10
Solar hydrogen production with nanostructured metal oxidesRoel van de Krol, * Yongqi Liang and Joop Schoonman Received 10th December 2007, Accepted 18th February 2008 First published as an Advance Article on the web 11th March 2008 DOI: 10.1039/b718969a The direct conversion of solar energy into hydrogen represents an attractive but challenging alternative for photo-voltaic solar cells. Several metal oxide semiconductors are able to split water into hydrogen and oxygen upon illumination, but the efficiencies are still (too) low. The operating principles of photo-electrochemical devices for water splitting, their main bottlenecks, and the various device concepts will be reviewed. Materials properties play a key role, and the advantages and pitfalls of the use of interfacial layers and dopants will be discussed. Special attention will be given to recent progress made in the synthesis of nanostructured metal oxides with high aspect ratios, such as nanowire arrays, which offers new opportunities to develop efficient photo-active materials for solar water splitting. Introduction Growing environmental concerns and an increasing energy demand drive the search for new, sustainable sources of energy. In 2001, the global energy consumption rate was 13.5 TW, of which more than 85% originated from fossil fuels. 1 To meet future demands, which are estimated to reach 27 TW in 2050, new sustainable energy sources have to be found. While no single source will be able to meet all our energy needs, solar energy may go a long way to meet this formidable challenge. 2 With 120 000 TW of solar energy striking the surface of the earth at any given moment, it is by far the most abundant clean energy source available. Most energy sources that are currently being scruti- nized (fossil fuels, wind, tide, hydropower, biomass) originate from solar energy. However, direct use of solar energy by e.g. solar thermal power or photo-voltaic solar cells is still limited, and only 0.04% of all energy is generated by photo-voltaics. By covering 0.16% of the surface of the earth with 10% efficient solar cells, 20 TW peak of power could be generated. 2 While 0.16% appears to be a small number, it should be realized that this corresponds to the surface areas of France and Germany combined. Covering such a large area with solar cells presents a daunting task, even when this is undertaken on a (de-central- ized) global scale. Interestingly, the solar cell market is currently growing by 35–40% per year, and is one of today’s fastest growing markets. With a total produced capacity of 1200 MW peak reported for 2004, 2 a sustained 35% annual growth rate would lead to a total installed capacity of 20 TW peak in 2036. One of the challenges with solar cells is that they only generate electricity during daytime. Hence, large-scale use of solar energy requires an efficient energy storage solution. So far, the only practical way to store such large amounts of energy is in the form of a chemical energy carrier, i.e., a fuel. Hydrogen is one of the prime candidates as a future energy carrier. It can be made by electrolysis of water, and upon combustion in a fuel cell it returns to its original form (water) without generating any harmful by-products. The future prospect of a so-called Hydrogen Economy has attracted much interest, and many research efforts are currently underway to develop new techno- logies for the production, storage, utilization, and transport of hydrogen. The most straightforward method to produce hydrogen from water and sunlight is by coupling an electrolyzer to a solar cell array. The current generation of commercially available electro- lyzers have electricity-to-hydrogen conversion efficiencies of up to 85%, which seems to make this an attractive route. However, to achieve a current density of 1 A cm 2 that is normally used in electrolyzers, a cell voltage of 1.9 V is required. Since the Dr Roel van de Krol received his PhD degree at Delft University of Technology (TUD). After a postdoc fellowship at MIT, he returned to TUD where he is currently an assistant professor. His main interest is in the solid state chemistry of metal oxides. His current research focuses on nanostructured photo-electrodes and photo-catalysts for water splitting and degradation of pollutants. Dr Yongqi Liang received his PhD degree on physical chemistry from Peking Univer- sity (Beijing, China) in 2005. He is currently a postdoctoral fellow at Delft University of Technology, where he works on photo-electrochemical H 2 generation with metal oxides. His main research inter- ests are on the application of nanostruc- tured materials (nanowires, nanotubes) in energy conversion technologies. Joop Schoonman is a professor of Inor- ganic Chemistry at Delft University of Technology. He has several honorary degrees at universities in Romania and Poland, and has been a visiting professor at Stanford University since 2004. His research interests include gas-phase synthesis and defect chemistry of nano- structured materials for energy conversion and storage, and ceramic membranes for gas separation. Delft University of Technology, Faculty of Applied Sciences, Department DelftChemTech / Materials for Energy Conversion and Storage (MECS), P.O. Box 5045, 2600 GA Delft, The Netherlands. E-mail: [email protected]; Fax: +31-15-2788668; Tel: +31-15-2782659 † This paper is part of a Journal of Materials Chemistry theme issue on hydrogen storage and generation. Guest editor: John Irvine. This journal is ª The Royal Society of Chemistry 2008 J. Mater. Chem., 2008, 18, 2311–2320 | 2311 FEATURE ARTICLE www.rsc.org/materials | Journal of Materials Chemistry Downloaded by University of Virginia on 19 August 2012 Published on 11 March 2008 on http://pubs.rsc.org | doi:10.1039/B718969A View Online / Journal Homepage / Table of Contents for this issue

Solar hydrogen production with nanostructured metal oxides

  • Upload
    joop

  • View
    215

  • Download
    1

Embed Size (px)

Citation preview

Page 1: Solar hydrogen production with nanostructured metal oxides

FEATURE ARTICLE www.rsc.org/materials | Journal of Materials Chemistry

Dow

nloa

ded

by U

nive

rsity

of

Vir

gini

a on

19

Aug

ust 2

012

Publ

ishe

d on

11

Mar

ch 2

008

on h

ttp://

pubs

.rsc

.org

| do

i:10.

1039

/B71

8969

AView Online / Journal Homepage / Table of Contents for this issue

Solar hydrogen production with nanostructured metal oxides†

Roel van de Krol,* Yongqi Liang and Joop Schoonman

Received 10th December 2007, Accepted 18th February 2008

First published as an Advance Article on the web 11th March 2008

DOI: 10.1039/b718969a

The direct conversion of solar energy into hydrogen represents an attractive but challenging alternative

for photo-voltaic solar cells. Several metal oxide semiconductors are able to split water into hydrogen

and oxygen upon illumination, but the efficiencies are still (too) low. The operating principles of

photo-electrochemical devices for water splitting, their main bottlenecks, and the various device

concepts will be reviewed. Materials properties play a key role, and the advantages and pitfalls of the

use of interfacial layers and dopants will be discussed. Special attention will be given to recent progress

made in the synthesis of nanostructured metal oxides with high aspect ratios, such as nanowire arrays,

which offers new opportunities to develop efficient photo-active materials for solar water splitting.

Introduction

Growing environmental concerns and an increasing energy

demand drive the search for new, sustainable sources of energy.

In 2001, the global energy consumption rate was 13.5 TW, of

which more than 85% originated from fossil fuels.1 To meet

future demands, which are estimated to reach 27 TW in 2050,

new sustainable energy sources have to be found. While no single

source will be able to meet all our energy needs, solar energy may

go a long way to meet this formidable challenge.2 With 120 000

TW of solar energy striking the surface of the earth at any given

moment, it is by far the most abundant clean energy source

available. Most energy sources that are currently being scruti-

nized (fossil fuels, wind, tide, hydropower, biomass) originate

from solar energy. However, direct use of solar energy by e.g.

solar thermal power or photo-voltaic solar cells is still limited,

and only 0.04% of all energy is generated by photo-voltaics. By

covering 0.16% of the surface of the earth with 10% efficient

solar cells, 20 TWpeak of power could be generated.2 While

0.16% appears to be a small number, it should be realized that

this corresponds to the surface areas of France and Germany

Dr Roel van de Krol received his PhD

degree at Delft University of Technology

(TUD). After a postdoc fellowship at

MIT, he returned to TUD where he is

currently an assistant professor. His main

interest is in the solid state chemistry of

metal oxides. His current research focuses

on nanostructured photo-electrodes and

photo-catalysts for water splitting and

degradation of pollutants.

