Polar solvation dynamics of H2O and D2O at the surface of zirconia nanoparticles

Article abstract of DOI:10.1021/jp991746q

Time-resolved fluorescence-upconversion spectroscopy has been used to study the polar solvation dynamics of H₂O and D₂O at the surface of zirconia (ZrO₂) nanoparticles. While an isotope effect is observed for the solvation dynamics of bulk D₂O, there is no isotope effect on the interfacial solvation dynamics. The interfacial solvation dynamics are the same for H₂O and D₂O and are faster than the bulk solutions. The bulk isotope effect is due to stronger hydrogen bonding in D₂O compared to H₂O, slowing the reorientation of the excited-state dipoles in the bulk D₂O. The lack of isotope effect for the interfacial dynamics is explained in terms of the solvent interacting with the ZrO₂ surface. ? 1999 American Chemical Society.

Full text of DOI:10.1021/jp991746q

7846
J. Phys. Chem. B 1999, 103, 7846-7852
Polar Solvation Dynamics of H2O and D2O at the Surface of Zirconia Nanoparticles
Debi Pant and Nancy E. Levinger*
Department of Chemistry, Colorado State UniVersity, Fort Collins, Colorado 80523-1872
ReceiVed: May 27, 1999; In Final Form: July 19, 1999
Time-resolved fluorescence-upconversion spectroscopy has been used to study the polar solvation dynamics
of H2O and D2O at the surface of zirconia (ZrO2) nanoparticles. While an isotope effect is observed for the
solvation dynamics of bulk D2O, there is no isotope effect on the interfacial solvation dynamics. The interfacial
solvation dynamics are the same for H2O and D2O and are faster than the bulk solutions. The bulk isotope
effect is due to stronger hydrogen bonding in D2O compared to H2O, slowing the reorientation of the excited-
state dipoles in the bulk D2O. The lack of isotope effect for the interfacial dynamics is explained in terms of
the solvent interacting with the ZrO2 surface.
I. Introduction
important because it allows electrons to approach the interface.
Thus it is important to probe solvent motion at the semiconduc-
tor interface.21
The interaction of liquids with solid surfaces finds application
in many different areas of chemistry including chromatography,
heterogeneous catalysis, and solar energy harvesting. Despite
their ubiquity, little is known about the structure of liquids at
solid surfaces and even less is known about dynamical aspects
of liquids interacting with solid surfaces. In this paper, we report
on the dynamics of H2O and D2O molecules interacting with
the surface of nanoparticulate ZrO2.
This system is interesting for several reasons. Among others,
it serves as a model for TiO2. In recent years, dye-sensitized
TiO2 nanoparticles have become a promising medium for the
conversion of solar energy into electrical energy in photoelec-
trochemical cells.1-5 Dye sensitization overcomes the large band
gap of TiO2 using an adsorbed molecule to inject the electron
from the molecule to the semiconductor. The electron injection
time from the adsorbed dye to the semiconductor has been
measured to be ultrafast, typically <200 fs, and is similar for
the all dye-semiconductor systems that have been meas-
ured.2,3,6-13 This ultrafast electron injection has been attributed
to strong electronic coupling between the dye and conduction
band energy levels in the semiconductor. The back reaction from
the semiconductor to the solution varies by many orders of
magnitude from system to system, and is believed to lie in the
Marcus inverted region.14-16
The motion of electrons in solution can be highly sensitive
to solvent motion. Indeed, barrierless ultrafast electron-transfer
reactions are often determined by the solvation dynamics.17,18
Relaxation times for isolated electrons in solution also correlate
directly to the ability of the solvent to equilibrate the newly
formed charge.19 The extremely fast electron injection from dye
molecules into semiconductor nanoparticles has been attributed
to the high density of states in the semiconductor conduction
band.20 Though solvent reorganization may not be required for
the dye molecule because injection occurs into the conduction
band, the nature of the conduction band in the nanoparticles
may lead to injection into discrete levels making the electron
injection analogous to barrierless electron transfer. Even if
injection occurs into a true continuum, solvent motion may be
Solvation dynamics have been studied for a wide range of
bulk liquids22 and for some restricted environments. Several
experimental techniques have been used to measure solvation
dynamics, including time-resolved fluorescence,23-25 time-
resolved optical Kerr effect,26-29 and photon echoes.30,31 For
all solvents studied there exist two components to the solvation
dynamics: an ultrafast, sub-100-fs inertial component and slow-
er diffusive subpicosecond components.32 The solvation dynam-
ics of liquid water have been studied in detail theoretically33-37
as well as using ultrafast experimental techniques.31,38-41
The solvation of a solute molecule (or ion) in water has been
shown to be bimodal with a Gaussian response at short times,
contributing 70%-80% to the total solvation energy, and an
exponential response at longer times. These two different
responses have been interpreted as different types of solvent
motion: an ultrafast, sub-50-fs underdamped component and
two slower diffusive components with subpicosecond time
constants.
Due to the interactions of solvent molecules with the surface,
one would expect the dynamics at interfaces to differ from bulk
solution. While the dynamics in bulk solutions have been
measured extensively, there are only few reports of measure-
ments at surfaces. Yanagimachi et al.42 studied solvation
dynamics of 1-butanol at silica surfaces. While the shorter
solvent relaxation time components could not be resolved due
to the >60 ps time-resolution of their instrument, the longer
time components increased for the interfacial butanol and the
solvent appeared to be influenced a substantial distance from
the interface. Bessho et al.43 explored fluorescence decays from
1,8-anilino-8-naphthalenesulfonic acid at the water-heptane
interface. Like Yanagimachi et al., the polar solvent response
observed was also slower at the interface than in bulk. Using
time-resolved optical Kerr effect measurements, Fourkas and
co-workers44,45 have probed liquid dynamics in porous silicate
sol-gel glasses. They demonstrate that for weakly wetting
solvents, such as CS2 and CH3I, the dynamics of the molecules
at the surface are slowed and determined by geometric confine-
ment. However, for more interacting solvents such as acetonitrile
the dynamics are influenced by molecular interactions with the
* Author to whom correspondence should be addressed. E-mail: levinger@
lamar.colostate.edu.
