Electrochimica Acta 125 (2014) 266274

Contents lists available at ScienceDirect

Electrochimica Acta
journal homepage: www.elsevier.com/locate/electacta

Behaviour of Titanium-based Fe2 O3 Photo-Anodes in

Photo-Electrochemical Reactors for Water Splitting
C.K. Ong, S. Dennison, S. Fearn, K. Hellgardt, G.H. Kelsall
Department of Chemical Engineering, Imperial College London, London SW7 2AZ, UK

a r t i c l e

i n f o

a b s t r a c t

Article history:
Received 8 October 2013
Received in revised form 10 January 2014
Accepted 14 January 2014
Available online 28 January 2014
Keywords:
Photo-electrochemical reactor
spray pyrolysis
titanium
iron oxide
hydrogen

The behaviour of titanium substrates was investigated as replacements for transparent uorine-doped
tin oxide-coated glass substrates, the low conductivity of which causes inhomogeneous spatial distributions of potential and current densities on scaling up photo-electrodes in photo-electrochemical reactors.
Hence, Fe2 O3 (ca. 23 nm thick) was deposited by spray pyrolysis onto the titanium surface at 450 C in
air, causing growth of (less photo-active) rutile by thermal oxidation, the thickness being measured as ca.
20 nm by secondary ion mass spectrometry. Anodising titanium in aqueous phosphoric acid enabled ca.
12 nm anatase TiO2 to be grown prior to spray pyrolysis, limiting the subsequent thermal growth of rutile,
so the overall TiO2 thickness was ca. 16 nm. 23 nm thick Fe2 O3 deposited on a 22 nm TiO2 | Ti substrate
after heat treatment at 500 C for 1 hour was found to produced greater current densities than Fe2 O3
deposited onto TEC-8 uorine-doped tin oxide coated on glass; uv absorption by TiO2 and reection by
the titanium substrate contributed to measured photo-current densities.
2014 Elsevier Ltd. All rights reserved.

1. Introduction
Photo-electrochemical production of hydrogen [1] from water
using solar energy and its subsequent use in fuel cells, could obviate
the predicted consequences of CO2 emissions from fossil-fuelled
power stations. The feasibility of such solar-powered processes is
well established; suitable semiconducting lms in contact with
water absorb photons with energies greater than their band gap
energy, produce electron-hole pairs (excitons) by:
semiconductor + h

absorption

semiconductor(hv+b + ecb
)

(1)

recombination

Suitably energetic electron holes in the semiconductors

valence band (vb) could then oxidise water to oxygen by reaction
(2) if Evb /e > UO2 /H2 and electrons in the semiconductors conduction band (cb) could reduce water to hydrogen by reaction (3),
if Ecb /e < UH + /H2 , where E represents energies of the band edges
and U represent the corresponding (relative) electrode potentials:
2H2 O + 4h+
O2 + 4H +
vb

(2)

2H + + 2ecb
H2

(3)

Corresponding author.
E-mail addresses: g.kelsall@imperial.ac.uk, ock86@yahoo.com (G.H. Kelsall).
0013-4686/$ see front matter 2014 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.electacta.2014.01.086

In the photo-electrochemical reactor described below, a n-type

photo-anode was connected electronically to a platinized titanium
cathode, with anolyte and catholyte separated by an ion-permeable
membrane. This design was intended to maximise the effective
charge yield of reaction (3), by minimising the rates of hydrogenoxygen recombination (c.f. electron-hole recombination by the
reverse of (1)) and/or the rates of the reverse of reactions (2) and (3)
at the cathode and anode, respectively. Such reactors would enable
photon energy harvesting and chemical energy storage, thereby
compensating for their diurnal operation.
Most research on photo-electrochemical processes has been
focussed on the properties and behaviour of semiconducting
photon-absorbing phases, with little development of reactors [2,3]
in which to deploy them. While relaxing the design constraints
for achieving more homogeneous photon uxes, the use of (transparent) uorine-doped tin oxide (FTO) substrates was predicted to
result in large spatial distributions of electrical potential and hence
current densities [2].
The objective of the work reported below was to determine
whether such effects can be mitigated by using metallic substrates,
such as titanium, which would be subject to subsequent thermal
oxidation, increasing interfacial resistance, during spray pyrolysis
of semiconducting photon absorbing phases such as iron(III) oxide
(-Fe2 O3 ). Fe2 O3 was chosen as the photo-anode material because
2.2 eV band gap energy), is
it absorbs visible light (ca. 560 nm
stable under water oxidation conditions, is cheap to fabricate and
abundant [4].

