Electron injection dynamics of RuII(dcbpy)2(SCN)2 on zirconia

Article abstract of DOI:10.1021/jp014160o

The electron injection dynamics of Ru^II(4,4a?2-dicarboxy-2,2a?2-bipyridine)₂(SCN) ₂ (the so-called N3-dye) adsorbed on colloidal ZrO₂ particles in room-temperature solution have been studied using time-resolved absorption polarization spectroscopy. The solvents used are acetonitrile and methanol. Electron injection is evidenced by the transient absorption anisotropy and/or the wavelength-dependent unpolarized absorption intensity. The results indicate that 630 nm excitation results in no N3* to ZrO₂ electron transfer in both acetonitrile and methanol solvents, 515 nm excitation results in electron injection in acetonitrile, and 495 nm excitation results in electron injection in both acetonitrile and methanol. These results may be understood in terms of simple energetic considerations: 630 nm excitation produces N3* with insufficient energy for electron injection while 495 nm excitation produces N3* with an energy just sufficient for electron injection. In all cases where electron injection is energetically favorable, it occurs very rapidly, <200 fs. There is no indication of rapid (< 5 ps) electron injection into deep trap states on the ZrO₂ surface; however, the possibility of rapid injection into shallow trap states cannot be excluded. The observed kinetics also indicate that, following photoexcitation of the adsorbed N3, the ZrO₂ surface relaxes on the 5-10 ps time scale.

Full text of DOI:10.1021/jp014160o

J. Phys. Chem. B 2002, 106, 6211-6219
6211
II
Electron Injection Dynamics of Ru (dcbpy)2(SCN)2 on Zirconia
C. M. Olsen, M. R. Waterland, and D. F. Kelley*
Department of Chemistry, Kansas State UniVersity, Manhattan Kansas 66506-3701
ReceiVed: NoVember 12, 2001; In Final Form: April 8, 2002
II
The electron injection dynamics of Ru (4,4′-dicarboxy-2,2′-bipyridine)
2
(SCN)
2
(the so-called N3-dye) adsorbed
2
on colloidal ZrO particles in room-temperature solution have been studied using time-resolved absorption
polarization spectroscopy. The solvents used are acetonitrile and methanol. Electron injection is evidenced
by the transient absorption anisotropy and/or the wavelength-dependent unpolarized absorption intensity. The
2
results indicate that 630 nm excitation results in no N3* to ZrO electron transfer in both acetonitrile and
methanol solvents, 515 nm excitation results in electron injection in acetonitrile, and 495 nm excitation results
in electron injection in both acetonitrile and methanol. These results may be understood in terms of simple
energetic considerations: 630 nm excitation produces N3* with insufficient energy for electron injection
while 495 nm excitation produces N3* with an energy just sufficient for electron injection. In all cases where
electron injection is energetically favorable, it occurs very rapidly, <200 fs. There is no indication of rapid
(
<5 ps) electron injection into deep trap states on the ZrO
into shallow trap states cannot be excluded. The observed kinetics also indicate that, following photoexcitation
of the adsorbed N3, the ZrO surface relaxes on the 5-10 ps time scale.
2
surface; however, the possibility of rapid injection
2
Introduction
Electron transfer can also occur from sensitizer dyes to
semiconductor surface states. In cases of large band gap
semiconductors, the conduction band is energetically inacces-
sible and electron transfer can occur only to localized surface
states. Despite the inaccessibility of the conduction band,
There has recently been great interest in interfacial electron
transfer between sensitizer dyes and semiconductor substrates.
Much of this interest has been focused on ruthenium based
sensitizer dyes and nanoporous TiO2, because of the application
in photovoltaic cells.1 The dye which has received the most
1
5
electron injection can occur very efficiently. This type of
electron transfer reaction may be quite sensitive to the sur-
rounding solvent for several reasons. An electron in a surface
state can be stabilized by the local polar solvent environment.
As a result, one might expect this type of electron transfer to
be more susceptible to solvent effects than an electron transfer
directly into the semiconductor conduction band. The nature of
the solvent can also affect the semiconductor surface chemistry,
and this can affect the dynamics of electron transfer from
adsorbates to semiconductor surface states. The solvent can
affect the depth and density of surface trap states and thus the
energetics of interfacial electron transfer. Furthermore, the nature
of the solvent can also affect adsorbate/surface binding, which
can alter the donor/acceptor electronic coupling.
-3
II
attention is the so-called “N3 dye”, Ru (dcbpy)2(NCS)2, where
4
dcbpy ) 4,4′-dicarboxy-2,2′-bipyridine. Numerous studies have
measured the electron injection rates from photoexcited N3 into
3,5-13
nanoporous TiO2 and other types of semiconductor surfaces.
These studies have shown that electron injection into TiO2 is
very fast, with most or all of the injection occurring within 100
fs. While many electron transfer rates are strongly dependent
on the nature of the surrounding solvent, the above result is
solvent independent. Specifically, fast electron injection is
observed for both dry TiO2 surfaces and TiO2 surfaces immersed
3
,7
in ethanol. This is because the dense manifold of conduction
band states is strongly coupled to the photoexcited N3 dye and
electron transfer is energetically favorable, independent of the
local solvent environment. Molecules that are similar to N3,
In addition to studies of electron transfer from relaxed
sensitizers, there has been considerable interest in the possibility
of electron injection from vibrationally unrelaxed states of the
except having less coupling to the surface, exhibit slower
electron injection.14 Electron injection rates from photoexcited
3,13,16-19
photoexcited dye.
In many cases, electron injection has
N3 into ZnO and SnO2 surfaces have been measured to be much
3,9,10
been shown to occur on a time scale comparable to or faster
than vibrational relaxation. Thus, it is a reasonable speculation
that a significant (and perhaps dominant) path for electron
injection is from the initially excited Franck-Condon state of
the dye, rather than from a relaxed state. The electron injection
dynamics of several different ruthenium-type dyes having
different excited state oxidation potentials have been studied,
and ultrafast electron injection has been shown to occur in
slower than those into TiO2 and nonsingle exponential.
This
result is explained in terms of the nature of the orbitals which
constitute the semiconductor conduction band. In the TiO2 case,
the orbitals comprising the conduction band have considerable
titanium d-orbital character. It is suggested that these orbitals
couple more strongly to the N3 π*-donor orbitals than do the
metal s-orbitals of ZnO or SnO2. This point has recently been
5
disputed, so the role of the orbital character in determining
3
several different cases. Dyes having oxidation potentials below
the extent of coupling is still somewhat unclear.
the TiO2 conduction band were excited to high levels from
which electron injection could occur in competition with
vibrational relaxation. The results indicate that relaxation and
*
To whom correspondence should be addressed. E-mail: dfkelley@
ksu.edu.