10.1021/jp991746q CCC: $18.00 © 1999 American Chemical Society
Published on Web 08/27/1999

Polar Solvation Dynamics of H2O and D2O on Zirconia
J. Phys. Chem. B, Vol. 103, No. 37, 1999 7847
surface. Recently, we have used ultrafast time-resolved fluo-
rescence techniques to probe the motion of water in reverse
micelles41,46-49 where we have shown that the water motion
inside the reverse micelles is significantly slowed compared to
the bulk solutions.
evaporation at a temperature lower than 40 °C. The resulting
white powder was dried under vacuum overnight. KOH was
placed in the desiccator to help remove the excess HNO3. The
ZrO2 particle radius measured 3.8 ( 0.5 nm via dynamic light
scattering (DynaPro-MSTC, Protein Solutions). For the solvation
dynamics measurements, we estimate the concentration of
nanoparticles in the solution was 1 × 10-3 M. Coumarin 343
(C343, Exciton), was used as a probe molecule without further
purification. The dye concentration was 3 × 10-4 M. At these
relative concentrations it is unlikely that multiple dye molecules
adsorb on a single nanoparticle. At higher relative concentrations
of dye to nanoparticle, that is, up to ∼20 dye molecules per
nanoparticle, the absorption spectra were significantly altered.
We discuss these results briefly.
Bulk samples for comparison with the particle suspension
were prepared from high-purity H2O (Milli-Q filtered, 18.2MΩ
cm) and D2O (99.5%, Aldrich). To study the solvent effect on
the dynamics at the surface of ZrO2, we have attempted to make
the samples in acetonitrile, ethanol, and 2-propanol but the ZrO2
did not remain suspended well with these solvents.
The fluorescence upconversion spectrometer used in these
experiments has been described in detail previously.41 Briefly,
a mode-locked Ti:sapphire laser produced output pulses centered
at 820 nm, with a duration of 160 fs, at 100 MHz repetition
rate, and energies of 6.5 nJ/pulse. Frequency doubled laser
pulses served to excite the sample, while the residual funda-
mental light was used to gate the sample fluorescence in a
nonlinear BBO crystal. The sum frequency of the fluorescence
and the gate pulse was detected as a function of the time delay
between excitation and gate pulses. The polarization of the pump
and gate pulses is kept at the magic angle to eliminate effects
from rotational diffusion. The upconverted signal passed through
a monochromator and was detected with a photomultiplier tube.
Signals were collected via photon counting interfaced to a
computer. Samples were circulated with a peristaltic pump fitted
with PTFE tubing.
Data were collected at nine different wavelengths from 455
to 575 nm at 15 nm intervals. The data were fit to a
multiexponential function using an iterative-reconvolution fit
program with the cross-correlation of the pump and gate pulses
as the instrument response function. From these fits, time-
resolved fluorescence spectra were reconstructed. The time
correlation function, C(t),
Compared to experimental studies, even fewer models exist
for solvent dynamics at any kind of liquid interface.50-52
Chandra et al.53 have developed a time-dependent density
functional theory to predict polarization relaxation at the solid-
liquid interface. They find that while the polarization is smaller
at the interface, the relaxation occurs at a faster rate than the
bulk relaxes. Ursenbach et al.50 have developed a model for
water at the surface of a silicon single crystal. Their results show
that the vibrational modes of water molecules at the surface
are virtually unchanged from their bulk values. However, they
suggest that ultrafast motions of the interfacial solvent, occurring
on the time scale of molecular vibrations, may impact interfacial
electron transfer. Benjamin has simulated the dynamics of water
at liquid/liquid interfaces. His molecular dynamics calculations
predict little change in the water dynamics at the interface and
that as close as one monolayer away from the interface, the
dynamics are completely bulklike.51 Using molecular dynamics
calculations, Hartnig et al.54 have simulated water motion within
model polar and nonpolar cylindrical pores. In nonpolar pores,
reduced hydrogen bonding at the surface results in a translational
motion and dipolar motion that is faster than bulk water.
However, in polar pores, generated by adding surface charges
to the model system, the translation motion and dipolar
relaxation times are longer than bulk. Using molecular dynamics
calculations, Shinto et al.55 have modeled the response of water
at NaCl surfaces. While their simulations show that the water
diffusivity is reduced at the surface, they also observe a strong
perturbation to the hydrogen-bonding network for solvent
molecules at the solid surfaces.
Recently, we studied the solvation dynamics of a 95:5% (v/
v) water/acetone mixture at the ZrO2 surface.56 The interfacial
solvent response we observed displayed two subpicosecond
diffusive components with the relaxation times similar to that
of bulk solution. However, the relative amplitude corresponding
to the shorter component was significantly larger for the ZrO2
solution than the bulk solution, which led to a faster average
solvation response for molecules at the ZrO2 surface. Several
questions were generated from this study, such as (i) how does
the minuscule acetone component affect the observed dynamics
at the surface of ZrO2 nanoparticle, and (ii) at the surface, do
the dynamics depend on the solvent used for the measurements?
Is the result universal, that is, would other systems display the
same response? To address these and other issues, we report
the solvation dynamics of pure H2O and D2O at the surface of
ZrO2 nanoparticles. The dynamics are measured for the bulk
solutions also in order to compare with the dynamics at the
surface of ZrO2 nanoparticles.
ν(t) - ν(∞)
ν(0) - ν(∞)
C(t) )
(1)
where ν(t), ν(0), and ν(∞) are the peak of the fluorescence
spectra at time ) t, instantaneously, and at equilibrium,
respectively, was calculated using the peak position of the
reconstructed fluorescence spectra fitted to log-normal line
shapes.
