Vibrational mode-specific photochemical reaction dynamics of chlorine dioxide in solution

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:

130.88.90.110 On: Sat, 20 Dec 2014 03:21:15

JOURNAL OF CHEMICAL PHYSICS

VOLUME 114, NUMBER 15

15 APRIL 2001

Vibrational mode-speciﬁc photochemical reaction dynamics of chlorine

dioxide in solution

Henk Fidder^a⁾

Department of Physical Chemistry, Uppsala Universitet, P.O. Box 532, S-751 21 Uppsala, Sweden

Frank Tschirschwitz, Oliver D u¨ hr, and Erik T. J. Nibbering

Max-Born-Institut f u¨ r Nichtlineare Optik und Kurzzeitspektroskopie, Max Born Strasse 2A,

D-12489, Berlin, Germany

͑

Received 25 April 2000; accepted 29 January 2001͒

We study the reaction dynamics of OClO in cyclohexane, acetonitrile, and water by femtosecond

pump–probe spectroscopy. In all solvents we observe a quantum beat in a 403 nm one-color

pump–probe experiment with 55 fs temporal resolution, that decays with a 1.3–1.5 ps time constant.

From this we conclude that, in contrast to previous reports, not all OClO molecules dissociate after

excitation with 403 nm light. In both cyclohexane and water we observe in the 403 nm experiment

an increase in stimulated emission between 0.5 and 2 ps that appears to be connected to the quantum

beat decay. We explain these results as the consequence of vibrational relaxation of the bending

mode of OClO. Relaxation from (␯ ,1,0) to (␯ ,0,0) leads to population of a state with a two times

1

higher transition dipole moment, which accounts for the increased stimulated emission. Further

proof that not all OClO molecules dissociate immediately after excitation is found in the

identiﬁcation of a stimulated emission contribution in femtosecond 400 nm pump/800 nm probe

experiments, which also decays with about a 1.5 ps time constant. Femtosecond 400 nm pump/267

nm probe measurements indicate that a fraction of the OClO molecules dissociate very rapidly, with

dissociation times of р60, 80, and 140 fs, in acetonitrile, water, and cyclohexane, respectively. An

anisotropy decay is resolved at 267 nm of the formed ClO in water and cyclohexane, with

anisotropy decay times of 0.17 and 0.27 ps, respectively. In all solvents a fraction of the ClOϩO

fragments recombine, with time constants of 1.2 and 4.1 ps in water, 6.0 ps in acetonitrile, and 8.9

ps in cyclohexane. In acetonitrile a secondary dissociation pathway is identiﬁed with a 2.1 ps time

constant. This pathway might also be responsible for the biexponentiality of the recombination

process in water. In particular, in acetonitrile and cyclohexane the data indicate cage escape of a

͓

DOI: 10.1063/1.1357202͔

I. INTRODUCTION

formed after recombination of ClOϩO. A further point of

importance is that OClO has only three vibrational degrees

of freedom. As a consequence the density of molecular levels

is relatively sparse, which strongly reduces the number of

intramolecular vibrational energy redistribution ͑IVR͒

The object of our studies is the ultrafast photo-induced

reaction dynamics of the three-atomic molecule OClO in liq-

uid solution ͑see Fig. 1͒. This molecule has met with consid-

erable interest over the last two decades, particularly due to

the possible involvement of OClO and its photochemical

products in the ozone depletion of the atmosphere. An as-

pect that makes the OClO molecule very interesting for stud-

ies of reaction mechanisms is that it has been found to break

31

pathways. Therefore, reaction dynamics characteristics of

individual vibronic levels might be identiﬁed in OClO. This

in contrast to typical dye molecules containing a few dozens

of atoms, where IVR and energy exchange with the solvent

typically lead to a rapid relaxation to the vibrationally unex-

1

,2

up in two ways, either giving ClOϩO, or ClϩO₂. Thus, in

a molecule with only three atoms a competition exists be-

tween at least two exit channels. In contrast, only one reac-

cited electronic state.

Electronic excitation of OClO occurs from the X B

ground state to the optically accessible A A state ͑Fig. 2͒.

2

1

2

32

2

3

4

5–10

11

tion channel exists in the studies of ICN, NaI, I ,

IBr,

2

The complexity of OClO photodissociation is partially due to

two other, optically inaccessible, electronic excited states

1

2

Ϫ 13,14

Ϫ 15

^C^H3^I^,I ,

BrI , or two indistinguishable channels in

2

Ϫ 16–20

21,22

22

23,24

the case of I₃,

HgI₂,

HgBr₂, and CH₂I₂.

On

2

with about the same internal energy as the A A₂state.

2

5

top of this OClO can also undergo isomerization to ClOO.

Strong interactions between these three excited electronic

2

6–29

27,28

In low temperature solid matrix ͑argon,

neon,

states are thus expected to occur. Peterson and Werner have

2

8

29

30

krypton, amorphous, and polycrystalline ice͒ studies

2

shown that the A A state interacts through spin–orbit cou-

2

ClOO proved to be the dominant end product, possibly

2

32–34

pling with the 1 A state.

Vibronic coupling exists be-

1

2

tween the 1 A and 1 B states. The magnitude of the cou-

1

2

^a^͒Electronic mail: henk.ﬁdder@fki.uu.se

pling between these states can be expected to be modulated

0021-9606/2001/114(15)/6781/14/$18.00

6781

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:

30.88.90.110 On: Sat, 20 Dec 2014 03:21:15

1

6782

J. Chem. Phys., Vol. 114, No. 15, 15 April 2001

Fidder et al.

tion. Along the asymmetric stretch coordinate (␯ ) dissocia-

3

tion can occur on all three PESs, the difference being that on

2

both the A A and the 1 A PES there is a barrier to cross,

2

1

whereas the 1 B PES does not have a barrier towards dis-

2

sociation.

The absorption spectrum of the OClO molecule is domi-

nated by progressions of vibronic lines (␯ ,␯ ,␯ ) in the

1

2

3

5,36

Franck–Condon active symmetric stretch (␯ ) mode,

as-

1

2

sociated with the X B →A A electronic transition ͑see

1

2

Fig. 3͒. Vibronic states (␯ ,0,0), where only ␯ is excited,

1

3

6

have the strongest transition moments. Combination bands

where the symmetric stretch is excited, together with the

bending mode, the asymmetric stretch mode, or both, are

also observed. These lead to additional progressions of the

type (␯ ,1,0), (␯ ,0,2), (␯ ,1,2), and (␯ ,0,4).

FIG. 1. Flow scheme of the possible photochemical reaction pathways of

OClO as previously known from the literature. The electronic state indices

1

2

Extensive experimental work on OClO in the gas phase

X B , A A , 1 A , and 1 B transform under symmetry operations in

1

2

1

2

37–49

39

the symmetry point group C₂_v; the indices 1 AЉ, 2 AЉ, 2 AЈ, and 1 AЈ

has been performed.

Richard and Vaida found that vi-

are the corresponding indices in the case of the point group C . We note

s

bronic states with excitation of the ␯ bending mode or ␯₃

2

that the asymmetric dissociation channel OClO→ClOϩO should be de-

asymmetric stretch mode have larger linewidths than states

scribed in the C group. For details see the text.

s

with only excitation of the ␯ symmetric stretch mode, which

1

suggests that these modes might expedite the predissociation

process. Bishenden and Donaldson⁴

0,41

found that the ClO

by the surrounding solvent. Studying the cuts through the

potential energy surfaces ͑PESs͒ along different vibrational

coordinates calculated by Peterson and Werner³²͑reproduced

in Fig. 2͒, we can get some idea of what might happen after

X(␯ϭ4) product yield ͑excitation 356 to 370 nm͒ is 2–3

times higher if the asymmetric stretching mode is excited,

than for the other vibronic transitions. Davis and Lee⁴²con-

cluded that the ClϩO2 channel reaches a maximum yield of

4% for 404 nm excitation. Below 400 nm ͑Ͼ3.1 eV͒ the

yield of this channel sharply decreased, indicating that a new

dissociation channel to ClOϩO becomes available. Further-

2

excitation to the A A₂PES. Along the symmetric stretch

vibrational coordinate (␯ ) the molecule cannot dissociate,

1

2

but relaxation might occur to the dark states 1 A and

1

2

1

B , which both are not dissociative along the symmetric

2

stretch coordinate ␯ . Along the bending coordinate (␯ ) the

more, they concluded that the yield of the ClϩO channel is

1

2

molecule is bound on the A A surface, which is crossed by

about ten times lower when additional energy is deposited in

2

43

the 1 A surface near the bottom. Relaxation to the 1 A

the asymmetric stretching mode. Both results indicate that

1

surface is a potential pathway for dissociation since this sur-

face appears to have only a small barrier towards dissocia-

the ␯₃asymmetric stretching mode promotes the ClOϩO

photodissociation channel. Energy partitioning into transla-

FIG. 2. Cuts through the ab initio potential energy surfaces as calculated by Peterson and Werner ͑taken from Ref. 32͒.

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:

130.88.90.110 On: Sat, 20 Dec 2014 03:21:15

J. Chem. Phys., Vol. 114, No. 15, 15 April 2001

Photochemical reaction of OClO in solution

6783

FIG. 3. Absorption spectra of OClO in

cyclohexane, acetonitrile, and water,

together with the gas phase stick spec-

trum constructed from the results of

Hubinger and Nee ͑Ref. 36͒. The spec-

trum of the pump pulse around 400

nm is depicted as the dashed line.

tional, vibrational, and rotational degrees of freedom of the

ClOϩO and conclude that almost all these fragments recom-

4

2–47

photodissociation products has been determined.

From

54

bine within 0.8 ps to form OClO again, vibrationally hot

these cw gas phase studies the picture has emerged that the

lifetimes of the excited state and outcome of the photochemi-

cal reactions depend on the vibronic mode that has been

excited.