1
0.1021/jp014160o CCC: $22.00 © 2002 American Chemical Society
Published on Web 05/25/2002

6
212 J. Phys. Chem. B, Vol. 106, No. 24, 2002
Olsen et al.
electron injection occur on comparable time scales and that some
In all cases, solvents were thoroughly dried and purified prior
to use and stored under dry nitrogen. Methanol was Spectral
grade and distilled over iodine-activated magnesium in a
nitrogen atmosphere. Acetonitrile was HPLC grade and distilled
over anhydrous P2O5 in a nitrogen atmosphere. N3 dye was
obtained from Dr. S. Ferrere at NREL and used without further
purification. Samples were held in 1 cm path length sealed cells,
and rapidly stirred with a magnetic stir bar. In all cases, samples
were filtered with 0.22 µm Millipore filters and degassed
immediately before being sealed in the optical cells. N3 dye
has labile thiocyanate ligands that are easily hydrolyzed, and
we have found that the above purifications are necessary to avoid
hydrolysis of the dye. Over the course of the experiments there
was no detectable change in the sample absorption spectrum,
and we conclude that the above procedures result in N3/ZrO2/
acetonitrile samples that are sufficiently dry to prevent N3
hydrolysis. We note, however, that while the above procedure
removes water that is not adsorbed onto the particle surface,
the particles are never heated under vacuum and therefore retain
much of the adsorbed water.
(but not all) of the excited state population undergoes electron
injection prior to relaxation. Although excitation wavelength
dependent studies have not been performed, excitation to the
lowest vibrational levels of the dye would have presumably
resulted in little or no electron injection. A similar conclusion
1
3
results from visible/near-IR time-resolved N3/TiO2 studies.
In this paper we examine the roles of excitation wavelength
and the solvent on the dynamics of N3*/ZrO2 interfacial electron
transfer. This is done by observing the relaxation and electron
injection dynamics of photoexcited N3 absorbed on colloidal
ZrO2 particles. The N3/ZrO2 particles are dissolved in either
methanol or acetonitrile solvents at room temperature, and the
spectra and dynamics are compared to those obtained for N3
in the neat solvent, lacking the ZrO2 particles. The dynamics
are observed to be critically dependent on the excitation
wavelength. We also observe somewhat different electron
transfer dynamics in the two solvents, despite using the same
ZrO2 particles in both solvents. The solvent dependent absorp-
tion spectra of N3 adsorbed on ZrO2 particles show that the
N3/ZrO2 electronic coupling is very strong in both solvents,
although slightly stronger in acetonitrile.
Time-resolved absorption measurements were made using the
femtosecond absorption spectrometer described in a previous
2
1
publication. Briefly, the femtosecond light source is based on
a Clark-MXR 2001. This produces 775 nm, 130 fs, 800 µJ
pulses at a repetition rate of 1 kHz. About 4% of the pulse
intensity is split off, attenuated, and either used for the sample
probe or used to generate white light continuum. In the latter
case, wavelength selection is accomplished using 10 nm band-
pass interference filters. The remainder of the 775 nm beam is
used to pump an OPA (Clark-MXR, vis-OPA). The output of
the OPA is approximately 10 µJ fs pulses, tunable throughout
the visible region of the spectrum. This tunable output is used
for sample excitation; specifically, results following 495, 515,
and 630 nm excitation are presented here. The pump beam is
typically focused to a spot size of 0.5-1.0 mm at the sample.
The power density can be varied by changing the position of
the sample with respect to the focal point of the pump beam.
In the results presented here, variation of the power density by
a factor of about 10 has no detectable effect on the observed
kinetics.
The low intensity probe beam is split into reference and
sample probe components. The intensity of the probe pulses is
less than 1 µJ at the sample. After the sample, the probe beam
is split into horizontal and vertical polarization components.
These beams and the I0 beam are imaged onto UDT Sensors
PIN 13DI photodiodes, biased at -15 V. The photodiode outputs
are amplified and input into an SRS gated integrator. The gated
integrator output is measured using an Analog Devices 16 bit
analog-to-digital (A/D) converter in the data acquisition com-
puter. The A/D converter and gated integrator triggering and
reset are synchronized with the CPA 2001 Q-switch and
controlled by home-built timing electronics. Data acquisition
is controlled by LabView software running on a Pentium II
computer.
Experimental Section
ZrO2 particles used in these studies are prepared by the acid
15
hydrolysis of ZrCl4, as previously described. Specifically, 2.5
g of ZrCl4 (Alfa) is dissolved in 40 mL of anhydrous methanol.
The solution is cooled to 0 °C, acidified with 0.125 mL of
concentrated HCl and 1.0 mL of distilled H2O, and allowed to
warm to room temperature while being stirred for 2 h. This
solution is then heated to reflux for 2 h. This procedure has
been reported to produce 10-20 nm particles. The solution is
then rotoevaporated to dryness. At this point the particles are
resuspended in either acetonitrile or anhydrous methanol. The
solvent is then evaporated off under vacuum at room temper-
ature. Several cycles of resuspension followed by solvent
evaporation are used to remove as much water as possible. The
particles are finally resuspended in either acetonitrile or
methanol to form ZrO2 stock solutions having concentrations
of approximately 25 mg/mL. This procedure results in a small
amount of particle aggregation. The larger particles result in
light scattering, which is a problem in the optical measurements.
The solutions are therefore filtered through 0.22 µm Millipore
filters prior to the addition of dissolved N3. Filtration removes
the larger particles, resulting in clear solutions. We note that
the acetonitrile and methanol solutions are derived from the same
particles. Each has simply been washed in the appropriate
solvent to remove water. The solutions are never neutralized,
and the particles are thus probably somewhat acidic.
N3/zirconia samples are made by combining equal volumes
of N3/acetonitrile or N3/methanol stock solutions (ca. 50 mM)
with the appropriate ZrO2 stock solutions. It is possible to assess
the fraction of N3 dyes that attach to the zirconia particles, as
opposed to remaining free in solution. This is done from
rotational diffusion measurements. Attachment to the relatively
large zirconia particles is expected to dramatically slow the N3
rotational diffusion rate. The zirconia particles are expected to
Results and Discussion
20
rotationally diffuse on a time scale of many nanoseconds, and
N3 attached to the particle surface will undergo rotational
diffusion on the same time scale. In contrast, N3 in solution
In this section we discuss the use of transient absorption
spectroscopy (polarized and unpolarized) to elucidate the
dynamics of electron injection. This discussion is based on the
known spectroscopic properties of the N3 metal-to-ligand charge
transfer (MLCT) excited state and cation and the geometry of
the N3 molecule. These spectroscopic and geometric consid-
erations are discussed below. This is followed by the use of
these considerations in the analysis of the time-resolved results
21
rotationally diffuses on the hundreds of picoseconds time scale.
The results presented below show no detectable rotational
diffusion on the hundreds of picoseconds time scale, indicating
that the vast majority of the N3 dyes are attached to the zirconia
particle surfaces.