Time-resolved fluorescence anisotropy measurements were
made for the dye in the ZrO2 and bulk solutions to quantify
dye adsorption to the nanoparticle surface. These measurements
were analyzed were via
II. Experimental Methods
The ZrO2 particle solutions were prepared following the
previously published synthetic methods for the synthesis of TiO2
nanoparticles.6,57 Briefly, 200 mg of Zr(IV) isopropoxide
(Aldrich) was dissolved into 25 mL of 2-propanol (Fisher
Scientific, 99.9%). This solution was slowly added (3 mL/min)
to 200 mL of acidified H2O (with 1.2 mL of concentrated HNO3)
at 1 °C under N2 (air was previously removed from the solution
by boiling for 5 min). Milli-Q quality water (18.2 MΩ cm
resistivity) was used to avoid impurities in the final ZrO2 sample.
After stirring for 3-4 h the solvent was removed by rotary
I||(t) - I (t)
I||(t) + 2I (t)
r(t) )
(2)
where I (t) and I (t) are the time-dependent intensities of the
||
upconverted fluorescence with the polarization of the pump and
probe pulses parallel and perpendicular to each other, respec-
tively. The initial anisotropy was ∼0.4 for all samples, showing
that initially they are completely isotropic.

7848 J. Phys. Chem. B, Vol. 103, No. 37, 1999
Pant and Levinger
changes in the polarization energies of the dye’s ground and
excited electronic state from the interaction of the dye with the
particle.
One possible explanation for this effect suggested by Martini
et al.58 is that the dye is influenced by an image dipole formed
in the semiconductor. To assess the validity of this argument,
we have calculated the potential due to the molecule at the
surface of the ZrO2 particle both with and without an image
dipole.59 We approximate the dipole as the sum of equal but
opposite charges separated by a distance estimated by the
structure of the dye60 to yield the known ground and excited-
state dipole moments.61 Because the positions selected for the
charges comprising the dipole are somewhat arbitrary, we
change only the magnitude of the charges, not their positions,
to change the dipole magnitude. Then, using the formalism
presented by Jackson59 to describe the potential due to a charge
in the presence of an image charge, the potential due to the
dipole and its image will be
2
qi
q′i
1
Φ )
+
(3)
∑
i
i
(
)
i)1 ꢀ
1 R1
R2
where q is the charge of in the material, q′ is the image charge,
i
i
at distance R1 and R2 from the charge and image charge,
respectively. The summation is over the individual charges
comprising the dipole. The image charge, q′ is given by
Figure 1. (a) Absorption and (b) emission spectra for C343 in bulk
H2O (s - s), bulk D2O (---), ZrO2 in H2O (s), and ZrO2 in D2O (---).
ꢀ2 - ꢀ1
q′ )
q
(4)
(
)
ꢀ1 + ꢀ2
Absorption spectra were recorded with a Cary 2400 UV-
vis-NIR spectrophotometer. Fluorescence spectra were mea-
sured with a home-built fluorometer.
where ꢀ1 and ꢀ2 are the dielectric constant of the water, 78,62
i
and ZrO2, 13.5,63 respectively. In cylindrical coordinates, R1
i
)
F2+(d 2-z2) and R2
)
F2+(d 2+z2), where di is the
distance of the charge from the interface. In the absence of the
image dipole, the potential due to the dye’s dipole is given by
x
x
i
i
III. Results and Discussion
Sample Characterization. Before time-resolved solvation
dynamics measurements were made, samples were characterized
via several different methods. First, the steady-state absorption
and emission spectra of C343 were measured in bulk H2O and
D2O, and H2O/ZrO2 and D2O/ZrO2 samples. These spectra of
C343 are shown in Figure 1. Clearly, both the absorption and
the emission spectra of C343 in the ZrO2 solutions shift
significantly to longer wavelength compared to the bulk
solutions. While the absorption spectra for C343 in the ZrO2
sample are slightly broader than the bulk samples, the shapes
of the emission spectra remain largely unchanged. Part of the
red shifting observed for the absorption spectra is most likely
due to a difference in the dye’s oxidation state. That is, in the
acidic ZrO2 solutions, the dye is found in its neutral form whose
spectrum peaks near 458 nm in aqueous solution compared to
the peak at 427 nm for the C343 anion in aqueous solution.46
The red shift in the steady-state spectra of C343 for the ZrO2
solutions implies that the environment sensed by the dye in the
ZrO2 solutions differs from bulk solution. Similar red shifts in
steady-state spectra have been observed for virtually every dye-
sensitized semiconductor system that has been probed. For
example, Cherepy et al.2 observed a red shift in the absorption
and emission spectra of a cyanine dye complexed to the ZrO2
nanoparticles. They attributed this red shift to the presence of
a charge transfer state for the cyanine dye complexed to ZrO2
in analogy to the cyanine dye complexed to TiO2. Martini et
al.6,58 have also measured red shifts in the absorption spectrum
of 9-anthracenecarboxylic acids on TiO2 and ZrO2. They
hypothesize that the shifts in the absorption spectra reflect
q1
q2
Φ )
+
(5)
ꢀR1 ꢀR2
where q is the charge, ꢀ is the dielectric constant of water, and
R is the distance from the charge. We calculated the potential
due to the ground- and excited-state dipole moments assuming
that the ground- and excited-state dipole moments of C343 are
10.41 and 15.70 D.61 A plot of the potential as a function of
distance from the interface is shown in Figure 2. This plot shows
that the image dipole generates changes peak-to-peak potential
distance by only ∼4% for both ground and excited states.
However, the largest deviation in the potential occurs near the
interface. Thus, it is possible that the image dipole could lead
to an overall red shift of the steady-state spectra.
