Selenium-modified titanium dioxide photochemical diode/electrolyte junctions: Photocatalytic and electrochemical preparation, characterization, and model simulations

Article abstract of DOI:10.1021/jp044156n

The photoelectrochemical behavior of TiO₂ thin film electrodes, photocatalytically modified with Se islands, is described. The TiO₂ thin films were electrodeposited on transparent conducting oxide glass substrates. The resultant electrode forms a n-TiO₂p-Se photochemical diode which, in turn, contacts an electrolyte phase. Both transient photocurrent profiles (in response to excitation light that is switched on or off) and steady-state current-potential curves in response to chopped irradiation are considered. We show that the relative dominance of the contributions from the TiO₂ and Se components to the overall response of the photochemical diode/electrolyte junction crucially depends on the wavelength distribution of the excitation light source. A simple equivalent circuit representation of this junction is presented, comprised of a photodiode in parallel with two photodiodes connected in series back-to-back. Simulations of the transient and steady-state photoelectrochemical response of this system are presented, and are shown to be in good agreement with the corresponding experimental profiles. ? 2005 American Chemical Society.

Full text of DOI:10.1021/jp044156n

J. Phys. Chem. B 2005, 109, 11953-11960
11953
Selenium-Modified Titanium Dioxide Photochemical Diode/Electrolyte Junctions:
Photocatalytic and Electrochemical Preparation, Characterization, and Model Simulations
Norma R. de Tacconi,* C. R. Chenthamarakshan, and Krishnan Rajeshwar*
Department of Chemistry and Biochemistry, The UniVersity of Texas at Arlington, Arlington, Texas 76019-0065
Eugenio J. Tacconi
Megger Engineering Group, Dallas, Texas 75237-1019
ReceiVed: December 22, 2004; In Final Form: April 25, 2005
The photoelectrochemical behavior of TiO2 thin film electrodes, photocatalytically modified with Se islands,
is described. The TiO2 thin films were electrodeposited on transparent conducting oxide glass substrates. The
resultant electrode forms a n-TiO2/p-Se “photochemical diode” which, in turn, contacts an electrolyte phase.
Both transient photocurrent profiles (in response to excitation light that is switched on or off) and steady-
state current-potential curves in response to chopped irradiation are considered. We show that the relative
dominance of the contributions from the TiO2 and Se components to the overall response of the photochemical
diode/electrolyte junction crucially depends on the wavelength distribution of the excitation light source. A
simple equivalent circuit representation of this junction is presented, comprised of a photodiode in parallel
with two photodiodes connected in series back-to-back. Simulations of the transient and steady-state
photoelectrochemical response of this system are presented, and are shown to be in good agreement with the
corresponding experimental profiles.
Introduction
trochemical behavior of the n-TiO2/p-Se/electrolyte junction are
simulated by a simple equivalent circuit model.
Novel configurations for semiconductor-electrolyte junctions
have relevance to a variety of practical applications ranging from
solar energy conversion to chemical sensors. Several types of
such monolithic structures have been reviewed.1 Perhaps the
simplest is the Schottky barrier2 type of structure resulting from
platinization of a semiconductor surface and its subsequent
immersion and use in a liquid electrolyte (see, for example,
Figure 12 in ref 1). Alternately, a n-type semiconductor can be
mated with a p-type semiconductor to afford a “photochemical
diode”.1,3 Even two different (particulate) semiconductors of the
same type (say, n) can be combined in a composite configura-
tion.4 Finally, an integrated, monolithic electrode design com-
prised of a p-n photovoltaic junction biasing a p-type semi-
conductor (e.g., p-GaInP2) has been deployed for photoelectro-
lyzing water.5
In previous papers from this laboratory,6,7 we showed how
photocatalysis could be used for preparing n-TiO2/p-Se diodes.
A similar approach was also used by another group, who
discussed the interesting properties of this diode in contact with
an electrolyte junction under UV illumination.8 In the present
paper, we further elaborate on the photoelectrochemical proper-
ties of this diode in contact with an inert supporting electrolyte
at fixed potentials (photocurrent-time transient profiles) and
under potentiodynamic (swept potential) conditions. We show
that the relative dominance of the contributions from the TiO2
and Se components to the overall photoelectrochemical behavior
of the diode crucially depends on the wavelength distribution
of the excitation light source employed. Finally, both the
measured transient profiles and the potentiodynamic photoelec-
Experimental Section
Chemicals and Materials. All chemicals were from com-
mercial sources and were the highest purity available. [Caution:
Selenium solutions are toxic; their handling and disposal must
be done with extreme care.] Deionized water (18 MΩ) was used
in all cases for making solutions. Either polycrystalline gold or
transparent conducting oxide (TCO) was used as the substrate
for the working electrode films. The latter consisted of F-doped
tin oxide (Nippon Sheet Glass Ltd.) coated on soda-lime glass.