Dr Yongqi Liang receiv

on physical chemistry fr

sity (Beijing, China)

currently a postdoctor

University of Technolo

on photo-electrochemi

with metal oxides. His m

ests are on the applica

tured materials (nano

in energy conversion te

Delft University of Technology, Faculty of Applied Sciences, DepartmentDelftChemTech / Materials for Energy Conversion and Storage(MECS), P.O. Box 5045, 2600 GA Delft, The Netherlands. E-mail:[email protected]; Fax: +31-15-2788668; Tel: +31-15-2782659

† This paper is part of a Journal of Materials Chemistry theme issue onhydrogen storage and generation. Guest editor: John Irvine.

This journal is ª The Royal Society of Chemistry 2008

combined. Covering such a large area with solar cells presents

a daunting task, even when this is undertaken on a (de-central-

ized) global scale. Interestingly, the solar cell market is currently

growing by 35–40% per year, and is one of today’s fastest

growing markets. With a total produced capacity of 1200

MWpeak reported for 2004,2 a sustained 35% annual growth

rate would lead to a total installed capacity of 20 TWpeak in 2036.

One of the challenges with solar cells is that they only generate

electricity during daytime. Hence, large-scale use of solar energy

requires an efficient energy storage solution. So far, the only

practical way to store such large amounts of energy is in the

form of a chemical energy carrier, i.e., a fuel. Hydrogen is one

of the prime candidates as a future energy carrier. It can be

made by electrolysis of water, and upon combustion in a fuel

cell it returns to its original form (water) without generating

any harmful by-products. The future prospect of a so-called

Hydrogen Economy has attracted much interest, and many

research efforts are currently underway to develop new techno-

logies for the production, storage, utilization, and transport of

hydrogen.

The most straightforward method to produce hydrogen from

water and sunlight is by coupling an electrolyzer to a solar cell

array. The current generation of commercially available electro-

lyzers have electricity-to-hydrogen conversion efficiencies of up

to 85%, which seems to make this an attractive route. However,

to achieve a current density of �1 A cm�2 that is normally used in

electrolyzers, a cell voltage of 1.9 V is required. Since the

ed his PhD degree

om Peking Univer-

in 2005. He is

al fellow at Delft

gy, where he works

cal H2 generation

ain research inter-

tion of nanostruc-

wires, nanotubes)

chnologies.

Joop Schoonman is a professor of Inor-

ganic Chemistry at Delft University of

Technology. He has several honorary

degrees at universities in Romania and

Poland, and has been a visiting professor

at Stanford University since 2004. His

research interests include gas-phase

synthesis and defect chemistry of nano-

structured materials for energy conversion

and storage, and ceramic membranes for

gas separation.

J. Mater. Chem., 2008, 18, 2311–2320 | 2311

Page 2: Solar hydrogen production with nanostructured metal oxides

Dow

nloa

ded

by U

nive

rsity

of

Vir

gini

a on

19

Aug

ust 2

012

Publ

ishe

d on

11

Mar

ch 2

008

on h

ttp://

pubs

.rsc

.org

| do

i:10.

1039

/B71

8969

A

View Online

thermodynamic potential required for splitting water into

hydrogen and oxygen is 1.23 eV, the overall energy conversion

efficiency has an upper limit of 65% (1.23/1.9). When combining

the electrolyzer with a 12% solar cell, the overall conversion

efficiency is limited to �8%.3

A potentially more attractive route is the use of photo-electro-

chemical cells. Here, the photo-active semiconductor is immersed

in water and the photo-generated electrons and electron-holes are

directly used to reduce and oxidize water, respectively. Since the

photo-active area is the same as the electrochemically active

area, much smaller current densities are allowed (10–30 mA

cm�2). This results in lower overpotentials and higher overall

energy conversion efficiencies.3 Moreover, the functions of solar

cell and electrolyser are now combined in a single device, which

is favourable in terms of packaging and overall systems costs.

This paper will start by describing the operating principles of

photo-electrochemical devices for water splitting. The require-

ments on the materials properties will be discussed in detail,

and we will outline how different device concepts, materials

combinations, and purpose-built nano-architectures can be

used to overcome some of the limitations imposed by conven-

tional materials.

Photo-electrochemical water splitting

The principle of operation for a photo-electrochemical cell

(PEC) for water splitting is presented in the energy diagram of

Fig. 1. When the material is illuminated with photons that

have an energy that is equal to or larger than the bandgap,

electrons are excited from the valence band into the conduction

band. In an n-type semiconductor, the electrons travel to the

back contact and are transported to the counter electrode where

they reduce water to form hydrogen gas. The electron-holes that

remain in the valence band migrate to the surface, where they

oxidize water to form oxygen gas.

The recombination of electrons and holes is prevented by the

presence of an electric field close to the surface of the semicon-

ductor. This field, indicated by a ‘bending’ of the energy bands

in Fig. 1, is formed during the formation of a Schottky-type

contact between the semiconductor and the aqueous electrolyte.4

During the formation of the Schottky contact, free electrons

from the bulk are trapped into surface states at the semicon-

ductor/electrolyte interface. This sheet of negative charge is

Fig. 1 Energy diagram of a PEC cell for the photo-electrolysis of water.

The cell is based on an n-type semiconducting photo-anode.

2312 | J. Mater. Chem., 2008, 18, 2311–2320

compensated by the remaining positive charges in the bulk,

i.e., the ionized (shallow) donors.

Assuming that all the current that flows through the outer

circuit corresponds to the water splitting reaction (i.e., absence

of any competing side reactions), the overall solar-to-hydrogen

conversion efficiency of the device can be determined from the

following expression:

hSTH ¼ jðVredox � VBÞPlight

(1)

Here, j is the photo-current density (A m�2), Plight is the incident

light intensity in W m�2, Vredox is the potential required for water

splitting, and VB is the bias voltage that can be added in series

with the two electrodes in order to assist the water splitting

reaction (Fig. 1). Vredox is usually taken to be 1.23 eV (at room

temperature) based on a Gibbs free energy change for water

splitting of 237 kJ mol�1. Alternatively, the enthalpy change

(286 kJ mol�1) is sometimes used, which corresponds to a redox

potential of 1.48 eV. There is some debate on which of these

values is the more appropriate one for photo-electrolysis.

Usually, the Gibbs free energy value is used which results in

more conservative efficiency numbers.

Suitable photo-electrode materials for efficient solar hydrogen

generation have to fulfil the following requirements:

1. Strong (visible) light absorption

2. High chemical stability in the dark and under illumination

3. Suitable band edge positions to enable reduction/oxidation

of water

4. Efficient charge transport in the semiconductor

5. Low overpotentials for the reduction/oxidation reactions

6. Low cost

The spectral region in which the semiconductor absorbs light

is determined by the bandgap of the material. The minimum

bandgap is determined by the energy required to split water

(1.23 eV) plus the thermodynamic losses (�0.4 eV)5 and the

overpotentials that are required at various points in the system

to ensure sufficiently fast reaction kinetics (�0.3–0.4 eV).6,7 As

a result, the bandgap should be at least 1.9 eV, which corre-

sponds to an absorption onset at �650 nm. The maximum value

of the bandgap is determined by the solar spectrum shown in

Fig. 2. Below 400 nm the intensity of sunlight drops rapidly,

imposing an upper limit of �3.1 eV on the bandgap. Hence,

the optimum value of the bandgap should be somewhere

between 1.9 and 3.1 eV, which is within the visible range of the

solar spectrum.