Electron Injection Dynamics of N3 on ZrO2
J. Phys. Chem. B, Vol. 106, No. 24, 2002 6213
Figure 2. Absorption spectra of N3 in methanol solution (dashed curve)
Figure 1. Absorption spectra of N3 in acetonitrile solution (dashed
and N3/ZrO
2
in methanol solution (solid curve). The lowest energy
2
curve) and N3/ZrO in acetonitrile solution (solid curve). The lowest
absorption maximum of each spectrum is indicated. Also shown is an
energy absorption maximum of each spectrum is indicated. Also shown
extrapolation of the red edge of the N3/ZrO
absorbance, 730 nm.
2
absorption to zero
2
is an extrapolation of the red edge of the N3/ZrO absorption to zero
absorbance, 778 nm.
removes this π* electron. Despite the fact that in both cases
the ruthenium is in the 3+ state, there are significant spectral
differences between the N3 MLCT excited state and the N3
cation. The spectra of both species in ethanol solution have been
on the N3/ZrO2/solvent systems. The central dynamical question
about the N3/zirconia systems is whether electron transfer occurs
from the MLCT excited state to the zirconia particles and, if
so, what the electron transfer rates are. There are several ways
that this question may be addressed. Most commonly, these
dynamics are addressed through the use of time-resolved infrared
spectroscopy. Alternatively, the electron transfer as well as the
intramolecular relaxation dynamics of the adsorbed dye may
be addressed by the use of time-resolved absorption polarization
spectroscopy. The analysis and interpretation of these results is
definitive and straightforward but not immediately obvious. We
therefore analyze and discuss the probe wavelength dependent
absorption intensity and polarization differences between the
N3 MLCT excited state and cation, below. From this analysis
and the comparison of the transient spectra obtained in a pure
solution (no zirconia) and adsorbed onto solution phase zirconia
particles, the electron transfer dynamics are elucidated.
6
,7,13,17,23
reported and carefully compared.
In both cases there
is a strong absorption in the red to near-infrared spectral region,
3
+
which is assigned to a thiocyanate to Ru ligand-to-metal
charge transfer (LMCT) transition. The presence of the electron
in the dcbpy π* orbital (i.e., the MLCT state) results in a
considerable shift in this absorption compared to that obtained
for the cation. The transient absorption spectrum of the MLCT
excited state exhibits an absorption maximum at 720 nm. The
N3 cation exhibits a broader, somewhat weaker absorption that
peaks at about 800 nm. The two species have the same
absorption intensity (i.e., the same extinction coefficients) at
about 780 nm. Thus, whereas the MLCT excited state and cation
have comparable absorbances at 780 nm, the MLCT state
exhibits a larger absorbance at bluer wavelengths. The N3
ground state also has significant absorbance at wavelengths
below 700 nm. Excitation bleaches this transition and thus
results in an isosbestic point at about 620 nm upon formation
of the MLCT state. Similarly, an isosbestic point is obtained
further to the red, at about 650 nm if the cation is formed. The
spectral differences between the MLCT excited state and the
N3 cation can be used to elucidate electron injection dynamics.
Specifically, the N3 MLCT state exhibits a strong absorption
in the 620-800 nm spectral region, and electron injection
reduces the intensity of this absorption. This is particularly true
in the bluest part of this region, 600-700 nm. The above
+
Unpolarized Spectroscopy of N3* and N3 . The absorption
spectra of N3 in acetonitrile and methanol solutions as well as
adsorbed on ZrO2 particles in these solvents are shown in
Figures 1 and 2. The spectrum of N3 in methanol solution (ca.
-
5
4
5
× 10 M) shows a lowest energy MLCT absorption peak at
30 nm. The spectrum of an analogous sample containing ZrO2
particles (at approximately the same N3 concentration) is also
shown. In this case, the MLCT peak shows a slight (=8 nm)
shift to the red. This is due to the perturbation of the MLCT
transition by the presence of the zirconia surface. The ZrO2
particles absorb below 425 nm, and this is also seen in the N3/
ZrO2 spectrum. Qualitatively similar, but somewhat larger
differences are observed in the case of acetonitrile solvent. In
this case there is an =9 nm red shift of the lowest energy MLCT
peak. This peak also shows considerable broadening when the
N3 is attached to the zirconia. Taken together, these spectra
indicate that the N3 binds to the zirconia, independent of the
nature of the surrounding solvent. They also show a slightly
greater spectral perturbation, indicating slightly stronger elec-
tronic coupling in the acetonitrile case.
+
differences between the N3* and N3 spectra are clearly
observed in time-resolved N3/TiO2 studies examining this
1
3
spectral range.
+
Polarization Spectroscopy of N3* and N3 . The polariza-
tion spectroscopy of the N3 MLCT excited state and cation may
be analyzed in terms of the pseudooctahedral geometry depicted
2
1
in Figure 3. Excitation results in an MLCT state in which an
electron is localized on a single bipyridine ligand, and polarized
excitation therefore photoselects one of the bipyridines. The
MLCT transition is therefore polarized along a vector from the
ruthenium, bisecting one of the bipyridines. The coordinate axes
in Figure 3 are chosen such that the photoselected bipyridine is
between the -x- and +z-axes. The red and near-IR transient
Excitation results in an MLCT excited state in which the
ruthenium is in the 3+ state and an electron is localized in a
π* orbital of one of the 4,4′-dicarboxy-2,2′-bipyridine (dcbpy)
ligands.22 Subsequent electron transfer to the zirconia surface

6214 J. Phys. Chem. B, Vol. 106, No. 24, 2002
Olsen et al.
Figure 3. Geometry of the N3 molecule. The axes are chosen such
that the photoselected bipyridine ligand (indicated with bold italics) is
on the -x- and z-axes. The syn- and anti-thiocyanate ligands are on
the y- and x-axes, respectively.
absorption is due to thiocyanate to ruthenium LMCT transitions
that are polarized along the thiocyanate to ruthenium axes, the
x- and y-axes. Thus, the angle (denoted as θ) between the MLCT
excitation vector and the LMCT vector associated with the syn-
thiocyanate (on the y-axis) is 90°. Similarly, the angle between
the MLCT excitation vector and the LMCT vector associated
with the anti-thiocyanate (on the x-axis) is 135°. Photoselection
theory allows calculation of the time-dependent anisotropy in
terms of the angle between pump and probe transition vectors,
Figure 4. Experimental plots of the 775 nm absorption anisotropy for
N3/ZrO /acetonitrile (solid line), N3/ZrO /methanol (dotted line), and
2
2
N3/acetonitrile (dashed line), following 630 nm excitation.
pendent. The presence of an electron in one of the bipyridine
π* orbitals (the N3 MLCT excited state) perturbs the LMCT
transitions, complicating the above analysis. The presence of
the π* electron shifts the LMCT transition about 80 nm to the
blue. This spectrum is the superposition of the two separate
2
4,25
θ. Specifically
r(t) ) (Apar - Aperp)/(Apar + 2Aperp) )
2/5)P (cos θ) exp(-t/τ ) (1)
(syn and anti) LMCT spectra. Simple electrostatic considerations
suggest that the transition associated with the anti-thiocyanate
should be shifted less than that associated with the syn ligand.
