Femtosecond ultraviolet-visible absorption study of all-trans→13-cis9-cis photoisomerization of retinal

Femtosecond ultraviolet-visible absorption study of all - trans 13- cis 9- cis

photoisomerization of retinal

Shoichi Yamaguchi and Hiro-o Hamaguchi

Citation: The Journal of Chemical Physics 109, 1397 (1998); doi: 10.1063/1.476692

View online: http://dx.doi.org/10.1063/1.476692

View Table of Contents: http://scitation.aip.org/content/aip/journal/jcp/109/4?ver=pdfcov

Published by the AIP Publishing

Articles you may be interested in

Matrix isolation and computational study of isodifluorodibromomethane ( F 2 CBr – Br ) : A route to Br 2

formation in CF 2 Br 2 photolysis

J. Chem. Phys. 132, 084503 (2010); 10.1063/1.3319567

Energetics of the all trans13cis isomerization of the retinal chromophore of bacteriorhodopsin: Electronic

structure calculations for a simple model system

AIP Conf. Proc. 1080, 157 (2008); 10.1063/1.3058976

Red luminescence of Cr in Ga 2 O 3 nanowires

J. Appl. Phys. 101, 033517 (2007); 10.1063/1.2434834

IR laser manipulation of cis trans isomerization of 2-naphthol and its hydrogen-bonded clusters

J. Chem. Phys. 124, 054315 (2006); 10.1063/1.2162164

Visible and ultraviolet Raman scattering studies of Si 1x Ge x alloys

J. Appl. Phys. 88, 2523 (2000); 10.1063/1.1287757

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:

130.39.62.90 On: Fri, 29 Aug 2014 14:13:24

JOURNAL OF CHEMICAL PHYSICS

VOLUME 109, NUMBER 4

22 JULY 1998

Femtosecond ultraviolet-visible absorption study of all-trans˜13-cis–9-cis

photoisomerization of retinal

Shoichi Yamaguchi^a)

Department of Basic Science, Graduate School of Arts and Sciences, The University of Tokyo, 3-8-1

Komaba, Meguro, Tokyo 153, Japan

Hiro-o Hamaguchi^b)

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

Japan

͑Received 31 December 1997; accepted 14 April 1998͒

The all-trans→13-cis•9-cis photoisomerization reaction of retinal in aerated nonpolar solvents

has been studied by femtosecond time-resolved ultraviolet-visible ͑UV-VIS͒ absorption

spectroscopy. The excited-state absorption spectra in the wavelength region 400–800 nm indicate

that there is no all-trans→13-cis•9-cis isomerization reaction pathway that is complete in the

electronic excited singlet manifold of S₁, S₂, and S₃. The ground-state bleaching recovery of

all-trans retinal monitored in the near UV ͑ultraviolet͒ wavelength region 310–390 nm shows that

*

a perpendicular excited singlet state (p ) takes part in the all-trans→13-cis•9-cis isomerization

*

reaction. The lifetime of p is about 7 ps, and the precursor of p is most probably the S₂state. The

isomerization quantum yield derived from the femtosecond UV absorption data agrees well with

those determined by the HPLC analysis of the photoproduct. The temperature dependence of the

isomerization quantum yield indicates the existence of a potential-energy barrier as high as (1.2

*

Institute of Physics. ͓S0021-9606͑98͒00728-4͔

respectively.^5,6Ganapathy and Liu investigated the relation

I. INTRODUCTION

between the photoisomerization quantum yield and the con-

centration of the all-trans isomer by using high performance

liquid chromatography ͑HPLC͒.⁶They found no concentra-

tion dependence of the all-trans→13-cis•9-cis isomeriza-

tion quantum yield in aerated solutions upon direct photoex-

citation. This result indicates that the reaction is

unimolecular. Both the excited singlet and triplet states can

take part in the photoisomerization. According to our nano-

second time-resolved Raman studies, the photoexcitation of

the all-trans, 7-cis, 9-cis, and 11-cis isomers in aerated hex-

ane results in an identical Raman spectrum at 20 ns after

photoexcitation.⁷This Raman spectrum has been

ascribed to the T₁all-trans isomer. Though the

7-cis(or 9-cis,11-cis)→all-trans isomerization proceeds

via the lowest excited triplet (T₁) manifold, the all-trans

isomer did not seem to isomerize in the T₁state ͑‘‘one-way’’

isomerization͒. This conclusion was supported by subsequent

picosecond time-resolved absorption⁸and picosecond

2D-CARS⁹studies. We have recently proved by submicro-

second time-resolved IR ͑infrared͒ spectroscopy that there is

no isomerization reaction taking place in the T₁state when

the all-trans isomer is the starting material.¹⁰According to

our result, T₁all-trans retinal is generated upon direct pho-

toexcitation within the time resolution of the experiment ͑50

ns͒ and decays exclusively to the all-trans isomer in the

ground state ͑S₀all-trans͒ with the time constant of 5 ␮s.

The S₀13-cis and S₀9-cis isomers are also generated within

the time resolution and do not decay at all. It is concluded

that the all-trans→13-cis•9-cis isomerization does not

Retinal, vitamin A aldehyde and the chromophore of

light-sensitive proteins known as rhodopsins, is one of the

well studied prototypical molecules exhibiting cis-trans pho-

toisomerization. The spectroscopic and photochemical prop-

erties of retinal strongly depend on its surrounding environ-

ment. Though organic solvents appear to provide simpler

environments than proteins, the excited electronic structure

of retinal in organic solvents seems to be more complicated

than that in rhodopsins. Thus, the cis-trans photoisomeriza-

tion reaction mechanism of retinal in organic solvents¹is

much less understood than that in proteins.^2,3It is of consid-

erable interest to look into the relationship between the

excited-state electronic structure and the isomerization

pathway/yield of retinal in organic solvents.

Retinal has four CvC double bonds in its polyenic

backbone; there are sixteen possible cis-trans isomers asso-

ciated with it including the all-trans and four mono-cis iso-

mers. It is known that all of these isomers in solution un-

dergo cis-trans isomerization upon UV irradiation.⁴In this

paper, we focus on the all-trans→13-cis•9-cis isomeriza-

tion in aerated nonpolar solvents. The quantum yield of the

all-trans→13-cis and the all-trans→9-cis isomerization

upon direct photoexcitation are ϳ0.1 and ϳ0.02,

a͒

Present address: Analytical Sciences Laboratory, Yokohama Research

Center, Mitsubishi Chemical Corporation, 1000 Kamoshida, Aoba, Yoko-

hama 227, Japan.

b͒

Author to whom correspondence should be addressed. Electronic mail:

hhama@chem.s.u.-tokyo.ac.jp

0021-9606/98/109(4)/1397/12/$15.00

1397

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:

130.39.62.90 On: Fri, 29 Aug 2014 14:13:24

1398

J. Chem. Phys., Vol. 109, No. 4, 22 July 1998

S. Yamaguchi and H. Hamaguchi

take place in the T₁state; in other words, it proceeds via the

excited singlet manifold.

͑2.5%͒. The same quantity of the 13-cis retinal sample was

as follows: 13-cis, 99.2% ͑97.9%͒; all-trans, 0.5% ͑1.7%͒;

9-cis, 0.0% ͑0.1%͒; 7-cis, 0.3% ͑0.3%͒.

Three excited singlet states are known to exist in the

low-energy region of all-trans retinal up to

The femtosecond time-resolved pump–probe UV-VIS

absorption spectra of the retinal isomers were measured in

the wavelength range 310–390 nm and 400–800 nm by us-

ing the femtosecond laser system described elsewhere.^28,29

Hexane or cyclohexane solution ͑2ϫ10^Ϫ3mol dm^Ϫ3, vol-

ume 0.2 dm^Ϫ3͒ was ejected from a jet nozzle to make a thin

ﬂat ﬁlm suitable for the transient absorption measurements.

The solution was photoexcited by the pump pulse at 400 nm

with excitation energy density of 2ϫ10 J m^Ϫ2. The VIS

wideband probe was generated by focusing intense 800 nm

femtosecond pulses into water. The UV wideband probe was

obtained by replacing the 800 nm pulses with the 400 nm.

