay wassignificantly faster in the monomer than in thedimer, 7.6 vs 100 ps. The same authors applied thePEPICO method to dynamics in ammonia clusters,investigating the dynamics of hydrogen transfer inelectronically excited (NH3)2.81 By monitoring the PEsignal associated with the (NH3)2

+ and NH4+ ions,

they were able to sort out the rather complicateddynamics associated with this system, finding timeconstants of 170 fs for the decay of (NH3)2* toNH4‚‚‚NH2, and 4 ps for dissociation of the lattercomplex to NH4 + NH2.

3.2. Time-Resolved Photoelectron AngularDistributions

In 1989, Y. Fujimura and co-workers prescientlyconsidered the effects of Coriolis coupling on ps time-resolved photoelectron angular distributions (PADs)in a model system.82 These authors showed thatb-axis Coriolis coupling leads to a k-coherence thatis directly observable in the time-resolved PADs. In1993, K. L. Reid independently proposed an approachto the study of excited-state dynamics through theuse of time-resolved PADs.83 Reid showed that thesensitivity of the PAD to the intermediate statealignment leads to evolution of the photoelectronangular distribution. Reid subsequently demon-strated this in a nanosecond-resolution experiment,using this time dependence to follow hyperfine de-polarization in the NO (A 2Σ+) state.84 These seminalstudies opened up the possibility of using the othercomponent of the electronic continuumsthat of thefree electronsas a probe of excited-state dynamics.

3.3. Theoretical DevelopmentsIn 1991, Seel and Domcke24,25 wrote pioneering

theoretical papers on femtosecond TRPES and pro-posed how it could be applied to non-adiabaticdynamics in polyatomic molecules, using S2 f S1internal conversion (IC) dynamics in pyrazine as anexample. These papers laid the foundation for thegeneral application of fs TRPES to problems regard-ing excited-state non-adiabatic dynamics. Of particu-lar significance, Domcke showed how the electronicstructure of the molecular ionization continuum andthe ionization correlations of the neutral configura-tions involved can be used to help understand the

Figure 5. Time-resolved PEPICO spectra of benzene monomer (left) and dimer (right) with λpump ) 200 nm and λprobe )267 nm. The ultrafast internal conversion process can be seen to be very different in the dimer than in the monomer. ThePEPICO technique allows complete separation of these channels. Reprinted with permission from ref 80. Copyright 1997Elsevier.


non-adiabatic dynamics in the neutral excited state.In 1993, Meier and Engel85 applied a treatmentsimilar to Domcke’s to TRPES of diatomic Na2,predicting many of the features associated withcoherent vibrational wave packet motion that werelater observed experimentally. These authors ex-panded these studies in 1994, looking at opportuni-ties for TRPES in the mapping of wave packetdynamics in double-welled potentials86 and develop-ing technically important approximation treatmentsof the photoionization dynamics, considerably sim-plifying the numerical task involved in computa-tions.30 In 1996, Meier and Engel studied the appli-cation of TRPES to non-adiabatic predissociationproblems involving the coherent decay of sets ofresonances,87 complementing Stolow’s 1996 experi-mental ion yield observation of these in the IBrmolecule.88 In 1997, Braun and Engel considered therole of initial temperature of the ground state inTRPES, using the hot cesium dimer as an example.89

In 1998, Meier and Engel explicitly considered theimportant role of frequency chirp (i.e., the timeordering of frequencies in a broad bandwidth pulse)in fs TRPES experiments, relevant to the discussionof quantum coherent control in this area.90

In 1997, T. Seideman developed a formal nonper-turbative theory of time-resolved PADs.26 This in-cluded exact (all orders) treatment of the matter-field interaction and included, in principle, the effectsof both nuclear and electronic dynamics on PADs.This work laid the foundation for the developmentof a research program, beginning in 1999, by Seide-man and co-workers on several aspects of time-resolved PADs.14

4. Experimental Methods

4.1. Photoelectron Spectroscopy of Neutrals andAnions

In any photoelectron spectroscopy measurement,the observables of interest are the electron kineticenergy (eKE) distribution and the photoelectronangular distribution (PAD). Spectrometers for fem-tosecond TRPES have modest energy resolutionrequirements as compared to modern standards forphotoelectron spectrometers. The bandwidth (fwhm)of a Gaussian 100 fs pulse is ∼150 cm-1. A pump-probe measurement involves the convolution of twosuch pulses, leading to an effective bandwidth of ∼25meV. This limits the energy resolution required inmeasuring the energy of the photoelectrons. How-ever, TRPES experiments are very data-intensive, asthey require the collection of many photoelectronspectra. As a result, most neutral and all anionTRPES experiments performed to date make use ofhigh-efficiency electron energy analyzers in which alarge fraction of the photoelectrons are collected.

The analyzer most commonly used in TRPESexperiments has been the magnetic bottle time-of-flight spectrometer.22,91 This technique uses a stronginhomogeneous magnetic field (1 T) to rapidly par-allelize electron trajectories to a flight tube and aconstant magnetic field (10 G) to guide the electronsto the detector.21,92 With careful design, the collection

efficiency of magnetic bottle spectrometers can exceed50%, while maintaining an energy resolution es-sentially equivalent to the fs laser bandwidth. Thehighest resolution is obtained for electrons createdwithin a small volume (φ < 100 µm) at the verycenter of the interaction region. In contrast with nslaser experiments, however, it is not desirable tofocus fs lasers to small spot sizes due to the inherenthigh intensity of such pulses, thus leading to mul-tiphoton ionization. Longer focal lengths mean thatthe Rayleigh range (the “length” of the focus) extendswell beyond the favorable region, leading in generalto an overall lower resolution.

Magnetic bottle analyzers have been used in manyneutral and anion TRPES experiments. They arerelatively simple, have high collection efficiency andrapid data readout, and, importantly, permit straight-forward photoelectron-photoion coincidence (PEPI-CO) measurements. Magnetic bottles suffer from thegeneral disadvantage that they can only be used todetermine eKE distributions; the complex electrontrajectories in magnetic bottle analyzers make itimpractical (if not impossible) to extract a PAD.There is an additional complication associated withthe use of magnetic bottle analyzers in anion TRPESexperiments. Mass-selected negative ion beams typi-cally have laboratory kinetic energies of 1 keV orhigher; hence, the velocity of the ion beam is nolonger negligible compared to that of the photoelec-trons. As a result, photoelectrons with the samespeed in the center-of-mass frame of reference butscattered at different polar angles with respect to theion beam will have different speeds in the laboratoryframe.93 This “Doppler broadening” significantly de-grades the energy resolution of a high-efficiencyanalyzer with no angular resolution, such as amagnetic bottle, yielding eKE resolution as poor asseveral hundred millielectronvolts. This resolution isoften sufficient for TRPES experiments but can beimproved (with loss of signal) to several tens ofmillielectronvolts by pulsed deceleration of the ionbeam, as has been demonstrated in several labora-tories,37,94-96 or, as shown recently by Cheshnovsky,97

by impulsive deceleration of the detached electrons.These limitations of magnetic bottle analyzers may

be overcome by the use of two-dimensional (2D)photoelectron imaging techniques, in which position-sensitive detection is used to measure the photoelec-tron kinetic energy and angular distributions simul-taneously. When used, as is common, with CCDcamera systems for image collection, one gives uprapid data readout, as multi-kilohertz readout ofCCD chips is very challenging. This precludes thepossibility of simple PEPICO measurements. Themost straightforward 2D imaging technique is pho-toelectron velocity-map imaging (VMI),98 a variantof the elegant photofragment imaging method devel-oped by Chandler and Houston.99 Photoelectron VMIwas first demonstrated by Eppink and Parker forneutrals98,100 and by Bordas101 and Sanov102 for nega-tive ions. Typically, a strong electric field projectsnascent charged particles onto a microchannel plate(MCP) detector. The ensuing electron avalanche fallsonto a phosphor screen, which is imaged by the CCD


camera. Analysis of the resultant image allows forthe extraction of both energy- and angle-resolvedinformation. In this case, a two-dimensional projec-tion of the three-dimensional distribution of recoilvelocity vectors is measured; various image recon-struction techniques103-105 are then used to recoverthe full three-dimensional distribution.

Photoelectron VMI thus yields close to the theo-retical limit for collection efficiency, along withsimultaneous determination of the photoelectron eKEand angular distributions. In addition, and of primeimportance to anion studies, VMI eliminates theDoppler broadening observed with magnetic bottleanalyzers, obviating the need for deceleration meth-ods. In anion time-resolved photoelectron imaging(TRPEI) studies utilizing velocity-map imaging, theresolution limit closely approaches the laser band-width.

This 2D particle imaging approach may be usedmost straightforwardly when the image is a projec-tion of a cylindrically symmetric distribution whosesymmetry axis lies parallel to the two-dimensionaldetector surface. This requirement usually precludesthe use of non-coincident pump and probe laserreference frames (i.e., other than parallel or perpen-dicular laser polarizations), a situation which mayprovide detailed information on intramolecular dy-namics through the alignment dependence of thePAD. It may be preferable to adopt fully three-dimensional (3D) imaging techniques based upon“time-slicing” 106,107 or full time- and position-sensitivedetection,108 where the full three-dimensional distri-bution is obtained directly without the introductionof mathematical reconstruction. In fs pump-probeexperiments where the intensities must be keptbelow a certain limit (often leading to single-particlecounting methods), “time-slicing” (which requiressignificant signal levels) may not be practical, leavingonly full time- and position-sensitive detection as theoption.

Modern MCP detectors can measure both spatialposition (x,y) on the detector face and time of arrival(z) at the detector face.108 In this situation, a weakelectric field is used to extract nascent chargedparticles from the interaction region. Readout of the(x,y) position yields information about the velocitydistributions parallel to the detector face, equivalentto the information obtained from 2D detectors. How-ever, the additional timing information allows mea-surement of the third (z) component of the laboratory-frame velocity, via the “turn-around” time of theparticle in the weak extraction field. Thus, thesedetectors allow for full 3D velocity vector measure-ments, with no restrictions on the symmetry of thedistribution or any requirement for image reconstruc-tion techniques. Very successful methods for fulltime- and position-sensitive detection are based uponinterpolation (rather than pixellation) using eithercharge-division (such as “wedge-and-strip”)109,110 orcrossed delay-line anode timing MCP detectors.111,112

In the former case, the avalanche charge cloud isdivided among three conductorssa wedge, a strip,and a zigzag. The (x,y) positions are obtained fromthe ratios of the wedge and strip charges to the zigzag

(total) charge. Timing information can be obtainedfrom a capacitive pick-off at the back of the last MCPplate. In the latter case, the anode is formed by apair of crossed, impedance-matched delay lines (i.e.,x and y delay lines). The avalanche cloud that fallson a delay line propagates in both directions towardtwo outputs. Measurement of the timing differenceof the output pulses on a given delay line yields thex (or y) positions on the anode. Measurement of thesum of the two output pulses (relative to, say, apickoff signal from the ionization laser or the MCPplate itself) yields the particle arrival time at thedetector face. Thus, direct anode timing yields a full3D velocity vector measurement. An advantage ofdelay line anodes over charge division anodes is thatthe latter can tolerate only a single hit per laser shot.This precludes the possibility of multiple coinci-dences, of interest in some experiments, and makesthe experiment very sensitive to background, suchas scattered UV light.

4.2. Photoelectron−Photoion CoincidenceMethods

Photoionization (photodetachment) always pro-duces two species available for analysissthe ion(neutral) and the electron. The extension of thephotoelectron-photoion coincidence (PEPICO) tech-nique to the femtosecond time-resolved domain wasfirst demonstrated by Stert and Radloff and wasshown to be very important for studies of dynamicsin clusters.79,81 In these experiments, a simple yetefficient permanent magnet design “magnetic bottle”electron spectrometer was used for photoelectrontime-of-flight measurements. A collinear time-of-flight mass spectrometer was used to determine themass of the parent ion. Using coincidence electronics,the electron time-of-flight (yielding electron kineticenergy) is correlated with an ion time-of-flight (yield-ing the ion mass). In this manner, TRPES experi-ments may be performed on neutral clusters, yieldingtime-resolved spectra for each parent cluster ion(assuming cluster fragmentation plays no significantrole). Signal levels must be kept low (much less thanone ionization event per laser shot) in order tominimize false coincidences.

Time- and angle-resolved PEPICO measurementsyielding photoion and photoelectron kinetic energyand angular correlations can shed new light onphotodissociation dynamics in polyatomic molecules.As shown by Continetti and co-workers for the caseof nanosecond laser photodetachment, correlatedphotofragment and photoelectron velocities can pro-vide a complete probe of the dissociation process.113,114

The photofragment recoil measurement defines theenergetics of the dissociation process and the align-ment of the recoil axis in the laboratory frame, thephotoelectron energy provides spectroscopic identi-fication of the products, and the photoelectron angu-lar distribution can be transformed to the recoilframe in order to extract vector correlations such asthe photofragment angular momentum polarization.The integration of photoion-photoelectron timingimaging (energy and angular correlation) measure-ments with femtosecond time-resolved spectroscopy


was demonstrated, using timing imaging detectors,in 1999 by C. C. Hayden and co-workers115,116 atSandia National Laboratories. This coincidence-imaging spectroscopy (CIS) method allows the timeevolution of complex dissociation processes to bestudied with unprecedented detail.

4.3. Femtosecond Laser TechnologyThe progress in femtosecond TRPES over the past

10 years derives from prior developments in femto-second laser technology, since techniques for photo-electron spectroscopy have been highly developed forsome time. There are several general requirementsfor such a femtosecond laser system. Most of theprocesses of interest are initiated by absorption of aphoton in the wavelength range ∼200-350 nm,produced via nonlinear optical processes such asharmonic generation, frequency mixing, and para-metric generation. Thus, the output pulse energy ofthe laser system must be high enough for efficientuse of nonlinear optical techniques and ideally shouldbe tunable over a wide wavelength range. Anotherimportant consideration in a femtosecond laser sys-tem for time-resolved photoelectron spectroscopy isthe repetition rate. To avoid domination of the signalby multiphoton processes, the laser pulse intensitymust be limited, thus also limiting the availablesignal per laser pulse. As a result, for many experi-ments a high pulse repetition rate can be morebeneficial than high energy per pulse. Finally, thesignal level in photoelectron spectroscopy is often lowin any case, and, for time-resolved experiments,spectra must be obtained at many time delays. Thisrequires that any practical laser system must runvery reliably for many hours at a time.

Modern Ti:sapphire-based femtosecond laser oscil-lators have been the most important technical ad-vance for performing almost all types of femtosecondtime-resolved measurements.117 Ti:sapphire oscilla-tors are tunable over a 725-1000 nm wavelengthrange, have an average output power of severalhundred milliwatts or greater, and can producepulses as short as 8 fs, but more commonly 80-130fs, at repetition rates of 80-100 MHz. Broadlytunable femtosecond pulses can be derived directlyfrom amplification and frequency conversion of thefundamental laser frequency.

The development of chirped-pulse amplificationand Ti:sapphire regenerative amplifier technologynow provides millijoule pulse energies at repetitionrates of >1 kHz with <100 fs pulse widths.118

Chirped-pulse amplification typically uses a gratingstretcher to dispersively stretch fs pulses from a Ti:sapphire oscillator to several hundred picoseconds.This longer pulse can now be efficiently amplified ina Ti:sapphire amplifier to energies of several milli-joules while avoiding nonlinear propagation effectsin the solid-state gain medium. The amplified pulseis typically recompressed in a grating compressor.

The most successful approach for generating tun-able output is optical parametric amplification ofspontaneous parametric fluorescence or a white lightcontinuum, using the Ti:sapphire fundamental orsecond harmonic as a pump source. Typically, an 800

nm pumped fs optical parametric amplifier (OPA) canprovide a continuous tuning range of 1200-2600nm.119 Noncollinear OPAs (NOPAs)120 pumped at 400nm provide microjoule-level ∼10-20 fs pulses whichare continuously tunable within a range of 480-750nm, allowing for measurements with extremely hightemporal resolution. A computer-controlled steppermotor is normally used to control the time delaybetween the pump and probe laser systems.

The development of femtosecond laser sources withphoton energies in the vacuum ultraviolet (VUV,100-200 nm), extreme ultraviolet (XUV, <100 nm),and beyond (soft X-ray) opens new possibilities forTRPES, including the preparation of high-lying mo-lecular states, the projection of excited states onto abroad set of cation electronic states, and, in the softX-ray regime, time-resolved inner-shell photoelectronspectroscopy. High harmonic generation in rare gasesis a well-established and important method for gen-erating fs VUV, XUV,121 and soft X-ray radiation.122-124

Harmonics as high as the ∼300th order have beenreported, corresponding to photon energies in excessof 500 eV. Both pulsed rare gas jets and hollow-coreoptical waveguides123,125 have been used for highharmonic generation. Sorensen and co-workers126

used the sixth harmonic (9.42 eV) of a Ti:sapphirelaser (generated from the third harmonic of the Ti:sapphire second harmonic) as the pump pulse inTRPES experiments. Even higher harmonics (17th)have been incorporated into TRPES experiments byLeone and co-workers and used for time-resolvedinner-shell photoelectron spectroscopy.127-129 As thesetechniques become more commonplace, the range ofapplicability of TRPES will be increased significantly.

Finally, it is worth noting that high harmonicgeneration provides the route to attosecond (10-18 s)pulse generation,130,131 generally producing wave-lengths within the soft X-ray regime. Although in itsinfancy, the first TRPES experiments using attosec-ond pulses have been reported132 and were used tostudy inner-shell electronic dynamics in atoms133

with sub-femtosecond time resolution. This providesa new and direct time domain approach to the studyof electron correlation in atoms and molecules.

5. Applications

In the following sections, we review recent applica-tions (1999-2003) of TRPES to problems in thedynamics of isolated molecules, anions, and theirclusters.

5.1. Non-adiabatic Intramolecular DynamicsAs discussed in the Introduction, a natural applica-

tion of TRPES is to problems in excited-state non-adiabatic coupling. Non-adiabatic dynamics involvea breakdown of the adiabatic (e.g., Born-Oppenhe-imer) approximation, which assumes that electrons“instantaneously” follow the nuclear dynamics. Thisapproximation is exact, provided that the nuclearkinetic energy is negligible, and its breakdown istherefore uniquely due to the nuclear kinetic energyoperator. Spin-orbit coupling, leading to intersystemcrossing, is not a non-adiabatic process in this


sense: the Born-Oppenheimer states could be cho-sen to be the fully relativistic eigenstates and hencewould be perfectly valid adiabatic states. Neverthe-less, the description of intersystem crossing as a non-adiabatic process is seen in the literature, and wetherefore include spin-orbit coupling problems inthis section.

In 1998-1999, Stolow considered and experimen-tally investigated the use of Koopmans’-type correla-tions toward distinguishing states involved in non-adiabatic transitions and delineating the two limitingcases for Koopmans’-type correlations in TRPESexperiments.41,42 Type I involves the neutral excitedstates R and â clearly correlating to different ionelectronic continua, as suggested by Figure 3. As anexample of Type I correlations, Stolow and co-workersconsidered ultrafast internal conversion in the linearpolyene all-trans-2,4,6,8-decatetraene (DT). The firstoptically allowed transition in DT is S2 (1 1Bu) rS0 (1 1Ag). The S2 state is a singly excited configura-tion. The lowest excited state is the dipole-forbiddenS1 (2 1Ag) state, which arises from configurationinteraction between singly and doubly excited Agconfigurations. Non-adiabatic coupling, leading toultrafast internal conversion from S2 to S1, is pro-moted by bu symmetry vibrational motions. As dis-cussed in more detail elsewhere,39,41 the S2 excitedstate electronically correlates with the D0 (1 2Bg)ground electronic state of the cation. The S1 state,by contrast, correlates predominately with theD1 (1 2Au) first excited state of the cation. As shownin Figure 4, a femtosecond pump pulse at 287 nm(4.32 eV) prepared the excited S2 state at its vibra-tionless electronic origin. It then evolves into avibrationally hot (∼0.7 eV) S1 electronic state viainternal conversion. Using a 235 nm (5.27 eV) UVprobe photon, the excited-state wave packet dynamicswere probed. As the non-adiabatic coupling proceeds,the evolving electronic character of the wave packetalters the electronic photoionization channel, leadingto large shifts in the time-resolved photoelectronspectrum.