C.K. Ong et al. / Electrochimica Acta 125 (2014) 266274

267

Nomenclature
Symbol
Ai
C
eE
H
I
I0
I
ji
kB
K
L
N
ND
Ri
ti
T
U
Wa
X
0
sc

Surface area of i (m2 )

Area specic capacitance (F m2 )
Electronic charge, 1.602 1019 (C)
Energy (J)
Characteristic length of electrochemical cell (m)
Current (A)
Incident light irradiance (W m2 )
Light intensity after absorption (W m2 )
Current density of reaction i (A m2 )
Boltzmann constant, 1.38 1023 (J K1 )
Ratio of light absorption for multiple and single
passes through absorber (A.8) (1)
Distance between current collectors on the same
electrode (m2 )
Number of passes through light absorbing semiconductor (1)
Donor density (m3 )
Area specic resistance of i ( m2 )
Thickness of material i (m)
Temperature (K)
Electrode potential vs. reference electrode (V)
Wagner number (1)
Distance at which the light is absorbed into the
semiconductor (m)
Permittivity of vacuum, 8.854 1012 (F m1 )
Relative permittivity of semiconductor (Fe2 O3 ), 26.4
(1)
Overpotential (E - Eeqm ) (V)
Electrical conductivity of phase i (1 m1 )
Absorption coefcient (m1 )

Subscript
A
Anode
Cb
Conduction band
Electrolyte
E
FTO
Fluorine-doped tin oxide substrate
Vb
Valence band

Fig. 1. Schematic of photo-electrochemical reactor system.

The current owing through the mid-point of the anode, I2 was

smaller than the current I1 owing near the current collector at the
edge:
I1
R2
=
>1
I2
R1

(4)

If the ratio of I1 and I2 is unity, the potential and current distribution would be uniform. The Wagner number [7] can be dened
by the ratio of area specic resistances Ra due to the faradaic reaction at the anode and that of the electrolyte Re , for the case when
the anode has metallic conductivity:

Ra
Wa =
=
Re

Aa da
dI

h/e

(5)

Where h is the characteristic length. If h1 h2 , then h = h2 . If

h2 h1 , then h = h1 . e is the conductivity of the electrolyte, while
Ra = Aa .da /dI. The larger Wa, the more uniform the current density
distribution; Wa = 0 corresponds to the primary current distribution for which effects of electrode kinetics can be ignored. When
the anode has nite conductivity, the Wagner number can be redened as the ratio of polarisation resistance and the sum of resistance
in the electrolyte and the resistance in the FTO at the mid-point
between two current collectors (Fig. 1):
Ra
Re + RFTO

Wa =

(6)

Since Re and RFTO can be expressed as:

2. Potential and current density distributions
Electrochemical processes are highly sensitive to electrode
potentials, the spatial distributions of which, together with the
resulting current density distributions, are major design considerations in the scale up and optimisation of electrochemical
systems in general, the objective being to minimise spatial inhomogeneities [5]. In photo-electrochemical water splitting research,
optically transparent materials, such as uorine-doped tin oxide
(FTO) coated glass, are commonly used as substrates, onto which
absorbing semiconductor photo-electrode materials are deposited.
The thin (ca. 300 nm [6]) TEC-8 FTO coating has low in-plane electrical conductivity, so scale up simply by increasing the geometric
photo-electrode area would result in large potential distributions
between the current collectors. A simple photo-electrochemical
reactor system consisting of two at plate electrodes is illustrated
in Fig. 1. Current is collected at the edges of the electrodes, of length
L, height h1 and distance apart h2 .
The equivalent circuit of the reactor system can be approximated as a resistance in series problem shown in Fig. 2. Since a
metallic cathode was used, its resistance was neglected by comparison with the other resistances, so it was treated as an equipotential
surface.

Re =

h
e

(7)

L
L 2
L h1 =
FTO tFTO
FTO tFTO h1

RFTO =

(8)

Where tFTO is the thickness of the FTO layer, which usually is

ca. 300 nm [6] and L is the longest distance away from the current
collector, in this case, L = L. Hence:
RFTO + Re =

L 2
h
L 2 e + h FTO tFTO
+
=
FTO tFTO
e
e FTO tFTO

(9)

On substituting equation (9) into equation (6), the modied

Wagner number is given by:
Wa =

 A d 
a
a
dI

e FTO tFTO
L 2 e + h FTO tFTO

(10)

So the ratio of the two Wagner numbers is:

Wa
=
Wa

L 2 e
1+
h FTO tFTO

(11)

Hence, potential and current density distributions are predicted

to be more uniform if:

