Controlling surface defects and photophysics in TiO2
nanoparticles
Manuel J. Llansola-Portoles§*, Jesse J. Bergkamp§, Daniel Finkelstein-Shapiro§, Benjamin D.
Sherman§, Gerdenis Kodis§, Nada M. Dimitrijevic†, Devens Gust§, Thomas A. Moore§, Ana L.
Moore§.

§

Department of Chemistry and Biochemistry, Center for Bioenergy and Photosynthesis,

Arizona State University, Tempe, Arizona 85287-1604, USA. *e-mail: mjllansola@asu.edu
†

Argonne National Laboratory, Chemical Sciences and Engineering Division and Center of
Nanoscale Materials, 9700 S. Cass Ave. Argonne, IL 60439, USA.

Corresponding Author
*Manuel J. Llansola-Portoles. e-mail: mjllansola@asu.edu

KEYWORDS: surface defects, Ti3+, electron transfer, perylenes, dye-sensitized solar cell, optical
properties.

1

ABSTRACT
Titanium dioxide (TiO2) is widely used for photocatalysis and solar cell applications, and the
electronic structure of bulk TiO2 is well understood. However, surface structure of
nanoparticulate TiO2, which has a key role in properties such as solubility and catalytic activity,
still remains controversial. Detailed understanding of surface defects structures may help explain
reactivity and overall materials performance in a wide range of applications. In this work we
address the solubility problem and surface defects control on TiO2 nanoparticles. We report the
synthesis and characterization of ~ 4 nm TiO2 anatase spherical nanoparticles that are soluble
and stable in a wide range of organic solvents and water. By controlling the temperature during
the synthesis, we are able to tailor the density of defect states on the surface of the TiO2
nanoparticles without affecting parameters such as size, shape, core crystallinity and solubility.
The morphology of both kinds of nanoparticles was determined by TEM. EPR experiments were
used to characterize the surface defects, and transient absorption measurements demonstrate the
influence of the TiO2 defect states on photoinduced electron transfer dynamics.

2

INTRODUCTION
Titanium dioxide (TiO2) has long been a workhorse of artificial solar energy research. It was
used in the first photocatalytic water splitting device,1 in electrodes in dye-sensitized solar cells2,
in bio-inorganic systems3 and as the semiconductor of choice for environmental remediation
applications.4-7 It owes its success to its high stability, non-toxicity, adequate alignment of its
energy bands edges and particular defect structure which endows it with highly reactive sites.8
The electronic structure of bulk TiO2 is now well understood,7-8 whereas the structure and
properties of the surface remain controversial.8-9 The TiO2 surface consists of high-energy bond
configurations in which the terminal Ti and O atoms rearrange so that the symmetry and
electronic structure of the resulting surface is different from that of the bulk10 creating states that
would be considered defects in the bulk material. These defect states are composed of
undercoordinated Ti4+ sites resulting from surface curvature, and oxygen vacancies resulting in
Ti3+ sites.11 Surface defects influence the interfacial charge-transfer kinetics,12-13 modify the
reactivity in catalytic processes14 and affect the gas-adsorption distribution constants (CO, CO2,
H2, and H2O).15 The energy levels resulting from changes in the electronic structure of
reconstructed surface species are found in the mid-bandgap region.16 The diversity of exposed
facets, surface terminations and defect types present within a given polymorph of TiO2 (anatase,
rutile and brookite) give rise to radically different behavior.8 Defects in the TiO2 surface act as
both anchoring sites and charge injection/recombination sites and thereby play a crucial role in
the electron injection and recombination dynamics of the material.5,11,17 In addition to controlling
the surface reactivity, the defects may affect the electronic structure in a significant fraction of

3

the atoms of TiO2. For example, in a 4 nm TiO2 particle roughly 31% of the atoms are on the
surface.18
There are several methods used for creating surface defects in TiO2: UV radiation,19 annealing
in vacuum,20 ion sputtering,20 plasma-treating21 and controlling the calcination atmosphere
during synthesis.15 Unfortunately, these methods affect not only surface properties, but also size,
shape, and crystallinity of nanoparticles. Several low-temperature techniques for synthesizing
TiO2 nanoparticles have been developed using the following approaches: sol-gel methods,22
formation from Ti-alkoxide precursors in aqueous23 or organic media,24 TiCl4 hydrolysis at low
temperature,25 and a non-hydrolytic combination of Ti-alkoxide and TiCl4.26 However, these
methods have given rise to issues such as controlling the reaction rates which lead to the loss of
microstructural control22 and obtaining nanoparticles which are not soluble in organic solvents.
Therefore, in order to perform experiments in different media than water, their dispersion can
only be achieved by constructing a complex system involving reverse micelles which have
intrinsic problems such as surface crowding, micelle-chromophore interactions and excess free
micelles in solution.27 The method described in this manuscript presents the possibility of
controlling the density of surface defects on the nanoparticles and allows to use TiO2
nanoparticles without micelles in non-aqueous environments, expanding the available conditions
for device optimization and for some applications for which organic solvent is required.2
Herein, we report a strategy for obtaining small TiO2 anatase nanoparticles, which offers
advantages due to their solubility in a wide range of organic solvents and water. By carrying out
the procedure in the presence of organic solvent, the surface becomes terminated by methyl
4

