Paper

1. Electrochimica Acta 50 (2005) 5278–5291
Photoelectrocatalytic degradation of diazo dyes on
nanostructured WO3 electrodes
Maria Hepel∗, Sandra Hazelton
Department of Chemistry, State University of New York at Potsdam, Potsdam, NY 13676, USA
Received 23 December 2004; received in revised form 3 March 2005; accepted 3 March 2005
Available online 12 July 2005
Abstract
Photoelectrocatalytic degradation of Remazol Black B (RBB) dye in aqueous solutions was investigated. The degradation was carried out
on a highly efficient n-WO3 photoelectrode, which was synthesized by electrochemical processing. Effects of various synthesis and post-
treatment conditions were investigated using electrochemical techniques and AFM imaging. Photocurrent measurements have shown that
Butler–Gartner equation is fulfilled for n-WO3 films, which allowed us to determine the flat-band potential (Efb = 0.35 V versus Ag/AgCl,
for pH = 0). The WO3 film electrode was found to be a better photelectrocatalyst than TiO2 with similar AFM surface roughness. Kinetic
measurements of photoelectrocatalytic RBB degradation have shown that a generalized Langmuir–Hinshelwood model applies to the system,
leading to the overall rate constant k = 1.1 × 10−9
mol cm−2
s−1
, with RBB adsorption equilibrium constant K = 3.35 × 104
mol−1
L and Gibbs
free energy of adsorption G◦
= −6.16 kcal/mol (−25.8 kJ/mol). In mechanistic considerations of possible lines of oxidative attack on RBB,
we looked at electrolyte composition and properties of electronic structure of RBB, by utilizing quantum mechanical calculations of molecular
orbitals. The ab initio SCF Hartree–Fock calculations of molecular orbitals were carried out to evaluate most vulnerable areas of RBB molecule
for electrophilic and nucleophilic attacks, as well as for a direct charge transfer oxidation process. Several electronic properties were mapped
onto the total electron density surface and analyzed to reveal initial reaction pathways for oxidative degradation of RBB.
© 2005 Elsevier Ltd. All rights reserved.
Keywords: Photoelectrocatalysis; Tungsten trioxide photoelectrode; Azo dyes; Remazol Black B; Degradation of pollutants; Photoelectrochemistry; Adsorption
1. Introduction
Maintaining the natural human habitat in the era of global
industrializationrequiresimmenseresearcheffortstodevelop
pollution-free technologies and advise efficient methods for
cleaning,alreadypolluted,environment.Manyindustrialpro-
cesses release every day tons of toxic, carcinogenic, and
degenerative chemicals. Some of the pollutants remain in
the environment for long time. One of the ubiquitous envi-
ronmental pollutants are dyes used in textile and other indus-
tries [1–21]. The textile dyes do not degrade easily [15–17].
The dye degradation schemes have been widely investigated
[18–20] and, in recent years, considerable improvement in the
degradation efficiency has been achieved. It has been reported
that several organic pollutants can be degraded on illuminated
∗ Corresponding author. Tel.: +1 315 267 2264; fax: +1 315 267 3170.
E-mail address: hepelmr@potsdam.edu (M. Hepel).
semiconductor powders [1–21]. On the other hand, pho-
toelectrodes developed recently for water photoelectrolysis
[22–25], can also be used as efficient catalysts for degrading
pollutants. There has been a growing interest in the use of
semiconductors with sensitizers for the complete oxidative
mineralization of pollutants [1,25]. In contrast to semicon-
ductor photomineralization proposed as a method of water
purification, the current technologies used in water industry
[26], such as the chlorination and ozonation, utilize strong
oxidants of a seriously hazardous and undesirable nature.
In a photooxidative degradation process, an n-type semi-
conductor is illuminated with radiation having energy at least
equal to the bandgap (Eg) of the semiconductor. The pho-
tons absorbed excite electrons from the valence band (VB)
to the conduction band (CB), while holes are left in the
valence band. The holes in a semiconductor have a high
oxidizing power and can oxidize organic molecules in an
aqueous environment. These photoelectrons and photoholes
0013-4686/$ – see front matter © 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.electacta.2005.03.067

2. M. Hepel, S. Hazelton / Electrochimica Acta 50 (2005) 5278–5291 5279
may recombine producing thermal energy, but they may be
separated and engaged in driving chemical reactions in pho-
tochemical processes. The rate of oxidation by holes has to be
balanced by the rate of the reduction by electrons. The trans-
fer of photogenerated holes from the valence band to organic
molecules may be either isoenergetic or inelastic via bandgap
surface states. A theoretical and experimental studies of elec-
tron transport phenomena in dye-sensitized nanocrystalline
photovoltaic cells were performed by Peter and cowork-
ers [27,28]. Parsons et al. [29] investigated quantum well
electrodes for photoelectrochemistry and Tomkiewicz [30]
determined surface states induced on chemically modified
TiO2 electrodes. Surface states can act as the recombination
centers for photogenerated electron–hole pairs but may also
participate in sub-bandgap excitation processes and help uti-
lize more energy from the solar spectrum.
To improve photochemical degradation effectiveness of
organic pollutants on semiconductor powders, Serpone et al.
[14] proposed adding hydrogen peroxide to the reactor slurry.
Considerable enhancement of the photodegradation rate was
observed for trichloroethylene and tetrachloroethylene in
a suspension of TiO2. The augmentation of the rate was
attributed to the increased number of OH• radicals formed
on illuminated TiO2 [14]. Mechanistic aspects of the photo-
oxidative degradation of organic pollutants have recently
been discussed [31–35]. A rotating disk photocatalytic reac-
tor with TiO2 catalyst was designed by Dionysiou et al. [36]
and proposed for destruction of organic pollutants in water.
Electrochemical and photoelectrochemical methods are
very efficient for degrading organic pollutants [37–41].
The efficiency of photoelectrochemical degradation for dyes
depends not only on the selection of a suitable supporting
electrolyte and pH values, but also on the electrode potential
and preparation conditions of the semiconductors involved.
In a photoelectrochemical system, photoelectrons and pho-
toholes can be separated under the influence of an applied
electric field. The problem of the separation of semiconductor
particles from the treated solution, so persistent in heteroge-
neous photolysis, is not an issue in photoelectrochemical sys-
tems. There are numerous semiconductors which can be used
as photoelectrocatalytic materials, such as TiO2, WO3, SnO2,
ZnO, CdS, diamond, and others. The advantages of photo-
electrochemical degradation of dyes in water also include
the following: (i) the end products can be controlled and are
harmless to the environment, (ii) the degradation process can
be turned on or off at any time, and (iii) there is a real pos-
sibility that it could be readily incorporated into existing UV
water purification systems.
Tungsten trioxide is an important photoelectrocatalytic
material. The photoelectrochemical behavior of WO3 films
was extensively studied in this laboratory for electrochromic
applications [42,43], solar energy conversion [44,45], and
degradation of pollutants [31,46]. The electrodeposited
Pt/WO3 catalysts have also been found to have high activity
toward oxidation of methanol [47,48] and formic acid [49],
higher than benchmark Pt catalysts. The WOx films with Pt,
Sn, and Ru centers were used by Bock and MacDougall [50]
for electrooxidation of HCOOH and (COOH)2. Kulesza and
Faulkner [51] described electrocatalysis of chlorate reduction
on mixed valency WO3−x electrodes. The electrocatalytic
activity of Pt/WO3 electrode in phosphoric acid fuel-cells
was studied by Savadogo [52,53]. Augustynski et al. [54],
investigating photooxidation of methanol on thermal WO3
films, have found that at high methanol concentrations, hole
scavenging is the predominant process while at low methanol
concentration, an indirect photooxidation with the forma-
tion of hydrogen peroxide species takes place. Modulation
of absorbance spectra of composite films WO3–TiO2 were
described by de Tacconi et al. [55]. The photoelectrodes were
synthesized by pulsed electrodeposition. Krasnov and Kol-
basov [56] have found that index of refraction of electrode-
posited a-WO3 films increases during the film drying process,
indicating on chemisorbed water content. The Raman and
FT-IR spectra of WO3 films deposited on IrO2-coated Ti
substrates were published by Habazaki et al. [57] who also
found hydration water in “as-deposited” films. The growth
kinetics of anodic WO3 films was investigated by Metikos-
Hukovic and Grubac [58] using electrochemical impedance
spectroscopy. Mixed WO3 + RuO2 were studied by Musiani
and coworkers [59] and synthesis of mixed MoO3/TiO2 and
WO3/TiO2 was reported by Lee and coworkers [60].