The relative concentration used for the solvation dynamics
measurements reported here yields an average of <1 dye
molecule/particle. However, because many dye sensitization
studies employ a significantly higher relative concentration, we
have probed the steady-state absorption and emission, and time-
resolved emission spectra as a function of the relative dye
concentration. To avoid effects due to the concentration of dye
in solution, we varied relative dye/particle concentration by
changing the nanoparticle concentration in solution while the
dye concentration was kept constant. In Figure 3, absorption
spectra of C343 are shown as a function of ZrO2 nanoparticle
concentration. At low ZrO2 concentration, the absorption spectra
are significantly blue shifted. As the particle concentration is

Polar Solvation Dynamics of H2O and D2O on Zirconia
J. Phys. Chem. B, Vol. 103, No. 37, 1999 7849
TABLE 1: Fit Parameters for Time-Resolved Anisotropy
Measurementsa
sample
R
τ (ps)
bulk H2O
bulk D2O
ZrO2 in H2O
ZrO2 in D2O
0.37
0.37
0.34
0.35
83
75
5700
5200
a Error bars are (10%.
a low probability for finding more than one dye molecule
adsorbed to any given nanoparticle to ensure that the dynamics
measured reflect solvent motion rather than complications from
aggregation.
We used time-resolved fluorescence anisotropy to measure
whether the dye adsorbed to the ZrO2 particles. Time-resolved
fluorescence anisotropy was therefore measured for each dif-
ferent solution of C343. The rotational correlation times are
obtained from multiexponential fits to the data, and the fitting
parameters are given in Table 1. The rotational correlation time
for the dye in H2O agrees well with the literature values for the
bulk water.41 While the viscosity of D2O is higher than that for
H2O, the emission anisotropies measured here for H2O and D2O
were, within experimental error, the same. In contrast, the large
increase in the rotation time for C343 in the ZrO2/H2O solution
(4 ns) as well as in the ZrO2/D2O solution (5 ns) indicates that
the dye adsorbs to the surface of the nanoparticles. If the dye
were free in the ZrO2 solution, it should display a rotational
correlation time that is the same as for the bulk solution. The
presence of nanoparticles in solution is insufficient to slow the
molecular rotation time. In aqueous SiO2 nanoparticle suspen-
sions, r(t) for the dye is indistinguishable from bulk solution
because the dye does not adsorb to the negatively charged
surface of the silica particles. This shows that mere presence
of the nanoparticles in solution is insufficient to change the
rotational correlation time and that the C343 dye adsorbs
strongly to the ZrO2 particle surface.
Figure 2. Electric potential as a function of distance from the interface
for the ground (heavy lines) and excited (light lines) states of Coumarin
343 at the water-ZrO2 interface. Solid lines represent calculations
including the image dipole, and dashed lines are without the image
dipole.
Figure 3. Absorption spectra of C343 in H2O with various ZrO2
nanoparticle concentrations.
In the studies reported here, the C343 molecule does not inject
electrons into the conduction band of ZrO2, because the excited-
state redox couple of C343 lies below the conduction band edge
of ZrO2.56 Evidence for the unfavorable electron injection can
be seen from the fluorescence intensity, which increases for the
dye in the ZrO2 solutions compared to the bulk solutions. If
the dye injected electrons into the ZrO2 solution, the fluores-
cence quantum yield, hence the intensity, should decrease
substantially. Thus, in contrast to C343 on TiO2,3 the excited-
state dynamics for C343 on ZrO2 solution reflect the solvation
dynamics rather than electron injection.
Solvation Dynamics, The solvation dynamics of H2O and
D2O at the surface of the ZrO2 nanoparticles and in bulk
solutions were measured. Fluorescence decays from which the
time-resolved fluorescence spectra were generated were normal-
ized to the steady-state spectrum at 5 ps. No additional Stokes
shift was observed beyond 5 ps for the longer time scans, hence
there was no additional contribution to the solvation dynamics
on a longer time scale. Figure 4 displays the time-correlation
function for bulk H2O and H2O at the surface of ZrO2 while
Figure 5 shows the time-correlation function for bulk D2O and
D2O at the surface of ZrO2. All solvation dynamics parameters
for these samples are given in Table 2. In all these solutions,
the decays fit well to a biexponential function characterized by
three different time scales.25 First, we calculate the inverse of
the decay rate or the “initial” time constant, τ0,
increased, the dye’s spectrum shifts to longer wavelength.
However, unlike smooth shifts we have observed for other
varying environments,41 these spectra also display significant
changes in the spectral shape.
As shown in Figure 3, the absorption spectra show a dramatic
change with dye concentration. At low relative concentration
(fewer than one dye molecule per nanoparticle) the absorption
spectrum reflects the spectrum of the protonated dye in bulk
solution. As the relative dye concentration increases, an ad-
ditional spectral feature appears to the blue of the original dye
absorption peak. This shoulder grows with increasing relative
concentration eventually dominating the spectrum. This behavior
is reminiscent of H-aggregation.64 In contrast to the dramatic
spectral shifts observed for the absorption spectrum, the steady-
state emission spectra vary only slightly over the extensive
relative concentration range. It appears that excitation into the
“aggregate” band yields emission from the monomer, or, at least,
emission that is indistinguishable from the monomer emission.
In contrast, the time-resolved fluorescence Stokes shift displays
significantly faster relaxation for the high relative dye concen-
tration than the low concentration. This large, fast red shift
depends on the relative concentration of dye molecules on the
surface. Thus, while we cannot verify unequivocally, these
results indicate that the time-resolved behavior observed at high
relative concentration of dye on the nanoparticle surface, that
is >40 molecules per particle, arises from energy transfer within
the aggregated molecules. Thus, all our solvation dynamics
experiments were conducted in the range where statistics predict
-1
τ0-1 ) a1τ1-1 + a2τ2
(6)

7850 J. Phys. Chem. B, Vol. 103, No. 37, 1999
Pant and Levinger
the time resolution for these measurements precluded full
resolution of the solvation dynamics inertial component. A
significant isotope effect is observed for the longer solvation
time component in D2O. The structure of D2O is reported to be
more ordered and stable with stronger hydrogen bonding than
H2O.66 Thus, the hydrogen-bond making and breaking processes
are slower in D2O than in H2O, leading to slower motions and
consequently slower solvation dynamics in D2O. Further, the
substantial isotope effect in the solvation dynamics of D2O can
simply reflect the increased mass hence the larger moment of
inertia of D2O compared to H2O.