The time resolution was ϳ0.3 ps both in the VIS and UV

wavelength region. A set of time-resolved spectra consisted

of 70 time-delay points from Ϫ4 to 120 ps. The exposure

time of the CCD ͑charge coupled device͒ was 1 s, and the

number of the exposures at each time delay point was ﬁve. A

chirp correction was performed by using the optical Kerr

effect ͑OKE͒ cross-correlation method²⁹to make the ﬁnal

time-resolved spectra. The VIS absorption measurements

were done at 293Ϯ2 K. As will be described later, the

all-trans→13-cis•9-cis isomerization total quantum yield

is obtained from the femtosecond UV absorption data. In

order to investigate the temperature dependence of the quan-

tum yield, the solution temperature in the UV absorption

measurements was set at four different points: 269, 273, 293,

and 298Ϯ2 K.

11–21

30 000 cm^Ϫ1

.

Though the state ordering is still unclear,

Ϫ

there are surely a B^ϩ_u(␲,␲ ) state, an A_g(␲,␲ ) state, and

*

an (n,␲ ) state in the low-lying excited singlet manifold.

Polarized UV-VIS absorption spectroscopic study of crystal-

line all-trans retinal by Drikos et al.^12–14and two-photon

ﬂuorescence excitation spectroscopy by Birge et al.^16,17sug-

gested that the lowest excited singlet (S₁) state was of

A^Ϫ_g(␲,␲ ) character. On the other hand, ﬂuorescence spec-

*

troscopy of hydrogen-bonded retinal by Takemura et al. pos-

*

tulated that an (n,␲ ) state was the lowest excited singlet

state.¹¹The kinetics of the excited singlet states have been

studied by picosecond time-resolved absorption and ﬂuores-

cence spectroscopies.^22–24Our recent femtosecond time-

resolved VIS absorption measurement has revealed the pres-

ence of two excited singlet states ͑S₃, S₂͒ that precede S₁.²⁵

Larson et al. suggested the assignments of the S₃, S₂, and

Ϫ

S₁states to B^ϩ_u(␲,␲ ), A_g(␲,␲ ), and (n,␲ ), respec-

tively, based on their femtosecond absorption kinetics

study.²⁶Very recently Takeuchi and Tahara performed fem-

tosecond time-resolved ﬂuorescence up-conversion spectros-

copy and found that the transient species of the shortest life-

time emitted the strongest ﬂuorescence. They concluded that

*

the S₃state of all-trans retinal was of B^ϩ_u(␲,␲ )

*

character.²⁷

The purpose of this paper is to clarify the all-trans

→13-cis•9-cis isomerization mechanism of retinal in aer-

ated nonpolar solvents. The femtosecond time-resolved VIS

absorption spectra of all-trans, 9-cis, and 13-cis retinal in

hexane have been measured. No cis singlet spectra were ob-

served when the all-trans isomer was photoexcited. This

ﬁnding motivated us to perform femtosecond time-resolved

absorption spectroscopy in the UV region which can monitor

the ground-state recovery of the starting all-trans isomer.

Two time constants were found in the bleaching recovery of

which one was coincident with the T₁lifetime. The other

time constant was longer than the S₂lifetime but shorter than

that of the S₁state. This time constant of the bleaching re-

covery is most likely to correspond to the lifetime of a per-

pendicular excited singlet state. Detailed discussion on the

all-trans→13-cis•9-cis isomerization reaction of retinal is

presented on the basis of these newly obtained experimental

data.

III. RESULTS

A. Femtosecond time-resolved VIS absorption

spectra

In Fig. 1, fourteen representative femtosecond time-

resolved VIS absorption spectra of all-trans retinal in aerated

hexane are shown. The spectra are shown in the forms of

difference spectra. The positive signals correspond to the

photoinduced increase of absorbance which is caused by

transitions from excited states to higher excited states. The

negative signals correspond to the photoinduced decrease of

absorbance which is caused by ground-state bleaching and/or

stimulated emission gain. Stimulated Raman loss ͑inverse

Raman͒ may also contribute to the positive signals and

stimulated Raman gain to the negative signals.

There are several salient features in the time-resolved

spectra in Fig. 1. First of all, the absorption band around 450

nm at 100 ps after the photoexcitation is very distinct and

undoubtedly assigned to the T_n←T₁absorption.^8,25,30,31As

mentioned in Sec. I, the T₁species has exclusively the all-

trans conﬁguration when the S₀all-trans isomer is the start-

ing material.^7–10,31The absorption band around 450 nm in

Fig. 1 is ascribed to the T₁all-trans isomer. The band shape

and the peak position of this band agree well with those in

the literature.^8,30,32,33Broad absorption bands in the subpico-

second time range are other features to be noted in the spec-

tra in Fig. 1. These bands of all-trans retinal were observed

ﬁrst by Hochstrasser et al. in 1976,²²and there was no fur-

II. EXPERIMENT

Hexane and cyclohexane ͑HPLC grade͒ were purchased

from Kanto Chemical Co. or Wako Chemical Co. and used

as received. All the solvents were aerated. Commercial reti-

nal ͑Sigma Chemical Co.͒ was used without further puriﬁca-

tion. The retinal samples were analyzed for isomeric compo-

sition before and after the time-resolved spectroscopic

measurements by using HPLC. The isomeric composition of

the all-trans retinal sample before ͑and after͒ the measure-

ments was as follows: all-trans, 98.3% ͑97.5%͒; 13-cis, 1.7%

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:

130.39.62.90 On: Fri, 29 Aug 2014 14:13:24

J. Chem. Phys., Vol. 109, No. 4, 22 July 1998

S. Yamaguchi and H. Hamaguchi

1399

FIG. 2. Singular values obtained from the SVD analysis of the time-

resolved VIS spectra of all-trans retinal in hexane.

temporal evolution of the ith SVD component. All the sin-

gular values are positive, and V ϾV if iϽj. s and t are

͕ ͖ ͕ ͖

i

j

i

t

both normalized orthogonal sets; s_is_jϭ␦_ijand t_it_jϭ␦_ij. In

Eq. ͑1͒, the contribution of the ith SVD component to A is

determined by V_i. SVD components with smaller singular

values are less signiﬁcant. We have only to take into account

the SVD components having large singular values. The

omission of the SVD components with small singular values

is effective in simplifying the analysis and removing noise.

This is one of the advantages of the SVD analysis.

FIG. 1. Femtosecond time-resolved VIS absorption spectra of all-trans reti-

nal in hexane at various time delays.

The SVD of the set of the time-resolved spectra in Fig. 1

yielded four principal singular values as shown in Fig. 2. The

ﬁfth and the higher singular values were almost zero and

regarded as noise. In order to obtain physically meaningful

temporal evolutions and the corresponding spectra from the

SVD components, we start with the following kinetics

scheme for the photophysics of all-trans retinal:

ther report until our letter was published in 1996.²⁵It is

likely that the excited singlet states contribute to these broad

bands, but no deﬁnitive assignments have been performed. A

detailed kinetics analysis is necessary to obtain more insight

into the singlet electronic structure of retinal from the time-

resolved spectra in Fig. 1.

When two or more transient species are involved in one

set of time-resolved spectra just like the present case, the

singular value decomposition ͑SVD͒ analysis combined with

ﬁtting procedures is often useful for its kinetics analysis. The

methods and the advantages of the SVD analysis have been

brieﬂy reviewed by Chen and Braiman³⁴and there have been

many examples in various ﬁelds which show the usefulness

of SVD.^{10,25,35–38}In SVD, a set of time-resolved spectra is

treated as a matrix. Let this matrix be written as A of which

the ith row vector corresponds to the spectrum at the ith time

delay point. The SVD analysis yields several SVD compo-

nents from A, and each SVD component has its singular

value ͑scalar͒, its spectrum ͑column vector͒, and its temporal

evolution ͑column vector͒. A is expressed as follows:

␾ ␥

1 1

3

2

S —— S —— S —— S —— T .