The experimental photoelectron kinetic energyspectra in Figure 4 (bottom) reveal a rapid shift ofelectrons from an energetic peak (ε1 ) 2.5 eV) to abroad, structured low-energy component (ε2). The 2.5eV band is due to ionization of S2 into the D0 ion state.The broad, low-energy band arises from photoioniza-tion of S1 that correlates with the D1 ion state. Itsappearance is due to population of the S1 state byinternal conversion. Integration of the two photoelec-tron bands directly reveals the S2-to-S1 internalconversion time scale of 386 ( 65 fs. It is importantto note that these results contain more informationthan the overall (integrated) internal conversiontime. The vibrational structure in each photoelectronband yields information about the vibrational dy-namics, which promote and tune the electronic popu-lation transfer. In addition, it gives a direct view of theevolution of the ensuing intramolecular vibrationalenergy redistribution (IVR) in the “hot molecule”which occurs on the “dark” S1 potential surface.41

In the other limiting case, Type II, the one-electroncorrelations, upon ionization, lead to the same cat-

ionic states. Stolow and co-workers went on tocharacterize the limiting cases of ionization correla-tions and to include, with important theoreticalcontributions from T. Seideman, the continuumelectron as part of the final state description.134 Anexample of Type II correlations is seen in the S2-S1internal conversion in the polyaromatic hydrocarbonphenanthrene (PH), discussed by Stolow and co-workers in more detail elsewhere.42 In the case of PH,both the S2 and the S1 states correlate similarly withthe electronic ground state as well as the first excitedstate of the cation. In PH, the non-adiabatic couplingof the bright electronic state (S2) with the densemanifold of zeroth-order S1 vibronic levels leads tothe nonradiative “decay” (dephasing) of the zeroth-order S2 state. The energy gap between these twoexcited states is large, and the density of S1 vibroniclevels is extremely large compared to the reciprocalelectronic energy spacing.

In this experiment, PH was excited from the S01A1

ground state to the origin of the S21B2 state with a

282 nm (4.37 eV) fs pump pulse and then ionizedafter a time delay ∆t using a 250 nm (4.96 eV) probephoton. The S2

1B2 state rapidly internally convertedto the lower lying S1

1A1 state at 3.63 eV, transform-ing electronic energy into vibrational energy. In PH,both the S2

1B2 and S11A1 states can correlate with

the D02B1 ion ground state. The time-resolved

photoelectron spectra for PH, shown in Figure 6,revealed a rapidly decaying but energetically narrowpeak at ε1 ≈ 1.5 eV due to photoionization of thevibrationless S2

1B2 state into the ionic ground stateD0

2B1, resulting in a decay time constant of 520 ( 8fs. A broad photoelectron band, centered at about∼0.7 eV, in these photoelectron spectra was due toionization of vibrationally hot molecules in the S1state, formed by the S2-S1 internal conversion. Attimes t > 1500 fs or so (i.e., after internal conversion),the photoelectron spectrum is comprised exclusivelyof signals due to S1 ionization. The S1 state itself islong-lived on the time scale of the experiment.Despite the fact that Type II molecules present anunfavorable case for disentangling electronic fromvibrational dynamics, in PH a dramatic shift in thephotoelectron spectrum was seen as a function oftime. This is due to the facts that PH is a rigidmolecule and the S2, S1, and D0 states all have similargeometries. The photoionization probabilities aretherefore dominated by small ∆v transitions. Hence,the 0.74 eV vibrational energy in the populated S1state should be roughly conserved upon ionizationinto the D0 ionic state. Small geometry changes favorconservation of vibrational energy upon ionizationand thereby permit the observation of the excited-state electronic population dynamics via a photoelec-tron kinetic energy analysis alone. In general, how-ever, significant geometry changes will lead tooverlapping photoelectron bands, hindering thedisentangling of vibrational from electronic dynamics.As discussed below, time-resolved photoelectron an-gular distributions provide another view of elec-tronically non-adiabatic processes, which may, infavorable situations, shed more light on this prob-lem.


The Koopmans’ picture is a simplification of theionization dynamics. Nevertheless, it provides auseful zeroth-order picture from which to consider theTRPES results. Any potential failure of this inde-pendent electron picture can always be experimen-tally tested directly through variation of the photo-ionization laser frequencysresonance structuresshould lead to variations in the form of the spectrawith electron kinetic energy, although this effect ismore likely to be prominent in PAD measurements.

In 1999, Suzuki and co-workers first applied their2D photoelectron imaging method to the study of psintersystem crossing (ISC) dynamics in the pyrazineS1

1B3u (nπ*) state and fs internal conversion in theS2

1B2u (ππ*) state.135,136 The importance of this meth-od is that it allowed the determination of both energyand angular distributions. This work on ps ISC nicelycomplements the 1995 energy-resolved photoelectronstudies of ISC dynamics in aniline and aminopyri-dines by Weber and co-workers.67 The additionalinformation provided by analysis of the PAD asym-metry parameters can be invaluable in determiningthe excited-state photophysical mechanisms. In theirimaging experiments, Suzuki and co-workers pre-pared the excited state with a pump laser and usedtwo-photon ionization as a probe. The S1 state was

seen to decay in ∼100 ps. The S2 state, lying about7000 cm-1 above the S1 origin, was also prepared withtheir pump pulses. As the lifetime of the pyrazine S2state is expected to be about 30 fs24,25 and thereported time resolution in Suzuki’s experiments wasabout 450 fs, these authors were able to observe thedecay of the highly excited S1 state formed by internalconversion. They report a lifetime of 22 ps for thisand, for comparison, a lifetime of 39 ps for theperdeuterated pyrazine-d4 molecule, also prepared atits S2 origin. As in Weber’s data, the buildup of theT1 triplet state formed by the intersystem crossingwas directly observed. The authors had initiallyassigned another peak in the energy-resolved spec-trum to the T2 triplet state, but this was correctedin a later publication. As Suzuki first demonstrated,the powerful combination of energy- and angle-resolved photoelectron spectra brings an importantnew tool into TRPES studies. The photoelectron an-gular distributions, which were extracted from thesemeasurements, are discussed in a following section.

In 2000, Radloff and co-workers applied TRPES tothe famous pyrazine S2 lifetime problem,137 connect-ing with Domcke’s earlier calculations. Using shortUV laser pulses (∼130 fs), Radloff excited pyrazineS2 with two different wavelengths: 200 (6.2 eV) and196 nm (6.31 eV). To avoid the complications ofintermediate resonances, dynamics were probed viasingle-photon ionization using pulses with wave-lengths of 266.7 (4.66 eV) and 261.3 nm (4.75 eV),respectively. As shown in Figure 7, the photoelectronspectra vary significantly with time during the cross-correlation of the laser pulses. These changes are dueto the ultrafast variation of the electronic characterfrom ππ* (S2) to nπ* (S1), leading to a switching ofthe Koopmans’-type ionization correlations from theD1 A 2B1g(π-1) to the D0 X 2Ag(n-1) cation electronic

Figure 6. (Top) Energy level scheme for TRPES ofphenanthrene, a molecule exhibiting Type II Koopmans’correlations. The pump laser prepares the optically brightstate S2. Due to ultrafast internal conversion, this stateconverts to the lower-lying state S1 with ∼0.74 eV ofvibrational energy. The expected corresponding Type IIKoopmans’ correlations are shown: S2 f D0 + e-(ε1) andS1 f D0 + e-(ε2). (Bottom) TRPES spectra of phenanthrenefor a pump wavelength of 282 nm and a probe wavelengthof 250 nm. The disappearance of the band ε1 at ∼1.5 eVand growth of the band at ε2 at ∼0.5 eV represents a directmeasure of the S2-S1 internal conversion time (520 fs).Reprinted with permission from Annual Reviews in Physi-cal Chemistry (http://www.AnnualReviews.org), ref 15.Copyright 2003 Annual Reviews.

Figure 7. TRPES spectra of pyrazine pumped at 267 nmand probed at 200 nm. The S2 and S1 contributions can bedeconvoluted. The ultrafast S2 internal conversion occursin about 20 fs, supporting earlier theoretical results. TheS1 state decays in 22 ps. Reprinted with permission fromref 137. Copyright 2000 American Institute of Physics.


states. By fitting the longer-lived components of thephotoelectron spectra, Radloff and co-workers wereable to deconvolute the S2 photoelectron spectra andobtained a lifetime of 20 ( 10 fs for this state,confirming the theoretical estimates of Domcke.Interestingly, the S2 photoelectron spectrum appearsto show correlations with both the D1 and D0 cationstates, although the D1 contribution is dominant, asexpected. This can be understood in terms of non-adiabatic coupling between the two cation states.138

These experiments serve to show that, with carefuldispersion management of fs UV pulses, extremelyshort time scale non-adiabatic processes may bedirectly observed.

In 2000, Stolow and co-workers investigated elec-tronic relaxation rates in a series of simple linear R,â-enones: acrolein (i.e., propenal), methyl vinyl ketone(i.e., R-methylpropenal), and crotonaldehyde (i.e.,γ-methylpropenal).139 Excited at their S2 (ππ*) ori-gins, these molecules internally convert to the S1(nπ*) on ultrafast time scales (30-50 fs). As is well-known in aldehyde and ketone photophysics, the low-lying nπ* state leads to ultrafast intersystem crossingrates and triplet photochemistry. The triplet productquantum yields depend on the competition, in the S1state, between intersystem crossing to triplets andinternal conversion to the S0 ground state. Verydramatic effects on the S1 relaxation rates due to thelocation of the methyl group were observed: upon S2(ππ*) excitation, the S1-S0 internal conversion ratein crotonaldehyde (γ-methyl) was ∼5 times morerapid than the same in either methyl vinyl ketone(R-methyl) or acrolein (no methyl), the latter twohaving very similar S1-S0 internal conversion rates.One might expect, in the Golden Rule sense, that themethyl substituent increases the density of vibra-tional states and therefore a faster decay rate ob-tains. The identical number of degrees of freedom inmethyl vinyl ketone and crotonaldehyde suggeststhat vibrational density of states is not the dominanteffect. The methyl effect on the torsional conicalintersections at the CdC bond could be due to achange in diabatic vs adiabatic torsional dynamicsat the conical intersection (an inertial effect due tothe lowering of the torsional frequency upon methy-lation) and, possibly, to a subtle re-ordering of theconical intersections upon methylation (an electroniceffect).140 The methyl group effect on CdC conicalintersections seen in the R,â-enones is of generalinterest because it illustrates how simple substitu-tions could be used to provide an important new typeof control over the electronic branching ratios.139

In 2001, G.Gerber, A. H. Zewail, T. Baumert, andco-workers studied the famous cis-stilbene photo-isomerization problem using a combination of massspectrometry and magnetic bottle TRPES.141 Zewailand co-workers had previously determined the S1lifetime to be 320 fs when the molecule was excitedwith a 285 nm pump pulse.142 In this study, a 267nm pump pulse excited the molecule to the S1 state.A 400 nm probe pulse was used for two-photonionization of the excited-state wave packet. Theresults indicated that two long-lived photoproductswere formed.

In 2001, Soep, Mestdagh, Visticot, and co-workersapplied fs-ns time scale spectroscopies to studyexcited-state dynamics in tetrakis(dimethylamino)-ethylene (TDMAE) over a wide range of time scales:from femtosecond to picosecond to nanosecond.143

Using a combination of fluorescence, ion, and pho-toelectron detection schemes, the authors followedthe complete evolution of the excited states. Pump-probe measurements were performed at 200, 266, and400 nm excitation wavelengths, with the Ti:sapphirefundamental (800 nm) used as the probe color. TheTDMAE molecule has an unusually low ionizationpotential of 5.5 eV. At 200 nm excitation, strongfragmentation of the parent ion was observed. At 266nm pump, the TRPES results showed significantvariation. At zero time delay, a sharp peak near zeroelectron kinetic energy is observed. By 300 fs, thisbecomes a double-peaked structure with maximanear 0 and 0.8 eV. The decay of the sharp peak andthe growth of the 0.8 eV peak were commensurate,with a time constant of ∼300 fs. At 400 nm pump, across-correlation signal was observed, indicative ofan extremely fast or nonresonant process, followedby a small but slowly decaying pedestal. At 266 nm,the dynamics were interpreted in terms of a transi-tion to a low-lying Rydberg state plus the ππ* valencestate. By analogy with ethylene, the 300 fs time scalewas assigned to the C-C twist upon approach to theN-V states’ conical intersection. Subsequent forma-tion of the zwitterionic π*π* Z state was inferred.This example is important because it shows theconnection of the fs TRPES results with the longertime scale fluorescence measurements in order todetermine the dynamics over multiple time scales.

In 2001, Radloff and co-workers applied the time-resolved PEPICO technique to the internal conver-sion dynamics of toluene and the toluene dimer.144

These were excited to a vibrationally excited statewithin the S2 manifold via 150 fs pump pulses at 202nm (6.14 eV), followed by fs single-photon ionizationat 267 nm. Toluene makes an interesting comparisonwith the earlier time-resolved PEPICO studies ofbenzene monomer and dimer79,80 due to its lowersymmetry. Both toluene monomer and dimer S2states were observed to decay with a time constantof ∼50 fs. Time-resolved photoelectron spectra coin-cident with the toluene dimer are shown in Figure8. The rapid decay of the S2 state and the formationof the long-lived S1 state can be seen. By analogy withbenzene, the very high S2 internal conversion rateswere attributed to a conical intersection near the S2origin, leading to the formation of vibrationallyexcited S1 states. This S1 lifetime in the toluenemonomer was determined to be 4.3 ps. Surprisingly,in the toluene dimer, the lifetime of the vibrationallyexcited S1 state was greater than 100 ps. Further-more, despite the large internal energy, little frag-mentation of the dimer was observed. This suggeststhat the S1 vibrational dynamics involve intramo-lecular dynamics that have very poor coupling to theintermolecular vibrations.

In 2001, Suzuki and co-workers applied the time-resolved 2D imaging technique to their continuingstudies of pyrazine excited-state dynamics.145 In that


paper, the authors investigated in detail the spec-troscopy and dynamics of the 3s and 3p Rydbergstates involved in their previous two-probe photonionization studies of the pyrazine S1 state.135 Throughvariation of the pump laser wavelength from 401 to330 nm, series of singlet 3s(π-1), triplet 3s(n-1), andtriplet 3pA(n-1) Rydberg states with sub-ps lifetimeswere observed. As discussed in a following section,the time-resolved photoelectron angular distributions(PADs) and their variation with pump-probe polar-ization geometry were cleanly observed via the 2Dimaging method, and this provided essential infor-mation on the state assignments. This work alsonicely illustrates the importance of understandingthe variety of laboratory-frame alignment effects thatcan occur in TRPES experiments on excited statesof polyatomic systems.

In 2001, G. Stock and co-workers developed a newcomputational scheme for the efficient calculation offs TRPES spectra in polyatomic molecules.146 Toavoid the “costly” discretization of the ionizationcontinuum, these authors made use of the fact thatthe ionization yield at a fixed photoelectron energyis formally equivalent to the transient absorptionspectrum into a single excited electronic state. Usewas then made of a new convolution scheme for the

rapid calculation of the transient absorption spec-tra.147 These authors applied their scheme to the four-mode vibronic coupling problem in S2 pyrazine,including three neutral states (S2, S1, and S0) and twoionic states (D0 and D1), and treated the finiteduration of the laser pulses in a numerically exactmanner. Compared with the earlier work of Dom-cke,24,25 this work included new ab initio data,incorporated another vibrational mode (ν9a) knownto be important in the dynamics, and explicitlyincluded D0-D1 vibronic coupling in the cation states.To model the experimental results of Radloff and co-workers,137 pulse durations of 100 fs were chosen. Thesimulations reproduced the ∼20 fs time scale for theinternal conversion and showed a significant red shiftof the photoelectron spectrum, reflecting the vibra-tional energy redistribution in the S1 state. A com-parison of the calculated vs experimental photoelec-tron spectra showed that the simulations were missingthe low-energy photoelectrons seen in the experi-ments. It was suggested that these could originatefrom the autoionization of Rydberg states. This workis important because it opens up new methods forthe direct comparison of experimental with theoreti-cal TRPES in real polyatomic systems.

In 2002, K. L. Reid and co-workers presented a psTRPES study of intramolecular dynamics in S1p-fluorotoluene (pFT).148 These researchers empha-sized the importance of the vibrational energy resolu-tion obtainable with ps pulses for the study ofintramolecular vibrational energy redistribution (IVR).As compared with p-difluorobenzene (pDFB), pFTexhibits anomalous IVR “lifetimes”: at ∼2000 cm-1

excess energy in S1, the IVR “lifetimes” are 150 and3.5 ps for pDFB and pFT, respectively.149 The methylgroup internal rotation was thought to play animportant role in this effect. In these TRPES studies,an excess S1 vibrational energy of 1200 cm-1 wasemployed. Photoelectron spectra recorded at 5, 10,30, and 60 ps time delays showed deterioration inthe resolvable vibrational structure, particularly athigher ion internal energies. This evolution wasattributed to IVR in the S1 manifold and supportedthe previous chemical timing studies of Parmenter.149

The well-resolved vibrational structure in the pho-toelectron spectra should permit further analysis interms of vibrational energy flow models.

In 2002, A. Stolow, I. C. Chen, and co-workersstudied substituent effects in electronic relaxationdynamics using a series of monosubstituted benzenesas model compounds.150 They focused on the first andsecond ππ* states of these aromatic systems and onsubstituents which were expected to have electroniceffects. Six benzene derivatives were studied: ben-zaldehyde (BZA), styrene (STY), indene (IND), aceto-phenone (ACP), R-methylstyrene (R-MeSTY), andphenylacetylene (φACT). This choice of substituentsaddressed several points: (i) the effect of the sub-stituent on the electronic states and couplings; (ii)the effect of the substituent on the rigidity or flop-piness of the molecular frame; (iii) a comparison ofType I with Type II Koopmans’ systems; and (iv) thepotential effects of autoionization resonances (i.e.,non-Koopmans’ behavior) on the observed dynamics.

Figure 8. Time-resolved PEPICO spectra of toluene dimerwith λpump ) 202 nm and λprobe ) 269 nm, showingvibrational structure. The ultrafast S2 internal conversionoccurs in ∼50 fs for both monomers (not shown) anddimers. By contrast, the S1 internal conversion takes only4.3 ps for the monomer but >100 ps for the dimer. Again,the PEPICO technique allows complete separation of thesechannels. Reprinted with permission from ref 144. Copy-right 2001 American Chemical Society.


Three electronically distinct substituents were cho-sen: CdO, CdC, and CtC. For the CdC, potentialoff-axis conjugation effects with the ring in STY werecontrasted with the lack of these in the CtC of φACT.For the heteroatomic substituent CdO, the influenceof the additional nπ* state on the ππ* dynamics wasinvestigated by comparing BZA with STY and ACPwith R-MeSTY. The effects of vibrational dynamicsand densities of states on the electronic relaxationrates were studied via both methyl (“floppier”) andalkyl (“more rigid”) ring substitution: STY wascompared with both the floppier R-MeSTY and themuch more rigid IND; BZA was compared with thefloppier ACP. Both BZA and ACP have Type IKoopmans’ ionization correlations, and the rest haveType II correlations, allowing for another comparisonof these two cases. To investigate the potential effectsof autoionization resonances, these authors variedthe ionization probe photon energy significantly (∼0.4eV). In these systems, the form of the photoelectronspectra and the fits to the lifetime data were invari-ant with respect to probe laser frequency.

A sample TRPES spectrum, STY at λpump ) 254.3nm and λprobe ) 218.5 nm, is shown in Figure 9 (top).The S1(ππ*) component grows in rapidly, correspond-ing to the ultrafast internal conversion of the S2(ππ*)state. The S1(ππ*) component subsequently decayson a much longer ps time scale. It can be seen that,despite STY being an unfavorable Type II case, thetwo bands are well enough resolved to allow forunambiguous separation of the two channels anddetermination of the sequential electronic relaxationtime scales. Energy integration over each band allowsfor extraction of the electronic relaxation dynamics.The time-dependent S2(ππ*) 00 photoelectron bandintegral yields for STY are also shown in Figure 9(bottom, a). The open circles represent the pump-probe cross-correlation (i.e., the experimental timeresolution) at these wavelengths. Integration over theS2(ππ*) 00 photoelectron band is shown as the solidcircles. The solid line is the best fit to the S2(ππ*)channel, yielding a lifetime of 52 ( 5 fs. In Figure 9(bottom, b), the time-dependent S1(ππ*) photoelectronband integral yields for STY are shown, obtainedfrom a fit to the long delay part of the data, yieldinga lifetime of 88 ( 8 ps for the state S1(ππ*). Overall,these results and the results obtained for the otherfive substituted benzenes demonstrate that the TRP-ES method is well-suited to the quantitative studyof electronic relaxation processes, producing facile,direct, and accurate measurements of electronicrelaxation rates that are in quantitative agreementwith the currently accepted understanding of aro-matic photophysics.

In 2003, Suzuki and co-workers applied the time-resolved 2D imaging method to the study of py-ridazine excited-state dynamics.151 These authorspumped pyridazine to its S1 (nπ*) origin and probedvia resonant two-photon ionization, obtaining a life-time of 323 ps attributed to internal conversion.Measured lifetimes were a strong function of S1excess vibrational energy, decreasing from 323 ps atthe origin to 235, 181, and 123 ps at 300, 900, and1900 cm-1 excess energy, respectively. The photo-

electron imaging method allowed identification of theintermediate states resonances as 3s(n-1) and 3p(n-1)Rydberg series. In contrast with the earlier studiesof excited-state dynamics in pyrazine,145 pyridazineinterestingly showed no evidence of intersystemcrossing dynamics (i.e., no triplet-state formation).Analysis of the PADs indicated mostly p-wave emis-sion from the 3s state, whereas the 3p photoemissionwas comprised of a mixture of partial waves. Carefulanalysis determined the term values of the 3s and3p series to be 5.69 ( 0.03 and 6.28 ( 0.04 eV,respectively. This work demonstrates the quantita-tive accuracy of the photoelectron imaging methodin the study of excited Rydberg states in polyatomicmolecules.