They were nominally ca. 400 nm thick and had a sheet resistance
of 10.3 Ω/0. Strips (0.65 × 2 cm2) were cut and cleaned in
four 5 min steps of cleansing in ultrasonicated acetone, ethanol,
1:1 H2O2/ammonia, and finally ultrapure water. The Au disk
electrodes (Bioanalytical Systems, 0.018 cm2 geometric area)
were mechanically polished to mirror finish using alumina
particles (Buehler) of successively finer size (1.0 µm, 0.3 µm,
and 0.05 µm) on a microcloth (Buehler No. 40-7212). They
were then ultrasonically cleaned in water for 1 min.
Electrosynthesis. Titanium dioxide films were cathodically
electrosynthesized on TCO substrates from a peroxotitanium
precursor bath containing nitrate ions.9 The bath preparation
has been described by other authors9 but briefly consists of
dissolving Ti powder in a solution containing H2O2 and NH3.
The gel initially formed is redissolved in HNO3. The pH of
this solution is then adjusted to 1.7 with dilute NH3. Film
formation occurs in this process by the initial electrogeneration
of base and an increase in the local pH at the substrate/solution
interface.10,11 The titanium hydroxide gel film was obtained on
the substrate by reaction of the electrogenerated base with the
Ti peroxo species in solution.12 Thus the titanium precursor
* Authors to whom correspondence should be addressed. E-mail: de
Tacconi, ntacconi@uta.edu; Rajeshwar, rajeshwar@uta.edu.
10.1021/jp044156n CCC: $30.25 © 2005 American Chemical Society
Published on Web 05/28/2005

11954 J. Phys. Chem. B, Vol. 109, No. 24, 2005
de Tacconi et al.
concentration (8-10 mM), bath pH (∼1.7), and applied potential
(-0.9 to -1.0 V) must be precisely controlled to secure good-
quality films. Considerable effort was initially expended to
optimize the film deposition variables. The precursor bath was
used immediately after preparation as incipient colloid formation
occurs in solutions aged for several hours. Both the precursor
concentration and medium pH are crucial variables for stabiliz-
ing the solutions against fast colloid formation. Nominal
deposition periods ranged from 20 to 25 min to afford
micrometer-thick films. These oxide films were compact and
visually very smooth (see Figure 1, for example). The deposited
films were given a thermal anneal of 400 °C for 30 min in all
the cases.
For comparison purposes, selenium layers were grown on
Au substrates from 10 mM SeO2 in 0.1 M Na2SO4 at a fixed
potential of -0.7 V (vs Ag|AgCl|satd. KCl reference; all
potentials in this study are quoted with respect to this reference)
for 15 min. From the electrodeposition charge an approximate
film thickness of 130-180 nm could be estimated for these
layers.
Photocatalytic Preparation of n-TiO2/p-Se Diodes. The
diodes were prepared according to procedures described by us
in previous articles.6,7 Briefly, the TiO2 film-coated TCO slides
were placed in a quartz cell containing an aqueous solution of
0.4 mM SeO2 and irradiated with the full output of a Xe arc
lamp (see specifications below) for 2 h. To avoid solution
heating, a 2.5 cm water filter (made of two parallel quartz plates)
was interposed between the quartz cell and the Xe lamp.
Radiation Sources. A Xe arc lamp and a tungsten (W)-
halogen source (with a quartz envelope) were used in this study.
Both lamps were operated at full output at their rated 75 and
100 W levels, respectively. For the Xe arc case, the photon flux
was ∼0.45 mW/cm2 at 340 nm. The W-halogen lamp had a
measured photon flux of ∼0.28 mW/cm2 at 500 nm. The
bandwidth in both cases was 4 nm. The Xe lamp was an Oriel
Model 66173 operated in conjunction with a power supply (Oriel
Model 68806). The W-halogen lamp was operated with an Oriel
Model 68735 power supply. The photon fluxes were measured
on an Oriel radiant power/energy meter (Model 70260) and the
fluxes quoted above are not corrected for reflection losses by
cell windows etc.
Figure 1. Top view (a) and cross-section (b) by SEM of an
electrodeposited TiO2 layer on TCO substrates. The layer was given a
thermal anneal (see Experimental Section) prior to SEM examination.
Film thickness is ca. 592 nm as provided by the SEM cross-section.
were immersed ∼4-5 mm in the etch bath and subjected to
4.8 V AC voltage for 4 min, using a platinum foil as counter
electrode. They were then progressively emersed under polariza-
tion to concentrate the etch current at the very low end of the
tip. The last step improves the tip sharpness significantly, which
was verified with an optical microscope. All STM images
consist of 256 × 256 data points and were taken in constant
height mode. X-ray photoelectron spectroscopy (XPS) used a
Perkin-Elmer/Physical Electronics Model 5000C system with
details as given elsewhere.13
Electrochemistry/Photoelectrochemistry. A standard single-
compartment, three-electrode electrochemical cell was used both
for film preparation and for the electrochemical/photoelectro-
chemical measurements. A large Pt coil and an Ag|AgCl| satd.