͑I͒

→

0

3

2

1

hv

Here it is assumed that only the S₃←S₀transition is one-

photon allowed. This assumption is based on the results of

femtosecond time-resolved ﬂuorescence up-conversion spec-

troscopy by Takeuchi and Tahara.²⁷According to their con-

clusion, the S₃state of all-trans retinal is of B^ϩ_u(␲,␲ )

*

character and the S₃←S₀transition contribute to about 96%

of the ground-state absorption. The contribution of the direct

S₂←S₀photoexcitation is small and disregarded in the

present analysis.³⁹The simplest relaxation schemes are as-

sumed among the three excited singlet states: The S₃state

relaxes to the S₂state with the quantum efﬁciency ␾ and

3

Aϭ

V t ^ts ,

͑1͒

the lifetime ␥^Ϫ₃¹, S₂to S₁with ␾ and ␥^Ϫ1, and S₁to T₁

͚

i i

i

2

iϭ1

with ␾ and ␥^Ϫ1. Since the maximum time delay after the

1

where V_iis the singular value of the ith SVD component, s_i

is the spectrum of the ith SVD component, and t_iis the

photoexcitation in the present experiments is about 100 ps, it

is unnecessary to take into account the relaxation of the T₁

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:

130.39.62.90 On: Fri, 29 Aug 2014 14:13:24

1400

J. Chem. Phys., Vol. 109, No. 4, 22 July 1998

S. Yamaguchi and H. Hamaguchi

state.⁹According to Kinetics Scheme ͑I͒, the time-delay de-

pendence of the population of each electronic state is ex-

pressed as follows:

t

V₁t₁

^tT_1pop

V₂t₂

^tS_1pop

t

ϭC

.

͑8͒

t

ͩ ͪ ͩ ͪ

V₃t₃

S_2pop

t

V₄t₄

^tS_3pop

S

ϭe^{Ϫ␥ t}

,

͑2͒

͑3͒

3

͓

͔

3

where C is a matrix whose (i, j) component is c_ij, S_3popis a

column vector whose ith component is the S₃population at

␾₃␥₃

3

ϭ

e^Ϫ␥²^tϪe^{Ϫ␥ t}͒,

͑

2

the ith time delay point, and so are S_2pop, S_1pop, T_1pop

.

␥₃Ϫ␥₂

The populations of the transient species are ﬁnally recom-

posed from t

͕ ͖

as follows:

i

1рiр4

␾ ␾₃␥₂␥₃

␥₃Ϫ␥₁

␥₃Ϫ␥₂

2

S

e^Ϫ␥¹^tϪ

e^{␥ t}

͓

͔

ͩ

1

^tT_1pop

^tS_1pop

V₁t₁

t

␥ Ϫ␥ ͒ ␥ Ϫ␥ ͒

͑

2

1

3

1

t

V₂t₂

ϭC^Ϫ1

.

͑9͒

␥₂Ϫ␥₁

␥₃Ϫ␥₂

e^{Ϫ␥ t}

,

͑4͒

ͩ

^tS_2popͪ ͩV₃t₃

ͪ

t

3

ϩ

ͪ

^tS_3pop

V₄t₄

t

␥₂␥₃e^{Ϫ␥ t}

1

The spectra of the transient species have to satisfy the fol-

lowing equation:

T

ϭ␾ ␾ ␾ 1Ϫ

͓

͔

ͭ

1

2

3

␥ Ϫ␥ ͒ ␥ Ϫ␥ ͒

͑

2

1

3

1

t

␥₁␥₃e^{Ϫ␥ t}

␥₁␥₂e^{Ϫ␥ t}

AϭS_3pop^tS_3speϩS_2pop^tS_2speϩS_1pop^tS_1speϩT_1popT_1spe

,

2

3

ϩ

Ϫ

, ͑5͒

ͮ

͑10͒

␥ Ϫ␥ ͒ ␥ Ϫ␥ ͒

͑

3 1 3 2

͑

2

1

3

2

where S_3speis a column vector and stands for the absorption

spectrum of the S₃state. The ith component of S_3specorre-

sponds to absorbance at the ith wavelength point, and the

same for S_2spe, S_1spe, and T_1spe. By using Eqs. ͑1͒, ͑9͒, and

͑10͒, the spectra of the transient species are recomposed from

where ␦-function-like photoexcitation at tϭ0 is assumed and

the S₃population directly generated by the photoexcitation is

set to be unity. All the populations are zero for tϽ0. There is

a relation among the populations of the ﬁve electronic states

involved in Kinetics Scheme ͑I͒:

s

͕ ͖

i

1рiр4

S

ϭϪ S Ϫ S Ϫ S Ϫ T ,

͔

1

͑6͒

^tT_1spe

^ts₁

͓

͔

͓

͔

͓

͔

͓

͔

͓

0

3

2

1

^tS_1spe

^ts₂

ϭ^tC

.

͑11͒

t

where the S population S is always negative. The term

͓

͔

0

ͩ ͪ ͩ ͪ

S_2spe

s₃

^tS_3spe

^ts₄

‘‘population’’ in this paper stands for the difference between

the population with and without photoexcitation. Though the

ﬁve electronic states are involved in Kinetics Scheme ͑I͒,

The SVD analysis of the 70 time-resolved spectra of

all-trans retinal in hexane gives the absorption spectra and

the population changes shown in Figs. 3͑a͒–3͑h͒. In the

analysis, ␾ was ﬁxed to 0.74, and ␾ and ␾ were put equal

four SVD components are enough because S is expressed

͓

͔

0

as a linear combination of the other four populations. The

number of SVD components that have to be taken into ac-

count does not always coincide with that of transient species,

since all the populations of transient species are not always

linearly independent just like the present case, while all the

temporal behaviors of SVD components have to be linearly

independent.

2

1

3

to unity. Note that the triplet quantum yield ͑which is equal

to ␾ ␾ ␾ ͒ of all-trans retinal in hexane at room tempera-

1

2

3

ture is going to be determined later in this paper to be 0.74.

It is also shown later that ␾ is equal to one. Nearly identical

1

spectra and populations were also obtained in the case of

all-trans retinal in cyclohexane. This means that hexane and

cyclohexane provide similar environments for the excited-

state dynamics of retinal. Time constants obtained from

twelve independent measurements and analyses are given in

Table I.

We now derive the populations and the spectra of the

transient species (S₃,S₂,S₁,T₁) from

t

i

and

of the SVD components: The product of the ith

1рiр4

͕ ͖

1рiр4

s

͕ ͖

i

singular value and the temporal evolution of the ith SVD

component, V_it_i, is ﬁtted to the following model function:

The spectrum in Fig. 3͑d͒ is identical with the known

T_n←T₁absorption spectrum of all-trans retinal in nonpolar

solvents.^8,25,30–33The T₁lifetime is much longer than 100 ps

and is ﬁxed to inﬁnity in the ﬁtting procedures. The spectrum

in Fig. 3͑c͒ corresponds to the S_n←S₁absorption spectrum,

though the band shape looks strange. It will be shown later

that the main band of the S_n←S₁absorption lies in the wave-

length range 350–400 nm, and only the band edge appears in

Fig. 3͑c͒. The S₁lifetime, 32Ϯ2 ps, agrees well with the

literature values.^8,22,24It also agrees with the T₁rise time,

indicating that the S₁state is the precursor of the T₁all-trans

isomer and that the S₁state is most likely to be in the all-

trans conﬁguration. The S₁rise time, which is equal to the

c_i4

S

ϩc

S

ϩc

S

ϩc

T ,

͔

1

͑7͒

͓

͔

͓

͔

͓

͔

͓

3

i3

2

i2

1

i1

of

which

the

ﬁtting

parameters

are

␥₁,␥₂,␥₃,c_i1,c_i2,c_i3,c_i4. The parameters ␥₁,␥₂,␥₃have

to be common to all the four SVD components. The quantum

efﬁciencies, ␾₁,␾₂,␾₃, are not determined by the ﬁtting

procedure. In the practical ﬁtting procedure, the model func-

tion of Eq. ͑7͒ is convoluted with a Gaussian function which

represents the instrumental response. The relation between

t

and the populations of the transient species is ex-

͕ ͖

i

1рiр4

pressed by using the converged ﬁtting parameters c_ijas fol-

lows:

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:

130.39.62.90 On: Fri, 29 Aug 2014 14:13:24

J. Chem. Phys., Vol. 109, No. 4, 22 July 1998

S. Yamaguchi and H. Hamaguchi

1401

FIG. 3. Spectra ͓͑a͒–͑d͔͒ and populations ͓͑e͒–͑h͔͒ of the four transient species obtained from the time-resolved VIS spectra of all-trans retinal in hexane.

clear idea why the S₂lifetime in the present study does not

agree with that in the ﬂuorescence up-conversion study²⁷

even if experimental errors are taken into account. The tran-

sient species exhibiting the spectrum in Fig. 3͑b͒ and the

population change in Fig. 3͑f͒ is the precursor of S₁all-trans

retinal and is naturally assigned to the S₂all-trans isomer.