Remazol Black B (RBB) is a complex textile diazo dye
and has a high photo- and thermal stability. It cannot be effi-
ciently degraded using conventional methods of oxidative
degradation such as ozonation. A recent study [61] reported
onZnO-assistedphotocatalyticdegradationofRBBandother
dyes. Various metal oxide semiconductors were used by
Poulios and Tsachpinis [62] in their RBB photodegradation
investigations. The decolorization of RBB on TiO2 and CdS
photocatalysts in aqueous suspensions was also investigated
[63]. Photooxidative degradation of other azo dyes has been
reported [31,64–66]. Reaction mechanisms for reduction of
several diazo dyes on mercury electrode were elucidated
by Zanoni and coworkers [67]. Biodegradability studies of
Ganesh et al. [68] have shown that RBB and other azo dyes
are not biodegradable under aerobic conditions.
In this paper, we present new data on photoelectrocatalytic
degradation of RBB diazo dye obtained using high efficiency
nanostructured WO3 film electrodes. The performance of
WO3 in relation to MoO3 and TiO2 has been evaluated. The
quantum mechanical calculations of electronic structure of
RBB and similar azo dye molecules were performed to elu-
cidate initial stages of the dye degradation process.
2. Experimental
2.1. Chemicals
Tungsten powder (Alfa, APS 1–5 ␮m, 99.9%) was
obtained from Johnson Matthey Company. Remazol Black B
(Reactive Black 5, CAS# 17095-24-8) was purchased from

3. 5280 M. Hepel, S. Hazelton / Electrochimica Acta 50 (2005) 5278–5291
Dystar (Charlotte, NC, USA). The formula of RBB dye is
shown as follows:
whereRattachedtoSO3 groupis CH2CH2 SO3
−.Other
chemicals were obtained from Aldrich. All reagents used in
experiments were of analytical grade purity.
2.2. Apparatus
A standard photoelectrochemical setup was employed
for dye degradation experiments and current–potential mea-
surements. It was composed of a microcomputer controlled
potentiostat, Model PS-1705 (Elchema, Potsdam, NY), and a
500 W Quartz Halogen Lamp, as the illumination source. The
electrochemical quartz crystal nanobalance (EQCN) tech-
nique [69] was applied to measure the effective mass changes
of the electrode during the deposition of WO3 films. A Model
EQCN-700 (Elchema) was employed to record these tran-
sients. All experiments were performed in three-electrode
electrochemical cells. A cell with quartz window was used
for UV–vis light induced degradation experiments. The semi-
conductor electrodes were used as working electrodes. A
saturated (KCl) calomel electrode (SCE) was used as the
reference electrode and a Pt wire was used as the counter elec-
trode. Photocurrent–potential characteristics were recorded
using Elchema PHOTOMAT-1 dark cell assembly with elec-
tronically controlled illumination modulation and DAQ-716v
Data Logger acquisition and control system.
Atomic force microscopy (AFM) imaging of WO3 films
on Au piezoelectrode substrates was done using a Veeco
Model Nanoscope IIIa SPM and an Elchema nanopositioner
Nanoscan 5.02.
The concentration of RBB dye in the solution was deter-
mined by measuring the absorbance of samples of the dye
solution using an HP-8452A Diode Array UV–vis Spec-
trophotometer (Hewlett-Packard). Solution samples were
taken every 60 s for fast RBB decay transients, and up to
every 10 min for slow decays.
2.3. Preparation of semiconductor film electrodes
2.3.1. Preparation of nanostructured WO3 films
Solutions for the electrodeposition of WO3 films were
prepared in a similar way as described previously [33,43].
Briefly, procedure involved dissolution of 2 g of tungsten
metal powder in 10 mL of aqueous 30% hydrogen peroxide
solution. After the exothermic reaction had ended, 100 mL
of distilled water and 30 mL of 2-propanol were added. The
cleaned Pt gauze electrode (real surface area: 66.1 cm2) was
immersed in the electrodeposition solution and biased at
a deposition potential Edep (selected in the range −300 to
−500 mV) for 30–60 min. The obtained WO3 films were
carefully washed with Milli-Q water and ethanol, then heat
treated for 1 h (the annealing temperature varied from 200
to 600 ◦C). The amount of the coated semiconductor on a Pt
gauze substrate was determined by weighing. The WO3 film
loading was typically 500 ␮g/cm2 corresponding to the film
thickness of ca. 700 nm. The SEM examination revealed that
the average size of WO3 grains in the film was 30 nm.
The preparation of WO3 films for AFM imaging was car-
ried out in the same way as described above but the AT-cut
quartz crystal resonator wafers were used as the substrate
instead. The resonator crystals had a nominal resonance fre-
quency of 10 MHz and were optically polished, which is
necessary for AFM measurements. The standard key-hole
electrode pattern (200 nm Au over 50 nm Cr) was deposited
on both sides of the quartz crystal. The working electrode
had a 5 mm diameter.
2.3.2. Synthesis of MoO3 films
The method of preparation of MoO3 semiconducting films
was similar to that described for WO3, by dissolving sample
of Mo0 powder in 10 mL of 30% H2O2 and water, yielding a
claudyyellowsolution.ThecathodicpotentialusedforMoO3
film deposition was Edep = −450 mV versus Ag/AgCl. The
washed and dried films were annealed at 450 ◦C for 1 h.
2.3.3. Synthesis of TiO2 films
The nanoparticulate TiO2 films on Pt gauze electrodes
were obtained by electrodeposition using TiCl3 solution
purified with amalgamated Zn. The Pt gauze electrode was
cleaned successively by 5 M HNO3, Milli-Q water, acetone,
ethanol, and Milli-Q water. The Pt gauze was biased at 0.0 V

4. M. Hepel, S. Hazelton / Electrochimica Acta 50 (2005) 5278–5291 5281
for 30 min in 50 mM TiCl3 solution adjusted to pH 2.2 with
deaerated solution of 0.5 M Na2CO3. The obtained TiO2 film
wascarefullywashedwithdeaerated10 mMHClandethanol.
The films were dried and finally annealed at 450 ◦C for 1 h.
2.4. Other procedures
All the potentials are quoted with respect to the saturated
Ag/AgCl except when otherwise noted (NHE, normal hydro-
gen electrode, reference used for energy level calculations).
Solutions were deoxygenated for 20 min before experiments
using purified nitrogen and a nitrogen purge was maintained
above the solution during experiments. The experiments were
performed at room temperature (∼20 ◦C).