Figure 9. (Top) TRPES spectrum shown for styrene (STY)with λpump ) 254.3 nm and λprobe ) 218.5 nm. The ener-getics and IP allow for assignment of the photoelectronbands to ionization of S2(ππ*) and S1(ππ*), as indicated.The S2 state decays on ultrafast time scales. The S1 statedecays on a much longer (ps) time scale. (Bottom) (a)Time-dependent S2(ππ*) 00 photoelectron band integralyields for STY, yielding a decay time constant of 52 fs. Opencircles represent the laser cross-correlation at these wave-lengths. (b) Time-dependent S1(ππ*) photoelectron bandintegral yields for STY obtained from a fit to the long timedelay part of the data (not shown). Reprinted with permis-sion from ref 150. Copyright 2002 American ChemicalSociety.


5.2. Excited-State Intramolecular Proton TransferExcited-state intramolecular proton transfer (ES-

IPT) processes are important for both practical andfundamental reasons.152 o-Hydroxybenzaldehyde(OHBA) is the simplest aromatic molecule displayingESIPT and serves as a model system amenable tocomparison with theory. Stolow and co-workers usedfs TRPES to study ESIPT in OHBA, monodeuteratedODBA, and an analogous two-ring system, hydroxy-acetonaphthone (HAN), as a function of pump laserwavelength, tuning over the entire enol S1(ππ*)absorption band of these molecules.153,154 The experi-mental scheme is depicted in Figure 10 (top), showingenergetics for the case of OHBA. Excitation with atunable pump laser hνpump forms the enol tautomerin the S1(ππ*) state. ESIPT leads to ultrafast popula-tion transfer from the S1 enol to the S1 keto tautomer.On a longer time scale, the S1 keto population decaysvia internal conversion to the ground state. Both theenol and keto excited-state populations are probedby ionization with a probe laser hνprobe, producing thetwo photoelectron bands ε1 and ε2.

In Figure 10 (bottom) are TRPES spectra of OHBAat an excitation wavelength of 326 nm. Two photo-electron bands, ε1 and ε2, with distinct dynamics wereobserved. Band ε1 is due to photoionization of theinitially populated S1 enol tautomer, and band ε2 isdue to the photoionization of the S1 keto tautomer.The decay of band ε1 yields an estimated upper limitof 50 fs for the lifetime of the S1 enol tautomer. Inexperiments with ODBA, no isotope effect was ob-served, suggesting that the barrier in the OH stretchcoordinate must be very small or nonexistent, con-sistent with ab initio calculations.155,156 As is commonin TRPES, these spectra also give insights into thedynamics on the “dark” S1 keto state. The picoseconddecay of band ε2 corresponds to S1 keto internalconversion to the ground state. The wavelength-dependent S1 keto internal conversion rates in OHBAand ODBA likewise revealed no significant isotopeeffect. The measured internal conversion rates arevery fast (1.6-6 ps over the range 286-346 nm),considering the large energy gap of 3.2 eV betweenthe ground and excited states. One possibility is thatfast internal conversion in such systems is due to anefficient conical intersection involving a πσ* state andlarge-amplitude hydroxy H-atom motion.155,156 Theobserved absence of an isotope effect on S1 ketointernal conversion in ODBA does not support thismechanism. Rather, the comparison with internalconversion rates in HAN suggests that interactionswith a nearby nπ* state may play an important rolein the keto internal conversion, as discussed in detailelsewhere.153 This example serves to illustrate howTRPES can be used to study the dynamics of biologi-cally relevant processes such as ESIPT and that itreveals details of both the proton-transfer step andthe subsequent dynamics in the “dark” state formedafter the proton transfer.

Related to ESIPT, intermolecular hydrogen-trans-fer reactions in ammonia157,158 and indole-ammo-nia159 clusters have been studied in detail by Radloffand co-workers using time-resolved PEPICO meth-ods. In a combined experimental-theoretical study

of the 200 nm excitation of the ammonia dimer,Radloff observed that the (NH3)2 A state evolves intothe NH4-NH2 complex via hydrogen transfer.157 Thetime-resolved photoelectron spectra coincident withthe dimer cation show a dramatic change from abroad (>1 eV) band at short times (100-500 fs) to amuch narrower and lower energy band by 2-4 ps.By contrast, the time-resolved photoelectron spectracoincident with the NH4

+ cation show broad features

Figure 10. (Top) Energetics for excited-state protontransfer in OHBA, showing the enol and keto forms.Excitation with a pump laser hνpump forms the enol tau-tomer in the S1(ππ*) state. ESIPT leads to ultrafastpopulation transfer from the S1 enol to the S1 keto tau-tomer. On a longer time scale, the keto S1 populationdecays via internal conversion to the keto ground state.Both the enol and keto excited-state populations are probedvia ionization with a probe laser hνprobe, producing the twophotoelectron bands ε1 and ε2. (Bottom) TRPES spectra ofOHBA at an excitation wavelength of 326 nm and a probelaser wavelength of 207 nm. Two photoelectron bands wereobserved: ε1 due to ionization of the S1 enol, and ε2 due toionization of the S1 keto. Band ε1 was observed only whenthe pump and probe laser beams overlapped in time,indicating a sub-50 fs time scale for the proton transfer.Band ε2 displayed a wavelength-dependent lifetime in thepicosecond range, corresponding to the energy-dependentinternal conversion rate of the dark S1 keto state formedby the proton transfer. Reprinted with permission from ref154. Copyright 2001 American Institute of Physics.


that do not evolve much with time. These wereattributed to fragmentation of vibrationally excited(NH3)2

+ cations and are therefore complementary tothe photoelectron spectra of the bound dimer cations.The results were interpreted in terms of a model inwhich the initially excited (NH3)2 A state proceedsvia N-N stretching to a conical intersection with thecharge-transfer state NH4

+NH2-, leading to NH4-

NH2 products. The narrowing of the photoelectronbands with time was attributed to the shorter pro-gressions expected for the NH4-NH2 f NH4

+NH2 +e- ionization channel, as opposed to the longerprogressions expected for the (NH3)2 A f (NH3)2

+ +e- channel.

Radloff and co-workers subsequently extendedtheir PEPICO studies to larger ammonia clusters aswell as deuterated clusters158 using a sophisticatedthree-pulse arrangement. The probing of the hydrogen-transfer state, prepared by 200 nm excitation, wasmediated by a near-infrared (832, 1200, or 1400 nm)control pulse which excited this state to a higherelectronically excited state, which could then beionized by a delayed fs 400 nm pulse. This helped todifferentiate the initially excited A state, whichdecays in 180-270 fs, from the longer-lived (ps)H-transfer state. The excited-state dynamics of the200 nm excited (NH3)3 A trimer was very differentfrom that of the dimer. Using an 832 nm controlpulse, the (NH3)NH4(3p)NH2 state was observed tohave a lifetime of 2.7 ps, which upon deuterationlengthened to 32 ps. The lifetime of the H-transferstate decreased dramatically with cluster size, di-minishing to ∼150 fs for the hexamer (NH3)6. Incontrast to the dimer, these results suggest a fastinternal conversion mechanism in which the (NH3)-NH4(3p)NH2 state decays to a lower-lying (NH3)NH4-(3s)NH2 state, which then rapidly dissociates.

Indole, the chromophore of tryptophan, is a veryimportant model system for studying biomoleculephotophysics. Calculations of the isolated moleculeand indole-water clusters suggest that a low-lyingπσ* state, populated via internal conversion from theinitially excited ππ* state, plays a crucial role in thephotophysics.160,161 Radloff and co-workers appliedthe time-resolved PEPICO method to study photo-induced H-transfer in indole-(NH3)n clusters.159 Com-plex dynamics involving internal conversion, hydro-gen transfer, rearrangement, and dissociation werediscerned, with different observations for small (n )1-3) as opposed to larger (n > 4) clusters. Inspiredby theory, an ultrafast ππ* f πσ* internal conversionmodel was used to interpret the results. For thesmaller clusters (n ) 1-3), the internal conversion(time scale of ∼150 fs) was followed by H-transfer,leading to a structural rearrangement on the tens tohundreds of ps time scale. A minor but competingdissociative channel operating on the 80-140 ps timescale was also observed. In the larger clusters (n >4), the initial ultrafast internal conversion was notseen, likely due to unfavorable Franck-Condon fac-tors. Nevertheless, structural rearrangements sub-sequent to H-transfer were observed. These resultsillustrate how time-resolved PEPICO can add avaluable new dimension to the study of complex

competing reaction and electronic relaxation chan-nels in molecular clusters.

5.3. Photophysics of Model Molecular SwitchesActive molecular-scale electronics9 involves the use

of molecules as transistors, switches, modulators, etc.and requires a dynamical process, which coupleschanges in optical or electrical properties to molec-ular rearrangements such as isomerization. As opti-cal or electrical properties are governed by theelectronic wave function and molecular rearrange-ments are vibrational in character, the underlyingphysical process in molecular electronic switches isthe non-adiabatic coupling of electronic with vibra-tional motions. It is the competition between fastmolecular electronic processesssome desired, somenotswhich will govern the efficiency and stability ofactive molecular-scale electronic devices, and there-fore the rational design of such devices should includeconsiderations of excited-state non-adiabatic pro-cesses.

Stolow and co-workers applied the TRPES methodto the study of electronic relaxation in trans-azoben-zene (AZ),162 often considered the prototypical fastmolecular electronic switch as its photoisomeriza-tion163 is the basis for numerous functional materials.Despite this great interest in azobenzene, the basicphotoisomerization mechanism remained disputed:164 in contrast to the expectations of Kasha’s rule, theisomerization quantum yields decrease rather thanincrease with increasing photon energy. In Figure 11(top), the two possible isomerization channels, pro-

Figure 11. (Top) Photoisomerization dynamics of trans-to cis-azobenzene, indicating torsional and inversion path-ways. (Bottom). TRPES spectra of trans-azobenzene excitedat 330 nm and probed at 207 nm. Two photoelectron bands,ε1 and ε2, were observed, having identical laser-limited risetimes but differing decay rates (τ1 ) 130 fs, τ2 ) 410 fs)and differing Koopmans’ ionization correlations. Theseresults indicate that there is a previously unrecognized ππ*state, S3, involved in the dynamics, suggesting a new modelof trans-azobenzene photoisomerization dynamics. Re-printed with permission from ref 162. Copyright 2003American Chemical Society.


ceeding via either a planar pathway (“inversion”) ora nonplanar, twisted pathway (“torsion”), are shown.Previous studies determined that isomerization inthe first excited state, S1, proceeds along the inver-sion coordinate.163 The second excited state, S2, isgenerally thought to be the NdN analogue of theCdC ππ* state in stilbene, and it is thought that,somehow, motion along the torsional coordinate inS2 is responsible for the observed reduction in isomer-ization yield. Recent time-resolved studies suggestedthat photoisomerization proceeds via the inversioncoordinate in S1. The role of the torsional isomeriza-tion pathway remains controversial. Theoretical stud-ies have supported both torsion and inversion path-ways but disagreed on the states involved in theexcited-state relaxation.

In Figure 11 (bottom), a time-resolved photoelec-tron spectrum for excitation of AZ to the origin of itsS2 state is shown.162 Two photoelectron bands, ε1 andε2, with differing lifetimes and differing Koopmans’correlations were observed. Due to these two differ-ences, the ε1 and ε2 bands must be understood asarising from the ionization of two different electronicstates. Furthermore, as both bands rise within thelaser cross-correlation, they are due to direct photo-excitation from S0 and not to secondary processes.Therefore, to account for different lifetimes, differentKoopmans’ correlations, and simultaneous excitationfrom S0, the existence of an additional state, S3, whichoverlaps spectroscopically with S2 must be invoked.162

According to the Koopmans’ analysis (based uponassignment of the photoelectron bands) and to high-level, large active space complete active space self-consistent field (CASSCF) calculations, this new stateS3 corresponds to ππ* excitation of the phenyl rings.162

The π* excitation in the ring does not directly “break”the NdN bond and therefore is expected to lead toreduced isomerization quantum yields. This mecha-nism differs greatly from that of all earlier modelsin that those always assumed that a single brightstate, the S2 (NdN ππ*) state, exists in this wave-length region.162 This example shows how TRPES canbe used to study competing electronic relaxationpathways in a model molecular switch, revealinghidden yet highly important electronic states that canbe very hard to discern via conventional means.

5.4. Superexcited StatesAs discussed above, for a single active electron,

photoionization transitions are restricted to molec-ular orbitals containing an electronic configurationdiffering only by a single-electron vacancy. In theouter valence region (i.e., within a few electronvoltsof the IP), two-electron transitions are rare but canoccur, usually weakly, because of electron correla-tion.23 This situation, however, can potentially changewhen an intermediate resonance is introduced. Anintermediate state can have very different ionizationcorrelations than the ground state. A possible out-come of this is the preparation of “superexcited”valence states: highly excited valence (as opposed toRydberg) states lying above the IP. From the pointof view of TRPES, superexcited states complicate theunderstanding of ionization correlations, PADs, and

Franck-Condon analysis. Therefore, the extent towhich such states contribute to the ionization dy-namics is worthy of consideration. In particular, in2001, P. Weber and co-workers investigated the roleof superexcited states in the fs multiphoton photo-electron spectroscopy of the S1 and S2 states ofphenol165,166 and the S2 state of 1,3-cyclohexadiene.167

As an illustration of effects due to superexcitedstates, Weber and co-workers excited phenol to theS1 level, whereupon a second photon excited it to atotal energy of 9 eV, about 0.5 eV above the IP.165 InFigure 12, a photoelectron spectrum is shown for fslaser ionization of phenol at 275 nm, via the S1 state.While the major features can be readily explainedin terms of the well-known two-photon photoelectronspectrum, a small peak is seen at higher electronkinetic energies due to three-photon ionization. We-ber and co-workers have examined this small three-photon signal in considerable detail for both phenoland cyclohexadiene. This signal is characterized interms of a superexcited valence state prepared by twopump laser photons. Superexcited states are oftenvery short-lived and can decay via autoionization,internal conversion to Rydberg series, and neutraldissociation. In these experiments, the superexcitedcomplex was ionized by a third time-delayed laserpulse. As long as autoionization is a minor channel,the two-photon photoelectron spectrum may be in-terpreted in the conventional manner. Analysis of thethree-photon spectrum requires, as Weber and co-workers show, considerable care. Their two-color,three-photon photoelectron spectra are composed ofmultiple overlapping components with varying life-times. Many of these were assigned to Rydberg series,and two clear series could be assigned to quantumdefects of 0.80 and 0.32, respectively, associated mostlikely with a ground-state core. The authors suggestthat the primary superexcited state is a valence statedue to the promotion of two electrons from theHOMO to the LUMO: in other words, the S1 statemust resonantly absorb the same photon as that usedfor the resonant S0 f S1 transition.

Figure 12. Photoelectron spectrum of phenol using 180fs, 275 nm laser pulses. Processes due to both two-photonand three-photon ionization are observed, the latter arisingfrom preparation and subsequent ionization of a superex-cited valence state. This decays on ultrafast time scales toRydberg manifolds, yielding characteristic structures. Re-printed with permission from ref 165. Copyright 2001American Chemical Society.


Weber subsequently studied superexcited states inphenol prepared via the S2 state.166 In this case,polarization and orbital energy studies suggest asuperexcited state also at 9 eV but with a configu-ration due to single-electron excitation of the [HO-MO - 2] orbital into the [LUMO + 1] orbital. As thesuperexcited states formed by S1 and S2 excitationwill be both of A1 symmetry, it is likely that thesetwo states will mix, thus forming a superexcitedvalence complex. In addition, Stolow and co-workersdetermined, via TRPES, the S2 lifetime of phenol tobe <50 fs,168 leading Weber to suggest that theadmixture of configurations in the superexcitedcomplex will vary with time delay, further complicat-ing the analysis.

In 2001, Weber and co-workers studied 9 eVsuperexcited states in 1,3-cyclohexadiene,167 preparedvia resonant two-color, three-photon ionization throughthe S2

1B2 state. The three-photon spectra revealedtwo prominent Rydberg series with quantum defectsof 0.58 and 0.98. In the analysis, the kinetic energyof the emitted electron was assumed independent ofthe vibrational excitation of the core, a reasonableassumption for high-lying Rydberg states. Weberconfirmed this by tuning over the broad S2 absorptionprofile in cyclohexadiene, showing that the Rydbergassignment was invariant with respect to the pumplaser frequency.

In 1999, Stolow and co-workers used the photoion-ization of superexcited states to help confirm timescales of non-adiabatic processes in all-trans-decatet-raene.39,41 In this case, 235 nm probe ionization ofthe 287 nm excited S2(1Bu) state was compared with352 nm probe ionization of this same state. In thesecases, the shape of the S2 f D0 photoelectron bandwas invariant, and identical decay times (400 fs)were extracted for the lifetime of the S2 state. In thecase of 352 nm ionization, however, a broad band athigher kinetic energies grew in with a time constantof 400 fs. This band is due to two-photon ionizationof the vibrationally excited S1(2 1Ag) state formedvia the 400 fs internal conversion. By contrast, notwo-photon ionization of the S2(1Bu) state was ob-served.

It is interesting to consider why, at invariant probelaser intensity, the 352 nm photoionization processswitches from single-photon ionization of S2 to two-photon ionization of S1: in both cases, the first 352nm photon is already above the ionization potential,and therefore the S1 ionization is due to absorptionof a second photon in the ionization continuum. Thisswitching can be rationalized through a considerationof the relative rates of two competing processes:second photon absorption vs direct or autoionization.For the case of S2, the Koopmans’ correlation is withD0, and therefore the ionization is direct. In otherwords, the ionization is extremely rapid, and secondphoton absorption cannot compete. For the case ofS1, the Koopmans’ correlation is with D1. The D1state, however, is one-photon energetically inacces-sible at 352 nm, and therefore the transition is intoa superexcited complex, likely including Rydbergseries converging on the D1 threshold. For these toemit an electron into the D0 continuum channel,

there must be an electronic rearrangement, whichwould have a finite autoionization rate (in fact, thisprocess was not observed, and the superexcited statelikely decays via neutral channels). In this case, itappears that the absorption of a second photoncompetes effectively with any autoionization. Thesetwo-photon experiments not only confirm the one-photon results but also demonstrate the symmetryselectivity of the photoionization process itself.

Perhaps the most general test available to experi-mentalists to prove that a TRPES spectrum is notadversely affected by the autoionization of superex-cited states is to considerably vary the probe photonfrequency, keeping the pump photon frequency fixed.Tuning across an autoionizing resonance is expectedto lead to significant changes in the form of thephotoelectron spectrum, precluding a simple Franck-Condon and Koopmans’-type analysis. In 2002, Stolowand co-workers investigated the potential role ofautoionization resonances through the study of aseries of six substituted benzenes, varying the probephoton wavelength from 219 to 208 nmsan over 2400cm-1 change in ionization energy.150 For example, theS2 00 lifetime of benzaldehyde was probed using thesetwo different probe wavelengths. Identical time con-stants of 440 ( 8 fs were obtained for both 208 and219 nm probes, suggesting that the ionization dy-namics are very likely direct and not significantlycontaminated by the autoionization of superexcitedstates in these systems.

Finally, we note that resonance phenomena in theionization continuum are likely to have a moresignificant effect on photoelectron angular distribu-tions than angle-integrated kinetic energy distribu-tions. An energy dependence of angular momentumbarriers to electron escape could lead to phase shiftsbetween partial waves, thus altering the angulardistributions while having minimal effects on thekinetic energy distributions.

5.5. Time-Resolved VUV/XUV/X-ray PhotoelectronSpectroscopy

The examples of TRPES discussed so far haveconsidered only outer valence shell photoionization.However, due to the development of fs high harmonicand soft X-ray sources, there is growing interest inthe development of time-resolved inner-shell photo-electron spectroscopies. These have been applied tosurface physics problems for some time.169

As opposed to using high harmonic radiation as aprobe, it is possible to use it as a pump of highlyexcited states in molecules. In 2000, Sorensen, Svens-son, L’Huillier, and co-workers used Ti:sapphire sixthharmonic radiation (9.42 eV) as the pump pulse in aTRPES study of the 1Σu

+ 3dπ3R′′′ν2 ) 1 Rydbergstates of C2H2.126 This was the first TRPES exampleof using high harmonic radiation to prepare a wavepacket in a very high lying electronic state of amolecule. These highly excited states are known topredissociate rapidly to C2H* + H. The TRPESmeasurements directly determined the lifetimes ofthe 3R′′′ states to be ∼150 fs.

