electronic continua in time-resolved photoelectron spectroscopy....

JOURNAL OF CHEMICAL PHYSICS VOLUME 114, NUMBER 3 15 JANUARY 2001

Electronic continua in time-resolved photoelectron spectroscopy.I. Complementary ionization correlations

V. Blanchet,a) M. Z. Zgierski, and Albert Stolowb)

Steacie Institute for Molecular Sciences, National Research Council, Ottawa, Ontario, Canada K1A 0R6

~Received 14 June 2000; accepted 20 October 2000!

We examine the role of electronic continua in time-resolved photoelectron spectroscopy studies ofpolyatomic nonadiabatic dynamics. We have investigated the two limiting cases for such studies.We consider here the limiting case of complementary ionization correlations where the twononadiabatically coupled excited electronic states~S2 andS1! correlate~in the Koopmans’ picture!to different cation electronic states. We show, using an example of ultrafast internal conversion alinear polyene, that this favorable case allows for disentangling of the electronic populationdynamics from the coupled vibrational dynamics. In the following paper, we investigate theunfavorable case of corresponding ionization correlations. ©2001 American Institute of Physics.@DOI: 10.1063/1.1331636#

inniano

es

rpia

hesteicabetin

ded

ilen

lue

g-analy—ct

to-

tionr-c-

ry

hichnic

rcsh-e

tos.for

mo-olu-

heter-n thethervo-

suc-aseo

iteri-sm

tee-

mayma

INTRODUCTION

The excited state dynamics of polyatomic moleculesdominated by the nonadiabatic coupling of vibrational aelectronics degrees of freedom. This mixing of electroand nuclear motions induces both charge redistributionenergy flow in molecules. It is the primary step in the phtochemistry of polyatomic molecules,1 photobiological pro-cesses such as vision2 and photosynthesis, and underlimany concepts in molecular electronics.3 The Born–Oppenheimer approximation~BOA! plays the pivotal role indefining the potential energy surface that allows us to repsent nuclear trajectories, thus permitting a mechanisticture of molecular dynamics. The BOA is exact provided ththe nuclear kinetic energy is negligible. Breakdown of tBOA, therefore, isuniquelydue to the motions of the atomand occurs at the intersections or near intersections of potial energy surfaces belonging to different electronconfigurations.4 The nonzero matrix elements of the nuclekinetic energy operator which induce these transitionstween zeroth order states are due to the so-called ‘‘promomodes.’’ These radiationless transitions5–10 often lead tocomplex, broadened absorption spectra due to the highsity of nuclear states and a strong variation of transitionpole with nuclear coordinate.

High resolution spectroscopy provides the most detainsights into these processes, as in the cases of the RenTeller and Jahn–Teller effects11 and conical intersections.12

In some larger molecules such as pyrazine, ‘‘exact’’ sotions to the radiationless transition problem have bedemonstrated.13 In general, however, it remains a challening problem when the state density becomes very highmulti-mode vibronic couplings are involved. Especially chlenging is the case of greatest interest to photochemistrwhen the zeroth order excited states are directly or indire

a!Present address: Lab. Collisions, Agre´gats, Re´activite, IRSAMC,UMR5589 du CNRS, Toulouse, France.

b!Author to whom correspondence should be addressed. [email protected]

1190021-9606/2001/114(3)/1194/12/$18.00

sdcd

-

e-c-t

n-

r-g

n-i-

der–

-n

d-

ly

coupled to a true continuum, leading to nonadiabatic phodissociation dynamics.14,15

Rapid electronic dephasing leading to a strong reducin transition dipole generally limits the spectrosopic ‘‘obsevation’’ of excited state dynamics to times within the eletronic dephasing time,T2* . A classic example of this is theS2←S0 absorption spectrum of butadiene in which the vebroad vibrational structure cannot be resolved.16 Doubleresonance, such as resonance Raman spectroscopy, wprojects the excited state back onto the ground electrostate, can reveal details of theinitial excited statedynamics17–19 but, importantly, only for times on the ordeof T2* . To obtain information about the molecular dynamiafter T2* , i.e., on the ‘‘dark’’ state, double resonance tecniques where the final state isnot the ground electronic statwill be useful.

In this paper we consider an alternative approachthese problems, through the use of time domain technique20

The large bandwidth of a femtosecond pump pulse allowspreparation of a coherent nonstationary superposition oflecular eigenstates—a wave packet. The wave packet evtion is monitored via a coherent~femtosecond! probe pulsewhich projects it onto a final state as a function of time. Tset of these coherent two-photon transitions contains inferences between degenerate transitions which depend oevolving phase factors of the molecular eigenstates; in owords, the signal modulations depend on the dynamical elution of the wave packet.

The femtosecond pump–probe technique has beencessfully applied to a wide range of problems in gas phchemical dynamics.20–31 The choice of the final state ontwhich the wave packet is projected is very important asaffects the information content and determines the expmental method~e.g., detection of photons versus particle!.The particular choice of the molecular ionization continuuas a final state has been argued32–35because the ground staof the ion is often well known, ionization is a universal dtection scheme~no ‘‘dark’’ states!, and multiply differentialdetection techniques such as photoelectron spectroscopyil:

4 © 2001 American Institute of Physics

e

o

y-te

d

eth

oononu

leecrretemun

nicllom

-thresx

.e

td

mfan

c

nra

onec-n.gu-

ale

ects

ic-pec-tlye-ronthee

u-

thethe

istonoton

icser

ndesb-

ans’-

alec-dis-

1195J. Chem. Phys., Vol. 114, No. 3, 15 January 2001 Electronic continua. I

be employed, increasing the information content of the msurement.

Femtosecond time-resolved photoelectron spectrosc~TRPES: for a recent review, see Ref. 36! is beginning to beapplied to a variety of problems. Its application to polatomic nonadiabatic dynamics was studied theoreticallyyears ago37 and first demonstrated for the case ofS1–S0

internal conversion in hexatriene.38 TRPES has been applieto wave packet dynamics in simple systems,32,39–42intramo-lecular vibrational energy redistribution,43–45 nonadiabaticintramolecular dynamics~internal conversion!,46–52 photo-dissociation dynamics,53,54 picosecond spin-orbit coupling~intersystem crossing!,55,56 intracluster reaction dynamics,57

excited state proton transfer dynamics,58 and model molecu-lar electronic switches.59 The outgoing photoelectron may bdifferentially analyzed as a function of time not only wirespect to kinetic energy but also angular distributions60–69

and even spin polarization.70 Additionally, time-resolvedphotoelectron-photoion coincidence~PEPICO!50 andphotoelectron-photoion coincidence-imaging71 spec-troscopies have been demonstrated.

We consider here the role of the electronic structurethe ionization continuum in time-resolved photoelectrspectroscopy experiments. The molecular ionization ctinuum contains two types of electronic structures: that dto the ion core and that due to the free electron. The etronic structure due to the ion core is simply the set of eltronic states of the cation. Electronic Koopmans-type colations upon photoionization of neutral excited stamolecular orbitals may or may not, depending on the symetries involved, allow for disentangling of electronic poplation dynamics from vibrational dynamics. In the presepaper we consider one of the limiting cases, that ofcomple-mentary ionization correlations, which favors the disentagling of complex vibronic dynamics. When the electroncorrelations are mutually exclusive, the nonadiabaticacoupled excited neutral states correlate to different ion ctinua provided that the coupled states are of differing symetry ~as is often the case!. In this limiting case, the electronic population dynamics can be separated fromcoupled vibrational wave packet dynamics even in the pence of large geometry changes upon nonadiabatic crosor ionization. We demonstrate this possibility using the eample of ultrafast internal conversion in a linear polyene

In the second, following paper, we consider the othlimiting case of electronic correlations, that ofcorrespondingionization correlations, which in general are expectedhinder the disentangling. In these cases, where the nonabatically coupled excited states correlate to thesameion con-tinua, considerations of the geometry changes~displace-ments! upon nonadiabatic crossing and ionization becoimportant. We discuss this case using examples of ultrainternal conversion in the polyaromatic hydrocarbophenanthrene and naphthalene.