The quantum mechanical calculations were performed
using a commercial Quantum CAChe 4.1 program (Fujitsu
and Oxford Molecular Ltd.) and homemade software on a
3.4 GHz Pentium 4 CPU.
3. Results and discussion
3.1. Effect of semiconductor catalyst on dye
photodegradation process
The primary role of an electrocatalyst for photooxidation
processes is to provide very low valence band edge energy,
EVB, and to make the surface electron–holes to act as power-
ful oxidizing sites for generating radical oxidants in reactions
with the medium. The direct oxidation of dye molecules
by high energy holes will occur as well, though the low
concentration of dye may diminish the contribution of this
reaction pathway. Thus, the main goals in catalyst develop-
ment is obtaining high energy holes and maximizing their
surface concentration. We have developed methods of syn-
thesis for several transition metal oxide semiconductors and
tested them for photoelectrocatalytic activity in various diazo
dye degradation processes.
In Fig. 1, the absorbance decay transients for Remazol
Black B diazo dye at peak wavelength λmax = 598 nm are
compared for three photoelectrodes: MoO3, WO3, and TiO2.
The initial concentration of the dye was 8.8 × 10−5 M in
0.1 M NaCl at pH 5.0. It is seen that the fastest degrada-
tion rate was achieved for WO3 photocatalyst, followed by
MoO3, and clearly much slower rates were observed for com-
monly used TiO2 catalyst. The AFM surface roughness was
similar for all catalysts. From these experimental data (and
also from data for degradation of other diazo dyes. obtained
on the same photoelectrodes, not shown here) we can infer
that WO3 and MoO3 semiconductor electrodes show higher
photocatalytic activity than TiO2.
Searching for possible reasons of high catalytic activity of
WO3 and MoO3 semiconductors, we analyzed energy dia-
grams (Fig. 2) for semiconductor electrodes in solution
containing azo dyes. The oxidizing power of electron–holes
is best represented by the potential of the upper valence
band edge VVB (in NHE scale). For TiO2, we can calculate
Fig. 1. Decay of RBB dye absorbance A (at λmax = 598 nm) with time during
photoelectrocatalytic degradation of RBB on: (1) TiO2, (2) MoO3, and (3)
WO3 film electrodes, under UV illumination in a 8.8 × 10−5 M RBB + 0.1 M
NaCl solution at E = 1.08 V.
VVB from the published values of Eg = 3.0 eV and Vfb = 0.05–
0.059 pH [70] (NHE). Thus, VVB = Eg + Vfb = 3.05 V at pH 0,
and VVB = 2.76 V at pH 5. In the same way, for WO3, we
have Eg = 3.35 eV [44] and Vfb = 0.55–0.059 pH, so that VVB
is 3.90 and 3.60 V at pH 0 and 5, respectively. For MoO3,
using Eg = 3.15 eV [71] and Vfb = 0.35–0.059 pH (NHE), we
obtain VVB = 3.50 V at pH 0, and VVB = 3.21 V at pH 5.
Therefore, the oxidizing power of electron–holes at the upper
edge of valence band is in the order: TiO2 < MoO3 < WO3,
paralleling that for the catalytic activity toward RBB degra-
dation. Although the VVB potentials are much higher than
potentials practically utilized in pure electrochemical mea-
surements (without illumination), they have to be compared
to the energy level of electrons in a molecular orbital of an
adsorbed dye, represented in Fig. 2 by a Gaussian with energy
of the maximum ED. The quantum mechanical calculations of
molecular orbitals and their energy for RBB will be described
in the next sections.
Fig. 2. Energy diagram for an n-type semiconductor photoelectrode in a
solution of organic dye D, shown for positive bias and a position of ED
allowing for direct charge transfer without dye photoexcitation.

5. 5282 M. Hepel, S. Hazelton / Electrochimica Acta 50 (2005) 5278–5291
Hence, one of the possible explanations of the higher pho-
toactivity of WO3 and MoO3 in relation to TiO2, is the ability
of these semiconductor materials to produce higher energy
holes. We can also evaluate, if there are any differences in
the density of states in these metal oxides. A higher density
of states may increase the probability of interfacial charge
transfer processes. Since the respective metal elements are
located in different rows of periodic table, considerable dif-
ferences are encountered. Thus, the density of states at the
upper valence band edge nVB is lowest in TiO2, higher in
MoO3, and the highest in WO3, which parallels the devel-
opment of electron shells 3d, 4d, and 5d, respectively. The
total number of electrons beyond the noble gas configuration
is increasing in the same order: [Ar]3d2s2 in Ti, [Kr]4d5s1 in
Mo, and [Xe]4f145d46s2 in W.
There are also other factors influencing characteristics of
semiconducting materials, which may then be reflected in
their photoelectrocatalytic activity. As examples can serve:
doping level, surface states, quantum grain size effect modi-
fying Eg, etc. Full analysis is beyond the scope of this work,
but we would like to emphasize the importance of one effect,
the material non-stoichiometricity, which is usually not very
well controlled (or tested) in low-temperature semiconduc-
tor film syntheses in contrast to precision high-temperature
doping processes. Its influence is through the charge trans-
fer probability, as illustrated below, but non-stoichiometricity
may also lead to recombination losses through surface traps.
The semiconductor non-stoichiometricity is necessary in
non-doped materials since the concentration of charge carri-
ers in an intrinsic semiconductor with a wide energy gap is
very low:
n = n0 = exp −
(ECB − EF)F
RT
(1)
where n is the equilibrium electron concentration in conduc-
tion band (CB) and EF is the Fermi level energy. To increase
the electronic conductance, the material is usually doped or
nonstoichiometric (n-type for photoanode semiconductors),
by adding Dn electron donors. With nearly total ionization
of the donor, concentration of electrons sharply increases
(n = Dn n0), leading to further decreasing the equilibrium
hole concentration p in the bulk semiconductor:
p =
n0p0
Dn
(2)
A pronounced increase in p is impressed by the band bending
V in the near-surface depletion layer, resulting from raising
the electrode potential V over Vfb. But it is the photoexcita-
tion process that generates excess non-equilibrium holes [72]
of high energy for a photocatalytic process. The photoexci-
tation efficiency depends on a number of factors including
nVB, Eg, absorption coefficient, thickness of the depletion
layer, recombination losses through surface states and bulk
electron–hole annihilation, and others.
In a direct charge transfer between the adsorbed dye and
semiconductor electrode, the hole is transferred to occupied
molecular orbital in the dye. In Fig. 2, the density of states of
a dye D is expressed by the Gaussian (cf. [72]):
ρe = ρe,0 exp −
(E − ED)F)
4λRT
(3)
where ED is the energy level of highest occupied molecular
orbital, λ the reorganization energy of the medium (λ ≈ 1 eV
[73]), and ρe is the formal density of states. The total oxi-
dation rate is obtained by integrating over available energy
levels for electrons in the dye and holes (E = Eh) in the semi-
conductor. Thus the oxidation rate depends on the density
of states in the semiconductor at the upper part of valence
band and hole distribution, which are both affected by non-
stoichiometricity.
In this conjecture, we expect WO3 to have high density
of states in valence band and a concentration of holes that is
not diminished by recombination losses. These conditions are
necessary to support and augment the high VVB of WO3 and
provideineffectahighlyefficientphotoelectrocatalyticmate-
rial for dye degradation. Owing to the use of electrochemical
control system, the electric field in the semiconductor and
the depletion layer thickness are controlled by the electrode
potential and no electron scavengers are needed in the solu-
tion to prevent electron–hole recombination.
In view of the above considerations and owing to favor-
able degradation rates for azo dyes, further experiments were
conducted using WO3 photoelectrocatalysts.