groups covalently bound to the lattice, making these nanoparticles, to the best of our knowledge,
the first micelle-free TiO2 nanoparticles highly soluble in organic solvents, as well as water. The
resulting nanoparticles are stable over long periods of time, and can be dried and redispersed in
the solvent of choice without the irreversible aggregation typical of other TiO2 nanoparticules.
Furthermore, we demonstrate control over the surface defect density while keeping parameters
such as the number of methyl groups on the surface, size, shape and crystalline phase essentially
constant, producing nanoparticles which span the range from minimal surface defects to highly
defective (with red-shifted absorption). The transient absorption experiments highlight the big
role that the surface defects plays on charge transfer dynamics. Hence, our system opens the
possibility of exploring the interaction of defects with sensitizers in a variety of solvents in
colloidal suspension.
EXPERIMENTAL SECTION
Materials. Titanium tetrachloride (≥97.0%), dichloromethane (anhydrous, 99.5%), acetonitrile
(anhydrous 99.9%) and methanol (anhydrous 99.8%) were purchased from Sigma-Aldrich and
used as received.
High Resolution Transmission Electron Microscopy. TEM micrographs were obtained using
a Philips CM200 TEM at 200kV, Cs 1.2 mm, PTP Resolution: 0.25nm Focused Probe: 0.5nm
and Imaging Modes: TEM/STEM. TEM micrographs were analyzed using Digital Micrograph™
software28
Electron Paramagnetic Resonance. Continuous wave electron paramagnetic resonance (CWEPR) spectra were acquired with a Bruker Elexsys E580 spectrometer equipped with an Oxford
5

CF935 helium flow cryostat with an ITC-5025 temperature controller. Samples were placed into
a quartz EPR tube and bubbled with argon for 15 min to remove air/oxygen. Excitation of the
sample in the EPR spectrometer was carried out with a broadband Xenon lamp equipped with
water as an IR cut-off filter.
Steady state absorption. Spectra were recorded in 1.0 cm path length cuvette with the
following spectrophotometers: Steady-state absorption spectra were measured on a Shimadzu
UV2550 UV-visible spectrometer.
Spectroelectrochemical experiments. Spectroelectrochemical experiments were conducted
using 1 cm quartz cuvettes under a constant Ar flow, a Shimadzu UV-3101PC
spectrophotometer, and a CH Instruments 650C potentiostat. An indium doped tin oxide (ITO)
working electrode, Ag+/Ag quasi-reference electrode, and platinum wire counter electrode were
set inside the quartz cuvette. Approximately 0.4 mL of dichloromethane containing 100 mM
TBAPF6 and dye was added to the cuvette, and the solution level was maintained during the
experiment by addition of neat dichloromethane to account for evaporative losses. Attachment of
a glass cover slip over the conductive face of the ITO with heat-shrink plastic allowed the cavity
between the ITO and cover slip to fill by capillary action when the sample was immersed in
solution, and thereby provided a small volume of analyte on the working electrode. The beam of
the spectrophotometer was manipulated to measure only the solution contained in the working
electrode and the small volume ensured bulk electrolysis of the solution in the path length during
the course of the experiment. The absorbance spectrum of the singly oxidized form of perylene
dye was measured by applying positive potential (determined by a prior CV scan of the setup)
for 15 min.
6

Time resolved absorption. Femtosecond to nanosecond transient absorption measurements
were acquired with a kilohertz pulsed laser source and a pump-probe optical setup. Laser pulses
at 800 nm (c.a. 100 fs) were generated from an amplified, mode-locked titanium sapphire
kilohertz laser system (Millennia/Tsunami/Spitfire, Spectra Physics). Part of the laser pulse
energy was sent through an optical delay line and focused onto a 3 mm sapphire plate to generate
a white light continuum for the probe beam. The remainder of the pulse energy was used to
pump an optical parametric amplifier (Spectra Physics) to generate excitation pulses at 650 nm,
which were selected using a mechanical chopper. The white light generated was compressed by
prism pairs (CVI) before passing through the sample. The polarization of the pump beam was set
to the magic angle (54.7°) relative to the probe beam and its energy was adjusted to 200 nJ using
a continuously variable neutral density filter. The beams were focused in a 2 mm path-length
quartz cuvette to a ~200 µm diameter spot. The white light probe was dispersed by a
spectrograph (300 line grating) onto a charge-coupled device (CCD) camera (DU420, Andor
Tech.). The final spectral resolution was about 2.3 nm for over a nearly 300 nm spectral region.
The instrument response function was ca. 100 fs. Perylene-TiO2 systems were dissolved in
acetonitrile to an absorbance of 0.8 at the excitation wavelength in 2 mm path-length cuvette.
The decay-associated spectra (DAS) were obtained by fitting globally the transient absorption
kinetic traces over a selected wavelength region simultaneously as described by equation (1)
(parallel kinetic model).29