(The reaction field from the solvent stabilizes the anti-LMCT
dipole, but is orthogonal to the syn-LMCT dipole.) Thus, in
the absence of interligand electron transfer or rotational diffu-
sion, the absorption anisotropy is expected to be positive of
-0.05 when probed to the red of 720 nm (the LMCT absorption
maximum) and approaching +0.10 at the furthest red wave-
lengths, at which only the anti-LMCT transition is probed.
Previous studies on N3 in solution have shown that the initial
775 nm probe anisotropy is about +0.04, in agreement with
(
2
rot
where Apar and Aperp are the absorption intensities having
polarizations parallel and perpendicular to that of the pump
pulse, P2(cos θ) is the second Legendre polynomial
2
P (cos θ) ) (1/2)(3 cos θ - 1)
2
θ is the angle between the pump and probe transitions and τrot
is the rotational diffusion time. The denominator in eq 1, Apar
21
this expectation. Interligand electron transfer interchanges the
roles of syn- and anti-LMCT transitions, and following inter-
ligand electron transfer, both are probed equally. Thus, following
interligand electron transfer (but prior to rotational diffusion),
an anisotropy of -0.05 is expected. An important conclusion
may be drawn from the above considerations: A positive
anisotropy at 775 nm indicates the presence of the N3 MLCT
excited state. This anisotropy may become negative as a result
of either cation formation or interligand electron transfer. Since
both interligand electron transfer and cation formation result in
negative anisotropy, no conclusion may be drawn from the
observation of a negative anisotropy.
+
2Aperp, is the total, unpolarized absorption and is independent
of interligand electron transfer or rotational diffusion. The initial
t ) 0) anisotropies of the LMCT transitions associated with
the syn- and anti-thiocyanate ligands are therefore -0.20 and
0.10, respectively. The observed anisotropy will be between
(
+
these extremes, and will depend on the relative intensities of
these LMCT transitions. If the absorption intensities of the two
LMCT transitions are the same, then a net anisotropy of -0.05
is obtained. This is the case in the N3 cation, in which the
thiocyanate ligands are equivalent. Since the two thiocyanate
ligands are not equivalent in the N3 MLCT state, the spectra
of the two LMCT transitions are not expected to be identical
and the observed anisotropy will be probe wavelength dependent
in this case. The rotational diffusion time depends on the
hydrodynamic radius of the chromophore. In the case of N3 in
simple solvents, this is on the order of 100 ps.21 The ZrO2
particles to which the N3 dyes are attached are comparatively
large (10-20 nm), and since the rotational diffusion time is
proportional to the particle volume,20 the ZrO2 particles and
attached dyes rotationally diffuse on a much longer time scale.
A simple conclusion follows from the above considerations:
prior to rotational diffusion, red and near-IR LMCT absorption
of the N3 cation will exhibit a negative anisotropy. In the ideal
case (no interfering transitions, no mixing with other states, no
saturation of the transition), this anisotropy is -0.05. This
negative anisotropy is expected to be probe wavelength inde-
Dynamics of N3 on Zirconia Following 630 nm Excitation.
Figure 4 shows the photoinduced 775 nm transient absorbance
anisotropy obtained following 630 nm excitation of N3 in
acetonitrile and adsorbed onto zirconia particles in acetonitrile.
A similar result, obtained in N3 on zirconia particles in
methanol, is also shown in Figure 4. The N3/methanol (not
shown) and N3/acetonitrile results are virtually identical to those
21
reported in a previous paper. The N3/methanol and N3/
acetonitrile results show an initially positive anisotropy that
subsequently decays as interligand electron transfer and rota-
tional diffusion occur. The anisotropy kinetics obtained for N3/
ZrO2/acetonitrile and N3/ZrO2/methanol show similar positive
anisotropies and no subsequent decay on the 100 ps time scale.
As mentioned above, the lack of any detectable anisotropy decay
due to rotational diffusion indicates that the vast majority of

Electron Injection Dynamics of N3 on ZrO2
J. Phys. Chem. B, Vol. 106, No. 24, 2002 6215
the N3 dyes are attached to the zirconia particles, rather than in
solution. Since either interligand electron transfer or electron
injection results in a negative anisotropy, this result shows that
these processes do not occur on the 100 ps time scale following
spectrum to zero absorbance, i.e., the absorption onset. The onset
of the N3 absorption spectrum in acetonitrile is at about 778
nm (see Figure 1). While the red edge of the absorption spectrum
is not exactly linear with wavelength, this procedure provides
a reasonable estimate of the lowest energy (vibrationally and
solvent relaxed) of the MLCT state. Excitation above that
wavelength puts energy into the vibrational and solvent polar-
ization degrees of freedom, which may facilitate electron
transfer. 630 nm excitation is about 0.37 eV over the absorption
onset. Thus, following 630 nm excitation, the N3 molecule has
an energy that is about 0.43 eV too low to transfer an electron
to the ZrO2 conduction band. It is also possible that there are
6
30 nm excitation. The absence of both interligand electron
transfer and of electron injection are easy to understand, and
are discussed below.
Electron injection does not occur because the ZrO2 conduction
band is energetically inaccessible following 630 nm excitation.
This can be seen from the N3 and ZrO2 electrochemical
potentials, and both potentials must be considered. The oxidation
potential of the relaxed N3 MLCT state is about -0.8 V (vs
4
energetically accessible trap states associated with the ZrO2
SCE). The redox potential of the ZrO2 conduction band is best
surface, into which electron transfer could occur. The results
determined in comparison to the more heavily studied material,
TiO2. The TiO2 conduction band potential follows Nernstian
behavior, and is given (in volts) by -0.400 - 0.06pH, versus
presented here indicate that on the time scale of solvent and
vibrational relaxation little or no electron transfer into these
states occurs. They also indicate that injection into very deep
trap states (0.8 V below the conduction band) does not occur
from relaxed N3 on the hundreds of picoseconds time scale.
This is consistent with spectroelectrochemical results indicating
that most of the surface electron trap states are within 0.1-0.2
2
6
SCE. The ZrO2 conduction band potential is about 1.15 V
27
more negative than that of TiO2. This difference is ap-
proximately pH independent. Thus, in the pH range of 0-2
(appropriate for the acidic conditions under which these particles
are synthesized), the ZrO2 potential is at about -1.55 to -1.65
2
8
V versus SCE.2
2,28
V of the conduction band edge.