The broad absorption bands in the subpicosecond time re-

gion in Fig. 1 is due to the S_n←S₂absorption spectrum. The

S₂rise time or the S₃lifetime is too short to be determined

by the present analysis. It is ﬁxed to the value, 0.03 ps,

reported by Takeuchi and Tahara.²⁷This means that we have

no clear evidence in the present study that the S₂state is

generated only from the S₃state, as assumed in Kinetics

Scheme ͑I͒. There remains a possibility that some portion of

the S₂population is directly generated by the photoexcita-

tion. The spectrum in Fig. 3͑a͒ is assigned to the S_n←S₃

absorption spectra. The band shape of the S_n←S₃absorption

is distinct from that of the S_n←S₂absorption. The vibra-

S₂lifetime, is obtained as 0.73Ϯ0.2 ps. This value is smaller

than the value reported in our previous letter ͑1.2 ps͒ and

25

that recently reported by Larson et al. ͑1.8 ps͒,²⁶but larger

than that determined by Takeuchi and Tahara (0.37

Ϯ0.02 ps) using the ﬂuorescence up-conversion technique.²⁷

In our previous letter, only three SVD components were

taken into account because of a lower signal to noise ratio

and a narrower wavelength range measured.²⁵An exponen-

tial time constant in the SVD analysis is strongly inﬂuenced

by the number of components.³⁴We believe that the present

value is more reliable than the previous one. We have no

TABLE I. Excited-state lifetimes of all-trans retinal in hexane at 293 K

obtained from the fetosecond time-resolved VIS absorption spectra.

State

Lifetime

S₃

S₂

S₁

T₁

0.03 ps ͑ﬁxed͒

0.73Ϯ0.2 ps

32Ϯ2 ps

*

tionally excited S₂state (S₂) does not seem to contribute to

the absorption band in Fig. 3͑a͒. Note that vibrational exci-

ϱ ͑ﬁxed͒

tation most often causes only broadening of an absorption

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:

130.39.62.90 On: Fri, 29 Aug 2014 14:13:24

1402

J. Chem. Phys., Vol. 109, No. 4, 22 July 1998

S. Yamaguchi and H. Hamaguchi

FIG. 5. Singular values obtained from the SVD analysis of the time-

resolved UV spectra of all-trans retinal in hexane at 298 K ͑solid circles͒

and 269 K ͑open circles͒.

ments were performed with a cyclohexane solution at 293 K

which gave nearly identical spectra with those in hexane at

298 K. This once again conﬁrms our idea that hexane and

cyclohexane give similar environments for all-trans retinal in

the excited electronic states.

FIG. 4. Femtosecond time-resolved UV absorption spectra of all-trans reti-

nal in hexane at various time delays at ͑a͒ 298 K and ͑b͒ 269 K.

The SVD analysis of the femtosecond time-resolved UV

absorption spectra in Figs. 4͑a͒ and 4͑b͒ gave two series of

singular values as shown in Fig. 5. We recognize two prin-

cipal singular values in both of these two series. The third

and higher SVD components are disregarded. As shown

above, the time-resolved VIS absorption spectra of all-trans

retinal have been successfully interpreted in terms of Kinet-

ics Scheme ͑I͒. The ﬁve transient species of which four are

linearly independent are involved in Kinetics Scheme ͑I͒. It

is now necessary to identify the two transient species that

contribute to the time-resolved UV spectra. One is deﬁnitely

the S₀state, because the negative signals in Fig. 4 must be

due to the ground-state bleaching. The other is most likely to

be the S₁state which has a 32 ps lifetime that corresponds to

the observed positive signals. The ﬁtting of the two principal

temporal evolutions of the SVD components were carried

out as in the case of the VIS spectra. It turned out that the

ﬁtting was successful only when the following model func-

tions were assumed for the S₁and S₀populations:

band.^28,40,41The small negative band at wavelength 454 nm

in Fig. 3͑a͒ is the impulsive stimulated Raman gain signal

assigned to the CH stretching modes of hexane. The Raman

gain process is also instantaneous in the present time resolu-

tion and has the same temporal evolution as that of the S₃

population. Two possibilities are considered for the negative

signals around 407 nm in Fig. 3͑a͒; the ground-state bleach-

ing or the stimulated emission gain. It will be shown later in

the femtosecond time-resolved UV spectra that negative sig-

nals due to the ground-state bleaching is not expected to be

large in the wavelength range longer than 400 nm. Therefore,

the stimulated emission gain is more likely than the ground-

state bleaching.

B. Femtosecond time-resolved UV absorption spectra

Fourteen representative femtosecond time-resolved UV

absorption spectra of all-trans retinal in hexane at 298 K and

those at 269 K are shown in Figs. 4͑a͒ and 4͑b͒, respectively.

The femtosecond time-resolved absorption spectra in the

near UV region are reported for the ﬁrst time to the best of

our knowledge. Since the ground-state absorption bands of

all-trans retinal exist in the wavelength range 310–390 nm,

we expect that negative signals due to the ground-state

bleaching dominate the spectra. The bleaching band shape at

100 ps after the photoexcitation is in agreement with that

measured by Veyret et al. with nanosecond time resolution.³³

In the time range 0.6–20 ps, positive signals are observed in

the wavelength range longer than 370 nm. Similar measure-

f_S1t͒ϭe^Ϫ␥¹^tϪe^{Ϫ␥ t}

,

.

͑12͒

͑13͒

2

͑

f_S0t͒ϭϪ1ϪAe^{Ϫ␥ t}

p

͑

Equation ͑12͒ is derived from Eq. ͑4͒ by using the relation

␥₃ӷ␥₂. The physical meaning of Eq. ͑13͒ and the new pa-

rameters A and ␥_pwill be discussed later.

Figure 6 shows the spectra and the population changes

of the S₁and S₀states of all-trans retinal in hexane at 298 K

obtained by the SVD analysis of the spectra in Fig. 4͑a͒. The

three time constants (␥^Ϫ₂¹,␥₁^Ϫ1,␥^Ϫ_p¹) and the parameter A

determined from eight independent measurements are as fol-

lows:

␥

^Ϫ1ϭ0.56Ϯ0.05 ps,

␥

^Ϫ1ϭ31Ϯ3 ps,

␥

^Ϫ1ϭ7.2

2

1

p

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:

130.39.62.90 On: Fri, 29 Aug 2014 14:13:24

J. Chem. Phys., Vol. 109, No. 4, 22 July 1998

S. Yamaguchi and H. Hamaguchi

1403

FIG. 6. Spectra ͓͑a͒ and ͑b͔͒ and populations ͓͑c͒ and ͑d͔͒ of the two transient species obtained from the time-resolved UV spectra of all-trans retinal in hexane

at 298 K.

Ϯ2 ps, Aϭ0.36Ϯ0.08. The two time constants, ␥^Ϫ₂¹and

^Ϫ1, are in good agreement with those determined from the

1Ϫ␾ ͒␥ ␥

2 3

␥₃Ϫ␥_p

␥₃Ϫ␥₂

͑

2

e^Ϫ␥^p^tϪ

e^{Ϫ␥ t}

2

*

p

ϭ

͔

͓

ͩ

␥

1

͑␥₂Ϫ␥_p͒͑␥₃Ϫ␥_p͒

VIS spectra within experimental uncertainties. It is con-

ﬁrmed that the transient species showing the spectrum in Fig.