∆A=
(λ, t )

7

n

∑ A ( λ ) exp ( −t / τ )
i =1

i

i

(1)

where ∆A(λ,t) is the observed absorption change at a given wavelength at time delay t and n is
the number of kinetic components used in the fitting. A plot of Ai(λ) versus wavelength is called
a Decay Associated Spectra (DAS) and represents the amplitude spectrum of the ith kinetic
component, which has a lifetime of τi. Random errors associated with the reported lifetimes
obtained from transient absorption measurements were typically ≤ 5%. Rigorous analysis of
absorption data in a non-homogeneous system requires the use of a model which considers a
distribution of multi-exponential decays to describe the kinetics of each transient species and
thus accounting for the distribution of available environments in the heterogeneous system. In
our analysis we used the minimum number of exponential components for an adequate fitting of
the experimental data within the experimental error. Thus it is possible that the reported decay
components correspond to a weighted mean value of the actual distribution of constants
associated with each species.
RESULTS AND DISCUSSION
The nanoparticles were prepared in argon atmosphere using 80 mL of dichloromethane as a
solvent. It was kept at 195 K for the preparation of TiO2 nanoparticles with low density of defect
states (L-TiO2) and at 313 K for TiO2 nanoparticles with high density of defect states (R-TiO2).
Under continuous stirring, 200 µL of TiCl4 were introduced into the reactor and 3 mL of
methanol were added dropwise and allowed to react for 2 h. The synthesis of both L-TiO2 and RTiO2 follow identical steps with the exception of the reaction temperature (see supporting
information). The morphologies of L-TiO2 and R-TiO2 were studied by high resolution
transmission electron microscopy (HRTEM). Figure 1 shows nearly spherical particles with the
8

size distribution presented in the inset histogram. Fitting the size distribution to a Gaussian
model, mean diameters of 3.8 nm (FWHM = 1.3 nm) and 3.7 nm (FWHM = 1.7 nm) for L-TiO2
and R-TiO2 were obtained, respectively. Although both samples have the same average size; RTiO2 particles have a larger size dispersion, which we attribute to the faster reaction rate during
the preparation, which does not allow the nanoparticles to grow as uniformly as they do at lower
temperature. Lattice planes were determined analyzing the profile of the nanoparticles of
HRTEM images in Figure 1 (profiles shown in supporting information). For both kinds of
nanoparticles lattice planes of 1.6 ± 0.1 Å and 2.4 ± 0.1 Å were found, which matches the
literature values of 1.6 Å and 2.4 Å reported for the (211) and (004) planes of tetragonal
anatase.30-32 Confirmation that anatase is the crystalline phase for both R-TiO2 and L-TiO2
nanoparticles was demonstrated by TEM results, EPR measurements and UV-vis spectra (vide
infra). Such results are in agreement with the observations of Durupthy and co-workers who
showed that anatase is the most stable phase for TiO2 nanoparticles smaller than 14 nm.33

9

Figure 1: HRTEM micrographs of TiO2 nanoparticles a) High magnification image showing
crystallinity in the L-TiO2. Inset: Histogram of size distribution and Gaussian fit (black line). b)
High magnification image showing crystallinity in the R-TiO2. Inset: Histogram of the size
distribution and Gaussian fit (black line).

10

Figure 2 shows the UV-Vis absorption spectra of R-TiO2 and L-TiO2 at a concentration of 250
mg/mL in acetonitrile. The band gap energy was determined using equation (2),34 (Tauc plot;
inset Figure 2) where α is the absorption coefficient, k is a constant, h is Planck's constant, υ
represents the frequency of incident light and Eg is the band gap energy in electron volts.

(α h=
υ ) 2 k (hυ − Eg )

(2)

The value obtained (3.25 ± 0.05) eV is the same for R-TiO2 (red dots) and L-TiO2 (black dots),
indicating that the core of both samples is crystalline phase anatase with no appreciable sizebased quantum effects.8,35-37 L-TiO2 shows negligible light absorption at energies lower than the
band gap energy, suggesting a low density of defect states, whereas R-TiO2 exhibits a strong
absorption in the region from 400 nm to 650 nm, which is attributed to transitions involving
defect states at the surface. Indeed, band gap absorption (λ < 400 nm) remains constant for all
solvents whereas the characteristics of absorption attributed to defect states (400 nm < λ < 650
nm) depend on the surrounding solvent. This behavior strongly suggests that the band gap
absorption is due to the anatase core (not in contact with the solvent), whereas the absorption due
to defect states arises in solvent-contacted surface regions (see Supporting Information, Figure
S7).