Although the above seems quite straight-
forward, the present studies were performed in nonaqueous
solvents and several points about the use of these redox
potentials need to be mentioned. The conduction band potentials
of metal oxide semiconductors are pH dependent and well-
defined only in an aqueous environment. Furthermore, the TiO2
redox potential is known to be strongly dependent on the nature
of the nonaqueous solvent.29 However, the zirconia samples are
prepared through the acidic hydrolysis of ZrCl4, and never
thoroughly dried. As a result, even in acetonitrile solutions there
is probably still a fair amount of water adsorbed on the zirconia
surfaces. (The presence of adsorbed water is also indicated by
the relaxation dynamics of adsorbed N3, discussed below.) Thus,
despite the nonaqueous environment, the above value of the
ZrO2 conduction band potential is a reasonable estimate. Recent
studies on the pH dependence of the energetics of N3/TiO2
electron injection also bear on the energetic considerations
The lack of interligand electron transfer may be understood
in terms of breaking of the symmetry between the two dcbpy
ligands. Attachment of the N3 to a ZrO2 surface is presumably
qualitatively similar to attachment to a TiO2 surface, which has
been studied. Three carboxylates can bind to the surface,
resulting in one of the dcbpy ligands being more strongly bound
than the other. Since binding to the surface shifts the absorption
to the red, we presume that the π* orbitals of the more strongly
bound ligand are lower in energy. 630 nm is on the red edge of
the absorption spectrum, and the MLCT state localized on the
lower energy ligand is preferentially excited. The observed
spectral shifts are large compared to thermal energies, suggesting
that the energy splitting between the two ligands is also large
compared to thermal energies, and we conclude that the weakly
bound ligand is more than kT higher in energy than the initially
populated ligand. Thus, following 630 nm excitation, the π*
electron starts out in the lower energy MLCT state and stays
there; both interligand electron transfer and electron injection
are energetically uphill.
3
0,31
here.
These studies show that, as the potential of the TiO2
conduction band changes with pH, the redox potential of
adsorbed N3 or similar molecules changes almost as much. This
is due to the fact that the ionic environment of the N3 changes
along with that of the metal oxide surface. The conclusion
resulting from the above considerations is that the ZrO2 band
edge is about 0.8 V above the reduction potential of the relaxed
N3 MLCT state and that electron injection is that amount
energetically unfavorable. This value is probably accurate to
It is of interest to note the N3/ZrO2/acetonitrile dynamics that
occur in the first 10-20 ps in Figure 4. The figure shows that
the anisotropy evolves (specifically, becomes more positive) on
the 5 ps time scale. This transient is too slow to be assigned to
vibrational relaxation, which typically takes place on a time scale
of less than a few picoseconds for large molecules in a polar
condensed phase. An assignment of this being due to electron
injection may also be ruled out based on the above polarization
considerations; electron injection would cause a decrease in the
anisotropy. Evidence of dynamics on the 5 ps time scale are
also seen in the kinetics of the total unpolarized (Apar + 2Aperp)
absorption, Figure 5. The unpolarized absorption of N3/ZrO2/
acetonitrile shows mostly a pulse width limited rise, followed
by a smaller amplitude rise on the 5 ps time scale. We note
that electron injection would result in a absorption decay,
contrary to the observed kinetics. For comparison, the total
absorption kinetics of N3/acetonitrile following 625 nm excita-
tion are also shown in Figure 5. The slight decay observed in
the N3/acetonitrile kinetics is due to interligand electron transfer
dynamics. The N3/acetonitrile kinetics closely correspond to
those presented and analyzed in a previous paper.21 The slow
rise is absent in the N3/acetonitrile (no ZrO2 particles) samples.
(The apparent negative absorption at t ) 0 is present in samples
2
7
about (0.1 V.
Photoexcitation of the N3 molecule does not, in general,
produce a relaxed MLCT state. The amount of vibrational and
solvent polarization energy that excitation puts into the N3
molecule may be estimated from the excitation wavelength and
the N3 absorption spectrum. This is easily done if the ground
and MLCT excited states are considered offset classical oscil-
lators with a thermal population on the lower curve. The offset
coordinate is taken to be the collective solvent polarization
coordinate. In this model, the width of the absorption peak is
determined by the thermal distribution in the ground state. The
21
N3 ground to excited state reorganization energy is about 1250
-
1
cm . This reorganization energy and a thermal energy of 208
-
1
cm results in excitation to the bottom of the MLCT curve
having about 5% of the peak absorption intensity. The wave-
length corresponding to this excitation may be approximated
by a linear extrapolation of the red edge of the absorption

6216 J. Phys. Chem. B, Vol. 106, No. 24, 2002
Olsen et al.
from the magnitudes of the total (unpolarized) absorbance
changes and their kinetics.
Figure 6 shows the photoinduced transient absorbance
changes at 775, 680, 640, and 600 nm obtained following 515
nm excitation. The figure shows results for N3 in acetonitrile
and methanol solvents as well as for N3 adsorbed onto zirconia
particles in these solvents. In the results presented in Figure 6,
the amplitudes of the transients have been normalized to the
amount of the pump intensity absorbed. Since the focal spot
sizes, pump energies, and 515 nm sample absorbances are held
very close to constant, this normalization corresponds to small
corrections.
Figure 6 shows that the 680, 640, and 600 nm transients
observed in the N3/methanol, N3/acetonitrile, and N3/ZrO2/
methanol samples all exhibit the same transient absorption
magnitudes. Very different results are obtained in the case of
the N3/ZrO2/acetonitrile sample, with this sample giving smaller
absorbances at 680, 640, and 600 nm (negative at 600 nm) than
the others. The negative absorbance change at 600 nm indicates
that 600 nm is to the blue of the isosbestic point in N3/ZrO2/
acetonitrile case and the absorbance change is dominated by
the N3 ground state bleach. These results indicate that a different
species is formed in the N3/ZrO2/acetonitrile case. Since the
excited MLCT state absorbs more strongly than the N3 cation
in the 600-700 nm spectral region, these results allow assign-
ment of the transient absorption in the N3/ZrO2/acetonitrile case
to the N3 cation. Thus, these comparisons show that electron
Figure 5. Unpolarized 775 nm absorption kinetics (Apar + 2Aperp) of
N3/ZrO /acetonitrile (solid line) and N3/acetonitrile (dotted line)
following 630 nm excitation.
