Sunteți pe pagina 1din 6

ARTICLES

PUBLISHED ONLINE: 8 MAY 2011 | DOI: 10.1038/NMAT3017

Highly active oxide photocathode for


photoelectrochemical water reduction
Adriana Paracchino1 , Vincent Laporte2 , Kevin Sivula1 , Michael Grtzel1 and Elijah Thimsen1 *
A clean and efficient way to overcome the limited supply of fossil fuels and the greenhouse effect is the production of hydrogen
fuel from sunlight and water through the semiconductor/water junction of a photoelectrochemical cell, where energy collection
and water electrolysis are combined into a single semiconductor electrode. We present a highly active photocathode for solar
H2 production, consisting of electrodeposited cuprous oxide, which was protected against photocathodic decomposition in
water by nanolayers of Al-doped zinc oxide and titanium oxide and activated for hydrogen evolution with electrodeposited
Pt nanoparticles. The roles of the different surface protection components were investigated, and in the best case electrodes
showed photocurrents of up to 7.6 mA cm2 at a potential of 0 V versus the reversible hydrogen electrode at mild pH. The
electrodes remained active after 1 h of testing, cuprous oxide was found to be stable during the water reduction reaction and
the Faradaic efficiency was estimated to be close to 100%.

he production of chemical fuels by solar energy conversion is


an attractive and sustainable solution to the energy problem.
Photoelectrochemical (PEC) splitting of water into hydrogen
and oxygen by the direct use of sunlight is an ideal, renewable
method of hydrogen production that integrates solar energy
collection and water electrolysis into a single photoelectrode. A
PEC cell is based on a semiconductor/liquid junction, where the
minority charges (electrons and holes for a p-type and an n-type
semiconductor, respectively) generated on light absorption in the
semiconductor are driven into the solution by the electric field
at the junction, where they can drive a redox reaction, such as
the reduction of H+ to H2 for a p-type semiconductor. Owing
to the formidable geographical areas that would be required for
widespread solar energy harvesting, the semiconductor device must
be low-cost and fabricated from abundant elements using scalable
processing techniques, which precludes the use of previously
reported, efficient solar water splitting systems14 . Binary copper
semiconductors, such as copper oxides, are attractive because
they are photoactive, copper is abundant and the materials
can be processed by industrially proven, low-cost methods such
as electrodeposition.
Cuprous oxide, Cu2 O, is an attractive p-type oxide for photoelectrochemical hydrogen production with a direct bandgap of 2 eV
and a corresponding theoretical photocurrent of 14.7 mA cm2
and a light-to-hydrogen conversion efficiency of 18% based on
the AM 1.5 spectrum and the lower heating value of hydrogen. In
recent years it has been explored as a photocatalyst for solar-driven
water splitting and H2 generation510 , as an electrode for lithium
ion batteries1113 and as a p-type semiconductor in a heterojunction
with n-type ZnO for photovoltaic applications1417 .
For solar water splitting, Cu2 O possesses favourable energy
band positions, with the conduction band lying 0.7 V negative
of the hydrogen evolution potential and the valence band lying
just positive of the oxygen evolution potential6 . As there is no
overpotential available for oxygen evolution, the photocathode can
drive half of the water splitting reaction and an external bias (which

can be applied by a dye-sensitized solar cell in tandem18 ) is required


for the other half (water oxidation). The main limiting factor for
the use of Cu2 O as a photocathode for water reduction is its poor
stability in aqueous solutions, because the redox potentials for the
reduction and oxidation of monovalent copper oxide lie within
the bandgap (Fig. 1). As the minority-carrier diffusion length in
electrodeposited Cu2 O (20100 nm; refs 6,19) is incommensurate
with the light absorption depth near the bandgap (approximately
10 m; refs 20,21), aspect ratios of 50250 would be necessary to
collect the theoretical maximum photocurrent. This requires the
use of a conformal coating to stabilize the surface, and although
one attempt has been made to stabilize it with a robust passivating
agent7 , no strategies have yet been developed that can be applied to
complex film morphologies while also ensuring that the surface is
catalytically active towards water reduction.
Here, the issue of Cu2 O instability in water under illumination is
addressed by depositing protective layers on the electrode surface by
atomic layer deposition22 (ALD), which is a scalable layer-by-layer
synthesis process that can be readily applied to deposit ultrathin
films on nanostructured surfaces at low temperatures with strict
control over thickness and composition at the angstrom length
scale. We note that ALD has had success in preventing anodic
corrosion in other nanostructured systems23 . In this work, the
effect of ultrathin protective n-type oxides, whose conduction band
levels lie between the Cu2 O conduction band and the hydrogen
evolution potential, on the magnitude and stability of the water
splitting photocurrent is presented. In addition to protecting
semiconducting materials in PEC devices, the multiple layer ALD
protection strategy presented here is expected to offer improved
corrosion protection for industrially relevant metals such as steel24
and brass25 and specialized materials employed in optics26 and
clinical27 applications.
The Cu2 O films used for this study were synthesized by
electrodeposition and had a constant thickness of 1.3 m.
Individual Cu2 O grains of the film were about 1 m in size with
a cubic morphology, and had a predominant (111) orientation

