JOURNAL OF CHEMICAL PHYSICS VOLUME 113, NUMBER 16 22 OCTOBER 2000

Resonance-state selective photodissociation of OCS „2 1S¿…: Rotational

and vibrational distributions of CO fragmentsRyuji Itakura,a) Akiyoshi Hishikawa, and Kaoru Yamanouchib)

Department of Chemistry, School of Science, The University of Tokyo, 7-3-1 Hongo, Bunkyo-ku,Tokyo 113-0033, Japan

~Received 6 June 2000; accepted 28 July 2000!

The rotational and vibrational state distributions of the CO fragments produced through thephotodissociation of OCS in the vacuum ultraviolet~VUV ! region ~150–155 nm!, OCS (21S1)→CO (X 1S1)1S(1S), are derived for the three lowest quasi-bound vibrational resonances (v*5022) in the 21S1 state. The rotational state distributions of the CO fragments in thevCO50 and1 vibrational states are determined, respectively, by the analysis of the rotational structures in thelaser-induced fluorescence~LIF! spectra of theA1P –X 1S1(0,0) and~1,1! transitions of CO. Therotational temperatures of CO in thevCO50 state are low~;100 K! for all the three resonances,while those in thevCO51 state are substantially higher, i.e., 2210, 940, and 810 K forv* 50, 1, and2, respectively. The vibrational state distributions of CO are derived from the Doppler spectroscopyof the counterpart S(1S) fragments. From the analysis of the observed Doppler profiles, it is foundfor all the three lowest vibrational resonances of OCS that the vibrational distributions arerepresented well by the Boltzmann-type distribution with a vibrational temperature of around 7000K. On the basis of these new findings, the energy partitioning in the photodissociation processthrough these three vibrational resonances in the 21S1 state is discussed. ©2000 AmericanInstitute of Physics.@S0021-9606~00!01740-2#

mo-osionrmuumo

eaaytan

of

itoni-tinth

tiotu

alaceein-pen-thecalard.

fasttionrgynedsid-ofticte

the

cs,nal

le-

-ng-

ionres-

I. INTRODUCTION

When molecules are irradiated by light in the vacuuultraviolet ~VUV ! wavelength region, they undergo photdissociation in most cases. In the case of polyatomic mecules, which have more than or equal to three degreevibrational freedom, the nuclear motion along the directperpendicular to the dissociation coordinate is often bouTherefore, even in the direct dissociation, they could foquasi-bound vibrational resonances appearing as structprofiles with broadened peaks in an absorption spectrfrom which the nuclear motion near the Franck–Cond~FC! region could be extracted.1,2 However, until recently,little effort has been made to interpret such rich spectral ftures of polyatomic molecules in the VUV region. This mbe partly due to the overlap of the transitions from the rotionally and vibrationally excited states, which broadens ablurs originally characteristic spectral structures.

In our recent studies,3–7 the photodissociation processOCS, OCS (21S1)→CO (X 1S1)1S(1S), in the 140–160nm wavelength region was studied by measuring the exction spectrum called the photofragment excitati~PHOFEX! spectrum under jet-cooled conditions by montoring the S(1S) fragments employing a tunable VUV lighgenerated by the four-wave difference frequency mixtechnique. As shown in Fig. 1, the PHOFEX spectra ofentire region of the 21S1 – 1 1S1 band of OCS exhibitedfive main peaks. These peaks were assigned to the transito the resonant vibrational states with respective quan

a!Research Fellow of the Japan Society for the Promotion of Science.b!Author to whom correspondence should be addressed.

6590021-9606/2000/113(16)/6598/10/$17.00

l-of

nd.

red,

n

-

-d

a-

ge

nsm

numbers,v* 50 – 4, representing the quasi-bound vibrationmotion near the FC region on the potential energy surf~PES! of the 21S1 state, which is mostly repulsive along thdissociation coordinate. This quasi-bound vibration wasterpreted as the in-phase CO and CS stretching mode perdicular to the dissociation coordinate. On the basis ofPHOFEX spectrum exhibiting characteristic asymmetripeak profiles with different peak widths, the ultrafast nuclemotion on the 21S1 PES near its FC region was discusse

Even though we were able to understand the ultranuclear dynamics near the FC region from the absorpfeatures, there still remains a question of how the eneimposed on molecules by the photoabsorption is partitiointo the fragments in the dissociation process. When conering that the five main peaks, which form the progressionthe vibrational resonances, exhibit different characterisasymmetric profiles, it would be interesting to investigahow the energy flow is influenced by the character ofinitially prepared resonant states.

In order to understand the energy partitioning dynamiit becomes necessary to probe the rotational and vibratiodistributions of the CO fragments. Since the first dipoallowed excited state,A1P, of CO is located in the energyrange of;65 000 cm21, we need to introduce another tunable VUV light source to probe the fragments while tunithe first VUV light source to selectively excite the vibrational resonance peaks in the 21S1 – 1 1S1 transition ofOCS.

In the present study, we probe the rotational distributof the CO fragments by measuring the laser-induced fluocence~LIF! spectrum of theA1P –X 1S1 bands of CO using

8 © 2000 American Institute of Physics

ofpw

f tth

n-hr to

torultoi

nathrgtsf

n-rgld

ursu

dC-o

o-in

,

er-o-

-o-e5,

in-

ll

thein-t ofe at

ofen-

er

ondof

V.45

elar

andpho-

the

6599J. Chem. Phys., Vol. 113, No. 16, 22 October 2000 Photodissociation of OCS

the second tunable VUV light source, and probe the S(1S)fragments by tuning the UV light near the3D° –1S transitionof the sulfur atoms to derive the vibrational distributionthe CO fragments through the measurements of the Dopprofiles. On the basis of the product state distributions,discuss the mechanism which governs the later stage ophotodissociation of OCS after it leaves the FC region of2 1S1 surface.

In spite of the long history of the pump–probe-type ivestigation of molecular dynamics using the laser ligsources, as far as we know, the present study reported fofirst time the pump–probe experiments in which two setstunable VUV laser light sources generated by a two-phoresonant four-wave mixing scheme are employed. It is tthat two- or three-photon LIF or resonantly enhanced muphoton ionization~REMPI! techniques could be used tprobe photofragments, but a nonlinearity of the process wrespect to the laser intensities as well as complex rotatiotransition selection rules for molecular fragments causingoverlapping rotational structure tends to introduce a laambiguity in the product state distribution of the fragmenTherefore, the present success of the one-photon probelowed by one-photon excitation in the VUV region is ecouraging in the sense that the number of the research taof photodissociation spectroscopy in the VUV region coube largely increased.