3.2. Effect of synthesis conditions on photocatalytic
activity of WO3
The photocatalytic activity of transition metal oxides
depends strongly on the method of synthesis, substrate struc-
ture, and heat treatment. It is important to realize that it is
not only the surface morphology that changes with process-
ing. Both the bandgap energy and flat band potential depend
strongly on materials processing as well. In previous works
[44], we have determined Vfb and Eg for WO3 synthesized by
thermal evaporation method. A strong dependence of Eg on
water content was found: Eg = 3.365, 2.90, and 2.80 eV for
a-WO3, WO3·(1/3)H2O, and WO3·2H2O, respectively. The
flat band potential for annealed WO3 film electrode is given
by:
Vfb = 0.55 − 0.059 pH (in V versus NHE) (4)
The experiments described below were carried out in con-
junction with electrochemical synthesis of WO3 photocata-
lysts followed by thermal processing.
During the initial stages of WO3 film formation, the nucle-
ation and growth of spherical nuclei is observed. The contact
mode AFM images of such films are presented in Fig. 3. At
low cathodic overpotential, Edep = −300 mV, nearly perfect
spherical monodisperse WO3 nanostructures are formed, as
illustrated in Fig. 3a. Because the nucleation rate at this poten-
tial is very low in relation to the growth rate, the size of WO3

6. M. Hepel, S. Hazelton / Electrochimica Acta 50 (2005) 5278–5291 5283
Fig. 3. Contact mode AFM images for initial stages of nucleation and growth of WO3 film on Au-piezoelectrode surface during electrodeposition at Edep (mV
vs. Ag/AgCl): (a) −300, (b) −450. Image size: (a) 2.5 × 2.5 ␮m2, (b) 650 × 650 nm2.
crystals is quite large, approaching 250 nm. By increasing
cathodic overpotential, the nucleation rate can be increased
to achieve higher nucleation density and lower crystal dimen-
sions. This situation is depicted in Fig. 3b, showing the AFM
image for a WO3 film deposited at Edep = −450 mV. As fol-
lows from the sectioning analysis, the average crystal size
is reduced to 70 nm. The crystallites are monodisperse and
spherical in shape. In addition to the considerable decrease
in crystal size, there is a clear long range ordering seen in the
image, which leads to linear structures, resembling lines of
slightly fused balls. This ordering does not come from epi-
taxial effects on the Au substrate, which is composed of a
random mosaic of grains with 80 nm diameter.
The WO3 film annealing is an essential treatment bring-
ing about structural and morphological changes. The “as-
deposited” films show amorphous structure with nearly
spherical grains. The annealing at a low temperature,
t = 200 ◦C, improves somewhat the photocatalytic activity,
but at least 350 ◦C is required to obtain partial recrystalliza-
tion of the film. At t = 400 ◦C, X-ray diffraction patterns show
a wide variety of crystal faces with most pronounced (0 0 2),
(0 2 0), (2 0 0), (2 0 2), and (1 1 2). The effect of annealing
temperature on catalytic activity of WO3 photoelectrodes
is illustrated in Fig. 4a. The catalytic activity curve, shown
in the inset, passes through a maximum at t = 400 ◦C and
further temperature increases are counterproductive. Care-
ful inspection of film morphology changes indicates that the
post-peak decrease in catalytic activity is due to the develop-
ment of cracks and film peeling off. Therefore, in dye degra-
dation experiments, measurements were mainly carried out
using photoelectrodes annealed at t = 400 ◦C. The absorbance
spectra for RBB solutions, with initial RBB concentration
of 6 × 10−5 M, after 20 min of photoelectrodegradation at
E = 1.08 V, using Au | WO3 film electrodes annealed at dif-
ferent temperatures, are presented in Fig. 4b. The absorbance
maximum at λmax = 598 nm is diminished considerably for
film annealed at 200 ◦C and disappears completely for film
annealed at 400 ◦C.
3.3. Photocurrent measurements
Linear scan voltammograms for a Au | WO3 electrode
show two potential regions with completely different opti-
cal properties of WO3. This is exemplified in Fig. 5. In the
lower potential region (shown in the inset of Fig. 5), from
E = +200 to −200 mV, the electrochromic behavior prevails
[42,43] and the intercalation of hydrogen into WO3 lattice
takes place, which is marked by blue film coloration on cath-
odization and bleaching on anodization. The higher poten-
tial region, from E = +250 to +1100 mV, is the double-layer
region, providing an ample potential window for carrying
out electrocatalytic processes. It is this region, where the
photocurrents useful for initiating dye degradation processes
are developing. According to the Butler–Wilson theory [74]
based on Schottky–Gartner [75] model, the photocurrent iph
is given by:
iph = αW0qϕ0 E − Efb (5)
whereαistheabsorptioncoefficient(α = 5.9 × 104 cm−1 [74]
at λ = 320 nm for WO3), W0 the depletion layer width for a
potential of 1 V across it, ϕ0 the photon flux, q the electron
charge, and Efb is the flat band energy.
The photocurrent measurements where performed in a
switchable system allowing to record voltammetric current
in total darkness and under UV illumination. The duration
of light pulses was 60 s, followed by dark current measure-
ment, also for 60 s. The obtained photocurrent characteristic
of a Au | WO3 electrode is presented in Fig. 6. Very slow
potential scan, v = 1 mV/s was applied due to high capacitive
currents of the WO3 (supercapacitance due to multivalency
of Wn+ centers in the oxide lattice). The slow scan allowed
the capacitive currents to decay completely during illumina-
tion modulation without affecting the current levels measured
under light-off and light-on conditions. To increase the film
firmness and reduce the possibility of electrical shorting,
three layers of WO3 were deposited on top of each other

7. 5284 M. Hepel, S. Hazelton / Electrochimica Acta 50 (2005) 5278–5291
Fig. 4. (a) Absorbance A vs. t transients for photoelectrocatalytic degrada-
tion of a 6 × 10−5 M RBB + 0.1 M NaCl solution on WO3 film electrodes
at 1.08 V, after film annealing at ta (◦C): (1) 25, (2) 200, (3) 300, (4) 400,
(5) 500, (6) 600. Inset: dependence of initial degradation rate v0 on ta, (b)
UV–vis absorbance spectra after 20 min of photoelectrocatalytic degrada-
tion using WO3 film electrodes annealed at ta (◦C): (1) 25, (2) 200, (3) 400,
(4) 600.
in these experiments. For each layer, after depositing, a heat
treatment at t = 400 ◦C for 60 min was applied.
The flat band potential of the Au | WO3 film electrode was
then determined by measuring the photocurrent onset poten-
tial. As found in our earlier studies [44], the Butler equation
can be applied:
E − Efb ∼
iph
αW0qϕ0
2
(6)
Fig. 7 displays the linear dependence of the square of the pho-
tocurrent versus applied potential. The linefit extrapolated
to zero photocurrent leads to the flat band potential value:
Efb = +293 mV versus Ag/AgCl in 0.1 M KCl at pH 1.6. This
value translated to pH 0 and NHE scale gives Efb = +0.55 V
(NHE) which is needed for the evaluation of hole energy level
Fig. 5. Linear scan voltammetric characteristics of a WO3 film electrode in
the double-layer region, marking the potential window accessible for pho-
toelectrocatalytic degradation of dyes. Inset: cathodic intercalation region.
and its comparison with molecular orbital energies for dye
molecules.
3.4. Kinetics of photodegradation processes
The kinetic equation that governs the azo dye photooxi-
dation process:
D + h+
→ P (7)
where D is the dye molecule and P represents the reac-
tion products, is generally assumed to arrive from a
Langmuir–Hinshelwood model [32,76] and is of the general
Fig. 6. The LSV i–E dependence for photocurrent measurements with
switchable ON-OFF illumination, recorded for an n-WO3 film electrode
in 0.1 M KCl solution, pH 1.6, at a scan rate v = 1 mV/s and illumination
pulse duration of 60 s.