11

Figure 2: Absorption spectra of L-TiO2 (black line) and R-TiO2 (red line) in acetonitrile at a
concentration of 250 mg/mL. Inset: Absorption onset value of the first direct allowed transition
for L-TiO2 (black line) and R-TiO2 (red line). Intersection of the lines on the x axis at y = 0
indicates the estimated band gap energy and blue lines represent the confidence bands of 99%
for the fitting using equation (2).
Continuous wave electron paramagnetic resonance (CW-EPR) measurements were
carried out at 4.5 K to characterize the structure and nature of the paramagnetic centers created in
R-TiO2 and L-TiO2 upon UV-vis irradiation. The EPR spectrum for nanoparticles without
illumination reveals the presence of Ti3+ and oxygen vacancies on the surface of R-TiO2, which
are absent in L-TiO2 (Figure 3a). This supports the observation from UV-vis spectroscopy that
R-TiO2 nanoparticles have indeed a higher density of defect states. Under UV-vis illumination
we find that the photogenerated charges localize on the surface in both type of nanoparticles.
Oxidizing equivalents exist as oxygen-centered radicals, (Ti4+-O•) and reducing equivalents as
12

Ti3+ centers (Figure 3b). The g-tensors of photogenerated charges (gz = 2.041, gy = 2.012, gx =
2.002 for O-centered radicals and g = 1.9344 for Ti3+ for L-TiO2, and gz = 2.047, gy = 2.016, gx =
2.002 for O-centered radicals and g = 1.9071 for Ti3+ for R-TiO2) agree with those of an anatase
structure. The trapping of electrons on the surface at 4.5 K is characteristic of anatase particles
with small domains of crystallinity and/or a high density of five-coordinated surface Ti4+ that are
formed as a result of high curvature of the nanoparticles.38-39 These undercoordinated sites
preferentially act as electron traps on the surface and become reduced to the paramagnetic center
Ti3+. Normalizing the signals to the oxygen-centered radicals clearly shows that the density of
photoexcited Ti3+ centers (arising from reduced uncoordinated Ti4+ on the surface) is larger for
R-TiO2 than L-TiO2. The double integral of the CW-EPR spectrum normalized to the value of
the oxygen radical signal indicates that the concentration of photoexcited states in R-TiO2 is 34%
higher than in L-TiO2. Moreover, the presence of oxygen vacancies at the surface of R-TiO2
affects the overall electron density distribution, as seen in different g-tensor values for
photogenerated charges localized on the surface of L-TiO2 and R-TiO2. It is likely that these
oxygen vacancies stabilize a larger number of undercoordinated Ti4+ in R-TiO2.
To further study the surface, we examined the adsorption of the superoxide anion and perylene
dye (vide infra) on both kinds of nanoparticles.38-39 The superoxide anion is created by exposure
of the nanoparticles to ozone for five h. The observed superoxide anion exhibits g-tensor values
of gz=2.0217, gy=2.0091, gx=2.0024 for R-TiO2 and gz=2.0193, gy=2.0091, gx=2.0024 for L-TiO2
(Figure 3c).39-41 Superoxide anion (O2–) is a 13-electron π radical characterized by the presence
of three electrons in two 2π* antibonding orbitals. Despite the presence of an unpaired electron
in this radical, the CW-EPR spectrum is only observed when the degeneracy of both π* orbitals
13

is removed by external perturbations, such as an external electric field. The splitting (ΔE) of the
π* levels determines the value of the gz component (gz = ge + 2λ/ΔE). Thus the gz value depends
on the nature of the adsorption site and its local electric field. The difference observed for the
values of the gz component confirms different local environments at the surface for the R-TiO2
and L-TiO2 nanoparticles.38,40-42 The system dye–R-TiO2 shows evidence of charge transfer (CT)
complex43 formation, whereas dye–L-TiO2 does not, suggesting the presence of strong electronic
coupling between near-by defects and the dye in the dark (see Supporting Information Figure
S6).