II. EXPERIMENT

The schematic diagram of the experimental setupshown in Fig. 2. A sample gas@OCS ~10%!/Ar# was ex-panded into a vacuum chamber from a pulsed valve~10 Hz!with the 0.8 mm orifice diameter with the stagnation pressof 1.5 atm. When the pulsed valve was operated, the presin the chamber was about 1.531024 Torr. The rotationaltemperature of OCS in the expanded free jet was estimatebe about 6 K, which was obtained for CO contained in Osample~Matheson, 97.5% purity! as the trace amount of impurity from the measurements of the VUV–LIF spectrathe A1P –X 1S1(0,0) transitions.

FIG. 1. The PHOFEX spectrum of the entire region of the 21S1 – 1 1S1

band of OCS measured in our previous study~Ref. 7!.

lerehee

thefnei-

thl-ee.ol-

ets

is

ere

toS

f

A tunable VUV laser light was generated by the twphoton resonant four-wave difference frequency mixingthe Xe gas.8 For the VUV light for the photolysis of OCSthe output beams of the two dye lasers~Lambda PhysikFL3002! simultaneously pumped by a XeCl excimer las~Lambda Physik LPX 205i! were used. The frequencydoubled output (v1) of one dye laser was tuned to the twphoton resonant frequency of the Xe 6p@1/2#0 ~80 119.474cm21!, and then the output of the other dye laser (v2 :13 700–15 385 cm21! was tuned to generate the VUV frequencies (2v1–v2) corresponding to the three lowest resnant peaks (v* 50 – 2) in the PHOFEX spectrum of th2 1S1 – 1 1S1 band of OCS in Fig. 1, appearing at 64 7465 595, and 66 380 cm21.7 The two light beams with thefrequencies ofv1 and v2 were overlapped coaxially by adichroic mirror and focused into a stainless steel cell containg the Xe gas~15–30 Torr! by a achromatic lens (f5300 mm). The VUV laser light generated in the Xe cewas collimated or mildly focused by a LiF lens (f5120 mm) and separated from thev1 andv2 light beams bya LiF prism placed in the separator chamber connected toXe cell. Then, only the generated VUV light beam wastroduced into the main chamber and it crossed the free jethe sample gas 12 mm downstream from the nozzle orifican angle of 45 degrees.

In order to probe the vibrational and rotational statesthe CO fragments, another tunable VUV laser light was gerated by using a dye laser~Lambda Physik Scanmate! andan OPO laser~Lambda Physik Scanmate OPPO! simulta-neously pumped by third harmonics of the Nd:YAG laswith the injection seeding~Coherent Infinity 40-100!. Thesame 2v1–v2 scheme was adopted to generate the secVUV laser to probe the CO fragments as in the generationthe photolysis VUV laser light. The bandwidth of the VUlaser light for both the photolysis and the probe was 0cm21 in FWHM.

The probe VUV laser light was introduced into thchamber so that it hits the crossing point of the molecu

FIG. 2. Schematic diagram of the experimental setup for the pumpprobe measurements using two tunable VUV laser light sources. Thetolysis light source excites OCS to the three lowest resonances of the 21S1

state and the probe light source probes the CO fragments thoroughA1P –X 1S1 transitions. For probing the S(1S) fragments, a tunable UVlight source is used in place of the VUV light source for the probe~see Sec.II !.

f 4th

onit

wa

(V

ee

e

rohn.it

et

-id-e

ec

te

o

-e

totro

dtr/oo

to

bugan

Tthng

tioe

si-

nd

sell-

nal-

byhel-OCS

o-

6600 J. Chem. Phys., Vol. 113, No. 16, 22 October 2000 Itakura, Hishikawa, and Yamanouchi

beam and the photolysis VUV laser beam at an angle odegrees with the molecular beam and at right angles withphotolysis VUV laser beam. The direction of the polarizatiof the photolysis VUV laser beam was set to be parallel wthat of the probe VUV laser beams. The fluorescencemonitored by a solar-blind photomultiplier~HamamatsuR1259! placed beneath the triple crossing point throughlight-collection optics composed of a pair of LiF lensesf530 mm). By scanning the wavelength of the probe VUlaser, the LIF spectra of theA1P –X 1S1(0,0) ~64 640–64 800 cm21! and~1, 1! ~63 900–64 150 cm21! bands of COwere measured.

In order to probe the S(1S) fragments produced from thphotolysis of OCS by the VUV light, the output of the dylaser~Lambda Physik Scanmate! pumped by third harmonicsof a Nd:YAG laser with injection seeding~Coherent Infinity40-100! was frequency doubled in a BBO crystal, and tunaround the frequency of the3D° –1S transition ~45 636cm21! of the S atoms. The probe UV laser beam was intduced into the main chamber so that it crosses with the ptolysis VUV laser beam and the free jet in the same manas the probe VUV laser for monitoring the CO fragments

By monitoring the atomic fluorescence at 148 nm emted from the3D° state to the3P state of the S atoms, thfrequency of the probe laser was scanned to measureDoppler profile of the S(1S) fragments. The intracavity etalon was used for the probe laser to make the bandw~FWHM! as narrow as 0.08 cm21. In order to set the polarization of the photolysis VUV laser to be parallel with thdirection of the propagation of the probe UV laser, the dirtion of the polarization of thev2 light was rotated by 90degrees by using a pair of Fresnel rhombs. For achievinghigh purity of the linear polarization of the VUV light, thlight beam with the frequency ofv2 was transmitted thougha Glan-laser polarizer after it passed through the pairFresnel rhombs.

For the purification of the OCS sample to eliminate impurities of CO, several freeze–pump–thaw cycles were pformed at the liquid N2 temperature. After the purification iwas confirmed that the LIF signal of the impurity CO did nappear with detectable intensities except the transition fthe j ,3 rotational levels of thevCO50 state.

The probe laser beam was introduced with a certainlay after the photolysis laser radiation to reduce the conbution from the scattered light of the photolysis laser andthe fluorescence from the impurity CO excited by the phtolysis laser. The delay time was about 30 ns when moniing S(1S) and about 160 ns when monitoring CO (X 1S1).