In 2001, Leone and co-workers applied a time-resolved soft X-ray photoelectron spectroscopy tech-


nique to the dissociation dynamics of a diatomicmolecule: Br2.128 Using high harmonic generation inAr, these researchers generated fs pulses at the 17thharmonic (26.4 eV, 47 nm) of the 800 nm Ti:sapphirelaser, compressed them in a grazing double-gratingarrangement, and combined them with fs 400 nmpump pulses for a time-resolved study. The 400 nmpulse excited ground-state Br2(X 1Σg

+) to its dissocia-tive C 1Πu state, correlating to two ground-stateBr(2P3/2) atoms. The time-resolved formation ofBr(2P3/2) atoms was directly observed in the X-rayphotoelectron spectra, with an estimated dissociationtime of ∼40 fs. In this energy regime, the analysis iscomplicated by both multiple cross-correlation fea-tures, including ATI, excited-state molecular photo-ionization, and product-state photoionization as wellas the numerous final states of the ion that areenergetically accessible. The photoelectron spectrumof the recoiling Br atom did not appear to shift withtime delay. It was also observed that the atomicfragments had much larger (∼40×) photoionizationcross sections than the parent molecule. In 2002,these same researchers presented a more extensivestudy of time-resolved X-ray photoelectron spectros-copy of Br2 dissociation dynamics.129 In that case,they used five different high harmonics (13th, 15th,17th, 19th, and 21st) and were able to elucidateseveral of the competing photoionization channels.Transient features associated with the molecule inthe process of dissociation were directly observed.Interestingly, this feature seems to have enhancedintensity as compared with the molecular groundstate. As Leone and co-workers have shown, theability to easily change the X-ray photoelectronenergy (from one harmonic to the next) brings apowerful new tool into the TRPES arsenal.

In a recent 2003 study, V. Blanchet, S. Sorensen,D. Gauyacq, A. L’Huillier, and their co-workers usedhigh harmonic radiation to pump the high-lying F40

2

and E4-502 Rydberg states (∼2 eV below the ioniza-

tion potential) in C2H2 and C2D2 molecules.170 TheF 1Σu

+(3dπg) Rydberg state absorbs most strongly inthis region and exhibits symmetric stretching excita-tion of the C-C triple bond. In this experiment, theseresearchers doubled their high-power Ti:sapphirepulse to 396 nm and then used third harmonicgeneration of this in xenon to generate fs VUV pulsestunable around 132 nm (9.4 eV). The authors dis-cussed the role of laser chirp in TRPES studies. Anychirp in the high harmonic pulses means that thetotal photoelectron energy shifts with time, leadingto energy shifts of the photoelectron bands during thecross-correlation overlap of the laser pulses. Theseauthors made detailed studies of the chirp of theirhigh harmonic pulses and used this information todevelop a new method for extracting time- andenergy-resolved information about the dissociationdynamics. Using their time-energy scale method, thepredissociation lifetimes for the C2H2 Rydberg stateswere measured to range from 20 to 200 fs, and forC2D2 from 150 to 600 fs. It would not be possible toextract information about the time scales of themolecular dynamics without a full analysis of thechirp. This study both serves as a warning and

indicates the potential opportunity for the use of laserchirp in TRPES.

5.6. Neutral Photodissociation DynamicsThe most interesting case for photochemistry is

that of unbound excited states or excited statescoupled to a dissociative continuum. The dissociativeelectronically excited states of polyatomic moleculescan exhibit very complex dynamics, usually includingnon-adiabatic processes. As illustrated by Blanchetand Stolow in 1998 in a study of the NO dimer,77

TRPES may be used to study the photodissociationdynamics of neutral polyatomic molecules.

In 1999, Hayden, Continetti, and co-workers ap-plied the newly developed coincidence-imaging spec-troscopy (CIS) technique to the photodissociationdynamics of NO2.115 The CIS method allows forenergy- and angle-resolved (i.e., full 3D momentumvector imaging) detection of both electrons and ionsin coincidence and represents the most differentialTRPES measurements made to date. These authorsstudied the dissociative multiphoton ionization ofNO2 at 375.3 nm. This was identified as a three-photon transition to a repulsive surface, correlatingwith NO(C 2Π) + O(3P) fragments. The NO(C) wassubsequently ionizing by a single photon, yieldingNO+(X 1Σ+). As a first illustration of the multiplydifferential information obtained via CIS, Haydenand co-workers used energy-energy correlations toplot photoelectron kinetic energy vs NO(C) photo-fragment kinetic energy, as a function of time, asshown in Figure 13. At early times, 0 and 350 fs,there is a negative correlation between electron and

Figure 13. Coincidence-imaging spectroscopy of dissocia-tive multiphoton ionization processes in NO2 throughenergy-energy correlations using ∼100 fs laser pulses at375.3 nm. The two-dimensional maps show, at time delaysof 0 fs, 350 fs, 500 fs, and 10 ps, the correlation betweenthe photoelectron kinetic energy (abscissa) and NO photo-fragment recoil energy (ordinate). The intensity distribu-tions change from a negative correlation at early times touncorrelated at later times, yielding information about themolecule as it dissociates. Reprinted with permission fromref 115. Copyright 1999 American Institute of Physics.


fragment recoil energy. This is the form expected fora molecule in the process of dissociating, where thereis a tradeoff between ionization energy and fragmentrecoil energy. At longer time delays, 500 fs and 10ps, the NO(C) fragment is no longer recoiling fromthe O atomsit is a free particlesand the photoelec-tron spectrum obtained is simply that of free NO(C);hence, the negative correlation vanishes.

In a subsequent study, Hayden, Continetti, and co-workers again applied the CIS method to the samemultiphoton NO2* f NO(C) + O process, but thistime making use of angle-angle correlations.116 Bymeasuring the angle of recoil of both photoelectronand photofragment in coincidence, the photoelectronangular distribution (PAD) may be transformed intothe photofragment recoil frame, for each recoil angleand time delay. As an example, in Figure 14, thetime-resolved recoil-frame PAD is shown for the caseof photofragments ejected parallel to the laser polar-ization axis. It can be seen that at early times, 0 and350 fs, the PAD is highly asymmetric. The breakingof forward-backward symmetry in the recoil frameoriginates from NO(C) polarization due to the pres-ence of the O atom from which it is recoiling. At

longer times, 1 and 10 ps, this forward-backwardasymmetry vanishes as the NO(C) becomes a freeparticle. This once again shows the power of CIS inobtaining highly detailed information about mol-ecules in the process of dissociating.

In 1999, Radloff and co-workers studied the 267nm photodissociation dynamics of CF2I2,171 using thetime-resolved PEPICO method. CF2I2 has a broadand featureless absorption spectrum in the 230-400nm region due to the overlap of several dissociativecontinua. Analysis of the photoelectron spectra re-vealed that three electronic states are involved andthat an ultrafast non-adiabatic process (∼30 fs)precedes dissociation (∼100 fs). The dissociationproducts were determined to be CF2 + I + I, as noCF2I fragments or photoelectron spectra were ob-served. Due to the intensity of the pump laser pulses,dissociative multiphoton processes involving both twoand three pump photons were also observed. At thetwo-photon level, direct formation of molecular I2 wasobserved, consistent with earlier work. At the three-photon level, highly excited I(nd) atom channels wereobserved in the photoelectron spectra. The coincidentdetection of photofragments and photoelectrons wascrucial to disentangling these various processes.

In 2001, Radloff and co-workers applied the time-resolved PEPICO method to the photodissociation ofClO2 using very short (∼50 fs) pump pulses at 398nm.172 This reaction had previously been studiedusing TRPES by Chen and co-workers at 386 nm.53

At 398 nm, two channels are open:

The yield of channel B is ∼4% at 398 nm (3.11 eV).At wavelengths below 399.2 nm (3.1 eV), channel Ais directly open. In addition, an ultrafast spin-orbitinduced internal conversion A 2A2 f A 2A1 process isallowed. The A 2A1 state can either dissociate viachannel A or undergo further internal conversion tothe 2B2 state. This latter state may then dissociatethrough both channels A and B. Using the time-resolved photoelectron spectra coincident with eachof OClO+, ClO+, O2

+, and Cl+ ions, Radloff and co-workers were able to unravel the complex dissocia-tion dynamics of this molecule, using various mul-tiphoton dissociation and ionization processes. Thespin-orbit-induced A 2A2 f A 2A1 transition wasdetermined to have a time constant of 7 ps. Thenascent A 2A1 state decayed by both dissociation andinternal conversion, with an overall time constant of∼250 fs. The 2B2 state formed by the subsequentinternal conversion then decayed on a 750 fs timescale. These latter two channels were unresolved inthe earlier TRPES study by Chen.53 This once againillustrates the great utility of time-resolved photo-electron-photoion coincidence spectroscopy in thestudy of complex reaction mechanisms.

Figure 14. Coincidence-imaging spectroscopy of dissocia-tive multiphoton ionization processes in NO2 with ∼100 fslaser pulses at 375.3 nm, using angle-angle correlations.The polar plots show, at time delays of 0 fs, 350 fs, 500 fs,1 ps, and 10 ps, the angular correlation between the ejectedelectron and NO photofragment when the latter is ejectedparallel to the laser field polarization vector. The intensitydistributions change from a forward-backward asym-metric distribution at early times to a symmetric angulardistribution at later times, yielding detailed informationabout the molecule as it dissociates. Reprinted withpermission from ref 116. Copyright 2000 American PhysicalSociety.

(A) OClO(X) 98hν1

OClO(A 2A2) f

ClO(X 2Π) + O(3P)

(B) OClO(X) 98hν1

OClO(A 2A2) f

Cl(2P) + O2(3Σg

-, 1∆g,1Σg

+)


In 2002, V. Engel and co-workers proposed the useof TRPES to study the complex photodissociationdynamics of the metal carbonyl Fe(CO)5.173,174 A long-standing problem in the study of metal carbonyl UVphotochemistry is the competition between neutraldissociation followed by ionization and the dissocia-tive ionization of both parent molecule and photo-fragments. As both types of processes produce frag-ment ions with broad and featureless kinetic energydistributions, it is very difficult to discern them viaion detection alone. Engel shows via calculation that,due to the differing ionization potentials of the Fe-(CO)n (n ) 2-4) fragments, shifts in the photoelec-tron spectra are expected as a function of time. Thisshould likely assist the disentangling of the variousionization/fragmentation channels, allowing for directmeasurements of the neutral dissociation channels.

In 2002, Meier and Engel used both classicaltrajectory and quantum wave packet calculations tomodel the direct dissociation of the H2O (A 1B1) stateprobed by fs TRPES.31 The authors calculated that,with a 6 fs VUV pump pulse at 155 nm, the dissocia-tion dynamics of H2O (A 1B1) is revealed directly byTRPES, using a 6 fs XUV probe pulse at 86.7 nm(14.3 eV). As the H atom separated from OH, thephotoelectron spectra shifted, in 10-20 fs, from abroad peak at 0.4 eV to a sharper peak at 0.1 eV.Importantly, the time-resolved photoelectron spec-trum could be calculated classically via an extensionof the impulse approximation. This points the waytoward future applications of classical mechanics inTRPES calculations in larger systems where quan-tum or semiclassical calculations are prohibitivelyexpensive.

Following the earlier study of 210 nm photodisso-ciation dynamics of the (NO)2 dimer,77 Hayden,Stolow, Shaffer, and co-workers applied the CISmethod to this same problem. Initial results usingboth energy- and angle-resolved correlation mapsrevealed new details of the dissociation dynamics.175

The previously proposed two-step non-adiabatic pho-todissociation mechanism at 210 nm was indepen-dently supported by these results. Suzuki and co-workers have further studied this system via time-resolved charged particle imaging toward deconvolvingnon-adiabatic and photodissociation dynamics. Aninitial study,176 performed with a 200.5 nm pumpwavelength, compared the valence-state decay life-time determined from time-dependent (NO)2

+ pho-toion intensity with (NO)2 3s Rydberg state popula-tion growth monitored via time-resolved photoelectronimaging. These time scales were found to be com-mensurate at ∼200 fs. The dimer Rydberg stateyields a more energetic and anisotropic photoelectronangular distribution relative to the valence state andcorrelates to the NO(X) + NO(A) photodissociationchannel. A series of time-resolved pump-probe ex-periments were performed with 200.5 nm pump and250, 270, and 290 nm probe pulses toward character-izing non-adiabatic decay of the initially excitedvalence state and the dissociation of the Rydbergstate through changes in measured PADs. Thoughthe dimer Rydberg and NO(A) states yield nearlyisoenergetic photoionization features, a subtle shift

of ionization energy, coupled to a change in aniso-tropy of this peak, indicated a ∼1 ps time scale forreaching the free NO(A) asymptote through dissocia-tion. In a subsequent study,177 these researchersextended investigation of the dimer valence-stateelectronic dephasing rate compared to the dimer 3sRydberg state growth following excitation at variouswavelengths between 200 and 235 nm. These timescales were found to match for excitation wavelengthsbelow 220 nm and indicated a lifetime contractionwith decreasing pump wavelength. While an effectivephotodissociation lifetime additionally depends on thedimer Rydberg lifetime, these results corroboratephotofragment imaging studies reported by Demy-anenko et al.178 that find higher fragment anisotropyfor higher excitation energies in this region, sug-gested to result from shortened dimer lifetimes.

5.7. Neutral Reaction DynamicsThe study of bimolecular reaction dynamics via the

femtosecond pump-probe technique is challenged bythe broad radial distribution function, g(r), found in(almost) any gas. Even if a bimolecular reaction isphotoinitiated by a short-pulse laser, this distributionof nearest-neighbor distances means that the reactivecollisions will occur over a broad range of times,completely blurring the short time scale of thecollision dynamics itself. The solution to this paradoxwas presented by Zewail and co-workers in 1987 ina seminal study of the H + CO2 reaction,179 namelyto use van der Waals clusters to hold reactants (inthis case HI and CO2) in close proximity and in awell-defined geometry, thus forcing a single “collisiontime”.

The applications of femtosecond TRPES to bimo-lecular reaction dynamics are still in their infancy,having been first demonstrated by Radloff and co-workers for the Ba + CH3F reaction using thePEPICO method.180 In this experiment, the Ba‚‚‚FCH3 van der Waals complex was the reactant state.Pumped at hν1 ) 618 nm, the Ba‚‚‚FCH3 complex isexcited to the A(1E) state, correlating to the Ba 1P(6s6p) state. Reactions proceed and were probed (hν2) 400 nm) as follows:

The A 1E f A′ 1A1 internal conversion (II) wasdirectly observed via TRPES (via bands ε1 and ε2),yielding a time constant of 270 fs. The harpooningreaction (III) occurred rapidly (∼50 fs), followinginternal conversion. At longer time delays (severalps), only photoelectron spectra of the BaF product(band ε3) were observed. This study lays the ground-

Ba‚‚‚FCH3 98hν1

Ba*‚‚‚FCH3 A(1E) 98hν2

Ba‚‚‚FCH3+ + e-(ε1) (I)

Ba*‚‚‚FCH3 A(1E) f Ba*‚‚‚FCH3 A′ (1A1) 98hν2

Ba‚‚‚FCH3+ + e-(ε2) [f BaF+ + CH3] (II)

Ba*‚‚‚FCH3 A(1E) f BaF + CH3 98hν2

BaF+ + CH3 + e-(ε3) (III)


work for future applications of TRPES in neutralclusters to the study of bimolecular reactions dynam-ics, a subject which is already well-established intime-resolved studies of anionic clusters (vide infra).

5.8. Anion TRPES: I2- Nuclear Wave PacketDynamics

As mentioned in the Introduction, anion TRPESallows one to track both excited- and ground-statedynamics, in contrast to most neutral experimentsthat are restricted to excited-state studies. Thecapabilities of anion TRPES are illustrated throughreference to discussion of diiodide anion (I2

-) dynam-ics. I2

- has many desirable properties from theperspective of anion TRPES experiments. Its con-stituent atoms are sufficiently heavy that its nucleardynamics can be followed with ∼100 fs laser pulses.It has two strong electronic transitions, the A′ 2Πg,1/2

r X 2Σu+ and B 2Σg

+ r X 2Σu+ bands, that may be

accessed using the fundamental (∼790 nm) andsecond harmonic (∼395 nm) of a Ti:sapphire femto-second laser. Finally, the photodetachment crosssections of I2

- and I- are quite large. For thesereasons, I2

- is an excellent model system for TRPESas well as several variations of the basic experiment,which are depicted pictorially in Figure 15 and areexplained in context below.

The first anion TRPES experiment probed thedissociation dynamics of I2

- excited to its A′ 2Πg,1/2state, schematically depicted in Figure 15a.36 I2

- wasexcited to this dissociative excited state by a broad-band ∼80 fs pulse at 780 nm, launching a wavepacket that propagates toward separated I + I-

products, and was subsequently probed by photode-tachment with a 260 nm pulse of ∼100 fs duration.Only the “atomic” I- f I(2P3/2)/I*(2P1/2) + e- detach-ment features were observed at the latest pump-probe delays (>425 fs), while additional transients,arising from detachment to the A′/A and B states ofI2, were observed between 0 and 200 fs at lowerbinding energies than the atomic features. Quantummechanical simulations of the experiment were un-dertaken using a wave packet propagation scheme29

and the best available potential energy curves for I2-

and I2. Experiment and simulation indicated dis-sociation by 100 fs.

A subsequent study,37 performed at higher resolu-tion and simulated with a more accurate experimen-tally derived anion ground-state potential,181 pro-duced an accurate potential for the A′ 2Πg,1/2 dissocia-tive state. These higher-resolution TRPE spectra aredisplayed in Figure 2. Here, a long-time shift of ∼10meV in the position of the nominal I- f I(2P3/2) + e-

feature was observed, revealing a long-range attrac-tive well in the dissociative state arising from charge-induced dipole, charge-induced quadrupole, anddispersion interactions. The excited A′ state wasempirically characterized to have a 17 ( 10 meV wellat 6.2 ( 0.6 Å by fitting to a simulation performedwith an efficient wave packet propagation scheme.182

While TRPES experiments nominally provide in-formation on the dynamics of the excited stateaccessed by the pump pulse, variations on thisexperiment can probe the ground state, generally by

producing coherent vibrational wave packet motionon the ground state that can be followed by TRPES,as depicted schematically in Figure 15b. In the firstexample of this type, the pump pulse used to excitethe A′ 2Πg,1/2 r X 2Σu

+ transition in I2- was also

shown to create a coherent superposition of the v )0 and v ) 1 vibrational levels on the ground X 2Σu

+

state, thereby creating a wave packet whose motioncould be detected by oscillations in the TRPE spec-trum at selected electron kinetic energies.181 Thisground-state excitation process, resonant impulsivestimulated Raman scattering (RISRS), could then beused to obtain the fundamental vibrational frequencyof I2

-, 110 cm-1.

Figure 15. Experimental versatility of TRPES applied toanion systems exemplified with I2

-. (a) Femtosecond pho-toelectron spectroscopy (FPES), whereby ultrafast (fs)dissociation dynamics initiated by a ∼780 nm pump pulse(A′ 2Πg,1/2 r X 2Σu

+) are subsequently probed via ∼260 nmdetachment at variable delays. (b) Resonant impulsivestimulated Raman scattering (RISRS) FPES, in whichground-state vibrational dynamics, initiated by scattering,are probed via FPES. (c) Femtosecond stimulated emissionpumping (FSEP) FPES, in which ground-state wave pack-ets are generated at various average vibrational quantathrough a ∼800 nm pump-near-IR dump process and aresubsequently probed with FPES.


In a more sophisticated ground-state wave packetexcitation scheme, femtosecond stimulated emissionpumping (FSEP),183 femtosecond pump and dumppulses create a ground-state wave packet with aver-age vibrational energy Eexc ) hνpump - hνdump (Figure15c). The pump pulse is resonant with the A′ 2Πg,1/2

r X 2Σu+ transition, while the dump pulse, applied

50-150 fs after the pump pulse, drives some of thedissociating excited-state wave packet back to theground state. By tuning the dump pulse, Eexc wasvaried from 0.262 eV up to 0.993 eV; the latter valueis within 2% of the I2

- bond dissociation energy. Wavepacket oscillations were monitored via detachmentwith a third time-delayed femtosecond pulse (UV),producing high-energy electrons from detachment tothe neutral ground state as the packet reaches theinner turning point (ITP) of the anion potential. ITPdetachment signal oscillations arising from I2

- vi-brational wave packet motion at low, intermediate,and high excitation energies are shown in Figure 16.Fourier transformation of these oscillations yieldedthe I2

- vibrational frequency as a function of Eexc,while wave packet rephasing times as a function ofEexc probed the anharmonicity of the ground-statepotential. The frequency and anharmonicity informa-tion from the RISRS and FSEP experiments wasused to construct an accurate I2

- ground-state po-tential that was improved relative to previouslycalculated curves.

TRPES, RISRS, and FSEP experiments have allbeen applied to clustered I2

- as well as the bareanion. Results from these experiments are describedin subsequent sections.

5.9. Dissociation Dynamics of Polyatomic AnionsExperiments with triatomics present the added

complication of energy-dependent ion-neutral prod-uct branching ratios. In such a pump-probe experi-ment, Gantefor and co-workers75 examined Au3

-

dissociation from 0 to 3600 ps following ∼3.0 eVexcitation. Photodetachment signal measured atearly delays was attributed to an activated complexpopulated following nuclear relaxation from theinitially excited configuration. This complex wasdetermined to decay subsequently toward separatedproducts with a lifetime of 1500 ( 200 ps and withpreferential production of the lowest energy productchannel, Au- + Au2. Pumping at 3.14 eV resulted ina significantly shorter-lived complex (<50 ps) with afactor of 2.5 increase in the Au2

-:Au- product ratio.Time-resolved PES of I3

-, photodissociated at 390nm, revealed both I- and I2

- photofragments,184 incontrast to results of similar experiments done in thesolution phase.185,186 The TRPE spectra showed dis-tinguishable contributions corresponding to depletionof I3

- reactant and growth of I- and I2- products.