Though the static dielectric constant and the index of
refraction of H2O and D2O are almost the same,67 dynamical
parameters such as the Debye relaxation time, τD, are different.68
Like bulk H2O, the solvation time correlation function for D2O
reveals two time components. The shorter relaxation times for
both solvents are similar and most likely reflect a degree of
inertial behavior.37 However, the longer time component for D2O
is significantly larger than bulk H2O. This leads to a larger
solvation time for D2O. Barbara and co-workers65 report time-
resolved fluorescence Stokes shift measurements using C343
in H2O and D2O. Their results showed the solvation dynamics
were approximately 30% slower in D2O than in H2O, essentially
what is predicted from a dielectric continuum model and the
kinetic isotope effect. Das et al.69 have reported that the solvation
dynamics of D2O are a factor of 1.2 slower than H2O when
confined to the interior of reverse micelles. Using the technique
of resonance heterodyne optical Kerr effect, Zolotov et al.70 have
measured the isotope effect for D2O compared to H2O. Their
results agree well with the effect of isotopic substitution on the
Debye relaxation time. They attributed the results to a specific
solvation due to the formation and breaking of an intermolecular
solute-solvent hydrogen bond. Using ultrafast near-infrared
pump/variable wavelength probe transient absorption spectros-
copy of the solvated electron, Kimura et al.19 have reported that
a small isotope effect exists for the longitudinal relaxation time
in water.
Figure 4. Normalized time-correlation function, C(t), displaying the
dynamic response for C343 in bulk H2O (2) and at the surface of ZrO2-
in H2O (b). Points are the experimental data and lines are fits to the
data via ∑i ai exp(-t/τi). Values for the fits are given in Table 2.
Figure 5. Normalized time-correlation function, C(t), displaying the
dynamic response for C343 in bulk D2O (2) and at the surface of ZrO2
in D2O (b). Points are the experimental data and lines are fits to the
data via ∑i ai exp(-t/τi). Values for the fits are given in Table 2.
Isotope effects have been observed in solvation dynamics for
solvents other than water. Pal et al.71 have investigated the
deuterium isotope effect on the solvation dynamics of aniline
using the dynamic Stokes shift method. They found a slower
solvation dynamics for the deuterated aniline than for normal
aniline and explained the isotope effect as due to the presence
of intermolecular hydrogen bonding. Using femtosecond fluo-
rescence upconversion, Shirota et al.72 found the solvation
dynamics to be slower for deuterated solvent than for normal
methanol.
TABLE 2: Fit Parameters for Solvation Dynamics
Measurements
∆ν
τ1
τ2 τ0 t1/e τs
sample
(cm-1
) R1 (ps) R2 (ps) (ps) (ps) (ps)
bulk water
800 0.34 0.07 0.66 0.63 0.17 0.40 0.44
700 0.32 0.14 0.68 0.68 0.30 0.44 0.51
500 0.18 0.06 0.82 1.01 0.26 0.81 0.84
150 0.73 0.09 0.27 0.64 0.12 0.14 0.24
water/acetonea
bulk D2O
ZrO2 + water
ZrO2 + water/acetonea 260 0.77 0.15 0.23 0.66 0.18 0.20 0.27
ZrO2 + D2O
190 0.63 0.06 0.37 0.56 0.09 0.12 0.25
The isotope effects in D2O solvent motion have been
explained via theory and simulations as well as experiments.
Schwartz and Rossky have modeled the H2O and D2O response
to the solvated electron.35,37,73 Their calculations mirror experi-
mental results. That is, the fastest, underdamped, “inertial”
response and the slowest solvent response are largely unaffected
by isotopic substitution. However, the intermediate time re-
sponse, ∼1 ps, corresponding largely to rotational motion,
displays the expected ∼30% isotope effect. They explain the
missing isotope effect in the fastest dynamics as due to
translational motion that would be less strongly influenced than
rotational motions. In contrast, Fonseca and Ladanyi74 have
shown via molecular dynamics simulations that the fast inertial
type component of solvation in methanol is slowed upon isotopic
substitution and attribute this to the free rotation of the hydroxyl
hydrogen about the CO bond. Svishchev and Kusalik75 have
also used molecular dynamics simulations to show that both
a Results from Pant and Levinger.56
where a1, and a2 are the relative amplitudes, and τ1, and τ2 are
the corresponding relaxation times. This time constant, τ0,
reflects short time behavior of the dynamics. Second, we use
the time required to reach 1/e (0.368) of the original amplitude,
which represents the overall relaxation time. Finally, we
calculate the average solvation time, τs ,
τs ) a1τ1 + a2τ2
(7)
characterizing the long time component of our data.
Before comparing the dynamics of bulk D2O with the
dynamics of D2O at the surface of ZrO2, we discuss the solvation
dynamics for the bulk H2O and D2O. Our measurements of
solvation dynamics for the liquid H2O and D2O (Table 2) are
in good agreement with the literature values,31,38-41,65 though

Polar Solvation Dynamics of H2O and D2O on Zirconia
J. Phys. Chem. B, Vol. 103, No. 37, 1999 7851
short as well as long time components of dielectric relaxation
process in water become slower on isotopic substitution of the
hydrogen by the deuterium or tritium atoms. Recently, Nandi
et al.76 have predicted solvation dynamics using molecular
hydrodynamic theory. In particular, they find that the longer
time component in D2O is slower than that in the H2O. Thus,
calculations would predict isotope effects in both the short and
long time response of the liquid.
of the water that is important for stabilizing the dye molecule,
it also perturbs the electronic environment in which the dye
sits.