The intensities of the respective photolysis and proVUV laser beams were monitored after they passed throthe molecular beam by being reflected by a quartz plateguided to a solar-blind photomultiplier~Hamamatsu R1259!placed in the housing connected to the vacuum chamber.probe UV laser intensity was monitored by detectingweak reflex from a surface of a prism or a window by usia photodiode~Hamamatsu S1336-5BQ! before the UV laserwas introduced into the vacuum chamber. The normalizaof the fluorescence intensity from the fragments was p

5e

hs

a

d

-o-er

-

he

th

-

he

f

r-

m

e-i-r-r-

ehd

hee

nr-

formed by the monitored photolysis and probe laser intenties.

III. RESULTS AND DISCUSSION

A. Rotational state distribution of CO „X 1S¿;v COÄ0,1…

1. Determination of rotational state distribution

After OCS was selectively excited to the quasi-bouresonances at 64 745 cm21 (v* 50), 65 595 cm21 (v* 51),and 66 380 cm21 (v* 52), the LIF spectra of theA1P –X 1S1(0,0) and~1,1! transitions of the CO fragmentwere measured. The observed LIF spectra exhibiting wresolved rotational structure are shown in Figs. 3~a! and 3~b!.On the basis of the rotational assignments, the rotatiostate distributions of the CO fragment in thevCO50 and 1states were derived using the unperturbed rotational linesdividing the observed rotational transition intensities by tcorresponding Ho¨nl–London factors. The derived rotationastate distributions of thevCO50 and 1 states of the CO fragments produced through the three lowest resonances ofare plotted in Fig. 4. It should be noted in Figs. 4~a!–4~f! thatthe populations derived from theQ branch transitions werealways smaller than those derived from theP andR branch

FIG. 3. The observed LIF spectra for the~a! A1P –X 1S1(0,0) and~b!A1P –X 1S1(1,1) bands of CO produced via the lowest vibrational resnance withv* 50 of the 21S1 state of OCS.

aeollyo

nd

aagohioio

-

th

ag

erygu-ld

-ring

ne,rep-

to

ontoO

cede

n.as-

so-

ser

rvedakalor-ereciesm

spe-,an-

een

is-n

ned

on-l-

p-

so-

ts

6601J. Chem. Phys., Vol. 113, No. 16, 22 October 2000 Photodissociation of OCS

transitions. As explained later, these discrepancies arecribed to the alignment effect of the CO fragments inducby the polarized photolysis laser light, i.e., the angular mmentum vector of the CO fragment is formed preferentiain the direction perpendicular to the polarization vectorthe photolysis light.9–12

The lifetimes of OCS in the lowest three quasi-bouvibrational resonances of 21S1 were determined to be133 fs (v* 50), 44 fs (v* 51), and 27 fs (v* 52) in ourprevious study.7 These three values are all much shorter tha rotational period of OCS, and therefore, the CO photofrments could not recoil isotropically with respect to the plarization direction of the photolysis laser light within sucan ultra-short time duration. In such an ultrafast dissociatprocess, a phenomenon originating from the correlatamong~i! the polarization vector of the photolysis light,~ii !the recoil vector, and~iii ! the angular momentum of the fragment can be observed.

From the conservation of the angular momentum,total angular momentum of OCS,J, before the dissociation isequal to the sum of the angular momentum of the CO frment,j , and the orbital angular momentum,L , of the recoil-ing fragments of S and CO, i.e.,J5 j1L . In our experiment,

FIG. 4. The observed rotational state distributions of the CO fragmentthe vibrational states ofvCO50 and 1 via the three lowest vibrational resnances ofv* 50 – 2, derived by taking account only of the Ho¨nl–Londonfactor from the peak intensities of theP~m!, Q~s!, andR~j! branches in theobserved LIF spectra. The length of the bar for the data points represenexperimental uncertainty~s!.

s-d-

f

n-

-

nn

e

-

OCS is prepared under the ultracold conditions at the vlow rotational temperature. The magnitude of the total anlar momentum,J, is considered to be very small and coube assumed to be zero, i.e.,J50, for simplification. WhenJ50, the recoil vector,v, is constrained to be on the molecular plane defined by the three atoms and fixed in space duthe dissociation process and bothj and L (52 j ) becomeperpendicular to the plane in the highj 5u j u limit. Since the2 1S1 – 1 1S1 transition is a parallel transition, the transitiomoment,mOCS, is located in the molecular plane. Thereforthe correlations among the aforementioned vectors areresented asv' j andmOCS' j .

In the high-j limit, the transition moment of CO,mCO,for theA1P –X 1S1 transition is considered to be parallelj for theQ branch transition and perpendicular toj for thePandR branches. In our experimental condition, the directiof the polarization of the photolysis laser light was parallelthe direction of the polarization of the probe light for the Cfragments. Therefore, the LIF intensities in theP and Rbranch transitions of the CO fragments should be enhancompared with those in theQ branch transitions due to thanisotropic angular momentum distribution of CO.

By taking account of themOCS–v– j vector correlationsand the detection geometrical factor,9,11 the Doppler line pro-files for the respectiveP, Q, andR branch transitions werecalculated in a semi-classical manner described by Dixo9

In the calculation, the recoil anisotropy parameter wassumed to beb51.8 for all the rotational states of the COfragments, which was previously derived for the photodisciation of OCS at 157 nm.13 After convoluting the calculatedDoppler profiles by a Gaussian function with the probe labandwidth (FWHM;0.45 cm21), the rotational spectrumwas synthesized, which can be compared with the obserotational spectrum. Then, by dividing the observed peintensities in the LIF spectrum by the simulated rotationtransition intensities of the corresponding peaks, the crected rotational state distributions of the CO fragments wobtained as shown in Fig. 5. The extent of the discrepanin Fig. 4 between the rotational distributions obtained frotheP andR branch transitions and those obtained from theQbranch transitions decreases considerably in Fig. 5, ecially in the high-j region of thevCO51 channels. Howevereven after the correction, there still remain some discrepcies, and the populations derived from theQ branch transi-tions tend to become larger than those derived from thPandR branch transitions. This tendency is most clearly sein the low-j region of thevCO50 channel. Although thevalue ofb was changed in the simulation to reduce the dcrepancies in the low-j region, these were not improved iany value ofb.