56

and on the ground state PES. Signals between 459 and

1

024 nm, after excitation at 400 nm, were described by VER

5

6

on the OClO ground state PES. The original bleach was

found to recover almost entirely, leaving a quantum yield of

Ultrafast time-resolved gas phase studies of OClO by

5

4

8

only about 7% for the ClϩO dissociation channel, a yield

2

Baumert et al. with pumping between 308–352 nm dem-

onstrated biexponential wavelength-dependent dissociation

dynamics, suggesting the existence of at least two dissocia-

tion channels. The two lifetimes varied in the range of ϳ30

to ϳ500 fs. Ludowise et al.⁴⁹produced a quantum beat sig-

nal from a superposition of the ͑7,0,0͒ and the ͑6,0,2͒ vi-

bronic states. They found a 4.2 ps and 2.0 ps lifetime for the

similar to what Dunn et al.⁵²had previously concluded. They

also showed that the signal spectrum after 40 ps corresponds

very well to the absorption spectrum of Cl radicals in water,

which indicated that in water Cl and O are the only surviv-

2

ing end products.⁵⁴A recent time-resolved resonance Raman

study by Thomsen et al.⁵⁷provided direct evidence for the

formation of ClOO after excitation of OClO in water, based

͑7,0,0͒ and the ͑6,0,2͒ state, respectively. From a 250 fs de-

lay for the rise of the ClO signal they concluded that the 4.6

ps relaxation is to an intermediate state with a 250 fs lifetime

Ϫ1

on the appearance of a vibrational band at 1442 cm , which

corresponds to one of the vibrational frequencies of ClOO.

This band was found to appear 12.7 ps delayed and with a

rise time of 27.9 ps, and the signal decayed with a time

constant of 398 ps.

2

(

1 A and/or 1 B₂states͒, from where dissociation to

1

ClOϩO takes place.

Several studies of ultrafast reaction dynamics of OClO

in solution have been reported. Most work has been on OClO

Apart from Dunn et al.,⁵²only Reid and co-workers⁵⁸^–⁶⁴

have published data on OClO in solvents other than water. In

a time-resolved subpicosecond study on OClO in water and

_i_n_w_a_t_e_r_._S_i_m_o_n_a_n_d_c_o_-_w_o_r_k_e_r_s5

0–53

performed the ﬁrst

time-dependent studies of OClO reaction dynamics in solu-

tions. Their ﬁrst studies⁵

0–52

had ϳ80 ps time resolution,

5

8,59

which is too low to capture the fast reaction dynamics. These

data mainly reﬂect the absorption changes due to the solvent-

dependent mixture of end products Cl, ClO, O, and

acetonitrile,

they found that their data in water supported

the interpretation of Keiding and co-workers.⁵

4–56

However,

in acetonitrile they assigned the signals for probing at 800

5

0–52

ClOO.

molecules dissociates to ClOϩO, and only 10% to ClϩO₂.

In a later study with femtosecond time resolution on OClO in

They concluded that in water 90% of the OClO

58,59

nm after excitation at 400 nm to vibrationally hot ClOO,

5

2

in line with the previous interpretation by Chang and

53

Simon for OClO in water. Reid and co-workers also pub-

5

3

water, with excitation at 395 nm, they observed transients

between 400 and 700 nm that they assigned to vibrational

energy relaxation ͑VER͒ of vibrationally hot ClOO, which

they claim is formed within 0.7 ps.

lished a number of papers on resonance Raman investiga-

6

0,61

tions of the OClO reaction dynamics in cyclohexane,

acetonitrile,⁶²and water.

62–64

Their main conclusions

60,61,63

_K_e_i_d_i_n_g_a_n_d_c_o_-_w_o_r_k_e_r_s5

4–56

are that the PES for the asymmetric stretch coordinate is

dramatically altered in solution compared to the gas phase,

and that depolarization of the symmetric stretch Raman line

concluded from studies

with femtosecond time resolution on OClO in water that

there are no strong indications for formation of ClOO. They

demonstrated that in water OClO mainly dissociates to

2

indicates inﬂuence of the X B →1 A transition that has

1

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:

30.88.90.110 On: Sat, 20 Dec 2014 03:21:15

1

6784

J. Chem. Phys., Vol. 114, No. 15, 15 April 2001

Fidder et al.

not been observed in any absorption experiment so far. In

time-resolved resonance Raman measurements, Reid and co-

workers have found that in the geminate recombination of

In all cases the pump–probe signal was measured after

2

optically pumping the X B →A A transition at 400 nm.

1

2

The pump–probe experiments were performed at room tem-

ClOϩO and the subsequent vibrational relaxation in the

perature on 500 ␮m thick ﬂow cells with 1 mm thick CaF₂

2

X B state the excess vibrational energy is deposited along

windows. Fused silica or BaF windows turned out to have

1

2

6

2,64

the asymmetric stretch coordinate.

lower damage thresholds than CaF when exposed to chlo-

2

Summarizing, we mention that the bulk of work on

OClO in solution has been done for the solvent water. A

coherent interpretation for the reaction dynamics in this sol-

vent has emerged. Nevertheless, due to our higher time reso-

lution and the larger spectral bandwidth of our laser pulses

we have observed and time-resolved new features, which

compel us to introduce some important modiﬁcations to the

prevailing interpretation. Our data on the solvents acetoni-

trile and cyclohexane provide additional information on

OClO reaction dynamics that we could not have determined

from the water data alone. Combined, these data lead to a

more complex picture of reaction dynamics of OClO in sol-

vents than exists so far. From our measurements we con-

clude that the photodissociation dynamics of OClO in liquid

rine dioxide and the radicals formed after photoexcitation.

The relative large sample thickness was necessary to obtain a

sufﬁciently good signal-to-noise ratio. Group velocity mis-

match ͑GVM͒ between 400 and 800 nm, and between 400

and 267 nm, decreased the temporal resolution to about 160–

250 fs, as indicated by the width of the multiphoton pro-

cesses signal at zero delay. These multiphoton peaks were

also detected in the pure solvent. However, a possible con-

tribution of the cell windows to the multiphoton peaks

should not be discarded, in particular for the 400 nm/267 nm

experiments, and the stated values should be considered as

an upper limit. The relevant signals, on the other hand, will

only experience GVM from OClO and the solvent. There-

fore, we estimate that the effective temporal resolution is

more likely to be on the order of 120–150 fs in the two-color

experiments. However, in our analysis of the two-color ex-

periments we will use the upper limit values characterized by

the multiphoton peaks.

solution is dependent on which vibronic level is excited

2

within the A A state.

2

II. EXPERIMENT

The laser beams were focused with an aluminum mirror

with Rϭ30 cm radius of curvature. Pump intensities were

Pump–probe measurements were performed on a home-

2

less than 10 GW/cm at the focus, well below the threshold

built 1-kHz ampliﬁed Ti:sapphire laser system, which has

_b_e_e_n_d_e_s_c_r_i_b_e_d_e_l_s_e_w_h_e_r_e_.6

5,66

of self-phase modulation. Measurements were performed

with lower pump intensities to check the linearity of the

response to pulse energy density.

Solutions of OClO in cyclohexane, acetonitrile ͑both

Merck, analytical grade͒, and doubly deionized water were

prepared using the method of Bray.⁶⁷Oxalic acid ͑Merck͒

and potassium chlorate ͑Aldrich͒ were mixed with a few

drops of water and heated to 60 °C. The resulting gaseous

Frequency conversion of the

laser output of 1.2 mJ 40 fs pulses centered at 800 nm in 100

m BBO cut at 27° resulted in 200 ␮J 40 fs pulses at the

␮

second harmonic frequency. These were used in the single-

color pump–probe experiments where the pulses were cen-

tered at 403 nm. In the two-color experiments half of the

fundamental energy was used for frequency conversion to

give 100 ␮J of the second harmonic; the other half was used

either as a probe or as a mixing beam for third harmonic

generation. Half of the energy of the second harmonic was

used as the pump beam for both the 400/800 and the 400/267

nm experiments. In the 400/267 nm pump–probe experi-

ments the other half of the second harmonic beam was up-

converted to the third harmonic with the remainder of the

fundamental beam in a 100 ␮m BBO crystal cut at 45°.

Cross correlation between the 400 and 800 nm beams in a

mixture of OClO and CO was bubbled through the used

2

solvents. The resulting solutions were used as obtained.

Typical optical densities at 400 nm were 0.5–0.6 for the 500

␮m thick samples.

Measurements of electronic absorption spectra were per-

formed before, during, and after the pump–probe experi-

ments to check for possible degradation of the used solu-

tions.

100 ␮m thick BBO crystal resulted in correlation widths of

70–75 fs, from which we conclude that the pulse widths are

50–60 fs. In gas phase experiments the cross-correlation

III. RESULTS AND DISCUSSION

width between the 400 and 267 nm beams was 65–70 fs.

This leads us to conclude that the third harmonic also had a

pulse width of 50–60 fs. Polarization rotators and Glan–

Taylor prisms were inserted in the probe beams to study

anisotropy effects. The inclusion of these additional optical

elements in the pump and probe beams resulted in only mi-

nor pulse lengthening effects. In these anisotropy measure-

ments the probe polarization was rotated to 45°, with respect

to the pump polarization. Probe transmission through the

sample was then measured simultaneously for both the par-

allel and perpendicular components ͑400/800 nm experi-

ments͒ or directly after each other ͑400/267 nm experi-

ments͒.

This section is structured as follows. In Sec. III A we

describe what contributions might be expected at the various

probe wavelengths following pumping at 400 nm. In Sec.

III B we discuss the origin of the quantum beat feature that

the 403 nm pump/403 nm probe traces have in common for

all solvents. In Sec. III C we present a ﬂow scheme of the

OClO reaction dynamics we obtain from our experiments,

and give particulars concerning the ﬁt functions that we used

to describe our data. In Secs. III D–III F we describe our data

and present our interpretation of the reaction dynamics of

OClO in cyclohexane, acetonitrile, and water, respectively.

In Sec. III G we compare the results for the different solvents

and draw some general conclusions.

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:

130.88.90.110 On: Sat, 20 Dec 2014 03:21:15

J. Chem. Phys., Vol. 114, No. 15, 15 April 2001

Photochemical reaction of OClO in solution

6785

A. Possible pump–probe signal contributions

In Fig. 3 parts of the electronic absorption spectra of

OClO in cyclohexane, acetonitrile, and water are given, to-

gether with a stick spectrum representing the gas phase data

by Hubinger and Nee.³⁶We have performed some prelimi-

nary ﬁts on the absorption spectra that indicate that the cy-

clohexane, acetonitrile, and water spectra are redshifted com-

pared to the gas phase spectrum by about 275, 0, and 150

Ϫ1

cm , respectively. In these ﬁts we broadened the gas phase

spectrum by convoluting the stick spectrum with a Gaussian.