With aid from two-dimensional wave packet simula-tions, the I- photoproduct was determined to arisefrom three-body dissociation along the symmetricstretch of I3

-. The I2- products were formed in a

coherent superposition of vibrational levels, observ-able through oscillations in the TRPE spectra. Simi-lar effects were seen in the solution-phase photodis-sociation of I3

- using transient absorption, but muchgreater vibrational excitation was seen in the gas-phase experiments. The I3

- features in the TRPEspectra showed oscillatory structure from RISRSexcitation of the symmetric stretch, yielding a vibra-tional frequency of 112 ( 1 cm-1.

5.10. Photodissociation, Recombination, andReaction Dynamics in Anionic Clusters

TRPES experiments on size-selected clusters inwhich an anion chromophore is solvated with aknown number of solvent species offer a detailedprobe of how clustering affects the dynamics ofrelatively simple systems. The pioneering work byLineberger and co-workers187-191 on I2

-(CO2)n andI2

-(Ar)n clusters, in which the I2- chromophore was

photodissociated and the masses of the resulting iondaughter fragments were measured, showed pro-found effects of clustering on the I2

- dissociationdynamics. The recoiling photofragments interact withthe surrounding solvent species, resulting in a size-dependent yield of caged daughter ions from solvent-induced I + I- recombination. Time-resolved experi-ments by Lineberger, in which the absorption recoverytime of the I2

- was measured, yielded overall timeconstants for recombination and vibrational relax-ation of the clustered I2

-. These aspects of thedynamics have been explored by Parson and Cokerin a series of theoretical simulations.192-197 TRPESexperiments on these clusters offer a more completepicture of their dynamics subsequent to photoexci-tation.

Greenblatt et al. investigated the TRPE spectra ofI2

-(Ar)ne20 and I2-(CO2)ne16 clusters in a series of

Figure 16. Vibrational wave packet dynamics measuredon the I2

- ground state (X 2Σu+) through FSEP-FPES.

Vibrational wave packets are generated in the anionground state through FSEP (Figure 15c) at various averagevibrational energies (0.262, 0.705, 0.965 eV) to within2% of the dissociation limit. Wave packet dynamics, probedby FPES, are monitored by oscillation of high eKE photo-electron signal, arising from detachment at the innerturning point (ITP) of the anion potential well. Oscillationperiods reveal average vibrational spacings within eachexcitation window, while recurrences reflect the anharmo-nicity of the diatomic potential. Reprinted with permis-sion from ref 183. Copyright 2000 American Institute ofPhysics.


papers.198-201 The maximum sizes in both casesrepresent complete solvent shells. In all the clusters,dissociation of the I2

- chromophore subsequent toexcitation of the A′ 2Πg,1/2 r X 2Σu

+ occurs on a timescale similar to that of bare I2

-, 200-300 fs, resultingin the production of solvated I- within the cluster.The extent of solvation at these short times, beforeany solvent rearrangement or other dynamics haveoccurred, can be determined from the positions of thefeatures in the PE spectra and confirms the anoma-lous charge-switching nature of the A′ 2Πg,1/2 statepredicted by Parson and co-workers,197 in which theelectron is localized on the less-solvated iodine atom.In small clusters (e9 Ar atoms, or e6 CO2 molecules),where no recombination occurs, the TRPE spectrashow relatively small shifts, reflecting solvent rear-rangement dynamics as the I and I- fragmentsseparate.

In the larger Ar clusters, recombination of I2- is

observed on the X and A 2Πg,3/2 states. The PE spectrawere analyzed to yield the number of solvent atomsremaining as a function of time for both states of therecombined I2

-, and the extent of vibrational relax-ation for clusters with I2

- in the X state. In I2-(Ar)20,

it was found that energy transfer to the solvent atomsvia vibrational relaxation was slightly faster (∼100ps) than energy loss from the cluster through Arevaporation, indicating temporary storage of energywithin the Ar cluster modes. Vibrational relaxationof the recombined I2

- in the larger I2-(CO2)n clusters

was found to be significantly faster (<5 ps) than inI2

-(Ar)20, in agreement with the absorption recoveryexperiments by Lineberger.190 However, solvent evapo-ration was clearly much slower, and comparison tothe tandem mass spectrometric data obtained byLineberger189 showed that evaporation was incom-plete for pump-probe delay times as long as 200 ps.The faster relaxation and slower evaporation timesare consistent with the stronger solute-solvent cou-pling and higher solvent density of states in I2

-(CO2)nclusters.

The effects of clustering on dissociation and recom-bination dynamics have also been studied in a seriesof TRPES experiments by Zewail and co-workers202-204

on molecular oxygen cluster anions, (O2-)ng3, and

clusters in which O6- is solvated by another species

(N2, Xe, N2O). Previous work by Hiraoka205 andJohnson206 showed that (O2)n

- clusters with n > 2have an intact O4

- core. Johnson found that the n >2 clusters undergo photodissociation in the near-infrared, a process attributed to charge transfer fromthe O4

- core to the O2 solvent(s). Zewail’s experimentsshowed that excitation of pure or mixed clusters at800 nm resulted in biexponential growth of the O2

-

photoproduct. The fast process (τ1 ) 100-500 fs) wasassigned to back electron transfer to the O4 core,followed by dissociation on an optically inaccessiblerepulsive state, while the slower process (τ2 ≈ 1-10ps) was attributed to vibrational predissociation of avibrationally excited O2

- solvent molecule formed bythe initial charge-transfer step. τ2 generally increasedwith cluster size, consistent with a statistical evapo-ration process. In the study of mixed clusters, it wasfound that dissociation via vibrational predissociation

was more prevalent for O6-(N2O) than for any other

comparably sized cluster, and that the time scale forthis channel was much slower, e.g., 7.7 ps vs 2.1 psfor O6

-(Xe). This result was attributed to rapid V-Venergy transfer from the transient O2

- solvent mol-ecule to the N2O.

The largest oxygen clusters showed evidence of apreviously unseen phenomenon in clusters: cage-induced recombination of molecular photofragments.204

As shown in Figure 17, TRPE spectra of (O2)n-

clusters with n > 7 exhibited recovery at highelectron binding energy at times intermediate (2-4ps) to electron transfer/direct dissociation and slowvibrational relaxation. This shift was assigned tocage-induced recombination of O2 and O2

- to formvibrationally excited O4

-, followed by O2 evaporation.The slow time scale of this molecule-molecule cagingrelative to the atom-atom recombination seen inclustered I2

- was attributed to the need for properorientation of the two molecular species for cagingto occur.

In a recent variation on the clustered I2- TRPE

experiments, motivated by neutral studies of bimo-lecular reactions utilizing cluster-constrained geom-etries, Wester et al.207 investigated the reactiondynamics of the symmetric SN2 reaction of I- withCH3I by TRPES of the precursor complex I2

-‚CH3I.The reaction was initiated by photodissociating theI2

- chromophore with a pump pulse at either 790 or395 nm. Immediately after the pump pulse, thereactants formed an I-‚CH3I complex, with an inter-

Figure 17. Time-resolved measurements of (O2)n-, n )

6-10, photofragment photoelectron spectral peak posi-tion: Circles, experiment; line, fit. Recovery at high bindingenergies, arising from O4

- photodetachment, for clustersn > 7 indicates O2-O2

- recombination with increasedcluster size. Reprinted with permission from ref 204.Copyright 2003 American Institute of Physics.


nal energy dependent on the energy of the pumpphoton. For 790 nm pump photons, only stablecomplexes were observed, while at 395 nm, abouttwo-thirds of the complexes decayed back to thereactants, I- + CH3I, on the time scale of theexperiment (<100 ps). These unstable complexesexhibit biexponential dissociation dynamics: 80%decayed with a fast time constant of about 0.75 psand 20% with a slow time constant of about 10 ps.This biexponential decay and the observed shift ofthe I-CH3I photoelectron peak during a time spanof about 5 ps are indicative of “apparent non-RRKMbehavior” of the complex, reflecting the evolution ofvibrational energy flow in the complex on a time scalecomparable to or slower than that of dissociation. Theobserved dynamics are similar to those seen inclassical trajectory calculations on entrance channelcomplexes formed in bimolecular Cl- + CH3Cl colli-sions.

5.11. Anion Vibrational Wave Packet andRelaxation Dynamics

Just as the TRPES experiments on the photodis-sociation of bare I2

- provide a basis for understandingthe clustered I2

- experiments described in the previ-ous section, RISRS and FSEP experiments on I2

- laythe groundwork for understanding vibrational wavepacket spectroscopy and dynamics in clusters of I2

-.Experiments in which vibrational wave packets aregenerated in clusters probe the conditions underwhich coherent wave packet motion can be detected,and how the populations and phases initially encodedinto the wave packet evolve with time, therebyproviding new insight into spectroscopy and relax-ation dynamics in clusters.

For example, Zanni et al.208 used RISRS to examinesolvent effects on the I2

- fundamental vibrationalfrequency. For I2

-(Ar)n clusters, no spectral shiftswere observed for n e 6, while the n ) 12 and 18clusters showed blue shifts of 1.5 and 3 cm-1,respectively. The clusters I2

-(CO2)n, n ) 4, 9, showedblue shifts of 2 and 4.5 cm-1, respectively. Hence, onlyblue shifts were observed, and these increased withthe number of solvent species and strength of thechromophore-solvent interaction.

Such explorations were subsequently expanded tofurther examine solvent-chromophore interactionsand solvent mediation of vibrational relaxation byexciting higher in the potential well with FSEP. Aseries of papers by Davis et al.209-211 explored dy-namics in I2

-(X)n clusters (X ) Ar and CO2) subse-quent to coherent vibrational excitation of the I2

-

chromophore at a series of excitation energies Eexc.The relaxation dynamics of the I2

- were followedusing two observables in the TRPE spectra. At shorttimes, wave packet recurrences at the inner turningpoint of the I2

- potential resulted in oscillations inthe signal at high electron kinetic energy, just as inbare I2

-. However, dephasing was more rapid andirreversible in the clusters (3-4 ps in I2

-(CO2)4,5,8-10 ps in I2

-(Ar)ne12). More significantly, the oscil-lation frequency increased with time. The oscillationfrequency reflects the spacing between adjacentvibrational energy levels, so the increase with time

is a measure of vibrational relaxation of the I2- in

an anharmonic potential, where the level spacingincreases as the vibrational quantum number drops.The increase in frequency is as large as 20 cm-1 forI2

-(CO2)4 at Eexc ) 0.57 eV; this implies that the wavepacket loses at least 0.35 eV of vibrational energy in∼3.5 ps while maintaining its coherence.

Once dephasing of the oscillations has occurred, theabove method can no longer be used. However, overa much longer time scale, the maximum eKE atwhich signal was observed shifted to lower energy,and this effect was used to follow vibrational relax-ation dynamics for times as long as 200 ps. Expo-nential fits yielded relaxation time constants τ ) 4-5ps for I2

-(CO2)4,5. Relaxation times for I2-(Ar)n clus-

ters were generally larger (e.g., τ ) 27 ps for I2-(Ar)9

at Eexc ) 0.57 eV), but showed much more variationon cluster size and excitation energy. This variationreflects the similar time scales for vibrational energyflow and solvent evaporation in I2

-(Ar)n clusters, asopposed to I2

-(CO2)n clusters, where vibrationalenergy flow is much faster than solvent evaporation.

5.12. Charge-Transfer-to-Solvent and SolvatedElectron Dynamics

Solvated halide anions have been known to exhibitbroad ultraviolet absorption bands arising from ejec-tion of excess electrons into the surrounding solvent,the charge-transfer-to-solvent (CTTS) bands.212-214

The dynamics of electron solvation subsequent toCTTS excitation is an active area in condensed-phasedynamics.215-218 Remarkably, features analogous tothe bulk CTTS bands have been observed in the pho-todetachment spectroscopy of relatively small clus-ters; Serxner et al.219 observed them in I-(H2O)n)2-4,while Cheshnovsky220,221 observed much narrower butanalogous features in I-(Xe)4-54 clusters. These re-sults motivated a series of TRPES studies of theCTTS excited states in solvated halide clusters,222-226

with the goal of understanding how an inherentlybulk phenomenon, electron solvation, manifests itselfin finite clusters.

The first such studies, by Lehr et al.,222,227 exam-ined CTTS precursors of I-(H2O)n and I-(D2O)nclusters, n ) 4-6. In these experiments, CTTSprecursor states were prepared by ∼100 fs UV laserpulses and subsequently probed by detachment atvarious delays at 790 nm. Significantly differentdynamics were observed within this size range. TheTRPE spectra for the n ) 4 clusters were indicativeof simple population decay of the CTTS state viaautodetachment. However, significant spectral shiftstoward lower eKE (or toward higher vertical bindingenergy, VBE) were seen in the TRPE spectra of thelarger clusters on a time scale of several hundred fs,as depicted in Figure 18, and these shifts wereattributed to stabilization of the diffuse electron inthe CTTS excited state by reorganization of the watermolecules.

An alternative interpretation of these experimentswas proposed by Chen and Sheu,228 based on elec-tronic structure calculations229 that indicated a re-pulsive interaction between the CTTS electron andthe I atom. They proposed that the spectral shifts


resulted from the I atom leaving the cluster ratherthan solvent reorganization. Davis et al.226 pointedout that this interpretation was difficult to reconcilewith the difference between the n ) 4 and largerclusters, and with the slightly faster dynamics inclusters of H2O vs D2O. Calculations by Vila andJordan230 provided further support for solvent reor-ganization immediately after CTTS excitation, whilecalculations by Timerghazin and Peslherbe231 onI-(H2O)3 indicate both solvent network dynamics andelongation of the iodine-to-solvent distance in sub-sequent relaxation. At this point, the detailed dy-namics of these clusters remain somewhat of an openquestion, but probably involve a combination ofsolvent dynamics and iodine motion.

CTTS clusters composed of other solvents withvarying polarities were investigated by Frischkornet al.225 and Davis et al.,224 revealing different dy-namics for various solvent types. Similar to hydratedclusters, CTTS-excited I-(NH3)n and I-(CH3OH)nexhibit a time-dependent shift in electron binding

energy, concomitant with an increase in electrondetachment cross section, associated with partialsolvation and a subsequent localization of the diffuseexcess electron. In contrast to I-(H2O)n, these clustersexhibit monotonically increasing VBE shifts withincreasing solvation rather than a size-critical VBEshift. I-(CH3OH)n clusters exhibited spectral evolu-tion similar to that of I-(NH3)n, indicating partialsolvation of the extra electron within a few picosec-onds, though this is followed by a long-time shift tolower binding energies, suggesting the possibility ofsolvent evaporation or the contribution of two solventconfigurations to the observed dynamics.

TRPES experiments on the CTTS states of I-(Xe)nclusters were performed by Zanni et al.223 for clusterswith as many as 38 Xe atoms. Two sets of transitionswere studied, corresponding to excitation of [I*(2P1/2)-Xen]- and [I*(2P1/2)Xen]- CTTS states. The [I(2P3/2)Xen]-

states decayed on a time scale of 500-1500 fs andshowed no VBE shifts, indicating no solvent rear-rangement. The decay was attributed to spin-orbit-induced autodetachment ([I(2P1/2)Xen]- f I(2P3/2)Xen+ e-); longer lifetimes for the larger clusters impliedsolvent screening of the delocalized excited electronfrom the spin-orbit-excited iodine core. The lower[I(2P3/2)Xen]- CTTS state (n ) 6-13), previouslysuggested to be bound with respect to detach-ment,220,221,232,233 exhibited no appreciable decay outto hundreds of picoseconds.

Another key aspect of the solvated electron problemis the dynamics of electronically excited hydratedelectrons. In aqueous solution, hydrated electronsexhibit a broad absorption band around 700 nm,234

and the development of a complete and coherentmodel describing the dynamics associated with thisband has been the subject of a vast experimental andtheoretical literature over the years.235,236 Theseconsiderations have motivated many investigationsof water cluster anions, (H2O)n

-, focusing on theirenergetics and spectroscopy.237-239 Weber et al.240

have performed one-color, two-photon photoelectronspectroscopy at 800 nm on (H2O)n

- anions with asmany as 100 water molecules. They observed a bandin the PE spectrum that was clearly attributable toresonant excitation of the excited electronic state inthese clusters, which is the cluster analogue to thehydrated electron excited state. As seen in Figure 19,this band “tunes” to the excitation wavelength withincreasing cluster size. Preliminary one-color pump-probe experiments at 800 nm showed the lifetime ofthis state to be less than 150 fs, the cross-correlationof the laser pulses. These experiments nicely comple-ment the solvated halide experiments describedabove, with the overall body of work providingconsiderable insight into the dynamics of diffuseelectrons in clusters.

5.13. Anion Intramolecular Electronic RelaxationDynamics

As discussed previously in this article, one of themost general applications of TRPES is to probe thedynamics of radiationless transitions, because of itsunique ability to track the lifetimes and relaxationpathways of excited electronic states in molecules.

Figure 18. Femtosecond photoelectron spectra of I-(D2O)n,n ) 4-6 (a-c), and I-(H2O)5 (d) plotted as 2D contour plotsof electron kinetic energy (eKE) with respect to pump-probe delay. Dramatic spectral shifts for clusters with n >4 suggest solvent reorganization dynamics following CTTSexcitation. Reprinted with permission from Science (http://www.aaas.org), ref 222. Copyright 1999 American Associa-tion for the Advancement of Science.


Most TRPES experiments on neutrals have been ofthis type, as are increasing numbers in anion TRPESexperiments. The major constraint in anions is therequirement of an excited electronic state below thedetachment continuum, or a sufficiently long-livedstate above the continuum with dynamics that aredistinguishable from direct detachment (such as anautodetaching state). Both classes of excited statesare generally less prevalent in anions than in neu-trals, owing to the relatively low detachment energiesand absence of Rydberg states in anions. On the otherhand, the low detachment energies enable one totrack radiationless transition pathways all the wayto the ground electronic state of the anion, whereasin neutral TRPES experiments one is generallyconfined to excited-state processes because the ion-ization potential of the ground state is too high.

Cn- clusters have been known to support multiple

bound, electronically excited states. Frequency-resolved absorption experiments of these anions inrare-gas matrices and through gas-phase resonantmultiphoton detachment (RMPD) have mapped ex-cited states of these anions, as needed for this classof TRPES experiment. Minemoto et al.241 studied thedecay of the carbon trimer anion electronically ex-cited to an above-threshold Feshbach resonance firstobserved in photodetachment cross-section measure-

ments.242 The 2∆u state of C3- at 3.07 eV was pumped

and probed at variable delay with the second har-monic of a fs Ti:sapphire amplifier such that thenature of the participating electronic states could begleaned from the evolving spectrum. A dominantfeature was assigned to a combination of direct one-photon C3

1Σg+ r C3

- 2Πu detachment and competingautodetachment from the resonantly excited state.Time-dependent peaks were assigned to direct de-tachment of the excited resonance to its 3Πu parentneutral state and a “shake-down” detachment result-ing in a nominally one-electron disallowed C3

1Σg+ r

C3- 2∆u detachment. Signal from direct detachment

of the excited state was determined to decay in 2.6( 0.7 ps.

A similar study by Frischkorn et al.243 examinedirreversible electronic relaxation in C6

- from a boundexcited electronic state to the manifold of lower-lyingstates. Here, the C 2Πg r X 2Πu origin at 2.04 eV waspumped and UV-probed at various delays to inves-tigate decay to lower-lying electronic states. Decayof the fastest electron signal occurred on a time scaleof 730 ( 50 fs, while intermediate features wereobserved to grow in on a comparable time scale andsubsequently decay on a 3.0 ( 0.1 ps time scale to afinal static signal. This temporal progression wasinterpreted, guided by the best available neutral andanion energetics, as arising from sequential internalconversion from the initially excited C state throughthe A and/or B states. The energy and shape of thestatic long-time feature, accompanied by depletionrecovery in the UV one-photon spectrum, suggestthat relaxation ends with a highly vibrationallyexcited anion in its ground electronic state.

The two studies above involve energy flow betweena sparse manifold of discrete electronic states. Incontrast, transition metal clusters have a muchhigher density of electronic states, leading to thepossibility that TRPES studies of such clusters willreveal the types of fast electronic relaxation processesobserved in bulk metals, described better by inelasticelectron-electron scattering244,245 than by transi-tions between discrete electronic states. These ideashave motivated a series of TRPES experiments ontransition metal cluster anions by Pontius and co-workers.246-250

For example, Pt3- was studied with TRPES using

pump and probe pulses with photon energies of 1.5eV.248 Near time zero, two-photon photoemission(2PPE) closely matches the contours of a one-photonphotoelectron spectrum taken at 3.0 eV, reflecting arelatively constant density of states among unoc-cupied levels within the excitation window. Spectrameasured at various delays from 0 fs to 500 ps exhibita decay of structured features at low binding energy,concomitant with a buildup of structureless signalat higher binding energies. This decay (<70 fs) wasassigned to inelastic electron-electron scatteringprocesses in which the initially excited state(s)transfer population/energy to a many-electron excitedstate of higher binding energy in a fast cascade.