The discrepancies between the populations obtaifrom the Q branch and those from theP and R branchescould be attributed to~i! the assumption ofJ50 and~ii ! thesemi-classical approximation. Even under the jet-cooled cdition (Trot56 K), the population of OCS in the rotationaground state withJ50 is only;5%. Consequently, the vector correlations ofv' j and mOCS' j do not always hold.12

When the angular momentum of the CO fragment,j , is muchlarger thanJ, the semi-classical approximation becomes a

in

the

en

re

-n

th

ealio

bua

Otri

thened

be-

-enaler-

dtum

is-

intons-

-

.g-

,

nthe

eso-

6602 J. Chem. Phys., Vol. 113, No. 16, 22 October 2000 Itakura, Hishikawa, and Yamanouchi

propriate and the deviations from the vector correlations tto be small. However, in the low-j region, it would not beappropriate any more, and this is the reason why the discancies between the populations obtained from theQ branchand those from theP andR branches are larger in the lowjmanifold as shown in Fig. 5 even after the correction cosidering the vector correlations.

On the other hand, as long as we concentrate ondistributions derived from either theQ branch or theP andRbranches, the effect originating from the correction considing the vector correlations was found to be negligibly smBecause the number of rotational states whose populatwere determined from theP and R branches were muchlarger than those fromQ branch lines, theP and R branchtransitions were adopted to derive the rotational distritions, which were obtained simply by dividing the rotationtransition intensities by the Ho¨nl–London factors.

By assuming that the rotational distribution of the Cfragment,C( j ), as that expressed by the Boltzmann disbution,

C~ j !5C0~2 j 11! exp~2ERot/kbTrot!, ~1!

with a proportional constantC0 independent ofj, the rota-

FIG. 5. The corrected rotational state distributions of the CO fragmeobtained from Fig. 4 by taking account of the vector correlations amongphotolysis laser polarization, the fragment recoil, and the angular momtum of the CO fragments.

d

p-

-

e

r-l.ns

-l

-

tional distributions for thevCO50 and 1 channels for thev* 50 – 2 resonances obtained from theP andR branch tran-sitions were fitted by the least-squares method to deriverotational temperatures of the CO fragments. The determirotational temperatures are listed in Table I.

For the photodissociation viav* (50 – 2), the rotationaltemperatures of the CO fragments invCO50 are all verylow, i.e., Trot;100 K, while those invCO51 are muchhigher than invCO50, i.e., Trot52210, 940, and 810 K forv* 50, 1, and 2 respectively, and exhibit a tendency tocome lower asv* increases.

2. Comparison with model distributions

a. Prior distribution. We will compare the observed rotational distributions with prior distributions, in which thCO fragments are populated with probabilities proportioto a square root of the translational energy in all the engetically allowed ro-vibrational states.14,15If the total angularmomentumJ of a parent triatomic molecule is approximateto be zero in its dissociation process, the angular momenof the diatomic fragment,j , is perpendicular to the directionof the recoil, and the projection ofj onto the recoil direction,mv , becomes zero. Therefore, when calculating a prior dtribution function, P0(vCO, j ), of the CO fragment in asingle vibrational and rotational state (vCO, j ), the (2j 11)degeneracy of the rotational state should not be takenaccount,16 and it is represented by a square root of the tralational energy as

P0~vCO, j !}~Eavail2Ev2Er !1/2 ~2!

with

Eavail5Ehn2D02ES , ~3!

where Ehn is the energy of the photolysis light,D0 is thedissociation energy of OCS in the electronic ground state,ES

is the electronic energy of the1S state of the S atom measured from the3P2 state, andEv andEr are the vibrationaland rotational energies of the CO fragment, respectively

From the rotational state distribution of the CO framents in their specific vibrational state,P(vCO, j ), derivedexperimentally in Sec. III A 1, and the prior distributionP0(vCO, j ), the rotational surprisal,I (gR), is obtained14–17

as

I ~gR!52 lnP~vCO, j !

P0~vCO, j !~4!

as a function of

tsen-

TABLE I. The rotational temperatures~in K! for the CO fragments in thevCO50 and 1 states produced through the three lowest vibrational rnances ofv* 50 – 2 of OCS in the 21S1 state.a

v* vCO50 vCO51

0 95 ~10! 2210 ~280!1 100 ~30! 940 ~120!2 90 ~15! 810 ~65!

aThe numbers in the parentheses represent a standard deviation~s! derivedfrom the least-squares fit to the Boltzmann-type rotational distribution.

iao-x

ncit

-

teotiben

.

ba

-cy

stthe

ta-

de-if-ion

ngrag-te

ngthee

entsoto-itsCe,h

ionofonor

ion.n

l,COf

ressCO-rntsso-g-

up-

eho-ta--ith

alhetht

wn

6603J. Chem. Phys., Vol. 113, No. 16, 22 October 2000 Photodissociation of OCS

gR5Er

Eavail2Ev5

Er /Eavail

12~Ev /Eavail!5

f R

12 f v. ~5!

The resulting plots of the rotational surprisal are shownFig. 6 for the CO fragments in the two lowest vibrationstates (vCO50,1) when the three lowest vibrational resnances (v* 50 – 2) of the parents OCS molecules are ecited. Then, the surprisal plots were fitted by a linear fution to derive a slope. In the case of the CO fragments wvCO50, the slopes obtained from theP branch transitions are119~32!, 149~42!, and 147~31! for v* 50, 1, and 2, respectively, while those obtained from theQ-branch are 123~23!,100~7!, and 145~24!. Even though the slopes are associawith relatively large uncertainties, it can be said that the plare well described by a linear function and that they exhconsistently a slope of 100–150. This suggests that thexists a common mechanism which governs the rotatiodistribution in thevCO50 for the three lowest resonances15

In the case of the surprisal plots forvCO51, their behav-ior is not as simple as that forvCO50. In the range belowgR50.015, where the data for bothvCO50 and 1 are avail-able, theP branch data forv* 50 and 2 ofvCO51 exhibit analmost flat distribution with respect togR , while those forv* 51 exhibit a negative slope. When it comes to the globehavior of the plot in the entire range, 0<gR<0.15, both