Finally, in a forthcoming third paper, the use of the eletronic structure of the free electron~the partial wave struc-ture of the continuum! in wave packet measurements is cosidered from a theoretical point of view, using the genenonperturbative formalism developed by Seideman.62,68 The

a-

py

n

f

-ec---

--t

-

yn--

es-ing-

r

oia-

ests

-

-l

partial wave distribution, that is to say, the photoelectrangular distribution, is also generally sensitive to the eltronic symmetry of the state undergoing photoionizatioThus the measurement of time-resolved photoelectron anlar distributions~PADS! gives another route to the potentidisentangling of complex vibronic molecular dynamics. Wexplore several different symmetry cases and discuss effimportant in the determination of PADS.

ION ELECTRONIC CONTINUA

We consider now in more detail the role of the electronsymmetry of the final~ion! state in pump–probe photoionization experiments. Photoelectron spectroscopy is a stroscopy of electronic configurations that remains direcsensitive to vibrational dynamics. A simplified but very usful picture is that emission of an independent outer electoccurs without simultaneous electronic reorganization ofion core ~known as the Koopmans’ or frozen corapproximation!.72,73 Partial ionization probabilities~i.e., ion-ization into specified ion electronic states! can differ drasti-cally with respect to the molecular orbital nature of the netral electronic state.74,75 If a given electronic configurationcorrelates, upon removal of the outermost electron, toelectronic configuration of the ground electronic state ofcation, then the corresponding photoionization probabilitymuch higher than if it does not. In other words, one-phoone-electron operations are more probable than one-phtwo-electron operations.

In Fig. 1 we show a picture of the excited state dynamrelevant to many polyatomic systems. A zeroth ord‘‘bright’’ state, a, is coherently prepared with a femtosecopump pulse. According to the Koopmans’ picture, it ionizinto the a1 continuum, the electronic state of the ion o

FIG. 1. A picture of polyatomic nonadiabatic dynamics. Thea state pre-pared by the pump laser decays into the lower lyingb state due to nonadia-batic coupling. Here we assume that for these two states the Koopmtype correlations upon ionization arecomplementary: The a statepreferentially ionizes into thea1 ground state ion continuum, whereas thebstate preferentially ionizes into theb1 ion continuum, here shown ascation excited state. This scheme should allow the disentangling of etronic from vibrational dynamics during nonadiabatic processes, ascussed in the text.

iole

nigh

B

at-

en

le

led

ultiimmenrodr

b

neitw

se

teelyin

on

st-

ei

tuin

sti-itedith

sed

ce

63

re-rmsn ison

edin.

rald

-

nsityw-de-sion

-nd

,rndthe

the

l-ere

etnghe

e ofote,ges

al

1196 J. Chem. Phys., Vol. 114, No. 3, 15 January 2001 Blanchet, Zgierski, and Stolow

tained upon single-photon, single active electron ionizatof the outermost electron. This process produces a photoetron bande1 . In this example, we have chosena1 to be theground electronic state of the ion. We now consider a nodiabatic process which transforms the zeroth order brstate into a zeroth order ‘‘dark’’ state,b, as induced by pro-moting vibrational modes of the appropriate symmetry.the same arguments, theb state should ionize into theb1

ionization continuum, producing a photoelectron bande2 .Here we discuss the case of complementary correlationsassume thatb1 is an electronically excited state of the caion, as would be the case, for example, ifb contained doublyexcited configurations. Therefore, if we use a sufficientlyergetic probe photon such that both thea1 andb1 continuaare open channels, we would expect a switching of the etronic ionization channel frome1 to e2 during the nonadia-batic process. This picture suggests that we should be abdirectly monitor both the changing electronically excitestate symmetry~i.e., the electronic population dynamics! andthe coupled vibrational wave packet dynamics duringtrafast nonadiabatic processes. The use of photoionizaelectronic correlations was considered in picosecond tscale electronic population flow during intersystecrossing.55 On femtosecond time scales, photoelectron sptroscopy remains a vibrationally resolved technique atherefore the accompanying nuclear dynamics which pmotes and tunes nonadiabatic dynamics can be observethe detailed structure of thee1 ande2 bands. Intramoleculavibrational energy redistribution~IVR! in the a state which,by definition does not alter electronic symmetries, canobserved via thea1 photoelectron bande1 . Similarly, IVRin the b state can be observed via theb1 bande2 . If thispicture is correct, we have in principle a method to disetangle vibrational from electronic population dynamics evthough they are coupled. This could, when combined wquantum dynamical calculations, yield unprecedented vieof the promoting and tuning mode dynamics as well adirect view of the extent of and time scale for IVR on th‘‘dark’’ potential energy surface. Recently, we demonstrathat such disentangling is possible via TRPES, using theample of femtosecond internal conversion in a linear poene,all trans decatetraene47,48 and discuss these resultsdetail below.

LINEAR POLYENE PHOTOPHYSICS

Linear polyenes are hydrocarbon chains that have lbeen an area of fundamental and applied research.76 Theirnonadiabatic dynamics leads to the fundamental procescis-trans photoisomerization. Polyenes form the lighharvesting antennae in vision~rhodopsin!2 and light-driventransmembrane proton pumps~bacteriorhodopsin!. The in-vestigation of polyene photophysics is central to our undstanding of electron delocalization and electron correlationmolecules. They have also been a test bench for quanchemical theory due to their lowest excited state containdoubly excited configurations.77

All trans 2,4,6,8, decatetraene~DT! provides a classicexample of internal conversion in a polyene,76,78,79 the keyphotoinitiated dynamics beingcis-trans isomerization.80 A

nc-

a-t

y

nd

-

c-

to

-one

c-d-via

e

-nhsa

dx--

g

of

r-nmg

great deal of the investigations on polyenes and their subtuted analogs have been devoted to the ordering of excelectronic states, revealing that for simple polyenes wmore than three double bonds81 the first excited state hadoubly excited character, whereas the first dipole-allowstate is the second excited state.82 In DT, the lowest excitedstate is the one-photon forbidden 21Ag (S1) state and thesecond excited state is the one-photon allowed 11Bu (S2)state @the HOMO~p!→LUMO~p* ! transition#. The1 1Bu (S2) state retains a planar structure.83 The electronicorigin of the S2 state appears in the jet-cooled fluorescenexcitation spectrum at 34 784 cm21 ~4.3 eV!. The electronicorigin of the S1 state in DT has been observed at 28 9cm21 ~3.6 eV! via two-photon fluorescence excitation.83

When the energy gap betweenS2 andS1 is large, the densityof S1 vibronic levels can be very large compared to theciprocal electronic energy spacing and the dark state foan apparently smooth quasicontinuum. Such a situatioknown as the statistical limit for the radiationless transitiproblem. TheS2–S1 energy gap in DT is 5764 cm21 ~0.71eV! placing this large molecule~66 modes! in this statisticallimit. The statisticalversusintermediate cases are discussin more detail in the following paper and references thereThe 22 cm21 bandwidth of theS2 origin places a lower~Lorentzian! limit on internal conversion time scale toS1 ofaround 200–300 fs.84 There is a rapid decreases in spectstructure at;2000 cm21 and no resolvable structure beyon4300 cm21 above theS1 origin. The lifetime ofS1 decreasesdrastically beyond 3000 cm21 above the origin, and is attributed totrans-cis isomerization.85–87 In comparison with oc-tatetraene, the methyl end groups increase the state deand, via interaction between the methyl torsion and lofrequency skeletal modes of the chain, likely lead to acrease in planarity and an increase in the internal converrate.~The torsional barrier is significantly reduced inS1 rela-tive to S2 .! It is well known that nonplanarity leads to extremely fast internal conversion in butadiene ahexatriene.88

In Table I we show thep molecular orbital occupancyelectronic origins~in eV!, and semi-empirical weights foneutral and cation electronic states of DT. The groustates of the neutral and cation were calculated viaB3P86/6-31G* and CASSCF~7,8!/6-31G* methods.CASSCF methods were also used for the calculation of2 1Ag (S1) and 11Bu (S2), as well as excited cationD1 andD2 states. The B3P86/6-31G* method was also used to caculate some higher cation excited states. The weights wobtained from the semi-empirical QCFF/PI1CISD methodin which all singly and doubly excited configurations in thp-orbital 434 MO space were used.89,90 It can be seen thathe S2 state correlates, upon removal of the highest lyielectron, with theD0 ground state of the cation, whereas tS1 state correlates predominately with theD1 excited state ofthe cation. Thus we can see that DT provides an examplthe case of complementary ionization correlations. We nas discussed in more detail below, that the geometry chanbetweenS2 and S1 and betweenD0 and S1 are significant~there is bond order inversion betweenS1 andS0!. This usu-ally leads to large amplitude motion and long vibration

e-ons., single3re


TABLE I. Molecular orbital configurations, semi-empirical weights, and term energies inall-trans 2,4,6,8-decatetraene. The optically bright stateS2 ~4.3 eV! state is singly excited, whereas the lower lying dipolforbidden excitedS1 ~3.6 eV! state results from the interaction of both singly and doubly excited configuratiThe last column shows the cation electron configuration expected upon a Koopmans’-type single-photonactive electron ionization. It can be seen thatS2 correlates with theD0 cation ground state configuration at 7.eV, whereasS1 correlates predominantly with theD1 cation excited state configurations at 8.5 eV. It is therefoexpected that the photoionization electronic channel should switch fromD01e2 to D11e2 during theS2

→S1 internal conversion. For details, see the text.

