Photolysis of triiodide studied by femtosecond pump-probe spectroscopy with emission detection

Article abstract of DOI:10.1021/jp0108521

The photodissociation of triiodide into diiodide and iodine in ethanol solution was studied by a novel femtosecond pump-probe technique. The dispersed emission of the triiodide sample was recorded as a function of the delay between a 407 nm pump pulse and

Full text of DOI:10.1021/jp0108521

© Copyright 2002 by the American Chemical Society
VOLUME 106, NUMBER 9, MARCH 7, 2002
ARTICLES
Photolysis of Triiodide Studied by Femtosecond Pump-Probe Spectroscopy with Emission
Detection
Peter Gilch,* Ingmar Hartl,† Qingrui An, and Wolfgang Zinth
Sektion Physik, Ludwig-Maximilians-UniVersita¨t, Oettingenstr. 67, D-80538 Mu¨nchen, Germany
ReceiVed: March 6, 2001; In Final Form: NoVember 19, 2001
The photodissociation of triiodide into diiodide and iodine in ethanol solution was studied by a novel
femtosecond pump-probe technique. The dispersed emission of the triiodide sample was recorded as a function
of the delay between a 407 nm pump pulse and a 815 nm probe pulse. The most prominent feature of the
resulting time dependent spectra is a strong anti-Stokes (with respect to the probe wavelength) contribution.
This contribution rises with a delay of ∼500 fs and decays on the 2 ps time scale. The decay of the anti-
Stokes spectrum is discussed in terms of the cooling dynamics of nascent diiodide for which the technique
presented here gives a more direct access than transient absorption spectroscopy.
-
2
Introduction
seconds to form diiodide I2 in its electronic ground state Σu
and an iodine I* atom in one of its spin-orbit states (2P3/2 or
Small iodine molecules and ions are popular guinea pigs for
the study of ultrafast processes in gas and condensed phase (see,
e.g., refs 1-5). They are well suited for such experiments,
because their electronically excited states are often repulsive
and their vibrational frequencies are very low.6 The first property
allows for an easy triggering of the ultrafast reactions by a
(femtosecond) laser pulse, and the second property facilitates
the observation of vibrational coherences with moderately short
laser pulses. One iodine system that has been studied in detail
by femtosecond transient absorption and resonance Raman
2P1/2
)
(see eq 1). Recently, in addition to this established
2,4
mechanism, a process competing with the dissociation has been
proposed by Ruhman et al.9,10 In this process, an intermediate
denoted X by the authors is formed, the nature of which is still
to be elucidated. It has been suggested that this intermediate
might be a bound excited state of triiodide. Ruhman et al.
observed the X intermediate when exciting triiodide in ethanol
solution with 307 nm light. At this wavelength, mainly the high-
frequency UV transition of triiodide is excited (see the spectrum
in Figure 1). The group of Vo¨hringer performed experiments
with 400 nm light exciting exclusively the low-energy band.4
They analyzed their 400 nm data solely in terms of the
dissociation reaction given in eq 1. When comparing 400 nm
excitation (low energy transition) with 266 nm excitation (high
energy transition), they found differences in the yield of the
diiodide formation.11 For 400 nm, they observed a yield of
100%, whereas it was determined to be 80% for 266 nm.
Therefore, the additional decay path way observed by the
Ruhman group might only be accessible when the higher
frequency band is excited.
spectroscopy is triiodide I3 in polar solution.2,4,7,8 The results
of these experiments are summarized as follows. When either
of two absorption bands of triiodide are excited, the following
reactions occur:2,9
-
hν
I3- 98 I3 * f I2- + I*
(1)
-
hν
I3- 98 I3 * f X
(2)
-
Excited triiodide I3-* dissociates within a few hundred femto-
The dissociation (eq 1) of triiodide after excitation of the low
energy band occurs within a few hundred femotoseconds.4
Because of this short reaction time, the photoproduct diiodide
* To whom correspondence should be addressed. Fax: +49-89/2180-
9202. E-mail: Peter.Gilch@physik.uni-muenchen.de.
† Present Address: 30 Systems, 26081 Avenue Hall, CA 91355.
10.1021/jp0108521 CCC: $22.00 © 2002 American Chemical Society
Published on Web 02/09/2002

1648 J. Phys. Chem. A, Vol. 106, No. 9, 2002
Gilch et al.
Figure 2. Jabloniski diagram of the photophysics of triiodide. Thick
horizontal lines symbolize electronic states, thin lines symbolize
vibrational states of diiodide. Dissociation continua are represented by
shaded areas. Solid arrows stand for absorption processes, dashed arrows
stand for fluorescence emission.