The full width at half maximum ͑FWHM͒ of the Gaussian

Ϫ1

required in the convolutions is 435, 680, and 705 cm , for

cyclohexane, acetonitrile and water, respectively. Keiding

5

6

and co-workers modeled the OClO absorption spectrum in

Ϫ1

water using a FWHM of 525 cm . For cyclohexane, which

has the most structured absorption spectrum, these prelimi-

nary ﬁtting results suggest that the relative strength of the

different transitions differs somewhat in cyclohexane from

the gas phase. In addition, in the onset range of the cyclo-

hexane absorption spectrum some of the simulated peaks fall

in between the experimental peaks, while the main progres-

sion, which is only related to the ␯ symmetric stretch vibra-

1

tional frequency, is accurately reproduced. This indicates

that the frequency of the symmetric stretch mode has not

changed noticeably in cyclohexane compared to the gas

phase; however, it suggests that the other modes might have

altered frequencies! Notice that since there are no actual pro-

gressions in the other modes, a difference is most likely to be

discerned in the onset range of the absorption spectrum. A

comparison of the absorption spectra in water of OClO and

the products Cl and ClO, as well as the gas phase spectrum

of the possible intermediate ClOO, can be found in Ref. 56.

FIG. 4. The 403 nm/403 nm pump probe data on OClO in cyclohexane

upper panel͒. The lower panel shows the signal from pure cyclohexane.

͑

3. Pump 400 nm—probe 800 nm

The signal at 800 nm after pumping at 400 nm is com-

posed of two contributions: a stimulated emission and an

increased absorption contribution ͑see Figs. 5–7͒. For OClO

in water the increased absorption signal at 800 nm has been

assigned by Keiding and co-workers⁵⁶to absorption from

vibronically hot OClO on the ground state PES. They con-

vincingly demonstrated, by modeling the decay dynamics

1. Pump 400 nm—probe 267 nm

This experiment allows for most options of contributing

data over the range of 400 to 1200 nm, that these signals are

entities. Near 267 nm the ClO radical has its absorption

2

related to vibrational relaxation on the X B ground state

6

8

1

maximum. The Cl radical, that peaks near 310 nm in H O,

2

PES of OClO. The almost complete recovery of the bleach at

still has about a three times higher extinction coefﬁcient at

4

00 nm in water after 20 ps also indicates that most OClO

2

67 nm than the ClO radical.⁶

8,56

The ClOO molecule has an

58

molecules are restored to their original state. The stimu-

2

7

absorption maximum at 247 nm in a neon matrix, but its

extinction coefﬁcient is still comparable to that of ClO at 267

nm. Only OClO itself has a negligible absorption at 267

lated emission contribution has not previously been ascribed

58,59

to the OClO molecule.

We assign this contribution to

2

stimulated emission from OClO molecules in the A A ex-

5

2

nm.

cited state to high vibrational levels on the ground state PES,

basically the reverse of the increased absorption transition.

2. Pump 403 nm—probe 403 nm

There seems to be good agreement between most

5

3–56,58,59

B. Origin of the quantum beat

authors

that signals at 403 nm after pumping around

4

03 nm are mainly related to ground state bleaching and

All 403 nm pump/403 nm probe signals show a quantum

beat during the ﬁrst few picoseconds, with an identical beat

frequency of 101 fs ͑ϳ330 cm ͒ and a decay constant for

2

stimulated emission from the excited A A state. The data

around 400 nm for OClO in water usually show a small

2

Ϫ1

residual contribution which is related to the ClϩO channel.

the amplitude of 1.3–1.5 ps ͑see Figs. 4–7͒. In Fig. 4 we

illustrate for OClO in cyclohexane that this signal is not

related to the solvent. The existence of this quantum beat

implies that, in contrast to what has been said in the literature

2

Below 390 nm the Cl–water charge transfer complex absorp-

tion exceeds this tiny contribution resulting in a net increased

5

absorption. By inspection of Figs. 4–7 it is clear that in all

solvents a quantum beat signal persists up to several picosec-

onds. This indicates that a signiﬁcant fraction of the OClO

molecules does not dissociate on a time scale of less than 1

for OClO in water, not all OClO molecules dissociate and

recombine within 0.7–0.8 ps.⁵

3,54

Since initially there are

only OClO molecules absorbing at this wavelength they have

to be responsible for the coherence. The quantum beat fre-

5

3–56,58,59

ps.

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:

130.88.90.110 On: Sat, 20 Dec 2014 03:21:15

6786

J. Chem. Phys., Vol. 114, No. 15, 15 April 2001

Fidder et al.

FIG. 5. Data and ﬁts for the transient pump–probe data obtained on OClO

in cyclohexane. The upper panel contains the 267 nm data for OClO in

cyclohexane, ﬁtted according to Eqs. ͑6͒ and ͑7͒ ͑solid lines͒. The solid dots

are the data for parallel polarization; the open squares are the data for

perpendicular polarization. These data have not been scaled to each other.

The spike around zero delay is due to a multiphoton absorption process. The

inset shows the rise due to ClO formation and anisotropy decay in greater

detail. The middle panel shows the 403 nm ﬁt on OClO in cyclohexane

using Eq. ͑8͒, together with the experimental data ͑dots͒. The inset shows

the strong coherent coupling between pump and probe in the data around

zero delay. The lower panel shows the strong coherent signal observed at

FIG. 6. Data and ﬁts for the transient pump–probe data obtained on OClO

in acetonitrile. The upper panel shows the ﬁt ͑solid line͒ according to Eq. ͑9͒

to our 267 nm data ͑solid dots͒ obtained on OClO in acetonitrile for parallel

pump and probe polarization. The open squares are the data for perpendicu-

lar pump probe polarization. The spike around zero delay is due to a mul-

tiphoton absorption process. The middle panel shows the ﬁt curve to our 403

nm data using Eq. ͑10͒ ͑solid line͒ together with the experimental curve

͑dots͒. The inset shows the strong coherent coupling between pump and

probe in the data around zero delay. The lower panel shows the ﬁt to our

800 nm data obtained on OClO in acetonitrile using Eq. ͑11͒ ͑solid line͒

together with the experimental data ͑dots͒. The spike around zero delay is

due to a multiphoton absorption process. The inset shows the longer time

behavior of the transient as predicted by Eq. 11 ͑compare to Ref. 58͒.

800 nm.

quency poses some difﬁculty, since it does not connect to

any vibrational frequency or accessible combination of vi-

bronic transitions as known from the gas phase absorption

1% or less, except for the excited state bending mode, where

3

5,36

Ϫ1

spectra.

In addition, a contribution from a ground state

increases by as much as 5% to 303.0 cm in Kr matrices

Ϫ1

wave packet created by a stimulated Raman excitation pro-

cess is also unlikely, since the lowest vibrational frequency

occur, compared to 288.1 cm in the gas phase. This clearly

opens the possibility for a further increase in vibrational fre-

quency of OClO dissolved in molecular liquids. We are

faced with the consequence that the vibrational frequency

not only has to deviate substantially from those derived for

the gas phase, but also this deviation has to be identical in all

the solvents we used. We propose that the bending mode

2

Ϫ1

in the X B ground state is 448 cm . Furthermore, we men-

1

2

tion that the X B ground state vibrational frequencies are

1

considerably larger than kT; therefore, OClO is mainly ex-

cited from the ͑0,0,0͒ vibrational mode.

It is possible that, going from the gas phase to the sol-

vent phase, vibrational frequencies and energy separations

are altered. We ﬁnd indications for this in the electronic ab-

sorption spectra ͑see Sec. III A͒ and the emission spectra

analysis of OClO in solid neon, argon, and krypton matrices

at low temperatures by Liu et al.³¹Nearly all modes appear

(␯ ) is the best suitable candidate, since it exhibits the larg-

2

est change in frequency in noble gas matrices, and because it

is the only mode that has a lower frequency in the gas phase

Ϫ1

͑ϳ288 cm ͒ than the observed 330 cm for the quantum

beat. The increase in frequency can be visualized as the re-

sult of a steepening of the PES of this vibration caused by a

lack of free space in the solvent to perform the vibrational

motion. This would explain why the increase is similar for

to have slightly higher vibrational frequencies in noble gas

2

matrices than in the gas phase, both in the X B state and

1

2

the A A state. In all cases the change in frequency is about

2

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:

130.88.90.110 On: Sat, 20 Dec 2014 03:21:15

J. Chem. Phys., Vol. 114, No. 15, 15 April 2001

Photochemical reaction of OClO in solution

6787

FIG. 8. Flow diagram representing dynamical pathways of OClO photo-

chemistry in solution based on the results discussed in this paper.

cies. This makes it likely that two similar beat frequencies

cause the quantum beat. Our time resolution, however, is

insufﬁcient to justify a detailed line shape analysis on the

Fourier transform. With our time resolution and the spectral

bandwidth of our laser pulses we are incapable of creating

beats between the ͑5,1,0͒ and ͑6,1,0͒ states or the ͑6,0,0͒ and

the ͑5,0,0͒ states.

Based on our data of the three solvents, we believe that

the 1.3–1.5 ps decay is caused by relaxation from states

(

␯ ,1,0) to states (␯ ,0,0). For the various solvents this

1 1

would include vibrational relaxation from ͑6,1,0͒ to ͑6,0,0͒

and from ͑5,1,0͒ to ͑5,0,0͒. A consequence of such relaxation

is that we should observe an increase of stimulated emission

accompanying this decay, the rise time of which should

match the decay time of the quantum beat, which is exactly

what we see in the 403 nm data of cyclohexane. The reason

for this is that the gas phase absorption cross section for

FIG. 7. Data and ﬁts for the transient pump–probe data obtained on OClO

in water. The upper panel shows the ﬁts of our 267 nm data obtained on

OClO in water according to Eqs. ͑12͒ and ͑13͒ ͑solid lines͒. The solid dots

are the experimental data for parallel polarization; the open squares are the

experimental data for perpendicular polarization. The spike around zero de-

lay is due to a multiphoton absorption process. The inset shows the rise due

to ClO formation and anisotropy decay in greater detail. The middle panel

shows our 403 nm data on OClO in water ͑dots͒ and the ﬁt described by Eq.

states (␯ ,0,0) is 1.6–2.1 times higher than for states

1

͑14͒ ͑solid line͒. Again, the inset shows the strong coherent coupling signal

36

(

␯₁,1,0). The observed increase in stimulated emission

in the data around zero delay. The lower panel shows our 800 nm data ͑dots͒

obtained on OClO in water, together with the ﬁt given by Eq. ͑15͒ ͑solid

line͒. The spike around zero delay is due to a multiphoton absorption pro-

cess.

will, however, also depend on the relative intensity of the

probe laser pulse at the respective frequencies for the relaxed

and unrelaxed state. In combination with the different sol-

vent shifts this explains the solvent dependence of the

change in stimulated emission due to this relaxation process.

the different solvents, since the amount of unoccupied space

does not vary dramatically from solvent to solvent.