In a similar study,250 Ni3- was pumped and probed

with 3.0 eV photons; fast decay of the lowest bindingenergies, arising from de-excitation of the initially

Figure 19. Resonant two-photon detachment photoelec-tron spectroscopy (R2PD-PES) of (H2O)n

- clusters (n ) 20-100) with ∼100 fs laser pulses. Absorption tunes to the∼1.5 eV (800 nm) excitation energy with increasing clustersize. Electron kinetic energy ranges associated with one-and two-photon absorption are labeled I and II, respec-tively. Direct two-photon detachment represents the clusteranalogue of excitation to the bulk conduction band of waterand competes with excited-state cluster relaxation. Re-printed with permission from ref 240. Copyright 2001Elsevier.


populated levels, was modeled by the optical Blochequations for a two-level system convoluted with theprobing response of the experiment. A mean inelastice-e scattering rate of 215 ( 50 fs was determined,while long-time evolution of the total 2PPE was fitwith an exponential decay with a constant of 450 (150 fs in order to estimate a mean electronic-vibrational coupling time scale.

The effect of size variation249 on inelastic electronic-electronic and electronic-vibrational scattering wasexamined with Pdn

-. The results of this study aresummarized in Figure 20, which depicts both the(normalized) time-evolving photoelectron spectra ob-tained for each cluster and the fit high-energy andtotal signal intensities that reflect electronic-elec-tronic and electronic-nuclear relaxation pathways.Electronic relaxation time scales for Pd3

-, Pd4-, and

Pd7- are 42 ( 19, 91 ( 17, and 25 ( 14 fs; the faster

electronic relaxation of Pd7- with respect to the

smaller clusters is in agreement with Fermi liquidtheory251 in view of the determined extended densityof unoccupied levels, whereas the faster relaxationof Pd3

- with respect to Pd4- is attributed to intricacies

of electronic structure for these species. Electronic-vibrational relaxation decay constants of 0.7 ( 0.3and 1.0 ( 0.3 ps were determined for Pd7

- and Pd4-,

with the faster time scale for the larger cluster owingto an increased vibrational density of states.

The similarity of these time scales with thosemeasured for the relaxation of mesoscopic and bulktransition metals244, 245 suggests that the high elec-tronic level density makes the largest contributionto these decay processes. In contrast, alkali and noblemetal clusters, with much lower level densities, arepostulated to have much longer decay lifetimes,owing to a lowered probability for electron-electronscattering. Recent experiments by Gantefor and co-workers252,253 on the electronic relaxation of excitedsp metal clusters Au3

-, Au6-, and Aln

- (n ) 6-15)reveal varied behavior relative to the above-men-tioned transition metal clusters. Aln

- clusters, withpn+1 configurations, were observed to decay on timescales 2-3 times longer (200-500 fs) than those ofPt3

- and Pd3-7-, though on time scales comparable

to that of Ni3-, consistent with the lower density of

electronic states in sp metals. However, rapid relax-ation of the “magic” cluster Al13

-, which has a 1.5 eVHOMO-LUMO gap, suggests the participation of aneffective relaxation pathway independent of elec-tronic structure. In contrast, Au3

- and Au6- (sn+1)

exhibit relaxation time scales between hundreds ofps to 1 ns, indicative of drastically reduced electronicdensities of states. This has been explained in termsof increased interatomic interaction between Auatoms relative to other metals due to relativisticeffects that increase molecular orbital splittings andbond enthalpies.

5.14. Photoelectron Angular DistributionsProgress in the study of time-resolved photoelec-

tron angular distributions (PADs) has been reviewedrecently.13,14,17 In the following, we briefly discussdevelopments since 1999.

Beginning in 1999, Seideman and Althorpe imple-mented a general nonperturbative theory26 for the

calculation of time-resolved PADs in a study ofmolecular axis alignment in nitric oxide.58 As dis-cussed above, short laser pulses are inherentlyintense, leading to nonperturbative effects such asmolecular axis alignment due to rotational Rabicycling. These authors show that PADs in particularare quite sensitive to these, and therefore a nonper-turbative theory could be required in the detailedcomparison of experiment with theory. The use ofeven stronger pump laser fields to significantlyenhance the laboratory-frame axis alignment can beincluded in a natural way. These authors also showthat the PADs are much less sensitive to the probelaser intensity, as Rabi cycling against the ionizationcontinuum is very difficult.

In 1999-2000, V. McKoy, K. Takatsuka, and co-workers carried out sophisticated calculations on theTRPES and time-resolved PADs of Na2 in which theexact photoionization transition amplitudes wereexplicitly calculated, within the Golden Rule pertur-bative limit, as a function of internuclear dis-tance.254-256 The molecules were assumed to be ro-tationless and “fixed in space” (i.e., molecular frame).In their calculation, the intermediate excited statewas the 2 1Σu

+ state of Na2, a double-minimum statein which the character of the electronic wave functionchanges significantly on either side of the barrier.These authors show that this affects both the pho-toelectron energy and angular distributions and thatthe assumption of a constant transition dipole forphotoionization or photodetachment is problematicwhen the state created by the pump pulse undergoeslarge-amplitude nuclear motion or, for that matter,dissociation. Baumert and co-workers257 experimen-tally verified this idea with the 2 1Σu

+ state of Na2,demonstrating a 4-fold increase in the photoioniza-tion probability from 3 to 9 Å within the double-minimum potential by comparing and varying anR-independent wave packet simulation to time-de-pendent angle-averaged photoelectron intensitiesarising from ionization to the bound 1 2Σg

+ and re-pulsive 1 2Σu

+ states of Na2+. In 2000, McKoy and co-

workers expanded their theoretical studies by apply-ing their formalism to the exact (but rotationless,fixed-in-space) calculation of TRPES and time-resolved PADs in the ionic-covalent curve-crossingproblem in NaI.258

In 1999, K. L. Reid and co-workers259 presented aps time-resolved PAD study of the S1 state of p-difluorobenzene (p-DFB). In cases of favorable ioniza-tion dynamics and in the absence of electronicallynon-adiabatic processes, PADs can be a direct moni-tor of the evolution of molecular axis distributionsin the laboratory frame. These authors demonstratethat the PADs can be sensitive to rotationally medi-ated IVR, which leads to “irregular” rotation, ascompared to a rigid rotor. Rotations about the a axisare expected to have recurrences about every 280 ps,in comparison to those about the b and c axes, aboutevery 44 ps. Significant deviations from rigid rotorbehavior were observed, indicative of IVR in the S1state.

Theoretical work by Reid and Underwood60 on theextraction of molecular axis alignment from the


evolution of time-resolved PADs was presented in2000. If the ionization dynamics remain invariantover the course of this evolution (i.e., the radial dipolematrix elements are constant), then it is possible toextract the rotational wave packet evolution from thePADs. Of particular interest to experimentalists,measurement of the PADs in two polarization geom-etries allows extraction of a quantity directly relatedto the laboratory-frame alignment. As an illustration,these authors calculate the time-resolved PADsexpected for pure rotational dynamics in an idealsymmetric top (as a model of p-DFB).

The relationship between the molecular and labo-ratory frames in PAD studies requires consideration.In the laboratory frame (for isotropic samples), theangular anisotropy is limited by the angular momen-tum of the photons. Specifically, the laboratory-framePAD of an unoriented molecule produced via absorp-tion of n photons, followed by ionization/detachmentwith m photons with arbitrary polarizations, may beexpressed with the angular expansion

in which L must be even, YLM(θ,φ) are sphericalharmonics, and BLM are the angular moments takingnonzero values with M e 2n and 2n + 2m e 2lmax,where lmax is the maximum angular momentumcomponent of the molecular frame PAD.17 Conse-quently, only B20 (∝â2) and B40 (∝â4) are relevant to1 + 1′ time-resolved laboratory-frame PAD studiesperformed with parallel linearly polarized light. Thisrestriction does not occur in the molecular frame,which can exhibit rich angular structures due to thehighly multipolar nature of the ion core. Generally,the molecular-frame PAD is most sensitive to anyintramolecular dynamics. Underwood and Reid con-sidered the ability of laboratory-frame PAD measure-ments to extract information about intramoleculardynamics.260 In their study, these authors consideredthe case of C3v molecules and showed how the PADdepends on the nature of the laboratory-frame align-ment created by the pump laser pulse. Specifically,different excited-state electronic symmetries yielddifferent laboratory-frame PADs, reaffirming theutility of this method for studying electronically non-adiabatic processes.

Beginning in 2000, Seideman and co-workers ap-plied their general formalism of time-resolved PADsto polyatomic molecules.134,261 Using a model of theS2

1Bu f S1 2 1Ag internal conversion in octatetraene(OT, C2h), they show that the photoelectron angulardistributions are very sensitive to changes in theshape and symmetry of the excited-state electronicwave function associated with the non-adiabaticdynamics and could therefore provide complementaryinformation to energy-resolved, angle-integrated mea-surements, especially in the case of Type II ionizationcorrelations.42,134 In the case of OT, the PADs re-quired the inclusion of partial waves up to l ) 6 or 7in order to be converged to within 10%. The asym-metry parameters for single-photon ionization of theexcited S2 and S1 states into the ground electronic

Figure 20. (a) Pump-probe photoelectron difference spec-troscopy of Pdn

- clusters (n ) 3, 4, 7) taken with ∼80 fs, 1.5eV laser pulses. Spectra have been normalized to totalintensity to display relative intensity evolution and have beenreferenced to excitation energy in excess of the cluster HOMO(1.5, 1.45, and 1.9 for 3, 4, and 7, respectively). (b) (Left)Integrated two-photon photodetachment intensity from 0.8 to1.5 eV (n ) 3, 4) or from 1.1 to 1.6 eV (n ) 7) above the HOMOvs pump-probe delay. Electronic relaxation within these clus-ters is seen to occur on a time scale comparable to the laserpulse duration, with the dashed curve in the n ) 4 plot indi-cating the autocorrelation calculated for 80 fs pulses. (Right)Integrated signal between 0.0 and 1.5 eV above the HOMO(all clusters) plotted versus pump-probe delay; exponentialfits for n ) 4, 7 at delays >100 fs give the electronic-nuclearrelaxation rates reported in the text. Reprinted with permis-sion from ref 249. Copyright 2000 American Institute of Physics.

I(θ,φ) ∝ ∑L)0

2n+2m

∑M)-L

L

BLMYLM(θ,φ) (5)


state of the cation were found to differ significantly,suggesting that PADs should be able to directlymonitor the non-adiabatic transition. Seideman andAlthorpe subsequently investigated in more detail theeffects of coherent rotational motion262 and centrifu-gal rotation-vibration coupling59 on the form of thetime-resolved PADs. Importantly. the nuclear coor-dinate dependence of the electronic bound-free tran-sition amplitudes did not obscure the effects ofrotation-vibration coupling on the PADs. A completediscussion of the formal nonperturbative theory forcalculating time-resolved PADs in polyatomic sys-tems, including some numerical simulations, waspresented by Seideman in 2001.263

In 2002-2003, Seideman and co-workers extendedtheir theoretical methods for calculating time-re-solved PADs to multivibrational-mode, polyatomicsystems. Their method takes rotations into exactaccount, treats the fields nonperturbatively, andcomputes the electronic dynamics from first prin-ciples using density functional theory combined witha spline function basis set.27,28 Importantly, thisapproximate method for the electronic structurecalculation should be scalable to larger systems. Asa first application, they studied the ultrafast S2 fS1 internal conversion dynamics of pyrazine, tocompare with the original 1991 calculations of Dom-cke.24,25 These authors observed that the energy-resolved asymmetry parameters do indeed map outthe electronic population dynamics, filtering out thefast vibrational motions that complicate the energy-resolved angle-integrated photoelectron spectra. Theyalso note, however, that the photoelectron energydependence of the ionization matrix elements leadsto significant and nontrivial effects on the observedasymmetry parameters. This indicates that energy-integrated angle-resolved probes will likely not trans-parently map out the electronic population dynamics.The use of filtering of wave packet signals in orderto extract information about intramolecular couplingtime scales was previously considered by Seideman,Stolow, and co-workers,264 and PADs provide anexample of this.

The development of photoelectron imaging tech-niques (section 4) has greatly facilitated the study oftime-dependent PADs. As discussed in section 5.1,in 1999 T. Suzuki and co-workers applied 2D time-resolved photoelectron imaging (TRPEI) to the studyof intersystem crossing (ISC) in the pyrazine S11B3u (nπ*) and S2

1B2u (ππ*) states.135,136 The â2anisotropy parameters at 60 ps delay were measuredto be 1.1 and 1.7 for the S1 and T1 states, respectively.The full analysis of the PADs could be complicatedby the expected energy dependence of the ionizationamplitudes for π electron ejection265 and by interme-diate resonances in the two-photon ionization scheme.The authors report that no dramatic evolution in theform of the PADs was observed in these experiments.

In 2001, Suzuki and co-workers applied theirTRPEI technique to the study of excited-state rota-tional wave packet dynamics in the pyrazine S11B3u (nπ*) state.266 Using a two-photon ionizationscheme, the authors were able to probe the evolvinglaboratory-frame S1 molecular axis distribution

through both a 3s (parallel transition) and a 3p(perpendicular transition) Rydberg state. As shownin Figure 21, beautiful rotational revivals are seenin both the signal intensity and asymmetry param-eters. The atomic-like 3s orbital produces predomi-nantly pπ-type electrons. The 3p ionization channel,by contrast, requires outgoing partial waves of agsymmetry. Interestingly, rotational revivals were alsoseen in the triplet manifold ionization channel. Thereduced intensity in the triplet revival structure wasattributed to the much larger number of energy levels(∼20) in the superposition after the intersystemcrossing. Using single-probe photon ionization (200nm), a characteristic 4-fold symmetry was seen in thePADs near t ) 0, which vanished by ∼3 ps, fromwhich time a rather featureless PAD without revivalstructures was observed. A more complete discussionof the ionization dynamics of the pyrazine 3s and 3pRydberg states followed in 2001.145

As discussed in section 5.6, in 2000 Hayden andco-workers used the 3D coincidence-imaging spec-

Figure 21. Time-resolved photoelectron imaging of rota-tional wave packet dynamics in pyrazine excited states,pumped at 323 nm and probed via two-photon ionizationat 401 nm. (a) Inverse Abel-transformed image of photo-electrons from pyrazine [1 + 2′] ionization at a time delayof 30 ps. (b) Time evolution of the intensity of the photo-electron spectra corresponding to the S1 state via a 3sresonance, the S1 state via a 3p resonance, and the tripletstate. (c) Time evolution of the normalized â20 asymmetryparameter for photoelectron spectra corresponding to theS1 state via a 3s resonance, the S1 state via a 3p resonance,and the triplet state. Rotational revivals are seen in boththe intensities and asymmetry parameters. The solid linesare fits to the rotational coherence. Reprinted from ref 266.Copyright 2001 American Physical Society.


troscopy (CIS) method to observe the evolution ofmolecular-frame PADs during the photodissociationof NO2.116 In 2002, Hayden and co-workers studiedthe S2

1E′ f S11A1′ internal conversion in triethyl-

enediamine (DABCO) using the CIS method. Thiswork represents the current state of the art instudying intramolecular dynamics via time-resolvedPADs. As shown in Figure 22, dramatic evolution ofthe PADs is seen for both the decaying S2 state (top)and the forming S1 state (bottom). When the probelaser is parallel to the pump laser (a), the initial S2state PAD is dumb-bell-shaped. For perpendicularionization (b), it is a quadrupolar structure. Both ofthese types of PADs evolve in time, reflecting theintramolecular dynamics. It is interesting to note thatthe quadrupolar structure seen in the S2 PADs usingperpendicular polarizations would not be observableusing the 2D photoelectron imaging technique. To-gether with the non-cylindrically symmetric PADsobserved in NO2 photodissociation, this work il-lustrates the technical advantages of 3D over 2Dparticle imaging methods.

The first anion TRPEI experiments were reportedby Davis et al.267 in 2003, in which I2

- dissociationalong the A′ 2Πg,1/2 state following absorption at793 nm was examined. With a dissociation lifetimewell-established from previous FPES studies, I2

-

presented a useful standard for this new application.Time-resolved radially-integrated PES reflect thesame dynamics as observed in the previous studies,and as the anion dissociation occurs on a time scalemuch shorter than molecular rotation, changes in the

PAD anisotropy as a function of pump-probe delayreflect evolution of the electronic structure throughdissociation. The leading anisotropy â2 drops signifi-cantly from 0 to 250 fs, corresponding to the timeinterval in the electron kinetic energy spectrum inwhich the I- photoproduct becomes evident. However,â2 continues to evolve at delay times as large as 700-800 fs, at which point the photofragments are sepa-rated by more than 10 Å, while at longer times (>1ps) the anisotropy matches that of I- (-0.520 (0.035), arising from s- and d-wave interference. Thechanges in â2 around 700-800 fs were attributed toan electric field-induced mixing between the A′ andX electronic states, with g and u symmetry, respec-tively, at large internuclear separations where thetwo states are nearly degenerate. This mixing leadsto a change in the character of the orbital in whichthe electron resides, from an orbital which has equalamplitude on two weakly interacting I atoms to onewhich is localized on a single I atom, so that photo-detachment occurs from truly separated I + I-

products.In another time-resolved anion photoelectron im-

aging experiment, similar in vein to Suzuki’s pyra-zine experiments,135 subtle temporal evolution in thephotoelectron anisotropy following detachment fromthe long-lived B 2Σg

+ state of C2- was examined by [1

+ 1′] pump-probe photodetachment.268 C2- PADs

and anisotropy modulations are depicted in Figure23. The PADs show a full revival at ∼8900 fs,consistent with the well-known rotational constant

Figure 22. Time-resolved photoelectron angular distributions (PADs) from fs ionization of DABCO. Here, the opticallybright S2

1E′ state internally converts to the dark S11A1′ state on ps time scales. (a) PADs at 200 fs time delay for pump

and probe polarization vectors, both parallel to the spectrometer axis. The difference in electronic symmetry between S2and S1 leads to significant changes in the form of the PAD. (b) PADs at 200 fs time delay for pump polarization paralleland probe polarization perpendicular to the spectrometer axis, showing the effects of laboratory-frame molecular alignment.(c,d) The PADs evolve as a function of time due to molecular axis rotational wave packet dynamics. Used with permissionfrom C. C. Hayden, unpublished.


for the B state of C2- (1.87 cm-1). This experiment is

the first application of rotational coherence spectros-copy to an anion. Furthermore, qualitative analysisof the anisotropy oscillation in reference to molecularalignment following a parallel transition in a simplediatomic, guided by ideas put forth by Sanov,102

Reed,269 and Reid and Underwood,60 provides someinsight into the molecular-frame detachment process,indicating a detachment wave with larger amplitudeperpendicular rather than parallel to the molecularaxis.

6. Conclusions and Future Directions

We have presented a comprehensive review oftime-resolved photoelectron spectroscopy as appliedto the dynamics of gas-phase neutral and anionicmolecules and clusters. TRPES and its variants arepowerful, relatively new techniques that lend them-selves to highly detailed time-resolved studies ofinter- and intramolecular dynamics. We have at-tempted to describe the utility of this technique forstudying a variety of fundamental photochemical andphotophysical processes. TRPES directly probes theevolution of nonstationary states (wave packets) viaprojection onto the “universal” final state of theelectronic continuum. With appropriate choice ofphotoionization/photodetachment wavelength, TRP-ES is capable of revealing electronic and nucleardynamics and their couplings, including dynamics in“dark” states, thus permitting identification of allinvolved states and time scales. The signature ofrotational wave packet dynamics associated withtime-evolving molecular axis alignment distributionsand the elucidation of vector correlations have beenobserved through angularly resolved variants ofTRPES. To date, TRPES has been applied to animpressive range of photophysical and photochemicalproblems.

Future applications of this technique will benefitfrom ongoing developments in detector technologies,femtosecond and attosecond laser sources, and non-

linear optical frequency conversion schemes as wellas synchrotron and free electron laser acceleratorphysics. These developments will push the scope ofTRPES research to include molecular-frame mea-surements, photofragment-photoelectron scalar andvector correlations, extreme time scales, and inner-shell dynamics. For example, the use of shaped,intense nonresonant laser fields to create field-freealignment in polyatomic systems270,271 will combinewith 2D and 3D imaging variants of TRPES to probemolecular-frame dynamics. Further development ofthe multiply differential photoelectron-photofrag-ment coincidence and coincidence-imaging methodswill permit highly detailed investigation of statisticaland nonstatistical photoinduced charge and energyflow, an area of fundamental dynamical interest andof interest in applications to the gas-phase photo-physics of biomolecules.272 The development of highaverage power femtosecond VUV/XUV sources andthe dawn of attosecond science present the possibilityof probing highly excited states, core dynamics, andelectron correlation in real time. Parallel theoreticaldevelopments in ab initio molecular dynamics meth-ods for studying non-adiabatic processes in poly-atomic molecules/anions and in new methods forcalculating photoionization/photodetachment differ-ential cross sections will play an increasingly impor-tant role in the future of TRPES. All of these areasof research present significant new experimental andtheoretical challenges, and we anticipate excitingdevelopments and contributions from many in thecoming years.