Electronic Energy Molecular orbital occupancyQCFF/PI1CISD Correlated

state ~ev! 1au 1bg 2au 2bg 3au 3bg weight ion state

Neutral

S0 , 1 1Ag 0.0 2 2 2 2 89% D0 , 1 2Bg

S1 , 2 1Ag 3.6 2 2 2 0 2 35% D1 , 1 2Au

2 2 1 2 1 19% D1 , 1 2Au

2 2 2 1 0 1 19% D0 , 1 2Bg

S2 , 1 1Bu 4.3 2 2 2 1 1 95% D0 , 1 2Bg

Cation

D0 , 1 2Bg IP57.3 2 2 2 1 96%D1 , 1 2Au 8.7 2 2 1 2 53%

2 2 2 0 1 41%D2 , 2 2Au 9.6 2 2 1 2 38%

2 2 2 0 1 43%D3 , 2 2Bg 9.7 2 1 2 2 56%

2 2 2 0 0 1 20%

py

one

l-atghd

hth

fs-umb

traecnre

eos

erie

ge.

hethreto-ese

sedum

e-

that

eri-

m-the

progressions. It is of interest to see how these effects hamthe desired disentangling of electronic from vibrational dnamics.

EXPERIMENT

The molecular beam pump–probe photoelectrphotoion spectroscopy technique used in these experimhas been previously described.34 Briefly, synchronized psNd:YAG oscillator pulses~phase-locked to a fs Ti:Sa oscilator! are regeneratively amplified at a 20 Hz repetition rto the 100 mJ level. After frequency doubling, the hipower ps 532 nm pulses are used to pump three prismcell amplifier chains. The first chain providesmJ level 75 fsTi:Sa pulses at 765 nm, which are used to generate two wlight continua. These are independently amplified inother two dye chains to provide broadly tunable 60–100pulses in the visible, with;0.5 mJ energies after compresion. Harmonic generation from these pulses gave the pand probe pulses required for the experiments. A variatemperature pulsed molecular beam valve (T,200 °C) wasused to introduce the seeded molecular beam into the extion region of a dual flight tube photoelectron-photoion sptrometer. As illustrated by Fig. 2, both ions and electrocould be monitored as a function of time. Isomerically puall-trans 2,4,6,8 decatetraene~DT! was synthesized from thWittig reaction between hexadienal and crotyltriphenylphphonium bromide.78

DT was sublimed at 30 °C into 600 Torr of He carrigas. This temperature was high enough to obtain sufficvapor pressure~;0.5 Torr! yet low enough to avoid rapidpolymerization of the sample. The sample was exchandaily and there was no evidence of clusters in the beam

er-

-nts

e

ye

itees

ple

c--s

-

nt

d

The 100 fs, 2mJ pump pulse was centered at 287 nm, telectronic origin of theS2 state. The probe pulse wavelengwas varied in order to explore the role of different ion coelectronic continua on the form of the pump–probe phoelectron spectra. The two probe wavelengths used in thexperiments were 235 nm~250 fs, 1mJ! and 352 nm~100 fs,3 mJ!. Both pump and probe laser pulses were recompresvia fused silica prism pairs before entering the vacuchamber.

The absoluteDt50 and laser cross correlation were dtermined via nonresonant 111 ionization of NO and 112ionization of Xe. Using af /80 spherical aluminum mirror tofocus the pulses in the interaction region, we estimatethe pump laser intensity was;1011W/cm2, whereas theprobe laser intensity was;1010W/cm2. At these intensitiesand frequencies, the ponderomotive broadening91 of the pho-

FIG. 2. A depiction of the time-resolved photoelectron spectroscopy expment. A molecular beam is crossed by copropagating pump~excitation! andprobe~ionization! fs laser pulses in the interaction region of the spectroeter. Both ions and photoelectrons may be collected as a function ofpump–probe delay timeDt.

thot5toet

o-

eo

ththewloheer

po

on-on-

for

r

eonsn.

he

ill

s-e

o-la-

iginbe-ck–ion

foal

ctra

ion

ronbe


toelectron spectra is less than the effective laser bandwidThe photoelectron kinetic energies were calibrated by phionizing NO via three-photon transitions at 287 nm and 3nm or a two-photon transition at 235 nm. The MCP detecsubtended an;8° solid angle for electron collection. Thpump and probe laser polarizations were each parallel toelectron time-of-flight axis for all experiments. The photelectron spectra were averaged over 104 laser shots with a200 MHz multichannel scaler and background signals duthe pump and probe laser alone were subtracted. The phelectron spectrometer resolution was;100 meV at 1 eV.

RESULTS AND DISCUSSION

Cation electronic structure

To test our idea of using the electronic structure ofionization continuum as a template, we first need to knowelectronic level spacings and symmetries for the first fexcited states of the DT cation. Unfortunately, to our knowedge, there is no spectroscopic information available abthe low lying electronic states of the DT cation or even tionization potential of DT. With a combination of one-lasmultiphoton photoelectron spectroscopy andab initio calcu-lation, we were able to determine these. The ionizationtential was measured by two-photon nonresonant~l5289.5nm! ionization from the neutral ground state to be 7.360.05

FIG. 3. ~a! Calculated CASSCF-DFT Franck–Condon progressionthe S2→D0 transitions in DT, convoluted with a 75 meV instrumentfunction. The structure is dominated by the 0-0 transition.~b! CalculatedCASSCF-DFT Franck–Condon progression for theS1→D1 transitions inDT, convoluted with a 75 meV instrumental function. With scaled~30.8!displacement parameters; the result is shown as a dashed line.

s.o-2r

he

toto-

ee

-ut

-

eV, as discussed below. Two ion excited state vertical iization potentials were determined via above threshold iization ~ATI ! using three-photon transitions.

The calculated B3P86/6-31G* ionization potential~IP!is 7.12 eV. Using CASSCF geometries and force fieldsS1 and S2 and DFT ~density functional theory! geometriesand force fields forD0 and D1 , we were able to calculatedisplacement parameters92 and Franck–Condon structure fothe S2→D0 band, shown in Fig. 3~a!, and for theS1→D1

band, shown in Fig. 3~b!. To match the resolution at thkinetic energy expected in our experiments, the calculatiare convoluted with a 75 meV instrumental width functioThe relatively small geometry changes upon ionization ofS2

relative to D0 involve four ag modes~expansion of C5Cbonds and contraction of C–C bonds!. The S2→D0 band isstrongly dominated by the 0-0 transition. By contrast, tS1→D1 band involves a much larger geometry change~S1

has bond order inversion with respect toS0!. We expectedthat the combination of DFT and CASSCF geometries woverestimate92 the Franck–Condon activity of theag modesfor the S1→D1 band. Therefore the result with scaled diplacement parameters~30.8! is shown by the dashed linand is compared with experiment below.