1 Institute

of Chemical Sciences and Engineering, Ecole Polytechnique Fdrale de Lausanne, Laboratory of Photonics and Interfaces, Station 6, CH-1015
Lausanne, Switzerland, 2 Interdisciplinary Centre for Electron Microscopy, Ecole Polytechnique Fdrale de Lausanne, Station 12, CH-1015 Lausanne,
Switzerland. Present address: Argonne National Lab, Material Science Division, Argonne, Illinois 60439, USA. *e-mail: ethimsen@anl.gov.
456

NATURE MATERIALS | VOL 10 | JUNE 2011 | www.nature.com/naturematerials

2011 Macmillan Publishers Limited. All rights reserved

NATURE MATERIALS DOI: 10.1038/NMAT3017

ARTICLES
a

E/V versus NHE

CB
CB
+0.00

2H+ + 2e <> H2

+0.47
+0.60

Cu2O + H2O + 2e <> 2Cu + 2OH


2CuO + H2O + 2e <> 2Cu2O+ 2OH
<> 2H2O
VB

2
0

4
J (mA cm2)

O2 +

4e

J (mA cm2)

+1.23

4H+ +

CB

5
6
VB
Cu2O

VB

ZnO

2
4
6
8

10

15

20

Time (min)

TiO2

Figure 1 | Overview of the energy band positions for the semiconductors


of the multilayer photocathode and redox levels of the involved chemical
reactions. CB, conduction band; VB, valence band.

0.25

0.30

0.35
0.40
0.45
E (V) versus RHE

0.50

0.55

b
0

(1)

The bare electrodes were not scanned further than 0.25 V


versus RHE because the photocurrent had already reached a
plateau. The inset in Fig. 2a shows the photocurrent stability
measurement on the Cu2 O electrode held at 0 V versus RHE.
Reductive decomposition of the electrode was visually observed
by the formation of a black circle where the electrode was
illuminated and also by scanning electron microscopy (SEM),
which revealed Cu metal nanoparticles on the surface of the Cu2 O
grains (Supplementary Fig. S1). The presence of the Cu metal
nanoparticles on the surface of the Cu2 O grains was verified by
X-ray photoelectron spectroscopy (XPS) and X-ray diffraction
(XRD), as described in the Supplementary Information. On the
basis of the SEM images and XPS measurements, the electrode
degradation reaction (1) occurred at the Cu2 O/electrolyte interface.
To suppress this reaction, we next sought to passivate the electrode
surface with a protective oxide layer to prevent contact with the
electrolyte while maintaining the band bending in the Cu2 O.
The protective oxide layer must meet several criteria. First, it
must have a staggered type-II band offset, so that photogenerated
electrons can flow from the Cu2 O through the protective layer to
the electrolyte to promote water reduction, whereas holes flow into
the Cu2 O bulk. Second, the n-type oxide should have a conduction
band above the water reduction potential while having no reductive
degradation reactions at potentials in the bandgap. Third, the
surface in contact with the electrolyte must have favourable reaction
kinetics towards the reduction reaction of interest, in this case
water reduction for H2 evolution. If a hydrogen evolution catalyst
such as Pt nanoparticles is used, TiO2 can meet these criteria.

2
3
2

J (mA cm2)

Cu2 O + 2e + 2H+ Cu + H2 O

J (mA cm2)

with exposed {100} facets (Supplementary Fig. S1). Under AM 1.5


illumination, the as-deposited, bare Cu2 O produced a photocurrent
of 2.4 mA cm2 at 0.25 V versus the reversible hydrogen electrode
(RHE) in a nitrogen-purged 1 M Na2 SO4 electrolyte buffered at a
pH of 4.9, which was the electrolyte used for all PEC experiments in
this work. The photocurrent was measured under chopped AM 1.5
illumination (100 mW cm2 ), so the dark and light current could
be monitored simultaneously. The corresponding currentvoltage
(J V ) plot of a bare Cu2 O electrode (sample 1) is presented in
Fig. 2a. A significant fraction of the photocurrent is due to the
reductive decomposition of the material according to reaction (1).

2
4
6

7
8

0.1

10
Time (min)

0.2
0.3
E (V) versus RHE

0.4

15

20

0.5

Figure 2 | The photoelectrochemical response for the bare and


surface-protected electrode. a,b, Currentpotential characteristics in 1 M
Na2 SO4 solution, under chopped AM 1.5 light illumination for the bare
Cu2 O electrode (a) and for the as-deposited Cu2 O/5 (4 nm
ZnO/0.17 nmAl2 O3 )/11 nm TiO2 /Pt electrode (b). E is the electrochemical
potential of the electrode. The insets show the respective photocurrent
transients for the electrodes held at 0 V versus RHE in chopped light
illumination with N2 purging.