7. Acknowledgments

A.S. thanks co-workers Prof. I. Fischer (Wurzburg),Dr. M. J. J. Vrakking (FOM, Amsterdam), Dr. V.Blanchet (Toulouse), Dr. S. Lochbrunner (Munchen),Dr. M. Schmitt (Wurzburg), Prof. J. P. Shaffer(Oklahoma), Dr. J. J. Larsen (Aarhus), Dr. D. M.Villeneuve (NRC), Dr. T. Schultz (Max-Born-Insti-tute, Berlin), Dr. J. G. Underwood (Open University

Figure 23. (Top left) One-photon C2- photodetachment PAD at 264 nm. (Bottom left) Two-color [1 + 1′]C2

- photodetachmentPAD (541 + 264 nm) measured with a ∼100 ps pump-probe delay. (Right) Second- and fourth-order anisotropy momentsmeasured as a function of pump-probe delay. Solid curves are quantitative fits to a calculated molecular alignment function.Reprinted with permission from ref 268. Copyright 2003 Elsevier.


UK), Prof. I.-C. Chen (Tsing-Hua), Dr. S.-H. Lee(Tsing-Hua), Dr. S. Ullrich (NRC), and especially Dr.M. Z. Zgierski (NRC). A.S. also thanks Prof. K.Muller-Dethlefs (York) for discussions on ZEKEphotoelectron spectroscopy, Prof. C. A. de Lange(Amsterdam) for discussions on magnetic bottle pho-toelectron spectroscopy, Dr. C. C. Hayden (SandiaNational Laboratories) for ongoing collaborations onfs time-resolved 3D coincidence-imaging spectros-copy, Prof. T. Seideman (Northwestern) and Prof. K.L. Reid (Nottingham) for many enlightening discus-sions on photoelectron angular distribution measure-ments, and Prof. W. Domcke (Munchen), Prof. M.Olivucci (Siena), and Prof. T. Martinez (Illinois) fornumerous helpful discussions on non-adiabatic dy-namics in polyatomic molecules. A.E.B. and D.M.N.gratefully acknowledge support from the NationalScience Foundation under Grant No. CHE-0092574and thank their many co-workers on the anion femto-second photoelectron spectroscopy project in theirlaboratory: B. Jefferys Greenblatt, Martin Zanni,Alison Davis, Christian Frischkorn, Rainer Weinkauf,Leo Lehr, Benoit Soep, Mohammed Elhanine, ArthurBragg, Roland Wester, Aster Kammrath, and JanVerlet.

8. References(1) Zewail, A. H. J. Phys. Chem. A 2000, 104, 5660.(2) Bixon, M.; Jortner, J. J. Chem. Phys. 1968, 48, 715.(3) Jortner, J.; Rice, S. A.; Hochstrasser, R. M. Adv. Photochem.

1969, 7, 149.(4) Henry, S. R.; Siebrand, W. In Organic Molecular Photophysics;

Birks, J. B., Ed.; Wiley: London, 1973; Vol. 1, p 152.(5) Freed, K. F. In Radiationless Processes in Molecules and

Condensed Phases; Fong, F. K., Ed.; Springer-Verlag: Berlin,1976; p 23.

(6) Stock, G.; Domcke, W. Adv. Phys. Chem. 1997, 100, 1.(7) Michl, J.; Bonacic-Koutecky, V. Electronic Aspects of Organic

Photochemistry; Wiley: New York, 1990.(8) Schoenlein, R. W.; Peteanu, L. A.; Mathies, R. A.; Shank, C. V.

Science 1991, 254, 412.(9) Jortner, J.; Ratner, M. A. Molecular Electronics; IUPAC, Black-

well: Oxford, 1997.(10) Haight, R.; Bokor, J.; Stark, J.; Storz, R. H.; Freeman, R. R.;

Bucksbaum, P. H. Phys. Rev. Lett. 1985, 54, 1302.(11) Hayden, C. C.; Stolow, A. In Advanced Physical Chemistry; Ng,

C.-Y., Ed.; World Scientific: Singapore, 2000; p 10.(12) Neumark, D. M. Annu. Rev. Phys. Chem. 2001, 52, 255.(13) Suzuki, T.; Whitaker, B. J. Int. Rev. Phys. Chem. 2001, 20, 313.(14) Seideman, T. Annu. Rev. Phys. Chem. 2002, 53, 41.(15) Stolow, A. Annu. Rev. Phys. Chem. 2003, 54, 89.(16) Stolow, A. Int. Rev. Phys. Chem. 2003, 22, 377.(17) Reid, K. L. Annu. Rev. Phys. Chem. 2003, 54, 397.(18) Pedersen, S.; Herek, J. L.; Zewail, A. H. Science 1994, 266, 1359.(19) Steadman, J.; Syage, J. A. Rev. Sci. Instrum. 1993, 64, 3094.(20) Kruit, P.; Read, F. H. J. Phys. E: Sci. Instrum. 1983, 16, 313.(21) Cheshnovsky, O.; Yang, S. H.; Pettiette, C. L.; Craycraft, M. J.;

Smalley, R. E. Rev. Sci. Instrum. 1987, 58, 2131.(22) de Lange, C. A. In High-Resolution Laser Photoionization and

Photoelectron Studies; Powis, I., Baer, T., Ng, C. Y., Eds.;Wiley: New York, 1995; p 195.

(23) Eland, J. H. D. Photoelectron Spectroscopy; Butterworth: Lon-don, 1984.

(24) Seel, M.; Domcke, W. J. Chem. Phys. 1991, 95, 7806.(25) Seel, M.; Domcke, W. Chem. Phys. 1991, 151, 59.(26) Seideman, T. J. Chem. Phys. 1997, 107, 7859.(27) Suzuki, Y.; Stener, M.; Seideman, T. Phys. Rev. Lett. 2002, 89,

233002.(28) Suzuki, Y.; Stener, M.; Seideman, T. J. Chem. Phys. 2003, 118,

4432.(29) Kosloff, R. Annu. Rev. Phys. Chem. 1994, 45, 145.(30) Braun, M.; Meier, C.; Engel, V. J. Chem. Phys. 1995, 103, 7907.(31) Meier, C.; Engel, V. Phys. Chem. Chem. Phys. 2002, 4, 5014.(32) Rubner, O.; Meier, C.; Engel, V. J. Chem. Phys. 1997, 107, 1066.(33) Fischer, I.; Villeneuve, D. M.; Vrakking, M. J. J.; Stolow, A. J.

Chem. Phys. 1995, 102, 5566.

(34) Fischer, I.; Vrakking, M. J. J.; Villeneuve, D. M.; Stolow, A.Chem. Phys. 1996, 207, 331.

(35) Assion, A.; Geisler, M.; Helbing, J.; Seyfried, V.; Baumert, T.Phys. Rev. A 1996, 54, R4605.

(36) Greenblatt, B. J.; Zanni, M. T.; Neumark, D. M. Chem. Phys.Lett. 1996, 258, 523.

(37) Zanni, M. T.; Batista, V. S.; Greenblatt, B. J.; Miller, W. H.;Neumark, D. M. J. Chem. Phys. 1999, 110, 3748.

(38) Cyr, D. R.; Hayden, C. C. J. Chem. Phys. 1996, 104, 771.(39) Blanchet, V.; Stolow, A. In Ultrafast Phenomena XI; Elsaesser,

T., Fujimoto, J. G., Wiersma, D. A., Zinth, W., Eds.; SpringerSeries in Chemical Physics 63; Springer-Verlag: Berlin, 1998;p 456.

(40) Blanchet, V.; Zgierski, M. Z.; Seideman, T.; Stolow, A. Nature1999, 401, 52.

(41) Blanchet, V.; Zgierski, M. Z.; Stolow, A. J. Chem. Phys. 2001,114, 1194.

(42) Schmitt, M.; Lochbrunner, S.; Shaffer, J. P.; Larsen, J. J.;Zgierski, M. Z.; Stolow, A. J. Chem. Phys. 2001, 114, 1206.

(43) Bandrauk, A. D. Molecules in Laser Fields; Marcel Dekker: NewYork, 1994.

(44) Gavrila, M. Atoms in Intense Laser Fields; Academic Press: SanDiego, 1992.

(45) Zavriyev, A.; Fischer, I.; Villeneuve, D. M.; Stolow, A. Chem.Phys. Lett. 1995, 234, 281.

(46) Freeman, R. R.; Bucksbaum, P. H. J. Phys. B: At. Mol. Opt. Phys.1991, 24, 325.

(47) Assion, A.; Baumert, T.; Helbing, J.; Gerber, G. In UltrafastPhenomena X; Fujimoto, J. G., Zinth, W., Barbara, P. F., Knox,W. H., Eds.; Springer Series in Chemical Physics 62; Springer-Verlag: Berlin Heidelberg, 1996; p 270.

(48) Assion, A.; Baumert, T.; Helbing, J.; Seyfried, V.; Gerber, G.Phys. Rev. A 1997, 55, 1899.

(49) Baumert, T.; Gerber, G. Phys. Scr. 1997, T72, 53.(50) Frohnmeyer, T.; Baumert, T. Appl. Phys. B 2000, 71, 259.(51) Wollenhaupt, M.; Assion, A.; Bazhan, O.; Liese, D.; Sarpe-

Tudoran, C.; Baumert, T. Appl. Phys. B 2002, 74, S121.(52) Frohnmeyer, T.; Hofmann, H.; Strehle, M.; Baumert, T. Chem.

Phys. Lett. 1999, 312, 447.(53) Ludowise, P.; Blackwell, M.; Chen, Y. Chem. Phys. Lett. 1996,

258, 530.(54) Sun, Z. G.; Liu, H. P.; Lou, N. Q.; Cong, S. L. Chem. Phys. Lett.

2003, 369, 374.(55) Seideman, T. J. Chem. Phys. 1995, 103, 7887.(56) Seideman, T. Phys. Rev. Lett. 1999, 83, 4971.(57) Ortigoso, J.; Rodriguez, M.; Gupta, M.; Friedrich, B. J. Chem.

Phys. 1999, 110, 3870.(58) Althorpe, S. C.; Seideman, T. J. Chem. Phys. 1999, 110, 147.(59) Althorpe, S. C.; Seideman, T. J. Chem. Phys. 2000, 113, 7901.(60) Reid, K. L.; Underwood, J. G. J. Chem. Phys. 2000, 112, 3643.(61) Pallix, J. B.; Colson, S. D. Chem. Phys. Lett. 1985, 119, 38.(62) Sekreta, E.; Reilly, J. P. Chem. Phys. Lett. 1988, 149, 482.(63) Song, X. B.; Wilkerson, C. W.; Lucia, J.; Pauls, S.; Reilly, J. P.

Chem. Phys. Lett. 1990, 174, 377.(64) Smith, J. M.; Lakshminarayan, C.; Knee, J. L. J. Chem. Phys.

1990, 93, 4475.(65) Syage, J. A. Chem. Phys. Lett. 1993, 202, 227.(66) Syage, J. A. J. Phys. Chem. 1995, 99, 5772.(67) Kim, B.; Schick, C. P.; Weber, P. M. J. Chem. Phys. 1995, 103,

6903.(68) Dobber, M. R.; Buma, W. J.; Delange, C. A. J. Chem. Phys. 1993,

99, 836.(69) Dobber, M. R.; Buma, W. J.; Delange, C. A. J. Phys. Chem. 1995,

99, 1671.(70) Schultz, T.; Fischer, I. J. Chem. Phys. 1997, 107, 8197.(71) Schultz, T.; Fischer, I. J. Chem. Phys. 1998, 109, 5812.(72) Schultz, T.; Clarke, J. S.; Gilbert, T.; Deyerl, H. J.; Fischer, I.

Faraday Discuss. 2000, 17.(73) Baumert, T.; Thalweiser, R.; Gerber, G. Chem. Phys. Lett. 1993,

209, 29.(74) Assion, A.; Baumert, T.; Helbing, J.; Seyfried, V.; Gerber, G.

Chem. Phys. Lett. 1996, 259, 488.(75) Gantefor, G.; Kraus, S.; Eberhardt, W. J. Electron Spectrosc.

Relat. Phenom. 1998, 88, 35.(76) Lopez-Martens, R.; Long, P.; Solgadi, D.; Soep, B.; Syage, J.;

Millie, P. Chem. Phys. Lett. 1997, 273, 219.(77) Blanchet, V.; Stolow, A. J. Chem. Phys. 1998, 108, 4371.(78) Kajimoto, O.; Honma, K.; Kobayashi, T. J. Phys. Chem. 1985,

89, 2725.(79) Stert, V.; Radloff, W.; Freudenberg, T.; Noack, F.; Hertel, I. V.;

Jouvet, C.; Dedonderlardeux, C.; Solgadi, D. Europhys. Lett.1997, 40, 515.

(80) Radloff, W.; Stert, V.; Freudenberg, T.; Hertel, I. V.; Jouvet, C.;Dedonderlardeux, C.; Solgadi, D. Chem. Phys. Lett. 1997, 281,20.

(81) Stert, V.; Radloff, W.; Schulz, C. P.; Hertel, I. V. Eur. Phys. J. D1999, 5, 97.

(82) Noguchi, T.; Sato, S.; Fujimura, Y. Chem. Phys. Lett. 1989, 155,177.


(83) Reid, K. L. Chem. Phys. Lett. 1993, 215, 25.(84) Reid, K. L.; Duxon, S. P.; Towrie, M. Chem. Phys. Lett. 1994,

228, 351.(85) Meier, C.; Engel, V. Chem. Phys. Lett. 1993, 212, 691.(86) Meier, C.; Engel, V. J. Chem. Phys. 1994, 101, 2673.(87) Braun, M.; Meier, C.; Engel, V. J. Chem. Phys. 1996, 105, 530.(88) Vrakking, M. J. J.; Villeneuve, D. M.; Stolow, A. J. Chem. Phys.

1996, 105, 5647.(89) Braun, M.; Engel, V. Z. Phys. D 1997, 39, 301.(90) Meyer, S.; Meier, C.; Engel, V. J. Chem. Phys. 1998, 108, 7631.(91) Lochbrunner, S.; Larsen, J. J.; Shaffer, J. P.; Schmitt, M.;

Schultz, T.; Underwood, J. G.; Stolow, A. J. Electron Spectrosc.Relat. Phenom. 2000, 112, 183.

(92) Wang, L. S.; Cheng, H. S.; Fan, J. W. J. Chem. Phys. 1995, 102,9480.

(93) Hotop, H.; Lineberger, W. C. J. Phys. Chem. Ref. Data 1985,14, 731.

(94) Markovich, G.; Perera, L.; Berkowitz, M. L.; Cheshnovsky, O.J. Chem. Phys. 1996, 105, 2675.

(95) Handschuh, H.; Gantefor, G.; Eberhardt, W. Rev. Sci. Instrum.1995, 66, 3838.

(96) Wang, L. S.; Ding, C. F.; Wang, X. B.; Barlow, S. E. Rev. Sci.Instrum. 1999, 70, 1957.

(97) Giniger, R.; Hippler, T.; Ronen, S.; Cheshnovsky, O. Rev. Sci.Instrum. 2001, 72, 2543.

(98) Eppink, A. T. J. B.; Parker, D. H. Rev. Sci. Instrum. 1997, 68,3477.

(99) Heck, A. J. R.; Chandler, D. W. Annu. Rev. Phys. Chem. 1995,46, 335.

(100) Parker, D. H.; Eppink, A. T. J. B. J. Chem. Phys. 1997, 107,2357.

(101) Bordas, C.; Paulig, F.; Helm, H.; Huestis, D. L. Rev. Sci. Instrum.1996, 67, 2257.

(102) Surber, E.; Sanov, A. J. Chem. Phys. 2002, 116, 5921.(103) Whitaker, B. J. In Imaging in Chemical Dynamics; Suits, A. G.,

Continetti, R. E., Eds.; American Chemical Society: Washington,DC, 2000; p 68.

(104) Vrakking, M. J. J. Rev. Sci. Instrum. 2001, 72, 4084.(105) Dribinski, V.; Ossadtchi, A.; Mandelshtam, V. A.; Reisler, H. Rev.

Sci. Instrum. 2002, 73, 2634.(106) Gebhardt, C. R.; Rakitzis, T. P.; Samartzis, P. C.; Ladopoulos,

V.; Kitsopoulos, T. N. Rev. Sci. Instrum. 2001, 72, 3848.(107) Townsend, D.; Minitti, M. P.; Suits, A. G. Rev. Sci. Instrum. 2003,

74, 2530.(108) Continetti, R. E.; Hayden, C. C. In Advanced Series in Physical

Chemistry: Modern Trends in Chemical Reaction Dynamics; Liu,K., Yang, X., Eds.; World Scientific: Singapore, 2003.

(109) Martin, C.; Jelinsky, P.; Lampton, M.; Malina, R. F.; Anger, H.O. Rev. Sci. Instrum. 1981, 52, 1067.

(110) Continetti, R. E.; Cyr, D. R.; Osborn, D. L.; Leahy, D. J.;Neumark, D. M. J. Chem. Phys. 1993, 99, 2616.

(111) Ali, I.; Dorner, R.; Jagutzki, O.; Nuttgens, S.; Mergel, V.;Spielberger, L.; Khayyat, K.; Vogt, T.; Brauning, H.; Ullmann,K.; Moshammer, R.; Ullrich, J.; Hagmann, S.; Groeneveld, K.O.; Cocke, C. L.; Schmidt-Bocking, H. Nucl. Instrum. MethodsB 1999, 149, 490.

(112) Hanold, K. A.; Luong, A. K.; Clements, T. G.; Continetti, R. E.Rev. Sci. Instrum. 1999, 70, 2268.

(113) Hanold, K. A.; Garner, M. C.; Continetti, R. E. Phys. Rev. Lett.1996, 77, 3335.

(114) Garner, M. C.; Hanold, K. A.; Resat, M. S.; Continetti, R. E. J.Phys. Chem. A 1997, 101, 6577.

(115) Davies, J. A.; LeClaire, J. E.; Continetti, R. E.; Hayden, C. C. J.Chem. Phys. 1999, 111, 1.

(116) Davies, J. A.; Continetti, R. E.; Chandler, D. W.; Hayden, C. C.Phys. Rev. Lett. 2000, 84, 5983.

(117) Krausz, F.; Fermann, M. E.; Brabec, T.; Curley, P. F.; Hofer,M.; Ober, M. H.; Spielmann, C.; Wintner, E.; Schmidt, A. J. IEEEJ. Quantum Elect. 1992, 28, 2097.

(118) Squier, J.; Salin, F.; Mourou, G.; Harter, D. Opt. Lett. 1991, 16,324.

(119) Nisoli, M.; Desilvestri, S.; Magni, V.; Svelto, O.; Danielius, R.;Piskarskas, A.; Valiulis, G.; Varanavicius, A. Opt. Lett. 1994,19, 1973.

(120) Wilhelm, T.; Piel, J.; Riedle, E. Opt. Lett. 1997, 22, 1494.(121) L’Huillier, A. J. Nonlinear Opt. Phys. Mater. 1995, 4, 647.(122) Spielmann, C.; Burnett, N. H.; Sartania, S.; Koppitsch, R.;

Schnurer, M.; Kan, C.; Lenzner, M.; Wobrauschek, P.; Krausz,F. Science 1997, 278, 661.

(123) Rundquist, A.; Durfee, C. G.; Chang, Z. H.; Herne, C.; Backus,S.; Murnane, M. M.; Kapteyn, H. C. Science 1998, 280, 1412.

(124) Bartels, R.; Backus, S.; Zeek, E.; Misoguti, L.; Vdovin, G.;Christov, I. P.; Murnane, M. M.; Kapteyn, H. C. Nature 2000,406, 164.

(125) Durfee, C. G.; Rundquist, A. R.; Backus, S.; Herne, C.; Murnane,M. M.; Kapteyn, H. C. Phys. Rev. Lett. 1999, 83, 2187.

(126) Sorensen, S. L.; Bjorneholm, O.; Hjelte, I.; Kihlgren, T.; Ohrwall,G.; Sundin, S.; Svensson, S.; Buil, S.; Descamps, D.; L’Huillier,A.; Norin, J.; Wahlstrom, C. G. J. Chem. Phys. 2000, 112, 8038.

(127) Nugent-Glandorf, L.; Scheer, M.; Samuels, D. A.; Bierbaum, V.;Leone, S. R. Rev. Sci. Instrum. 2002, 73, 1875.

(128) Nugent-Glandorf, L.; Scheer, M.; Samuels, D. A.; Mulhisen, A.M.; Grant, E. R.; Yang, X. M.; Bierbaum, V. M.; Leone, S. R.Phys. Rev. Lett. 2001, 87, 193002.

(129) Nugent-Glandorf, L.; Scheer, M.; Samuels, D. A.; Bierbaum, V.M.; Leone, S. R. J. Chem. Phys. 2002, 117, 6108.

(130) Corkum, P. B.; Burnett, N. H.; Ivanov, M. Y. Opt. Lett. 1994,19, 1870.

(131) Ivanov, M.; Corkum, P. B.; Zuo, T.; Bandrauk, A. Phys. Rev. Lett.1995, 74, 2933.

(132) Hentschel, M.; Kienberger, R.; Spielmann, C.; Reider, G. A.;Milosevic, N.; Brabec, T.; Corkum, P.; Heinzmann, U.; Drescher,M.; Krausz, F. Nature 2001, 414, 509.