Thus, we suggest that the lifetime of the single excited bend-

2

ing mode vibration on the A A PES is 1.3–1.5 ps. Vibra-

2

The quantum beat reﬂects the interference of the polar-

izations on two different vibronic transitions. Based on gas

tional relaxation is known to occur on the picosecond time

69

scale, and the value found for the bending mode in the

4

0–43

phase work

and data presented in Secs. III D–III F we

2

electronic excited A A state can be compared to vibrational

2

propose that states of the type (␯ ,␯ ,2) dissociate within a

7

1

2

relaxation time scales of other small molecules like I ,

2

few hundred femtoseconds after creation by a 403 nm pulse.

We propose that the quantum beat is generated between

states (␯ ,1,0) and (␯ ,0,0). In this case the beat is mainly

Ϫ

14,17–19

Ϫ 70,71

Ϫ 72

Ϫ

73

I ,

CN ,

ClO , and N .

2

3

1

between the states ͑6,1,0͒ and the ͑6,0,0͒, and to a lesser

extent the ͑5,1,0͒ and ͑5,0,0͒ states. However, in the gas

C. Flow scheme for the reaction pathways

As guidance to the discussion of our results we ﬁrst

present in Fig. 8 a ﬂow scheme of the OClO photodissocia-

tion reaction based on our experimental results. As is clear

from this scheme we describe the different processes as a

series of consecutive ﬁrst-order kinetic reaction steps:

phase the separation between the ͑6,0,0͒ and the ͑5,0,0͒ state

Ϫ1 36

was found to be about 681 cm

,

which is only slightly

more than twice the bending mode frequency. This indicates

also that a beat between the ͑6,0,0͒ and the ͑5,1,0͒ states is

generated with almost the same frequency. The Fourier

transform of the quantum beat typically has a FWHM of

␶1

␶2

␶3

Ϫ1

1

5–20 cm , and shows some asymmetry to higher frequen-

A →B →C →D.

͑1͒

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:

130.88.90.110 On: Sat, 20 Dec 2014 03:21:15

6788

J. Chem. Phys., Vol. 114, No. 15, 15 April 2001

Fidder et al.

Based on the scheme in Fig. 8 and information in Sec. III A,

the signal at 267 nm is mainly fed by steps a and e, and

decreases through step b for the ClO contribution; the feed-

ing of a possible Cl–solvent charge transfer complex contri-

bution at this wavelength is related to step f. The signal at

32%–35% in states (␯ ,␯ ,2), and about 15%–20% in states

1

2

(␯ ,1,0) ͑see Table I͒. The above ﬂow scheme therefore sug-

1

gests that in our experiments only one third of the excited

molecules enters the direct dissociation channel associated

with step a. In the next paragraphs we discuss the dynamics

of OClO in solution for each solvent separately.

403 nm is inﬂuenced by the decay times ␶ of the steps a, c, d,

e, and f, where steps a, c, e, and f lead to a reduction in probe

transmittance, and step d can lead to both an increase and

reduction in probe transmittance. The stimulated emission

D. Cyclohexane

2

For OClO in cyclohexane no time-resolved measure-

ments have been reported previously. However, Reid and

co-workers have performed cw resonance Raman measure-

ments on OClO in cyclohexane.⁶

signal at 800 nm we assign to OClO molecules on the A A₂

PES, that do not immediately dissociate, and therefore can

connect to steps d, e, and f. The positive contribution at 800

0,61

†

nm we ascribe to absorption of vibrationally hot OClO on

2

the X B surface to the A A PES, that is related to steps b

1

2

1. Pump 400 nm—probe 267 nm

and c. Note, however, that the signal at 800 nm is observed

halfway through the cascade of vibrational relaxation steps

that brings the recombined hot OClO to the relaxed ground

state. Therefore, the feeding time of step b is a lower limit

for the feeding time of this signal, and the time for the

ground state recovery should be slower than the ‘‘decay time

out’’ ͑step c͒ at 800 nm.

The different ﬁts we use to describe our experimental

data are composed of signal contributions that decay expo-

nentially after excitation

The signals with probing at 267 nm are shown in the

upper panel of Fig. 5. In accordance with conclusions from

†

previous work on OClO in water,⁵

3–56,58,59

we assign these

signals mainly to the ClO fragment ͑see Sec. III A͒. We note

further that chlorine atoms in the gas phase do not absorb

between 140 and 370 nm,⁷⁴and we do not know if chlorine

also forms a charge transfer complex with cyclohexane. The

following equations provided excellent ﬁts to the data.

Pump and probe polarization parallel:

ϱ

A͑t͒ϭexp͑Ϫt/␶ ͒,

͑2͒

1

S

ʈ

͑␶͒ϭ

͵

dt

͕

͓0.57•͑1.02•͑e^Ϫ^t^/⁸^.⁹Ϫe^Ϫ^t^/⁰^.¹⁴͒͒

Ϫϱ

contributions that are fed by one rate (␶ ) and decay by

another (␶₂)

1

ϩ0.43•͑1ϪeϪt/0.14_͒_]

^␶2

•H t •A tϪ␶ ϩ67•A ␶ ;

͑

͒

͑

͒

͖

͑ ͒

͑5͒

͑6͒

B͑t͒ϭ

•͓exp͑Ϫt/␶ ͒Ϫexp͑Ϫt/␶ ͔͒,

͑3͒

2

1

␶₂

Ϫ␶₁

Pump and probe polarization perpendicular:

and contributions that are fed after two consecutive ﬁrst-

ϱ

1

2

3

S_Ќ͑␶͒ϭ

͵

dt

͕

͓0.49•͑1.02•͑e^Ϫ^t^/⁸^.⁹Ϫe^Ϫ^t^/⁰^.²⁰͒͒

order reaction steps (␶ ,␶ ) and decay by a third (␶ )

Ϫϱ

␶₃

␶₂

␶₁

C͑t͒ϭ

•

ͫ

• exp͑Ϫt/␶ ͒Ϫ

ϩ0.51•͑1Ϫe^Ϫ^t^/⁰^.²⁰͒]

H͑t͒•A͑tϪ␶͒

2

␶₂

Ϫ␶₁␶₂Ϫ␶₃

␶₁Ϫ␶₃

•

͖

ϩ27•A͑␶͒,

␶₂

␶₁

•

exp͑Ϫt/␶ ͒

ͬ

Ϫ

ͫ

Ϫ

ͬ

1

␶₂

Ϫ␶₃␶₁Ϫ␶₃

where H(t) is the Heaviside function ͓H(tу0)ϭ1, and

H(tϽ0)ϭ0], and ␶ is the pulse delay time. The response is

convoluted with the autocorrelation response function A(␶),

which has a FWHM of 226Ϯ15 fs for these measurements.

The ﬁt for both parallel and perpendicular probe polarization

contains an exponential decay with 8.9 ps decay time and an

offset, both with similar weights, to describe the decay rates

of the ClO contribution. The only difference is the rise time

that describes the formation of the ClO. We ﬁnd a rise time

of 140 fs for parallel and 200 fs for perpendicular probe

polarization. The anisotropy decay function we constructed

from these decays according to the formula

•

exp͑Ϫt/␶ ͒.

͑4͒

3

These expressions are all normalized; that is, A(t)ϩB(t)

ϩC(t)ϩD(t)ϭ1 holds at all times ͑D is formed after the

relaxation step with time constant ␶ ). This implies that the

3

relative weight they obtain in a ﬁt function composed of a

number of these expressions corresponds to the relative

amount of molecules that have followed the corresponding

pathway. The appropriate sum of contributions is convoluted

with the Gaussian response function, and for the 267 and 800

nm probe data this Gaussian response function is also added

at time-zero delay to account for the coherence ͑multiphoton

absorption͒ contribution during time overlap. This contribu-

tion is not included in the ﬁts of the 403/403 nm data, where

a strong bleach before and a strong increased absorption after

zero time delay clearly indicates that more complicated mul-

tiphoton processes contribute in the case of identical pump

and probe pulse wavelengths.

S ͑␶͒ϪS ͑␶͒

Ќ

ʈ

r͑t͒ϭ

,

͑7͒

S ͑␶͒ϩ2S ͑␶͒

Ќ

ʈ

shows a decay time of ϳ0.27 ps. These data suggest that ClO

fragments are formed in cyclohexane with a 140 fs time con-

stant ͑step a, Fig. 8͒, and that the ClO fragments obtain ro-

tational momentum during the dissociation, which causes a

depolarization decay with a 0.27 ps time constant. The 8.9 ps

decay time we assign to recombination of ClOϩO to OClO

The 403 nm pump pulse we used in our experiments

creates about 45%–53% population in states (␯ ,0,0), about

1

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:

130.88.90.110 On: Sat, 20 Dec 2014 03:21:15

J. Chem. Phys., Vol. 114, No. 15, 15 April 2001

Photochemical reaction of OClO in solution

6789

͑

step b͒. Of the three solvents we studied, cyclohexane

ate this 0.1 ps decay with the ClO formation ͑0.14 ps͒. Our

calculations indicate that about 32% of the excited state

population is in states (␯ ,␯ ,2), which we expect to cause

clearly has the slowest recombination time. If the remaining

offset is only caused by ClO absorption the data suggest that

for the formed ClOϩO there is about a 43% chance that a

fragment escapes from the solvent cage, when pumping at

1

2

this component. The second part of the ﬁt describes the in-

crease of stimulated emission, with a 1.3 ps rise time, that

becomes noticeable after 300 fs. This 1.3 ps rise time is

considerably faster than the 8.9 ps decay time observed at

267 nm, and is therefore not related to recombination of

ClOϩO. The 1.3 ps rise time exactly matches the decay time

of the quantum beat, strongly suggesting that the two are

4

00 nm. Note that we do not know if a possible chlorine–

cyclohexane charge transfer complex contributes to the sig-

nal. More experiments are needed to clarify this issue.