Several authors have made predictions for the maximum

attainable efficiency based on the bandgap of the material, the

solar spectrum, and estimated values for the thermodynamic

losses, overpotentials, optical reflection losses, etc.5–7 Most

recently, Murphy et al. predicted a maximum efficiency of

16.8% for a hypothetically ideal material with a bandgap of

2.03 eV.7 Higher efficiencies can be obtained by using

a multiple-bandgap system (see section on tandem cells below).

The second requirement, the stability against (photo-)corro-

sion, is a severe one that limits the usefulness of many

photo-active materials. Most non-oxide semiconductors, such

as Si, GaAs, GaP, CdS, etc. either dissolve or form a thin oxide

film which prevents electron transfer across the semiconductor/

electrolyte interface. Almost all metal oxide photo-anodes are

This journal is ª The Royal Society of Chemistry 2008

Page 3: Solar hydrogen production with nanostructured metal oxides

Fig. 2 Intensity of sunlight versus wavelength for AM1.5 conditions.

The grey area represents the part of the spectrum that can be absorbed

by a hypothetically ideal bandgap of 2.03 eV.7

Dow

nloa

ded

by U

nive

rsity

of

Vir

gini

a on

19

Aug

ust 2

012

Publ

ishe

d on

11

Mar

ch 2

008

on h

ttp://

pubs

.rsc

.org

| do

i:10.

1039

/B71

8969

A

View Online

thermodynamically unstable, which means that the photo-gener-

ated holes are able to oxidize the semiconductor.4 However, if

the kinetics of charge transfer across the interface (oxidation of

water) are faster than the anodic decomposition reaction,

photo-corrosion is avoided. For example, TiO2 and SnO2 show

excellent stability over a wide range of pH values and applied

potentials, while ZnO always decomposes in aqueous environ-

ments upon illumination. Fe2O3 represents an intermediate

case, for which the stability appears to depend on the presence

of dopants, pH, and oxygen stoichiometry.8,9 The general trend

is that wide-bandgap metal oxide semiconductors are stable

against photo-corrosion, while small bandgap semiconductors

are not. While this is in obvious conflict with the requirement

Fig. 3 Energy band positions for various semiconductors at pH 14.4,10–16 W

were extrapolated using �59 mV per pH unit. It should be noted that values

a volt. Most values shown here are for polycrystalline films, obtained using cap

between the flatband potential and the conduction band.

This journal is ª The Royal Society of Chemistry 2008

of visible-light absorption, it does not represent a fundamental

limitation.

The third requirement implies that the conduction and valence

band edges should ‘straddle’ the reduction and oxidation poten-

tials of water. Specifically, EC should be above Ered(H2/H+) and

EV should be below Eox(OH�/O2). Fig. 3 shows the band-edge

positions of various semiconductors, along with the reduction

and oxidation potentials of water. It should be noted that the

data in Fig. 3 are drawn for a pH of 14, and that the reduction

and oxidation potentials of water vary with �59 mV per pH

unit. Most metal oxides (and even several non-oxide semicon-

ductors) also show a �59 mV pH�1 variation in the flatband

potential.17 For these materials, the band positions do not

move with respect to Ered(H2/H+) and Eox(OH�/O2) and the

diagram can also be used at different pH values. Fig. 3 indicates

that most non-oxide semiconductors are able to reduce, but not

oxidize water. Conversely, most oxide semiconductors are able

to oxidize, but not reduce water. Since the stability criterion

favors oxide semiconductors, the reduction of water appears to

be challenging. In most cases, an externally applied voltage or

a separate electrode compartment with a different pH is neces-

sary to assist the reduction reaction.18 Indeed, only a few metal

oxides have been demonstrated to be able to evolve both oxygen

and hydrogen, and only SrTiO3 has been shown to photo-cleave

water in a two-electrode system without any assistance.19 The

efficiency was, however, less than 1% due to the large bandgap

of the material (3.2 eV).

The fourth requirement, that of efficient charge transport, is

easily fulfilled by some materials, while in others it is the main

cause of poor overall conversion efficiencies. The intrinsic

electron and hole mobilities are determined by the electronic

structure of the material.20 The conduction and valence band

of most (but not all) metal oxides are primarily composed of

hen no experimental data were available for this pH, the band positions

reported in the literature show significant scatter, up to a few tenths of

acitance (Mott–Schottky) measurements and corrected for the difference

J. Mater. Chem., 2008, 18, 2311–2320 | 2313

Page 4: Solar hydrogen production with nanostructured metal oxides

Dow

nloa

ded

by U

nive

rsity

of

Vir

gini

a on

19

Aug

ust 2

012

Publ

ishe

d on

11

Mar

ch 2

008

on h

ttp://

pubs

.rsc

.org

| do

i:10.

1039

/B71

8969

A

View Online

metal 3d orbitals and oxygen 2 p orbitals, respectively.21 Hence,

extensive overlap of the metal 3d orbitals will lead to high

electron mobilities, while the amount of overlap between the

O-2p orbitals determines the hole mobility. While orbital overlap

arguments may provide some initial guidelines for evaluating the

charge transport properties of a material, extrinsic factors such

as the presence of defects often play a much more important

role. Shallow donors or acceptors enhance the conductivity of

a semiconductor, whereas defects with deeper lying energy levels

usually act as recombination centers.

The fifth requirement implies that hole transfer across the

n-type semiconductor/electrolyte interface should be sufficiently

fast in order to compete with the anodic decomposition reaction.

Furthermore, it should also be fast enough to avoid the accumu-

lation of holes at the surface, as this would lead to a decrease of

the electric field and a concomitant increase in electron–hole

recombination. To improve the kinetics of hole transfer, cataly-

tically active surface species can be added. Perhaps the best

known example of an effective oxygen evolution catalyst is

RuO2, whereas Pt is usually employed as a catalyst for hydrogen

evolution.4,22

Fig. 4 Scanning electron micrographs of: A) fractal-shaped TiO2

deposited by chemical vapour deposition; B) Fe2O3 nanobelts obtained

by oxidizing Fe foil at 800 �C in air; C) and D) Fe2O3 nanoflakes

obtained by oxidizing Fe foil at 450 �C in air; E) 100 nm Fe2O3 film

obtained by thermal oxidation of an electrodeposited Fe film at 450 �C

in air; F) same as E) but for a film with twice the thickness.

Photo-anode materials

In the above, we have outlined the basic working principles of

photo-electrochemical cells and identified the main material’s

requirements. Despite extensive research efforts, mainly in the

1970s and early 1980s, no photo-active material has yet been

found that fulfils all these requirements. However, the exciting

progress made in the field of nanostructured materials during

the last decade offers new opportunities for solar water splitting.

We will illustrate this by discussing recent developments for two

widely studied photo-anode materials, TiO2 and Fe2O3.

TiO2 has been the most extensively investigated material for

photo-electrochemical and photo-catalytic applications. The

main reason for its popularity is its excellent chemical stability,

its low cost, and the fact that it is widely available. It is by far

the most popular material for photo-catalytic oxidation of

organic compounds in e.g. water purification. Efforts to use

this material in a photo-electrochemical device date back to

1972, with the publication of Fujishima and Honda’s landmark

paper in which they first demonstrated the possibility to split

water by shining light on a semiconducting photo-anode.18 Since

TiO2 only absorbs UV light due to its bandgap of 3.2 eV, most

research efforts were aimed at shifting the optical absorption

towards the visible part of the spectrum. Some success has

been achieved by doping the material with various transition

metal ions, such as Fe and Cr.23,24 These dopants introduce

additional energy levels in the bandgap, and excitation from

these levels to the conduction band requires less energy than

excitation of valence band electrons.