However, as described below, TiO2 alone did not stabilize the


electrodes for these ALD films, and a ZnO buffer layer was
required to afford chemical stability. A schematic representation
of the electrode structure and an SEM image of the Cu2 O grains
covered with a 5 (4 nm ZnO/0.17 nm Al2 O3 )/11 nm TiO2 ALD
protective layer and Pt nanoparticles is presented in Fig. 3 and
the energy band positions are shown in Fig. 1. The notation 5
(4 nm ZnO/0.17 nm Al2 O3 )/11 nm TiO2 represents five bilayers of
4 nm of ZnO and 0.17 nm of Al2 O3 followed by 11 nm of TiO2 .
Table 1 summarizes the photoelectrochemical response afforded
by different protective layers. In Table 1, the photocurrent maximum refers to the value at 0 V versus RHE during the linear sweep
voltammetry under chopped illumination (J V plot). The stability

NATURE MATERIALS | VOL 10 | JUNE 2011 | www.nature.com/naturematerials

2011 Macmillan Publishers Limited. All rights reserved

457

NATURE MATERIALS DOI: 10.1038/NMAT3017

ARTICLES
a

TiO2

ZnO:Al

Pt

Cu2O
FTO

Au
500 nm

Figure 3 | The surface-protected Cu2 O electrode. a, Schematic


representation of the electrode structure. b, Scanning electron micrograph
showing a top view of the electrode after ALD of 5 (4 nm ZnO/
0.17 nm Al2 O3 )/11 nm TiO2 followed by electrodeposition of
Pt nanoparticles.

of the photocurrent was evaluated with chronoamperometric


measurements at 0 V versus RHE in chopped light over 20 min and
was quantified as the percentage of the photocurrent at the end on
the last light cycle (J ) compared with the value at the end of the first
light cycle (J0 ). All electrodes had Pt nanoparticles on the surface.
With no catalyst, the measured photocurrent was only several
microamperes per square centimetre, because H2 evolution does
not occur readily at the surface of uncatalysed semiconductors28 .
The 20 nm ZnO buffer layer between the Cu2 O and 10 nm TiO2
was necessary to obtain stable electrodes. This is clearly seen by
comparing J /J0 for samples 2 and 3, which are 0% and 14% for
the TiO2 alone, and the ZnO/TiO2 case, respectively. In the absence
of the ZnO layer, even increasing the TiO2 thickness to 30 nm did
not have a significant effect on the stability, and still resulted in a
decay of the photocurrent to 0% of its initial value after 20 min (data
not shown here). This behaviour indicates incomplete coverage
and photoreduction of the Cu2 O at pinholes in the TiO2 layer.
TiO2 grown by ALD from titanium tetraisopropoxide and H2 O is
known to have issues with growing non-uniformly29 , and the ZnO
buffer probably provides a more uniform hydroxylated surface for
TiO2 to grow on, while also forming a rectifying junction with
Cu2 O. A local electrostatic field formed at the Cu2 O/ZnO pn
junction could also assist in extracting photogenerated electrons
from the Cu2 O. Inserting monolayers of Al2 O3 periodically into the
ZnO was necessary to further improve the stability of the electrode
without changing the photocurrent, which indicates that the band
bending in Cu2 O was unaffected. By carefully analysing samples 37
(Table 1), it can be seen that aluminium doping increases J /J0 if
inserted in the ZnO layer or at the ZnO/TiO2 interface, but not
if inserted at the Cu2 O/ZnO interface. From this observation we
conclude that aluminium doping stabilizes the ZnO layer, although
the mechanism is unclear (Supplementary Information). Increasing
the aluminium content too much (sample 5), however, had a
deleterious effect on the initial photocurrent, presumably owing
to reduced electron mobility (Supplementary Fig. S3a) or tunnel

barriers in the protective layer resulting from the high Al2 O3 content
of the structure (expected 6.6 at.% for Al in sample 5).
It is known that Al3+ doping in ZnO can increase the electron
concentration30,31 , but we do not believe this is the origin of
the phenomenon observed here. The carrier concentration in the
ZnO layer did increase with Al2 O3 incorporation (Supplementary
Fig. S3c), but the initial photocurrents of sample 3 and sample
4 were approximately the same. Furthermore, by comparing
the measured carrier concentration to the expected Al3+ atom
concentration (Supplementary Information), we estimate that 90%
of the Al3+ atoms were not donating electrons to the conduction
band, and must therefore be serving a different role.
When the electrodes with the protective layers were subjected to
a mild heat treatment at 200 C in air for 45 min before platinum
deposition, the stability was further improved and accompanied
by a drop in the initial photocurrent (samples 8 and 9). The
effect of aluminium addition to the ZnO remained after annealing,
and the electrode with 0.17 nm Al2 O3 layers spaced every 4 nm
exhibited an increased stability of 78% after 20 min (sample 9),
compared with 50% for the ZnO-alone case (sample 8). The
observed decrease in initial photocurrent can be explained by the
increased resistivity for both ZnO and Al:ZnO (Supplementary
Fig. S3b) compared with the corresponding as-deposited films,
especially for the undoped ZnO/TiO2 film, which was too resistive
for the Hall coefficient to be measured by our instrument. The
increased stability could have resulted from removal of residual
water from the ALD layers at 200 C, or could have derived from
restructuring of the layers at the interface. It is worth noting that to
the knowledge of the authors, the initial photocurrents of sample 9
and the as-deposited electrode with the same configuration
(sample 4) exhibited higher photocurrents, 5.7 mA cm2 and
7.6 mA cm2 respectively, than any oxide-based photoelectrode
previously published under AM 1.5 illumination. The incident
photon-to-current efficiency for an annealed electrode held at
0 V versus RHE is presented in Supplementary Fig. S6 with the
corresponding J V plot. The electrode showed a photoresponse
from 350 to 600 nm, and the incident photon-to-current efficiency
was 40% between 350 and 480 nm.
The production of H2 gas was confirmed by the formation
of bubbles evolving off the photocathode immediately after
illumination and during the stability measurement (Supplementary
Fig. S5). After passing 0.58 C through the photocathode under
continuous illumination at 0 V versus RHE, an amount of
75 10 l of H2 gas was detected by volume displacement
measurements, which was consistent with the expected amount of
67 l, corresponding to a Faradaic efficiency of close to 100%.
The evidence is consistent with chemical stabilization of
Cu2 O and indicates that the photocurrent decay was not due
to the degradation of the photoactive material. We carried out