An additional effect may contribute to the reduced overall
Stokes shift. One way to reduce the overall Stokes shift would
be for the water molecules at the surface to be prealigned. If
the water molecules are close to the position required to solvate
the dye’s excited state then we would anticipate a smaller degree
of reorientation leading to a smaller Stokes shift. It is most likely
that a combination of all three effects, that is, the partial blocking
of the dye’s solvation shell, prealignment of the surface water,
and the image dipole lead to the overall smaller solvation energy
for the dye at the particle surface.
The interfacial dynamics differ from the bulk dynamics in
several ways. First, all of the solvation dynamics data show
that the molecules at the ZrO2 surface move faster than they do
in the bulk solutions (see Table 2). Second, while the two
subpicosecond time components comprising the interfacial
solvent response are the same, if not shorter, than the corre-
sponding components in bulk solutions the relative amplitude
of the shorter component dominates the interfacial dynamics.
This leads all three characteristic time responses, τ0, t1/e, and
τs , to be shorter by at least a factor of 2 for surface solvent
molecules than for the corresponding bulk solution. Third, the
overall fluorescence Stokes shift displayed by the dye adsorbed
to the particle surface is considerably smaller than the shift
observed in bulk solvent. This implies that the solvation energy
is reduced for the dye adsorbed to the ZrO2 surface and that
the extent of molecular motion of the water molecules is smaller
than in bulk. Previously, we reported similar results for 95:5%
water:acetone solutions.56 The results for water solutions
reported here show that the presence of 5% acetone, by volume,
in our previous work plays only a minor a role in the measured
interfacial and bulk water dynamics. The solvation dynamics
for bulk acetone/water and neat water virtually identical for both
bulk and at the surface of ZrO2. Finally, we observe no isotope
effect for the solvation dynamics at the surface of ZrO2.
Surprisingly, all three characteristic time responses, τ0, t1/e, and
τs , are essentially the same for both H2O and D2O at the ZrO2
surface.
We feel that the most plausible reason for the faster response
and missing isotope effect at the surface is that the hydrogen
bonding is disrupted allowing water to move more freely than
it does in bulk solutions. The ZrO2 solutions are slightly acidic
and the point of zero charge for ZrO2 is near pH ) 7.78 As a
result, the surface chemistry is most likely modified from the
standard O-Zr(O)-O moiety to O-Zr(OH)-O (O-Zr(OD)-
O) in the H2O(D2O) solutions. Thus, the solvation dynamics
probe H2O(D2O)-OH(OD) interactions rather than interactions
with surface O atoms. If the interfacial water forms weaker
hydrogen bonds with the surface OH(D) groups than it does
with itself, then it could reorient more quickly. Furthermore,
interfacial motion could reflect librational motion of the surface
hydroxyls in addition to surface water motion.
Prealignment of the surface water and a reduced solvation
shell could also potentially shorten the relaxation time. Pre-
alignment could reduce the solvation response; because the
extent of the motion would be smaller, the time for the motion
to occur would be shorter. A smaller number of water molecules
in the dye’s solvation shell could also lead to a shorter relaxation
time if collective motion were responsible for the observed
relaxation. However, neither of these effects can account for
the missing isotope effect. Therefore, the shorter solvation times
measured at the particle surface cannot be solely due to these
effects.
There exist plausible explanations for the faster solvation
dynamics, smaller solvation energy, and missing isotope effect
we observe at the ZrO2 surface. The smaller solvation energy
is most likely due to several different effects in concert. First,
the nanoparticle surface blocks the part of the dye molecule, so
the relative number of solvent molecules surrounding the C343
molecule is smaller than the number surrounding it in the bulk
solution. Thus fewer solvent molecules would respond to the
new electric field leading to smaller solvation energy. A simple
geometric calculation of the total number of water molecules
surrounding the C343 molecule shows that approximately 12.5
water molecules are in the first solvation shell. If each water
molecule contributed the same amount to the overall Stokes
shift, then the observed Stokes shift for C343 adsorbed on the
ZrO2 surface would lead us to conclude that the molecule is
solvated by at most four water molecules. Geometrically, it is
impossible for the surface to block that much of the dye
molecule. However, spectroscopic measurements of the related
molecule, coumarin 151 (C151) show that a single water
molecule located near the cyclic ester functionality of the
molecule is responsible for a large fraction of the Stokes shift.77
If C343 behaves similarly to C151, then the overall Stokes shift
would be smaller because the C343 attaches to the nanoparticle
via this group. Thus, the semiconductor would block the position
for the water molecule that causes the most significant portion
of the Stokes shift. Additionally, our image dipole calculation
shows that the potential due to the molecule and its image differs
most dramatically in the region where the molecule attaches to
the particle. Thus, the ZrO2 particle not only blocks the position
The intermolecular structure of water at the particle surface
and near the dye could be perturbed by the presence of the dye
molecule. Since the dipole moment of the dye molecule points
toward the nanoparticle, it is aligned with the anticipated
orientation of water at the surface, assuming that the ZrO2 is
hydroxylated. Thus the presence of the dye could disrupt the
normal water surface structure causing the water to align relative
to the dye rather than the surface. While this means that the
dynamics we measure could be significantly different from the
dynamics in the absence of the dye molecule, this result is
inconsequential for comparison with electron injection by the
C343 into the semiconductor because it is only dye sensitization
that makes electron injection possible. Thus the solvation
dynamics reported here are most relevant for comparison with
electron injection into the semiconductor particles.