KCl reference electrode (Microelectrode Inc.), along with the
working electrode, completed the cell setup. Electrodeposition
was carried out on a Model 100A Electrochemical Analyzer
[Bioanalytical Systems (BAS), West Lafayette, IN]. Photovol-
tammetry and transient photocurrent-time profiles were re-
corded on a Model CV-27 BAS voltammograph. All the
electrolyte solutions (both for film preparation and for the
photoelectrochemical measurements) were sparged with ultra-
pure N2 for at least 20 min prior to use. All measurements
described below were performed at the laboratory ambient
temperature (25 ( 2 °C).
Other Instrumentation. For scanning electron microscopy
(SEM), a JEOL Model JSM 35C instrument was used, operating
at 15 kV. The samples were coated with Au-Pd alloy prior to
examination under the microscope. Scanning tunneling micros-
copy (STM) measurements were made with a Nanoscope E
(Digital Instruments, Santa Barbara, CA) Model ECM-NS3
instrument, using Pt/Ir tips (Nanotips; Digital Instruments) that
were electrochemically etched in a 20% KCN solution. The tips
Results and Discussion
Film Morphology and Topography. Figure 1 contains top
and cross-sectional views of electrodeposited TiO2 layers as seen
by SEM. The film is seen to be quite compact, and the grains
are nanometer-sized (see top SEM picture) in agreement with
STM data. Representative STM images of the electrodeposited
TiO2 film before (Figure 2a) and after (Figure 2b) photocatalytic
modification with selenium are contained in Figure 2. Table 1
contains data on the grain height, grain diameter, and grain
footprint (base area) computed from these microscopy data for
the two types of samples. Importantly, the grain footprint for
TiO2/Se films is seen to be smaller than that of the TiO2 host
pointing to an island topography for the deposited selenium.
Interestingly, this trend coincides with the island growth

Behavior of Se-Modified TiO2 Thin Film Electrodes
J. Phys. Chem. B, Vol. 109, No. 24, 2005 11955
Figure 3. Comparative SEM (a) and STM (b) images of electrode-
posited Se film on polycrystalline gold substrate. Films were grown
by a potentiodynamic scan between 0 and -0.6 V at 20 mV/s (a) and
at a constant potential of -0.7 V (b). Both films were prepared from
10 mM SeO2 + 0.1 M Na2SO4. For the STM image: Etip ) 1000 mV,
It ) 300 pA, and f ) 10 Hz.
Figure 2. STM images of the electrodeposited TiO2 film before (a)
and after (b) photocatalytic modification with selenium. For TiO2 the
tip bias (Etip) was 470 mV, the tunneling current (It) was 135 pA, and
the scan frequency (f) was 3.69 Hz. For TiO2/Se: Etip ) -4780 mV,
It ) 496 pA, and f ) 5.09 Hz.
TABLE 1: Morphological Data from Scanning Tunneling
Microscopy
Se was clearly seen. The Ti 2p1/2 and 2p3/2 signals were located
at 459.8 and 465.4 eV with a peak separation of 5.6 eV, almost
in exact agreement with literature values.14 The signal for Se
3d at 56.1 eV is to be compared with 55.3 eV reported in the
literature.14 From the high-resolution XPS data, a concentration
ratio, Ti:Se, of 89:11 (atom %) could be estimated.
grain height
(nm)
grain diameter
(nm)
grain footprint
(nm2)
sample
a
TiO2
3.8
2.5
10
9.6
6.5
17
72
33
229
TiO2/Sea
Au/Se
Photoelectrochemical Behavior. Photocurrent/potential pro-
files under chopped irradiation (0.1 Hz) for the individual
components, TiO2 and Se, are presented in Figure 4. The
photocurrent profiles for these two components will be useful
in the analysis of the behavior of the TiO2/Se bilayer (see
below). Both photocurrent-potential curves were taken in the
same supporting electrolyte, 0.1 M Na2SO4. The n- and p-type
behavior of TiO2 and Se is clearly seen in this comparison, with
the photocurrent signals being “up” (anodic) for TiO2 and
“down” (cathodic) for Se, respectively. The electrochemical
processes contributing to these photocurrents presumably are
the oxidation of water (for TiO2) and the reduction of protons
a Film supported by transparent conducting oxide substrate; see the
Experimental Section.
behavior of electrosynthesized selenium on gold electrodes
(Figure 3) as revealed by both SEM (Figure 3a) and STM
(Figure 3b) data. Also included for comparison in Table 1 are
the STM-derived grain morphology data for these samples. The
discontinuous nature of the Se deposits on the TiO2 surface
has important ramifications for device modeling as elaborated
later.