Although cation doping indeed results in a visible-light

response, the absorption coefficient is usually very small due to

low dopant concentrations (limited solubility) and small absorp-

tion cross-sections. To address this, work in our laboratory has

focused on developing highly structured TiO2 photo-elec-

trodes.25 The SEM image in Fig. 4a shows an example of a Cr

and Fe co-doped TiO2 film made by chemical vapor deposition

(CVD). Extensive light scattering in such a morphology

2314 | J. Mater. Chem., 2008, 18, 2311–2320

enhances the absorption coefficient, while the high aspect ratio

of these fractal-like structures ensures that the photo-generated

holes have to travel only short distances before reaching the

surface. This minimizes the chance for charge trapping and

recombination. The first fractal-shaped TiO2 electrodes did not

generate any photo-current under visible light, although sensiti-

sation with Ru-based dye molecules resulted in a solar cell with

an efficiency of 2.7%.25 Recently, however, we showed that

fractal-shaped TiO2 can also generate photo-currents under

visible light illumination by co-doping with Fe and Cr.26 Fig. 5

reveals that fractal-shaped TiO2 films have a remarkably

enhanced visible light response compared to undoped, dense

TiO2. This enhancement is many times higher than observed

for dense, smooth TiO2 films doped with Fe and Cr, indicating

that the nanostructured morphology plays a crucial role. Fig. 5

also shows that the enhanced absorption of visible light comes

at the cost of lower absorption at wavelengths below 350 nm.

This is attributed to the shorter penetration depth of the light

at these wavelengths, which leads to longer distances that the

electrons have to travel before reaching the back contact and

a higher chance of recombination at defect sites. An interesting

(though unfavourable) phenomenon observed for these fractal-

shaped films is that extensive degradation occurs during simulta-

neous exposure to light and to H2O or O2. Under these

conditions the color changes from the original cinnamon-brown

to white, and the optical absorption reverts back to that of

undoped TiO2. While these films are of limited use for

This journal is ª The Royal Society of Chemistry 2008

Page 5: Solar hydrogen production with nanostructured metal oxides

Fig. 5 Incident photon-to-current conversion efficiency (IPCE) of

a fractal-shaped TiO2 photo-anode doped with Cr and Fe and deposited

by CVD, compared to an undoped dense TiO2 film. The data for dense

TiO2 were recorded at 0 V vs. Ag/AgCl in an aqueous 0.1 M KOH

electrolyte (pH 13). Since the fractal-shaped TiO2 was unstable in water,

its (short-circuit) photo-current was measured in a two-electrode cell

using 0.5 M KI + 0.05 M I2 in propylene carbonate as the electrolyte.

The optical absorption of the nano-TiO2 film shows a similar spectral

shape as its IPCE.

Dow

nloa

ded

by U

nive

rsity

of

Vir

gini

a on

19

Aug

ust 2

012

Publ

ishe

d on

11

Mar

ch 2

008

on h

ttp://

pubs

.rsc

.org

| do

i:10.

1039

/B71

8969

A

View Online

photo-electrochemical water splitting, it does illustrate that one

can tune the material’s properties by adopting suitable

nanostructured morphologies.

Although substitutional doping with metal ions can indeed be

used to sensitize wide-bandgap semiconductors to visible light, it

does not significantly improve the overall efficiency of

photo-electrodes. This is because the dopants introduce defect

states deep in the bandgap which act as recombination centers.

In 2001, research interest in the sensitization of wide-bandgap

metal oxides by doping returned after a paper by Asahi et al.

on anion-doped TiO2.27 Quantum-chemical calculations

indicated that the wave functions of anion dopants, such as

nitrogen and carbon, show significant overlap with the valence

band wavefunctions. The sub-bandgap defect levels are,

therefore, less localized than for cation dopants, and recombina-

tion should be less severe. Nitrogen- and carbon-doping of TiO2

particles have indeed been shown to result in higher photo-

catalytic activities.28–30

Attempts to incorporate carbon or nitrogen into thin film

TiO2 photo-anodes have met with limited success. Spray

pyrolysis of TiO2 under a CO2 atmosphere31 or a high-tempera-

ture treatment of TiO2 in a hexane-containing environment32

result in carbon concentrations that are too low for a significant

change in the absorption spectrum. Interestingly, even a low

concentration of carbon shifts the anatase-to-rutile phase

transformation temperature from 600 �C to beyond 800 �C.

This can be useful in the synthesis of TiO2-based photo-catalysts,

since the anatase phase is the more active one. To increase the

concentration of carbon in TiO2, the oxidative annealing of

TiC films may be a more promising route.33 Sputter-deposition

in a CO2 containing atmosphere may also lead to enhanced

carbon concentrations. Lindgren et al. published some inter-

esting work on nitrogen-doped TiO2 films deposited by reactive

DC magnetron sputtering.34,35 An optimal concentration of

This journal is ª The Royal Society of Chemistry 2008

nitrogen improved the white-light response by a factor of 200

compared to undoped TiO2. However, nitrogen doping also

decreases the UV response,34 in a similar manner as observed

for the metal-doped TiO2 films in Fig. 5. This suggests that

recombination centers are difficult (if not impossible) to avoid

when doping wide-bandgap semiconductors, regardless of

whether anion or cation dopants are used.

An alternative approach is to start with a material that has an

intrinsically small bandgap, such as a-Fe2O3 (hematite). This

material has an orange-brown color and with a bandgap of

2.1–2.2 eV it readily absorbs light with wavelengths below

�600 nm. Moreover, it is stable against photo-corrosion in

aqueous solutions. Older work on Fe2O3 has already shown

high photo-conversion efficiencies for polycrystalline Fe2O3

pellets9 and single crystalline Fe2O3 samples.36 For practical

applications, however, thin films are preferred in order to avoid

electrical conductivity limitations and to reduce the cost for

fabrication. Several authors have reported the beneficial

effects of dopants, especially Ti and Si, on the photo-electro-

chemical properties of a-Fe2O3.37–40 Recently, an exceptionally

high photo-current of 2.7 mA cm�2 was reported at a bias

of 1.23 V vs. reversible hydrogen (RHE) under simulated

AM1.5 sunlight for Si-doped porous films synthesized using

atmospheric pressure chemical vapor deposition (APCVD).41,42

The key ingredients for the high performance of these films

seem to be i) Si doping, which is thought to induce a favorable

morphology during film growth, ii) the presence of an ultra-

thin (�1 nm) SiO2 interfacial layer between the Fe2O3 and the

transparent conducting substrate, and iii) the addition of a cobalt

catalyst.

To investigate the effects of Si doping in more detail, we have

deposited Si-doped Fe2O3 films by spray pyrolysis of an

ethanolic solution of Fe(AcAc)3.43 First, the effect of a 5 nm

SnO2 interfacial layer between the transparent conducting

substrate and the Fe2O3 was investigated. Although this layer

is significantly thicker than the 1 nm SiO2 interfacial film

mentioned above42 (and non-insulating), a similar beneficial

effect was observed. The current–voltage characteristics shown

in Fig. 6 (left) reveal that the SnO2 layer shifts the photo-current

onset potential to �0.2 V more negative values. This results in

a photo-current increase from 0.04 to 0.33 mA cm�2 at a potential

of 0.2 V vs. Ag/AgCl. The right-hand side of Fig. 6 shows that the

optimal Si concentration in the spray solution is �0.2% and �1%

for films with and without the interfacial layer, respectively.

Another effect of the interfacial layer is a significant improvement

of the reproducibility of the photo-response, especially for

undoped samples. Without the interfacial layer, photo-current

variations of up to a factor of �5 or more are observed between

samples made under the same conditions. With the interfacial

layer, these variations are within a factor of 2.