Table 1 | Summary of PEC performance for the different ALD layers investigated.
Sample ID

ALD protective layer

Cu2 O
thickness
(m)

ALD
post-treatment

J at 0 V
versus RHE
(mA cm2 )

J/J0 after
20 min at 0 V
versus RHE (%)

1
2
3
4
5
6
7
8
9

Bare
11 nm TiO2
21 nm ZnO/11 nm TiO2
5 (4 nm ZnO/0.17 nm Al2 O3 )/11 nm TiO2
10 (2 nm ZnO/0.17 nm Al2 O3 )/11 nm TiO2
21 nm ZnO/0.9 nm Al2 O3 /11 nm TiO2
0.9 nm Al2 O3 /21 nm ZnO/11 nm TiO2
21 nm ZnO/11 nm TiO2
5 (4 nm ZnO/0.17 nm Al2 O3 )/11 nm TiO2

1.3
1.3
1.3
1.3
1.3
1.3
1.3
1.3
1.3

As-deposited
As-deposited
As-deposited
As-deposited
As-deposited
As-deposited
As-deposited
200 C, 45 min, air
200 C, 45 min, air

2.4 (0.25 V versus RHE)


1.8
7.8
7.6
4.7
2.7
2.4
5.2
5.7

0
0
14
33
53
52
17
50
78

458

NATURE MATERIALS | VOL 10 | JUNE 2011 | www.nature.com/naturematerials

2011 Macmillan Publishers Limited. All rights reserved

NATURE MATERIALS DOI: 10.1038/NMAT3017


a

Dark scan 2
Dark scan 3
Light scan 2
Light scan 3

Dark scan 2
Dark scan 3
Light scan 2
Light scan 3

J (mA cm2)

J (mA cm2)

ARTICLES

1
0

0.2

0.4
0.6
0.8
E (V) versus RHE

1.0

1.2

0.2

0.4

0.6

0.8

1.0

1.2

E (V) versus RHE

Figure 4 | Cyclic voltammetry of the bare and surface-protected Cu2 O electrode. a,b, Cyclic voltammetry on an as-deposited bare Cu2 O electrode (a)
and on a Cu2 O/5 (4 nm ZnO/0.17 nm Al2 O3 )/11 nm TiO2 /Pt electrode after PEC characterization (b). The absence of any features other than
photoreduction of water in the cyclic voltammogram of the protected electrode shows its chemical stability.

cyclic voltammetry in the dark and in the light on bare Cu2 O,


and on sample 9 after stability testing. Figure 4 shows the
cyclic voltammetry results. Whereas the bare electrode exhibited
prominent features from both photooxidation (peaks near 0.9 V
versus RHE) and photoreduction (vertical offset to 0.5 mA cm2 ),
the protected electrode showed only one feature corresponding to
photoreduction of water and no corrosion peaks. The chemical
stabilization of Cu2 O was also confirmed by the position of
the Auger Cu LMM signal obtained after XPS depth profiling
(Supplementary Fig. S8). The Cu LMM peak for an electrode
measured for 80 min under irradiation at 0 V versus RHE
(Supplementary Fig. S8a) showed the same position as for a bare
Cu2 O that was not exposed to a PEC measurement (Supplementary
Fig. S8b). Moreover, no Cu 2p signal was detected through the ALD
layers of a PEC-measured electrode of the configuration of sample 9
(Supplementary Fig. S7). XPS data were collected after cycles of Ar+
sputtering and only Pt, TiO2 and ZnO were all visible and appeared
in the expected order during the depth profile analysis (Fig. 5).
This indicates ALD film continuity and no Cu+ migration in the
biased electrode, meaning no Cu2 O exposure to the electrolyte.
The Zn signal appeared after 150 s of sputtering, corresponding
to approximately 10 nm below the electrode surface, as expected
(Fig. 5). This indicates that the TiO2 layer was continuous and
the ZnO was not exposed to the electrolyte. The Pt and Ti signals
decreased with sputtering time but did not disappear, presumably
because the rough surface had grains (1 m) much smaller than
the analysis area (1 mm2 spot). Interestingly, very little Al was
detected. The Al should have been readily detectable given the
predicted concentration of 3.4% based on the cycle numbers and
growth rate per cycle (GPC) values according to the rule of
mixtures30 . One possible explanation for the lack of Al signal could
be its very localized distribution, in the form of dopant layers rather
than homogeneous dopant dispersion in the ZnO matrix32 .
The XRD spectrum of the as-deposited bare Cu2 O was identical
to the protected electrode after PEC stability measurement, as
opposed to the bare Cu2 O electrode, which showed reflections
from Cu and CuO after PEC stability measurement (Supplementary
Fig. S4). Moreover, the charge passed through the photocathode
during 20 min of stability testing was more than enough to reduce
all of the Cu2 O to Cu (1.8 C, 1.4 turnovers by reaction (1))
and the protected electrodes were measured several times without
showing any colour change (Supplementary Fig. S5), which would