Surface Analysis of TiO2/Se Films by XPS. Aside from the
expected intense signals from Ti and O (in TiO2), a peak for

11956 J. Phys. Chem. B, Vol. 109, No. 24, 2005
de Tacconi et al.
Figure 4. Photocurrent-potential profiles at 0.1 Hz chopped irradiation
for electrodeposited TiO2 (top) and Se (bottom) films in 0.1 M Na2-
SO4. The two films were electrodeposited on TCO and Au substrates,
respectively (see the Experimental Section). Photovoltammograms were
obtained at 2 mV/s by using the full output of a 75 W Xe lamp. See
the Experimental Section for the film electrodeposition conditions.
Figure 6. Comparison of the effect of excitation wavelengths on the
photoresponse of the n-TiO2/p-Se/electrolyte junction as provided by
transient photocurrent profiles: (a) Xe lamp and (b) W-halogen lamp.
Figure 7. Effect of bias potential on the transient photoresponse for
the n-TiO2/p-Se/electrolyte junction in 0.1 M Na2SO4 under Xe arc
lamp irradiation.
Figure 5. Photocurrent-potential profiles at 0.1 Hz chopped irradiation
for n-TiO2/p-Se/electrolyte junction diodes under irradiation with the
Xe arc lamp (a) and with a W-halogen lamp (b). The inserts are
schematic sketches of the interface and the type of irradiation used.
Other conditions are as in Figure 4.
Further insights into the effect of excitation wavelengths on
the photoresponse of the n-TiO2/p-Se/electrolyte junction are
provided by the transient photocurrent profiles in Figure 6. In
Figure 6a for Xe arc lamp irradiation, the dominant photoanodic
response is from the TiO2 component and only a trace of the
faster photocathodic (spiked) response, attributable to the Se
component, is seen. On the other hand, this faster photocathodic
spike is much more prominent under W-halogen excitation
(Figure 6b). The two sets of data in Figure 6 were obtained at
a fixed potential of -0.14 V.
The effect of polarization potential (of the n-TiO2/p-Se/
electrolyte junction) on the transient photoresponse is
brought out by the data contained in Figure 7 with Xe arc
irradiation. Clearly, the net influence of the two variables
(excitation wavelength and polarization potential) on the
photoresponse is not dissimilar (cf., Figures 6 and 7). A
systematically more negative bias potential (driving the
electrons to the electrolyte side of the junction) (Figure 7)
provokes a very similar type of photoresponse to exciting
dominantly the p-Se side of the n-TiO2/p-Se/electrolyte
junction.
(for Se) although no attempts were made to analyze for the
generated products in the two cases. The photooxidation of water
on the TiO2 surface is a well-established reaction.1,15,16 For the
Se case, other potentially competitive processes to the hydrogen
evolution reaction such as the photocorrosion of Se itself can
be ruled out on the basis of previous work in the literature.17
Figure 5 contains corresponding data for the n-TiO2/p-Se/
electrolyte junction under irradiation with the Xe lamp (Figure
5a) and using a W-halogen lamp (Figure 5b). The effect of the
much smaller absorption cross-section of the TiO2 component
with the visible light output of the lamp is clearly seen in two
ways: (a) The photoanodic output in the range of potentials
from -0.40 to ∼0 V is almost nonexistent in Figure 5b relative
to Figure 5a. (b) The S-shaped photoanodic envelope (that is a
signature for n-type semiconductor/electrolyte junction, refs 18
and 19) at potentials more positive than 0 V is drastically
diminished in Figure 5b relative to Figure 5a.

Behavior of Se-Modified TiO2 Thin Film Electrodes
J. Phys. Chem. B, Vol. 109, No. 24, 2005 11957
Figure 8. Equivalent circuit representation of the n-TiO2/p-Se/
electrolyte junction. VT is the total applied bias and iT is the total
photocurrent containing contributions from the two parallel circuit
branches.
Model Simulations. A very simple equivalent circuit model
consisting of photodiodes in series and parallel was used to
simulate the measured transient photoresponse profiles (Figures
6 and 7). Since the TiO2 surface is not completely covered by
Se (cf., Figures 2 and 3, and XPS assay), allowance must be
made for a parallel photocurrent flow pathway across the bare
TiO2/electrolyte interfaces. This is done in the simple equivalent
circuit representation of the n-TiO2/p-Se/electrolyte junction in
Figure 8. The following equations were used for simulating the
transient photocurrent profiles in Figures 6 and 7:
Figure 9. Simulation of the transient photoresponse for pristine TiO2
(a) and the effect of selenium on the simulated transient photoresponse
of TiO2 (b) with use of eqs 1 and 2. Both cases pertain to Xe arc
irradiation. The parameters used were (a) a1 ) 10 µA cm-2, b1 ) 2.5
jph ) a1[1 - e-b t] - a2[1 - e
]
(1)
(2)
-b2t
1
-b2t
s-1, a′ ) 10 µA cm-2, and (b) a1 ) 10 µA cm-2, b1 ) 2.5 s-1, a2 )
jd ) a′[1 - e-b t] - a′[1 - e
]
1
1
3.4 µA cm-2, b2 ) 0.7 s-1, a′ ) 10 µA cm-2, a′ ) 3.4 µA cm-2
.