268

C.K. Ong et al. / Electrochimica Acta 125 (2014) 266274

Fig. 2. Equivalent analogue circuit of photo-electrochemical reactor describing potential distribution due to the sheet resistance of FTO.

a) The distance between the mid-point of the electrode to the current collector, L is smaller;
b) The conductivity of the substrate is greater;
c) The thickness of the FTO substrate is increased/its resistance is
decreased;
d) The electrolyte conductivity is smaller;
e) The distance between the anode and cathode is larger.
Option (c) does not apply to transparent conductive oxides glass,
as thicker material would make the substrate opaque and defeats
the purpose of having the photoactive material on a transparent
substrate.
Options (d) and (e) are not desirable, as lower electrolyte conductivity and increasing the distance between the anode and
cathode will increase the ohmic potential drop between the two
electrodes, and hence increase the required cell potential difference
for oxygen and hydrogen evolution reaction at constant current
density, or decrease current densities at constant potential difference.
The ratio of the Wagner numbers is an important scale-up
parameter to maintain active area of the photo-electrode. One possible means to obviate the problem of potential distribution due to
the FTO substrates sheet resistance is to replace it with a metallic
substrate, such as titanium and stainless steel. The choice of metallic substrate would be determined by the stability of such metals
in alkaline solution and under oxygen evolution conditions.
In the derivation of these Wagner numbers, light absorption by
a photo-electrode was not considered. It was assumed that the
optical window would be larger than the semiconductor used as
photo-electrode, and the photon ux to the semiconductor would
be uniform and impeded by any obstruction between the light
source and semiconductor. Shadows cast on a photo-electrode
surface would generate spatially inhomogeneous photocurrents
and photovoltages, the effects of which were ignored in this
analysis.

water, then immersed in 0.5 M oxalic acid for 15 minutes to minimise the thickness of the surface oxide layer by its dissolution as
(predominantly) TiO(C2 O4 ) and [TiO(C2 O4 )2 ]2 species. The TiO2 /Ti
substrates were then clamped to the hot plate for Fe2 O3 deposition
by spray pyrolysis [8]. In order to study the effect of the interfacial
TiO2 thickness, TiO2 was grown electrochemically on an oxalic acid
treated titanium plate in H3 PO4 , prior to Fe2 O3 being deposited and
its photo-electrochemical behaviour determined [9].
3.2. Preparation of Fluorine-doped Tin Oxide on Glass Substrates
The behaviour of Fe2 O3 spray-pyrolysed onto TEC-8 uorinedoped tin oxide (FTO) on glass [Hartford Glass Inc., USA] was
compared with that on titanium substrates. The FTO was rst
cleaned with acetone and then with high purity water in the ultrasonic bath for 15 minutes.

3. Experimental
3.1. Preparation of Titanium substrate
A 0.89 mm thick titanium foil (Alfa Aesar) was polished rst
with p600 and p1200 silicon carbide paper, then with 0.3 m
Al2 O3 and 50 nm Al2 O3 , until a mirror-like nish was achieved.
The titanium foil was then cleaned with acetone and high purity

Fig. 3. Photo-electrochemical reactor operating under solar simulator.

C.K. Ong et al. / Electrochimica Acta 125 (2014) 266274

269

Fig. 4. Cyclic voltammograms (10 mV s1 ) of non-heat treated Fe2 O3 on titanium

substrate in dark and when illuminated under 42 W m2 white (Xe) light.
Fig. 6. SIMS depth prole of non-heat treated Fe2 O3 | Ti electrode with ca. 7 nm of
TiO2 .

3.3. Spray pyrolysis

Fe2 O3 was deposited by spray pyrolysis onto both FTO and
titanium substrates, using 0.1 M FeCl3 + 6 104 M SnCl2 as the
precursor solution [10,11]. Fe2 O3 was deposited at a measured
surface temperature of 450 C (Fluke 560 IR thermometer) and
precursor delivery ow rate of 2 cm3 min1 for 40 layers (ca.
600700 nm). The deposited samples were then placed in an oven
for post-deposition heat treatment at 500 C in air for 1 hour to
ensure complete oxidation of SnII and crystallisation [12] of the
Fe2 O3 .
3.4. Photo-electrochemical measurements
The deposited Fe2 O3 on FTO and titanium substrates were then
masked with acrylic lacquer (RS Components, UK), leaving an
exposed surface area of ca. 3 3 mm2 . The photo-anodes were
then inserted into a photo-electrochemical cell containing 1.0 mol
dm3 NaOH electrolyte solution, together with a platinised titanium mesh cathode and HgO|Hg reference electrode.
A potentiostat/galvanostat (MetroOhm PGSTAT302) was used
for voltammetric experiments with and without illumination from
a 300 W xenon lamp (LOT-Oriel Omni- 150). Light intensities were