The observednonresonant two-photon photoelectronspectrum is compared with calculation in Fig. 4. In Fig. 4~a!we show the Franck–Condon activity betweenS0 and D0 ,showing two different resolutions~2.5 meV and 100 meV!.In Fig. 4~b! we show the experimental nonresonant twphoton photoelectron kinetic energy spectrum. Both calcution and experiment are characterized by a strong orband, supporting the calculated small geometry changetween the neutral and cation ground states. Weak FranCondon activity is revealed in the calculated high resolut

r

FIG. 4. One-color nonresonant two-photon ionization photoelectron speof DT. The fs UV pulse was centered at 289.5 nm~4.28 eV!, just below theS2 band origin.~a! Calculated CASSCF-DFT Franck–Condon progressfor the S0→D0 transitions, at two different resolutions~2.5 meV and 100meV!. ~b! The one-color, nonresonant two-photon ionization photoelectspectrum of DT at 289.5 nm. The ionization potential was measured to7.3 eV.

wd.ma

befsk–ond

onin

en

eethe

esoni

e2

all

nn

ha

ing

doneob-olor

re,

ve-yur-

to

ton,

n-h

a-ionza-lec--

ingion

ofllyns

ashisss-

ightnt

ed.ee ofen-

allyl

pe

on

th,

FTind 5.


spectrum via short progressions in theag symmetry C5Cand C–C stretches. Very weak activity connected to lofrequency CCCbu symmetry bending vibrations is predicteThe ionization potential is estimated to be 7.12 eV. To copare with experiment, the 2.5 meV resolution spectrum wbroadened to 100 meV, shown as the dashed line in Fig. 4~a!.The experimental ionization potential was determined to7.3 eV and the overall width and shape of the spectra arreasonable agreement. The relatively small 170 meV ofin the ionization potential, together with the similar FrancCondon profile, gave us confidence in using the calculatito help assign the three-photon photoelectron spectrum,cussed below.

In Fig. 5~a! the experimental three-photon photoelectrspectrum is plotted as a function of kinetic energy, revealboth ionization into the ion ground state (ek,1.26 eV) andother very weak bands at higher kinetic energy (2,ek

,5 eV), signatures of ATI processes. To improve theergy resolution over this range, a retardation voltage~1 V!was applied. This spectrum is shown in Fig. 5~b!, plotted asfunction of the electron binding energy, assuming a thrphoton ionization. The two maxima observed suggestpresence of two cation excited states at 8.55 eV and 9.67respectively. The electronic symmetry expected for thtwo states is2Au since they are populated by a three-phottransition from the ground state. The calculations shownTable II suggest a low lying2Au state at 1.57 eV above thion ground state and a trio of near-degenerate states ateV, of which one is of2Au symmetry. Therefore, two2Au

excited state origins of the DT cation are predicted to lie8.69 eV and 9.69 eV, respectively. The two experimentaobserved peaks are in good agreement. TheD1 andD2 statesof the cation consist of the same electronic configuratiothe first being the symmetric and the second being the asymmetric combination. It should be noted, however, t

FIG. 5. One-color nonresonant multi-photon ionization photoelectron stra of DT. The fs UV pulse was centered at 289.5 nm~4.28 eV!, just belowthe S2 band origin.~a! Photoelectron spectrum revealing both two-photionization, as in Fig. 4, and three-photon above-threshold ionization~ATI !.~b! The three-photon ATI photoelectron spectrum, showing transitions tofirst D1 ~at 8.55 eV! and secondD2 ~at 9.67 eV! 2Au cation electronic statesin agreement with theab initio results given in Table I.

-

-s

einet

sis-

g

-

-eV,e

n

.57

ty

s,ti-t

due to poor energy resolution, weak signals, and overlappbands~around 9.7 eV!, it is difficult to comment on the bandshapes, their relative intensities, and their Franck–Conactivity. In Table II we summarize the origin bands of thion states predicted and the assignment of the bandsserved in photoelectron spectra obtained in these one-cmultiphoton ionization experiments.

Pump–probe experiments at 10.6 eV total energy

In all pump–probe experiments on DT presented hethe pump pulse excited theS2 ~0,0! electronic origin at 287.5nm. The first pump–probe experiment had a probe walength ~235 nm! sufficient to give 2.3 eV excess energabove the ionization threshold. Therefore, dynamics occring in the coupledS2–S1 neutral states were projected onboth the ion ground (D0,12Bg) and first excited (D1,12Au)electronic states. As suggested by Table I, for single-phosingle active electron emission, theS2 state correlates withD0 ground state producing photoelectrons with kinetic eergy e1 , whereas theS1 state correlates predominately witthe D1 excited state~at 8.55 eV!, yielding photoelectron ki-netic energye2 . This is the case of complementary ioniztion correlations, depicted in Fig. 1. For DT, this assumptis supported by theoretical work suggesting that the ionition energies used here are low enough for independent etron approximations93 and confirmed by our experimental results, below. Therefore, as the nonadiabatic couplproceeds, we expected a switching of the photoionizatelectronic channel. We expect that transitions to theD2 state~at 9.67 eV! are difficult to see due to the combinationpoor Franck–Condon overlap with the highly vibrationaexcitedS1 state and poor transmission of very slow electro(ek,200 meV) through our spectrometer.

In Fig. 6 we show the measured DT parent ion signalfunction of time delay between pump and probe pulses. Tsignal shows the rise time expected from the laser crocorrelation~Gaussian FWHM 290620 fs! and a steplike be-havior at positive time delays. There appears to be a sloverall decay of the parent ion signal, with time consta9606300 fs. We note that no fragment ions were observThe lifetime of the zeroth orderS2 state was expected to baround 0.5 ps. As might be expected, the time dependencthe total parent ion signal in this experiment appears inssitive to the electronic symmetry change induced byS2–S1

internal conversion.By contrast, as shown in Fig. 7~b!, the time-resolved

photoelectron kinetic energy spectrum shows a dramaticdifferent result. Figure 7~a! shows the relevant energy leve

c-

e

TABLE II. The vertical ionization energies predicted by CASSCF and Dmethods~see the text for details! and assignment of the peaks observedthe one-color multiphoton ionization photoelectron spectra of Figs. 4 an

D0 1 2Bg D1 1 2Au D2 2 2Au D3 2 2Bg D4 3 2Bg

Theory~eV!

7.118~I.P.!

8.69 9.62 9.71 9.91

Expt.~eV!

7.29 8.55 9.67 ¯ ¯

m

-idd

n--es

wd:tr

thio

is

vth

sion

o

atgapall

re-

dseur

turerydy-tionkettndatee

. T

als

n wths.

lec-

ser,

c-nm.

to

tionand


scheme. We expect a shift in photoelectron spectrum frohigher energy band (e1<2.4 eV) to lower energy band (e2

<1.1 eV) uponS2–S1 internal conversion. The photoelectron spectra of Fig. 7~b! are indeed characterized by a rapshift from a band centered at;2.4 eV to a broad, structurelow energy band ranging from;1.6 eV to 0 eV. This rapidshift is thedirect signature of the changing electronic cofigurations induced byS2–S1 nonadiabatic coupling. The energetic component (e1'2.4 eV) is readily assigned to thphotoionization ofS2 into D0 . The broad low energy band iassigned to photoionization ofS1 into D1 and roughly agreesin width with the scaled Franck–Condon calculation shoin Fig. 3b. As shown in Fig. 8, integration of these two bandirectly yields the S2–S1 internal conversion times scale402665 fs. This clear result obtains despite the geomechanges involved.

These results demonstrate the strong selectivity ofionization dynamics for cases of complementary ionizatcorrelations and, importantly, show thatdisentanglingofelectronic population dynamics from vibrational dynamicspossible during ultrafast nonadiabatic processes.48 If vibra-tional wave packet dynamics were present inS2 ~e.g., byexciting above the electronic origin! they would be revealedby the vibrational structure in thee1 band. Likewise, vibra-tional wave packet dynamics inS1 is seen in thee2 band.The electronic populations dynamics accompanying thebrational dynamics are synchronously observed viaswitching frome1 to e2 .