1
2
1
2
In eqs 1 and 2, jph is the photocurrent density (jph ) iph/A, where
A is the geometric photoelectrode area) and jd is the corre-
sponding dark current density. The first terms in eqs 1 and 2
represent the contribution from the bare TiO2/electrolyte contacts
and the second terms account for the parallel photodiode-in-
series branch representing the n-TiO2/p-Se/electrolyte junction
(Figure 8). The opposite signs for the two terms in eqs 1 and 2
are consistent with the fact that a n-type semiconductor (TiO2)
contacts the electrolyte in the first branch while a p-type
semiconductor (Se) contacts the electrolyte in the second
(parallel) branch of the overall circuit. The dominance of one
or the other photodiode in this second branch depends on the
polarization potential: The top photodiode representing the p-Se/
electrolyte junction conducts at potentials more negatiVe than
its photocurrent turn-on potential (ca. 0.20 V) while the bottom
one associated with the n-TiO2/p-Se junction conducts at
potentials more positiVe than its corresponding turn-on potential
(ca. -0.20 V).
Figure 9a is what would be expected for the photoelectrochemi-
cal behavior of an irradiated n-type semiconductor (moderately
reverse-biased) in contact with an electrolyte. That is, the
photogenerated holes are driven to the electrolyte with conse-
quent generation of an anodic photocurrent. For the simulations
in Figure 9b, the contribution of n-TiO2/p-Se/electrolyte is
“turned on”, with inclusion of the second terms in eqs 1 and 2.
The effect of the parallel (photodiodes-in-series) branch
appears as a slow decrease of the stationary photocurrent ending
with a sharp transient at the light “off” point (Figure 9b). The
parameter values for the simulations in Figure 9b are again
contained in the corresponding caption. The simulated profiles
in Figure 9b may be compared with their experimental
counterpart in Figure 7 (bottom frame) while the corresponding
simulated profile for pristine TiO2 (Figure 9a) is entirely in
accord with experimental data (not shown here). Further, the
experimental profile for a reverse-biased case, shown in the top
frame of Figure 7, clearly approaches a situation where the
current is predominantly driven through the n-TiO2/electrolyte
branch.
Four (adjustable) parameters are contained in the above
equations to simulate the photocurrents and the dark currents
respectively, with the “a” terms (a1, a2, a′, and a′2) accounting
Figure 10 simulates the transient photocurrent profiles for
the two experimental situations (Figure 6) where the light source
was switched from a Xe arc to a W-halogen lamp. Thus Figures
6a and 10a are comparable as are Figures 6b and 10b; the
agreement between experiment and model simulation is satisfac-
tory in both cases. The values for the simulation parameters
are once again contained in the caption for Figure 10.
For simulating the experimental current-potential profiles
in Figure 5, the approach taken was to first consider the
photoelectrochemical behavior of the individual n-TiO2/
electrolyte and p-Se/electrolyte components and then combine
1
for the stationary current flow (i.e., plateau adjustment) and the
“b” terms (b1 and b2) representing the carrier transit time
constants across the two parallel current flow branches in Figure
8. All simulations were performed with MATHCAD 2000
software.
The behavior of pristine TiO2 (i.e., oxide surface containing
no Se islands), in contact with the electrolyte, is simulated in
Figure 9a. For this simulation, only the first terms in eqs 1 and
2 were deployed. The values for the simulation parameters are
contained in the caption. The resultant simulated profile in

11958 J. Phys. Chem. B, Vol. 109, No. 24, 2005
de Tacconi et al.
TABLE 2: Nominal Valuesa of the Various Simulation
Parameters (see eqs A1-A3 in the Appendix)
value
parameter (units)
TiO2
Se
R (cm-1
)
4 × 105
4 × 10-6
8 × 10-5
0.025
2 × 105
4 × 10-7
4 × 10-6
0.075
W0 (cm/V1/2
L (cm)
)
D (cm2 s-1
)
k1 (cm s-1
k2 (cm s-1
kd (cm s-1
)
)
)
3 × 10-1
2 × 10-3
3 × 10-2
2 × 10-2
4 × 10-4
1 × 10-2
a For some simulations, the values shown were perturbed, see, for
example, Figure 12. These perturbations are identified in the corre-
sponding figure captions.