8. M. Hepel, S. Hazelton / Electrochimica Acta 50 (2005) 5278–5291 5285
Fig. 7. Determination of the flat-band potential from Gartner-type i2
ph vs. E
dependence for an n-WO3 film electrode in 0.1 M KCl solution at a scan rate
v = 1 mV/s.
form:
v =
k KC0
1 + KC0
(8)
where C0 is the initial dye concentration, K the Langmuir-
type adsorption equilibrium constant, and k is the forward
reaction rate constant. The above equation fits very well to
a situation in which dye molecules adsorb reversibly on the
catalyst surface and their oxidation is solely controlled by
the reaction of adsorbed dye with electron–holes from semi-
conductor, i.e. by the first stage of the overall degradation
process. However, the real situation of dye photodegradation
involves also parallel reactions, such as the formation of oxi-
dants from the components of the medium, participation of
these oxidants in the dye degradation in solution, etc. The
reaction mechanisms discussed in literature [1,14,31,32,76]
take into account all these complexities. The experimen-
tal data can usually be fitted with an equation of the form
of expression (8), but due to complexities just mentioned,
the meaning of constants may not be exactly that of a sim-
ple Langmuir–Hinshelwood model. For instance, K may not
represent the adsorption equilibrium of the dye but rather
adsorption of electrolyte anions or adsorption of powerful
OH radicals formed on the photocatalyst surface. Eq. (8) is
usually tested in the form of reciprocal rate versus reciprocal
concentration:
1
v
=
1
k
+
1
k KC0
(9)
and the rate constant k is determined by extrapolating to infi-
nite C0. The deviation from the model equation can be readily
discernedwhentheexperimentaldependence1/vversus1/C0
is not linear. For very low coverages θ, when C0 1/K, the
determination error increases considerably making the anal-
ysis impractical. We have found that in such cases, the use of
Fig. 8. Determination of kinetic parameters for photoelectrocatalytic degra-
dation of RBB on a WO3 film electrode using Langmuir–Hinshelwood
reciprocal rate dependence 1/v0 vs. 1/[RBB].
first order kinetic equation:
v = k1C0 (10)
where k1 = k K, is a better choice, though it does not allow
for estimation of K. If K can be determined from other exper-
iments (e.g. adsorption data), then k can still be determined.
In Fig. 8, the reciprocal rate dependence 1/v0 versus
1/C0 for RBB photoelectrocatalytic degradation on Pt | WO3
electrode in 0.5 M NaCl solution is presented. The values
of initial degradation rate v0 were determined from initial
slopes of the concentration decay C–t plots. The photoelec-
trode was biased at 1.04 V versus Ag/AgCl. From linear
least-square fitting, the values of k and K are obtained as
follows: k = 1.10 × 10−9, K = 3.35 × 104 L/mol. The Gibbs
free energy for the adsorption process:
G◦
ads = −RT ln K (11)
is then G◦
ads = −0.59165 × 2.302 log(3.35 × 104) = −6.16
kcal/mol (−25.8 kJ/mol) which, most likely, falls into the
electrosorption category. The analysis of Langmuir adsorp-
tion equilibrium with constant K indicates that one half of
the electrode surface is covered with adsorbate (θ = 0.5) at
C = 3×10−5 M. Because at the applied bias, the electrode is
charged positively and RBB is an anion, the high surface
coverage is expected. However, in the presence of strong
chemisorption, a much higher value of K is usually obtained.
Further studies would be needed to evaluate if there is any
competitive adsorption from Cl− ions. Since Cl− ions have
been found to desorb from a Pt electrode at potentials of oxide
formation, it is likely that interactions between WO3 surface
and Cl− ions are also weak. The implication of this would
be a low K value for chloride adsorption. Hence, the K value
observed experimentally is most likely due to RBB adsorp-
tion and fulfillment of Langmuir–Hinshelwood model. The
strong chemisorption interaction of an azo compound with

9. 5286 M. Hepel, S. Hazelton / Electrochimica Acta 50 (2005) 5278–5291
transition metal oxide was postulated by Bauer et al. [77] on
the basis of interactions of a model azo dye, Acid Orange
7, with a TiO2 substrate. In the adsorption model proposed
by them, the oxygen atoms from SO3
− moiety participate
in an inner-sphere-type complex with the surface Tiiv atoms.
Since RBB molecule has four SO3
− groups, similar interac-
tions with TiO2 could be expected. However, the interactions
of RBB with WO3 seem to be much weaker.
3.5. Effect of the dye electronic structure on
photoelectrocatalytic degradation
In aromatic azo compounds, under reductive conditions,
cleavage of an azo bond proceeds usually through hydrazone
tautomersandleadstotheformationoftwoaminocompounds
[67,78]. Under oxidative conditions, many reaction pathways
can be considered, including paths leading to complete min-
eralization of the dye. Despite of the extensive literature on
the degradation of dye pollutants, it is not well understood at
present in which areas of the dye molecule most of the vul-
nerability to oxidative degradation is concentrated and elec-
trons from which regions participate in initial stages of the
process. Therefore, we have begun investigations of the elec-
tronic structures of dyes and attempted correlating them with
experimental degradation rates. We have performed quantum
mechanical calculations for a wide range of azo and diazo
dyes. In this work, we have examined the mechanism of pho-
toelectrocatalytic degradation of RBB using self-consistent
field Hartree–Fock molecular orbital calculations [79,80].
The ab initio calculations were performed for the equilib-
rium geometry and electrons in the ground state using basis
set 6-31G*. In Fig. 9a–d, the electron density map (as a space
model) and two highest occupied orbitals (HOMO−1 and
HOMO) for undissociated RBB molecule are presented. The
RBB molecule has all the basic properties of RBB, only the
inactive side chains ( CH2CH2SO3 ) in terminal aryl sul-
fonate groups were removed to decrease computer memory
requirements and reduce computation time (to several days
on a 3.4 GHz CPU). The lowest unoccupied molecular orbital
(LUMO) of RBB is also included (Fig. 9e). The electronic
structural information was then utilized in evaluating reactiv-
ity of the dye toward oxidants. In general, the dye oxidation
may proceed through a mechanism of either an electrophilic
or nucleophilic attack by strong oxidants adsorbed on the
electrode or dissolved in solution in vicinity of electrode sur-
face. The dye molecule may also undergo oxidation by a
direct charge transfer with the electrode:
D + h+
= D+
(12)
These three mechanisms are discussed below.
3.5.1. Electrophilic attack
In the electrophilic attack, an oxidant would interact with
atoms and bonds of the molecule where the most external
electrons are present, the HOMO orbital. It is seen in Fig. 9c
that there are several regions where the RBB molecule is vul-
nerable to such an attack and the degradation process may
proceed through all of these paths. One of these paths is the
cleavage of azo bonds, weakened by the removal of HOMO
electrons on oxidation. There are also evident weaknesses in
the central naphthalene system and NH2 substituent but not
in the peripheral aryl groups. It has to be noted that all the
apparently separate regions of HOMO belong to one orbital
and the electron density from all these regions would disap-
pear if HOMO electrons are removed on oxidation. However,
it is to be expected that an electrophilic oxidant would attempt
to approach RBB molecule from the side of highest concen-
tration (highest density) of HOMO electrons. The oxidants
that might follow the electrophilic attack mechanism could
include HClO, Cl
•
, ClO2
•, SO4
•−, H2O2. On the basis of sys-
tematic kinetic studies, Oakes and Gratton [81] have found
out that HClO acts through the electrophilic attack mecha-
nism during oxidation of model Orange I azo compounds.