FIG. 6. The rotational surprisal plots,I (gR)s, for the CO fragments in thevibrational states ofvCO50 and 1 produced via the quasi-bound vibrationresonances ofv* 50, 1, and 2. The data derived from the intensities of tobservedP~m!, Q~s!, andR~j! branches in the LIF spectra are plotted widifferent marks. The straight lines drawn in~a!–~c! represent the best-filinear lines for the surprisal plots forvCO50. ForvCO51, the three plots donot exhibit simple linear behavior and, therefore, no straight line is dra

nl

--h

dsitreal

l

the P andQ branch data forv* 50 and 1 exhibit a shallowminimum at gR50.06– 0.08 andgR50.02– 0.05, respectively. For v* 52, it would be hard to describe the tendendue to the small data number in the plot, but theP branchdata abovegR50.015 exhibit a clear positive slope. It iuncertain whether thev* 52 data have a minimum, but ican be said that the transition from the flat region topositive slope region occurs atgR50.01– 0.02. The complexbehavior of the surprisal plots described above for the rotional distributions of the CO fragments in thevCO51 state,suggests that a simple dynamical constraint could notscribe the rotational distributions. It is probable that two dferent kinds of dynamics govern respectively the productof the low-j and high-j CO fragments.

b. Rotational Franck–Condon model. When the PES ofthe excited state of OCS around the FC region is flat alothe bending angle and no torque is imposed on the CO fments to drive their rotational motion, the rotational stadistribution is determined by the projection of the bendiwave function of OCS in the electronic ground state ontorotational wave function of the CO fragments. Within thframework of this rotational FC model, Freed and Mors18

showed that the rotational distribution of diatomic fragmecan be represented by a Boltzmann distribution when phdissociation of a linear triatomic molecule occurs fromJ50 and vbend50 state. By adopting the rotational Fmodel, Strausset al.13 calculated the rotational temperaturTrot , of Trot5380 K for the CO fragments produced througOCS (21S1). However, the PES of the 21S1 state of OCScould have a certain slope along the bending directaround the FC region which induces the bending motionOCS and eventually imposes the torque for the rotationCO. Whether the rotation of the CO fragment is promotedsuppressed depends on the anisotropy of the 21S1PESalong the bending angle in the course of the dissociatTherefore, the deviation from the Boltzmann distributiowith Trot5380 K, derived from the rotational FC modecould be ascribed to the anisotropy of the PES. Since thefragments withvCO50 have a rotational temperature oaround 100 K for all the three low-lying resonances ofv*50 – 2, the CO fragments withvCO50 are considered to beformed through the anisotropic PES, which tends to suppthe rotation of the fragments. On the other hand, thefragments withvCO51 exhibit much higher rotational temperatures thanTrot5380 K; i.e., 2210, 940, and 810 K fov* 50, 1, and 2, respectively. Therefore, the CO fragmewith vCO51 are considered to be formed through the anitropic PES, which tends to promote the rotation of the framents though the extent of the rotational excitation is spressed for the higher resonance state of OCS.

3. Effect of resonances in the FC region

Previously, Strausset al.13 studied the rotational statdistribution of the CO fragments produced through the ptolysis of OCS at the fixed wavelength of 157 nm. The rotional distributions forvCO50 – 3 were found to be well reproduced respectively by the Boltzmann distributions wthe rotational temperatures ofTrot51350, 1300, 980, and 770K. This decreasing tendency ofTrot for a largervCO shows a

.

hrll

n

prula

ileoblap

he

lyide

ustlln

elo

tiolo

texxc

ne

hetharoi

.tgbe

istht

le

theo

heant

ateffi-anthe

heng-

eresame

6604 J. Chem. Phys., Vol. 113, No. 16, 22 October 2000 Itakura, Hishikawa, and Yamanouchi

marked contrast with the present results obtained at the tlowest vibrational resonances for which the substantiahigher rotational temperatures~1000–2000 K! are achievedfor vCO51 than forvCO50(;100 K).

As was reported in our previous study,7 a weak con-tinuum is located at 157 nm under the jet-cooled conditioAt the peaks of the vibrational resonances (v* 50 – 2) thereis a large contribution from the resonance component resenting the nuclear motion along the direction perpendicto the dissociation coordinate, while at 157 nm theremostly the off-resonance component representing the nucmotion along the dissociative coordinate. Therefore, theserved difference in thevCO dependence of the rotationatemperature could be attributed to the difference in the chacter of the nuclear motion near the FC region on the upstate PES.

As shown in Fig. 1, the vibrational resonances in tPHOFEX spectrum of the 21S1 – 1 1S1 band exhibit char-acteristic peak profiles; thev* 50 resonance shows a largeasymmetrical profile shading towards the lower energy swith a narrow peak width, while the peak profile becommore symmetrical and broader asv* increases to thev*51 and 2 resonances. It would be natural to expect that sa character of the resonances influences the productdistributions, because the extent of the asymmetry as wethe peak width reflects the extent of the mixing of the cotinuum and bound characters.19 However, it seems that thmemory of the character of the resonance state is mostlyin the rotational state distribution for thevCO50 channel,suggesting the existence of a strong exit-channel interacbetween the FC region and the asymptotic region. Theof the memory of the resonance states is also seen invibrational state distributions of CO described in the nsection. On the other hand, the extent of the rotational etation for thevCO51 channel is more suppressed asv* in-creases from 0 to 2~see Table I and Fig. 6!. It is possible thatthe remnants of the character of the resonances appear iform of the difference in the rotational excitation for thvCO51 channel.

B. Vibrational state distribution of CO „X 1S¿…

1. Vibrational state distribution of CO from Dopplerprofiles of S„1S…

In order to derive the vibrational state distribution of tCO fragments, we measured the Doppler profiles ofS(1S) fragments. Since a1S state has no electronic angulmomentum, the Doppler profile reflects directly the recmomentum of the S(1S) fragments,9,10,20from which the ki-netic energy distribution of the S(1S) fragments is obtainedThen, on the basis of the conservation of the energy andmomentum, the internal energy distribution of the CO framents including their vibrational state distribution canestimated.