The decaying part of thee2 band ~7206240 fs! and,similarily, that of the ion parent signal~Fig. 6, 9606300 fs!are likely the signature of the subsequent internal converfrom S1 to S0 . We expect relatively poor Franck–Condofactors for ‘‘hot’’ ground stateS0 ionization since the probepulse gives a maximum of 2.3 eV excess energy in the ctinuum, whereas the nascent ‘‘hot’’S0 ground state would

FIG. 6. Excited state dynamics at theS2 origin of all trans decatetraene~DT! as measured by time-resolved photoionization mass spectrometryprobe ionization laser~235 nm! had sufficient energy to access both theD0

ground andD1 first electronic states of the cation. Parent ion sign(C10H14

1 ) are shown as function of time delay between pump~287.5 nm! andprobe pulses. No fragments were observed. The laser cross-correlatiotL5290620 fs. An essentially steplike behavior is seen, showing thation signal in this experiment does not reveal the nonadiabatic dynamic

a

ns

y

en

i-e

n

n-

contain up to 4.3 eV of vibrational energy. We note ththere is a nonzero signal near 1.6 eV, seemingly in thebetween the two relevant ion states. This is due to smdeviations from our simple Koopmans’ picture and corsponds toS1 photoionization into theD0 continuum.

A central point in this paper is that photoelectron banin Fig. 7 contain much more information than simply th‘‘rate constant’’ of the nonadiabatic process. Even with olow energy resolution, we can see some vibrational strucin thee2 photoelectron band. This, in principle, contains vedetailed information about the state-to-state vibrationalnamics which promotes and tunes the electronic populatransfer, as well as the ensuing IVR which the wave pacundergoes on the ‘‘dark’’S1 potential surface. We note thathe S2 state is prepared here at its vibrationless origin atherefore any vibrational dynamics observed must originfrom theS1 state. All signals decay to zero on a longer tim

he

ase

FIG. 7. Time-resolved photoelectron spectra revealing vibrational and etronic dynamics during internal conversion inall trans decatetraene~DT!.~a! Level scheme in DT for one-photon probe ionization. The pump laprepares the optically bright stateS2 . Due to ultrafast internal conversionthis state converts to the lower lying stateS1 with 0.7 eV of vibrationalenergy. The expected ionization propensity rules are shown:S2→D0

1e2(e1) andS1→D11e2(e2). ~b! Femtosecond time-resolved photoeletron kinetic energy spectra of DT pumped at 287 nm and probed at 235There is a rapid~;400 fs! shift in the distribution: from (e1) an energeticpeak at 2.5 eV due to photoionization ofS2 into the D0 cation groundelectronic state; to (e2) a broad, structured band at lower energies duephotoionization of vibrationally hotS1 into theD1 cation first excited elec-tronic state. These results show a disentangling of electronic populadynamics from vibrational dynamics. The structure in the low energy breflects the vibrational dynamics inS1 .

ouioroa

rg-isa

s

nsl-

tio

ew6

rgeiaratthathtestiaav

n-er-gthm

at

ar-ed.

cayan-

ionon

e to

inrgy


scale, presumably due toS1–S0 internal conversion. Al-though we are not able to extract detailed information abthe vibrational dynamics at the present time, a first indicatthat one might be able to do so eventually can be seen fFig. 9, where we have replotted the photoelectron spectrFig. 7, clearly showing the changing shape of thee2 band asa function of time. We see a general shift from higher eneto lower energy electrons~i.e, higher cation vibrational energy! within this band, as a function of time. This shiftanalogous to a time-dependent fluorescence red shiftsuggests that we are observing an ongoing IVR procesthe ‘‘dark’’ S1 potential energy surface.

Many photoinduced polyatomic unimolecular reactioare based on vibrationally ‘‘hot’’ molecules formed via utrafast internal conversion.94 A fundamental approximationin unimolecular reaction rate theory is that of the assumpof statistical energy redistribution due to IVR~i.e., the com-plete sampling of phase space! being very fast comparedwith reaction.95 Many models and interpretations are bason the supposition of fast, complete IVR. Increasingly, hoever, this assumption has come into question. DT hasinternal degrees of freedom and 0.7 eV of internal eneafter conversion toS1 and is therefore expected to be in th‘‘statistical limit.’’ Nevertheless, we are able to observe, vthe photoelectron spectrum, some evolution in the vibtional energy distribution on theS1 surface. We believe thaour experimental method could present new views ofextent of and time scale for the onset of statisticality inisolated, energized molecule. The detailed extraction ofvibrational wave packet dynamics associated with this innal conversion and its subsequent vibrational energy redibution relies on modeling the multi-dimensional nonadbatic coupling and calculating overlaps between this w

FIG. 8. Electronic population dynamics in DT at theS2 origin. ~a! Theenergy integratede1 photoelectron band~1.8 eV–3.2 eV! of Fig. 7, plottedas a function of time delay, showing the decay of theS2 state. An exponen-tial fit directly reveals the internal conversion rate: 402665 fs. ~b! Theenergy integratede2 photoelectron band~,1.8 eV! of Fig. 7, plotted as afunction of time delay, showing the formation and~on a longer time scale!decay of theS1 state. This signal rises in 400 fs, due to internal conversfrom S2 and decays;700 fs, presumably due to subsequent internal cversion toS0 .

tnmof

y

ndon

n

d-6y

-

ener-ri--e

packet and the relevant states of the ion~work presently inprogress!.

Pump–probe experiments at 7.8 eV total energy

To confirm these ideas about the selectivity of the ioization continuum for specific electronic symmetries, we pformed a second experiment by tuning the probe wavelento 352 nm to yield lower excess energy in the continuu~now 0.54 eV!. This is below theD1 threshold and thereforeallowing one-photon ionization into theD0 ground ion stateonly. In contrast with the results of Fig. 6, we predicted ththe parent ion signal would now decay in;400 fs since theformed S1 state should not correlate favorably with theD0

continuum. In Fig. 10 we show that the time-dependent pent ion signal for 352 nm ionization is indeed well describby a decay rate of 410615 fs, confirming the above resultsThe fact that the parent ion signal in Fig. 10 does not decompletely to zero indicates that there remains a small tr

-

FIG. 9. Time-resolved photoelectron spectra from Fig. 7, replotted herclearly show the changing shape of thee2 band (ekin,1.5 eV) as a functionof time. A shift from higher energy to lower energy photoelectrons withthee2 band as a function of time is suggestive an ongoing vibrational eneredistribution process on the ‘‘dark’’S1 potential energy surface.

th

n

nselelyy-

asw

up

umrigyenthn

ili

nheecb

wo-

y,

ne

ontic

then,omleiza-

s in

on-

erle-

ove

umhe

y.

istioati

ent,ser

s375

s to


sition probability betweenS1 and D0 . The poor Franck–Condon factors associated with this transition are due to0.54 eV vibrational energy available inD0 , as comparedwith the 0.7 eV vibrational energy inS1 , as well as signifi-cant geometry changes betweenS1 andD0 . Comparison ofthe time-dependent parent ion signals at 235 nm with 352probe laser wavelength clearly demonstrates that the formthe parent ion signal dependsstronglyon the specific photo-ionization dynamics and, in order to avoid misleading coclusions, must be analyzed for each specific case. Thissitivity of the integrated parent ion signal to the molecuspecific photoionization dynamics was previousdiscussed54 in a study of the nonadiabatic dissociation dnamics of (NO)2 .

In contrast with the 235 nm probe experiment, the mspectrum at 352 nm now shows fragmentation. In Fig. 11show the time dependence of the main fragment ion C9H11

1

~m/e5119!, which corresponds to loss of a methyl end grofrom DT1. It is important to note that theD0 , S2 , andS1

states are all energetically stable at these energies of pand probe excitation and therefore the dissociation must afrom a two-photon probe transition, yielding a total enerof 11.3 eV. We note that the pump laser intensity was idtical to that in the 235 nm probe experiment. However,probe laser intensity was approximately an order of magtude higher at 352 nm, leading us to consider the possibof above threshold ionization~ATI !. A probe laser powerstudy supported this point, yielding a quadratic dependefor the C9H11

1 fragment ion and a linear dependence for tparent ion signal. This is confirmed by photoelectron sptroscopy results, discussed below. Interestingly, it canseen that the C9H11

1 signal begins at almost zero nearDt50 and rises with a time constant of 3756120 fs. This

FIG. 10. Excited state dynamics at theS2 origin of all trans decatetraene~DT! as measured by time-resolved photoionization mass spectrometrcontrast with Fig. 6, the probe ionization laser~352 nm! could access onlythe ground electronic state of the cation. Parent ion signals (C10H14

1 ) areshown as function of time delay between pump~287.5 nm! and probepulses. The signal risetime~the laser cross-correlation half-width! was tL

58464 fs. The parent ion signal decays with a time constant of 410615 fs,confirming the results of Fig. 8. A behavior very different from Fig. 6seen, showing the crucial important of understanding the photoionizadynamics in time-resolved ionization experiments involving nonadiabdynamics.

e

mof

-n-

-

se

pse

-ei-ty

ce

-e

strongly suggests that the fragmentation arises from tphoton ionization of the vibrationally excitedS1 state formedby the internal conversion andnot from two-photon ioniza-tion of S2 . This is confirmed by photoelectron spectroscopas discussed below.