3.5.2. Nucleophilic attack
In the case of a nucleophilic attack, the evaluation of most
electron-deficient spots on RBB molecule is of primary inter-
est. The evaluation can be done by considering the LUMO
orbital, which is unoccupied and spreads over areas of the
molecule that are most deficient in electrons. There are sev-
eral low electron density areas of RBB molecule, where most
likely the nucleophilic attacks could be concentrated. While
the HOMO orbital is clearly stabilizing the N N bonds, parts
of the LUMO electron clouds (unoccupied) are spread over
the C N bonds indicating on electron deficiency there. If a
nucleophilic attack would be directed toward benzene ring
carbons, cleavage of nitrogen bonds to aromatic rings could
occur, leading to the isolation of N2 molecule, provided that
nitrogens of azo bonds are not simultaneously oxidized to
higher oxidation states (viz., nitrogen oxides). It is seen from
Fig. 9e that electron deficiency is also encountered in the
naphthol area and the aryl ring adjacent to the azo bond close
to OH substituent. The other aryl ring seems to be protected
by scavenging electrons from the distant electron donating
NH2 group. The oxidants that would follow the nucleophilic
attack mechanism could include OH• and HO2
−. According
to Oakes and Gratton [81], HO2
− can act through the nucle-
ophilic attack mechanism during oxidation of some monoazo
compounds. It remains to be seen if this can have a more gen-
eral significance and be expanded to other azo compounds.
3.5.3. Direct charge transfer
A direct charge transfer between the dye molecule and
the electrode can occur if two basic conditions are fulfilled:
(i) the energy level of HOMO (or lower lying molecu-
lar orbitals) is higher than or equal to the energy of an
electron–hole in the semiconductor electrode and (ii) the
distance between the two centers exchanging electron is
sufficiently small so that the tunneling probability is high
enough to effect the transfer. It follows from our calculations
that the energy of HOMO for RBB is ED = −8.25 eV. This

10. M. Hepel, S. Hazelton / Electrochimica Acta 50 (2005) 5278–5291 5287
Fig. 9. Electronic structure of RBB obtained using SCF Hartree–Fock molecular orbital calculations: (a) total electron density surface Sd for
ρe = 0.0135 electrons/ ˚A3, (b) HOMO−1, (c) HOMO, (d) HOMO−1 and HOMO, (e) LUMO; orbitals (b)–(e) were obtained for a fixed ρe = 0.2 electrons/ ˚A3.
level should be compared to the energy of holes available
at the upper VB edge in the semiconductor. For WO3, we
have calculated in preceding sections: VVB = 3.60 V. Thus
EVB = ENHE − VVB = −4.5 − 3.60 = −8.1 eV. This value is
slightly higher than that for ED. However, owing to three
effects listed below, the elastic charge transfer is possible:
(i) due to the band bending, there is still a high concen-
tration of holes at lower energies, especially at high bias
potential, when V can approach 0.7 V (lowering the hole
energy to −8.8 eV in the bulk semiconductor; only part of
V is actually effective as the hole concentration p decreases
rapidly with distance from the interface); (ii) the reorga-
nization energy of the medium λ (λ ≈ 0.5–1.5 eV [72,82])
increases the width of the density-of-states peak for the dye
(Fig. 2) so that a higher energy electrons are also available
for charge transfer; (iii) thermal holes occupy lower energy
levels (below VB), on the order of 0.025–0.05 eV, which can
also be active in the direct charge transfer mechanism. There-
fore, for the RBB molecule and WO3 photocatalyst, the direct
oxidative charge transfer is possible. The areas of RBB with
the highest probability of such a transfer are concentrated
in locations with the highest HOMO electron density, i.e.
the azo bonds, the naphthalene carbons, and the nitrogen of
NH2 substituent. The azo bond close to NH2 substituent
is more vulnerable than the azo bond close to OH, which is
protected by hydrogen bonding N· · ·H· · ·O.
The effect of various substituents on the degradation rate is
important. The electron donating substituents in ortho posi-
tion tend to increase HOMO electron density over the azo
bond, thus protecting it against nucleophilic attacks. How-

11. 5288 M. Hepel, S. Hazelton / Electrochimica Acta 50 (2005) 5278–5291
ever, the oxidation by electrophilic oxidants is then enhanced.
The effect of electron withdrawing substituents in the
ortho position is in general opposite. They tend to decrease
HOMO electron density over the azo bond, thus slowing the
dye oxidation by electrophilic oxidants. At the same time, the
rate of oxidation by nucleophilic oxidants is enhanced.
3.6. Mapping the properties of electronic structure on
electron density surface
Further analysis of the effects of electronic structure of
RBB on kinetics of photoelectrocatalytic oxidative degrada-
tion was performed by considering electronic properties at
the outskirts of the molecule, i.e. in places where an oxidant
would normally commence its interactions with the dye. The
outer limits of the dye molecule were defined by computing
the total electron density surface Sd for a low electron density,
0.0135 electrons/ ˚A3, and mapping the following properties
on this surface: HOMO (MH), LUMO (ML), electrostatic
potential (MV), and ionization potential (MI). The obtained
results are presented in Fig. 10. The first property map, MH,
represents a cross-sectional view of HOMO orbital calcu-
lated up to the molecule periphery where the actual HOMO
electron density was plotted in the color scale from red to
blue (low to high electron density). In comparison to MH,
the regular HOMO electron density plot of Fig. 9c is lim-
ited to high electron densities only (0.2 electrons/ ˚A3) and
shows a region of space inside the molecule, not extending
to the molecule borders, albeit the actual HOMO electron
cloud extends farther, though with decrementing probability
of finding electrons there. The significance of the MH repre-
sentation is that it shows regions on Sd surface where HOMO
electrons with energy EH can be found with lower or higher
probability. In other words, the MH map shows where HOMO
electrons on the molecule ‘surface’ can be found.
From the HOMO map MH presented in Fig. 10a, it is
seen that HOMO electrons are easily approachable in the
region of naphthalene aromatic system, which is then most
vulnerable to an electrophilic attack. The ring of the naph-
thalene with NH2 substituent is particularly vulnerable. On
the other hand, the HOMO electrons which are spread over
the azo bonds (seen clearly in Fig. 9c) are located deep
inside RBB molecule and are not readily accessible at the
outskirts of the molecule. This means that in order to inter-
Fig. 10. Mapping of electronic properties on the total electron density surface Sd: (a) MH, (b) ML, (c) MV, (a) MI.

12. M. Hepel, S. Hazelton / Electrochimica Acta 50 (2005) 5278–5291 5289
act with azo bond electrons, the oxidant would need a close
encounter (a collision) with RBB molecule. The azo bond
adjacent to NH2 substituent is more conspicuous than the
one in vicinity of OH, which is protected by hydrogen
bonding.
The LUMO mapping was done similar to the HOMO map-
ping. The utility of ML is based on the fact that it points
to the areas of molecule surface with highest electron defi-
ciency. As illustrated in Fig. 10b, the electron deficiencies
are concentrated in the central naphthol system (except for
the NH2 substituent) and the C N bonds of azo anchors on
both sides of the naphthol group, but not on the N N bonds
themselves. Particularly strong electron deficiency is seen in
the area of the C N bonds of azo anchor adjacent to the OH
substituent. In consequence of these electron deficiencies,
any nucleophilic attack is expected to be directed to the cen-
tral naphthol group and to the anchors of azo bonds. These
conclusions basically confirm the conjectures from analyzing
LUMO of Fig. 9e, namely the vulnerability of the naphtha-
lene center group and the azo bonds anchors (but not the azo
bonds). The inequality of the aryl groups is also clearly seen
in Fig. 10b, with higher electron deficiency in aryl group on
the side of OH substituent than that on the side of NH2
substituent.