In the case of the 21S1 – 1 1S1 transition,21,22 the elec-tronic transition moment is parallel to the molecular axand the fragments are expected to eject preferentially indirection along the laser polarization. Therefore, in orderobserve the Doppler effect in a form of a clearer Dopp

eey

s.

e-r

sar-

r-er

es

chateas-

st

nssheti-

the

e

l

he-

,e

or

profile, the UV probe laser light to detect the S(1S) frag-ments was introduced along the direction parallel withelectric vector of the photolysis laser light so that the twlaser beams cross at right angles with each other.

The observed Doppler profiles obtained by exciting tcentral position of the peak for the three lowest resonstates, i.e., 64 745(v* 50), 65 595(v* 51), and 66 380cm21 (v* 52), are shown in Figs. 7~a!–7~c!. All the Dop-pler profiles exhibit a clear dip in the center, indicating ththe released energies in the dissociation process areciently transferred into the translational motion rather thinto the rotational and vibrational degrees of freedom ofCO fragments.

From the rotational analysis of the LIF spectra of tA 1P –X 1S1(0,0) and~1, 1! bands of the CO fragments iSec. III A 1, the rotational state distributions of the CO framents in the vibrational states ofvCO50 and 1 were found to

FIG. 7. The observed~open circles! and best-fit~solid lines! Doppler pro-files of the S(1S) fragments for the photodissociation of OCS (21S1) viathe quasi-bound states of~a! v* 50, ~b! v* 51, and~c! v* 52. In the trial-and-error simulation, the vibrational distributions of the CO fragments wassumed to be represented by the Boltzmann-type distribution and theoptimized vibrational temperatures ofTvib57000 K were derived for all thethree profiles~see text for details!.

isi

o--

Cibteth

s-v

o

oo

he

ithera-r a

ra-ra-

ro-cha-ra-he

ys

onr

orureerp-

l

e

inghe

nts

of

five

6605J. Chem. Phys., Vol. 113, No. 16, 22 October 2000 Photodissociation of OCS

be approximated by the Boltzmann distributions for the dsociation through the three low-lying resonance states wv* 50 – 2. Therefore, in the simulation of the Doppler prfiles of the S(1S) atomic fragments, the rotational distributions for the energetically accessible vibrational states ofwere assumed to be described by the Boltzmann distrtions. The rotational temperatures for the vibrational staexcept forvCO50 and 1 were assumed to be the same ascorresponding values forvCO51; i.e., Trot52210, 940, and810 K were adopted for the fragment CO (vCO.1) producedthrough the three lowest resonances withv* 50, 1, and 2,respectively. The recoil anisotropy parameterb was assumedagain to beb51.8.

The Doppler profile of S(1S), obtained when OCS isexcited to the lowest vibrational resonance ofv* 50, isshown in Fig. 8~a!, and the Doppler profiles calculated asuming that the CO fragments are populated in a singlebrational state in the range betweenvCO50 and 8 are shownin Fig. 8~b!. For comparison with the observed Doppler prfile, the calculated Doppler profiles forvCO50 and 8 are alsodrawn in Fig. 8~a! in dotted lines. When OCS is excited tv* 50 and 1, the maximum vibrational quantum numberCO, which can be produced energetically, isvCO58, andwhen v* 52, the maximum becomesvCO59. It is clearly

FIG. 8. ~a! The observed Doppler profile of the S(1S) fragments when OCSwas excited to the lowest quasi-bound resonances withv* 50 ~solid line!.The calculated Doppler profiles are shown for comparison for the lowvibrational state (vCO50) and the highest vibrational state (vCO58) ~dottedlines! when the CO fragments are assumed to be populated only in the svibrational state.~b! The calculated Doppler profiles are shown for all tvibrational states,vCO50 – 8, which are energetically accessible.

-th

Ou-se

i-

-

f

seen in Figs. 8~a! and 8~b! that the contribution from thehighly excited CO must be relatively small to reproduce tcentral dip in all the observed Doppler profiles forv*50 – 2.

In the course of the simulation of the Doppler profiles,was found that the Boltzmann-type distribution in which tvibrational population exponentially decreases as the vibtional energy increases was found to give the best fit. Aftetrial-and-error-type simulation using the vibrational tempeture, Tvib , as one variable parameter, the optimized vibtional temperature ofTvib5700062000 K was obtained forall the three resonances ofv* 50 – 2. As shown in Fig. 7, thebest-fit Doppler profiles reproduced well the observed pfiles. This result suggests that there exists a common menism for all the three resonances which governs the vibtional population of CO by affecting in a later stage after tFC region regardless of the different excitation energies.

2. Comparison with prior distributions

A prior distribution of vibrational states is obtained bthe summation ofP0(vCO, j ) over all the rotational statewhich are energetically accessible as

P0~vCO!5(j

P0~vCO, j !. ~6!

In Fig. 9, the best-fit Boltzmann vibrational state distributiof Tv57000 K, Pobs(vCO), to describe the observed Doppleprofile is compared with the prior distribution calculated fthev* 50 resonant state of OCS. It can be seen in this figthat the CO fragments tend to be populated in the lowvibrational states than in the prior distribution. The discreancy between the observed distribution,Pobs(vCO), and theprior distribution,P0(vCO), is described by the vibrationasurprisal,I ( f v),14,17

I ~ f v!52 lnP~vCO!

P0~vCO!, ~7!

st

le

FIG. 9. The normalized vibrational state distributions of the CO fragmeproduced after the photodissociation of OCS (21S1); ~a! the Boltzmanndistribution of 7000 K which reproduces all the observed Doppler profilesthe S(1S) fragments when thev* 50 – 2 resonances of OCS are excited,~b!the prior distribution calculated for thev* 50 resonance, and~c! the vibra-tional distribution forvCO50 – 3 obtained by Strausset al. ~Ref. 13! for thephotodissociation at 157 nm. For the normalization for~c!, the amount ofthe population in each ofvCO54 – 8 is assumed to be 5% ofvCO50, whichwas estimated in Ref. 13 as an upper limit of the population for thesevibrational states.