From gas phase work it has been concluded that for ex-

citation energies near 3.1 eV the dissociation process

changes, which has been attributed to depositing sufﬁcient

energy into the OClO molecule for direct dissociation over a

barrier. This strongly suggests that full understanding of

the OClO reaction dynamics involves knowledge over the

entire four-dimensional space set up by the three vibrational

coordinates and the potential energy. In other words: disso-

ciation channels could depend on combinations of vibra-

Ϫ1

connected. In cyclohexane, which causes a 275 cm redshift

of the OClO absorption spectrum compared to the gas phase,

the excited state superposition created by our pump pulse is

dominated by the ͑6,0,0͒ and the ͑5,02͒ mode ͑see Fig. 3͒. To

a lesser extent the ͑6,1,0͒, ͑5,1,0͒, and ͑4,1,2͒ modes contrib-

ute. Since we assume that states (␯ ,␯ ,2) are associated

4

2

1

2

with the nearly instantaneously formed ClO, the dynamics at

403 nm after a few hundred femtoseconds has to be related

to states (␯ ,1,0) and (␯ ,0,0). The strongest contributions

tional motions. The population is initially created on the

A A PES. In Sec. III B we proposed that the quantum beat

2

1

is between states (␯ ,1,0) and (␯ ,0,0). Given PES cross

sections calculated by Werner and Petersen ͑Fig. 2͒, states

to this beat should come from the superposition of the ͑6,1,0͒

and the ͑6,0,0͒ mode, and the superposition of the ͑5,1,0͒ and

the ͑6,0,0͒ mode. The relaxation from ͑6,1,0͒ to ͑6,0,0͒ re-

sults in the creation of a state with a 2.1 times higher oscil-

lator strength,³⁶and located close to the maximum of the

laser pulse. Therefore, this relaxation should give rise to a

strong increase of stimulated emission, which explains the

rising of the bleach with the same time constant as the decay

of the quantum beat. Since this rising contribution does not

decay noticeably over the ﬁrst 4 ps, the lifetime of the ͑6,0,0͒

state has to exceed 10 ps in cyclohexane. Consequently, a

large part of the nondecaying offset signal is also related to

stimulated emission from the ͑6,0,0͒ state.

1

3

2

that include excitation of the ␯ coordinate are most likely to

3

2

lead to rapid direct dissociation from the A A state, and are

2

therefore probably connected to the rise of the ClO absorp-

tion signal at 267 nm. This is also in agreement with conclu-

sions from gas phase work, which indicate that the ClO yield

is three times higher,⁴⁰and the Cl yield ten times lower, for

43

modes with similar total excitation energy if the ␯ mode is

3

also excited.

2. Pump 403 nm—probe 403 nm

During the ﬁrst picosecond OClO in cyclohexane shows

a strong coherent artifact, followed by a decrease in bleach-

3. Pump 400 nm—probe 800 nm

ing up to ϳ300 fs, after which the bleach increases ͑see Figs.

Probing at 800 nm we only observe a very strong coher-

ence spike ͑Fig. 5, lower panel͒. In previous work on OClO

4, 5͒. In contrast to water ͑vide infra͒, the bleach does not

decay during the ﬁrst 4 picoseconds and the signal reaches a

minimum after about 4 ps. We also see a strong quantum

beat superimposed on the signal with a beat period of 101 fs,

in water and acetonitrile, signals were observed for this type

53–56,58,59

of experiment.

For OClO in water, these signals

were attributed to ͑increased͒ absorption from vibrationally

Ϫ1

which corresponds to an energy of 330 cm . The amplitude

54–56,58,59

2

hot OClO on the X B surface to the A A surface.

1

2

of the quantum beat decreases with a decay time of 1.3 ps.

The data are well described by the function ͑see Fig. 5,

middle panel͒

This vibrationally hot OClO was believed to be formed by

recombination of ClOϩO to OClO. The absence of a clear

signal could be due to the longer recombination time. Obvi-

ously our data do not give evidence for this interpretation.

ϱ

Ϫt/1.3

S͑␶͒ϭ

͵

dt͓Ϫ0.27•͑e

͒•cos͑2␲t/0.101͒

Ϫϱ

E. Acetonitrile

Ϫ0.60•e^Ϫ^t^/⁰^.¹Ϫ0.40Ϫ0.29•͑1Ϫe^Ϫ^t^/¹^.³͔͒

1

. Pump 400 nm—probe 267 nm

•

H͑t͒•A͑tϪ␶͒.

͑8͒

Reid and co-workers have previously studied OClO re-

The width of the autocorrelation function was 55 fs. This ﬁt

consists of two parts. The ﬁrst part is a two-component decay

function that starts at time tϭ0, containing a decay time of

action dynamics in acetonitrile, using subpicosecond pump

5

8,59

probe spectroscopy

and time-resolved resonance Raman

spectroscopy.⁶²Their data showed a delayed rise, which they

connected to a 2.2 ps decay time with negative amplitude,

0.1 ps and an offset, with an oscillation superimposed on it

with a time period of 101 fs and amplitude decay time of 1.3

ps. Due to the strong coherent artifact the ultrafast 0.1 ps

decay time is not accurately determined. However, its exis-

tence is clear since the bleaching still drops between 200 and

followed by a slow decay of ϳ15% of the signal during the

ﬁrst 20 ps, leaving a large offset.⁵

8,59

Even though we notice

the similarity between our signals at 267 nm ͑Fig. 6, upper

panel͒ and those of Philpott et al.,⁵

8,59

there are some signiﬁ-

3

00 fs, while the coherent artifact contribution after tϭ0 has

cant differences in the dynamics we extract from it. This is

mainly a result of our more than three times better time

an increased absorption signature. It seems logical to associ-

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:

130.88.90.110 On: Sat, 20 Dec 2014 03:21:15

6790

J. Chem. Phys., Vol. 114, No. 15, 15 April 2001

Fidder et al.

resolution. In contrast to Philpott et al.⁵⁸^,⁵⁹we ﬁnd that the

main rise of the signal shows no signiﬁcant delay: after ϳ0.2

ps the signal has nearly reached maximum. Only an addi-

tional 7% increase appears to occur during the following 2.5

ps. This 7% signal rise is of comparable size to the slow

noise oscillations between 18 and 24 ps delay; thus, it is not

decays with a time constant of 37 ps. Note that the decay

time for the quantum beat is similar to that observed for the

data in cyclohexane.

Since there is no redshift of OClO in acetonitrile com-

pared to the gas phase, our pulse mainly excites the ͑6,0,0͒

and ͑5,1,0͒ modes, and to a lesser extent the ͑4,1,2͒, ͑5,0,0͒,

͑4,0,2͒, ͑5,0,2͒, and the ͑6,1,0͒ modes. Due to the larger

spectral broadening more states are signiﬁcantly populated

by the pump pulse in acetonitrile and water than in cyclo-

hexane. Following our interpretation for cyclohexane, that

the quantum beat decay is related to relaxation of the bend-

1

00% certain that this is a real feature! Given that Phillpott

5

8,59

et al.

observed a similar time delay ͑2.2 ps decay time͒

before reaching maximum, we are inclined to believe it is a

real feature.

The data for both parallel and perpendicular pump and

probe polarization show the same behavior. However, since

the signal with perpendicular probe polarization has a lower

signal/noise ratio, it cannot provide a critical test to the pa-

rameters obtained for parallel polarization. Therefore, we

present only the ﬁt data for parallel polarization ͑see Fig. 6͒.

ing mode ␯ , seems to agree well with more excitation of

2

the ͑5,1,0͒ than the ͑6,1,0͒ mode. The relaxed ͑5,0,0͒ state is

located at a part of the probe pulse spectrum with a much

lower intensity, which certainly compensates for the higher

oscillator strength, and thereby could lead to no apparent

increase of stimulated emission upon this relaxation. The off-

Pump and probe polarization parallel:

set should then be due to stimulated emission and bleach.

ϱ

58,59

[0.22•(1.54•(e_Ϫ_t_/₆_.₀_Ϫ_eϪt/2.1))ϩ0.78]

The slow 37 ps decay obtained by Philpott et al.

at 400

S

ʈ

͑␶͒ϭ

͵

dt

͕

Ϫϱ

nm could partially be due to restoration of the ground state

absorption stemming from recombination of ClOϩO to

OClO. This decay process should be slower than the 6.0 ps

recombination time found at 267 nm. Relaxation from the

excited state to the ground state not involving dissociation

could also contribute to this 37 ps decay.

•

H(t)•A(tϪ␶)

͖

ϩ38•A͑␶͒,

͑9͒

where A(␶) has a FWHM of 240 fs. This ﬁt suggests that

78% of the ClO fragments are formed almost instantaneously

and do not recombine due to cage escape ͑the offset͒, formed

immediately after excitation by the pump beam. Basically

the rise time of this direct ClO formation is unresolvable

with our time resolution, which indicates that it is in the

range of 0–60 fs. The remaining 7% of the observed signal,

which is related to ϳ22% of the dissociating molecules ac-

cording to the ﬁt function, is formed with a 2.1 ps time

constant and disappears through recombination with a 6.0 ps

time constant. Here, we point out that this ﬁt is not unique.

For instance the 2.1 ps/6.0 ps/22% combination can be re-

placed by 1.5 ps/7.0 ps/25%, a ﬁt of comparable quality.

Alternatively, part of the instantaneously formed ClO could

also recombine, which we also found to give acceptable ﬁts

although requiring more ﬁt parameters. A preference for

which model to choose cannot be made with these data only.

An appealing motivation for a lower recombination yield of

the instantaneously formed fraction over the secondary frac-

tion ͑found for both descriptions͒ is that ‘‘impulsively’’

formed ClOϩO probably contains more translational kinetic

energy, and therefore is more likely to escape the solvent

cage. This argument is supported by the faster ClO formation

time in acetonitrile compared to cyclohexane.

In contrast to cyclohexane, in acetonitrile the initial fast

.25 ps decay component is slower than the ClO formation

0

time observed at 267 nm. At this point it is important to

stress that the multiphoton coherence signal around zero de-

lay time, which signiﬁcantly distorts this signal during the

ﬁrst 200 fs, is not included in this ﬁt. Therefore, care should

be taken with the interpretation of this fast component.

3

. Pump 400 nm—probe 800 nm

The function that describes the decay at 800 nm is ͑see

Fig. 6, lower panel͒

ϱ

Ϫt/1.5

S͑␶͒ϭ

͵

dt

͕

͓Ϫ0.48•e

Ϫϱ

ϩ0.52•͑8.49•eϪt/7.2_ϩ₀_.₇₆_•_e_Ϫ_t_/₂_.₁_Ϫ₉_.₂₅_•_e_Ϫ_t_/₆_.₀_͒_͔

•

H͑t͒•A͑tϪ␶͒

͖

ϩ6.7•A͑␶͒.

͑11͒

The autocorrelation function A(␶) has a FWHM of 162 fs.