In Fig. 10c, the electrostatic potential map MV for RBB is
presented. The green color represents intermediate potential,
the red color indicates more negative potential, and the blue
color stands for more positive potential. The map is domi-
nated by polar sulfonate groups. The electrostatic potential
distribution together with the direction and strength of the
dipole moment are the most important factors in interacting
of RBB with solid surfaces, especially when electrosorp-
tion is involved, as concluded from relatively low value of
G◦
ads = −6.16 kcal/mol determined in this work. The dipole
moment calculated for RBB is 6.37 D (Debye). It lies in
the plane of naphthalene rings and is directed as depicted
with an arrow in Fig. 10c (from positive to negative charge).
Therefore, the orientation of RBB on a positively charged
electrode surface (V > Vfb) will most likely be vertical tilted,
with counterions (cations) screening the upper part of the
molecule.
For a direct charge transfer between a dye molecule and an
electrode surface, perhaps the most attractive is the mapping
ofionizationpotentialonelectrondensitysurface.Suchamap
for RBB is presented in Fig. 10d. The ionization potentials
are all positive. The color scale is from red (low ionization
potential) to blue (high ionization potential). Evidently, the
lowest ionization potentials are found for aromatic rings of
naphthyl and aryl groups, followed by amino group. High
ionization potential is in areas of azo bonds and OH sub-
stituent. In the regions just mentioned, except for the aryl
groups, there is a full agreement between the HOMO map-
ping and ionization potential mapping. The low ionization
potential of aryl groups has no counterpart in HOMO map-
ping (Fig. 10a) and in HOMO orbital (Fig. 9c). It is likely
that this disparity is due to the strong show up on the Sd sur-
face of lower lying orbitals, HOMO−1 (Fig. 9b) and others
(not shown), characterized with lower (more negative) energy
levels.
Due to the high energy of VB holes in WO3, the degrada-
tionofdiazodyescontinuesuptothecompletemineralization
of all reaction intermediates. Further experimental mecha-
nistic studies, based on chromatographic and spectroscopic
analysis of intermediates [71] will be published elsewhere.
4. Conclusions
We have synthesized highly efficient n-WO3 film elec-
trodes for photoelectrocatalytic degradation of azo dye pol-
lutants. The comparison of WO3 catalytic activity with that of
MoO3 and TiO2 and analysis of electronic properties of these
semiconductor materials indicates that the high performance
of WO3 is correlated with EVB energy which is consid-
erably lower (more negative) than that for other transition
metal oxides. Moreover, the HOMO energy level for RBB
obtained from ab initio SCF Hartree–Fock molecular orbital
calculations(ED = −8.25 eV)isveryclosetoEVB = −8.10 eV
of WO3, in solution under study, making the direct charge
transfer between the dye and the electrode possible, without
considering any dye photoexcitation. This is less likely for
MoO3 and impossible for TiO2. The favorable level of EVB
for WO3 will also aid in raising probability for charge trans-
fer process in other oxidant generation, such as OH•, Cl2
•−,
Cl•, SO4
•−, etc.
Kinetic measurements of photoelectrocatalytic RBB
degradation have shown that a generalized Langmuir–
Hinshelwoodmodelappliestothesystem,leadingtotheover-
all rate constant k = 1.1 × 10−9 mol cm−2 s−1, with RBB
adsorption equilibrium constant K = 3.35 × 104 mol −1 L and
Gibbs free energy of adsorption G◦ = −6.16 kcal/mol
(−25.8 kJ/mol). The relatively low adsorption free energy
indicates that, most likely, the electrosorption is involved.
Further adsorption studies utilizing other techniques would
be required. At this point, there is no information avail-
able concerning co-adsorption of electrolyte anions and their
effect on the dye degradation process.
The quantum mechanical calculations of RBB orbitals and
their analysis have shown that important information can be
gained from them for molecule reactivity in the initial stages
of oxidative degradation. These conclusions are augmented
by mapping electronic properties on a total electron density
surface Sd. The most likely reaction paths for electrophilic,
nucleophilic, and direct charge transfer oxidative RBB degra-
dation have been elucidated.
Acknowledgments
This work was supported by the National Science Founda-
tion Grant No. CCLI-0126402 and the Petroleum Research
Fund of ACS, Grant No. 33190-B5.

13. 5290 M. Hepel, S. Hazelton / Electrochimica Acta 50 (2005) 5278–5291
References
[1] C. Comninellis, E. Plattner, in: S. Stucki (Ed.), Process Technologies
for Water Treatment, Plenum Press, New York, 1988, p. 205.
[2] E.J. Weber, R.L. Adams, Environ. Sci. Technol. 29 (1995)
1163.
[3] K. Vinodgopal, P.V. Kamat, CHEMTECH (1996) 18.
[4] K. Vinodgopal, P.V. Kamat, Environ. Sci. Technol. 30 (1996)
1660.
[5] A. Mills, R.H. Davies, D. Worsley, Chem. Soc. Rev. (1993) 417.
[6] A.E.H. Machado, A.J. Gomes, C.M.F. Campos, M.G.H. Terrones,
D.S. Perez, R. Ruggiero, A. Castellan, J. Photochem. Photobiol. A:
Chem. 110 (1997) 99.
[7] A. Hagfeldt, M. Graetzel, Chem. Rev. 95 (1995) 49.
[8] M. Schiavello, Electrochim. Acta 38 (1993) 11.
[9] H. Hidaka, J. Zhao, E. Pelizzetti, N. Serpone, J. Phys. Chem. 96
(1992) 2226.
[10] R.W. Matthews, J. Chem. Soc., Faraday Trans. I 85 (1989) 1291.
[11] S. Lakshmi, R. Renganathan, S. Fujita, J. Photochem. Photobiol. A:
Chem. 88 (1995) 163.
[12] B. Singhal, A. Porwal, A. Sharma, R. Ameta, S.C. Ameta, J. Pho-
tochem. Photobiol. A: Chem. 108 (1997) 85.
[13] H. Gerischer, Photochem. Photobiol. 16 (1972) 243.
[14] N. Serpone, P. Maruthamuthu, P. Pichat, E. Pelizzetti, H. Hidaka, J.
Photochem. Photobiol. A: Chem. 85 (1995) 247.
[15] S. Ito, T. Deguchi, K. Imai, M. Iwasaki, H. Tada, Electrochem.
Solid-State Lett. 2 (1999) 440.
[16] A.L. Linsebigler, G. Lu, J.J.T. Yates, Chem. Rev. 95 (1995) 735.
[17] S.K. Lee, P.K.J. Robertson, A. Mills, D. McStay, J. Photochem.
Photobiol. A: Chem. 122 (1999) 69.
[18] K. Tanaka, T. Hisanaga, K. Harada, N. J. Chem. 13 (1989) 5.
[19] S. Naskar, S.A. Pillay, M. Chanda, J. Photochem. Photobiol. A:
Chem. 113 (1998) 257.
[20] C. Nasr, K. Vinodgopal, L. Fisher, S. Hotchandani, A.K. Chattopad-
hyay, P.V. Kamat, J. Phys. Chem. 100 (1996) 8436.
[21] K. Vinodgopal, I. Bedja, S. Hotchandani, P.V. Kamat, Langmuir 10
(1994) 1767.
[22] A. Fujishima, K. Honda, Nature 238 (1972) 37.
[23] M. Gratzel (Ed.), Energy Resources Through Photochemistry and
Catalysis, Academic Press, New York, 1983.
[24] K. Rajeshwar, N.R. deTacconi, in: A. Wieckowski (Ed.), Interfa-
cial Electrochemistry. Theory, Experiment, and Applications, Marcel
Dekker, New York, 1999, p. 721.