Figure 1. Absorption and emission spectra of triiodide in ethanol. For
the emission spectrum, the excitation was tuned to 407 nm. The Raman
lines of the ethanol solvent have been subtracted, the spectral sensitivity
of the spectrometer has been corrected for, and the spectrum has been
multiplied by the ν˜-2 factor to account for the constant wavelength
(and not wavenumber) resolution of the spectrograph.
function for high quantum numbers. Simulations of Banin et
al.13 demonstrate that these two maxima occur at ∼1100 and
∼560 nm, respectively, for a broad vibrational excitation
centered at ν ) 17; that is, as stated above, a very broad
detection window is necessary to monitor high vibrational
excitations. If a smaller detection window is used, it will be
difficult to monitor the high vibrational excitation expected
during the first several 100 fs by time-dependent absorption
spectroscopy.
I2- is vibrationally hot and coherently excited, storing initially
a part of the excess energy of the reaction.
The value of this energy depends on the spin-orbit state in
2
which the iodine (2P3/2 or P1/2) atom is formed. The energies
of these states differ by 7600 cm-1 12
If the low energy
.
absorption band of triiodide is excited, presumably the upper
spin-orbit state (2P1/2) of iodine is formed.4,7,11 Assuming that
the upper state is formed, the excess energy of the photoproducts
amounts to 4000 cm-1, otherwise it amounts to 11600 cm-1
.
The vibrational wave packet motion mentioned above can
give access to the vibrational cooling during that period. Because
of the anharmonicity of the diiodide ground-state potential, the
oscillation period of this motion should decrease during cooling.
This has indeed been observed.4,17 Analysis of this frequency
shift yielded a relaxation time of ∼400 fs with an amplitude of
1400 cm-1; that is, at early times, the vibrational relaxation
seems to be much faster than at later times. This frequency shift
has later been examined in more detail by the same group using
impulsive stimulated Raman scattering.17 An LPSVD analysis
of their data revealed three oscillatory components. Two
components have frequencies close to the fundamental (113
cm-1) and the first harmonic of diiodide. These frequencies are
observed nearly immediately after the photolysis of triiodide
and change weakly and smoothly with time. The third compo-
nent has initially a frequency of ∼70 cm-1 and rises quite
abruptly to the value of the fundamental within some 100 fs.
The behavior of the first two components can be attributed to
the cooling of diiodide, and the third component calls for another
mechanism. It has been attributed to a strengthening of the
diiodide force constant which occurs while the photofragments
separate which is in conflict with the cooling mechanism
mentioned above. Therefore, other techniques allowing the study
of the cooling dynamics during the first few 100 fs are desirable.
We here present an alternative and more direct method to
study the dynamics of the formation and the vibrational
relaxation of diiodide. In the present experiment, the fluores-
This energy is partially stored in the vibrational mode of the
I2 ion. The decay of this vibrational excitation has been
-
examined by several groups. Ruhman et al. used transient
absorption spectroscopy13 and impulsive stimulated Raman
scattering14 to monitor the cooling dynamics after 307 nm
excitation of triiodide. Barbara et al. photoexcited diiodide and
recorded the recovery of the ground-state bleach of diio-
dide.3,15,16 All of these studies indicate that the relaxation is
biphasic. Most of the vibrational energy seems to be dissipated
in some 100 fs, whereas relaxation at the bottom of diiodide
potential well occurs in 3-4 ps. These results have been
confirmed in principle by Vo¨hringer et al. They worked with
400 nm pump light to excite triiodide which makes their results
most comparable with the experiments described here. They
recorded the absorption spectrum as a function of pump-probe
delay4 and used the shape of these spectra as an indicator for
the vibrational excitation of nascent diiodide. They found that
1.5 ps after excitation only a small amount of energy of ∼200
cm-1 corresponding to roughly two vibrational quanta of I2- is
still stored in its vibrational mode. The decay time of this
vibrational energy was determined to be 3.5 ps.
The study of the short time dynamics (<1.5 ps) of the
vibrational excitation by absorption methods is difficult for the
following two reasons. At early times, wave packet dynamics
are superimposed on the signal and obscure the reconstruction
of the time-dependent spectra. To monitor high vibrational
excitations of diiodide via transient absorption, a spectrally very
broad detection window is necessary3,14 which might cover
transitions to other electronic states. The simulated spectra in
ref 3 indicate that around the maximum of cold diiodide the
spectra of vibrationally excited diiodide molecules are essentially
flat. Therefore, for a precise determination of the excitation,
one has to observe the two maxima in the spectra which arise
from the two dominant antinodes of the vibrational wave
-
cence of the sample is observed after illumination of a I3
solution with two laser pulses: A 407 nm pump pulse excites
I3- and initiates its photodissociation. A delayed 815 nm probe
pulse excites the generated I2- molecules, and their fluorescence
-
emission is used to monitor the vibrational excitation of I2
.