Fig. 5. Picture of Fe2 O3 deposited on titanium substrate, before and after heat
treatment.

measured with an EPP2000-UVN-SR integrating sphere (StellarNet,

Tampa, Florida). The light was chopped at 87 Hz with a mechanical
chopper (Stanford Research Instruments, SR 540) and photocurrents were measured with a lock-in amplier (Stamford Research
Instruments, SR 830) interfaced to a computer. Electrochemical
impedance spectra were measured with a root mean square potential of 5 mV over the frequency range 10 Hz to 50 kHz. The reported
electrode potentials were controlled relative to a OH- | HgO| Hg reference electrode, the equilibrium potential of which was calculated
as: E (SHE)/V = 0.9262 - 0.0592.pH, so has a value of 0.093 V (SHE)
at pH 14.
0.1 0.1 m2 photo-anodes were mounted in a photoelectrochemical reactor, designed and fabricated in the authors
department, and using a highly reective mirror (4-6 Wave Mirror, Edmund Optics, UK), was illuminated by a solar simulator
(Abet Technologies, Model 11016A, 550 W m2 output; Milford,
Connecticut, USA), as illustrated schematically in Fig. 3.
3.5. Materials Characterisation
The surface morphologies of photo-anodes were imaged with
a high resolution eld emission gun scanning electron microscope (FEGSEM, LEO Germini 1525) at an accelerating voltage of
5 kV.

Fig. 7. SIMS depth prole of heat-treated Fe2 O3 | Ti electrode. Comparison with

data in Figure 6 shows that the original 7 nm of TiO2 increased to 22 nm during heat
treatment.

270

C.K. Ong et al. / Electrochimica Acta 125 (2014) 266274

Fig. 8. Cyclic photo-voltammograms (10 mV s1 ) for 1 mol NaOH dm3 | Fe2 O3 | TiO2
| Ti electrode, under 40 W m2 white (Xe) light illumination.

The elemental depth proles of the samples were carried out

using a secondary ion mass spectrometer (Ion-ToF SIMS5 ). Samples
were sputtered using a 2 keV Cs+ ion beam at 45 to the sample
surface. The ion beam current was 150 nA and rastered over an
area of 300 300 m2 . The analytical region was gated to the central 100 100 m2 of the crater bottom and analysed using a Bi+
ion beam in high current bunched mode (HCBM) for optimal mass
resolution, with an ion beam current of 1 pA. The CsM+ molecular signals were followed in order to avoid matrix effects at the
interfaces between the layers of the samples.
The nal crater depths were measured using a stylus prolometer (Tencor Alphastep 200 Automatic Step Proler) and depth
calibrations carried out assuming a constant sputter rate throughout the samples being analysed. Each layer thickness on the samples
was subsequently measured at 50 % of their maximum count value.
4. Results and Discussion
4.1. Characterisation of Fe2 O3 photo-anodes on titanium
substrates
Fe2 O3 was deposited onto metallic substrates such as titanium
in order to mitigate problems with potential distributions in

Fig. 9. Mott-Schottky plots for 1 mol NaOH dm3 | Fe2 O3 | TiO2 | Ti electrode at
10 kHz.

Fig. 10. Mott-Schottky plots (at 10 kHz) for 1 mol NaOH dm3 | Fe2 O3 | TiO2 | Ti
electrodes with 7 nm and 15 nm thick TiO2 interfacial layers.

substrates upon scaling up electrode areas, as predicted for FTO.

For an electrode area of 0.003 0.003 m2 , Fig. 4 shows that under
42 W m2 white xenon light illumination, the non-heat treated
23 nm Fe2 O3 | (non-deliberately grown) 22 nm TiO2 | Ti electrode
produced a photocurrent density of 0.4 A m2 at applied electrode
potential 0.5 V (vs. HgO|Hg) and exhibited a photo-current onset
at 0.2 V (vs. HgO|Hg).
4.1.1. Effect of heat treatment on Fe2 O3 deposited onto Ti
substrates
It has been reported that post-deposition heat treatment [13] is
necessary to enhance photo-current densities; this had signicant
effects on the interfacial TiO2 thickness and on subsequent photocurrent densities. Fig. 5 shows the colour of the electrode changed
as a result of heating the material for 1 hour at 500 C. The colour
changed from brownish orange to blue, most probably due to the
different thickness ratio of Fe2 O3 to TiO2 producing different interference colours [14]. Secondary ion mass spectrometry (SIMS) was
carried out to measure the thickness of interfacial TiO2 layer, which
was found to have increased from 7 nm (Fig. 6) to 22 nm (Fig. 7) after
heating for 1 hour at 500 C.