(133) Drescher, M.; Hentschel, M.; Kienberger, R.; Uiberacker, M.;Yakovlev, V.; Scrinizi, A.; Westerwalbesloh, T.; Kleineberg, U.;Heinzmann, U.; Krausz, F. Nature 2002, 419, 803.

(134) Blanchet, V.; Lochbrunner, S.; Schmitt, M.; Shaffer, J. P.; Larsen,J. J.; Zgierski, M. Z.; Seideman, T.; Stolow, A. Faraday Discuss.2000, 33.

(135) Suzuki, T.; Wang, L.; Kohguchi, H. J. Chem. Phys. 1999, 111,4859.

(136) Wang, L.; Kohguchi, H.; Suzuki, T. Faraday Discuss. 1999, 37.(137) Stert, V.; Farmanara, P.; Radloff, W. J. Chem. Phys. 2000, 112,

4460.(138) Seidner, L.; Domcke, W.; Vonniessen, W. Chem. Phys. Lett. 1993,

205, 117.(139) Shaffer, J. P.; Schultz, T.; Schmitt, M.; Underwood, J. G.; Stolow,

A. In Ultrafast Phenomena XII; Elsaesser, T., Mukamel, S.,Murnane, M., Scherer, N., Eds.; Springer Series in ChemicalPhysics 66; Springer-Verlag: Berlin, 2000; p 338.

(140) Olivucci, M. Private communication.(141) Baumert, T.; Frohnmeyer, T.; Kiefer, B.; Niklaus, P.; Strehle,

M.; Gerber, G.; Zewail, A. H. Appl. Phys. B 2001, 72, 105.(142) Pedersen, S.; Banares, L.; Zewail, A. H. J. Chem. Phys. 1992,

97, 8801.(143) Soep, B.; Mestdagh, J. M.; Sorgues, S.; Visticot, J. P. Eur. Phys.

J. D 2001, 14, 191.(144) Farmanara, P.; Stert, V.; Radloff, W.; Hertel, I. V. J. Phys. Chem.

A 2001, 105, 5613.(145) Song, J. K.; Tsubouchi, M.; Suzuki, T. J. Chem. Phys. 2001, 115,

8810.(146) Hahn, S.; Stock, G. Phys. Chem. Chem. Phys. 2001, 3, 2331.(147) Hahn, S.; Stock, G. Chem. Phys. Lett. 1998, 296, 137.(148) Davies, J. A.; Reid, K. L.; Towrie, M.; Matousek, P. J. Chem.

Phys. 2002, 117, 9099.(149) Moss, D. B.; Parmenter, C. S. J. Chem. Phys. 1993, 98, 6897.(150) Lee, S. H.; Tang, K. C.; Chen, I. C.; Schmitt, M.; Shaffer, J. P.;

Schultz, T.; Underwood, J. G.; Zgierski, M. Z.; Stolow, A. J. Phys.Chem. A 2002, 106, 8979.

(151) Matsumoto, Y.; Kim, S. K. K.; Suzuki, T. J. Chem. Phys. 2003,119, 300.

(152) Scheiner, S. J. Phys. Chem. A 2000, 104, 5898.(153) Lochbrunner, S.; Schmitt, M.; Shaffer, J. P.; Schultz, T.; Stolow,

A. In Ultrafast Phenomena XII; Elsaesser, T., Mukamel, S.,Murnane, M., Scherer, N., Eds.; Springer Series in ChemicalPhysics 66; Springer-Verlag: Berlin, 2000; p 642.

(154) Lochbrunner, S.; Schultz, T.; Schmitt, M.; Shaffer, J. P.; Zgierski,M. Z.; Stolow, A. J. Chem. Phys. 2001, 114, 2519.

(155) Sobolewski, A. L.; Domcke, W. Chem. Phys. 1994, 184, 115.(156) Sobolewski, A. L.; Domcke, W. Phys. Chem. Chem. Phys. 1999,

1, 3065.(157) Farmanara, P.; Radloff, W.; Stert, V.; Ritze, H. H.; Hertel, I. V.

J. Chem. Phys. 1999, 111, 633.(158) Farmanara, F.; Ritze, H. H.; Stert, V.; Radloff, W.; Hertel, I. V.

Eur. Phys. J. D 2002, 19, 193.(159) Lippert, H.; Stert, V.; Hesse, L.; Schultz, C. P.; Hertel, I. V.;

Radloff, W. J. Phys. Chem. A 2003, 107, 8239.(160) Sobolewski, A. L.; Domcke, W. Chem. Phys. Lett. 1999, 315, 293.(161) Sobolewski, A. L.; Domcke, W. Chem. Phys. Lett. 2000, 329, 130.(162) Schultz, T.; Quenneville, J.; Levine, B.; Toniolo, A.; Lochbrunner,

S.; Schmitt, M.; Shaffer, J. P.; Zgierski, M. Z.; Stolow, A. J. Am.Chem. Soc. 2003, 125, 8098.

(163) Rau, H. In Photochromism, Molecules and Systems; Durr, H.,Buas-Laurent, H., Eds.; Elsevier: Amsterdam, 1990; p 165.

(164) Tamai, N.; Miyasaka, H. Chem. Rev. 2000, 100, 1875.(165) Schick, C. P.; Weber, P. M. J. Phys. Chem. A 2001, 105, 3725.(166) Schick, C. P.; Weber, P. M. J. Phys. Chem. A 2001, 105, 3735.(167) Cheng, W.; Evans, C. L.; Kuthirummal, N.; Weber, P. M. Chem.

Phys. Lett. 2001, 349, 405.(168) Schmitt, M.; Stolow, A. Private communication.(169) Probst, M.; Haight, R. Appl. Phys. Lett. 1997, 71, 202.(170) Zamith, S.; Blanchet, V.; Girard, B.; Andersson, J.; Sorensen,

S.; Hjelte, I.; Bjorneholm, O.; Gauyacq, D.; Norin, J.; Mauritson,J.; L’Huillier, A. J. Chem. Phys. 2003, 119, 3763.

(171) Farmanara, P.; Stert, V.; Ritze, H. H.; Radloff, W. J. Chem. Phys.2000, 113, 1705.

(172) Stert, V.; Ritze, H. H.; Nibbering, E. T. J.; Radloff, W. Chem.Phys. 2001, 272, 99.


(173) Erdmann, M.; Rubner, O.; Shen, Z.; Engel, V. Chem. Phys. Lett.2001, 341, 338.

(174) Erdmann, M.; Rubner, O.; Shen, Z.; Engel, V. J. Organomet.Chem. 2002, 661, 191.

(175) Shaffer, J. P.; Schultz, T.; Underwood, J. G.; Hayden, C. C.;Stolow, A. In Ultrafast Phenomena XIII; Miller, R. D., Murnane,M. M., Scherer, N. F., Weiner, A. M., Eds.; Spring Series inChemical Physics 71; Springer-Verlag: Berlin, 2002; p 73.

(176) Tsubouchi, M.; de Lange, C. A.; Suzuki, T. J. Chem. Phys. 2003,119, 11728.

(177) Tsubouchi, M.; Suzuki, T. Chem. Phys. Lett. 2003, 382, 418.(178) Demyanenko, A. V.; Potter, A. B.; Dribinski, V.; Reisler, H. J.

Chem. Phys. 2002, 117, 2568.(179) Scherer, N. F.; Khundkar, L. R.; Bernstein, R. B.; Zewail, A. H.

J. Chem. Phys. 1987, 87, 1451.(180) Stert, V.; Farmanara, P.; Ritze, H. H.; Radloff, W.; Gasmi, K.;

Gonzalez-Urena, A. Chem. Phys. Lett. 2001, 337, 299.(181) Zanni, M. T.; Greenblatt, B. J.; Neumark, D. M. J. Chem. Phys.

1997, 107, 7613.(182) Batista, V. S.; Zanni, M. T.; Greenblatt, B. J.; Neumark, D. M.;

Miller, W. H. J. Chem. Phys. 1999, 110, 3736.(183) Zanni, M. T.; Davis, A. V.; Frischkorn, C.; Elhanine, M.;

Neumark, D. M. J. Chem. Phys. 2000, 112, 8847.(184) Zanni, M. T.; Greenblatt, B. J.; Davis, A. V.; Neumark, D. M. J.

Chem. Phys. 1999, 111, 2991.(185) Kuhne, T.; Vohringer, P. J. Chem. Phys. 1996, 105, 10788.(186) Banin, U.; Ruhman, S. J. Chem. Phys. 1993, 98, 4391.(187) Papanikolas, J. M.; Gord, J. R.; Levinger, N. E.; Ray, D.; Vorsa,

V.; Lineberger, W. C. J. Phys. Chem. 1991, 95, 8028.(188) Papanikolas, J. M.; Vorsa, V.; Nadal, M. E.; Campagnola, P. J.;

Buchenau, H. K.; Lineberger, W. C. J. Chem. Phys. 1993, 99,8733.

(189) Vorsa, V.; Campagnola, P. J.; Nandi, S.; Larsson, M.; Lineberger,W. C. J. Chem. Phys. 1996, 105, 2298.

(190) Vorsa, V.; Nandi, S.; Campagnola, P. J.; Larsson, M.; Lineberger,W. C. J. Chem. Phys. 1997, 106, 1402.

(191) Sanov, A.; Lineberger, W. C. PhysChemComm 2002, 165.(192) Papanikolas, J. M.; Maslen, P. E.; Parson, R. J. Chem. Phys.

1995, 102, 2452.(193) Batista, V. S.; Coker, D. F. J. Chem. Phys. 1997, 106, 7102.(194) Delaney, N.; Faeder, J.; Maslen, P. E.; Parson, R. J. Phys. Chem.

A 1997, 101, 8147.(195) Faeder, J.; Delaney, N.; Maslen, P. E.; Parson, R. Chem. Phys.

Lett. 1997, 270, 196.(196) Margulis, C. J.; Coker, D. F. J. Chem. Phys. 1999, 110, 5677.(197) Parson, R.; Faeder, J.; Delaney, N. J. Phys. Chem. A 2000, 104,

9653.(198) Greenblatt, B. J.; Zanni, M. T.; Neumark, D. M. Faraday Discuss.

1997, 108, 101.(199) Greenblatt, B. J.; Zanni, M. T.; Neumark, D. M. Science 1997,

276, 1675.(200) Greenblatt, B. J.; Zanni, M. T.; Neumark, D. M. J. Chem. Phys.

2000, 112, 601.(201) Greenblatt, B. J.; Zanni, M. T.; Neumark, D. M. J. Chem. Phys.

1999, 111, 10566.(202) Paik, D. H.; Bernhardt, T. M.; Kim, N. J.; Zewail, A. H. J. Chem.

Phys. 2001, 115, 612.(203) Paik, D. H.; Kim, N. J.; Zewail, A. H. J. Chem. Phys. 2003, 118,

6923.(204) Kim, N. J.; Paik, D. H.; Zewail, A. H. J. Chem. Phys. 2003, 118,

6930.(205) Hiraoka, K. Chem. Phys. 1988, 125, 439.(206) Deluca, M. J.; Han, C. C.; Johnson, M. A. J. Chem. Phys. 1990,

93, 268.(207) Wester, R.; Bragg, A. E.; Davis, A. V.; Neumark, D. M. J. Chem.

Phys. 2003, 119, 10032.(208) Zanni, M. T.; Greenblatt, B. J.; Neumark, D. M. J. Chem. Phys.

1998, 109, 9648.(209) Davis, A. V.; Wester, R.; Bragg, A. E.; Neumark, D. M. J. Chem.

Phys. 2002, 117, 4282.(210) Davis, A. V.; Zanni, M. T.; Frischkorn, C.; Elhanine, M.;

Neumark, D. M. J. Electron Spectrosc. Relat. Phenom. 2000, 112,221.

(211) Davis, A. V.; Wester, R.; Bragg, A. E.; Neumark, D. M. J. Chem.Phys. 2003, 119, 2020.

(212) Franck, J.; Sheibe, G. Z. Phys. Chem. A 1928, 139, 22.(213) Blandamer, M.; Fox, M. Chem. Rev. 1970, 70, 59.(214) Jortner, J.; Ottolenghi, M.; Stein, G. J. Phys. Chem. 1964, 68,

247.(215) Long, F. H.; Shi, X. L.; Lu, H.; Eisenthal, K. B. J. Phys. Chem.

1994, 98, 7252.(216) Kloepfer, J. A.; Vilchiz, V. H.; Lenchenkov, V. A.; Bradforth, S.

E. Chem. Phys. Lett. 1998, 298, 120.(217) Martini, I. B.; Barthel, E. R.; Schwartz, B. J. Science 2001, 293,

462.(218) Sheu, W. S.; Rossky, P. J. Chem. Phys. Lett. 1993, 202, 186.(219) Serxner, D.; Dessent, C. E. H.; Johnson, M. A. J. Chem. Phys.

1996, 105, 7231.(220) Becker, I.; Cheshnovsky, O. J. Chem. Phys. 1999, 110, 6288.

(221) Becker, I.; Markovich, G.; Cheshnovsky, O. Phys. Rev. Lett. 1997,79, 3391.

(222) Lehr, L.; Zanni, M. T.; Frischkorn, C.; Weinkauf, R.; Neumark,D. M. Science 1999, 284, 635.

(223) Zanni, M. T.; Frischkorn, C.; Davis, A. V.; Neumark, D. M. J.Phys. Chem. A 2000, 104, 2527.

(224) Davis, A. V.; Zanni, M. T.; Frischkorn, C.; Neumark, D. M. J.Electron Spectrosc. Relat. Phenom. 2000, 108, 203.

(225) Frischkorn, C.; Zanni, M. T.; Davis, A. V.; Neumark, D. M.Faraday Discuss. 2000, 115, 49.

(226) Davis, A. V.; Zanni, M. T.; Weinkauf, R.; Neumark, D. M. Chem.Phys. Lett. 2002, 353, 455.

(227) Zanni, M. T.; Lehr, L.; Greenblatt, B. J.; Weinkauf, R.; Neumark,D. M. In Ultrafast Phenomena XI; Elsaesser, T., Fujimoto, J.G., Wiersma, D., Zinth, W., Eds.; Springer Series in ChemicalPhysics 63; Springer-Verlag: Berlin, 1998; p 474.

(228) Chen, H.-Y.; Sheu, W.-S. Chem. Phys. Lett. 2001, 335, 475.(229) Chen, H.-Y.; Sheu, W.-S. J. Am. Chem. Soc. 2000, 122, 7534.(230) Vila, F. D.; Jordan, K. D. J. Phys. Chem. A 2002, 106, 1391.(231) Timerghazin, Q. K.; Peslherbe, G. H. J. Am. Chem. Soc. 2003,

125, 9904.(232) Lenzer, T.; Furlanetto, M. R.; Asmis, K. R.; Neumark, D. M. J.

Chem. Phys. 1998, 109, 10754.(233) Lenzer, T.; Furlanetto, M. R.; Pivonka, N. L.; Neumark, D. M.

J. Chem. Phys. 1999, 110, 6714.(234) Hart, E. J.; Boag, J. W. J. Am. Chem. Soc. 1962, 84, 4090.(235) Migus, A.; Gauduel, Y.; Martin, J. L.; Antonetti, A. Phys. Rev.

Lett. 1987, 58, 1559.(236) Rossky, P. J.; Schnitker, J. J. Phys. Chem. 1988, 92, 4277.(237) Coe, J. V.; Lee, G. H.; Eaton, J. G.; Arnold, S. T.; Sarkas, H. W.;

Bowen, K. H.; Ludewigt, C.; Haberland, H.; Worsnop, D. R. J.Chem. Phys. 1990, 92, 3980.

(238) Ayotte, P.; Johnson, M. A. J. Chem. Phys. 1997, 106, 811.(239) Ayotte, P.; Weddle, G. H.; Bailey, C. G.; Johnson, M. A.; Vila,

F.; Jordan, K. D. J. Chem. Phys. 1999, 110, 6268.(240) Weber, J. M.; Kim, J.; Woronowicz, E. A.; Weddle, G. H.; Becker,

I.; Cheshnovsky, O.; Johnson, M. A. Chem. Phys. Lett. 2001, 339,337.

(241) Minemoto, S.; Muller, J.; Gantefor, G.; Munzer, H. J.; Boneberg,J.; Leiderer, P. Phys. Rev. Lett. 2000, 84, 3554.

(242) Tulej, M.; Fulara, J.; Sobolewski, A.; Jungen, M.; Maier, J. P. J.Chem. Phys. 2000, 112, 3747.

(243) Frischkorn, C.; Bragg, A. E.; Davis, A. V.; Wester, R.; Neumark,D. M. J. Chem. Phys. 2001, 115, 11185.

(244) Knoesel, E.; Hotzel, A.; Hertel, T.; Wolf, M.; Ertl, G. Surf. Sci.1996, 368, 76.

(245) Hertel, T.; Knoesel, E.; Wolf, M.; Ertl, G. Phys. Rev. Lett. 1996,76, 535.

(246) Pontius, N.; Bechthold, P. S.; Neeb, M.; Eberhardt, W. Appl.Phys. B 2000, 71, 351.

(247) Pontius, N.; Bechthold, P. S.; Neeb, M.; Eberhardt, W. Phys. Rev.Lett. 2000, 84, 1132.

(248) Pontius, N.; Bechthold, P. S.; Neeb, M.; Eberhardt, W. J. ElectronSpectrosc. Relat. Phenom. 2000, 106, 107.

(249) Pontius, N.; Luttgens, G.; Bechthold, P. S.; Neeb, M.; Eberhardt,W. J. Chem. Phys. 2001, 115, 10479.

(250) Pontius, N.; Neeb, M.; Eberhardt, W. Phys. Rev. B 2003, 67, 1.(251) Pine, D.; Nozieres, P. The Theory of Quantum Liquids; Ben-

jamin: New York, 1966.(252) Gerhardt, P.; Niemietz, M.; Kim, Y. D.; Gantefor, G. Chem. Phys.

Lett. 2003, 382, 454.(253) Niemietz, M.; Gerhardt, P.; Gantefor, G.; Kim, Y. D. Chem. Phys.

Lett. 2003, 380, 99.(254) Arasaki, Y.; Takatsuka, K.; Wang, K.; McKoy, V. Chem. Phys.

Lett. 1999, 302, 363.(255) Arasaki, Y.; Takatsuka, K.; Wang, K. H.; McKoy, V. J. Electron

Spectrosc. Relat. Phenom. 2000, 108, 89.(256) Arasaki, Y.; Takatsuka, K.; Wang, K.; McKoy, V. J. Chem. Phys.

2000, 112, 8871.(257) Wollenhaupt, M.; Assion, A.; Graefe, O.; Liese, D.; Sarpe-

Tudoran, C.; Wintner, E.; Baumert, T. Chem. Phys. Lett. 2003,376, 457.

(258) Takatsuka, K.; Arasaki, Y.; Wang, K.; McKoy, V. FaradayDiscuss. 2000, 1.

(259) Reid, K. L.; Field, T. A.; Towrie, M.; Matousek, P. J. Chem. Phys.1999, 111, 1438.

(260) Underwood, J. G.; Reid, K. L. J. Chem. Phys. 2000, 113, 1067.(261) Seideman, T. J. Chem. Phys. 2000, 113, 1677.(262) Seideman, T.; Althorpe, S. C. J. Electron Spectrosc. Relat.

Phenom. 2000, 108, 99.(263) Seideman, T. Phys. Rev. A 2001, 6404, 042504.(264) Resch, K.; Blanchet, V.; Stolow, A.; Seideman, T. J. Phys. Chem.

A 2001, 105, 2756.(265) Piancastelli, M. N.; Keller, P. R.; Taylor, J. W.; Grimm, F. A.;

Carlson, T. A. J. Am. Chem. Soc. 1983, 105, 4235.(266) Tsubouchi, M.; Whitaker, B. J.; Wang, L.; Kohguchi, H.; Suzuki,

T. Phys. Rev. Lett. 2001, 86, 4500.(267) Davis, A. V.; Wester, R.; Bragg, A. E.; Neumark, D. M. J. Chem.

Phys. 2003, 118, 999.


(268) Bragg, A. E.; Wester, R.; Davis, A. V.; Kammrath, A.; Neumark,D. M. Chem. Phys. Lett. 2003, 376, 767.

(269) Reed, K. J.; Zimmerman, A. H.; Andersen, H. C.; Brauman, J.I. J. Chem. Phys. 1976, 64, 1368.

(270) Peronne, E.; Poulson, M. D.; Bisgaard, C. Z.; Stapelfeldt, H.;Seideman, T. Phys. Rev. Lett. 2003, 91, 043003.

(271) Underwood, J. G.; Spanner, M.; Ivanov, M. Y.; Mottershead, J.;Sussman, B. J.; Stolow, A. Phys. Rev. Lett. 2003, 90, 223001.

(272) Weinkauf, R.; Schermann, J. P.; de Vries, M. S.; Kleinermanns,K. Eur. Phys. J. D 2002, 20, 309.

CR020683W


Documents

Femtosecond Time-Resolved Photoelectron Spectroscopybromine.cchem.berkeley.edu/grppub/fpes28.pdf · many concepts in molecular electronics.9 Femtosecond time-resolved methods involve