Caaser

nlg

nder

u-eo

ath

yt

elab

n

thebem isvial

bents

there-

a-is

ns

ione

tionthe

ra-ion

theonaled

icer-Once

-

hatl-o-ra-

icCSof

iclec-et-dleco-edti-di-wn

FCg inthe

tion

e

6606 J. Chem. Phys., Vol. 113, No. 16, 22 October 2000 Itakura, Hishikawa, and Yamanouchi

as a function off v5Ev /Eavail. The surprisal plots of the COfragments obtained for the three lowest resonances of Owith v* 50 – 2 are presented in Fig. 10. All three surprisplots exhibit a linear slope in the range of the vibrationstates of CO withvCO50 – 6, to which most of the fragmentare populated. By the fit to a straight line, the slopes wdetermined to be 1.90~13!, 2.14~12!, and 2.36~13! for v*50, 1, and 2, respectively. This linear behavior commoidentified for the dissociation through the three low-lyinresonances with the close values of the best-fit slopes icates the existence of a common mechanism which govthe vibrational energy partitioning process.15

3. Effect of resonances in the FC region

As shown in Fig. 9, the calculated vibrational distribtion of CO at 7000 K, which reproduces well the observDoppler profiles obtained via the vibrational resonancesv* 50 – 2, have the larger populations in the low vibrationstates than the vibrational distributions obtained fromphotodissociation at 157 nm by Strausset al.13 Because theavailable energy atv* 50 – 2 is larger than that obtained bthe 157 nm photodissociation, this difference is contrarythe statistical model in which the population in the highvibrational states of the fragments increases as the avaienergy becomes larger.

Since the photolysis at 157 nm by Strausset al.13 wasperformed under jet-cooled conditions and the CO fragme

FIG. 10. The vibrational surprisal plots,I ( f v)s, for the CO fragments ob-tained via the quasi-bound resonances;~a! v* 50, ~b! v* 51, and ~c! v*52. The slopes of the fitted lines to the surprisal plots in the range betwvCO50 and 6 are 1.90~13!, 2.14~12!, and 2.36~13! for v* 50, 1, and 2,respectively.

Sll

e

y

i-ns

df

le

orle

ts

formed in the cold region are considered to be sampled,vibrational state distribution reported in Ref. 13 couldregarded as that obtained when the off-resonant continuuexcited. It is interesting to note that the photodissociationthe three resonances,v* 50 – 2, afforded a lower vibrationatemperature than the photodissociation at 157 nm. It cansaid that the excitation of the vibration of the CO fragmeis suppressed when the dissociation proceeds throughresonance states of the quasi-bound vibration in the FCgion, while the energy partitioning into the fragment vibrtion becomes efficient when the off-resonant continuumexcited.

The observation that the similar vibrational distributioare obtained for all the three resonance states,v* 50 – 2,indicates that the quasi-bound vibration in the FC regwould not affect so much the vibrational distribution of thCO fragments. This means that the product state distribuis determined in the region between the FC region andasymptotic region. In other words, the memory of the vibtion in the FC region tends to be lost when the dissociatproceeds.

In the cases of the photodissociation of FNO (S1)23,24

and ClNO (T1 ,S1),25–27 their PHOFEX spectra exhibiteddistinct vibrational peaks and they were interpreted asvibrational Feshbach resonances. Therefore, the vibratidynamics of FNO and ClNO in their FC region is considerto have a certain similarity to OCS (21S1). However, con-trary to OCS, the vibrational distributions of the diatomphotofragments produced from FNO and ClNO were intpreted by an adiabatic picture, in which most of the Nfragments produced via the vibrational Feshbach resonawith the quantum number,v* , are populated in the vibrational state with the quantum number,vNO, equal tov* orv* 21. This type of an adiabatic distributions suggests tthe vibrational motion of the NO moiety of the parent moecules in the FC region is very close to the vibrational mtion of the NO fragments and keeps its nature of local vibtion in the course of the dissociation process.

As far as the vibrational distributions of the diatomfragments are concerned, the photodissociation of O(2 1S1) are closer to the symmetric triatomic moleculesH2O (A,1B1)28–32 and O3 ~Chappuis Band33–35 and HartleyBand36–39! rather than theX– NO-type molecules eventhough the broad absorption features of H2O and O3 exhibitno vibrational progression. When these symmetric triatommolecules are excited to the dissociative PES of the etronically excited state, they start to move along the symmric stretch coordinate from the FC point towards the sadpoint. Though the potential along the symmetric stretchordinate is bound, this motion is unstable and is deforminto the mixture of the translational motion along the ansymmetric stretch coordinate and the vibration of theatomic fragment, which is expected to cause the breakdoof the vibrational adiabaticity.

In the case of OCS, the vibrational motion near theregion is considered to be close to the in-phase stretchinboth the O–C and C–S bonds as was demonstrated bywavepacket calculation in our previous study.7 Therefore, inthe course of the dissociative process, the vibrational mo

en

lands

thatre

uss

hehe

o

heer-

n

ithnenadthc

e

thofunn

stisrm

onty

Fara

-astsh

rch

rendor-ci-

,

m.

lat.

a-

J.

urn,

m.

.

ys.

.

B.hys.

hys.

.

r,

.

6607J. Chem. Phys., Vol. 113, No. 16, 22 October 2000 Photodissociation of OCS

in the FC region would be transferred into both the transtional motion and the vibration of the CO fragments, aduring this process the vibrational adiabaticity would be loresulting in the similar vibrational state distribution to O3 andH2O.

IV. CONCLUSION

We investigated the photodissociation of OCS via2 1S1 state by measuring the rotational and vibrational stdistributions of the CO fragments produced through the thlowest vibrational resonances in the 21S1 – 1 1S1 band.Since the dissociation, OCS (21S1)→CO(X 1S1)1S(1S), proceeds on a single PES of the 21S1 state, thisband system affords us an ideal opportunity to study the pnuclear dynamics in the course of the dissociation proce

~1! The rotational distributions of the CO fragments in tvCO50 and 1 vibrational states were derived for tthree lowest resonance peaks,v* 50 – 2, of OCS bymeasuring the rotational structures of the LIF spectrathe one-photon COA 1P –X 1S1(0,0) and~1, 1! transi-tions. The rotational state distributions invCO50 forv* 50 – 2 were all represented approximately by tBoltzmann distribution of around 100 K. On the othhand, the CO fragments invCO51 were much more excited rotationally than CO (vCO50) and the distribu-tions were approximated by the Boltzmann distributiowith Trot522106280, 9406120, and 810665 K forv* 50, 1, and 2, respectively. From the comparison wthe rotational temperatures of the CO fragments obtaiby the photolysis at 157 nm and those by the rotatioFC model, the characteristic rotational temperaturesrived here were considered to reflect the character ofvibrational resonances in the FC region. The differenin the rotational distribution of thevCO50 andvCO51channels was also identified as the difference in thsurprisal plots.