Fig. 11. Cyclic photo-voltammograms (10 mV s1 ) for 1 mol NaOH dm3 | Fe2 O3 |
TiO2 | Ti electrode, under 40 W m2 white (Xe) light illumination.

C.K. Ong et al. / Electrochimica Acta 125 (2014) 266274

271

Table 1
Summary of measured thicknesses of interfacial TiO2 of non-heat treated and heat
treated Fe2 O3 | TiO2 | Ti electrodes.
Pre-grown TiO2 lm
thickness/nm

0
6
12

Fig. 12. Cyclic voltammograms (10 mV s1 ) of 23 nm heat treated Fe2 O3 deposited

on different thicknesses of TiO2 at TiO2 | Ti substrate under white (Xe) light illumination.

As a result of the heat treatment of the Fe2 O3 | 7 nm TiO2 |

Ti electrode forming a 22 nm interfacial TiO2 layer, photocurrent
densities increased and the on-set potential for oxygen evolution
decreased (Fig. 8). The shift of the onset potential for the light
driven oxygen evolution reaction from ca. 0.4 V to 0.2 V (vs. HgO
| Hg) corresponded to the change in the at band potential (0.46 V
to0.65 V) calculated from the Mott-Schottky equation:
1
2
=
e 0 sc ND
C2

E EFB

kB T
e


(11)

where sc = 26.4 [15]. The plot is shown in Fig. 9. However, it was

still uncertain whether the shift in at band potential was a result
of the heat treatment of the Fe2 O3 or the increase in TiO2 thickness
or both.
The results for the non-heat treated Fe2 O3 suggested that, having increased the thickness of interfacial TiO2 from 7 nm to 15 nm
by electrochemical pre-oxidation of Ti, the at band potential did
not change (Fig. 10) and there was a slight increase in photocurrent
densities (Fig. 11) which could be related to the higher donor density, ND , determined in the thicker anatase TiO2 (Fig. 10). Fig. 11
shows the second of three consistent cyclic voltammograms of the
Ti | TiO2 | Fe2 O3 photo-anode in 1 mol NaOH dm3 under 40 W m2
white light illumination. Thicker anatase TiO2 , having an apparently
higher donor density, may have been due to more TiO2 being doped
by Fe2 O3 at the Fe2 O3 -TiO2 interface, which has been reported to
have photocatalytic properties [1618]. The absorption coefcient

Measured TiO2 thickness/nm

Non-heat treated

Heat treated

8
7
15

20
22
16

of Fe2 O3 at 450 nm was about 1.6 107 m1 [19], corresponding to

absorption depth of ca. 62 nm. Hence, all the incident photon ux at
the Fe2 O3 | TiO2 | Ti composite photo-anode would not have been
absorbed fully by the Fe2 O3 , so the residual photons with wavelengths < 390 nm (Eg (TiO2 ) = 3.2 eV [20]) would have been absorbed
by the anatase TiO2 .
4.1.2. Effect of Interfacial TiO2 Layer Thickness
As discussed in section 4.1.1, both the temperature during deposition and subsequent heat treatment would increase the thickness
of TiO2 . The actual thickness of the TiO2 was measured by secondary
ion mass spectrometry (SIMS) and the effect of TiO2 thickness was
investigated. Data in Table 1 implies that pre-grown TiO2 inhibited
the subsequent thermal growth of TiO2 during the spray pyrolysis
deposition and heat treatment.
Fig. 12 shows the effect of the TiO2 lm thickness on the normalised current density-potential prole of heat treated Fe2 O3 |
TiO2 | Ti under illumination by white (Xe) light. Highest (normalised) photocurrent densities were produced with the thickest
interfacial TiO2 thickness, because TiO2 also absorbed photons and
injected photo-generated holes into Fe2 O3 , so increasing photocurrent densities. The electrode with 16 nm thick interfacial TiO2
produced greater photocurrents than those of the 20 nm TiO2 sample, because the former had a pre-grown anatase phase [9], while
the thermally grown TiO2 at 500 C would have been mostly the
less photo-active rutile phase [21].
Although a thicker (anatase) TiO2 lm produced greater photocurrent densities, this does not imply that a thicker TiO2 is always
desirable. In the present case, TiO2 could act only as a secondary
photo-anode (like a tandem cell) because the Fe2 O3 was very thin.
The absorption depth of Fe2 O3 would have been 63 nm (absorption
coefcient 1.6 107 m1 ) corresponding to a wavelength of 450 nm
[19]. This means that with 23 nm thick Fe2 O3 , only 31 % of photon
energies would have been absorbed by Fe2 O3 . However, in principle, TiO2 is also not a good choice to build a tandem cell structure

Fig. 13. Band structure of n-n semiconductor junction when A) applied electrode potential U is positive to Fermi energy levels/e of Fe2 O3 and TiO2 ; B) applied electrode
potential is positive to the Fermi energy level/e of TiO2 but negative to that of Fe2 O3 .