The fluorescence emission is generated by the different inter-
mediates of the reaction sequence (see Figure 2): (i) The
-
-
excitation of I3 by the pump pulse populates the state I3
*

Photolysis of Triiodide
J. Phys. Chem. A, Vol. 106, No. 9, 2002 1649
which may fluoresce with photons hνFl3,blue. (ii) The probe pulse
with an optical path length of 1 mm and 0.1 mm. The 407 nm
excitation wavelength corresponds to the red edge of the lowest
absorption band of I3-. The emission spectrum corrected for
Raman contribution of the solvent and the spectral sensitivity
of the setup is depicted in Figure 1. The emission spectrum is
broad and structureless and peaks at 21 000 cm-1. For the
determination of the fluorescence quantum yield φf, the fluo-
rescence intensity was referenced to the fluorescence of phena-
zine in methanol which was chosen because of its absorption
band lying close to that of triiodide and its low emission
quantum yield. The yield of phenazine in turn was determined
to be 0.06% by comparing its integrated emission with that of
coumarin 47 in ethanol (φf ) 73%22). The validity of this yield
was checked by an estimate based on the Strickler-Berg relation
and the published lifetime of phenazine23 which yields the same
magnitude. The fluorescence quantum yield of I3- obtained this
way is 5.8 × 10-5. An estimate for the excited-state lifetime τs
of triiodide can be calculated by the relation
may interact via excited-state absorption with I3-* leading to
-
emission from higher excited states of I3 (hνFl3,red, hνFl3,uv).
The promotion of I3-* to a higher excited state of course
depopulates this state and, therefore, reduces emission from I3-*.
-
(iii) When I2 is formed, the probe pulse, being strongly
absorbed by I2-, brings this molecule to one of its excited
electronic states I2-*. Contribution i does not depend on the
pump-probe delay, and contributions ii and iii are sensitive to
this delay and, therefore, allow us to monitor the decay of I3-*,
the formation of I2-, and (as we will explain later on) the
-
vibrational cooling of nascent I2
.
Experimental Section
The basic parts of the experimental system have been
described in refs 18 and 19. The frequency-doubled output (407
nm, 5 µJ) of a 1 kHz titanium sapphire laser/amplifier system
was used to excite the triiodide sample. A part (10 µJ) of the
815 nm fundamental of the laser system constitutes the probe
beam which crosses the pump beam in the sample under a small
angle of ∼5°. The emission of the sample was collected by
reflective optics, dispersed by an imaging spectrograph, and
recorded by a back-illuminated liquid-nitrogen-cooled CCD
camera (for details on the detection setup, see ref 18). Scattered
laser light at the fundamental and the second harmonic was
suppressed by suitable notch filters. The spectral sensitivity of
the spectrometer-detector setup has been corrected for with a
blackbody radiator as a reference. It should be stressed that in
this experiment detection of the emission was not gated. The
experimental response function (cross correlation function) was
determined by a two photon absorption method. The sample
was replaced by a 0.1 mm thick cuvette containing neat
1-methylnaphthalein. This liquid is transparent at 815 and 407
nm. Multiphoton absorption processes of the pump and the probe
pulse induce a fluorescence emission that is enhanced by ∼60%
when both pulses coincide. Recording of this fluorescence
(spectrally integrated) as a function of the delay time yields a
measure for the cross correlation function. For the experiments
described here, the cross correlation time was determined to be
300 fs (fwhm). The present setup can also be used to determine
emission lifetimes via an inverse Raman effect. A detailed
description will be given in a forthcoming publication.20 The
time-resolved absorption measurements were performed in the
group of M. E. Michel-Beyerle at the TU Mu¨nchen (for details
of the setup, see ref 21). The sample was prepared by dissolving
100 mg of I2 and 100 mg of KI in 50 mL ethanol. This
corresponds to a concentration of 8 mM I3- and an absorbance
at 407 nm (excitation wavelength) of 5.6 OD in the 1 mm
cuvette used. The sample solution was flown rapidly through
the cuvette to prevent local heating of the solution. Measure-
ments were performed at room temperature and with magic
angle polarization between pump and probe beams.
τs
φf )
(3)
τ0
where τ0 is the radiative lifetime. This lifetime was determined
via the Strickler-Berg relation24 using the absorption and
emission data to be 6.9 ns which in turn yields (eq 3) an excited
state lifetime of 400 fs. Of course, this value only gives the
order of magnitude of the excited state lifetime, particularly
because of some conceptional difficulties when applying the
Strickler-Berg relation to transitions involving repulsive states.
A more precise result from a time-resolved experiment will be
given below.
Time-Dependent Emission Spectra. The emission spectrum
of the I3- sample depends on the relative temporal delay between
407 nm pump and 815 nm probe pulse. If the probe pulse
precedes the pump pulse (large negative delay times), the
emission spectrum IFl(-∞) is a superposition of the pump
induced fluorescence of I3- and the Raman signal of the ethanol
solvent induced by the probe pulse (see insert in Figure 3a).
For positive delay times, the emission of the sample is enhanced
(see Figure 3a). The maximum enhancement amounts to ∼10%
of the total emission and has a distinctly different spectral
signature than the emission observed at negative delay times.
Details of the enhancement can be seen if the emission recorded
at large negative delay times IFl(-∞) is subtracted leading to
the enhanced emission signal ∆IFl(tD) ) IFl(tD) - IFl(-∞).
The Stokes Part of the Emission. The enhanced emission
on the Stokes side is broad and nearly structureless (Figure 3).