Figure 11. Simulations of the photocurrent-potential behavior for the
individual n-TiO2/electrolyte (dashed red line), p-Se/electrolyte (dashed
green line), and n-TiO2/p-Se/electrolyte (solid blue line). Parameters
are given in Table 2.
Figure 10. Simulation of the transient photocurrent profiles for the
two experimental situations shown in Figure 6 where the light source
was switched from a Xe arc (a) to a W-halogen lamp (b). The
parameters used were (a) a1 ) 20 µA cm-2, b1 ) 2.9 s-1, a2 ) 4.0 µA
cm-2, b2 ) 5.0 s-1, a′ 20 µA cm-2, a′ ) 0, and (b) a1 ) 1.6 µA
in this figure is the composite curve (solid line) for the two
photodiodes connected in series back-to-back. The simulation
of the n-TiO2/p-Se/electrolyte configuration was performed for
the cathodic and anodic branches of the current-potential profile
by using four i-V segments that correspond to the dark and
irradiated condition of the two semiconductors. Thus, for the
cathodic currents, the photocurrent originating from Se and the
dark current of TiO2 are combined and the related potentials
)
1
2
cm-2, b1 ) 2.9 s-1, a2 ) 1.4 µA cm-2, b2 ) 10 s-1, a′ ) 1.6 µA cm-2
,
1
a2′ ) 0.4 µA cm-2
.
them appropriately (see below). While many models are
available for simulating the current-potential profiles for
semiconductor/electrolytes,1,18-26 we chose the one by Wilson21
for its simplicity, transparency, and versatility. Thus the
underlying model assumptions are the same as those in the
original Wilson treatment (uniform doping level in the semi-
conductor, carrier recombination confined to the surface, etc.;
see refs 21 and 22). The current-potential relationship for each
component interface (see above) is computed as follows:
TiO2
Se
that satisfy the condition id
) iph were inserted in the
equation for iphSe and plotted as the cathodic photocurrent branch
(cathodic solid line). In the same way, in the anodic side, the
TiO2
condition to satisfy is iph
) idSe; again the resulting potentials
were used to obtain the combined current response but now for
the positive current, i.e., the anodic photocurrent branch (anodic
solid line). This equivalence of anodic and cathodic current
branches is implicit in considerations of particulate semiconduc-
tor suspensions,27,28 and in fact is a concept inspired by the
corrosion community.29,30
TiO2
TiO2
iTiO ) iph
+ id
(3)
(4)
2
iSe ) iphSe + id
Se
Recalling that the Wilson model18,19,21,22 contains an electron-
transfer constant, k1, Figure 12 illustrates the effect of varying
this model parameter for n-TiO2/electrolyte (Figure 12A) and
p-Se/electrolyte (Figure 12B). The corresponding effect on the
composite n-TiO2/p-Se/electrolyte response is also shown in the
figure as the solid lines. While the “dark” current branches are
unaffected, the variation in k1 manifests as a shift in the
photocurrent branch along the potential axis.
For fitting the above model to experimental data, the parallel
current branch attributable to the exposed n-TiO2 regions (Figure
8) must be combined with the composite response obtained from
the n-TiO2/p-Se/electrolyte junction. Thus the solid curves in
Figures 11 and 12 are in turn combined with the simulated
In eqs 3 and 4, i is the total current, iph is the photocurrent, and
id is the dark current and the “TiO2” and “Se” subscripts/
superscripts denote the two interfaces, respectively. We utilized
the conventional diode expression2,18,19 for the dark current flow,
i.e., exponential dependence of current on potential (see the
second terms in the model expressions in the Appendix). The
Appendix enumerates the model equations underlying the
simulations discussed next. Table 2 contains values for the
various parameters for the simulations.
Figure 11 contains simulations of the photocurrent-potential
behavior for the individual n-TiO2/electrolyte and p-Se/
electrolyte interface components (dashed curves). Also included

Behavior of Se-Modified TiO2 Thin Film Electrodes
J. Phys. Chem. B, Vol. 109, No. 24, 2005 11959
Figure 13. Comparison of the simulated (solid red lines) and the
experimental photocurrent-potential curves for n-TiO2/p-Se/electrolyte.
The latter are reproduced from Figure 5. The equivalent circuit shown
in Figure 8 was used and the parameters are listed in Table 2 for the
two photodiodes back-to-back. The parallel current branch for the
exposed n-TiO2 regions to the electrolyte uses S ) 1 (see eq A3).
in regions not covered by Se. The existence of a space charge
layer underpins the classical Wilson model albeit for the single
crystal case. In our nanostructured film case, these layers
probably encompass several TiO2 grains. The close similarity
of the voltammogram to that seen for an irradiated single crystal
semiconductor electrode18,19 is also indicative of the fact that
the bias potential acts across the semiconductor in a not too
dissimilar fashion in the two cases.