272

C.K. Ong et al. / Electrochimica Acta 125 (2014) 266274

Fig. 14. Cyclic voltammogram (10 mV s1 ) of Fe2 O3 | TiO2 | Ti and Fe2 O3 | FTO
electrodes under 42 W m2 white (Xe) light irradiation, in 1 M NaOH solution.

with Fe2 O3 , because the energy of the TiO2 conduction band edge
lies at a greater energy level than the conduction band edge of Fe2 O3
[22], creating a Schottky barrier as shown schematically in Fig. 13.
Fig. 13 shows 2 different possible band structures at the n-n
semiconductor junction. In both cases, photo-generated electrons
would need to overcome the resulting Schottky barrier. However,
depending on the applied electrode potential, which determines
the net Fermi energy level, the majority of the electrons could
accumulate in the bulk of Fe2 O3 (Fig. 13A), or at the n-n junction
(Fig. 13B) and recombine with photo-generated holes. The ideal
structure would be fabricated by depositing ca. 63 nm of Fe2 O3 onto
a Ti substrate with minimum thickness (< 1 nm to allow electron
tunnelling [23]) of TiO2 at the interface. Spray pyrolysis may not
be the best choice of Fe2 O3 deposition technique, as the growth of
TiO2 is dependent on the temperature and time [21] for which the
titanium substrate remains on the hot plate.
4.1.3. Comparison of Fe2 O3 on FTO and Titanium Substrates
Fig. 14 shows a comparison of the photo-electrochemical performances of Fe2 O3 | TiO2 | Ti and Fe2 O3 | FTO, with the former
photo-anode producing higher normalised photocurrents (since
photocurrents were linearly proportional to the light intensity [24])
at electrode potentials greater than 0.3 V (vs. HgO|Hg). At 0.5 V (vs.
HgO|Hg), the normalised photocurrent density of Fe2 O3 | TiO2 | Ti
was ca. 3 102 A W1 , which was 3 times greater than the normalised photocurrent density of Fe2 O3 | FTO material with almost
the same thickness (ca. 23 nm) of Fe2 O3 . An exponential increase
in current density for Fe2 O3 | FTO above 0.65 V (vs. HgO|Hg) was
detected, due to the increase in dark current.

Fig. 16. Cyclic voltammograms (10 mV s1 ) of heat treated Fe2 O3 deposited on

22 nm TiO2 | Ti substrate under white (Xe) light illumination, with and without
UV lter and correction for reectance.

The SEM image in Fig. 15A shows that the Fe2 O3 on FTO had an
open structure, and by contrast, the Fe2 O3 | TiO2 | Ti (Fig. 15B) was
a rather dense material. The morphological differences probably
resulted from the differences in thermal conductivities, and hence
in local temperature gradients during spray pyrolysis, between the
FTO and Ti substrates.
Anatase TiO2 is an intrinsic n-type semiconductor (band gap
3.2 eV [20]), which absorbs light at wavelengths < 390 nm, so would
also have contributed to the measured photocurrent. Since the
Fe2 O3 was only ca. 23 nm thick (Fig. 7), most of the incident photons would have passed through it, been absorbed by the TiO2 and
reected by the mirror-like titanium substrate for further absorption by Fe2 O3 .
Fig. 16 shows that with a UV lter and using the Beer-Lambert
law to correct current densities for absorption due to reection,
current densities of Fe2 O3 | 22 nm TiO2 | Ti matched those of Fe2 O3
| (TEC-8) FTO after absorption in the second pass of light (Fig. A.1.),
assuming the absorption coefcient of Fe2 O3 to be 1.6 107 m1
[19]. Calculations of multiple pass light absorption can be found in
supplementary data.
Predictions of the extent of potential distributions developed in section 2 can be applied to photo-anodes on FTO
and titanium substrates for the following parameter values:
e = 23 S m1 [25], FTO = 4.5 105 S m1 [6], Ti = 2.5 106 S m1
[25], tFTO = 3 107 m [6], tTi = 8.9 104 m, h = 2 102 m. For
electrodes with L = 103 m, the ratio of Wagner numbers Wa/Wa
for FTO and titanium substrates are predicted by equation (11) as
1.01 and 1.00, respectively, whereas for L = 5 102 m, those ratios
become 22.3 and 1.00, respectively. Hence, the use of FTO substrates

Fig. 15. SEM image of 23 nm of Fe2 O3 deposited onto A) TEC-8 FTO substrate; B) 22 nm TiO2 | Ti substrate.