The increase of the emission when approaching the probe
wavelength suggests that a maximum falls into this spectral
region. The inevitable blocking of the scattered probe laser light
prevents a determination of this maximum. Except for a weak
shoulder at 850 nm (Figure 3b), which decays in some 100 fs,
the Stokes spectrum virtually does not change its shape with
time. This applies to the 1 ps time scale on which the vibrational
cooling dynamics is expected to occur (Figure 3a) and to the
10-100 ps time scale on which the recovery of ground-state
triiodide takes place (Figure 3 parts b and c).
Results
Fluorescence Spectrum and Quantum Yield of I3-. Al-
-
though I3 has been thoroughly characterized by the usual
spectroscopic methods, to our knowledge, no fluorescence
spectrum of this ion has been published. This spectrum and the
fluorescence quantum yield allow to obtain a rough estimate of
the excited-state lifetime of I3-. The spectrum was recorded
with the same setup as used for the time-resolved measurements
except for the blocked probe beam and a reduced sample
concentration of 2.7 mM. Inner filter effects on the emission
spectrum could be excluded by comparing spectra of samples
The Stokes signal rises instantaneously, i.e., within the
temporal resolution of our setup (see integrated Stokes signal
in Figure 4). The signal then falls off slightly to a minimum
around 700 fs and, after reaching a second maximum around 2
ps, decays on the 10 ps time scale. The time dependence of the
Stokes signal strongly resembles that of the transient absorption
recorded in the same spectral region (Figure 4, open circles).

1650 J. Phys. Chem. A, Vol. 106, No. 9, 2002
Gilch et al.
Figure 4. Time dependence of the integrated Stokes part (850-990
nm) of the induced emission depicted in and of the transient absorption
recorded at 825 nm.
Figure 5. Time dependence of the anti-Stokes part of the induced
emission spectra. The time traces were obtained by integrating over
14 nm broad ranges of the spectra given in Figure 3. The center
wavelengths of these portions are given in the legend. The solid lines
describe the results of a fit as described in the text.
absorption data is accompanied by a change in the Stokes
emission spectrum (the decay of the 850 nm shoulder) suggest-
ing that during that decay a change of the electronic character
of the emitter occurs. Concerning the ensuing slight rise of the
absorption from 700 fs until 2 ps and the subsequent decay the
Stokes emission give no further information, contrary to the
anti-Stokes emission to which we will turn now.
The Anti-Stokes Part of the Emission. The rise of the anti-
Stokes part of the induced emission spectrum is delayed with
respect to the instantaneous rise of the Stokes part or equiva-
lently the transient absorption (Figure 3b). The time dependence
of the anti-Stokes emission for some wavelengths is compared
with the absorption data in Figure 5. The delay of several 100
fs for the rise of the anti-Stokes emission is clearly visible. In
addition, a negative signal at time zero is observed. This negative
signal consists of a broad and weak contribution extending
throughout the anti-Stokes part and a stronger, sharp feature at
660 nm. A broad negative signal indicates a reduction of the
I3- fluorescence emission induced by the probe pulse. A possible
mechanism for such a reduction works as follows: during the
lifetime of I3-*, the probe pulse can promote excited triiodide
to higher lying states Sn (see Figure 1). This process depletes
the S1 state and concomitantly reduces the S1 f S0 emission
Figure 3. Induced emission spectra of triiodide as a function of delay
time. The pump wavelength was set to 407 nm, whereas the probe
wavelength was tuned to 815 nm. The induced emission spectra were
obtained by subtracting the emission spectrum at large negative delay
times from the spectrum at the given delay time (for the absolute
emission spectra, see inset in a). (a) Decay of the induced emission
spectra (1.1-13.9 ps). (b) Rise of the induced emission spectra (0-
1.1 ps). (c) Long-term evolution of the Stokes part of the induced
emission spectrum. For the sake of comparison, both spectra have been
scaled to the same height.
This absorption change has been assigned to the excited-state
-
absorption of I3 at early delay times and to the ground-state
absorption of I2-.4 On the basis of the absorption data alone, it
-
is difficult to pinpoint the decay of I3-* and the rise of I2
.
Our results show that the decay of the first spike of the transient

Photolysis of Triiodide
J. Phys. Chem. A, Vol. 106, No. 9, 2002 1651
which in turn should cause a negative signal following the
-
emission spectrum of I3
.
The sharp spectral feature around 660 nm with a width of 15
nm (not shown) which is not present in the emission spectrum
of triiodide must have a different origin. This wavelength of
660 nm corresponds to a shift relative to the probe pulse of
∼2900 cm-1. The shift coincides with the most intense Raman
line of the ethanol solvent.25 We, therefore, attribute this
reduction to an inverse Raman scattering.26 In a stimulated
Raman process, fluorescence photons of I3- with a wavelength
of ∼660 nm are annihilated, whereas simultaneously, photons
with the wavelength of the probe pulse (815 nm) are created.