In Fig. 12~b! we show the time-resolved photoelectrospectrum for 352 nm probe laser ionization. Initially, thspectrum is characterized by a low energy band,e1 at 0.56eV which decays with time. As indicated in Fig. 12~a!, thee1

band is due to one-photon ionization ofS2 into D0 and cor-responds exactly with the 235 nme1 band of Fig. 7, simplyshifted to lower energy by the reduction in probe photenergy. This peak is narrower due to the improved kineresolution at low energy. A broad energetic band,e2 , rang-ing from 0.6 eV to 4 eV grows with time as thee1 banddecays and therefore arises from photoionization offormedS1 state. Thee2 band must, via energy conservatioarise from two-photon probe ionization. As can be seen frFig. 12~a!, due to the symmetry of the two-photon dipooperator, the ion continua accessed via two-photon iontion may also includeD0 , D3 , and D4 . This explains thebroad range and high kinetic energy of the photoelectronthe e2 band. Integration of thee1 band, shown in Fig. 13~a!,provides yet another independent confirmation of theS2–S1

internal conversion time scale, 377647 fs, fully in agree-ment with the previous results. The decaying part of thee2

component integration~7836109 fs!, Fig. 13~b!, as in Figs. 6and 7, is likely the signature of the subsequent internal cversion between theS1 andS0 states.

It is interesting to consider why, at invariant probe lasintensity, the photoionization process switches from singphoton ionization ofS2 to two-photon ionization ofS1 . Wenote that in the both cases, the first photon is already abthe ionization potential and therefore theS1 ionization is dueto absorption of a second photon in the ionization continu~ATI !. This can be rationalized by a consideration of t

In

nc

FIG. 11. A time-resolved photoionization mass spectrometry experimshowing a fragment ion signal C9H11

1 , due to two probe photon ionizationwhich was determined simultaneously with the data of Fig. 10. The lacross-correlation half-width wastL58464 fs. The fragment ion signal risemuch more slowly than the parent ion signal, with a time constant of6120 fs. This suggests that it derives from two-photon ionization of theS1

state formed by internal conversion. On longer time scales, it appeardecay~;2 ps! presumably due to internal conversion ofS1 to S0 .

a

lyr t

foinghere. Iffts

rateit-

oly-ionthe

nndfhed inketWe

aly-averuc-

tron-m-

c-itinghesed

-lta-

rna

m

ded

of

s.otoonni

thete:


relative rates of two competing processes: second photonsorptionversusautoionization. For the case ofS2 , the photo-ionization correlation is withD0 and therefore the ionizationis direct. In other words, the ‘‘autoionization’’ is extremerapid and second photon absorption cannot compete. Focase ofS1 , the photoionization correlation is withD1 . TheD1 state, however, is energetically inaccessible and therethe transition is most likely into Rydberg series convergon theD1 threshold. For these to emit an electron into topenD0 continuum channel, there must be an electronicarrangement, for which there is a finite autoionization ratethis case, the absorption of a second photon competes etively with autoionization. These two-photon experimen

FIG. 12. Time-resolved vibrational and electronic dynamics during inteconversion forall trans decatetraene~DT! via two-photon ionization.~a!Level scheme in DT for one- and two-photon probe ionization. The pulaser is identical to that in Fig. 7 and prepares the sameS2 state wavepacket. The expected ionization propensity rules are:S2→D01e2(e1) forone-photon ~u↔g, as in Fig. 7! ionization and S1→→D0 , D3 , D4

1e2(e2) for two-photon (g↔g) ionization.~b! Femtosecond time-resolvephotoelectron kinetic energy spectra of DT pumped at 287 nm and prob352 nm, using both one- and two-photon probes. At 352 nm, theD1 ionstate is not energetically accessible from theS1 state via a single-photontransition. Confirming the results of Fig. 7, there is a rapid shift~'400 fs! inthe distribution: from (e1) a peak at 0.4 eV due to one-photon ionizationS2 into the D0 cation ground electronic state; to (e2) a broad, structuredband at higher energies~1–3.5 eV! due to two-photon ionization of thevibrationally hotS1 into theD0 cation ground and excited electronic stateThe photoionization channel switches from a one-photon to a two-phprocess during the internal conversion indicating again that the electrstructure of the ionization continuum is selective of the evolving electrosymmetry in the neutral state.

b-

he

re

-nec-

not only confirm the one-photon results, but also demonstthe symmetry selectivity of the photoionization processself.

CONCLUSION

Nonadiabatic processes, ubiquitous in excited state patomic dynamics, lead to complex wave packet evolutand strongly varying transition dipole moments due tomixing of electronic and vibrational degrees of freedom. Aimportant goal of our laboratory is to try to use femtosecowavepacket methods todisentanglethese coupled degrees ofreedom in order to recover a picture of the evolution of tzeroth order Born–Oppenheimer states. We consideredetail the nature of the final state in polyatomic wave pacdynamics experiments using photoionization detection.have argued that the molecular ionization continuum isvery interesting final state for the study of nonadiabatic poatomic wavepackets. The vibrational aspects of the wpacket dynamics can be observed via the vibrational stture of the ionization continuum~i.e., vibrational states of thecation!. The electronic structure of the continuum~i.e., theset of electronic states of the cation and the free elecpartial wave structure! is sensitive to the electronic population dynamics~based on electronic propensity rules and symetry properties of the free electron partial waves!. We in-vestigated the role of electronic continua~the cationelectronic states! in femtosecond time-resolved photoeletron spectroscopy measurements and considered two limcases of Koopmans’-type photoionization correlations. Tfirst case, complementary ionization correlations, discusin this paper, is one in which the coupled states~e.g.,S2 andS1! correlate todifferent cation electronic states. We suggested that this favorable situation can allow for a simu

l

p

at

nicc

FIG. 13. Electronic population dynamics in DT at theS2 origin via two-photon ionization.~a! The energy integratede1 photoelectron band~,0.56eV! of Fig. 12, plotted as a function of time delay, showing the decay ofS2 state. An exponential fit directly reveals the internal conversion ra377647 fs, confirming the previous results.~b! The energy integratede2

photoelectron band~0.56 eV–4 eV! of Fig. 12, plotted as a function of timedelay, showing the formation and~on a longer time scale! decay of theS1

state. This signal rises in 400 fs, due to internal conversion fromS2 andagain decays;700 fs, confirming the previous results.

ic

nle

-o

.ra

s t-egerv

err

-dinntdi

tronta

secto

t ic

ro

ene

roree

weth

thtele

trn

ia-

in-weics

theics

ledder-to-

ute

cke,T.

-

nce

-

sed

u.

,

. Z.

nd

n-

, J.

m.

ys.


neous monitoring of both the electronic population dynamand the coupled vibrational wave packet dynamics.48 Thesecond limiting case, corresponding ionization correlatiodiscussed in the following paper, is one in which the coupstates~e.g.,S2 andS1! correlate to thesamecation electronicstates.