Basically Eq. ͑11͒ consists of only two contributions. The

ﬁrst is a stimulated emission contribution that decays with a

1

.5 ps decay time. The second contribution, a rising in-

2

. Pump 403 nm—probe 403 nm

creased absorption, is related to the ClOϩO formation and

recombination. Apart from the weight of 0.52, the param-

eters of this contribution are based on the rise and decay time

of the second fraction of ClO observed at 267 nm, and the

7.2 ps decay time has been determined by Philpott et al.⁵⁸

from a measurement that spans a 40 ps range. All other pa-

rameters of the second contribution are calculated from the

The ﬁt describing the 403 nm one-color pump probe

signal in acetronitrile ͑Fig. 6͒ is

ϱ

Ϫt/1.5

S͑␶͒ϭ

͵

dt͓Ϫ0.20•͑e

͒•cos͑2␲t/0.101͒

Ϫϱ

Ϫt/0.25

Ϫ0.34•e

Ϫ0.66͔•H͑t͒•A͑tϪ␶͒.

͑10͒

58

expression for C(t) in Eq. ͑4͒. Philpott et al. previously

The FWHM of the function A(tϪ␶) is 55 fs. Philpott

et al.

does not noticeably decay during the ﬁrst 6 ps eventually

assigned the negative stimulated emission signal contribution

to a coherent artifact, which is highly unlikely given the long

time it takes before the 800 nm negative signal changes into

5

8,59

have shown that one third of the ͑offset͒ signal that

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:

130.88.90.110 On: Sat, 20 Dec 2014 03:21:15

J. Chem. Phys., Vol. 114, No. 15, 15 April 2001

Photochemical reaction of OClO in solution

6791

an increased absorption signal. Given the agreement of the

ﬁt to our data with two decay exponents, 1.2 and 4.1 ps, and

an offset. This biexponential decay we ascribe to recombina-

tion of ClOϩO to form vibrationally hot OClO on the

ground state PES. Signals for both parallel and perpendicular

polarization have been measured, that can be described using

the same decay parameters, with only small modiﬁcations in

the weights of the different components ͑see Fig. 7, upper

panel͒. The ﬁt functions that describe our data are:

1

.5 ps decay time of this stimulated emission signal with the

.5 ps decay time of the quantum beat, it is very tempting to

2

connect both to the same vibrational motion on the A A₂

PES. Note that our time resolution in the 800 nm data is

almost three times lower than for the 403 nm data; therefore,

we can not resolve the quantum beat in the 800 nm data.

The increased absorption signal at 800 nm is assigned by

Philpott et al.

than OClO, to which similar signals in water are

assigned.

5

8,59

to vibrational relaxation of ClOO, rather

Pump and probe polarization parallel:

5

4–56,58,59

Their main argument is that the increased

ϱ

absorption signal at 800 nm is somewhat stronger in aceto-

nitrile ͑the peak is about 50% higher͒, while the cage escape

clearly is higher, and therefore the recombination yield

͓0.32•͑1.02•͑e_Ϫ_t_/₄_.₁_Ϫ_eϪt/0.08͒͒

S

ʈ

͑␶͒ϭ

͵

dt

͕

Ϫϱ

ϩ0.51•͑1.07•͑e^Ϫ^t^/¹^.²Ϫe^Ϫ^t^/⁰^.⁰⁸͒͒

ϩ0.17•͑1Ϫe

H͑t͒•A͑tϪ␶͒

5

8,59

should be lower.

First, we remark that this argument de-

pends on the conclusion that all OClO molecules initially

dissociate, which clearly is in contradiction with our results.

Second, a comparison of the apparent signal strength versus

the actual amplitude in the ﬁt function for our acetonitrile

and water ͑vide infra͒ data at 800 nm, indicates that the same

amount of recombined molecules give rise to about three

times more increased absorption signal for acetonitrile. This

is a consequence of two factors: ͑1͒ The very different ratio

of the rise and decay time at 800 nm for the two solvents; ͑2͒

The increased absorption contribution in water peaks at an

earlier delay time and the apparent peak height is therefore

much stronger reduced by the stimulated emission contribu-

tion. Thus, assuming that the increased absorption signals in

both water and acetonitrile are connected to the same spe-

cies, our data indicate that in water twice as many molecules

are responsible for the signal. Stated differently: the gemi-

nate recombination yield in water is two times higher than in

Ϫt/0.08

͔͒

•

͖

ϩ26•A͑␶͒,

͑12͒

Pump and probe polarization perpendicular:

ϱ

S_Ќ͑␶͒ϭ

͵

dt

͕

͓0.41•͑1.03•͑e^Ϫ^t^/⁴^.¹Ϫe^Ϫ^t^/⁰^.¹³͒͒

Ϫϱ

ϩ0.39•͑1.12•͑e^Ϫ^t^/¹^.²Ϫe^Ϫ^t^/⁰^.¹³͒͒

ϩ0.20•͑1Ϫe

•H͑t͒•A͑tϪ␶͒

Ϫt/0.13

͔͒

͖

ϩ13•A͑␶͒.

͑13͒

The FWHM of the autocorrelation A(␶) is 245 fs. A signiﬁ-

cant difference is found in the rise time of the two signals: 80

fs for parallel pump and probe polarization and 130 fs for

perpendicular polarization. Similar to the difference we ob-

served in cyclohexane, we again ascribe it to the ClO frag-

ment picking up rotational momentum during the bond-

breaking process. We extract a 0.17 ps anisotropy decay time

for the ClO fragment in water, and a 80 fs ClO formation

time.

acetonitrile. This demonstrates that it is not required to in-

voke the formation of ClOO in acetonitrile⁵

8,59

versus OClO

in water to explain a larger maximum signal strength at 800

nm in acetonitrile.

F. Water

1. Pump 400 nm—probe 267 nm

The reaction dynamics of OClO in solvents has been

2

. Pump 403 nm—probe 403 nm

5

2–59,62–64

studied most intensively in water.

Data by the

5

4,56

Keiding group

with probing in the range of 250 to 320

The signal at 403 nm shows a strong coherent artifact

nm gave no indication of the formation of any signiﬁcant

fraction of ClOO in water. Recently, Thomsen et al.⁵⁷dem-

onstrated with time-resolved resonance Raman experiments

that some ClOO is formed after excitation of ClOO in water.

The ClOO formation process is 13 ps delayed and rises with

a time constant of 28 ps. This indicates that ClOO does not

inﬂuence the dynamics measured at 267 nm during the ﬁrst

followed by a bleaching that decreases for 0.7 ps, and then

increases again up to about 1.5 ps, whereafter it decays with

an average time constant of 4 ps ͑Fig. 7, middle panel͒. The

initial drop followed by an increase has not previously been

identiﬁed by others.⁵

3–56,58,59

It has been shown that at longer

delay times the signal levels off to a tiny residual offset

related to the yield of the ClϩO channel, which has been

2

5

4–56

53–55,58

1

0 ps. In accordance with Keiding and co-workers

and

estimated at 7%–10%.

Superimposed on the decay we

5

7

the result of Thomsen et al., we assign the dynamics at 267

nm mainly to ClO.

observe a quantum beat, with an amplitude decay time of 1.5

Ϫ1

ps, and a beat frequency corresponding to 330 cm ͑101 fs͒,

The signal with probing at 267 nm ͑Fig. 7͒ shows a rapid

decay with a time constant of about 2.5 ps, which levels off

to about 20% of the maximum amplitude, after which no

signiﬁcant decay is observed up to 50 ps ͑cf. the data in Refs.

an identical frequency and comparable decay time as that

observed in cyclohexane and acetonitrile.

Due to the rise and decay on comparable time scales, the

ﬁt of these data becomes quite complicated. An example of

an equation that describes the data fairly well is ͑compare

Fig. 7͒

5

5, 56, 58͒. This residual offset signal has been assigned to

5

4

the Cl–water charge transfer complex. We obtain a better

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:

130.88.90.110 On: Sat, 20 Dec 2014 03:21:15

6792

J. Chem. Phys., Vol. 114, No. 15, 15 April 2001

Fidder et al.

TABLE I. Parameters concerning the initial degree of excitation of the

different vibronic states.

ϱ

Ϫt/1.5

S͑␶͒ϭ

͵

dt͓Ϫ0.28•͑e

͒•cos͑2␲t/0.101͒

Ϫϱ

Initial population in states

Ϫt/0.3

Ϫt/5.1

Ϫ0.57•e

Ϫ0.43•e

Solvent

(␯ ,0,0)

(␯ ,1,0)

(␯ ,␯ ,2)

1 2

1

Ϫt.2.0

Ϫe^Ϫ^t^/¹^.⁵͔͒͒

Ϫ1.27•͑4.0•͑e

Cyclohexane

Acetonitrile

Water

53%

45%

15%

20%

32%

35%

•

H͑t͒•A͑tϪ␶͒.

͑14͒

In our opinion, ﬁtting these data requires too many similar

decay times. Therefore, we do not wish to attach too much

signiﬁcance to most numbers in this ﬁt. The main exception

is the 1.5 ps quantum beat decay time, since the beat can be

separated from the total dynamics. This decay we again as-

ably not required since the recombination time was found to

be 6.0 ps, ﬁve times longer than the main recombination time

in water. The decay of the increased absorption is also per-

fectly described if we only use the contribution from the 1.2

ps decay at 267 nm instead. Even though we do not see a

justiﬁcation for omitting the 4.1 ps related fraction, we

present here the equation that gives a perfect ﬁt to the data

͑see Fig. 7͒

sign to vibrational relaxation from states (␯ ,1,0) to states

␯₁,0,0), which also accounts for the increase of stimulated

emission after 0.7 ps. The dynamics described by the ﬁt con-

sists of: ͑1͒ dissociation from (␯ ,␯ ,2) states ͑0.3 ps com-

1

(

1

2

ponent͒; ͑2͒ a 5.1 ps ground state absorption recovery time

8,59

͑

obtained by Philpott et al.⁵

from ﬁtting over a 60 ps

range͒; ͑3͒ a 1.5 ps decay time of (␯ ,1,0) states ͑responsible

for quantum beat decay and a rising stimulated emission͒.

1

ϱ

Ϫt/1.5

S͑␶͒ϭ

͵

dt

͕

͓Ϫ0.25•e

•H͑t͒ϩ0.75

Ϫϱ

The 2.0 ps decay time of states (␯ ,0,0) populated by relax-

1

ation from (␯ ,1,0) should not be taken too literally. How-

Ϫ͑tϪ0.55͒/0.08

Ϫ͑tϪ0.55͒/1.2

1

•

͑0.08•e

ϩ1.50•e

ever, a (␯ ,1,0) to (␯ ,0,0) relaxation faster than 2 ps could

1

Ϫ͑tϪ0.55͒/0.7

not cause a second minimum at ϳ1.5 ps, and a decay of

␯₁,0,0) states slower than ϳ4 ps is inconsistent with a 5.1

Ϫ1.58•e

͒•H͑tϪ0.55͔͒

(

•

A͑tϪ␶͒

͖

ϩ10•A͑␶͒.