Figure 3. a) Dry L-TiO2 and R-TiO2 in the dark. We observe the presence of oxygen vacancies
and (Ti3+)surf centers in R-TiO2. The sharp feature (*) is an artifact of the cavity signal
14

subtraction. b) CW-EPR of dry L-TiO2 and R-TiO2 upon UV illumination. Both oxygen centered
radicals, (Ti4+-O•), and surface (Ti3+)surf radicals are observed. Spectra are normalized to the
oxygen radical signal. c) CW-EPR of adsorbed superoxide on R-TiO2 and L-TiO2 in the dark.
Superoxide is obtained by exposing the surface to ozone for five hours.
In order to determine the role of defect states in photoinduced electron transfer processes, a
perylene dye bearing a phosphonic acid group as a binding moiety (dye in Scheme 1) was
synthesized and attached to the surface of the nanoparticles.44 Scheme 1 depicts the relative
energetics for the oxidation of the ground and excited states of the dye, the conduction band edge
(CB), valence band edge (VB) and defect state energy levels of TiO2.45 The potential of the
defect states of TiO2 were calculated using the UV-vis absorption spectra of R-TiO2 in
acetonitrile. The potential for the first oxidation (D•+/D) of the dye was measured
electrochemically (Supporting Information, Figure S5). The potential for the oxidation of the
excited state (D•+/D*) was calculated with the equation D•+/D* = D•+/D ‒ E00, where E00 is the
zero-zero optical excitation energy estimated from the intersection of the normalized absorption
and emission spectra in acetonitrile (Supporting information, Figure S8).

15

Scheme 1. Diagram depicting the potentials for the ground and the lowest excited singlet
states of the perylene dye, TiO2 conduction band energy (CB), TiO2 valence band energy (VB)
and TiO2 energy levels due to defect states (red lines). Broad arrows indicate that electrons can
be injected from the exited state of the dye into different trap states and that electrons can
recombine from different trap states into the ground state of the oxidized dye.
As shown in Scheme 1, the excited singlet dye species is thermodynamically capable of
photoinjecting an electron into the conduction band and/or defect states of TiO2. Theoretical
work affirms that defect states may act as electron acceptors and thereby increase the driving
force for the photoinduced electron transfer process.46-47 Nevertheless, there are no experimental
studies comparing the dynamics of the electron transfer processes in cases where the density of
defect states is the only variable in the system. It is expected that for L-TiO2, the CB is the most
probable electron-accepting energy level and electron transfer would follow the pathway
16

depicted by the blue arrows in Scheme 1: injection from the excited dye into the CB and
recombination from the CB to the oxidized dye. For R-TiO2, electron injection can take place
into both CB, indicated by the blue arrow, and into a wide distribution of defect states as
indicated by the red broad arrow.
Femtosecond transient absorption experiments were performed to explore the role of defect
states in the kinetics of the photoinduced electron transfer reactions in the dye–TiO2 systems in
acetonitrile solution. The global analysis of transient absorption decay kinetics for the dye alone
(Figure 4a) gives two decay-associated spectra (DAS) with lifetimes of 1.6 ps and 1036 ps. The
1036 ps DAS shows negative amplitudes at around 700 and 770 nm corresponding to the ground
state bleaching and stimulated emission of the dye, respectively; therefore it can be assigned to
the decay of the singlet excited state. The 1.6 ps DAS shows a shift of the stimulated emission
band from about 730 nm (negative amplitude) to 780 nm (positive amplitude). This DAS can be
assigned to solvation (relaxation) of the excited state of the dye. Inset 5a shows the difference
absorption spectrum of the singly-oxidized dye cation radical (dye•+) obtained by
spectroelectrochemistry of the dye free in solution. It has strong induced absorption on both sides
of the ground state bleaching (~ 700 nm) of the dye.
The global analysis of the transient absorption data for the dye-associated L-TiO2 (Figure 4b)
gives four DAS with lifetimes of ~200 fs, 5 ps, 65 ps and 463 ps. The 200 fs DAS can be
attributed to electron injection from the singlet excited state of the dye into the conduction band
of the L-TiO2. It shows the decay of stimulated emission around 720-800 nm and simultaneously
the formation of dye radical cation indicated by the dye•+ absorption characteristics presented
17

around 540-600 nm (see 4a inset). The 5 ps DAS is similar to the difference absorption spectrum
of the dye•+ obtained by spectroelectrochemistry in solution and therefore can be attributed to the
recombination of the charge separated state that was formed with a time constant of 200 fs. The
65 ps DAS shows positive amplitudes around 610 and 720 nm, these features are not
characteristics of the dye•+ however this DAS can be attributed to dye solvation (relaxation) and
cooling after very fast (5 ps) ground state recovery. For the 65 ps DAS, the ground state
solvation can be attributed to the blue shift of ground state bleaching (observed negative
amplitude at 690 nm and positive amplitude at 610 nm) and the ground state cooling can be
attributed to induced absorption from the ground state at wavelengths characteristic of the
stimulated emission (observed positive amplitude at 720 nm). The 463 ps DAS can be attributed
to the decay of the singlet excited state of the weakly attached (physisorbed) and aggregated dye.
It shows ground state bleaching, stimulated emission, and featureless induced absorption below
620 nm due to absorption by the singlet excited state which is similar to that observed for the
free dye in solution (1036 ps DAS, Figure 4a). The dye•+–TiO2(e–) formation and recombination
can also be followed by the characteristic induced absorption at 990 nm due to the free carriers
(injected electrons) in the TiO2 semiconductor