2
lacking N3 and is due to a four-wave mixing process involving
the solvent, and especially the zirconia particles.) We conclude
that the absorption spectrum of the N3 MLCT excited state
evolves on the 5 ps time scale when the N3 is adsorbed on
zirconia particles in an acetonitrile solution. Acetonitrile relaxes
on the time scale of a few picoseconds or less,32 and this is
why little evidence of solvent relaxation is apparent in the N3/
acetonitrile kinetics shown in Figure 5. Therefore, the observed
injection occurs in the N3/ZrO /acetonitrile case, but not in the
2
N3/ZrO /methanol case. As a control, the N3/ZrO /acetonitrile
2
2
5
ps relaxation cannot be associated with surrounding bulk
and N3/acetonitrile absorbance changes at 775 nm are also
compared. Figure 6 shows that these absorbance transients have
approximately the same intensity. Since the MLCT excited state
and cation have very nearly the same extinction coefficients at
775 nm, the same absorbance change is expected, independent
of whether electron injection occurs. This indicates that the
results presented here are consistent with those that show an
acetonitrile solvent. There are two possible explanations for the
presence of this 5 ps transient. First, it may be that acetonitrile
relaxes much more slowly in the proximity of the zirconia
surface. However, the observed results would require that the
relaxation slow by more than an order of magnitude. Solvent
dynamics on zirconia surfaces have been measured for the case
3
3
+
6,7,13,17,23
of H2O and D2O, and much smaller effects were reported.
isosbestic point near 775 nm in the N3* to N3 reaction.
Similarly, the reorientation dynamics of CS2 in confined spaces
are observed to slow down compared to the bulk solution, but
by less than a factor of 2.34 Thus, an order of magnitude slowing
of the acetonitrile relaxation in the presence of the zirconia
surface seems unlikely. Second, it may be that the relaxation is
associated with the zirconia surface and that rotational motion
of surface hydroxyl groups and/or adsorbed water contribute
to this relaxation. Many hydrogen bonding solvents have
It is also important to note that the N3/methanol and N3/
acetonitrile samples give essentially identical absorbance changes,
indicating that the observed differences in the N3/ZrO2/
acetonitrile sample are not simply due to solvent-induced
spectral shifts.
The conclusion that N3/ZrO2/acetonitrile undergoes electron
injection following 515 nm excitation but does not following
6
30 nm excitation may be understood in terms of the N3 and
32,35
relaxation components on this time scale,
and this second
ZrO2 energetics. 515 nm excitation prepares N3 with 0.81 eV
of excess (solvent polarization and vibrational) energy. This
excess energy permits electron injection into states that would
otherwise be energetically inaccessible. This can be thought of
as a transient shift of the N3 MLCT state oxidation energy from
possibility seems most likely. Consistent with this explanation,
very similar dynamics are seen in the N3/ZrO2/methanol results,
also shown in Figure 4. The similarity of these results, along
with the absence of this transient without the ZrO2, suggests
that the observed 5-10 ps transient is associated with relaxation
of the polar, hydrogen bonding surface environment.
-0.80 to -1.61 V (vs SCE). Since the ZrO conduction band
2
is at about -1.6 V, this electron transfer process is very close
Dynamics of N3 on Zirconia Following 515 and 495 nm
Excitation. The results obtained following 515 and 495 nm
excitation are considerably different than those obtained fol-
lowing 630 nm excitation. While the 630 nm excitation results
show a positive anisotropy, the 515 and 495 nm results show a
negative anisotropy. As discussed above, the interpretation of
the positive anisotropy is unambiguous: no electron injection
or interligand electron transfer occurs. However, a negative
anisotropy observed following excitation at 515 or 495 nm could
be caused by either electron injection or interligand electron
transfer: no conclusive statement can be made. Thus, the
dynamics following 515 or 495 nm excitation must be inferred
to energetically neutral.
The observed kinetics allow us to put an upper limit on the
electron injection time scale. If electron injection in the N3/
ZrO /acetonitrile case occurred on the hundreds of femtoseconds
2
or longer time scale, then the absorbance would initially have
an intensity comparable to that observed for N3/acetonitrile,
and subsequently decay as electron injection occurs. This
behavior is not observed, and we conclude that electron injection
takes place on the 200 fs time scale or faster. Furthermore,
solvent and vibrational relaxation occur on the 100 fs to 1 ps
time scale. Since excitation at lower energies does not result in
electron injection, these energy relaxation processes would be

Electron Injection Dynamics of N3 on ZrO2
J. Phys. Chem. B, Vol. 106, No. 24, 2002 6217
Figure 6. Unpolarized absorption kinetics (Apar + 2Aperp) with probe wavelengths of 775, 680, 640, and 600 nm are displayed. The different
samples are indicated as N3/acetonitrile (dotted line), N3/ZrO /acetonitrile (solid line), N3/methanol (dot-dash line), and N3/ZrO /methanol (dashed
line). Excitation is at 515 nm. Note that with probe wavelengths of 600, 640, and 680 nm only the N3/ZrO /acetonitrile (solid line) absorbances are
2
2
2
smaller, indicating formation of the N3 cation in this case. Also note that the 775 nm probe absorbances for the N3/ZrO
acetonitrile samples are the same, indicating that N3* and N3 have similar absorbances at this wavelength.
2
/acetonitrile and N3/
+
expected to quench the electron injection process. These
considerations also indicate that electron injection occurs very
rapidly: within the time resolution of this apparatus, in less
than 200 fs.
injection for N3/ZrO2/methanol is very close to energetically
neutral, or perhaps slightly energetically unfavorable. It is known
that the surface charge of a metal oxide particle can affect its
conduction band redox potential. We conclude that very slight
differences in sample preparation result in small differences in
the ZrO2 particle potential and hence the inconsistent results
following 515 nm excitation. The observation that some samples
exhibit partial electron injection indicates that there are signifi-
cant inhomogeneities within a sample, and perhaps on the
surface of each particle. The sample is apparently sufficiently
inhomogeneous that there are adsorbed N3 dyes which are, and
other absorbed N3 dyes which are not, able to undergo electron
injection. This does not appear to occur following 495 nm
excitation. Excitation at 495 nm results in the N3 molecule
initially having about 0.1 eV more energy than following 515
nm excitation. This is enough that all samples examined undergo
electron injection following 495 nm excitation. A comparison
of the 640 nm transient absorption kinetics following excitation
at 515 and 495 nm is shown in Figure 7. The figure shows that
a larger absorbance is obtained for the N3/methanol than for
the N3/ZrO2/methanol sample, indicating electron injection.
Comparison of the excitation wavelength dependent N3/ZrO2/
methanol kinetics shows somewhat more absorbance is observed
The results obtained for N3/ZrO2/methanol are more prob-
lematic, and perhaps more informative. The results presented
in Figure 6 indicate that, following 515 nm excitation, no
electron injection occurs in that particular N3/ZrO2/methanol
sample. This result, however, is not consistently reproducible.