͑15͒

ps ground state recovery. A single exponential ﬁt between 2

and 6 ps time delay gives a 3.8 ps decay time, indicating a

faster than 5.1 ps decay during the ﬁrst 6 ps. Therefore, it is

As for acetonitrile we propose that the stimulated emission

originates from population on the OClO A A excited state

2

justiﬁed to conclude that from the (␯ ,0,0) state OClO either

PES. This signal is very likely connected to the population

fraction that is responsible for the quantum beat in the 403

nm/403 nm data, that also decays with a 1.5 ps time constant.

1

dissociates followed by recombination or directly relaxes to

the OClO ground state with a 2–4 ps decay rate. The disso-

ciation option is appealing since it also could explain the

biexponentiality of the 267 nm decay.

With regard to the second contribution, we follow the inter-

pretation by Keiding and co-workers⁵

4,56

that this absorption

Thøgersen et al.⁵⁵found a ground state recovery time of

2

is related to a transient vibrationally hot species on the X B₁

surface, that apparently transits the 800 nm window with an

average decay time of 0.7 ps ͓see Eq. ͑15͔͒.

10 ps using 390 nm for both pump and probe. This suggests

that the actual numbers are very sensitive to the excitation

wavelength. Here, the signal does not reach maximum before

2

ps. This rising behavior is not addressed by them;⁵⁵how-

ever, this feature might be a reason for their conclusion of

instantaneous dissociation followed by recombination within

G. Comparison for the different solvents and

conclusions

0

.8 ps.

In Tables I, II, and III we summarize some of the param-

eters that we extracted from the data analysis in Secs. III D–

III F. The error margins on the wavelength of the laser pulse

and the redshift of the absorption spectra can only noticeably

inﬂuence the relative population of the three fractions indi-

cated in Table I for the cyclohexane data. This is a conse-

3

. Pump 400 nm—probe 800 nm

The data at 800 nm are shown in the lower panel of Fig.

. We have ﬁtted the data using a stimulated emission con-

7

tribution, that decays with 1.5 ps, and an increased absorp-

tion contribution. As with acetonitrile, only the weight factor

is a ﬁt parameter for the increased absorption contribution.

Unlike for the acetonitrile data, the ﬁt quality is far from

perfect. An explanation for this is that the 267 nm data de-

scribe the recombination time of ClOϩO to OClO, which is

not the same as the appearance time in the ground state vi-

brational levels where absorption to the excited state can

occur with 800 nm light. A perfect ﬁt of the changeover from

stimulated emission to increased absorption is obtained if the

increased absorption contribution is delayed by 0.55 ps.

Given the above argument, such a delay could be justiﬁed.

For acetonitrile the introduction of such a delay was prob-

TABLE II. Parameters concerning the recombination yield of those mol-

ecules that dissociated to ClOϩO.

Fraction of recombination of

initially dissociated molecules

1

st

2nd

fraction

Fraction of

cage escape

Solvent

Total

fraction

Cyclohexane

Acetonitrile

Water

у57%

у22%

у83%

57%

0%

51%

¯

43%

78%

17%

22%

32%

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:

130.88.90.110 On: Sat, 20 Dec 2014 03:21:15

J. Chem. Phys., Vol. 114, No. 15, 15 April 2001

Photochemical reaction of OClO in solution

6793

TABLE III. Values for the decay parameters in the different solvents, corresponding to the dynamical steps a–e

presented in Fig. 5. The time constant ␶_gdenotes a secondary dissociation process mentioned in the text. The

ground state recovery time is indicated by ␶_r_e_c_o_v_e_r

.

␶_c

␶_d

Solvent

␶_a

␶_b

(␭ϭ800 nm)

(ϭ␶_Q_B

)

␶_e

␶_g

␶_r_e_c_o_v_e_r

Cyclohexane 140Ϯ30 fs

8.9Ϯ0.9 ps

?

1.3Ϯ0.15 ps ?͑Ͼ10 ps͒

?

Acetonitrile 0 fs

6.0Ϯ1.5 ps 7.2 ps

͑Ref. 58͒

1.5Ϯ0.2 ps

?͑Ͼ10 ps͒ 2.1Ϯ0.6 ps 36.9 ps

͑Ref. 58͒

͑р60 fs͒

Water

80Ϯ15 fs a͒ 1.2Ϯ0.2 ps 0.7Ϯ0.3 ps 1.5Ϯ0.2 ps

b͒ 4.1Ϯ0.5 ps ͑р1.4 ps͒

‘‘2.0 ps’’

͑2–4 ps͒

5.1 ps

͑Ref. 58͒

quence of the much smaller spectral broadening of the cy-

clohexane absorption spectrum by solvent–solute

interactions.

The observation of a quantum beat in the 403 nm data

for all solvents proves beyond any doubt that not all OClO

molecules dissociate within the ﬁrst picosecond after excita-

tion. The 1.3–1.5 ps decay we ascribe to the relaxation from

time ͓51% in Eq. ͑12͔͒ and the offset ͓17% in Eq. ͑12͔͒ are

related to ClO formed from the (␯ ,␯ ,2) states ͑i.e., 68% of

the 267 nm signal corresponds to 35% of the initial popula-

tion; see Table I͒, than the fraction with the 4.1 ps recombi-

nation time ͓32% in Eq. ͑12͔͒ should correspond to about

1

2

1

6% (35•32/68) of the initial excited state population. Com-

bining these numbers indicates that in water 49% (100Ϫ35

Ϫ16) of the molecules have not dissociated.

(

␯ ,1,0)→(␯ ,0,0). This relaxation process explains at the

1

same time the rise of the stimulated emission in cyclohexane

and water, that for cyclohexane can be shown to occur with

the same time constant as the decay of the quantum beat. For

water the ﬁt is too complicated to allow us to draw such a

strong conclusion.

For the 800 nm signal we argue that twice as many mol-

ecules contribute to the increased absorption contribution of

the 800 nm signal in water than in acetonitrile, even though

this signal appears to be stronger in acetonitrile. Based on the

numbers above, in water 42% of the initially excited mol-

ecules dissociate and recombine ͑26% from the 1.2 ps re-

combination, and 16% from the 4.1 ps recombination͒. In

acetonitrile this number is only 10%, so there is a discrep-

ancy of a factor of 2! This line of argumentation does not

include contributions from radiationless relaxation through

The 267 nm data demonstrate that a fraction of the mol-

ecules dissociates very fast. We ascribe the almost instanta-

neously rising signal to molecules dissociating from states

(

␯ ,␯ ,2). This implies that in all cases about one third of

1 2

the excited state molecules ͑see Table I͒ dissociates within a

few hundred femtoseconds. In this respect our interpretation

deviates from the conclusions drawn in previous studies of

OClO photodissociation in solution, where all molecules are

the dark states to the ground state. The data of Phillpott

58,59

et al.

on the ground state absorption recovery at 400 nm

supposed to dissociate within much less than

1

show a residual offset ͑after 150 ps͒ of about 11% of the

amplitude at time zero for water and 67% for acetonitrile.

Depending on whether one assumes that the 67% offset con-

sists of only bleach or both bleach and stimulated emission,

this offset implies that after 150 ps 33% or 50% of the mol-

ecules have recovered to the ground state in acetonitrile. For

water, estimates of the ground state recovery are in the range

of 80%–90%. These numbers indicate that 2–3 times more

OClO molecules recover in water than in acetonitrile, in ac-

cordance with the previously mentioned factor of 2. The in-

creased absorption at 800 nm in both water and acetonitrile

is well described using decay time parameters extracted from

the 267 nm data.

5

3–56,58,59

picosecond.

In Table III we show that this direct

dissociation step occurs with a time constant (␶ ) of 140, 80,

a

and р 60 fs, for cyclohexane, water, and acetonitrile, respec-

tively. In the case of cyclohexane and water an anisotropy

decay time of the polarization was extracted of 0.27 and 0.17

ps, respectively. This conﬁrms that at 267 nm we deal with

an increased absorption contribution of a reaction product,

ClO, that acquires rotational momentum during formation.

In acetonitrile we identify a second slower dissociation

fraction, with a 2.1 ps (␶ in Table III͒ formation time. As-

g

suming that the 78% of dissociated OClO ͓fast fraction; see

Eq. ͑9͔͒ that does not recombine in acetonitrile is related to

the 35% population in states (␯ ,␯ ,2) ͑see Table I͒, then

1

2

this second slower fraction corresponds to 10% of the origi-

nal excited state population. This fraction is either formed in

competition with the relaxation from (␯ ,1,0)→(␯ ,0,0), or

A second proof that not all OClO molecules dissociate

within the ﬁrst picosecond was found in the 800 nm data for

water and acetonitrile. In both solvents these data clearly

contain a stimulated emission contribution that decays with

1

2

by dissociation from the dark 1 B or 1 A states. These

2

1

numbers suggest that in acetonitrile 55% (100Ϫ35Ϫ10) of

the molecules do not dissociate at all. We have not found

evidence for a secondary dissociation process in cyclohex-

ane. Therefore, if only (␯ ,␯ ,2) states dissociate in cyclo-

1

.5 ps. This signal we assigned to the same optical transition

that leads to the increased absorption at 800 nm.

In conclusion: We have found clear indications that the

fate of the excited electronic state of OClO in solution de-

pends on which vibronic state is created. All in all, the reac-

tion dynamics scheme of OClO in solution that emerges

from our data is far more complex than presented so far.

1

2

hexane, 68% of the OClO molecules should not dissociate.

In water a second channel probably also exists in view of the

clearly biexponential recombination decay ͑step b͒ found in

the 267 nm data. If the fraction with the 1.2 ps recombination

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:

130.88.90.110 On: Sat, 20 Dec 2014 03:21:15

6794

J. Chem. Phys., Vol. 114, No. 15, 15 April 2001

Fidder et al.

3

6

7

ACKNOWLEDGMENTS

S. Hubinger and J. B. Nee, Chem. Phys. 181, 247 ͑1994͒.

S. Michielsen, A. J. Merer, S. A. Rice, F. A. Novak, K. F. Freed, and Y.

Hamada, J. Chem. Phys. 74, 3089 ͑1981͒.