Concomitant with this wavelength shift, the 2900 cm-1 vibration
of ethanol is excited. This inverse Raman process can only occur
when the fluorescence light and the probe pulse overlap in time;
in other words, the negative signal follows the fluorescence
-
lifetime of I3 convoluted with the experimental response
function. This allows us to determine the fluorescence lifetime
by fitting the spectrally integrated inverse Raman signal with
an exponential convoluted with a Gaussian (fwhm 300 fs, see
the Experimental Section) representing the experimental re-
sponse function. We thereby determine the fluorescence lifetime
to be 240 fs. It is worth noting that the temporal width of the
inverse Raman signal coincides with the spike seen in the
absorption measurement. The fluorescence lifetime of 240 fs is
shorter than the rise time of the anti-Stokes emission (Figure
5). A numerical analysis shows that this rise time increases from
500 to 700 fs when moving from the blue edge (680 nm) to the
red edge (780 nm) of the anti-Stokes spectrum. At a first glance,
Figure 5 seems to indicate the contrary. This apparent contradic-
tion is due to the fact that the amplitudes of negative contribution
are large at shorter wavelengths.
The decay of the anti-Stokes emission occurs on the time
scale of ∼2 ps. The decay time depends on wavelength and
increases when moving to the red. For instance, at the blue edge
of the anti-Stokes spectrum, the decay time is 1.4 ps, whereas
it is 2.3 ps at the red edge. This observation is equivalent to a
red-shift of the anti-Stokes spectrum with time, which is clearly
seen in Figure 3a.
Figure 6. Schematics of possible potential energy surfaces relevant
to the interpretation of the induced emission spectra. (a) Lower and
upper surfaces are bound. In the excited-state, vibrational relaxation
occurs prior to fluorescence emission. The emission quantum yield is
large, and the emission spectrum is not influenced by the vibrational
excitation. (b) The lower surface is bound, and the upper surface is
repulsive. Fluorescence light is predominately emitted prior to vibra-
tional relaxation and contains information on the vibrational excitation
E
Exc. The emission quantum yield is small. (c) Both surfaces are
repulsive.
vibrational cooling is terminated, Phot will be promoted to an
electronically excited state, which will again be vibrationally
excited even if the probe pulse is tuned to the 0-0 transition
of the product state (see Figure 6). It is the emission of this
state which is detected in our difference emission measurements.
The information content of the spectrally resolved P* f P
emission depends crucially on the shape of the electronic
potential surfaces involved in the transition. We consider the
following three classes: (i) The surfaces of P and P* are bound
(Figure 6a). The hot molecule Phot is transferred to the P* state
where vibrational relaxation occurs. Fluorescence emission
predominately originates from the vibrational ground state of
P* because vibrational relaxation in polar solvents typically
occurs on the picosecond time scale,27 whereas radiative
lifetimes are much longer.24 For the time integrated fluorescence
detection of our setup, the signal would be dominated by the
fluorescence from the vibrational ground state of P*. The only
information on the vibrational cooling of Phot would be due to
changes of the absorption of Phot in the course of the cooling
process. The shape of the emission spectrum would not be
influenced, and the experiment should yield information similar
to transient absorption spectroscopy. (ii) The P surface is bound,
and the P* surface is repulsive. A hot Phot molecule promoted
to the P* surface by the probe beam starts to dissociate
immediately. This reaction is accompanied by a drastic decrease
of Franck-Condon overlap with the ground state.6 Therefore,
fluorescence emission occurs prior to significant motion on the
P* surface and has a very small quantum yield. In other words,
the initial (vibrational) state of the emission is identical to the
final state reached in the absorption process of the probe pulse.
A molecule Phot with an excess energy of Eexc, excited by a
probe pulse of the energy hνprobe can, therefore, emit fluores-
cence light up to a frequency of (Eexc/h) + Vprobe; the emission
spectrum will contain a Stokes and an anti-Stokes part (see
Figure 6b). The anti-Stokes fluorescence is a direct indicator
of the amount of vibrational excitation; as Phot cools the blue
Discussion
As pointed out in the Introduction, there is an uncertainty
about the photoproducts of the 400 nm photolysis of triiodide.
Because two products, diiodide and the X intermediate, dis-
cussed in the literature, absorb at our probe wavelength of 815
nm,9 it is crucial for the interpretation of our data to clarify
whether the X intermediate is also formed under our experi-
mental conditions. Ruhman et al.9 determined the lifetime of
the X intermediate to be 45 ps at room temperature. The Stokes
emission spectrum recorded at such long delay times (tD ) 180
ps) has virtually the identical shape as the spectrum recorded
∼10 ps after excitation, i.e., before the decay of the X
intermediate (Figure 3c). So either the X intermediate is not
formed in the 400 nm photolysis of triiodide or this intermediate
is not detectable with our method. In either case, if X is not
formed here or our method is insensitive to X, our results can
be interpreted assuming that diiodide and iodine are the only
photoproducts (eq 1).