The fact that these films were dense (i.e., non-porous) allowed

us to study the electrical properties by quantitative impedance

spectroscopy. A donor density of 1 � 1020 cm�3 was found for

films deposited from a spray solution containing 0.9% of Si,

which is three orders of magnitude higher than for undoped

Fe2O3 films (�1 � 1017 cm�3).43 This shows that in addition to

the structure-directing effect reported by Cesar et al.,41 Si also

acts as a donor-type dopant in Fe2O3. This is consistent with

front- and back-side illumination experiments, which reveal

J. Mater. Chem., 2008, 18, 2311–2320 | 2315

Page 6: Solar hydrogen production with nanostructured metal oxides

Fig. 6 Left: I–V curves in the dark and under simulated 80 mW cm�2 AM1.5 sunlight for Fe2O3 doped with 0.2% Si with (A) and without (B) a 5 nm

SnO2 interfacial layer. Right: dependence of the photo-current on the Si doping concentration at 0.23 VAg/AgCl for Fe2O3 films with and without the

interfacial layer. In both cases, an aqueous solution of 1 M KOH (pH 14) was used as an electrolyte.

Dow

nloa

ded

by U

nive

rsity

of

Vir

gini

a on

19

Aug

ust 2

012

Publ

ishe

d on

11

Mar

ch 2

008

on h

ttp://

pubs

.rsc

.org

| do

i:10.

1039

/B71

8969

A

View Online

that electron transport is rate limiting for undoped Fe2O3 films,

whereas hole transport limits the photo-current in Si-doped

films.43

Poor hole transport is one of the main factors that limit the

photo-response of Fe2O3 photo-anodes. An elegant solution

for this is to use high aspect-ratio nanowire electrodes, as

illustrated in Fig. 7. By ensuring that the wires have a radius

that is less than the hole diffusion length of Fe2O3 (2–4 nm38

or 20 nm),44 hole transport limitations can be avoided. The large

specific surface area of a nanowire electrode addresses the

intrinsically slow water oxidation kinetics by valence band

holes.44 Since recombination at grain boundaries was found to

limit the photo-response in particle-based photo-anodes,45 each

wire should consist of a single crystallite.46 When the nanowires

are sufficiently narrow, one-dimensional quantum confinement

can occur. Vayssieres et al. have reported a significant blue-shift

for a-Fe2O3 nanorod arrays with a primary rod diameter of

4–5 nm.47 If the blue-shift manifests itself in an upward shift

of the conduction band, direct reduction of water may

become feasible without the need for an additional bias (cf.

Fig. 3).47 Thus far, however, no one has succeeded to split

Fig. 7 Optimized morphology for an a-Fe2O3 photo-anode for water

splitting. The small diameter of the nanowires ensures short hole

diffusion path lengths.

2316 | J. Mater. Chem., 2008, 18, 2311–2320

water using a-Fe2O3 photo-anodes without assistance of a bias

voltage.

In addition to hydrothermal methods for Fe2O3 nanowire

growth on conducting substrates,47,48 thermal oxidation of

metallic Fe can also result in high aspect-ratio morphologies.49–54

Oxidation of iron foil in air at 800 �C leads to the nanowires

with a diameter of 50–100 nm, as shown in Fig. 4B. At 450 �C,

a similar treatment results in the formation of high aspect-ratio

nanoflakes (Fig. 4C and D). These features have been observed

at different labs, and can be readily reproduced. Photo-electro-

chemical measurements on thermally oxidized Fe foil show very

high background currents and a photo-response of a few tens of

mA cm�2 under AM1.5 illumination.55 This is due to exposure of

the underlying metallic Fe/Fe3O4 substrate to the electrolyte,

and can be avoided by using thin films of Fe that can be fully

oxidized. However, for such films the thermal growth of nanowires

seems to be less straightforward than for Fe foil. While

sputter-deposited films have been reported to form nanowires

upon oxidation,50,54 we have found different morphologies after

oxidation of electrodeposited metallic Fe films, as can be seen in

Fig. 4E and F.55 The 100 nm Fe2O3 film in Fig. 4E shows porous

micron-sized features. When the thickness is doubled (Fig. 4F),

the feature size becomes significantly smaller and some high

aspect-ratio outcroppings are observed. The latter may be

interpreted as the onset of nanoflake growth, and suggests

a required minimum film thickness for nanowire growth by

thermal oxidation. Clearly, further progress in this area requires

a more detailed understanding of the growth mechanism during

thermal oxidation.

Tandem cells

A closer look at Fig. 2 and 3 reveals that metal oxide semicon-

ductors absorb only a small portion of the solar spectrum. By

using a tandem cell approach, the remaining part of the spectrum

can be used to provide the additional bias voltage required for

the reduction reaction (i.e., to satisfy the criterion of suitable

band edge positions). Augustynski and Gratzel used a dye-sensi-

tized solar cell (DSSC) to provide a bias voltage to a WO3 or

Fe2O3 photo-anode. An efficiency of 4.5% has been published

This journal is ª The Royal Society of Chemistry 2008

Page 7: Solar hydrogen production with nanostructured metal oxides

Fig. 8 Left: one-pot system containing a suspension of photo-catalyst

particles in water. Illumination occurs either from the side or from the

top of the reaction vessel. The hydrogen and oxygen need to be separated

to avoid explosive mixtures. Right: energy diagram for a particle-based

photo-catalyst for water splitting. For very small particles (<100 nm),

the electric field inside the particles is negligible. To avoid electron–

hole recombination, co-catalysts (here shown only for the reduction

reaction) are added to enhance the surface reaction rates.

Dow

nloa

ded

by U

nive

rsity

of

Vir

gini

a on

19

Aug

ust 2

012

Publ

ishe

d on

11

Mar

ch 2

008

on h

ttp://

pubs

.rsc

.org

| do

i:10.

1039

/B71

8969

A

View Online

for the WO3-based photo-anodes,56 which is close to the

theoretical efficiency that has been reported for this material.7

By using highly efficient dendritic a-Fe2O3 photo-anodes, overall

solar-to-hydrogen efficiencies of 2.2% are within reach.42

Although biasing with a separate solar cell is attractive from an

efficiency point of view, the fact that it involves two separate

devices, each with its own electrodes and electrolyte (in the case

of a DSSC), complicates the system and increases the cost. Several

efforts have been made to design and fabricate a monolithic

tandem cell, where multiple functional layers are combined into

a single plate. When this plate is immersed in a solution and illu-

minated, hydrogen should evolve from one side while oxygen is

formed on the other side. Turner et al. have successfully fabricated

a monolithic tandem cell based on a p-type GaInP2 photo-cathode

biased by a GaAs p–n junction solar cell.57 The solar-to-chemical

conversion efficiency is 12.4% (based on the Gibbs free energy

change of water splitting) under an illumination intensity of �11

suns. As pointed out by the authors, the use of concentrated

sunlight is necessary due to the high cost of the device. The main

problem, however, is that the lifetime of the tandem device is

measured in hours instead of years due to severe photo-corrosion.

Licht et al. addressed the photo-corrosion problem by using

efficient co-catalysts for H2 and O2 evolution (Pt and RuO2,

respectively).22,58 They used a complex multiple bandgap system

based on AlxGa1�xAs and Si p–n junctions to achieve a record

18.3% efficiency for water splitting. However, it should be noted

that this system is composed of a small area tandem cell coupled

to a large-area electrode for electrolysis. As this state-of-the-art

device is composed of about 10 layers of high-quality thin film

semiconductors, the main challenges for this approach are

scale-up and cost reduction.56 It is not a priori clear whether

such a system will be cost effective in a solar concentrator

system. Practical design issues and light scattering by evolving

gas bubbles limit the illumination intensity to an estimated

10–20 suns.57 For comparison, photo-voltaic concentrator

systems based on these materials usually operate at light intensi-

ties exceeding 100 suns.59

Miller et al. used a triple-junction amorphous silicon photo-

voltaic cell in combination with efficient hydrogen and oxygen

evolution catalysts.60 Amorphous silicon (a-Si) is much cheaper

than GaAs-based materials and requires less energy to process.