6͑a͒ and the population in Fig. 6͑c͒ is the same S₁all-trans

isomer as that identiﬁed in the analysis of the VIS data. As

seen from Fig. 6͑a͒, a strong S_n←S₁absorption is located in

the wavelength range 360–390 nm. The negative population

in Fig. 6͑d͒ is ascribed to the ground-state bleaching. Though

the instantaneous decrease of the S₀population in Fig. 6͑d͒

is consistent with the depletion of the S₀all-trans isomer by

the photoexcitation, its recovery with the time constant of 7.2

ps (ϭ␥^Ϫ_p¹) has to be accounted for. The most simple and

reasonable modiﬁcation of Kinetics Scheme ͑I͒ is to assume

that an intermediate having a lifetime ␥^Ϫ_p¹exists and that it

␥₂Ϫ␥_p

3

ϩ

e^{Ϫ␥ t}

.

͑15͒

ͪ

␥₃Ϫ␥₂

By substituting Eqs. ͑2͒–͑5͒ and ͑15͒ for Eq. ͑14͒ and using

the relations ␥₃ӷ␥_pand ␥₂ӷ␥_p, Eq. ͑14͒ is reduced to Eq.

͑13͒. Thus, the kinetics assumed in the SVD analysis of the

UV absorption spectra, Eqs. ͑12͒ and ͑13͒, are derived from

Kinetics Scheme ͑II͒. We also have a relation between ␾

and A as follows:

2

1

␾₂ϭ

.

͑16͒

1ϩA

*

exclusively decays to S₀. The symbol p is hereafter used to

The S₁and T₁quantum yields are given by 1/(1ϩA), and

*

designate this intermediate. As discussed later, the p state is

considered to be the ‘‘perpendicular excited singlet state’’

which plays an essential role in the process of isomerization.

*

that of p by A/(1ϩA). There is no S₁→S₀relaxation path-

way (␾₁ϭ1) as indicated by the constant S₀population for

tϾ20 ps.

*

Either the S₃or the S₂state can be the precursor of p , since

the lifetime of p (␥^Ϫ_p¹) is longer than those of S₃and S₂

The SVD analyses of the femtosecond UV spectra of

all-trans retinal in hexane at lower temperatures were per-

formed in similar ways. It was found in the actual ﬁtting

procedures that ␥_pwas nearly independent of temperature

while A was not. Both ␥_pand A were determined by a small

fraction of the data in the time range 0–20 ps ͓see Fig. 6͑d͔͒,

and it was not possible to determine these two quantities

simultaneously with small uncertainties. Therefore, ␥^Ϫ_p¹was

ﬁxed to 7.2 ps in order to determine the temperature depen-

dence of A as exact as possible. The spectra and the popula-

tions of the S₁and S₀all-trans species in hexane at 269 K

are shown in Fig. 7. The amplitude A of the S₀recovery in

Fig. 7͑d͒ is much smaller than that in Fig. 6͑d͒. The S₁decay

in Fig. 7͑c͒ is slower than that in Fig. 6͑c͒. The converged

values of the ﬁtting parameters, ␥^Ϫ₂¹, ␥₁^Ϫ1, A, obtained from

several independent measurements at four temperatures are

given in Table II.

*

but is shorter than those of S₁and T₁. Here we assume that

the S₂state is the precursor of p . This is to assume that the

isomerization proceeds from the S₂state. Then, we need to

add to Kinetics Scheme ͑I͒ an extra relaxation pathway of the

S₂state

*

␥

␾ ␥

␥

1

3

2

S —— S —— S —— S ——

T

1

→

0

3

2

1

hv

͑1Ϫ␾ ͒␥

␥

p

2

*

S ——

p

—— S .

͑II͒

→

2

0

*

The S₀and p populations in Kinetics Scheme ͑II͒ are given

as follows:

*

ϭϪ S Ϫ S Ϫ S Ϫ T Ϫ p ,

͔

S

͑14͒

͓

͔

͓

͔

͓

͔

͓

͔

͓ ͔

͓

0

3

2

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:

130.39.62.90 On: Fri, 29 Aug 2014 14:13:24

1404

J. Chem. Phys., Vol. 109, No. 4, 22 July 1998

S. Yamaguchi and H. Hamaguchi

FIG. 7. Spectra ͓͑a͒ and ͑b͔͒ and populations ͓͑c͒ and ͑d͔͒ of the two transient species obtained from the time-resolved UV spectra of all-trans retinal in hexane

at 269 K.

IV. DISCUSSION

ing that the femtosecond time-resolved VIS absorption spec-

tra of all-trans retinal are fully explained by Kinetics Scheme

͑I͒. This means that no 13-cis excited-state species as a pho-

toisomerization product is observed in the all-trans VIS

spectra. We have no evidence for the all-trans

→13-cis•9-cis photoisomerization pathway that is com-

plete in the electronic excited states. It is, therefore, neces-

sary to ﬁgure out another reaction mechanism that is not

complete in the excited singlet states.

All the femtosecond time-resolved spectra, both in the

VIS and UV wavelength regions, have been successfully in-

terpreted by the SVD analysis based on Kinetics Scheme ͑II͒.

This scheme is fully consistent with the assignment of the S₃

state to B^ϩ_u(␲,␲ ) which was recently veriﬁed by femto-

*

second

time-resolved

ﬂuorescence

up-conversion

spectroscopy.²⁷The other two excited singlet states, S₂and

S₁, are most likely to be A^Ϫ_g(␲,␲ ) and (n,␲ ), respec-

*

Now we consider an isomerization scheme shown in Fig.

tively. It has been shown by an electron spin echo experi-

*

9, which involves the p state as the key intermediate. In this

*

ment that the T₁state of all-trans retinal is of (␲,␲ )

character.⁴²The very fast S₁→T₁intersystem crossing ͑32

scheme, the isomerization proceeds from the S₂state to the

1

3

*

perpendicular excited singlet state, p , which ﬁnally gives

*

ps͒ is indicative of a direct process (n,␲ )→ (␲,␲ ),

the trans and cis isomers in the ground state in a nearly

50/50 ratio. The isomerization scheme shown in Fig. 9 is

fully consistent with Kinetics Scheme ͑II͒ as will be shown

later. All-trans retinal in nonpolar solvents undergoes

isomerization to both the 13-cis and 9-cis isomers. There-

fore, strictly speaking, we should consider two different per-

pendicular conﬁgurations of which one is twisted about the

C₁₃vC₁₄double bond and the other about the C₉vC₁₀

double bond. However it is sufﬁcient to assume only the

former perpendicular state because ͑i͒ the 13-cis quantum

yield is about 5–10 times higher than the 9-cis quantum

yield, and ͑ii͒ the signal to noise ratio in the present femto-

second time-resolved UV absorption data is not sufﬁcient to

assume two different perpendicular states with different life-

times. Therefore, the isomerization scheme in Fig. 9 does not

distinguish the 13-cis and 9-cis isomers. The reaction mecha-

nism is then the same as that of the trans→cis photoi-

*

suggesting that the S₁state is of (n,␲ ) character. The S₂

state is then considered to be of A^Ϫ_g(␲,␲ ) character.

*

Kinetics Scheme ͑I͒ starts from the S₀state with the

all-trans conﬁguration and ends with the T₁state having the

same conﬁguration. It is, therefore, natural to assume that the

three excited singlet states have also the all-trans conﬁgura-

tion. In order to check whether other cis excited states are

involved in the all-trans VIS spectra ͑Fig. 1͒ as minor com-

ponents, the products of the spectra and the populations of

the S₃, S₂, S₁, and T₁states ͑Fig. 3͒ were subtracted from

the raw all-trans VIS spectra. The resultant residual spectra

are shown in Fig. 8 together with the raw femtosecond time-

resolved VIS absorption spectra of 13-cis retinal in hexane.

There are no meaningful bands in the residual spectra, show-

TABLE II. Converged ﬁtting parameters in the analyses of the femtosecond

time-resolved UV absorption spectra of all-trans retinal in aerated nonpolar

solvents at four different temperatures.