be expected if the Cu2 O were reduced to Cu (Supplementary


Fig. S1). We do note that although the evidence presented is
consistent with chemical stabilization of Cu2 O, it does not rule out
the possibility of side reactions that could degrade the electrode over
timescales longer than those investigated in this work.
Despite all of the evidence for chemical stability of the Cu2 O,
the photocurrent still decreased with illumination time for the
protected electrodes. The reason for the observed photocurrent
decay without structural failure of the protecting layers or apparent
chemical degradation of the Cu2 O was probably due to the presence
of Ti3+ traps in the TiO2 layer, as observed elsewhere in the case of
water splitting33,34 or under ultraviolet illumination in the absence
of air3537 . Photoelectrons generated in Cu2 O probably flowed
without hindrance into Al:ZnO and to TiO2 , but as the Fermi level
of TiO2 in the dark is very close to the water reduction potential,
they were not readily injected into the electrolyte and accumulated
in the protective layer as long-lived Ti3+ states. The formation of
these traps was probably exacerbated by the amorphous structure
of the ALD TiO2 used in this study. The Pourbaix diagram
for TiO2 in our electrolyte is shown in Supplementary Fig. S10
and indicates that the photoelectrons flowing from Cu2 O to the
protective layer could have reduced TiO2 to Ti(OH)3 as a parasitic
side reaction. Evidence of Ti3+ formation is given in the XPS for
Ti 2p of Supplementary Fig. S11, where the shoulder at 457 eV is
characteristic for Ti3+ in the TiO2 layer. This shoulder is also present
in the depth profile XPS of an unmeasured sample (Supplementary
Fig. S11a) because of preferential oxygen sputtering by the Ar+
beam, but the area of the Ti3+ peak relative to the Ti4+ peak
is larger in the measured sample (Supplementary Fig. S11b),
indicating that the TiO2 is reduced during the PEC measurement.
In Supplementary Fig. S9 a chronoamperometric experiment is
shown, where the surface of the electrode was oxidized after 20 min
of stability measurement by air-purging and then measured again
for 20 min after removal of the residual dissolved oxygen. The
initial photocurrent was restored after the air treatment, which
is consistent with reduction of TiO2 in the light as the origin of
the photocurrent decay. However, the TiO2 reduction was not
completely reversible in the dark with air purging, and after a few
regeneration cycles the photocurrent decay became irreversible,
perhaps because a critical concentration of Ti3+ was reached. Even
with this drawback, good photocurrent was observed after 1 h using
the regeneration technique (Supplementary Fig. S9). This is an

NATURE MATERIALS | VOL 10 | JUNE 2011 | www.nature.com/naturematerials

2011 Macmillan Publishers Limited. All rights reserved

459

NATURE MATERIALS DOI: 10.1038/NMAT3017

ARTICLES
Zn 2p

1,050

Binding energy (eV)

1,040
1,035
1,030

Intensity (cps)

1,045

250
500
750
1,000
1,250
1,500
1,750
2,000
2,250
2,500
2,750

1,025
1,020
0

30

60

90 120 150 180


Sputtering time (s)

210 240 270

Pt 4f

82

Binding energy (eV)

78
76

Intensity (cps)

80

100
200
300
400
500
600
700

74
72
70
68
66
0

30

60

90 120 150 180


Sputtering time (s)

210 240 270

Ti 2p

460

455

Intensity (cps)

Binding energy (eV)

465

250
500
750
1,000
1,250
1,500
1,750
2,000
2,250
2,500
2,750
3,000
3,250
3,500
3,750

450
0

30

60

90 120 150 180


Sputtering time (s)

210 240 270

Figure 5 | Contour plots for the XPS signals of Zn, Pt and Ti from the
barrier layer. XPS spectra with depth profile analysis on a Cu2 O/5
(4 nm ZnO/0.17 nm Al2 O3 )/11 nm TiO2 /Pt electrode after
photoelectrochemical characterization. The XPS spectra acquisition was
carried out after 15 s cycles of Ar+ sputtering. The data were interpolated
between each XPS spectra, but were not smoothed.

extraordinary improvement over the bare electrode, which gave


photocurrent for only about 1 min.
Despite the limitations of TiO2 in our protecting strategy, we
have shown that it is possible to ameliorate decomposition of
460

Cu2 O in aqueous electrolytes. Photocurrents up to 7.6 mA cm2


at 0 V versus RHE were demonstrated using electrodes protected
with catalysed layers of n-type oxides with the structure 5
(4 nm ZnO/0.17 nm Al2 O3 )/11 nm TiO2 . To our knowledge this
is the highest photocurrent ever measured for an oxide-based
photoelectrode for PEC water splitting under AM 1.5 illumination.
We found a Faradaic efficiency close to 100% for water reduction
based on volumetric measurements of the evolved gas, and stability
testing showed good photocurrents even after 1 h of testing. The
photocurrent did decrease with time, however, which was probably
due to electron accumulation in the TiO2 layer, whose Fermi level
was not optimally positioned for water splitting. We intend to
pursue alternatives to amorphous TiO2 in the future, and expect
that a new material will afford stability on the timescale of weeks
or even years. Finally, because our protection strategy can easily be
extended to high-aspect-ratio nanostructures, it will no doubt be
beneficial for future nanostructured Cu2 O electrodes designed to
harvest all of the absorbed solar energy.