An additional factor lends further credence to our model. The
twin-plateau voltammogram profile that is experimentally
observed for the current-potential profiles for the n-TiO2/p-
Se/electrolyte interface (Figure 5) shows that the influence of
the n-TiO2/electrolyte and n-TiO2/p-Se/electrolyte interfaces in
the overall composite photoelectrode system is additiVe. Note
that this twin-plateau profile is conspicuously absent for the
pristine n-TiO2/electrolyte interface containing no Se islands
(Figure 4a).
Figure 12. Simulated effect of varying the electron-transfer constant
on the resulting photocurrent-potential curves for TiO2 (A) and Se
(B); these are shown as dashed curves. The composite response for
the photodiode/electrolyte junction is shown as solid lines. Simulation
values for k1 are the following: (A) k1 ) 3 × 100 (a), 3 × 10-1 (b),
Ti
3 × 10-2 cm s-1(c); (B) k1 ) 2.0 × 10-3 (a), 2.0 × 10-2 (b), 2.0 ×
Se
10-1 cm s-1 (c). Other parameters are given in Table 2.
response of the n-TiO2/electrolyte interface to afford the
simulated response of the entire assemblage comprised of p-Se
islands atop a (contiguous) n-TiO2 surface in contact with the
electrolyte. The resulting model simulations are shown as fits
(solid lines enveloping the data spikes) to the experimental data
(reproduced from Figure 5) in Figure 13. The agreement
between experiment and our first-order model is satisfactory.
Clearly, the twin-plateau signature in the experimental photo-
current-potential curves (e.g., Figure 5a) is only accommodated
by taking into account the exposed n-TiO2/electrolyte regions
(i.e., those not covered by p-Se) of the interface. This second
plateau in the anodic photoresponse (Figures 5a and 13a) can
be viewed as arising from the effect of the reverse-biased n-TiO2
component of the overall n-TiO2/p-Se (photochemical diode)/
electrolyte interface.
Concluding Remarks
We have shown how photocatalysis and electrodeposition can
be profitably combined to afford novel monolithic composite
photoelectrodes comprised of a contiguous and compact TiO2
layer with isolated p-Se regions. These “photochemical diodes”
in contact with an inert electrolyte (e.g., Na2SO4) show
interesting transient photoresponses whose temporal profiles
depend on the wavelength distribution of the excitation light
source. A simple equivalent circuit model was used to simulate
both the transient photoresponse and the potentiodynamic
photoelectrochemical behavior.
General Discussion
Considering both the simplicity and first-order nature of our
model (see Figure 8), the good agreement between experiment
and simulations (cf., Figures 6 and 10, Figures 7 and 9, and
Figure 13) is gratifying, and undoubtedly the efficacy of our
model is rooted in the compactness of the underlying TiO2 layer
(see Figure 1). The lack of an open, microporous network
(permeated also by an electrolyte) within the TiO2 phase
presumably facilitates the development of space-charge layers
at the interfaces with the p-Se regions and with the electrolyte
Acknowledgments. This research was supported, in part, by
a grant from the U.S. Department of Energy, Office of Basic
Energy Sciences. The authors thank Ms. Sashikala Soma-
sundaram for technical assistance. We also thank Dr. Thierry
Pauporte´ (Laboratoire d’EÄ lectrochimie et Chimie Analytique,
UMR-CNRS 7575, EÄ cole Nationale Supe´rieure de Chimie,

11960 J. Phys. Chem. B, Vol. 109, No. 24, 2005
de Tacconi et al.
Paris) for the SEM data in Figure 1. We thank an anonymous
reviewer for constructive criticisms of earlier versions of the
manuscript.
Values for the various parameters that were used for the
simulations in Figures 12-14 are contained in Table 2. Values
of the “turn-on” potential for TiO2 and Se were taken as -0.45
and 0.20 V; these were obtained from experimental data as in
Figure 4.
Appendix
The following expressions were used for simulating the
current-potential behavior of the individual n-TiO2/electrolyte
and p-Se/electrolyte components. The expressions are adapted
from the Wilson model, ref 21.
References and Notes
(1) Nozik, A. J.; Memming, R. J. Phys. Chem. 1996, 100, 13061.
(2) Rhoderick, E. H. Metal-Semiconductor Contacts; Clarendon
Press: Oxford, UK, 1980.
(3) Nozik, A. J. Appl. Phys. Lett. 1977, 30, 567.
(4) Rajeshwar, K.; de Tacconi, N. R.; Chenthamarakshan, C. R. Chem.
Mater. 2001, 13, 2765.
(5) Khaselev, O.; Turner, J. A. Science 1998, 280, 425.
(6) Chenthamarakshan, C. R.; Ming, Y.; Rajeshwar, K. Chem. Mater.
2000, 12, 3538.
(7) Somasundaram, S.; Chenthamarakshan, C. R.; de Tacconi, N. R.;
Ming, Y.; Rajeshwar, K. Chem. Mater. 2004, 16, 3846.