We will now turn to the question under which circumstances
our pump-probe emission experiment can yield additional
information on the vibrational excitation. Assume that a pump
pulse prepares a vibrationally hot product state Phot via electronic
excitation of a reactant and subsequent ultrafast chemical
reaction. When the probe pulse interacts with Phot before the

1652 J. Phys. Chem. A, Vol. 106, No. 9, 2002
Gilch et al.
edge of the anti-Stokes spectrum moves to the red. Of course,
in a quantitative analysis of this excitation, the Franck-Condon
factors have to be taken into account. In short, the vibrational
excitation can be probed by an emission technique without
gating because the repulsive nature of the upper electronic state
provides an “internal” gating. (iii) Both states are repulsive
(Figure 6c). Here the appearance of an anti-Stokes emission
depends on the relative slope of the two surfaces. An anti-Stokes
signal can only be expected if the slope of the upper state surface
is more shallow than that of the lower.
values for the vibrational excitation than the one discussed above
are required if the vibrational energy is not localized in
symmetric stretch vibration. An involvement of the antisym-
metric stretch, the two degenerate bending vibrations, and the
modes of the surrounding solvent molecules increase this value
considerably. Second, the transient absorption spectrum recorded
by Ku¨hne et al.4 1.5 ps after excitation strongly resembles the
spectrum of diiodide and shows no contributions of a hot
triiodide ground state absorption. Third, by spectrally integrating
this absorption band, they came to the conclusion that geminate
recombination occurs on a time scale of some picoseconds or
longer,4 whereas the rise time of the anti-Stokes signal is well
below 1 ps. We, therefore, at the present stage of analysis, do
not consider contributions of hot triiodide to the anti-Stokes
signal and assign this signal to hot diiodide.
The three molecular species I3-, I3-*, and I2- present in our
pump-probe experiment on I3- are now classified accordingly.
Triiodide in its ground-state belongs to class ii because its
excited states are repulsive as the numerous transient absorption
studies discussed in the Introduction demonstrate. The first
excited state S1 of triiodide can be promoted to a higher lying
excited state SN by the 815 nm probe pulse. Energetically, this
SN state could be identical with the excited state directly
accessible by 270 nm excitation (see the absorption spectrum
in Figure 1) which is known to be repulsive (see above). In
this case, the probe pulse would induce a transition between
two repulsive surfaces; that is, I3-* belongs to class iii. Diiodide
as a stable molecular ion has a bound ground state. Its excited
states are known to be repulsive or weakly bound.28,3 There
The decay time of I3-* was determined here to be 240 fs
employing the inverse Raman effect. This is consistent with
-
the estimate based on the fluorescence quantum yield. The I3
*
decay time is shorter than the rise time of the anti-Stokes signal
which spans the range of 0.5-0.7 ps. We associate this rise
time with the formation of diiodide which as a class ii entity
can show an anti-Stokes emission if it is vibrationally hot. The
observation that the decay time of I3-* emission is shorter than
-
the formation time of I2 is in apparent contradiction to the
-
are experimental indications that when I2 is photoexcited to
simple reaction scheme given in eq 1. However, decay and rise
times of a one step reaction are only equal for rate-type
(exponential) processes. The experimental observation of vi-
brational wave packets even in the diiodide products2,4 gives
clear evidence that the dissociation of I3-* should not be viewed
as an exponential process but rather as a motion of a vibrational
wave packet on a repulsive surface. Because the Franck-
Condon factors vary strongly during such a motion, the time
constant of 240 fs may be related to motion out of Franck-
Condon region of the I3-* surface. The appearance of the anti-
Stokes emission then reflects the formation of hot I2-. The dip
in the transient absorption data depicted in Figure 4 supports
this interpretation. The rise times range from 0.5 to 0.7 ps
depending on the wavelength. This time scale compares well
with the appearance of the diiodide vibrational wave packet
observed in transient absorption spectroscopy13,11 which is one
indicator for the formation time of diiodide. A similar time is
obtained by quantum dynamics and molecular dynamics cal-
culations using empirical LEPS surfaces.11,31,32 There are,
however, other experimental results which indicate that nascent
diiodide still changes its electronic character after that formation
time of ∼400 fs. The transient absorption signal slightly
increases until approximately 2 ps after the photolysis pulse.4,11
In addition, the initial anisotropy of diiodide as measured in a
photolysis-pump-probe experiment decays from 0.7 to 0.4
within the same time.17 As the common origin of these signals,
the authors propose a change in the electronic structure of
diiodide during the separation of the diiodide and the iodine
atom. In other words, it takes about 2 ps until “free” diiodide
is formed. The rise of anti-Stokes does not reflect this change
of electronic structure. The formation time of diiodide deter-
mined here is very close to the appearance time of vibrational
wave packet motion seen in transient absorption experiments
(see above).
2
its repulsive excited states the bound state Π3/2,g is populated
via recombination of the photofragments29,16 (the population of
a bound state is, however, questioned in ref 30). The question
arises how this bound state influences the emission pattern
observed here. Population of this state is only expected to occur
via excitation of diiodide by the probe pulse; there are no
indications that excited diiodide is formed during the 400 nm
photolysis of triiodide.11 The 815 nm probe pulse places the
diiodide vibrational wave packet well above the dissociation
threshold of the 2Π3/2,g state (see the potential curves in ref 28)
even if I2- is vibrationally cold. Diiodide promoted to this state
by the probe pulse will first dissociate and then because of
caging partially recombine.16 Emission from recombined excited
diiodide will be strongly red-shifted with respect to the probe
pulse (the energy difference between the minima of the 2Π3/2,g
and the ground state corresponds to a wavelength of 1250 nm).