P. A. McDonald and K. K. Innes, Chem. Phys. Lett. 59, 562 ͑1978͒.

E. C. Richard and V. Vaida, J. Chem. Phys. 94, 163 ͑1991͒.

E. Bishenden and D. J. Donaldson, J. Chem. Phys. 99, 3129 ͑1993͒.

E. Bishenden and D. J. Donaldson, J. Chem. Phys. 101, 9565 ͑1994͒.

H. F. Davis and Y. T. Lee, J. Chem. Phys. 105, 8142 ͑1996͒.

H. F. Davis and Y. T. Lee, J. Phys. Chem. 96, 5681 ͑1992͒.

R. F. Delmdahl, S. Baumg a¨ rtel, and K.-H. Gericke, J. Chem. Phys. 104,

H.F. gratefully acknowledges ﬁnancial support by the

Swedish Natural Science Research Council. This work has

beneﬁtted from ﬁnancial support through the EU-TMR Pro-

gram ͑Contract Number MBI-97-005-01͒.

3

4

8

9

0

1

2

1

43

44

V. Vaida and J. D. Simon, Science 268, 1443 ͑1995͒.

V. Vaida, S. Solomon, E. C. Richard, E. Ruhl, and A. Jefferson, Nature

2

͑

London͒ 342, 405 ͑1989͒.

2883 ͑1996͒.

3

4

45

46

M. J. Rosker, M. Dantus, and A. H. Zewail, Science 241, 1200 ͑1988͒.

M. J. Rosker, T. S. Rose, and A. H. Zewail, Chem. Phys. Lett. 146, 175

͑

M. Roth, C. Maul, and K.-H. Gericke, J. Chem. Phys. 107, 10582 ͑1997͒.

A. Furlan, H. A. Scheld, and J. R. Huber, J. Chem. Phys. 106, 6538

1988͒.

͑

1997͒.

5

6

7

8

R. M. Bowman, M. Dantus, and A. H. Zewail, Chem. Phys. Lett. 161, 297

1989͒.

M. Dantus, R. M. Bowman, and A. H. Zewail, Nature ͑London͒ 343, 737

1990͒.

A. L. Harris, J. K. Brown, and C. B. Harris, Annu. Rev. Phys. Chem. 39,

41 ͑1988͒.

R. Zadoyan, Z. Li, C. C. Martens, and V. A. Apkarian, J. Chem. Phys.

01, 6648 ͑1994͒.

47

48

49

50

C. J. Kreher, R. T. Carter, and J. R. Huber, Chem. Phys. Lett. 286, 389

1998͒.

T. Baumert, J. L. Herek, and A. H. Zewail, J. Chem. Phys. 99, 4430

1993͒.

P. Ludowise, M. Blackwell, and Y. Chen, Chem. Phys. Lett. 273, 211

1997͒.

R. C. Dunn, E. C. Richard, V. Vaida, and J. D. Simon, J. Phys. Chem. 95,

060 ͑1991͒.

͑

3

͑

1

6

9

0

C. Lienau and A. H. Zewail, J. Phys. Chem. 100, 18629 ͑1996͒.

C. J. Bardeen, J. Che, K. R. Wilson, V. V. Yakovlev, V. A. Apkarian, C.

C. Martens, R. Zadoyan, B. Kohler, and M. Messina, J. Chem. Phys. 106,

51

52

R. C. Dunn and J. D. Simon, J. Am. Chem. Soc. 114, 4856 ͑1992͒.

R. C. Dunn, B. N. Flanders, and J. D. Simon, J. Phys. Chem. 99, 7360

1

͑

1995͒.

8

486 ͑1997͒.

53

54

Y. J. Chang and J. D. Simon, J. Phys. Chem. 100, 6406 ͑1996͒.

J. Tho”gersen, C. L. Thomsen, J. Aa. Poulsen, and S. R. Keiding, J. Phys.

Chem. A 102, 4186 ͑1998͒.

J. Tho”gersen, P. U. Jepsen, C. L. Thomsen, J. Aa. Poulsen, J. R. Byberg,

and S. R. Keiding, J. Phys. Chem. A 101, 3317 ͑1997͒.

J. A. Poulsen, C. L. Thomsen, S. R. Keiding, and J. Tho”gersen, J. Chem.

Phys. 108, 8461 ͑1998͒.

C. L. Thomsen, M. P. Philpott, S. C. Hayes, and P. J. Reid, J. Chem. Phys.

112, 505 ͑2000͒.

M. J. Philpott, S. Charalambous, and P. J. Reid, Chem. Phys. Lett. 281, 1

͑1997͒.

M. J. Philpott, S. C. Hayes, and P. J. Reid, Chem. Phys. 236, 207 ͑1998͒.

A. P. Esposito, C. E. Foster, R. A. Beckman, and P. J. Reid, J. Phys.

Chem. A 101, 5309 ͑1997͒.

P. J. Reid, A. P. Esposito, C. E. Foster, and R. A. Beckman, J. Chem.

Phys. 107, 8262 ͑1997͒.

S. C. Hayes, M. P. Philpott, S. G. Mayer, and P. J. Reid, J. Phys. Chem.

A 103, 5534 ͑1999͒.

C. E. Foster and P. J. Reid, J. Phys. Chem. A 102, 3514 ͑1998͒.

S. C. Hayes, M. J. Philpott, and P. J. Reid, J. Chem. Phys. 109, 2596

1

2

3

4

5

M. J. J. Vrakking, D. M. Villeneuve, and A. Stolow, J. Chem. Phys. 110,

474 ͑1996͒.

L. Poth, Q. Zhong, J. V. Ford, and A. W. Castleman, Jr., J. Chem. Phys.

09, 4791 ͑1998͒.

2

55

1

J. M. Papanikolas, V. Vorsa, M. E. Nadal, P. J. Campagnola, J. R. Gord,

and W. C. Lineberger, J. Chem. Phys. 97, 7002 ͑1992͒.

P. K. Walhout, J. C. Alfano, K. A. M. Thakur, and P. F. Barbara, J. Phys.

Chem. 99, 7568 ͑1995͒.

5

6

7

8

E. Gershgoren, E. Gordon, and S. Ruhman, J. Chem. Phys. 106, 4806

͑1997͒.

1

6

7

8

U. Banin, A. Waldman, and S. Ruhman, J. Chem. Phys. 96, 2416 ͑1992͒.

U. Banin and S. Ruhman, J. Chem. Phys. 98, 4391 ͑1993͒.

G. Ashkenazi, U. Banin, A. Bartana, R. Kosloff, and S. Ruhman, Adv.

Chem. Phys. 100, 229 ͑1997͒.

T. K u¨ hne, R. K u¨ ster, and P. V o¨ hringer, J. Chem. Phys. 105, 10788 ͑1996͒.

T. K u¨ hne, R. K u¨ ster, and P. V o¨ hringer, J. Phys. Chem. A 102, 4177

5

6

9

0

6

1

2

1

2

9

0

͑

1998͒.

2

1

2

3

R. M. Bowman, M. Dantus, and A. H. Zewail, Chem. Phys. Lett. 156, 131

1989͒.

6

3

4

͑

N. Pugliano, D. K. Palit, A. Z. Szarka, and R. M. Hochstrasser, J. Chem.

Phys. 99, 7273 ͑1993͒.

U. Marvet, Q. Zhang, E. J. Brown, and M. Dantus, J. Chem. Phys. 109,

͑

1998͒.

6

5

6

O. D u¨ hr, E. T. J. Nibbering, and G. Korn, Appl. Phys. B: Lasers Opt. 67,

525 ͑1998͒.

4

415 ͑1998͒.

2

4

5

6

F. Tschirschwitz and E. T. J. Nibbering, Chem. Phys. Lett. 312, 169

͑1999͒.

W. Bray, Z. Phys. Chem., Stoechiom. Verwandtschaftsl. 54, 569 ͑1906͒.

¨

U. K. Klaning and T. Wolff, Ber. Bunsenges. Phys. Chem. 89, 243 ͑1985͒.

J. C. Owrutsky, D. Raftery, and R. M. Hochstrasser, Annu. Rev. Phys.

Chem. 45, 519 ͑1994͒.

E. J. Heilweil, F. E. Doany, R. Moore, and R. M. Hochstrasser, J. Chem.

Phys. 76, 5632 ͑1982͒.

P. Hamm, M. Lim, and R. M. Hochstrasser, J. Chem. Phys. 107, 10523

Q. Zhang, U. Marvet, and M. Dantus, J. Chem. Phys. 109, 4428 ͑1998͒.

A. Arkell and I. Schwager, J. Am. Chem. Soc. 89, 5999 ͑1967͒.

K. Johnsson, A. Engdahl, P. Ouis, and B. Nelander, J. Phys. Chem. 96,

67

68

5

778 ͑1992͒.

2

7

8

9

69

70

71

72

73

74

H. S. P. M u¨ ller and H. Willner, J. Phys. Chem. 97, 10589 ͑1993͒.

L.-H. Lai, C.-P. Liu, and Y.-P. Lee, J. Chem. Phys. 109, 988 ͑1998͒.

C. J. Pursell, J. Conyers, P. Alapat, and R. Parveen, J. Phys. Chem. 99,

1

0433 ͑1995͒.

3

0

1

C. J. Pursell, J. Conyers, and C. Denison, J. Phys. Chem. 100, 15450

1996͒.

͑

͑1997͒.

C.-P. Liu, L.-H. Lai, Y.-Y. Lee, S.-C. Hung, and Y.-P. Lee, J. Chem.

Phys. 109, 978 ͑1998͒.

K. A. Peterson and H.-J. Werner, J. Chem. Phys. 96, 8948 ͑1992͒.

K. A. Peterson and H.-J. Werner, J. Chem. Phys. 105, 9823 ͑1996͒.

K. A. Peterson, J. Chem. Phys. 109, 8864 ͑1998͒.

M. Lim, S. Gnanakaran, and R. M. Hochstrasser, J. Chem. Phys. 106,

3485 ͑1997͒.

P. Hamm, M. Lim, and R. M. Hochstrasser, Phys. Rev. Lett. 81, 5326

͑1998͒.

CRC Handbook of Chemistry and Physics, 63rd edition ͑CRC, Boca Ra-

ton, 1982͒.

3

2

3

4

5

E. C. Richard and V. Vaida, J. Chem. Phys. 94, 153 ͑1991͒.

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:

30.88.90.110 On: Sat, 20 Dec 2014 03:21:15

1

Article Doi

Vibrational mode-specific photochemical reaction dynamics of chlorine dioxide in solution

DOI: 10.1063/1.1357202

Source and publish data:

Authors:

Article abstract of DOI:10.1063/1.1357202

Full text of DOI:10.1063/1.1357202

Products guided by the article

R&D Labs maybe for 10049-04-4

Relevant to this article

Hot Product