Some nominally identical N3/ZrO2/methanol samples indicate
that electron injection does occur. Still others show only slight
differences between the N3/ZrO2/methanol and N3/methanol
kinetics, indicating that some fraction of the N3 undergoes
electron injection. One’s initial thought is that in some samples
the N3 does not adsorb to the zirconia particles, and is simply
in solution. However, the static absorption spectra all show the
same shifts in the absorbance maxima, indicating that in all cases
N3 is indeed adsorbed onto the particles. These inconsistent
results may be understood in terms of the same energetic
analysis that was used to understand the N3/ZrO2/acetonitrile
results. In the methanol case, the absorption onset is at about
7
30 nm and 515 nm excitation results in an N3 molecule with
about 0.71 eV of excess energy. This is about 0.1 V less than
in the acetonitrile case. This analysis indicates that electron

6218 J. Phys. Chem. B, Vol. 106, No. 24, 2002
Olsen et al.
injection results observed for N3 on TiO2 surfaces.6,8,9,11,36
Similar binding and therefore similar electronic couplings are
expected for N3 on TiO2 and ZrO2. In the TiO2 case, the energy
of the N3 MLCT state is above the semiconductor conduction
band potential, resulting in very fast (<200 fs) electron injection.
The results presented here show that the same thing happens
when N3 is excited above the ZrO2 conduction band.
In the present case, photoexcitation at 630 nm results in an
N3 excited state which is about 0.5 V below the conduction
band. The results presented here indicate that electron transfer
to low lying surface states is slow compared to the 5 ps time
scale of solvent relaxation. This result is in contrast to those
reported for the alizarin/ZrO2 system, for which fast electron
15
transfer into the ZrO2 surface states is reported. The difference
cannot be due to weak coupling in this case: very fast injection
into ZrO2 conduction band is observed in the present case.
Furthermore, the ZrO2 particles are synthesized in exactly the
same manner and would be expected to have similar surface
chemistries, and therefore similar densities of trap states. The
reason for this discrepancy is unknown.
Figure 7. Unpolarized (Apar + 2Aperp) 640 nm absorption kinetics of
N3/methanol and N3/ZrO /methanol following 495 and 515 nm
excitation. The data are indicated as follows: N3/methanol, 515 nm
excitation (dashed line); N3/ZrO /methanol, 515 nm excitation (solid
line); N3/methanol, 495 nm excitation (dotted line); N3/ZrO /methanol,
/methanol
2
2
The above results may also be compared to those obtained
for N3 on ZrO2 nanocrystalline thin films in the presence of
2
4
5
95 nm excitation (dot-dash line). Note that the N3/ZrO
2
6
ethylene carbonate/propylene carbonate (1:1). No electron
15 nm excitation (solid line) and 495 nm excitation (dot-dash line)
curves are lower, indicating formation of the N3 cation in these cases.
transfer was observed following 605 nm excitation, in agreement
with results presented here. These studies also indicate that, at
most, limited electron injection occurs following 510 nm
excitation. Thus, little or no electron injection is observed for
N3/ZrO2 following excitation at either wavelength in the
presence of ethylene carbonate/propylene carbonate. This is
consistent with the methanol, but not the acetonitrile results
presented here. The reason for this difference is not clear. The
difference may be due to solvent-dependent shifts in the N3
energetics and that the solvent environment of ethylene carbon-
ate/propylene carbonate does not solvate the adsorbed N3 as
well as acetonitrile. These factors may result in electron injection
being energetically uphill in ethylene carbonate/propylene
carbonate as it is in most methanol samples. Alternatively, the
difference may be due to the preparation and history of the ZrO2
surfaces. We emphasize that 510 or 515 nm excitation produces
an N3 excited state that is energetically very close to the ZrO2
conduction band edge. Very slight energetic differences may
put this state slightly above or below the band edge, dramatically
changing the electron injection dynamics.
following 515 nm, consistent with the assertion that more
electron transfer occurs following 495 nm excitation.
It is of interest to note the kinetics that are observed when
electron injection is barely energetically favorable, i.e., the 515
nm excitation kinetics in Figure 7. The spike in the absorbance
near t ) 0 may reflect electron injection occurring on the time
scale of the instrument response function (about 200 fs).
However, there is no evidence of slow (>200 fs) electron
injection, which would be indicated by a resolvable absorption
decay. It could be argued that this is simply because relaxation
very quickly makes electron injection energetically unfavorable.
The relaxation dynamics of N3 in methanol have been studied
carefully,21 and bear on this possibility. Relaxation of the N3/
methanol absorption following excitation in this wavelength
range is due to polarization relaxation of the solvent and occurs
on the 5 ps time scale. This is in agreement with the known
time scales of methanol polarization relaxation dynamics, which
occurs on time scales ranging from subpicoseconds to several
3
2
picoseconds, with an average relaxation time of about 5 ps.
The above observations for both acetonitrile and methanol
are consistent with the idea that the electron-donating orbitals
on the excited state N3 are strongly coupled to a high density
of electronic states at the conduction band edge. If the coupling
The photoexcited N3 may therefore retain the excess energy
and be slightly above the conduction band energy for several
picoseconds, facilitating electron injection. All of the above
observations indicate that when electron transfer occurs, it does
so very rapidly. Electron transfer is energetically favorable only
prior to solvent and vibrational relaxation and therefore must
compete with these processes to occur efficiently. The large
difference between the absorbance changes in the N3/solvent
and N3/ZrO2/solvent samples indicates the electron transfer is
quite efficient, and we conclude that it occurs within a few
hundred femtoseconds.
is sufficiently strong, then the electron transfer is in the adiabatic
37-40
regime and is essentially barrierless.
The result can be very
rapid electron transfer, <200 fs. This appears to be the case in
the N3/ZrO2/solvent systems.
Conclusions
Several conclusions may be drawn from the results presented
here.
To this point we have tacitly assumed that electron injection
occurs into the ZrO2 conduction band. However, metal oxide
1. N3 adsorbs onto colloidal ZrO2 particles, resulting in a
significant perturbation in the absorption spectrum. This indi-
cates strong electronic coupling between N3 and the ZrO2
particle. This coupling is slightly stronger in the presence of
acetonitrile than in methanol.
2. The electron injection reaction from the excited state N3
into colloidal ZrO2 may be followed using time-resolved
polarized absorption spectroscopy of the thiocyanate to ruthe-
nium LMCT band. Both the total absorbance and the absorption
semiconductors have a high density of surface trap states 0.1-
6,28
0
.2 V below the conduction band.2
If the density of these
trap states is sufficiently high, then electron injection into these
states may also be occurring in the present case. The electron
transfer energetics are known only to within 0.1 or 0.2 V, which
is insufficiently accurate to exclude this possibility.
Comparison with Other Electron Injection Processes. The
results presented here are consistent with the fast electron

Electron Injection Dynamics of N3 on ZrO2
J. Phys. Chem. B, Vol. 106, No. 24, 2002 6219
anisotropy can reveal the electron injection dynamics, and the
two types of measurements are complementary.
(8) Heimer, T. A.; Heilweil, E. J. J. Phys. Chem. B 1997, 101, 10990.
(
9) Asbury, J. B.; Ellingson, R. J.; Ghosh, N.; Ferrere, S.; Nozik, A.