48

(see Supporting Information, Figure S7). It

should be emphasized that because of strong dye–TiO2 system heterogeneity the observed DAS
cannot be attributed to pure transient species and can have contributions from other species
decaying with the same lifetime. For example, the 5 ps DAS most likely has some contribution
from the physisorbed dye excited state solvation observed for the dye alone in solution and the
463 ps DAS has some contribution from the dye ground state relaxation and possibly from some
species that undergo slower charge separated state recombination.
18

The global analysis of the transient absorption for the dye–R-TiO2 (Figure 4c) gives a
satisfactory fit with five decay lifetimes of ~100 fs, 1.5 ps, 11 ps, 228 ps and a component that
does not decay within the 3.5 ns measurement time window. The dye–R-TiO2 system shows
even higher heterogeneity. The ~100 fs DAS can be attributed to the formation of charge
separated states involving the TiO2 CB and shallow defect states. Induced absorption at the blue
and red edges of the spectrum shows that some charge separation already happened on a much
faster time scale, presumably involving the lower energy (deeper) defect states, with more
driving force. The 11 ps, 228 ps and nondecaying DAS are associated with recombination of
charge separated species. The 1.5 ps DAS most likely shows the excited state solvation of the
weakly attached dye as was observed for the free dye in solution (Figure 4a). However the
negative amplitude around 590 nm it must also have some contribution from species which form
charge separated state on a slower time scale, presumably from weakly coupled species at
specific defect sites. The 228 ps DAS apart from features characteristic of the charge separated
state as observed in the nondecaying DAS, which corresponds to the pure charge separated
species, has features characteristic of the dye ground state solvation and cooling as was observed
for the dye–L-TiO2 system. In addition, it has a contribution from the singlet excited state decay
of physisorbed aggregated dye (negative amplitude due to stimulated emission at ~780 nm).

19

Figure 4. Decay-associated-spectra in acetonitrile solution measured by fs transient
absorption upon excitation at 650 nm for a) dye, b) dye–L-TiO2, c) dye–R-TiO2. Inset:
spectroelectrochemical difference spectrum in dichloromethane showing the difference in
absorption of the singly oxidized form of dye from its neutral ground state form.
The overall transient absorption results discussed above show evidence that dye–L-TiO2 and
dye–R-TiO2 systems both undergo photoinduced electron transfer to yield the species dye•+–LTiO2(e–) and species dye•+–R-TiO2(e–). The apparent forward (ki1) and reverse (kr1) rate constants
20

for the formation and decay of the dye•+–L-TiO2(e–) state can be estimated as ki1 ~ 5 × 1012 s–1
(1/200 fs) and kr1 ~ 2 × 1011 s–1 (1/5 ps), respectively. The apparent forward and reverse rate
constants for the formation and decay of the dye•+–R-TiO2(e–) state are difficult to estimate
because of the heterogeneity of the system, nevertheless the rates can be roughly approximated
as ki2 ~ 1 × 1013 s–1 (1/100 fs), kr2 ~ 9.1 × 1010 s–1 (1/ 11 ps (50%)) and kr2 ~ 4.4 × 109 s–1 (1/ 228
ps (46%)), respectively (there is some even slower recombination, 4%, associated most likely
with some deeply trapped electrons). The faster injection for dye–R-TiO2 than for dye–L-TiO2
could be explained by the larger driving force, the larger density of states and stronger electronic
coupling at the defect sites (as shown in EPR measurements Figure S6, Supporting Information).
Stronger electronic coupling has been observed in other nanoparticulated TiO2 systems.49 This
electronic coupling possibly results in the formation of charge transfer excitons which facilitate
injection into the CB and/or distant defect states. Recombination is slower for the dye–R-TiO2
because the electron can escape from the initially formed charge separated state (CT exciton) and
migrate into other surface defect states, where it eventually gets trapped. This effect is consistent
with the well-known slower mobility of electrons in trap states than in the CB.50 On the other
hand, for dye–L-TiO2 the electrons have higher mobility in the conduction band and are more
readily available to recombine. The presence of a higher density of defect states leads to a tighter
coupling of the dye with the surface, increases the density of accepting states with larger driving
force for electron injection, and therefore facilitates injection and helps to stabilize the charge
separated state.