Methods
Cu2 O electrodeposition. The Cu2 O thin films were deposited by electrodeposition
from a basic solution of lactate-stabilized copper sulphate. The deposition substrate
(working electrode) was TEC-15 fluorine-doped tin oxide (FTO, NGS glass)
coated with 200 nm of Au, deposited by thermal evaporation. An adhesion layer
of 20 nm of Cr was deposited by thermal evaporation on the FTO substrate before
gold deposition. Au-coated substrates were chosen over indium tin oxide or FTO
because the deposition reproducibility was better. The plating bath was 0.2 M
CuSO4 (Sigma Aldrich) and 3 M lactic acid (Fisher Scientific) solution in deionized
water with 0.5 M K2 HPO4 (Sigma) buffer. The bath pH was adjusted to pH 12
by adding a controlled amount of 2 M KOH. The temperature of the bath was
maintained at 30 C using a heating plate with an in situ temperature probe. The
Cu2 O thin films were deposited at a constant current density of 0.1 mA cm2
(Galvanostatic mode) for 220 min using a source meter (2400, Keithley) in a
two-electrode configuration (a Pt mesh served as the counter electrode), which
resulted in a film thickness of 1.3 m, as measured by a profilometer (Alpha-step,
Trencor). The parameters for Cu2 O electrodeposition from Cu (ii) lactate solution
were optimized for photoelectrochemical response and the details will be given
in a separate article.
Atomic layer deposition. Ultrathin protective layers of n-type oxides were
deposited on the surface of the Cu2 O thin films using a thermal ALD system
(Savannah 100, Cambridge Nanotech). Before ALD, to remove any residual
organic contaminants remaining from the electrodeposition38 and to saturate the
surface with hydroxyl groups in preparation for ALD, the Cu2 O samples were
dipped in 30% H2 O2 in water and immediately rinsed with copious amounts of
deionized water. The H2 O2 rinsing procedure was repeated twice. The ALD was
carried out at a substrate temperature of 120 C. This temperature was chosen
because a slight discolouration (and decrease in activity) of the Cu2 O thin film
was observed if deposition was carried out at higher temperatures of 200 C. Zinc
oxide (ZnO) was deposited using diethyl zinc (Tprecursor = room temperature) and
H2 O (Tprecursor = room temperature) as the Zn and O precursors, respectively.
Each precursor was held in the chamber for 2.0 s, followed by a 15.0 s nitrogen
purge. The GPC was measured by ellipsometry on films deposited on optically
polished silicon wafers with a native surface oxide, and for ZnO was 2.0 .
Aluminium-doped ZnO (Al:ZnO) was deposited by running 1 cycle of trimethyl
aluminium (Tprecursor = room temperature) and water after the specified number of
diethyl zinc and water cycles, either 20 or 10. For reference, Al2 O3 grown at these
conditions exhibited a GPC of 1.7 . Titanium dioxide (TiO2 ) was deposited using
titanium tetraisopropoxide (Tprecursor = 80 C) and H2 O as the Ti and O precursors,
respectively. Each precursor was held in the chamber for 2.0 s followed by a 15.0 s
nitrogen purge. The GPC for TiO2 was 0.28 . The cycle numbers were varied to
control the film thickness, which was calculated using the GPC values. Following
ALD, the samples were platinized either as-deposited or after heat treatment at
200 C for 45 min in air.
Cocatalyst deposition. To enhance the kinetics of the water reduction reaction,
platinum nanoparticles were potentiostatically electrodeposited in the dark from a
solution of 1 mM H2 PtCl6 in deionized water at 0.1 V versus Ag/AgCl for 15 min.
Epoxy resin was applied before platinization to cover any exposed FTO or Au.
Structural characterization. The electrodes were characterized by SEM, XRD
and XPS. The morphology of the films was characterized using a high-resolution
scanning electron microscope (FEI XL30 SFEG). XRD patterns were acquired with
a Bruker D8 Discover diffractometer in the BraggBrentano mode, using Cu K
radiation (1.540598 ) and a Ni -filter. Spectra were acquired with a linear
NATURE MATERIALS | VOL 10 | JUNE 2011 | www.nature.com/naturematerials