We studied ultrafastS22S1 internal conversion dynamics in all trans 2,4,6,8-decatetraene as an examplecomplementary ionization correlations. TheS2 state is singlyexcited and correlates with theD0 ground state of the cationTheS1 state, by contrast, contains doubly excited configutions and correlates preferentially with theD1 first excitedstate of the cation. By pumping the molecule to theS2 originand then probing with a photon sufficiently energetic so ainclude both theD1 andD0 cation states, we obtained timeresolved photoelectron spectra which revealed the changelectronic symmetry implicit in internal conversion, allowina direct determination of the internal conversion timscale—400 fs. Furthermore, these spectra allowed obsetion of vibrational wave packet dynamics on the ‘‘dark’’S1

potential energy surface.We confirmed these results with an independent exp

ment using two-photon probe laser ionization, where the fiprobe photon was above theD0 ionization threshold but below the D1 state. Again, the same picture emerges andirect measurement of the internal conversion time obtaWe also demonstrated that although the time-resolved igrated parent ion signals in pump–probe studies of nonabatic dynamics may have very different forms~depending onthe specifics of the photoionization dynamics!, they can bereadily interpreted in terms of their associated photoelecspectra. The measurement of integrated parent ion sigalone may be potentially misleading for cases of excited spolyatomic nonadiabatic dynamics.

In the following paper, we consider the unfavorable caof corresponding ionization correlations in which we expthe disentangling of electronic from vibrational dynamicsbe more challenging. In these cases, geometry changes~dis-placements! become an important aspect of the problem. Ialso interesting to study the other component of the molelar ionization continuum, the free electron. As photoelectangular distributions ~PADS! are directly sensitive tochanges in excited state electronic symmetry, experimare presently underway both in our laboratory and elsewhto investigate this possibility.

We hope that the time-resolved photoelectron spectcopy method will prove useful in many situations whethere are significant charge rearrangements during thecited state dynamics. In our current experimental work,are applying time-resolved photoelectron spectroscopy tocited state proton transfer, yielding both time scales andenergy dependencies of the intramolecular processes.96 In thearea of molecular electronics, azobenzene serves asmodel of an ultrafast molecular switch. We have studiednonadiabatic intramolecular dynamics of azobenzene, demining time scales and shedding new light on the compprimary photophysical processes.97 Of photochemical inter-est, the excited states of aldehydes and ketones, by conwith polyenes studied here, can have important carbo

s

s,d

f

-

o

in

a-

i-st

as.e-a-

nalste

et

su-n

tsre

s-

x-ex-e

theer-x

astyl

p-p* /n-p* interactions. We have examined the nonadbatic dynamics in a series of dieneones,98 showing the im-portance of inductive effects on the competition betweenternal conversion and intersystem crossing. In the future,plan to investigate excited state electron transfer dynamusing time-resolved photoelectron spectroscopy, withhope of directly observing the coherent vibrational dynamassociated with the charge transfer.99 Together with the workof many other laboratories, we are optimistic that a detainew view of the complex nonadiabatic processes that unpin our understanding of photochemistry, material phosciences, and molecular electronics will emerge.

ACKNOWLEDGMENTS

We thank P. Arya and M. Barnes of the Steacie InstitChemical Biology Program for the synthesis ofall-trans2,4,6,8 decatetraene. We thank T. Seideman, W. DomW. J. Buma, B. S. Hudson, M. Schmitt, J. P. Shaffer,Schultz, and J. G. Underwood for helpful discussions.

1J. Michl and V. Bonacic-Koutecky,Electronic Aspects of Organic Photochemistry~Wiley, New York, 1990!.

2R. W. Schoenlein, L. A. Peteanu, R. A. Mathies, and C. V. Shank, Scie254, 412 ~1991!.

3J. Jortner and M. A. Ratner,Molecular Electronics~IUPAC, BlackwellOxford, 1997!.

4G. Herzberg and H. C. Longuet-Higgins, Faraday Discuss.35, 77 ~1963!.5W. Siebrand, J. Chem. Phys.47, 2411~1967!.6M. Bixon and J. Jortner, J. Phys. Chem.48, 715 ~1968!.7J. Jortner, S. A. Rice, and R. M. Hochstrasser, Adv. Photochem.7, 149~1969!.

8B. R. Henry and W. Siebrand,Radiationless Transitions in Organic Molecular Photophysics, Vol. 1, edited by J. B. Birks~Wiley, London, 1973!.

9K. F. Freed, inRadiationless Processes in Molecules and CondenPhasesedited by F. K. Fong~Springer, Berlin, 1976!, p. 23.

10G. Stock and W. Domcke, Adv. Chem. Phys.100, 1 ~1997!.11G. Herzberg,Molecular Spectra and Molecular Structure, Vol. 3~Van

Nostrand, 1945, New York!.12D. R. Yarkony, Acc. Chem. Res.31, 511 ~1998!.13J. Kommandeur, W. A. Majewski, W. L. Meerts, and D. W. Pratt, Ann

Rev. Phys. Chem.38, 433 ~1987!.14R. Schinke,Photodissociation Dynamics~Cambridge University Press

Cambridge, 1993!.15L. J. Butler, Annu. Rev. Phys. Chem.49, 125 ~1998!.16G. Orlandi, F. Zerbetto, and M. Z. Zgierski, Chem. Rev.91, 867 ~1991!.17W. Siebrand and M. Z. Zgierski, J. Chem. Phys.71, 3561~1979!.18R. J. Sension, R. J. Boudryushi, B. S. Hudson, F. Zerbetto, and M

Zgierski, J. Chem. Phys.96, 2617~1992!.19M. K. Lawless, S. D. Wickam, and R. A. Mathies, Acc. Chem. Res.28,

493 ~1995!.20A. H. Zewail, Femtochemistry: Ultrafast Dynamics of the Chemical Bo

~World Scientific, Singapore, 1994!.21T. Baumert, B. Bu¨hler, M. Grosser, R. Thalweiser, V. Weiss, E. Wiede

mann, and G. Gerber, J. Phys. Chem.95, 8103~1991!.22J. C. Owrutsky and A. P. Baronavski, J. Chem. Phys.108, 6652 ~1999!;

ibid. 110, 11206~1999!.23S. A. Trushin, W. Fuss, T. Schikarski, W. E. Schmid, and K. L. Kompa

Chem. Phys.106, 9386~1997!.24H. Schwoerer, R. Pausch, M. Heid, V. Engel, and W. Kiefer, J. Che

Phys.107, 9749~1997!.25R. Lopez-Martens, T. W. Schmidt, and G. Roberts, J. Chem. Phys.111,

7183 ~1999!.26T. Shibata, H. Li, H. Katayanagi, and T. Suzuki, J. Phys. Chem. A102,

3643 ~1998!.27H. Ruppe, S. Rutz, E. Schreiber, and L. Woste, Chem. Phys. Lett.257,

356 ~1996!; 257, 365 ~1996!.28V. Blanchet, M. A. Bouchene, G. Cabrol, and B. Girard, Chem. Ph

Lett. 233, 491 ~1995!.

J

ge

m.

.

x

l.

C.

an

A

tt.

nom.

D.

em.

ec-

-

hys.

R.

L.

R.

J.

s.

sc.

nd


29S. I. Ionov, G. A. Brucker, C. Jaques, L. Valachovic, and C. Wittig,Chem. Phys.99, 6553~1993!.

30V. Vorsa, P. J. Campagnola, S. Nandi, M. Larsson, and W. C. LineberJ. Chem. Phys.105, 2298~1996!.

31U. Marvet and M. Dantus, Chem. Phys. Lett.256, 57 ~1996!.32I. Fischer, D. M. Villeneuve, M. J. J. Vrakking, and A. Stolow, J. Che

Phys.102, 5566~1995!.33I. Fischer, D. M. Villeneuve, M. J. J. Vrakking, and A. Stolow, inFem-

tochemistry: The Lausanne Conference, edited by M. Chergui~World Sci-entific, Singapore, 1996!, p. 43.

34I. Fischer, M. J. J. Vrakking, D. M. Villeneuve, and A. Stolow, ChemPhys.207, 331 ~1996!.

35I. Fischer, D. M. Villeneuve, M. J. J. Vrakking, and A. Stolow, inUl-trafast Phenomena X, edited by P. F. Barbara, J. G. Fujimoto, W. H. Knoand W. Zinth~Springer, Berlin, 1996!, p. 187.

36C. C. Hayden and A. Stolow, inAdv. Series in Physical Chemistry Vo10: Photoionization and Photodetachment, edited by C. Y. Ng~WorldScientific, Singapore, 2000!.