21

CONCLUSIONS
We are reporting a method for the preparation of TiO2 nanoparticles, which form stable
suspensions in a wide range of organic solvents at concentrations up to 250 mg/mL. By
controlling the temperature during synthesis, these 3-5 nm diameter TiO2 nanoparticles can be
prepared with either a high or a low density of defect states without affecting parameters such as
size, shape, core crystallinity and solubility. Due to the ability to control the defect states and the
excellent optical properties in a variety of organic solvents and water, these materials provide a
good platform for studies of photoinduced electron transfer reactions at interfaces. The important
role of surface defects in electron transfer dynamics has been proved. We show that there is a
well-marked difference in the behavior of the electron transfer dynamics as a function of the
density of defect sites. When the density of surface defects is low the electron follow a single
injection/recombination pathway. However, in high density surface defects nanoparticles there
are multiple injection paths and several recombination rates depending on the stabilization effect
of the surface trap on the charge separated state.
ACKNOWLEDGEMENTS
The work at ASU was supported by the Center for Bio-Inspired Solar Fuel Production, an
Energy Frontier Research Center funded by the U.S. Department of Energy, Office of Science,
Office of Basic Energy Sciences under Award Number DE-SC0001016, National Science
Foundation CHE-1124895 and the research at ANL was supported by the Department of Energy,
Office of Science, Office of Basic Energy Sciences, under Contract No DE-AC02-06CH11357.

22

ASSOCIATED CONTENT
Supporting information includes a reaction mechanism for formation of TiO2 nanoparticles,
FTIR-ATR, TEM latice fringes profiles, thermogravimetry, perylene dye synthesis and
characterization, binding of perylene mechanism to TiO2, electron paramagnetic resonance
experiments with dye binded to TiO2 and extended photophysical characterization. “This
material is available free of charge via the Internet at http://pubs.acs.org.” ko
AUTHOR INFORMATION
Corresponding Author
*To whom correspondence should be addressed to: mjllansola@asu.edu
Notes
The authors declare no competing financial interest.
REFERENCES
(1)
Fujishima, A.; Honda, K. Nature 1972, 238, 37.
(2)
O'regan, B.; Grätzel, M. Nature 1991, 353, 737.
(3)
Mershin, A.; Matsumoto, K.; Kaiser, L.; Yu, D.; Vaughn, M.; Nazeeruddin, M. K.;
Bruce, B. D.; Graetzel, M.; Zhang, S. Scientific Reports 2012, 2.
(4)
Fujishima, A.; Zhang, X. C. R. Chim. 2006, 9, 750.
(5)
Linsebigler, A. L.; Lu, G.; Yates, J. T. Chem. Rev. (Washington, DC, U. S.) 1995, 95,
735.
(6)
Fujishima, A.; Rao, T. N.; Tryk, D. A. J. Photochem. Photobiol., C 2000, 1, 1.
(7)
Hoffmann, M. R.; Martin, S. T.; Choi, W.; Bahnemann, D. W. Chem. Rev. (Washington,
DC, U. S.) 1995, 95, 69.
(8)
Diebold, U. Surf. Sci. Rep. 2003, 48, 53.
(9)
Thompson, T. L.; Yates, J. T. Chem. Rev. (Washington, DC, U. S.) 2006, 106, 4428.
(10) Diebold, U. Nat. Mater. 2010, 9, 185.
(11) Vijayan, B.; Dimitrijevic, N. M.; Rajh, T.; Gray, K. J. Phys. Chem. C 2010, 114, 12994.
23