C.K. Ong et al. / Electrochimica Acta 125 (2014) 266274

Fig. 17. Cyclic voltammograms (10 mV s1 ) of heat treated Fe2 O3 , deposited on

0.1 0.1 m2 Ti substrate, operating in dark (solid line) and under solar simulator
(- - - -).

273

would approach the condition of an equipotential electrode surface.

The choice of electrode substrate would greatly affect the sensitivity of this modied Wagner number to the distance between the
current collectors and the active surface area of the electrode.
As a result of using titanium as the substrate for Fe2 O3 photoanode, the irradiance corrected photocurrent generated was ca.
3 A W1 (vs. HgO | Hg), while the irradiance corrected photocurrent
generated from Fe2 O3 -FTO photoanode was ca. 1.5 A W1 (vs. HgO |
Hg). The use of metallic substrates such as titanium could mitigate
the problem of lateral potential distributions, but their oxidation
introduces additional resistance normal to electrode | electrolyte
interfaces. Results showed that titanium with a mirror-like nish,
coated with layers of Fe2 O3 of thickness less than its absorption
length, would result in photon reection at the Ti | TiO2 interface,
back through the Fe2 O3 , so increasing the extent of photon absorption per unit thickness of the semiconductor. Titanium could be a
possible substitute for FTO as substrate, but the decision on which
substrate to use depends, inter alia, on the costs of titanium sheets
and of FTO, the latter also requiring embedded and encapsulated
silver wire grids as current collectors [26].
Acknowledgements
The authors thank the UK Engineering and Physical Sciences
Research Council for a grant EP/F00270X1 funding a studentship
for CKO.
Appendix A. Supplementary data
Supplementary data associated with this article can be found,
in the online version, at http://dx.doi.org/10.1016/j.electacta.
2014.01.086.
References

Fig. 18. Voltammograms (10 mV s1 ) of heat treated Fe2 O3 deposited on 0.1 0.1
m2 FTO substrate, operating in dark (solid line) and under solar simulator (- - - -).

is predicted to result in greater inhomogeneities in potential and

current distributions upon scaling up photo-anode surface areas,
requiring careful design of (encapsulated) current collectors.
4.2. Effect of scaling up Fe2 O3 -Ti photo-anodes
Sn-doped, heat treated Fe2 O3 was deposited onto a 0.1 0.1
m2 Ti substrate and subsequently produced the promising current
density-potential response shown in Fig. 17. The prole was not
exactly the same as that for the 0.003 0.003 m2 electrode (Fig. 8),
partly due to temperature gradient during the spray pyrolysis process, and minor curvature of the surface. However, comparison of
Fig. 17 with the corresponding compared to the current densitiespotential prole of Fe2 O3 on a 0.1 0.1 m2 FTO substrate (Fig. 18),
demonstrated that the Fe2 O3 -Ti electrode produced photo-current
densities more than double those on the FTO substrate.
5. Conclusions
Potential and current density distributions are important considerations when scaling-up electrochemical processes. A modied
Wagner number was derived, which includes the effect of substrate
sheet resistance on potential and current density distributions, and
provides insight of the extent of spatial inhomogeneities in those
distributions. Results shown that if the modied Wagner number which is closer to unity, the current and potential distribution