37M. Seel and W. Domcke, J. Chem. Phys.95, 7806~1991!.38D. R. Cyr and C. C. Hayden, J. Chem. Phys.104, 771 ~1996!.39V. Engel, Chem. Phys. Lett.178, 130 ~1991!; Ch. Meier and V. Engel,

ibid. 212, 691 ~1993!; J. Chem. Phys.101, 2673~1994;!; Phys. Rev. Lett.73, 3207~1994!.

40T. Baumert, R. Thalweiser, and G. Gerber, Chem. Phys. Lett.209, 29~1993!.

41A. Assionet al., Phys. Rev. A54, R4605~1996!.42B. J. Greenblatt, M. T. Zanni, and D. M. Neumark, Chem. Phys. Lett.258,

523 ~1996!.43J. M. Smith, C. Lakshminarayan, and J. L. Knee, J. Chem. Phys.93, 4475

~1990!.44X. Zhang, J. M. Smith, and J. L. Knee, J. Chem. Phys.100, 2429~1994!.45C. Lakshminarayan and J. L. Knee, J. Phys. Chem.99, 1768~1995!.46W. Radloff, V. Stert, Th. Freudenberg, I. V. Hertel, C. Jouvet,

Dedonder-Lardeux, and D. Solgadi, Chem. Phys. Lett.281, 20 ~1997!.47V. Blanchet and A. Stolow, inUltrafast Phenomena XI, edited by T.

Elsaesser, J. G. Fujimoto, D. A. Wiersma, and W. Zinth~Springer, Berlin,1998!, p. 456.

48V. Blanchet, M. Z. Zgierski, T. Seideman, and A. Stolow, Nature~Lon-don! 401, 52 ~1999!.

49C. P. Schick, S. D. Carpenter, and P. M. Weber, J. Phys. Chem. A103,10470~1999!.

50V. Sert, W. Radloff, C. P. Schultz, and I. V. Hertel, Eur. Phys. J. D5, 97~1999!.

51M. Schmitt, J. J. Larsen, S. Lochbrunner, J. P. Shaffer, M. Z. Zgierski,A. Stolow, J. Chem. Phys.1206–1213, ~2000!, following paper.

52J. P. Shaffer, T. Schultz, M. Schmitt, M. Zgierski, and A. Stolow~unpub-lished!.

53M. R. Dobber, W. J. Buma, and C. A. de Lange, J. Phys. Chem.99, 1671~1995!.

54V. Blanchet and A. Stolow, J. Chem. Phys.108, 4371~1998!.55B. Kim, C. P. Schick, and P. M. Weber, J. Chem. Phys.103, 6903~1995!.56T. Suzuki, L. Wang, and H. Kohguchi, J. Chem. Phys.111, 4859~1999!.57B. J. Greenblatt, M. T. Zanni, and D. M. Neumark, Science276, 1675

~1997!.58S. Lochbrunner, J. P. Shaffer, M. Schmitt, T. Schultz, M. Zgierski, and

Stolow ~unpublished!.59T. Schultz, J. P. Shaffer, M. Schmitt, M. Zgierski, and A. Stolow~unpub-

lished!.60K. L. Reid, Chem. Phys. Lett.215, 25 ~1993!.61K. L. Reid, S. P. Duxon, and M. Towrie, Chem. Phys. Lett.228, 351

~1994!.62T. Seideman, J. Chem. Phys.107, 7859~1997!.63S. C. Althorpe and T. Seideman, J. Chem. Phys.110, 147 ~1999!.

.

r,

d

.

64K. L. Reid, T. A. Field, M. Towrie, and P. Matousek, J. Chem. Phys.111,1438 ~1999!.

65T. Suzuki, L. Wang, and H. Kohguchi, J. Chem. Phys.111, 4859~1999!.66Y. Arasaki, K. Takatsuka, K. Wang, and V. McKoy, Chem. Phys. Le

302, 363 ~1999!.67S. C. Althorpe and T. Seideman, J. Chem. Phys.~submitted!.68T. Seideman, J. Chem. Phys.~submitted!.69T. Seideman and S. C. Althorpe, J. Electron Spectrosc. Relat. Phe

~submitted!.70M. Aeschlimann, M. Bauer, S. Pawlik, W. Weber, R. Burgermeister,

Oberli, and H. C. Siegmann, Phys. Rev. Lett.79, 5158~1997!.71J. A. Davies, J. E. LeClaire, R. E. Continetti, and C. C. Hayden, J. Ch

Phys.111, 1 ~1999!.72J. H. D. Eland,Photoelectron Spectroscopy~Butterworths, London, 1984!.73J. Berkowitz,Photoabsorption, Photoionization, and Photoelectron Sp

troscopy~Academic, New York, 1979!.74C. A. de Lange, inHigh Resolution Laser Photoionization & Photoelec

tron Studies, edited by I. Powis, T. Baer and C. Y. Ng~Wiley, New York,1995!, p. 195.

75S. T. Pratt, P. M. Dehmer, and J. L. Dehmer, J. Chem. Phys.81, 3444~1984!.

76B. S. Hudson, B. E. Kohler, and K. Schulten,Linear Polyene ElectronicStructure and Potential Surfaces. Excited States 6edited by E. C. Lim~Academic, New York, 1982!.

77G. Orlandi, F. Zerbetto, and M. Z. Zgierski, Chem. Rev.91, 867 ~1991!.78H. Petek, A. J. Bell, K. Yoshihara, and R. L. Christensen, J. Chem. P

95, 4739~1991!.79F. Buda, H. J. M. de Groot, and A. Bifone, Phys. Rev. Lett.77, 4474

~1996!.80M. F. Granville, G. R. Holtom, and B. E. Kohler, J. Chem. Phys.72, 4671

~1980!; H. Petek, A. Bell, Y. S. Choi, K. Yoshihara, B. A. Tounge, andL. Christensen,ibid. 98, 3777~1993!.

81W. J. Buma, B. E. Kohler, and K. Song, J. Chem. Phys.94, 6367~1991!.82J. R. Andrews and B. S. Hudson, Chem. Phys. Lett.57, 600 ~1978!.83H. Petek, A. J. Bell, Y. S. Choi, K. Yoshihara, B. A. Tounge, and R.

Christensen, J. Chem. Phys.102, 4726~1995!.84W. G. Bouwman, A. C. Jones, D. Phillips, P. Thibodeau, C. Friel, and

L. Christensen, J. Chem. Phys.94, 7429~1990!.85W. J. Buma, B. E. Kohler, and T. A. Shaler, J. Chem. Phys.96, 399

~1992!.86S. Choi, T-S Kim, H. Petek, K. Yoshihara, and R. L. Christensen,

Chem. Phys.100, 9269~1994!.87W. J. Buma and F. Zerbetto, J. Phys. Chem. A103, 2220~1999!.88F. Zerbetto and M. Z. Zgierski, J. Chem. Phys.98, 4822~1993!; 93, 1235

~1990!.89A. Warshel and M. Karplus, J. Am. Chem. Soc.94, 5612~1972!.90F. Zerbetto, M. Z. Zgierski, F. Negri, and G. Orlandi, J. Chem. Phys.89,

3681 ~1988!.91A. Zavriyev, I. Fischer, D. M. Villeneuve, and A. Stolow, Chem. Phy

Lett. 234, 281 ~1995!.92See F. Negri and M. Z. Zgierski, J. Chem. Phys.107, 4827~1997!.93G. Fronzoni, P. Decleva, A. Lisini, and G. Alti, J. Electron Spectro

Relat. Phenom.69, 207 ~1994!.94For example, N. Nakashima and K. Yoshihara, J. Phys. Chem.93, 7763

~1989!.95W. Forst, Theory of Unimolecular Reactions~Academic, New York,

1973!.96S. Lochbrunner, T. Schultz, M. Schmitt, J. P. Shaffer, M. Z. Zgierski, a

A. Stolow, J. Chem. Phys.~submitted!.97T. Schultz, A. Stolowet al. ~unpublished!.98J. P. Shaffer, A. Stolowet al. ~unpublished!.99M. Bixon and J. Jortner, J. Chem. Phys.107, 1470~1997!.

electronic continua in time-resolved photoelectron spectroscopy....

Documents