2011 Macmillan Publishers Limited. All rights reserved

NATURE MATERIALS DOI: 10.1038/NMAT3017


silicon strip Lynx Eye detector from 2 = 20 80 at a scan rate of 3 min1 ,
step width of 0.02 and a source slit width of 4 mm. Reflection patterns were
matched to the PDF-4+ database (ICDD). XPS data were collected by an Axis Ultra
instrument (Kratos Analytical) under ultrahigh vacuum (<108 torr) and using a
monochromatic Al K X-ray source (1,486.6 eV), in the Surface Analysis Facility
of the Interdisciplinary Centre for Electron Microscopy at EPFL. The source
power was maintained at 150 W and the emitted photoelectrons were sampled
from a 200 m area. The photoelectron take-off angle was 90 . The analyser
pass energy was 80 eV for survey spectra and 40 eV for high-resolution spectra.
As the chemical state of copper is difficult to determine using the Cu 2p binding
energy, the Auger Cu LMM signal was also recorded. The adventitious carbon
1s peak was calibrated at 285 eV and used as an internal standard to compensate
for any charging effects. Sputtering of the surface by a 3 keV argon ion beam was
used to depth profile the samples with an estimated crater size of 1 1 mm2 . The
etch rate was approximately 4 nm min1 . Both curve fitting of the spectra and
quantification were carried out with the CasaXPS software, using relative sensitivity
factors given by Kratos.
Photoelectrochemical measurements. The photoelectrochemical performance of
the electrodes was evaluated in a three-electrode configuration under front-side
simulated AM 1.5 illumination using an Ivium Potentiostat/Galvanostat. The
electrolyte was a 1.0 M Na2 SO4 solution buffered at pH 4.9 with potassium
phosphate (0.1 M). The reference electrode was Ag/AgCl in saturated KCl, and a Pt
wire was used as the counter electrode. The photoresponse was measured under
chopped irradiation from a 450 W Xe lamp (Osram) equipped with an ultraviolet
filter (KG3 filter, 3 mm, Schott), calibrated with a Si diode to simulate AM 1.5
illumination (100 mW cm2 ). The scan rate for the linear sweep voltammetry and
for the cyclic voltammetry was 10 mV s1 . Photocurrent stability tests were carried
out by measuring the photocurrent produced under chopped light irradiation
(light/dark cycles of 30 s) at a fixed electrode potential of 0 V versus RHE. During
linear sweep voltammetry (J V plots) and chronoamperometry (stability plots) the
electrolyte was continuously bubbled with N2 to remove oxygen and thus eliminate
erroneous signals arising from oxygen reduction.

Received 9 August 2010; accepted 28 March 2011;


published online 8 May 2011

References
1. Khaselev, O. & Turner, J. A. A monolithic photovoltaicphotoelectrochemical device for hydrogen production via water splitting.
Science 280, 425427 (1998).
2. Aharon-Shalom, E. & Heller, A. Efficient p-InP(Rh-H alloy) and p-InP(ReH
alloy) hydrogen evolving photocathodes. J. Electrochem. Soc. 129,
28652866 (1982).
3. Marsen, B., Cole, B. & Miller, E. L. Photoelectrolysis of water using thin
copper gallium diselenide electrodes. Sol. Energy Mater. Sol. Cells 92,
10541058 (2008).
4. Fernandez, A. M. et al. Photoelectrochemical characterization of the
Cu(In,Ga)S2 thin film prepared by evaporation. Sol. Energy Mater. Sol. Cells
85, 251259 (2005).
5. Hara, M. et al. Cu2 O as a photocatalyst for overall water splitting under visible
light irradiation. Chem. Commun. 357358 (1998).
6. de Jongh, P. E., Vanmaekelbergh, D. & Kelly, J. J. Photoelectrochemistry of
electrodeposited Cu2 O. J. Electrochem. Soc. 147, 486489 (2000).
7. Siripala, W., Ivanovskaya, A., Jaramillo, T. F., Baeck, S. H. & McFarland, E. W.
A Cu2 O/TiO2 heterojunction thin film cathode for photoelectrocatalysis.
Sol. Energy Mater. Sol. Cells 77, 229237 (2003).
8. Hu, C. C., Nian, J. N. & Teng, H. Electrodeposited p-type Cu2 O as photocatalyst
for H-2 evolution from water reduction in the presence of WO3. Sol. Energy
Mater. Sol. Cells 92, 10711076 (2008).
9. Nian, J. N., Hu, C. C. & Teng, H. Electrodeposited p-type Cu2 O for H-2
evolution from photoelectrolysis of water under visible light illumination.
Int. J. Hydrog. Energy 33, 28972903 (2008).
10. Barreca, D. et al. The potential of supported Cu2 O and CuO nanosystems in
photocatalytic H-2 production. ChemSusChem 2, 230233 (2009).
11. Morales, J. et al. Electrodeposition of Cu2 O: An excellent method for obtaining
films of controlled morphology and good performance in Li-ion batteries.
Electrochem. Solid State Lett. 8, A159A162 (2005).
12. Lee, Y. H. et al. Fabrication and characterization of Cu2 O nanorod
arrays and their electrochemical performance in Li-ion batteries.
Electrochem. Solid State Lett. 9, A207A210 (2006).
13. Bijani, S. et al. Nanostructured Cu2 O thin film electrodes prepared by
electrodeposition for rechargeable lithium batteries. Thin Solid Films 515,
55055511 (2007).
14. Katayama, J., Ito, K., Matsuoka, M. & Tamaki, J. Performance of Cu2 O/ZnO
solar cell prepared by two-step electrodeposition. J. Appl. Electrochem. 34,
687692 (2004).
15. Akimoto, K. et al. Thin film deposition of Cu2 O and application for solar cells.
Sol. Energy 80, 715722 (2006).