~2! From the measurements of the Doppler profiles ofS(1S) atomic fragments, the vibrational distributionsthe CO fragments were determined and they were foto be well approximated by the Boltzmann distributiowith the temperature of 700062000 K. The common vi-brational distribution obtained for all the three loweresonance states with similar shapes in their surprplots indicates that the dissociation channels which fothe CO fragments in the differentvCO values are coupledstrongly with each other after OCS leaves the FC regileading to the breakdown of the vibrational adiabatici

ACKNOWLEDGMENTS

The authors are grateful to Professor Geerd H.Diercksen for his valuable discussion. The authors also thTakashi Kawauchi, Nobuhiro Hirano, and Tetsuya Kimu~Marubun Co.! for their technical support for the Infinitybased dye-laser and OPO-laser system and the LPX-bdye-laser system which were used to generate the two setunable vacuum UV laser light beams. The present study

-

t,

eee

re.

f

s

dl

e-ee

ir

e

d

al

,.

.nk

edof

as

been supported by the Grant-in-Aid for Scientific Resea~No. 08640636! and that for Priority Area~No. 07240106!from the Ministry of Education, Science, Sports and Cultuand by CREST~Core Research for Evolutionary Science aTechnology! fund from Japan Science and Technology Cporation. R.I. acknowledges the fellowship from Japan Soety for the Promotion of Science.

1R. Schinke,Photodissociation Dynamics~Cambridge University PressNew York, 1993!.

2E. J. Heller, Acc. Chem. Res.14, 368 ~1981!.3C. D. Pibel, K. Ohde, and K. Yamanouchi, J. Chem. Phys.101, 836~1994!.

4K. Yamanouchi, K. Ohde, A. Hishikawa, and C. D. Pibel, Bull. CheSoc. Jpn.68, 2459~1995!.

5K. Ohde, A. Hishikawa, and K. Yamanouchi, J. Electron Spectrosc. RePhenom.79, 433 ~1996!.

6K. Yamanouchi, K. Ohde, and A. Hishikawa, Adv. Chem. Phys.101, 789~1997!.

7A. Hishikawa, K. Ohde, R. Itakura, S. Liu, K. Yamanouchi, and K. Ymashita, J. Phys. Chem. A101, 694 ~1997!.

8K. Yamanouchi and S. Tsuchiya, J. Phys. B28, 133 ~1995!.9R. N. Dixon, J. Chem. Phys.85, 1866~1986!.

10P. L. Houston, J. Phys. Chem.91, 5388~1987!.11C. H. Green and R. N. Zare, J. Chem. Phys.78, 6741~1983!.12G. E. Hall, N. Sivakumar, D. Chawla, P. L. Houston, and I. Burak,

Chem. Phys.88, 3682~1988!.13C. E. Strauss, G. C. McBane, P. L. Houston, I. Burak, and J. W. Hepb

J. Chem. Phys.90, 5364~1989!.14R. D. Levine and R. B. Bernstein, Acc. Chem. Res.7, 394 ~1974!.15J. I. Steinfeld, J. S. Francisco, and W. L. Hase,Chemical Kinetics and

Dynamics~Prentice–Hall, Englewood Cliffs, NJ, 1989!.16J. T. Muckerman, J. Phys. Chem.93, 179 ~1989!.17E. Zamir, R. D. Levine, and R. B. Bernstein, Chem. Phys. Lett.107, 217

~1984!.18M. D. Morse and K. F. Freed, J. Chem. Phys.74, 4395~1981!.19U. Fano, Phys. Rev.124, 1866~1961!.20R. N. Zare, Mol. Photochem.4, 1 ~1972!.21J. W. Rabalais, J. M. McDonald, V. Scherr, and S. P. McGlynn, Che

Rev.71, 73 ~1971!.22M. I. McCarthy and V. Vaida, J. Phys. Chem.90, 6759~1986!.23S. A. Reid, J. T. Brandon, and H. Reisler, J. Phys. Chem.97, 540 ~1993!.24J. T. Brandon, S. A. Reid, D. C. Robie, and H. Reisler, J. Chem. Phys97,

5246 ~1992!.25C. X. W. Qian, A. Ogai, L. Iwata, and H. Reisler, J. Chem. Phys.92, 4296

~1990!.26A. Ogai, C. X. W. Qian, and H. Reisler, J. Chem. Phys.93, 1107~1990!.27C. X. W. Qian, A. Ogai, J. T. Brandon, Y. Y. Bai, and H. Reisler, J. Ph

Chem.95, 6763~1991!.28P. Anderson, G. S. Ondrey, B. Titze, and E. W. Rothe, J. Chem. Phys80,

2548 ~1984!.29V. Engle, V. Staemmler, R. L. Vander Wal, F. F. Crim, R. J. Sension,

Hudson, P. Andersen, S. Hennig, K. Weide, and R. Schinke, J. PChem.96, 3201~1992!.

30D. W. Hwang, X. Yang, and X. Yang, J. Chem. Phys.110, 4119~1999!.31V. Engle, R. Scihnke, and V. Staemmler, J. Chem. Phys.88, 129 ~1988!.32M. L. Doublet, G. J. Kroes, E. J. Baerends, and A. Rosa, J. Chem. P

103, 2538~1995!.33D. S. Moore, D. S. Bomse, and J. J. Valentini, J. Chem. Phys.79, 1745

~1983!.34H. B. Levene, J.-C. Nieh, and J. J. Valentini, J. Chem. Phys.87, 2583

~1987!.35B. Woywod, M. Stengle, W. Domcke, H. Flo¨thmann, and R. Schinke, J

Chem. Phys.87, 2583~1987!.36J. J. Valentini, D. P. Gerrity, D. L. Phillips, J.-C. Nieh, and K. D. Tabo

J. Chem. Phys.86, 6745~1987!.37C. Leforestier, F. LeQue´re, K. Yamashita, and K. Morokuma, J. Chem

Phys.101, 3806~1994!.38B. R. Johnson and J. L. Kinsey, Phys. Rev. Lett.62, 1607~1989!.39B. R. Johnson and J. L. Kinsey, J. Chem. Phys.91, 7638~1989!.