[1] K. Rajeshwar, Hydrogen generation at irradiated oxide semiconductor-solution

interfaces, J. Appl. Electrochem. 37 (2007) 765.
[2] C. Carver, Z. Ulissi, C.K. Ong, S. Dennison, G.H. Kelsall, K. Hellgardt, Modelling
and development of photoelectrochemical reactors for H2 production, Int. J.
Hydrogen Energy 37 (2012) 2911.
[3] S. Haussener, C. Xiang, J.M. Spurgeon, S. Ardo, N.S. Lewis, A.Z. Weber, Modeling, Simulation, and Design Criteria for Photoelectrochemical Water-Splitting
Systems, Energy & Environ.Sci 5 (2012) 9922.
[4] K. Sivula, F. Le Formal, M. Grtzel, Solar Water Splitting: Progress Using
Hematite (-Fe2O3) Photoelectrodes, ChemSusChem 4 (2011) 432.
[5] P.M. Robertson, The variation of current density and electrode potential with
electrode resistance and its role in cell design, Electrochim.Acta 22 (1977) 411.
[6] Q. Jia, Conductive Metal Oxide Thin Film, in Handbook of Thin Films Materials
Vol. [4,] Ch. 13, Academic Press, San Diego, 2002, p. 677.
[7] N. Ibl, Current Distributions in: E. Yeager, J.OM Bockris, B.E. Conway and S.
Sarangapani (Eds.), Comprehensive Treatise of Electrochemistry, Vol. 6, Ch 4,
Plenum Press, New York, 1983, p. 239.
[8] Y. Liang, R. van de Krol, Inuence of Si dopant and SnO2 interfacial layer on
the structure of the spray-deposited Fe2O3 lms, Chem. Phys. Lett 479 (2009)
86.
[9] J.F. McAleer, L.M. Peter, Photocurrent spectroscopy of anodic oxide lms on
titanium, Faraday Discuss 70 (1980) 67.
[10] C.J. Sartoretti, M. Ulmann, B.D. Alexander, J. Augustynski, A. Weidenkaff, Photoelectrochemical oxidation of water at transparent ferric oxide lm electrodes,
Chem. Phys. Lett 376 (2003) 194.
[11] C.J. Sartoretti, B.D. Alexander, R. Solarska, I.A. Rutkowska, J. Augustynki, R.
Cerny, Photoelectrochemical oxidation of water at transparent ferric oxide lm,
electrodes, J, Phys. Chem. B 109 (2005) 13685.
[12] F.S. Yen, W.C. Chen, M.M. Yang, C.T. Hong, Crystallite size variation of nano size,
Fe2O3 powders during - to - phase transformation, Nano Lett 2 (2002) 245.
[13] R. Morrish, M. Rahman, J.M.D. MacElroy, C.A. Wolden, Activation of Hematite,
Nanorod Arrays for Photoelectrochemical Water Splitting, ChemSusChem 4
(2011) 474.
[14] M.V. Diamanti, B.D. Curto, M. Pedeferri, Interference Colours of Thin Oxide,
Layers on Titanium, Color Res. Appl 33 (2008) 221.
[15] R.A. Lunt, A.J. Jackson, A. Walsh, Dielectric response of Fe2O3 crystals and thin,
lms, Chem. Phys. Lett. (2013), in press.
[16] C.L. Luu, Q.T. Nguyen, and S.T. Ho, Synthesis and characterization of Fe-doped
TiO2, photocatalyst by the sol-gel method, 1 (2010) 15008.

274

C.K. Ong et al. / Electrochimica Acta 125 (2014) 266274

[17] S.H. Othman, S.A. Rashid, T.I.M. Ghazi, N. Abdullah, Fe-doped TiO2, Nanoparticles produced via MOCVD: Synthesis, Characterization and Photocatalytic,
Activity, J. Nanomater 2011 (2011) 8.
[18] M. Effendi, Bilalodin, Effect of Doping Fe on TiO2 Thin Films Prepared by Spin,
Coating Method, International Journal of Basic and Applied Science 12 (2012)
107.
[19] A. Kay, I. Cesar, M. Grtzel, New Benchmark for Water Photooxidation by,
Nanostructured -Fe2O3 Films, J. Am. Chem. Soc 128 (2006) 15714.
[20] B. ORegan, M. Grtzel, A low cost, high efciency solar cell based on dye sensitized colloidal TiO2 lms, Nature 353 (1991) 737.
[21] E. Gemelli, A. Scariot, N.H.A. Camargo, Thermal characterization of commercially, pure titanium for dental applications, Mat. Res. 10 (2007) 241.

[22] Y. Xu, M. Schoonen, The absolute energy position of conduction and valence
bands, of selected semiconducting minerals, Am Mineral. 85 (2000) 543.
[23] S. Rhle, D. Cahen, Electron Tunneling at the TiO2/Substrate Interface Can,
Determine Dye-Sensitized Solar Cell Performance, J.Phys.Chem. B 108 (2004)
17946.
[24] A.J. Moul, K. Meerholz, Intensity-dependent photocurrent generation at the
anode, in bulk-heterojunction solar cells, Appl. Phys. B-Lasers O (2) (2008) 209.
[25] W.M. Haynes and D.R. Lide, CRC handbook of chemistry and physics, 93rd
edition., Boca Raton, CRC, Press, Inc., Florida, 2012-2013.
[26] W.J. Lee, P.S. Shinde, G.H. Go, E. Ramasamy, Ag grid induced photocurrent,
enhancement in WO3 photo-anodes and their scale-up performance toward,
photoelectrochemical H2 generation, Int. J. Hydrogen Energy (9) (2011) 5262.