Sputter-deposited NiMo and Fe:NiOx were used as hydrogen

and oxygen evolution catalysts, respectively. These materials

are pure electrocatalysts, i.e., they are not photo-active. A

solar-to-hydrogen conversion efficiency of 7.8% was achieved

for a small-area prototype device (<1 cm2). Both catalysts

showed excellent resistance against (photo-)corrosion. It should

be noted that the photo-active area is not the same as the electro-

chemically active area, since the catalysts are not transparent for

visible light. To address this, one of the a-Si junctions is replaced

by Fe2O3 or WO3.61 So far, the a-Si/WO3 hybrid photo-elec-

trodes show solar-to-hydrogen conversion efficiencies of 1% in

outdoor tests.62 A particular challenge for these systems is to

deposit the metal oxides at sufficiently low temperatures

(<300 �C) in order to avoid degradation of the underlying a-Si

junctions. Aerosol spray pyrolysis requires a temperature of

�400 �C and cannot be used. With reactive sputtering the

substrate temperature could be kept below 200 �C at the cost

of being a more expensive technique.

This journal is ª The Royal Society of Chemistry 2008

Photo-catalyst particles in suspensions

From the previous sections it becomes clear that the search for

suitable photo-anodes for water splitting is severely hampered

by the many requirements that the materials have to fulfil. An

alternative approach is to simply disperse a photo-catalyst

powder in water. This eliminates the need for a conducting

substrate so that conventional high-temperature solid-state

synthesis routes can be used, allowing complex, single-phase

compounds to be made with relative ease. Co-catalysts (see

below) can be easily added in the desired quantities by simple

mixing and firing. Scale-up is readily accomplished and allows

accurate chemical analysis of the evolved gasses, even at very

low efficiencies. As such, these systems also provide a convenient

screening method for selecting suitable photo-anode materials.

For particulate systems, the reduction and oxidation reactions

both have to take place at the surface of a single particle, as illus-

trated in Fig. 8. To enhance the kinetics and to avoid chemical

recombination of H2 and O2, small amounts of co-catalysts,

such as Pt, NiOx or RuO2 are deposited onto the photo-catalyst’s

surface to ensure that H2 and O2 evolve at spatially separated

sites. Typical loadings are 1–3 g of catalyst powder per 1000

ml H2O, and co-catalyst loadings of 0.5–1% vs. the amount of

photo-catalyst. The aqueous slurry is contained in a closed pyrex

glass cell which is illuminated using fluorescent tubes (‘black-

light’), or high-power xenon or high-pressure mercury lamps.

By connecting the cell to a closed gas circulation system, the

amounts of evolved hydrogen and oxygen can be determined

on-line with gas chromatography.

In many cases, the photo-catalyst is only able to evolve either

hydrogen or oxygen gas, not both at the same time. If only

hydrogen (oxygen) evolves, a sacrificial electron donor

(acceptor) must be present to ensure the stoichiometric consump-

tion of electrons and holes. While sacrificial systems are of

considerable technical interest as they permit the study of

electrode reactions, they cannot be used as a commercial

hydrogen source due to the consumption of high-value (in terms

of energy and cost) compounds, such as methanol, EDTA,

formic acid, etc. Examples of oxygen evolution photo-catalysts

J. Mater. Chem., 2008, 18, 2311–2320 | 2317

Page 8: Solar hydrogen production with nanostructured metal oxides

Table 1 Overview of particle-based photo-catalyst systems reported in the literature that are able to split water in stoichiometric ratios

Photo-catalyst Co-catalyst Lamp, electrolytea Efficiency (solar-to-hydrogen) Stability Absorption Ref.

Single particle systems:M2Sb2O7 M]Ca, Sr RuO2 L1 — >4 h UV 72NaTaO3 NiO L2, 10�3 M NaOH 20% @ 270 nm >14 h UV 73BaIn0.5Nb0.5O3 NiOx L3 >40 h UV/VIS 74In0.9Ni0.1TaO4 NiOx L3 0.66% @ >420 nm >120 h UV/VIS 75(Ga1�xZnx)(N1�xOx) Cr–Rh oxide L2/L3, pH 4.5 (H2SO4) 2.5% @ 420 nm >35 h UV/VIS 71TiSi2 TiO2/SiO2 350 or 540 nm lamps, spectral

width �60 nm3.9% @ 540 nm >1000 h UV/VIS 76

Layered materials:Ba:La2TiO7 NiO L2, >12 mM NaOH 50% in UV — UV 77Sr2Ta2O7 NiO L3 12% @ 270 nm >20 h UV 78Composite particles:Zn:Lu2O3/Ga2O3 NiO L2 — >220 h UV 79Cr:Ba2In2O5:In2O3 NiOx L2 4.2% @320 nm >400 h UV 80Particle mixtures, Z-scheme:WO3/(Ca,Ta):SrTiO3 Pt/Pt L3, 100 mM NaI 0.1% @ 420 nm >250 h UV/VIS 70an-TiO2/ru-TiO2 Pt/Pt L2, 0.1 M NaI, pH 11 — >100 h UV 81

a L1 ¼ 200 W Hg-Xe; L2 ¼ 400–500 W high pressure Hg; L3 ¼ 300 W Xe.

Dow

nloa

ded

by U

nive

rsity

of

Vir

gini

a on

19

Aug

ust 2

012

Publ

ishe

d on

11

Mar

ch 2

008

on h

ttp://

pubs

.rsc

.org

| do

i:10.

1039

/B71

8969

A

View Online

are BiVO463,64 and WO3.65 AgNO3 or Ce4+(aq) are often used as

electron acceptors (or electron scavengers) in these systems.

Oxygen gas is sometimes used as an electron scavenger in other

photo-oxidation reactions. Hydrogen evolution photo-catalysts

appear to be more common, examples include InVO4,66

Bi2RNbO7 (R ¼ Y, rare earth),67 and BaCr2O4.68 Methanol is

almost always used as a sacrificial electron donor (hole

scavenger), but other hole scavengers such as EDTA and formic

acid may also work.

Table 1 presents an overview of recently reported suspension-

based photo-catalyst systems that are able to split water in a H2

: O2 ratio of 2 : 1 without the need for any sacrificial agents.

Co-catalysts are required in all cases and the reported efficiencies

are low, especially in the visible-light region. In addition to the

straightforward single component systems, several other systems

have been studied. In all cases, the main goals are to i) spatially

separate hydrogen from oxygen to prevent chemical recombina-

tion and ii) provide different surfaces with different catalytic

activities for hydrogen and oxygen evolution. In the layered

perovskite compounds (Table 1), water molecules are able to

diffuse spontaneously into the crystal structure of the material

where photo-oxidation takes place. The oxygen diffuses back to

the surface through the interlayers, while photo-reduction takes

place at the surface of the co-catalyst.69 In the composite particles,

an n–n heterojunction forms between two different particles. Due

to the different energetic positions of the conduction and valence

band edges, the electrons are transported to one particle, while the

holes move to the other particle. In the mixed-particle system, this

concept is taken a step further by performing the oxidation and

reduction of water on two different particles that are not in direct

contact with each other. To ensure a complete redox reaction at

each individual particle, an I�/IO3� redox mediator is used to

transport electrons from the oxygen evolving particles to their

hydrogen evolving counterparts.70

There are also some disadvantages of using powder photo-

catalysts. It is often difficult to determine the actual number of

photons that is absorbed by the suspension, which prohibits

accurate efficiency measurements. When reactivation of the

2318 | J. Mater. Chem., 2008, 18, 2311–2320

catalyst is necessary, the filtration process needed to separate

the catalyst from the water is not very convenient. Furthermore,

hydrogen and oxygen are produced in the same reactor volume,

forming an explosive mixture, and need to be separated immedi-

ately afterwards. This not only represents a safety issue, the

separation also costs energy and lowers the overall efficiency of

the process. Ritterskamp et al.76 recently published some

interesting work on TiSi2 photo-catalysts for which the mixing

of H2 and O2 is no longer a problem. Upon irradiation only

hydrogen evolves, while oxygen remains adsorbed at the TiSi2surface. It can be released by heating the catalyst to 100 �C,

allowing easy separation. A conversion efficiency of �4% has

been reported for this system, which does not take into account

the amount of energy needed to heat the suspension.