Therefore, the bound excited state should not contribute to the
anti-Stokes emission, and diiodide can be assigned to class ii.
Directly after photoexcitation the only transient species is
I3-*. Experimentally, we observe only a weak anti-Stokes
-
emission during the first 100 fs. This is expected because I3
*
as a class iii species should not give rise to a large anti-Stokes
signal even if I3-* contained a large amount of excess energy.
In addition, the excitation at 407 nm takes place in the red-
-
wing of the lowest absorption band of I3 depositing only a
small amount of excess energy. It is only after the decay of
I3-* that a large amount of excess energy is stored in the
vibrational degrees of freedom of the molecular system which
could either be the mode of nascent diiodide or the modes of
geminately (hot) recombined triiodide. Concerning the contribu-
tion of the hot triiodide to the anti-Stokes signal, an estimate
based on the empirical potential energy surfaces of triiodide as
first parametrized by Benjamin et al.31 indicates that it requires
a vibrational excitation of g6000 cm-1 to shift its ground-state
absorption from 370 nm to the 815 nm probe wavelength. In
principle, such an amount of excess energy is available upon
recombination of diiodide and iodine (see above). However,
there are several arguments which render a contribution of hot
triiodide to the anti-Stokes signal rather unlikely. First, larger
There is, however, no doubt that nascent diiodide is vibra-
tionally hot. The degree of vibrational excitation of nascent
diiodide is imprinted in the anti-Stokes part of the emission
spectra presented here. A larger anti-Stokes shift monitors a
higher vibrational state. The rise and decay times determined
by fitting portions of the anti-Stokes emission, therefore, give

Photolysis of Triiodide
J. Phys. Chem. A, Vol. 106, No. 9, 2002 1653
an estimation of the population dynamics of segments of the
vibrational ladder of diiodide. In the instant of formation,
diiodide should have its highest vibrational excitation. Indeed,
the rise time at the blue edge (0.5 ps) of the anti-Stokes signal
is shorter than at the red edge (0.7 ps). Not only does the red
rise time reflect the formation dynamics of diiodide but also
the population transfer from higher vibrationally excited states
to lower ones. The maximum of the anti-Stokes emission is
reached ∼1 ps after photoexcitation. At that time, the anti-Stokes
spectrum ranges roughly from -2800 cm-1 (relative to central
frequency of the probe laser) to 0 cm-1. As the vibrational
Formation of nascent diiodide was found to occur within 0.5
ps. The emission of excited triiodide decays in 0.24 ps. The
difference between these two values is related to the nonexpo-
nential wave packet type character of the dissociation. The
amount of the vibrational excitation of nascent diiodide could
be estimated from the anti-Stokes part of the fluorescence
spectrum. The peak energy stored in the vibrational mode is
∼1400 cm-1 at 1 ps. The vibrational excitation decays on the
2 ps time scale.
Acknowledgment. We thank Prof. M. E. Michel-Beyerle
from the TU Mu¨nchen for the transient absorption data recorded
in her lab.
frequency of diiodide is 114 cm-1 33 the 2800 cm-1 shift
,
corresponds to a vibrational excitation of ∼24 quanta. The center
of the anti-Stokes spectrum at 1 ps is ∼1400 cm-1 giving a
rough estimation of the vibrational excitation at that time. The
anti-Stokes spectrum then decays and shifts to the red on a time
scale of 2 ps. In light of the results on the change of electronic
structure of nascent diiodide on the same time scale17 discussed
above, a possible contribution of these changes on the anti-
Stokes emission has to be taken into consideration. Such changes
might go along with variation of the energy gap and thereby
the excess energy available for the anti-Stokes emission. We
consider such a contribution as not very likely for the following
two reasons. First, the rise of anti-Stokes emission is faster than
the decay of initial anisotropy which was taken as the indicator
of electronic changes (see above). If the rise is insensitive to
this change, it is unlikely that the decay is. Second, the transient
absorption signal at the probe wavelength changes only very
weakly during the decay of the anti-Stokes signal (see Figure
5). If there were a large change in the energy gap, a more
pronounced change of this signal should occur. As a conse-
quence, the analysis of the anti-Stokes decay is based on the
assumption of a constant energy gap. The decay time of the
anti-Stokes signal increases with decreasing anti-Stokes shift
as expected for a vibrational cooling process. For instance, the
time constant is 1.4 ps for a shift of 2400 cm-1, whereas it is
2.3 ps for 400 cm-1. A characteristic time for the cooling is on
the order of ∼2 ps which is close to the somewhat longer time
constant of 3.5 ps determined by transient absorption spectros-
copy.4 However, the time constant of ref 4 is supposed to have
a small amplitude of only 250 cm-1 of vibrational excitation,
whereas our results strongly suggest that the picosecond cooling
time constant has a much larger amplitude. In the transient
absorption measurement, the dominant amplitude was reported
to be 1400 cm-1 with a decay time of 400 fs. Such a contribution
is clearly missing in our data.