*

somerization of stilbene. The p state decays to the perpen-

dicular ground state, p, which is then deactivated to give the

trans and cis isomers in the ground state in a 50/50 ratio. The

Temperature/K

S₂lifetime/ps

S₁lifetime/ps

A

298

293

273

269

0.56Ϯ0.05

0.47Ϯ0.05

0.53Ϯ0.05

0.49Ϯ0.02

31Ϯ3

31Ϯ1

42Ϯ2

42Ϯ1

0.36Ϯ0.08

0.29Ϯ0.04

0.21Ϯ0.01

0.18Ϯ0.02

*

decay rate of p is assumed to be much larger than that of p ,

(7.2 ps)^Ϫ1. The lifetime of the perpendicular excited singlet

state of stilbene is thought to be much shorter than a

picosecond^43–45while that of trans-1, 1 -biindanylidene

Ј

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:

130.39.62.90 On: Fri, 29 Aug 2014 14:13:24

J. Chem. Phys., Vol. 109, No. 4, 22 July 1998

S. Yamaguchi and H. Hamaguchi

1405

FIG. 10. Temperature dependence of all-trans→13-cis•9-cis photoi-

somerization total quantum yield obtained from the femtosecond UV data

͑solid circles͒. The isomerization quantum yields upon direct photoexcita-

tion measured by HPLC are shown together: Open circles, W. H. Waddell

and K. Chihara ͑Ref. 5͒; solid triangle, S. Ganapathy and R. S. H. Liu ͑Ref.

6͒.

͑stiff stilbene͒ is about 10 ps.⁴⁶The excited singlet state of

tetraphenylethylene, for which the perpendicular conﬁgura-

tion is conﬁrmed by nanosecond time-resolved Raman

spectroscopy,⁴⁷has a lifetime of 3 ns.⁴⁸The 7 ps lifetime of

*

the p state of retinal is well in the range of the reported

FIG. 8. ͑a͒ Residual spectra of all-trans retinal in hexane at various time

delays obtained by subtracting the contribution of the four transient species

in Fig. 3 from the raw spectra in Fig. 1. ͑b͒ Femtosecond time-resolved VIS

absorption spectra of 13-cis retinal in hexane at various time delays.

lifetimes of the excited singlet perpendicular states of oleﬁns.

According to the isomerization scheme in Fig. 9, the

total quantum yield of the all-trans→13-cis•9-cis photoi-

somerization, ␾_iso, can be derived from the femtosecond

UV data as follows:

1

2

A

␾

_isoϭ

•

.

͑17͒

1ϩA

*

Note that ␾ is half of the quantum yield of p . The tem-

iso

perature dependence of the all-trans→13-cis•9-cis pho-

toisomerization total quantum yield is obtained from Eq. ͑17͒

and Table II as shown in Fig. 10. The total quantum yields

upon direct photoexcitation measured by HPLC^5,6are also

shown together. The total quantum yields at the four differ-

ent temperatures obtained in the present time-resolved spec-

troscopic measurements agree well with the literature values.

This agreement strongly supports the present isomerization

scheme in Fig. 9.

The same S₀recovery kinetics ͓Eq. ͑13͔͒ is obtained if

we assume that the S₃state rather than the S₂state is the

*

precursor of p . This means that we cannot unequivocally

determine the isomerization pathway solely from the present

femtosecond UV data. However, the fact that the isomeriza-

tion quantum yield decreases with decreasing temperature

means that thermal excitations are involved in the process of

*

the p formation. The S₃lifetime, which has been reported

FIG. 9. All-trans→mono-cis photoisomerization reaction scheme of reti-

nal in aerated nonpolar solvents.

as 0.03 ps,²⁷is too short to be associated with any kind of

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:

130.39.62.90 On: Fri, 29 Aug 2014 14:13:24

1406

J. Chem. Phys., Vol. 109, No. 4, 22 July 1998

S. Yamaguchi and H. Hamaguchi

dependence expression of ␥_2→pinstead of Eq. ͑19͒. At the

*

present stage, it is appropriate to ascribe the temperature de-

pendence of the quantum yield to the existence of an energy

barrier.

In Table II, we see the temperature dependence of the S₂

and S₁lifetimes. Since ␥_2→1is about four times larger than

␥

, the S₂lifetime depends on T only weakly. The S₂

2→p

*

lifetime change on going from 269 to 298 K is estimated as

0.05 ps by using the relation ␥₂ϭ␥_2→1ϩa_2→p

*

exp(ϪE_a/k_BT). The experimental errors of the S₂lifetime in

Table II are too large to know whether the S₂lifetime de-

pends on T or not. The observed S₁lifetime change is larger

than the experimental errors. The S₁lifetime becomes longer

with decreasing temperature. This temperature dependence

of the S₁lifetime is already well-known. The nonradiative

T₁→S₀intersystem crossing rate generally decreases with

decreasing temperature. On the assumption of

␥

1

ϰexp(ϪE_S1→T/k_BT), the activation energy of the S₁→T₁in-

tersystem crossing (E_S1→T) is estimated as 6ϫ10²cm^Ϫ1. It

seems that the intersystem crossing from the S₁state to the

T₁state in all-trans retinal is accompanied with a structural

change along a coordinate other than that of the double-bond

isomerization.

FIG. 11. Arrhenius plot to obtain the activation energy of all-trans

→13-cis•9-cis isomerization. Two dashed lines correspond to the error

range of the activation energy.

*

thermal excitations. It is considered that the precursor of p

There are two independent ways to obtain temperature-

dependent S₁quantum yield, ␾_S1, from the femtosecond UV

is the S₂state. Then, the total quantum yield is determined

by two rate constants in the following way:

data. Using Eq. ͑16͒, ␾ is directly obtained from A whose

S1

temperature dependence is given in Table II. In addition, the

1

2

␥

2→p

*

␾

_isoϭ

•

,

͑18͒

relative value of ␾ is also obtained from the ratio of the

S1

␥

_2→1ϩ␥_2→p

*

area of the S₁absorption band ͓Figs. 6͑a͒ and 7͑a͔͒ to that of

where ␥_2→1is the S₂→S₁internal conversion rate constant

the S₀͓Figs. 6͑b͒ and 7͑b͔͒. The temperature dependent ␾

S1

*

and ␥_2→pis the S₂→p decay rate constant, and the equa-

values obtained by the former method and those by the latter

with a proper scaling factor are compared in Fig. 12. They

approximately coincide with each other. The coincidence of

*

tion ␥₂ϭ␥_2→1ϩ␥_2→phas to be satisﬁed. The temperature

*

dependence of ␾ in Fig. 10 means that ␥_2→pdecreases as

iso

*

the temperature decreases. The most straightforward way to

the independently obtained ␾ values strongly supports the

S1

introduce the temperature dependence of ␥_2→pis to use a

present isomerization scheme. Note that ␾ in the present

*

S1

transition-state expression as follows:

scheme is equal to the triplet quantum yield ␾_T. The ob-

tained ␾ or ␾ value at 298 K is 0.74Ϯ0.04 which is

T

S1

E_a

␥

ϭa_2→pexp Ϫ

,

͑19͒

slightly higher than the reported ␾ values which ranged

ͩ

ͪ

2→p

T

*

k_BT

from 0.43 to 0.7.^{32,33,54–56}All the reported ␾ values were

T

where a_2→pis the rate at a high-temperature limit, E_ais an

activation energy, and T is temperature. On the assumption

that ␥_2→1is independent of T, an Arrhenius-type relation is

obtained by using Eqs. ͑18͒ and ͑19͒:

obtained by means of the T_n←T₁absorption measurements

with nano- and microsecond time resolution. In the analysis

of the T_n←T₁absorption spectrum, it was necessary to as-

sume the molar absorption coefﬁcient of T₁all-trans retinal

and the effects of multiple excitation had to be taken into

account. The femtosecond UV absorption analysis does not

*

1

E_a

␥

2→1

log

Ϫ1 ϭ

ϩlog

.

͑20͒

ͩ

ͪ

ͩ

ͪ

2␾

k_BT

a_2→p

iso

*

need any assumption, and therefore, the present ␾ value is

T

Plotting log(2^Ϫ1

␾

^Ϫ_iso¹Ϫ1) versus T^Ϫ1as shown in Fig. 11, we

more trustworthy than the previously reported values.

obtain E_aas (1.2Ϯ0.6)ϫ10³cm^Ϫ1(14Ϯ7 kJ mol^Ϫ1). The

data of Waddell and Chihara gave 7.6ϫ10²cm^Ϫ1. The

present E_avalue is similar to the activation energy for the

trans→perpendicular reaction pathway of S₁stilbene

We now consider how the isomerization scheme in Fig.