Finally, it should be realized that the efficiencies for powder

photo-catalysts are significantly lower than for photo-anodes

in a photo-electrochemical cell. A very good powder photo-cata-

lyst for visible light water splitting generates perhaps 200 mmol

H2 per gram of catalyst per hour,71 although much smaller values

are also considered promising.66 In contrast, a 1 mm thick Fe2O3

film that shows a photo-current of 2 mA cm�2 (ref. 42) produces

hydrogen at a rate of �70 mmol g�1 h�1, i.e. 350 times more.

Conclusions

The operating principles of photo-electrochemical cells for solar

water splitting have been described. The severe requirement of

stability against (photo-)corrosion in an aqueous environment

limits the choice of the photo-anode material to metal oxides.

TiO2 photo-anodes show excellent chemical stability and charge

transport properties, but the large bandgap limits the efficiency.

Doping may enhance the optical absorption, but not the photo-

electrochemical performance since it appears impossible to avoid

the presence of deep defects that act as recombination centers.

a-Fe2O3 has a nearly ideal bandgap, absorbs visible light and

shows good chemical stability under the right conditions. Its

performance can be enhanced by moderate use of (shallow)

dopants to improve the electronic conductivity, and by

This journal is ª The Royal Society of Chemistry 2008

Page 9: Solar hydrogen production with nanostructured metal oxides

Dow

nloa

ded

by U

nive

rsity

of

Vir

gini

a on

19

Aug

ust 2

012

Publ

ishe

d on

11

Mar

ch 2

008

on h

ttp://

pubs

.rsc

.org

| do

i:10.

1039

/B71

8969

A

View Online

introducing a thin interfacial layer that presumably improves the

alignment of the band edges. High aspect-ratio morphologies,

such as nanowires, can be used to overcome the poor hole trans-

port properties and the slow surface reaction kinetics. A growing

interest in complex oxides is expected to lead to a much broader

choice in photo-active materials, and current research efforts on

powder photo-catalysts offer a convenient route towards the

selection of suitable photo-anode materials. Further improve-

ments in the field of solar hydrogen production can be expected

as new, low-cost and low-temperature techniques for the

controlled synthesis of metal oxides are developed.

Acknowledgements

The authors gratefully acknowledge the contributions of the

following postdocs, students, and other co-workers to this

work: Dr J.C. van ’t Spijker, E.L. Maloney (Everest Coatings),

C.S. Enache, and T.J. Olthof. Financial support for this work

was provided by the Delft Institute for Sustainable Energy,

SenterNovem, and the NWO/ACTS Sustainable Hydrogen

program (project nr. 053.61.009). RvdK also acknowledges the

Netherlands Organization for Scientific Research (NWO) for

a VENI grant which made part of these investigations possible.

References

1 N. S. Lewis and D. G. Nocera, Proc. Natl. Acad. Sci. U. S. A., 2006,103, 15729.

2 Basic research needs for solar energy utilization (report), USDepartment of Energy, Office of Basic Energy Sciences, 2005.

3 O. Khaselev, A. Bansal and J. A. Turner, Int. J. Hydrogen Energy,2001, 26, 127.

4 R. Memming, Semiconductor Electrochemistry, Wiley, Weinheim,2000.

5 M. F. Weber and M. J. Dignam, Int. J. Hydrogen Energy, 1986, 11,225.

6 J. R. Bolton, S. J. Strickler and J. S. Connolly, Nature, 1985, 316, 495.7 A. B. Murphy, P. R. F. Barnes, L. K. Randeniya, I. C. Plumb,

I. E. Grey, M. D. Horne and J. A. Glasscock, Int. J. HydrogenEnergy, 2006, 31, 1999.

8 R. Shinar and J. H. Kennedy, J. Electrochem. Soc., 1983, 130, 860.9 R. Shinar and J. H. Kennedy, Sol. Energy Mater., 1982, 6, 323.

10 S. U. M. Khan and J. Akikusa, J. Phys. Chem. B, 1999, 103, 7184.11 D. E. Scaife, Sol. Energy, 1980, 25, 41.12 J. M. Bolts and M. S. Wrighton, J. Phys. Chem., 1976, 80, 2641.13 R. Van de Krol, A. Goossens and J. Schoonman, J. Electrochem.

Soc., 1997, 144, 1723.14 P. E. de Jongh, D. Vanmaekelbergh and J. J. Kelly, J. Electrochem.

Soc., 2000, 147, 486.15 M. F. Finlayson, B. L. Wheeler, N. Kakuta, K. H. Park, A. J. Bard,

A. Campion, M. A. Fox, S. E. Webber and J. M. White, J. Phys.Chem., 1985, 89, 5676.

16 M. Nanu, J. Schoonman and A. Goossens, Adv. Mater., 2004, 16,453.

17 S. R. Morrison, Electrochemistry of Semiconductor and OxidizedMetal Electrodes, Plenum Press, New York, 1980.

18 A. Fujishima and K. Honda, Nature, 1972, 238, 37.19 J. G. Mavroides, J. A. Kafalas and D. F. Kolesar, Appl. Phys. Lett.,

1976, 28, 241.20 S. M. Sze, Physics of Semiconductor Devices, John Wiley & Sons, New

York, 1981.21 R. Hoffmann, Solids and surfaces – a chemist’s view of bonding in

extended structures, Wiley-VCH, New York, 1988.22 S. Licht, B. Wang, S. Mukerji, T. Soga, M. Umeno and H. Tributsch,

J. Phys. Chem. B, 2000, 104, 8920.23 H. Kato and A. Kudo, J. Phys. Chem. B, 2002, 106, 5029.24 H. P. Maruska and A. K. Ghosh, Sol. Energy Mater., 1979, 1, 237.

This journal is ª The Royal Society of Chemistry 2008

25 A. Goossens, E. L. Maloney and J. Schoonman, Chem. Vap.Deposition, 1998, 4, 109.

26 J. C. van’t Spijker, E. L. Maloney, J. Schoonman and R. Van de Krol,Unusually high photocurrents in anatase TiO2 thin films with fractalmorphologies, presented at the 16th International Conference onPhotochemical Conversion and Storage of Solar Energy (IPS-16),Uppsala, Sweden, 2006.

27 R. Asahi, T. Morikawa, T. Ohwaki, K. Aoki and Y. Taga, Science,2001, 293, 269.

28 S. Sakthivel and H. Kisch, Angew. Chem., Int. Ed., 2003, 42, 4908.29 S. Sakthivel and H. Kisch, ChemPhysChem, 2003, 4, 487.30 X. B. Chen, Y. B. Lou, A. C. S. Samia, C. Burda and J. L. Gole, Adv.