In summary, we have measured the solvation dynamics for
H2O and D2O both in bulk solution and at the surface of
nanoparticulate ZrO2 using the time-resolved fluorescence
Stokes shift technique. For bulk solutions, the dynamics follow
the anticipated behavior.38,65 The dynamics at the particle surface
depart from bulk behavior in several different ways. First, the
solvation dynamics at the surface of the nanoparticles are faster
than dynamics in bulk solution. Additionally, dynamics at the
surface show no isotope effect. Finally, the overall Stokes shift
for the probe dye molecule at the surface is smaller than in
bulk solution. These results suggest that the hydrogen-bonding

7852 J. Phys. Chem. B, Vol. 103, No. 37, 1999
Pant and Levinger
(33) Maroncelli, M.; Fleming, G. R. J. Chem. Phys. 1988, 89, 5044.
(34) Bader, J. S.; Chandler, D. Chem. Phys. Lett. 1989, 157, 501.
(35) Schwartz, B. J.; Rossky, P. J. J. Chem. Phys. 1994, 101, 6902.
(36) Schwartz, B. J.; Rossky, P. J. J. Phys. Chem. 1995, 99, 2953.
(37) Schwartz, B. J.; Rossky, P. J. J. Chem. Phys. 1996, 105, 6997.
(38) Jimenez, R.; Fleming, G. R.; Kumar, P. V.; Maroncelli, M. Nature
1994, 369, 471.
(39) Jarzeba, W.; Walker, G. C.; Johnson, A. E.; Kahlow, M. A.;
Barbara, P. F. J. Phys. Chem. 1988, 92, 7039.
(40) Castner, E. W., Jr.; Chang, Y. J.; Chu, Y. C.; Walrafen, G. E. J.
Chem. Phys. 1995, 102, 653.
network normally found in these solutions is substantially
disrupted at the particle surface. Further experiments are
currently underway to probe the universality of these results
for other surfaces, other solvents and pH values.
Acknowledgment. We gratefully acknowledge useful con-
versations with Prof. G. V. Hartland and Dr. I. B. Martini during
the course of this work. This work was supported by the
National Science Foundation. N.E.L. is the recipient of a NSF
Young Investigator award.
(41) Riter, R. E.; Willard, D. M.; Levinger, N. E. J. Phys. Chem. B
1998, 102, 2705.
(42) Yanagimachi, M.; Tamai, N.; Masuhara, H. Chem. Phys. Lett. 1992,
200, 469.
(43) Bessho, K.; Uchida, T.; Yamauchi, A.; Shioya, T.; Teramae, N.
Chem. Phys. Lett. 1997, 264, 381.
(44) Loughnane, B. J.; Farrer, R. A.; Fourkas, J. T. J. Phys. Chem. B
1998, 102, 5409.
(45) Loughnane, B. J.; Fourkas, J. T. J. Phys. Chem. B 1998, 102, 10288.
(46) Riter, R. E.; Undiks, E. P.; Levinger, N. E. J. Am. Chem. Soc. 1998,
120, 6062.
(47) Willard, D. M.; Riter, R. E.; Levinger, N. E. J. Am. Chem. Soc.
1998, 120, 4151.
(48) Pant, D.; Riter, R. E.; Levinger, N. E. J. Chem. Phys. 1998, 109,
9995.
(49) Pant, D.; Levinger, N. E. J. Phys. Chem. B 1999, in preparation.
(50) Ursenbach, C. P.; Calhoun, A.; Voth, G. A. J. Chem. Phys. 1997,
106, 2811.
References and Notes
(1) Hagfeldt, A.; Gra¨tzel, M. Chem. ReV. 1995, 95, 49.
(2) Cherepy, N. J.; Shreve, A. P.; Moore, L. J.; Boxer, S. G.; Mathies,
R. A. J. Phys. Chem. B 1997, 101, 3250.
(3) Rehm, J. M.; Mclendon, G. L.; Nagasawa, Y.; Yoshihara, K.;
Moser, J.; Gratzel, M. J. Phys. Chem. 1996, 100, 9577.
(4) Kamat, P. V. Chem. ReV. 1993, 93, 267.
(5) Burfeindt, B.; Hannappel, T.; Storck, W.; Willig, F. J. Phys. Chem.
1996, 100, 16463.
(6) Martini, I.; Hodak, J. H.; Hartland, G. V. J. Phys. Chem. B 1998,
102, 602.
(7) Tachibana, Y.; Moser, J. E.; Gra¨tzel, M.; Klug, D. R.; Durrant, J.
R. J. Phys. Chem. 1996, 100, 20056.
(8) Hannappel, T.; Burfeindt, B.; Storck, W.; Willig, F. J. Phys. Chem.
B 1997, 101, 6799.
(9) 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.
(10) Ghosh, H. N.; Asbury, J. B.; Lian, T. J. Phys. Chem. B 1998, 102,
6482.
(11) Hilgendorff, M.; Sundstro¨m, V. Chem. Phys. Lett. 1998, 287, 709.
(12) Heimer, T. A.; Heilweil, E. J. Book of Abstracts 217th ACS National
Meeting, 1999.
(13) Thompson, D. W.; Kelly, C. A.; Farzad, F.; Meyer, G. J. Langmuir
1999, 15, 650.
(14) Moser, C. C.; Sension, R. J.; Szarka, A. Z.; Repinec, S. T.;
Hochstrasser, R. M.; Dutton, P. L. Chem. Phys. 1995, 197, 343.
(15) Martini, I.; Hodak, J. H.; Hartland, G. V.; Kamat, P. V. J. Chem.
Phys. 1997, 107, 8064.
(51) Benjamin, I. In Reaction Dynamics in Clusters and Condensed
Phases; Jortner, J., Eds.; Kluwer Academic Publishers: Netherlands, 1994;
p 179.
(52) Michael, D.; Benjamin, I. J. Phys. Chem. B 1998, 102, 5145.
(53) Chandra, A.; Senapati, S.; Sudha, D. J. Chem. Phys. 1998, 109,
10439.
(54) Hartnig, C.; Witschel, W.; Spohr, E. J. Phys. Chem. B 1998, 102,
1241.