9 becomes compatible with Kinetics Scheme ͑II͒ in which

isomerization is not included. An ideal SVD analysis of the

femtosecond UV data based on the isomerization scheme in

Fig. 9 requires three components, the S₁all-trans, S₀all-

trans, and S₀13-cis isomers, but in fact only two SVD com-

ponents were obtained experimentally. It is already shown by

the present SVD analysis of the UV data that the S₁and S₀

all-trans species contribute to the two leading SVD compo-

nents. It is unlikely that all the contribution of the S₀13-cis

isomer to the SVD components was neglected in the present

SVD analysis, because the S₀absorption spectra of the all-

(1200 cm^Ϫ1).^49–53However, ␥_2→pat 298 K is estimated to

*

be about (2 ps)^Ϫ1which is more than ten times larger than

the trans→perpendicular reaction rate of S₁stilbene.^50,52,53

It is of great importance to check whether such an ultrafast

process can still be interpreted in terms of the ordinary

transition-state formalism that assumes the equilibrium be-

tween the reactant and the transition state. The present data

are not good enough to bring up a new temperature-

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:

130.39.62.90 On: Fri, 29 Aug 2014 14:13:24

J. Chem. Phys., Vol. 109, No. 4, 22 July 1998

S. Yamaguchi and H. Hamaguchi

1407

at least ten times better than in the present experiment, in

order to include the contribution of the 13-cis–all-trans dif-

ference spectrum correctly in the SVD analysis with the

three components. Owing to both the small difference be-

tween the S₀all-trans and 13-cis spectra and the low quan-

tum yield of the all-trans→13-cis•9-cis photoisomeriza-

tion, the SVD analysis of the UV data has been soundly

performed with only two components.

*

No transient absorption band ascribable to p was ob-

served either in the VIS region ͑400–800 nm͒ or in the near

UV region ͑310–390 nm͒, though it has been known that the

*

p

absorption band of tetraphenylethylene is located at 420

nm⁴⁸and that of stiff stilbene is located at 350 nm.⁴⁶It is

*

possible that a p absorption band of retinal lies in the wave-

length range shorter than 300 nm or longer than 800 nm. It is

*

also noted that the p quantum yield of retinal has proved to

be as small as 0.26Ϯ0.04 in hexane at room temperature,

while that of tetraphenylethylene or stiff stilbene is nearly

equal to one. Such a small quantum yield makes it difﬁcult to

FIG. 12. Temperature dependence of the S₁quantum yield obtained from

the femtosecond UV data by means of two independent methods. One af-

fords absolute values by using isomerization quantum yields ͑solid circles͒,

and the other gives relative values from absorbance ratio of S₁to S₀͑open

circles͒.

*

detect the p retinal absorption. Since the molar absorption

*

coefﬁcient of p is neither known nor predicted, it is not

possible to estimate the signal to noise ratio of time-resolved

*

spectra which enables the identiﬁcation of the p absorption

by means of the SVD analysis.

Finally, two additional remarks are made on the pro-

posed isomerization scheme in Fig. 9. First, it is possible to

ascribe the 7 ps lifetime to the perpendicular ground state, p.

A potential minimum at the perpendicular conﬁguration is

necessary to account for the lifetime as long as 7 ps. How-

ever, existence of such a potential minimum in the ground

state has never been rationalized either theoretically or ex-

perimentally. The perpendicular ground state whose lifetime

is 7.2 ps is much less likely than the perpendicular excited

state. Secondly, it is also possible to ascribe the 7 ps lifetime

trans and 13-cis isomers are nearly identical. On the con-

trary, it is highly likely that most of the contribution of the

S₀13-cis isomer has already been taken into account in the

present SVD analysis, and what was neglected is the contri-

bution of the difference spectrum between the S₀absorption

spectra of the 13-cis and all-trans isomers. When the tempo-

ral evolution of the S₀all-trans isomer is expressed as f_S0(t)

of Eq. ͑13͒, that of the difference spectrum between the 13-

cis and all-trans isomers is written as follows:

A

*

to the vibrationally excited electronic ground state (S₀).

p

f_difft͒ϭ

1Ϫe^{Ϫ␥ t}͒.

͑21͒

͑

2

*

However, the assignment of p to the S₀state is not consis-

tent with Fig. 10, which leads to an excellent agreement

between the isomerization quantum yields determined by the

two independent methods.

The contribution of each transient species to the SVD com-

ponents is determined by the product of the area of its ab-

sorption spectrum and that of its temporal evolution. The

ratio of the area of the S₀all-trans spectrum to that of the

13-cis—all-trans difference spectrum is obtained from the

ground-state absorption spectra as 22.5. The ratio of the area

of the S₀all-trans temporal evolution to that of the temporal

evolution of the 13-cis—all-trans difference spectrum is ob-

tained from Eqs. ͑13͒ and ͑21͒ as 14.5. Accordingly the ratio

of the contribution of the S₀all-trans isomer to that of the

13-cis—all-trans difference spectrum is calculated to be 326.

The ratio of the contribution of the S₀all-trans isomer to that

of the S₁all-trans isomer is calculated in the same way to be

1.77 from the spectra and the populations in Fig. 6. Conse-

quently the contributions of the S₀all-trans isomer, that of

the S₁all-trans isomer, and that of the 13-cis—all-trans dif-

ference spectrum are in the ratio 1:0.57:0.0031. The ﬁrst

through the fourth singular values derived from the femto-

second UV data ͑Fig. 5͒ are in the ratio 1:0.44:0.030:0.026,

and the third and higher SVD components are dominated by

noise. The expected contribution of the 13-cis—all-trans dif-

ference spectrum is ten times smaller than the present noise

level. Therefore, the signal to noise ratio has to be improved

V. SUMMARY

The excited-state dynamics and the photoisomerization

reactions of all-trans retinal in aerated nonpolar solvents

were studied by using femtosecond time-resolved UV-VIS

absorption spectroscopy. It was found in the time-resolved

VIS spectra of all-trans retinal that four transient species

͑S₃, S₂, S₁, T₁͒ were generated in sequence after the 400

nm photoexcitation. On the assumption that the S₃state is of

ϩ

*

B_u(␲,␲ ) character, the S₂and S₁states were character-

ized as A^Ϫ_g(␲,␲ ) and (n,␲ ), respectively. Comparison

of the time-resolved VIS spectra of the all-trans and 13-cis

isomers provided no evidence for an isomerization reaction

pathway complete within the excited singlet manifold. The

femtosecond time-resolved UV absorption spectra of all-

trans retinal were successfully analyzed in terms of an

isomerization scheme involving a perpendicular excited sin-

*

glet state, p . The p lifetime was found to be 7.2Ϯ2 ps and

was much longer than the S₃and S₂lifetimes but was shorter

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:

130.39.62.90 On: Fri, 29 Aug 2014 14:13:24

1408

J. Chem. Phys., Vol. 109, No. 4, 22 July 1998

S. Yamaguchi and H. Hamaguchi

than the S₁lifetime. It is proposed that the precursor of p is

the S₂state. The S₁state does not take part in the isomer-

ization reaction. The quantum yields of the all-trans

→13-cis•9-cis isomerization in the temperature range

269–298 K obtained from the femtosecond time-resolved

UV spectra were in good agreement with the reported values

measured by HPLC. The S₁quantum yields obtained from

the femtosecond UV data by way of two independent meth-

ods were also in good agreement. These agreements strongly

support the proposed isomerization scheme. The height of

the energy barrier on the reaction pathway from the S₂state

²⁴T. Tahara and H. Hamaguchi, Chem. Phys. Lett. 234, 275 ͑1995͒.

²⁵S. Yamaguchi and H. Hamaguchi, J. Mol. Struct. 379, 87 ͑1996͒.

²⁶E. J. Larson, L. A. Friesen, and C. K. Johnson, Chem. Phys. Lett. 265, 161

͑1997͒.

*

²⁷S. Takeuchi and T. Tahara, J. Phys. Chem. A 101, 3052 ͑1997͒.

²⁸S. Yamaguchi and H. Hamaguchi, Chem. Phys. Lett. 227, 255 ͑1994͒.