Funct. Mater., 2005, 15, 41.31 C. S. Enache, J. Schoonman and R. Van de Krol, J. Electroceram.,

2004, 13, 177.32 C. S. Enache, J. Schoonman and R. Van de Krol, Appl. Surf. Sci.,

2006, 252, 6342.33 H. Irie, Y. Watanabe and K. Hashimoto, Chem. Lett., 2003, 32, 772.34 T. Lindgren, J. M. Mwabora, E. Avendano, J. Jonsson, A. Hoel,

C. G. Granqvist and S. E. Lindquist, J. Phys. Chem. B, 2003, 107,5709.

35 T. Lindgren, J. Lu, A. Hoel, C. G. Granqvist, G. R. Torres andS. E. Lindquist, Sol. Energy Mater. Sol. Cells, 2004, 84, 145.

36 C. Sanchez, M. Hendewerk, K. D. Sieber and G. A. Somorjai, J. SolidState Chem., 1986, 61, 47.

37 J. H. Kennedy, M. Anderman and R. Shinar, J. Electrochem. Soc.,1981, 128, 2371.

38 J. H. Kennedy and K. W. Frese, Jr., J. Electrochem. Soc., 1978, 125,709.

39 C. J. Sartoretti, B. D. Alexander, R. Solarska, W. A. Rutkowska,J. Augustynski and R. Cerny, J. Phys. Chem. B, 2005, 109, 13685.

40 J. A. Glasscock, P. R. F. Barnes, I. C. Plumb and N. Savvides,J. Phys. Chem. C, 2007, 111, 16477.

41 I. Cesar, A. Kay, J. A. G. Martinez and M. Gratzel, J. Am. Chem.Soc., 2006, 128, 4582.

42 A. Kay, I. Cesar and M. Gratzel, J. Am. Chem. Soc., 2006, 128,15714.

43 Y. Liang, C. S. Enache and R. Van de Krol, Int. J. Photoenergy, 2008,in press.

44 M. P. Dare Edwards, J. B. Goodenough, A. Hamnett andP. R. Trevellick, J. Chem. Soc., Faraday Trans. 1, 1983, 79, 2027.

45 U. Bjorksten, J. Moser and M. Gratzel, Chem. Mater., 1994, 6, 858.46 N. Beermann, L. Vayssieres, S. E. Lindquist and A. Hagfeldt,

J. Electrochem. Soc., 2000, 147, 2456.47 L. Vayssieres, C. Sathe, S. M. Butorin, D. K. Shuh, J. Nordgren and

J. H. Guo, Adv. Mater., 2005, 17, 2320.48 L. Vayssieres, N. Beermann, S. E. Lindquist and A. Hagfeldt,

Chem. Mater., 2001, 13, 233.49 Y. Y. Fu, R. M. Wang, J. Xu, J. Chen, Y. Yan, A. Narlikar and

H. Zhang, Chem. Phys. Lett., 2003, 379, 373.50 T. Yu, Y. W. Zhu, X. J. Xu, K. S. Yeong, Z. X. Shen, P. Chen,

C. T. Lim, J. T. L. Thong and C. H. Sow, Small, 2006, 2, 80.51 C. H. Kim, H. J. Chun, D. S. Kim, S. Y. Kim, J. Park, J. Y. Moon,

G. Lee, J. Yoon, Y. Jo, M. H. Jung, S. I. Jung and C. J. Lee, Appl.Phys. Lett., 2006, 89.

52 R. M. Wang, Y. F. Chen, Y. Y. Fu, H. Zhang and C. Kisielowski,J. Phys. Chem. B, 2005, 109, 12245.

53 X. G. Wen, S. H. Wang, Y. Ding, Z. L. Wang and S. H. Yang,J. Phys. Chem. B, 2005, 109, 215.

54 Y. L. Chueh, M. W. Lai, J. Q. Liang, L. J. Chou and Z. L. Wang,Adv. Funct. Mater., 2006, 16, 2243.

55 T. J. Olthof, BSc thesis, Delft University of Technology, 2007.56 M. Gratzel, Nature, 2001, 414, 338.57 O. Khaselev and J. A. Turner, Science, 1998, 280, 425.58 S. Licht, J. Phys. Chem. B, 2001, 105, 6281.59 C. Algora, E. Ortiz, I. Rey-Stolle, V. Diaz, R. Pena, V. M. Andreev,

V. P. Khvostikov and V. D. Rumyantsev, IEEE Trans. ElectronDevices, 2001, 48, 840.

60 R. E. Rocheleau, E. L. Miller and A. Misra, Energy Fuels, 1998, 12, 3.61 E. L. Miller, R. E. Rocheleau and X. M. Deng, Int. J. Hydrogen

Energy, 2003, 28, 615.62 E. L. Miller, D. Paluselli, B. Marsen and R. E. Rocheleau, Sol.

Energy Mater. Sol. Cells, 2005, 88, 131.63 A. Kudo, K. Omori and H. Kato, J. Am. Chem. Soc., 1999, 121,

11459.

J. Mater. Chem., 2008, 18, 2311–2320 | 2319

Page 10: Solar hydrogen production with nanostructured metal oxides

Dow

nloa

ded

by U

nive

rsity

of

Vir

gini

a on

19

Aug

ust 2

012

Publ

ishe

d on

11

Mar

ch 2

008

on h

ttp://

pubs

.rsc

.org

| do

i:10.

1039

/B71

8969

A

View Online

64 K. Sayama, A. Nomura, T. Arai, T. Sugita, R. Abe, M. Yanagida,T. Oi, Y. Iwasaki, Y. Abe and H. Sugihara, J. Phys. Chem. B,2006, 110, 11352.

65 G. R. Bamwenda and H. Arakawa, Sol. Energy Mater. Sol. Cells,2001, 70, 1.

66 J. H. Ye, Z. G. Zou, M. Oshikiri, A. Matsushita, M. Shimoda,M. Imai and T. Shishido, Chem. Phys. Lett., 2002, 356, 221.

67 Z. G. Zou, J. H. Ye and H. Arakawa, Top. Catal., 2003, 22, 107.68 D. Wang, Z. Zhigang and Y. Jinhua,Chem. Phys. Lett., 2003, 373, 191.69 T. Takata, Y. Furumi, K. Shinohara, A. Tanaka, M. Hara,

J. N. Kondo and K. Domen, Chem. Mater., 1997, 9, 1063.70 K. Sayama, K. Mukasa, R. Abe, Y. Abe and H. Arakawa,

Chem. Commun., 2001, 23, 2416.71 K. Maeda, K. Teramura, D. L. Lu, T. Takata, N. Saito, Y. Inoue and

K. Domen, Nature, 2006, 440, 295.72 J. Sato, N. Saito, H. Nishiyama and Y. Inoue, J. Photochem.

Photobiol., A, 2002, 148, 85.

2320 | J. Mater. Chem., 2008, 18, 2311–2320

73 H. Kato and A. Kudo, J. Phys. Chem. B, 2001, 105, 4285.74 J. Yin, Z. G. Zou and J. H. Ye, J. Phys. Chem. B, 2003, 107, 61.75 Z. G. Zou, J. H. Ye, K. Sayama and H. Arakawa, Nature, 2001, 414,

625.76 P. Ritterskamp, A. Kuklya, M. A. Wustkamp, K. Kerpen,

C. Weidenthaler and M. Demuth, Angew. Chem., Int. Ed., 2007, 46,7770.

77 J. Kim, D. W. Hwang, H. G. Kim, S. W. Bae, J. S. Lee, W. Li andS. H. Oh, Top. Catal., 2005, 35, 295.

78 A. Kudo, H. Kato and S. Nakagawa, J. Phys. Chem. B, 2000, 104,571.

79 D. F. Wang, Z. G. Zou and J. H. Ye, Chem. Phys. Lett., 2004, 384,139.

80 D. F. Wang, Z. G. Zou and J. H. Ye, Chem. Mater., 2005, 17,3255.

81 R. Abe, K. Sayama, K. Domen and H. Arakawa, Chem. Phys. Lett.,2001, 344, 339.

This journal is ª The Royal Society of Chemistry 2008