(8) Tan, T. T. Y.; Zaw, M.; Beydoun, D.; Amal, R. J. Nanopart. Res.
2002, 4, 541.
(9) Natarajan, C.; Nogami, G. J. Electrochem. Soc. 1996, 143, 1547.
(10) Shen, P. K.; Syed-Bokhari, J.; Tseung, A. C. C. J. Electrochem.
Soc. 1991, 138, 2778.
(11) Switzer, J. A. Am. Ceram. Bull. 1987, 66, 1521.
(12) Karuppuchamy, S.; Nonomura, K.; Yoshida, T.; Sugiura, T.;
Minoura, H. Solid State Ionics 2002, 151, 19.
(13) Han, L. M.; Rajeshwar, K.; Timmons, R. B. Langmuir 1997, 13,
5941.
(14) Wagner, C. D.; Riggs, W. M.; Davis, L. E.; Mulder, J. F.;
Mattenberg, G. E. Handbook of X-Ray Photoelectron Spectroscopy; Perkin-
Elmer Corp. (Physical Electronics Division): Eden Prairie, MN, 1979.
(15) Rajeshwar, K.; Singh, P.; DuBow, J. Electrochim. Acta 1978, 23,
1117.
(16) Rao, M. V.; Rajeshwar, K.; Pai Verneker, V. R.; DuBow, J. J.
Phys. Chem. 1980, 84, 1987.
(17) Frese, K. W., Jr. J. Appl. Phys. 1982, 53, 1571.
(18) Rajeshwar, K. In Electron Transfer in Chemistry; Balzani, V., Ed.;
Wiley-VCH: Weinheim, Germany, 2001; Vol. 4, p 279.
(19) Rajeshwar, K. In Encyclopedia of Electrochemistry: Semiconductor
Electrodes and Photoelectrochemistry; Licht, S., Ed.; Wiley-VCH: Wein-
heim, Germany, 2002; Vol. 6, p 1.
(20) Butler, M. A. J. Appl. Phys. 1977, 48, 1914.
(21) Wilson, R. H. J. Appl. Phys. 1977, 48, 4292.
(22) Wilson, R. H. CRC Crit. ReV. Solid State Mater. Sci. 1980, 10, 1.
(23) Reichman, J. Appl. Phys. Lett. 1980, 36, 574.
(24) Reiss, H. J. Electrochem. Soc. 1978, 125, 937.
(25) Fonash, S. J. Solar Cell DeVice Physics; Academic Press: San
Diego, CA, 1981.
(26) Hammett, A. In ComprehensiVe Chemical Kinetics; Compton, R.
G., Hancock, G., Eds.; Elsevier: Amsterdam, The Netherlands, 1999; pp
223-280. See also references therein.
(27) Miller, D. S.; Bard, A. J.; McLendon, G.; Ferguson, J. J. Am. Chem.
Soc. 1981, 103, 5336.
(28) Rajeshwar, K.; Ibanez, J. G. J. Chem. Educ. 1995, 72, 1044.
(29) Wagner, C.; Traud, W. Z. Elektrochem. 1938, 44, 391.
(30) Steigerwald, R. F. Corrosion 1968, 24, 1.
Ti
k1
RTiLTi
1 + RTiLTi
TiWTi
iTiO
)
1 - e-R
+ e-RTiWTi
×
Ti
(
)
k1Ti + k2
2
[
LTi
Ti
+ kdTi[1 - e-k(V-V )] (A1)
on
DTi
STi
L +
(
)
]
Ti
-kS1e
RSeLSe
1 + RSeLSe
SeWSe
iSe
)
1 - e-R
+ e-RSeWSe
×
(
)
kS1e + kS2e
[
LSe
Se
- kdSe[1 - e-k(V-V )] (A2)
on
DSe
L
(
+
)
]
Se
SSe
In eqs A1 and A2, k1 and k2 are the electron transfer and
recombination constants (units of cm s-1, ref 21), R is the
absorption coefficient (cm-1), W is the depletion layer width
[W ) W0(V - Von)1/2], L is the diffusion length of minority
carriers (holes for TiO2 and electrons for Se) in the semiconduc-
tor bulk [L ) (Dτ)1/2, where D is the diffusion coefficient (cm2
s-1) and τ is the transit time (s)], kd is the dark current constant,
k ) q/kT ) 38.65 V-1 (where q is the electronic charge, k is
the Boltzmann constant, and T is the absolute temperature), and
Von is the photocurrent “turn-on” potential, a parameter whose
value is close to the flat-band potential. In each of the
expressions, S is a boundary-condition parameter given by:21
S ) (k1 + k2)ek(V-V
-
)
on
D(1 - e-RW
)
(1 - e-RW
RL
)
-1
1 +
(A3)
RL
RL + 1
Le-RW
e-RW
[
]
(
)
[
]
RL + 1