References and Notes
(1) Dantus, M.; Rosker, M. J.; Zewail, A. H. J. Chem. Phys. 1987, 87,
2395.
(2) Banin, U.; Waldman, A.; Ruhman, S. J. Chem. Phys. 1992, 96,
2416.
(3) Kliner, D. A. V.; Alfano, J. C.; Barbara, P. F. J. Chem. Phys. 1993,
98, 5375.
(4) Ku¨hne, T.; Vo¨hringer, P. J. Chem. Phys. 1996, 105, 10788.
(5) Pugliano, N.; Szarka, A.Z.; Hochstrasser, R. M. J. Chem. Phys.
1996, 104, 5062.
(6) Herzberg, G. Molecular Spectra and Molecular Structure; Krieger
Publishing Company: Malabar, Florida, 1950.
(7) Johnson, A. E.; Myers, A. J. Chem. Phys. 1995, 102, 3519.
(8) Johnson, A. E.; Myers, A. J. Phys. Chem. 1996, 100, 7778.
(9) Gershgoren, E.; Banin, U.; Ruhman, S. J. Phys. Chem. A 1998,
102, 9.
(10) Wang, Z.; Wasserman, T.; Gershgoren, E.; Vala, J.; Kosloff, R.;
Ruhman, S. Chem. Phys. Lett. 1999, 313, 155.
(11) Ku¨hne, T.; Ku¨ster, R.; Vo¨hringer, P. Chem. Phys. 1998, 233, 161.
(12) Murov, S. L., Carmichael, I., Hug, G. L., Eds. Handbook of
Photochemistry; Marcel Dekker: New York, 1993.
(13) Banin, U.; Kosloff, R.; Ruhman, S. Isr. J. Chem. 1993, 33, 141.
(14) Banin, U.; Kosloff, R.; Ruhman, S. Chem. Phys. 1994, 183, 289.
(15) Alfano, J. C.; Kimura, Y.; Walhout, P. K.; Barbara, P. F. Chem.
Phys. 1993, 175, 147.
(16) Walhout, P. K.; Alfano, J. C.; Thakur, K. A. M.; Barbara, P. F. J.
Phys. Chem. 1995, 99, 7568.
(17) Hess, S.; Bu¨rsing, H.; Vo¨hringer, P. J. Chem. Phys. 1999, 111,
5461.
(18) Hartl, I.; Zinth, W. Opt. Comm. 1999, 160, 184.
(19) Hartl, I.; Gilch, P.; Zinth, W. Appl. Phys. B 2000, 71, 397.
(20) An, Q.; Zinth, W.; Gilch, P. Opt. Comm. 2002, in press.
(21) Gilch, P.; Po¨llinger-Dammer, F.; Steiner, U. E.; Michel-Beyerle,
M. E. Chem. Phys. Lett. 1997, 275, 339.
(22) Jones, G., II.; Jackson, W. R.; Choi, C.-Y. J. Phys. Chem. 1985,
89, 294.
(23) Hirata, Y.; Tanaka, I. Chem. Phys. Lett. 1976, 43, 568.
(24) Strickler, S. J.; Berg, R. A. J. Chem. Phys. 1962, 37, 814.
(25) Schrader, B., Meier, W., Eds. Raman/IR Atlas of Organic Com-
pounds; Verlag Chemie: Weinheim, Germany, 1975.
(26) McLaren, R. A.; Stoicheff, B. P. Appl. Phys. Lett. 1970, 16, 140.
(27) Laubereau, A.; Kaiser, W. ReV. Mod. Phys. 1978, 50, 607.
(28) Chen, E. C.; Wentworth, W. E. J. Phys. Chem. 1985, 89, 4099.
(29) Papanikolas, J. M.; Vorsa, V.; Nadal, M. E.; Campagnola, P. J.;
Gord, J. R.; Lineberger, W. C. J. Chem. Phys. 1992, 97, 7002.
(30) Parson, R.; Faeder, J.; Delaney, N. J. Phys. Chem. A 2000, 104,
9653.
Conclusions
A new technique is used for the study of the photoreaction
of triiodide. Monitoring the induced emission in a pump-probe
experiment gives additional information on product dynamics
as compared to conventional pump-probe absorption spectros-
copy. The transition of excited triiodide into diiodide which is
accompanied by only minor changes in the transient absorption
spectrum is clearly visible in the induced emission spectrum.
(31) Benjamin, I.; Banin, U.; Ruhman, S. J. Chem. Phys. 1993, 98, 8337.
(32) Ashkenazi, G.; Kosloff, R.; Ruhman, S.; Tal-Ezer, H. J. Chem.
Phys. 1995, 103, 10005.
(33) Tripathi, G. N. R.; Schuler, R. H.; Fessenden, R. W. Chem. Phys.
Lett. 1985, 113, 563.