Green, A. N. M.; Palomares, E.; Haque, S. A.; Kroon, J. M.; Durrant, J. R. J. Phys.
Chem. B 2005, 109, 12525.
(13) Gregg, B. A.; Chen, S.-G.; Ferrere, S. J. Phys. Chem. B 2003, 107, 3019.
(14) Gong, X.-Q.; Selloni, A.; Batzill, M.; Diebold, U. Nat. Mater. 2006, 5, 665.
(15) Suriye, K.; Praserthdam, P.; Jongsomjit, B. Appl. Surf. Sci. 2007, 253, 3849.
(16) Gutierrez, C.; Salvador, P. J. Electroanal. Chem. Interfacial Electrochem. 1982, 138,
457.
(17) Hurum, D. C.; Agrios, A. G.; Gray, K. A.; Rajh, T.; Thurnauer, M. C. J. Phys. Chem. B
2003, 107, 4545.
(18) Chen, L. X.; Rajh, T.; Wang, Z.; Thurnauer, M. C. J. Phys. Chem. B 1997, 101, 10688.
(19) Mezhenny, S.; Maksymovych, P.; Thompson, T. L.; Diwald, O.; Stahl, D.; Walck, S. D.;
Yates Jr, J. T. Chem. Phys. Lett. 2003, 369, 152.
(20) Pan, J. M.; Maschhoff, B. L.; Diebold, U.; Madey, T. E. J. Vac. Sci. Technol., A 1992, 10,
2470.
(21) Nakamura, I.; Negishi, N.; Kutsuna, S.; Ihara, T.; Sugihara, S.; Takeuchi, K. J. Mol.
Catal. A: Chem. 2000, 161, 205.
(22) Brinker, C. J.; Scherer, G. W. Sol-Gel Science: The Physics and Chemistry of Sol-Gel
Processing; Academic Press: San Diego, 1990.
(23) Kotov, N. A.; Meldrum, F. C.; Fendler, J. H. J. Phys. Chem. 1994, 98, 8827.
(24) Skinner, D. E.; Colombo, D. P.; Cavaleri, J. J.; Bowman, R. M. J. Phys. Chem. 1995, 99,
7853.
(25) Serpone, N.; Lawless, D.; Khairutdinov, R. J. Phys. Chem. 1995, 99, 16646.
(26) Arnal, P.; Corriu, R. J. P.; Leclercq, D.; Mutin, P. H.; Vioux, A. Chem. Mater. 1997, 9,
694.
(27) Hernandez, L. I.; Godin, R. P.; Bergkamp, J. J.; Llansola-Portoles, M. J.; Sherman, B. D.;
Tomlin, J.; Kodis, G.; Méndez-Hernández, D. D.; Bertolotti, S.; Chesta, C.; Mariño-Ochoa, E.;
Moore, A. L.; Moore, T. A.; Cosa, G.; Palacios, R. E. J. Phys. Chem. B 2013, 117, 4568.
(28) Mitchell, D. R.; Schaffer, B. Ultramicroscopy 2005, 103, 319.
(29) Van Stokkum, I. H. M.; Larsen, D. S.; Van Grondelle, R. Biochim. Biophys. Acta 2004,
1657, 82.
(30) Zhao, X.; Jin, W.; Cai, J.; Ye, J.; Li, Z.; Ma, Y.; Xie, J.; Qi, L. Adv. Funct. Mater. 2011,
21, 3554.
(31) Anthony, J. W.; Bideaux, R. A.; Bladh, K. W.; Nichols, M. C. In Handbook of
Mineralogy; Mineral Data Publishing: 1997; Vol. 3.
(32) Shi, J.; Sun, C.; Starr, M. B.; Wang, X. Nano Lett. 2011, 11, 624.
(33) Durupthy, O.; Bill, J.; Aldinger, F. Cryst. Growth Des. 2007, 7, 2696.
(34) Johnson, E. J. In Optical Properties of III-V Compounds, Semiconductors and
Semimetals; R.K. Willardson, A. C. B., Ed.; Academic Press: New York, 1967; Vol. 3, p 153.
(35) Vayssieres, L.; Persson, C.; Guo, J.-H. Appl. Phys. Lett. 2011, 99.
(36) Sant, P. A.; Kamat, P. V. Phys. Chem. Chem. Phys. 2002, 4, 198.
(37) Hegazy, A.; Prouzet, E. Chem. Mater. 2011, 24, 245.
(38) Green, J.; Carter, E.; Murphy, D. M. Chem. Phys. Lett. 2009, 477, 340.
(39) Maurelli, S.; Vishnuvarthan, M.; Berlier, G.; Chiesa, M. Phys. Chem. Chem. Phys. 2012,
14, 987.
24
(12)

Attwood, A. L.; Murphy, D. M.; Edwards, J. L.; Egerton, T. A.; Harrison, R. W. Res.
Chem. Intermed. 2003, 29, 449.
(41) Berger, T.; Sterrer, M.; Diwald, O.; Knözinger, E. ChemPhysChem 2005, 6, 2104.
(42) Berger, T.; Sterrer, M.; Diwald, O.; Knözinger, E.; Panayotov, D.; Thompson, T. L.;
Yates, J. T. J. Phys. Chem. B 2005, 109, 6061.
(43) Dimitrijevic, N. M.; Poluektov, O. G.; Saponjic, Z. V.; Rajh, T. J. Phys. Chem. B 2006,
110, 25392.
(44) Guerrero, G.; Mutin, P. H.; Vioux, A. Chem. Mater. 2001, 13, 4367.
(45) Grätzel, M. Nature 2001, 414, 338.
(46) Martsinovich, N.; Troisi, A. Phys. Chem. Chem. Phys. 2012, 14, 13392.
(47) Jakubikova, E.; Snoeberger Iii, R. C.; Batista, V. S.; Martin, R. L.; Batista, E. R. J. Phys.
Chem. A 2009, 113, 12532.
(48) Pankove, J. I. Optical Processes in Semiconductors; Courier Dover Publications, 1975.
(49) Varaganti, S.; Ramakrishna, G. J. Phys. Chem. C 2010, 114, 13917.
(50) Richter, C.; Schmuttenmaer, C. A. Nat. Nanotechnol. 2010, 5, 769.
(40)

TABLE OF CONTENTS

25