ARTICLES
16. Jeong, S. S., Mittiga, A., Salza, E., Masci, A. & Passerini, S.
Electrodeposited ZnO/Cu2 O heterojunction solar cells. Electrochim. Acta 53,
22262231 (2008).
17. Cui, J. B. & Gibson, U. J. A simple two-step electrodeposition of Cu2 O/ZnO
nanopillar solar cells. J. Phys. Chem. C 114, 64086412 (2010).
18. Graetzel, M. Photoelectrochemical cells. Nature 414, 338344 (2001).
19. de Jongh, P. E., Vanmaekelbergh, D. & Kelly, J. J. Cu2 O: Electrodeposition and
characterization. Chem. Mater. 11, 35123517 (1999).
20. Sculfort, J. L., Guyomard, D. & Herlem, M. Photoelectrochemical
characterization of the p-Cu2 O-non aqueous electrolyte junction. Electrochim.
Acta 29, 459465 (1984).
21. Engel, C. J., Polson, T. A., Spado, J. R., Bell, J. M. & Fillinger, A.
Photoelectrochemistry of porous p-Cu2 O films. J. Electrochem. Soc. 155,
F37F42 (2008).
22. George, S. M. Atomic layer deposition: An overview. Chem. Rev. 110,
111131 (2010).
23. Hwang, Y. J., Boukai, A. & Yang, P. High density n-Si/n-TiO2 core/shell
nanowire arrays with enhanced photoactivity. Nano Lett. 9, 410415 (2009).
24. Shan, C. X., Hou, X. H. & Choy, K. L. Corrosion resistance of TiO2 films
grown on stainless steel by atomic layer deposition. Surf. Coat. Technol. 202,
23992402 (2008).
25. Guidi, F. et al. Electrochemical anticorrosion performance evaluation of
Al2 O3 coatings deposited by MOCVD on an industrial brass substrate.
Electrochim. Acta 50, 46094614 (2005).
26. Du, X. H., Zhang, K., Holland, K., Tombler, T. & Moskovits, M. Chemical
corrosion protection of optical components using atomic layer deposition.
Appl. Opt. 48, 64706474 (2009).
27. Galliano, P., De Damborenea, J. J., Pascual, M. J. & Duran, A. Solgel
coatings on 316L steel for clinical applications. J. SolGel Sci. Technol. 13,
723727 (1998).
28. Heller, A. Hydrogen-evolving solar cells. Science 223, 11411148 (1984).
29. Standridge, S. D., Schatz, G. C. & Hupp, J. T. Toward plasmonic solar cells:
Protection of silver nanoparticles via atomic layer deposition of TiO2 . Langmuir
25, 25962600 (2009).
30. Elam, J. W., Routkevitch, D. & George, S. M. Properties of ZnO/Al2 O3 alloy
films grown using atomic layer deposition techniques. J. Electrochem. Soc. 150,
G339G347 (2003).
31. Agashe, C. et al. Efforts to improve carrier mobility in radio frequency sputtered
aluminium doped zinc oxide films. J. Appl. Phys. 95, 19111917 (2004).
32. Lee, D-J. et al. Structural and electrical properties of atomic layer deposited
Aldoped ZnO films. Adv. Funct. Mater. 21, 448455 (2011).
33. Yasomanee, J. P. & Bandara, J. Multi-electron storage of photoenergy
using Cu2 OTiO2 thin film photocatalyst. Sol. Energy Mater. Sol. Cells 92,
348352 (2008).
34. Xiong, L. B. et al. Visible-light energy storage by Ti3+ in TiO2 /Cu2 O bilayer
film. Chem. Lett. 38, 11541155 (2009).
35. ORegan, B., Grtzel, M. & Fitzmaurice, D. Optical electrochemistry I:
Steady-state spectroscopy of conduction-band electrons in a metal oxide
semiconductor electrode. Chem. Phys. Lett. 183, 8993 (1991).
36. Howe, R. F. & Gratzel, M. EPR observation of trapped electrons in colloidal
titanium dioxide. J. Phys. Chem. 89, 44954499 (1985).
37. Kaise, M. et al. Electron spin resonance studies of photocatalytic interface
reactions of suspended M/TiO2 (M = Pt, Pd, Ir, Rh, Os, or Ru) with alcohol
and acetic acid in aqueous media. Langmuir 10, 13451347 (1994).
38. Nadesalingam, M. P. et al. Effect of vacuum annealing on the surface chemistry
of electrodeposited copper(i) oxide layers as probed by positron annihilation
induced auger electron spectroscopy. Langmuir 23, 18301834 (2007).

Acknowledgements
The research leading to these results has received funding from the European
Communitys Seventh Framework Programme (FP7/2007-2013) under grant agreement
no. 227179 (NanoPEC). We would also like to thank the Swiss Federal Office of Energy
(Project number 102326, PECHouse) and the Energy Center at EPFL for financial
support. We acknowledge N. Xanthopoulos from the Interdisciplinary Centre for
Electron Microscopy (CIME) at EPFL for helping in the XPS characterization. E.T. would
also like to thank A. Martinson at Argonne National Laboratory for helpful discussions
about aluminium-doped zinc oxide synthesized by ALD.

Author contributions
A.P. and E.T. designed the experiments. A.P. carried out electrodeposition, PEC
measurements, SEM and XRD. ALD was carried out by E.T. Faradaic efficiency
measurements were carried out by K.S. XPS was carried out by V.L. M.G. supervised the
project. All of the authors discussed and analysed the data.

Additional information
The authors declare no competing financial interests. Supplementary information
accompanies this paper on www.nature.com/naturematerials. Reprints and permissions
information is available online at http://www.nature.com/reprints. Correspondence and
requests for materials should be addressed to E.T.

NATURE MATERIALS | VOL 10 | JUNE 2011 | www.nature.com/naturematerials

2011 Macmillan Publishers Limited. All rights reserved

461

S-ar putea să vă placă și