(55) Shinto, H.; Sakakibara, T.; Higashitani, K. J. Phys. Chem. B 1998,
102, 1974.
(56) Pant, D.; Levinger, N. E. Chem. Phys. Lett. 1998, 292, 200.
(57) Colombo, D. P., Jr.; Roussel, K. A.; Saeh, J.; Skinner, D. E.;
Cavaleri, J. J.; Bowman, R. M. Chem. Phys. Lett. 1995, 232, 207.
(58) Martini, I.; Hodak, J. H.; Hartland, G. V. J. Phys. Chem. B 1998,
102, 9508.
(16) Martini, I.; Hartland, G. V.; Kamat, P. V. J. Phys. Chem. B 1997,
101, 4826.
(17) Barbara, P. F.; Meyer, T. J.; Ratner, M. A. J. Phys. Chem. 1996,
100, 13148.
(18) Gauduel, Y. J. Mol. Liq. 1995, 63, 1.
(19) Kimura, Y.; Alfano, J. C.; Walhout, P. K.; Barbara, P. F. J. Am.
Chem. Soc. 1994, 98, 3450.
(20) Lazafame, J. M.; Palese, S.; Wang, D.; Miller, R. J. D.; Muenter,
A. A. J. Phys. Chem. 1994, 98, 11020.
(21) Miller, R. J. D.; McLendon, G. L.; Nozik, A. J.; Schmickler, W.;
Willig, F. Surface Electron-Transfer Processes; VCH Publishers: New
York, 1995.
(22) Stratt, R. M.; Cho, M. H. J. Chem. Phys. 1994, 100, 6700.
(23) Castner, E. W., Jr.; Maroncelli, M.; Fleming, G. R. J. Chem. Phys.
1987, 86, 1090.
(24) Kahlow, M. A.; Jarzeba, W.; Kang, T. J.; Barbara, P. F. J. Chem.
Phys. 1989, 90, 151.
(25) Horng, M. L.; Gardecki, J. A.; Papazyan, A.; Maroncelli, M. J.
Phys. Chem. 1995, 99, 17311.
(59) Jackson, J. D. Classical Electrodynamics, 2nd ed.; John Wiley and
Sons: New York, 1975.
(60) Song, X. Y.; Chandler, D.; Marcus, R. A. J. Phys. Chem. 1996,
100, 11954.
(61) Mccarthy, P. K.; Blanchard, G. J. J. Phys. Chem. 1993, 97, 12205.
(62) Riddick, J. A.; Bunger, W. B.; Sakano, T. K. Organic SolVents,
4th ed., Vol. 2; John Wiley and Sons: New York, 1986.
(63) Sanesi, M.; Cremante, G.; Pizzini, S.; Wagner, V. Z. Naturforsch.
A 1971, 26, 159.
(64) Herz, A. H. AdV. Coll. Interface Sci. 1977, 8, 237.
(65) Barbara, P. F.; Walker, G. C.; Kang, T. J.; Jarzeba, W. Proc. SPIE
1990, 1209, 18.
(66) Nemethy, G.; Scheraga, H. J. Chem. Phys. 1964, 41, 680.
(67) The Structure and Properties of Water; Eisenberg, D., Kauzman,
W., Eds.; Oxford University Press: London, 1969.
(68) Kaatze, U. Chem. Phys. Lett. 1993, 203, 1.
(69) Das, S.; Datta, A.; Bhattacharyya, K. J. Phys. Chem. A 1997, 101,
3299.
(70) Zolotov, B.; Gan, A.; Fainberg, B. D.; Huppert, D. Chem. Phys.
Lett. 1997, 265, 418.
(71) Pal, H.; Nagasawa, Y.; Tominaga, K.; Kumazaki, S.; Yoshihara,
K. J. Chem. Phys. 1995, 102, 7758.
(26) Cho, M.; Rosenthal, S. J.; Scherer, N. F.; Ziegler, L. D.; Fleming,
G. R. J. Chem. Phys. 1992, 96, 5033.
(27) Chang, Y. J.; Castner, E. W., Jr. J. Chem. Phys. 1993, 99, 113.
(28) Neelakandan, M.; Pant, D.; Quitevis, E. L. J. Phys. Chem. A 1997,
101, 2936.
(29) Neelakandan, M.; Pant, D.; Quitevis, E. L. Chem. Phys. Lett. 1997,
265, 283.
(30) Joo, T. H.; Jia, Y. W.; Yu, J. Y.; Lang, M. J.; Fleming, G. R. J.
Chem. Phys. 1996, 104, 6089.
(31) Lang, M. J.; Jordanides, X. J.; Song, X.; Fleming, G. R. J. Chem.
Phys. 1998, 110, 5884.
(32) Maroncelli, M.; Kumar, P. V.; Papazyan, A.; Horng, M. L.;
Rosenthal, S. J.; Fleming, G. R. In Ultrafast Reaction Dynamics and SolVent
Effects; Gauduel, Y., Rossky, P. J., Eds.; American Institute of Physics:
Woodbury, NY, 1994; p 310.
(72) Shirota, H.; Pal, H.; Tominaga, K.; Yoshihara, K. J. Phys. Chem.
1996, 100, 14575.
(73) Schwartz, B. J.; Rossky, P. J. J. Chem. Phys. 1994, 101, 6917.
(74) Fonseca, T.; Ladanyi, B. M. J. Mol. Liq. 1994, 60, 1.
(75) Svishchev, I. M.; Kusalik, P. G. J. Phys. Chem. 1994, 98, 728.
(76) Nandi, N.; Roy, S.; Bagchi, B. J. Chem. Phys. 1995, 102, 1390.
(77) Pryor, B. A.; Palmer, P. M.; Andrews, P. M.; Berger, M. B.; Topp,
M. R. J. Phys. Chem. A 1998, 102, 3284.
(78) Mandel, F. S.; Spencer, H. G. J. Colloid Interface Sci. 1980, 77,
577.