J.; Lian, T. J. Phys. Chem. B 1999, 103, 3110.
10) Asbury, J. B.; Wang, Y.; Lian, T. J. Phys. Chem. 1999, 103, 6644.
3
. The excited MLCT state of N3 undergoes rapid (e200 fs)
(
electron injection if excited at a sufficiently high energy that
the nascent MLCT state is energetically above the ZrO2
conduction band level. Rapid electron injection into very shallow
(11) Ellingson, R. J.; Asbury, J. B.; Ferrere, S.; Ghosh, H. N.; Sprague,
J. R.; Lian, T.; Nozik, A. J. J. Phys. Chem. B 1998, 102, 6455.
(
12) Tachibana, Y.; Haque, S.; Mercer, I. P.; Durrant, J. R.; Klug, D.
R. J. Phys. Chem. B 2000, 104, 1198.
13) Benk o¨ , G.; Kallioinen, J.; Korppi-Tommola, J. E. I.; Yartsev, A.
(
<0.2 eV) trap states may also occur.
(
4
. If the excitation energy is more than several tenths of an
P.; Sundstr o¨ m, V. J. Am. Chem. Soc. 2002, 124, 489.
(14) Asbury, J. B.; Hao, E.; Wang, Y.; Lian, T. J. Phys. Chem. B 2000,
04, 11957.
electronvolt below the ZrO2 conduction band, then rapid electron
injection is not observed. If electron injection does occur into
deeper trap states, it must be sufficiently slow that it does not
compete with solvent and vibrational relaxation.
1
(15) Huber, R.; Sp o¨ rlein, S.; Moser, J. E.; Gr a¨ tzel, M.; Wachtveitl, J. J.
Phys. Chem. B 2000, 104, 8995.
(16) Wang, Y.; Asbury, J. B.; Lian, T. J. Phys. Chem. A 2000, 104,
291.
4
5
. The spectrum of N3 adsorbed on ZrO2 in acetonitrile
(
(
(
17) Moser, J. E.; Gratzel, M. Chimia 1998, 52, 160.
18) Ferrere, S.; Gregg, B. A. J. Am. Chem. Soc. 1998, 120, 843.
19) Lenzmann, F.; Krueger, J.; Burnside, S.; Brooks, K.; Gratzel, M.;
solvent evolves on the 5-10 ps time scale following MLCT
excitation. Since relaxation in pure acetonitrile occurs on a faster
time scale, this indicates that the ZrO2 surface relaxes on the 5
ps time scale. We speculate that rotational motion of surface
hydroxyl groups and/or adsorbed water or methanol contributes
to this relaxation.
Gal, D.; Ruhle, S.; Cahen, D. J. Phys. Chem. B 2001, 105, 6347.
20) Fleming, G. R. Chemical Applications of Ultrafast Spectroscopy;
Oxford: New York, 1986.
(
(
(
(
21) Waterland, M. R.; Kelley, D. F. J. Phys. Chem. A 2001, 105, 4019.
22) Hagfeldt, A.; Gratzel, M. Chem. ReV. 1995, 95, 49.
23) Das, S.; Kamat, P. V. J. Phys. Chem. B 1998, 102, 8954.
Acknowledgment. This work was supported by a grant from
the U. S. Department of Energy (Grant DE-FG03-00ER15037).
M.R.W. also thanks the Foundation for Research, Science and
Technology of New Zealand for a New Zealand Science and
Technology Postdoctoral Fellowship (Contract No. KSU-901).
The authors also thank Dr. S. Ferrere for the sample of N3 dye.
(24) Tao, T. Biopolymers 1969, 8, 609.
(25) Albrecht, A. J. Mol. Spectrosc. 1961, 6, 84.
(26) Rothenburger, G.; Fitzmaurice, D.; Gratzel, M. J. Phys. Chem. 1992,
9
9
1
6, 5983.
(27) Butler, M. A.; Ginley, D. S. J. Electrochem. Soc. 1978, 125, 228.
(28) Kay, A.; Humphry-Baker, R.; Gratzel, M. J. Phys. Chem. 1994,
8, 952.
(
(
29) Redmond, G.; Fitzmuarice, D. J. Phys. Chem. 1993, 97, 1426.
30) Zaban, A.; Ferrere, S.; Sprauge, J.; Gregg, B. A. J. Phys. Chem.
References and Notes
997, 101, 55.
(
(
(
(
31) Zaban, A.; Ferrere, S.; Gregg, B. J. Phys. Chem. B 1998, 102, 452.
32) Maroncelli, M. J. Mol. Liq. 1993, 57, 1.
33) Pant, D.; Levinger, N. E. J. Phys. Chem. B 1999, 103, 7846.
34) Loughnane, B. J.; Farrer, R. A.; Scodinu, A.; Reilly, T.; Fourkas,
(
(
(
1) Oregan, B.; Gratzel, M. Nature 1991, 353, 737.
2) Hagfeldt, A.; Gratzel, M. Acc. Chem. Res. 2000, 33, 269.
3) Asbury, J. B.; Hao, E.; Wang, Y.; Ghosh, H. G.; Lian, T. J. Phys.
Chem. B 2001, 105, 4545.
4) Nazeeruddin, M. K.; Kay, A.; Rodicio, I.; Humphry-Baker, R.;
Muller, E.; Liska, E. P.; Vlachopoulos, N.; Gratzel, M. J. Am. Chem. Soc.
993, 115, 6382.
5) Bauer, C.; Boschloo, G.; Mukhtar, E.; Hagfeldt, A. J. Phys. Chem.
B 2001, 105, 5585.
J. T. J. Phys. Chem. B 2000, 104, 5421.
35) Horng, M. L.; Gardecki, J. A.; Papazyan, A.; Maroncelli, M. J.
Phys. Chem. 1995, 99, 17311.
36) Hannappel, T.; Burfeindt, B.; Storck, W.; Willig, F. J. Phys. Chem.
(
(
1
(
(
B 1997, 101, 6799.
(
6) Tachibana, Y.; Moser, J. E.; Gratzel, M.; Klug, D. R.; Durrant, J.
(37) Gao, Y. Q.; Marcus, R. A. J. Chem. Phys. 2000, 112, 3358.
(38) Gao, Y. Q.; Marcus, R. A. J. Chem. Phys. 2000, 113, 6351.
(39) Smith, B. B.; Halley, J. W.; Nozik, A. J. Chem. Phys. 1996, 205,
47.
R. J. Phys. Chem. 1996, 100, 20056.
(
7) Moser, J. E.; Noukakis, D.; Bach, U.; Tachibana, Y.; Klug, D. R.;
Durrant, J. R.; Humphry-Baker, R.; Gratzel, M. J. Phys. Chem. B 1998,
02, 3649.
1
(40) Smith, B. B.; Nozik, A. J. Chem. Phys. 1996, 205, 245.