²⁹S. Yamaguchi and H. Hamaguchi, Appl. Spectrosc. 49, 1513 ͑1995͒.

³⁰W. Dawson and E. W. Abrahamson, J. Phys. Chem. 66, 2542 ͑1962͒.

³¹Y. Mukai, Y. Koyama, Y. Hirata, and N. Mataga, J. Phys. Chem. 92, 4649

͑1988͒.

³²M. M. Fisher and K. Weiss, Photochem. Photobiol. 20, 423 ͑1974͒.

³³B. Veyret, S. G. Davis, M. Yoshida, and K. Weiss, J. Am. Chem. Soc.

100, 3283 ͑1978͒.

³⁴W. G. Chen and M. S. Braiman, Photochem. Photobiol. 54, 905 ͑1991͒.

³⁵T. Iwata and J. Koshoubu, Appl. Spectrosc. 48, 1453 ͑1994͒.

³⁶M. Lupu and D. Todor, Chemometrics and Intelligent Laboratory Systems

29, 11 ͑1995͒.

3

Ϫ1

*

to p was estimated as (1.2Ϯ0.6)ϫ10 cm

.

¹H. Hamaguchi, in Vibrational Spectra and Structure, edited by J. R. Dur-

ing ͑Elsevier, Amsterdam, 1987͒, p. 227.

³⁷W. H. A. M. Vandenbroek, D. Wienke, W. J. Melssen, C. W. A. Decrom,

and L. Buydens, Anal. Chem. 67, 3753 ͑1995͒.

²R. M. Hochstrasser and C. K. Johnson, in Ultrashort Laser Pulses-

Generation and Applications, edited by W. Kaiser ͑Springer-Verlag, Ber-

lin, 1993͒, p. 396.

³⁸K. Mohanalingam and H. Hamaguchi, Chem. Lett. 1997, 157.

³⁹We note that the inclusion of the direct S₂←S₀photoexcitation in the

SVD analysis does not affect the resultant time constants and the quantum

yields of the S₁and T₁states.

³R. A. Mathies, in Ultrafast Processes in Chemistry and Photobiology,

edited by M. A. El-Sayed, I. Tanaka, and Y. Molin ͑Blackwell Science,

Oxford, 1995͒, p. 215.

⁴S. Ganapathy and R. S. H. Liu, Photochem. Photobiol. 56, 959 ͑1992͒.

⁵W. H. Waddell and K. Chihara, J. Am. Chem. Soc. 103, 7389 ͑1981͒.

⁶S. Ganapathy and R. S. H. Liu, J. Am. Chem. Soc. 114, 3459 ͑1992͒.

⁷H. Hamaguchi, H. Okamoto, M. Tasumi, Y. Mukai, and Y. Koyama,

Chem. Phys. Lett. 107, 355 ͑1984͒.

⁴⁰A. Samanta, C. Devadoss, and R. W. Fessenden, J. Phys. Chem. 94, 7106

͑1990͒.

⁴¹H. Miyasaka, M. Hagihara, T. Okada, and N. Mataga, Chem. Phys. Lett.

188, 259 ͑1992͒.

⁴²M. Ros, M. A. Hogenboom, P. Kok, and E. J. J. Groenen, J. Phys. Chem.

96, 2975 ͑1992͒.

⁸Y. Hirata, N. Mataga, Y. Mukai, and Y. Koyama, Chem. Phys. Lett. 134,

166 ͑1987͒.

⁴³R. J. Sension, S. T. Repinec, A. Z. Szarka, and R. M. Hochstrasser, J.

Chem. Phys. 98, 6291 ͑1993͒.

⁹T. Tahara, B. N. Toleutaev, and H. Hamaguchi, J. Chem. Phys. 100, 786

͑1994͒.

⁴⁴S. Abrash, S. Repinec, and R. M. Hochstrasser, J. Chem. Phys. 93, 1041

͑1990͒.

¹⁰T. Yuzawa and H. Hamaguchi, J. Mol. Struct. 352–353, 489 ͑1995͒.

¹¹T. Takemura, P. K. Das, G. Hug, and R. S. Becker, J. Am. Chem. Soc.

⁴⁵J. Saltiel, A. S. Walker, and J. D. F. Sears, J. Am. Chem. Soc. 115, 2453

͑1993͒.

100, 2626 ͑1978͒.

12

¨

G. Drikos, P. Morys, and H. Ruppel, Photochem. Photobiol. 40, 133

͑1984͒.

⁴⁶F. E. Doany, E. J. Heilweil, R. Moore, and R. M. Hochstrasser, J. Chem.

Phys. 80, 201 ͑1984͒.

13

¨

G. Drikos and H. Ruppel, Photochem. Photobiol. 40, 93 ͑1984͒.

G. Drikos, H. Ruppel, W. Sperling, and P. Morys, Photochem. Photobiol.

14

⁴⁷T. Tahara and H. Hamaguchi, Chem. Phys. Lett. 217, 369 ͑1994͒.

⁴⁸B. I. Greene, Chem. Phys. Lett. 79, 51 ͑1981͒.

⁴⁹J. A. Syage, P. M. Felker, and A. H. Zewail, J. Chem. Phys. 81, 4706

͑1984͒.

¨

40, 85 ͑1984͒.

¹⁵W. H. Waddell, A. M. Schaffer, and R. S. Becker, J. Am. Chem. Soc. 95,

8223 ͑1973͒.

¹⁶R. R. Birge, J. A. Bennett, B. M. Pierce, and T. M. Thomas, J. Am. Chem.

Soc. 100, 1533 ͑1978͒.

⁵⁰H. P. Good, U. P. Wild, E. Haas, E. Fischer, E.-P. Resewitz, and E.

Lippert, Ber. Bunsenges. Phys. Chem. 86, 126 ͑1982͒.

⁵¹J. Saltiel and J. T. D’Agostino, J. Am. Chem. Soc. 94, 6445 ͑1972͒.

⁵²M. Sumitani, N. Nakashima, K. Yoshihara, and S. Nagakura, Chem. Phys.

Lett. 51, 183 ͑1977͒.

¹⁷R. R. Birge, J. A. Bennett, L. M. Hubbard, H. L. Fang, B. M. Pierce, D. S.

Kliger, and G. E. Leroi, J. Am. Chem. Soc. 104, 2519 ͑1982͒.

¹⁸R. S. Becker, Photochem. Photobiol. 48, 369 ͑1988͒.

¹⁹S. Alex, H. L. Thanh, and D. Vocelle, Can. J. Chem. 70, 880 ͑1992͒.

²⁰J. Papanikolas, G. C. Walker, V. A. Shamamian, R. L. Christensen, and J.

C. Baum, J. Am. Chem. Soc. 112, 1912 ͑1990͒.

⁵³H. Hamaguchi and K. Iwata, Chem. Phys. Lett. 208, 465 ͑1993͒.

⁵⁴R. Bensasson, E. J. Land, and T. G. Truscott, Photochem. Photobiol. 21,

419 ͑1975͒.

²¹M. Merchan and R. Gonzalez-Luque, J. Chem. Phys. 106, 1112 ͑1997͒.

²²R. M. Hochstrasser, D. L. Narva, and A. C. Nelson, Chem. Phys. Lett. 43,

15 ͑1976͒.

⁵⁵R. Bensasson, E. J. Land, and T. G. Truscott, Photochem. Photobiol. 17,

53 ͑1973͒.

⁵⁶T. Rosenfeld, A. Alchalel, and M. Ottolenghi, J. Phys. Chem. 78, 336

͑1974͒.

²³A. G. Doukas, M. R. Jnnarkar, D. Chandra, R. R. Alfano, and R. H.

Callender, Chem. Phys. Lett. 100, 420 ͑1983͒.

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:

130.39.62.90 On: Fri, 29 Aug 2014 14:13:24

Article Doi

Femtosecond ultraviolet-visible absorption study of all-trans→13-cis9-cis photoisomerization of retinal

DOI: 10.1063/1.476692

Source and publish data:

Authors:

Article abstract of DOI:10.1063/1.476692

Full text of DOI:10.1063/1.476692

Products guided by the article

R&D Labs maybe for 472-86-6

Relevant to this article

Hot Product