[25] R. Ganesh, G.D. Boardman, D. Michelson, Water Res. 12 (1994)
1367.
[26] P.R. Gogate, A.B. Pandit, Adv. Environ. Res. 8 (2004) 501.
[27] L.M. Peter, E.A. Ponomarev, G. Franco, N.J. Shaw, Electrochim.
Acta 45 (1999) 549.
[28] G. Hodes, I.D.J. Howell, L.M. Peter, J. Electrochem. Soc. 139 (1992)
3136.
[29] C.A. Parsons, M.W. Pterson, B.R. Thacker, J.A. Turner, A.J. Nozik,
J. Phys. Chem. 94 (1990) 3381.
[30] M. Tomkiewicz, Surf. Sci. 101 (1980) 286.
[31] M. Hepel, J. Luo, Electrochim. Acta 47 (2001) 729.
[32] E. Pelizzetti, C. Minero, Electrochim. Acta 38 (1993) 47.
[33] H. Al-Ekabi, N. Serpone, E. Pelizzetti, C. Minero, M.A. Fox, R.B.
Draper, Langmuir 5 (1989) 250.
[34] N. Serpone, R.F. Khairutdinov, Semiconductor Nanoclusters, Elsevier
Science, Amsterdam, 1996.
[35] E. Brillas, B. Boye, I. Sires, J.A. Garrido, R.M. Rodriguez, C. Arias,
P.L. Cabot, C. Comninellis, Electrochim. Acta 49 (2004) 4487.
[36] D.D. Dionysiou, G. Balasubramanian, M.T. Suidan, A.P. Kho-
dadoust, I. Baudin, J.M. Laine, Water Res. 34 (2000) 2927.
[37] C. Comninellis, A. Nerini, J. Appl. Electrochem. 25 (1995) 23.
[38] C. Comninellis, C. Pulgarin, J. Appl. Electrochem. 21 (1991)
703;
C. Comninellis, C. Pulgarin, J. Appl. Electrochem. 23 (1993) 7108.
[39] S. Stucki, R. Kotz, B. Carcer, W. Suter, J. Appl. Electrochem. 21
(1991) 99.
[40] P. Canizares, J. Garcia-Gomez, C. Saez, M.A. Rodrigo, J. Appl.
Electrochem. 33 (2003) 917.
[41] L. Quattara, L. Duo, T. Diaco, A. Ivandini, K. Honda, T. Rao, A.
Fujishima, C. Comninellis, New Diamond Front. Carbon Technol.
13 (2003) 97.
[42] I. Shiyanovskaya, M. Hepel, J. Electrochem. Soc. 145 (1998)
1023.
[43] I. Shiyanovskaya, M. Hepel, J. Electrochem. Soc. 145 (1998)
3981.
[44] I. Shiyanovskaya, M. Hepel, J. Electrochem. Soc. 146 (1999)
243.
[45] I. Shiyanovskaya, M. Hepel, E. Tewksburry, J. N. Mater. Elec-
trochem. Syst. 3 (2000) 241.
[46] J. Luo, M. Hepel, Electrochim. Acta 46 (2001) 2913.
[47] P.K. Shen, A.C.C. Tseung, J. Electrochem. Soc. 141 (1994)
3082.
[48] P.K. Shen, K.Y. Chen, A.C.C. Tseung, J. Chem. Soc., Faraday Trans.
90 (1994) 3089.
[49] P.K. Shen, K.Y. Chen, A.C.C. Tseung, J. Electroanal. Chem. 389
(1995) 223.
[50] C. Bock, B. MacDougall, Electrochim. Acta 47 (2002) 3361.
[51] P.J. Kulesza, L. Faulkner, Colloids Surf. 41 (1989) 123.
[52] O. Savadogo, P. Beck, J. Electrochem. Soc. 143 (1996) 3842.
[53] O. Savadogo, E. Essalik, J. Electrochem. Soc. 143 (1996)
1814.
[54] J. Augustynski, C.J. Sartoretti, M. Ulmann, Proceedings of the Elec-
trochemical Society Meeting, vol. PV96-1, Los Angeles, May 5–10,
1996, p. 694.
[55] N.R. de Tacconi, C.R. Chenthamarakshan, K.L. Wouters, F.M.
MacDonnell, K. Rajeshwar, J. Electroanal. Chem. 566 (2004)
249.
[56] Y.S. Krasnov, G.Y. Kolbasov, Electrochim. Acta 49 (2004)
2425.
[57] H. Habazaki, Y. Hayashi, H. Konno, Electrochim. Acta 47 (2002)
4181.
[58] M. Metikos-Hukovic, Z. Grubac, J. Electroanal. Chem. 556 (2003)
167.
[59] C. Baruffaldi, S. Cattarin, M. Musiani, Electrochim. Acta 48 (2003)
3921.
[60] K.Y. Song, M.K. Park, Y.T. Kwon, H.W. Lee, W.J. Chung, W.I. Lee,
Chem. Mater. 13 (2001) 2349.
[61] C.A.K. Gouvea, F. Wypych, S.G. Moraes, N. Duran, N. Nagata, P.
Peralta-Zamora, Chemosphere 40 (2000) 433.
[62] L. Poulios, J. Tsachpinis, J. Chem. Technol. Biotechnol. 74 (1999)
349.
[63] L.B. Reutergardh, M. Iangphasuk, Chemosphere 35 (1997) 585.
[64] E. Stathatos, D. Tsiourvas, P. Lianos, Colloids Surf. A 149 (1999)
49.
[65] J. Bandara, V. Nadtochenko, J. Kiwi, C. Pulgarin, Water Sci. Technol.
35 (1997) 87.
[66] J. Luo, J. Yartym, M. Hepel, J. N. Mater. Electrochem. Syst. 5 (2002)
315.
[67] C.C.I. Guaratini, A.G. Fogg, M.V.B. Zanoni, Electroanalysis 13
(2001) 1535.
[68] R. Ganesh, G.D. Boardman, D. Michelsen, Water Res. 28 (1994)
1367.
[69] M. Hepel, in: A. Wieckowski (Ed.), Interfacial Electrochemistry.
Theory, Experiment, and Applications, Marcel Dekker, New York,
1999, p. 599.
[70] T. Hepel, M. Hepel, R. Osteryoung, J. Electrochem. Soc. 129 (1982)
2132.
[71] M. Hepel, S. Hazelton, in press.
[72] H. Gerischer, in: B.O. Seraphim (Ed.), Solar Energy Conversion,
Springer Verlag, Berlin, 1979, p. 115.
[73] H. Gerischer, Electrochim. Acta 38 (1993) 3.

14. M. Hepel, S. Hazelton / Electrochimica Acta 50 (2005) 5278–5291 5291
[74] M.A. Butler, J. Appl. Phys. 48 (1977) 1914.
[75] W.W. Gartner, Phys. Rev. 116 (1959) 84.
[76] C.S. Turchi, D.F. Ollis, J. Catal. 122 (1990) 178.
[77] C. Bauer, P. Jacques, A. Kalt, Chem. Phys. Lett. 307 (1999) 397.
[78] Z. Mandic’, B. Nigovic’, B. Simunic’, Electrochim. Acta 49 (2004)
607.
[79] P.W. Atkins, R.S. Friedman, Molecular Quantum Mechanics, Oxford,
1997.
[80] W.J. Hehre, L. Radon, P.R. Schleyer, J.A. Pople, Ab-initio Molecular
Orbital Theory, Wiley, New York, 1985.
[81] J. Oakes, P. Gratton, J. Chem. Soc., Perkin Trans. 2 (1998) 1857.
[82] L.P. Sonntag, M.J. Spitler, J. Phys. Chem. 89 (1985) 1453.