of 15
Energy &
Environmental
Science
www.rsc.org/ees
ISSN 1754-5692
SPECIAL COLLECTION
Editorial by Eric Miller, with contributions from Ager
et al
., Fabian
et al
.,
Coridan
et al
., Smith
et al
. and Esposito
et al
.
Photoelectrochemical Water Splitting
Volume 8 Number 10 October 2015 Pages 2799–3050
Published on 24 March 2015. Downloaded on 29/10/2015 17:25:58.
View Article Online
View Journal
| View Issue
This journal is
©
The Royal Society of Chemistry 2015
Energy Environ. Sci.,
2015,
8
, 2811--2824 |
2811
Cite this:
Energy Environ. Sci.,
2015,
8
,2811
Experimental demonstrations of spontaneous,
solar-driven photoelectrochemical water
splitting
Joel W. Ager,*
ab
Matthew R. Shaner,
cd
Karl A. Walczak,
ab
Ian D. Sharp
ae
and
Shane Ardo
f
Laboratory demonstrations of spontaneous photoelectrochemical (PEC) solar water splitting cells are
reviewed. Reported solar-to-hydrogen (STH) conversion efficiencies range from
o
1% to 18%. The
demonstrations are categorized by the number of photovoltaic junctions employed (2 or 3), photovoltaic
junction type (solid–solid or solid–liquid) and the ability of the systems to produce separated reaction
product streams. Demonstrations employing two photovoltaic (PV) junctions have the highest reported
efficiencies of 12.4% and 18%, which are for cells that
, respectively, do and do not contain a semiconductor–
liquid junction. These devices used PV components b
ased on III–V semiconductors; recently, a number of
demonstrations with
4
10% STH efficiency using potentially less cos
tly materials have been reported. Device
stability is a major challenge for the field, as evidenc
ed by lifetimes of less than 24 hours in all but a few
reports. No globally accepted protocol for evaluating and certifying STH efficiencies and lifetimes exists. It is
our recommendation that a protocol similar to that used by the photovoltaic community be adopted so that
future demonstrations of solar PEC water s
plitting can be compared on equal grounds.
Broader context
There is significant recent interest in solar-driven photoelectrochemical water splitting to produce hydrogen as a potential carbon-neutral tran
sportation fuel.
Renewable energy technologies must provide a positive monetary and net energy balance over their lifetimes to be viable for large scale deployment. T
echno-
economic analyses have suggested that solar photoelectrochemical water splitting could provide hydrogen at a cost that is competitive with energy d
erived from
fossil fuels. Thus, economical solar water splitting represents a goal with broad-reaching appeal. One specific implementation of this concept is a
n integrated
or monolithic solar-to-fuel conversion device that operates spontaneously, without added external electrical bias. Experimental demonstration
s of such systems
date back to the early 1970s, when Fujishima and Honda first reported solar water splitting using single-crystal TiO
2
. This inspired considerable research in the
field and to-date there have been over 40 reported demonstrations of spontaneous, solar-driven photoelectrochemical water splitting. These have l
ed to
increased fundamental and functional understanding and to increases in the overall energy-conversion efficiency. Herein, we compile reported solar
-to-
hydrogen conversion efficiencies and longevities. This information can be used to evaluate progress in the field and to target technical areas for futu
re
development.
Introduction
There is considerable interest in developing technologies which
could provide a sustainable alternative to the combustion of
fossil fuels to meet the current and future energy demands
of the planet.
1
Conversion of abundant sunlight to storable
energy is an attractive approach. This concept underlies biofuel
production,
2–4
as well as a number of solar-to-fuel or ‘‘artificial
photosynthesis’’ approaches.
5–7
This review concentrates on
approaches that use sunlight to split water into hydrogen and
oxygen,
8
noting the recent review by Ronge
́
et al.
,
9
which also
covers solar-driven carbon dioxide reduction. Hydrogen is a
storable fuel that can be used as a feedstock for fuel cells that
a
Joint Center for Artificial Photosynthesis, Lawrence Berkeley National Laboratory,
Berkeley, CA, USA. E-mail: JWAger@lbl.gov
b
Materials Sciences Division, Lawrence Berkeley National Laboratory, Berkeley, CA,
USA
c
Joint Center for Artificial Photosynthesis, California Institute of Technology,
Pasadena, CA, USA
d
Division of Chemistry and Chemical Engineering, California Institute of
Technology, Pasadena, CA, USA
e
Physical Biosciences Division, Lawrence Berkeley National Laboratory, Berkeley,
CA, USA
f
Department of Chemistry, and Department of Chemical Engineering and Materials
Science, University of California Irvine, Irvine, CA, USA
Electronic supplementary information (ESI) available: Additional analysis of
reported STH conversion efficiency and longevity as a function of electrolyte pH
and device configuration. See DOI: 10.1039/c5ee00457h
Received 10th February 2015,
Accepted 24th March 2015
DOI: 10.1039/c5ee00457h
www.rsc.org/ees
Energy &
Environmental
Science
REVIEW
Published on 24 March 2015. Downloaded on 29/10/2015 17:25:58.
View Article Online
2812
|
Energy Environ. Sci.,
2015,
8
, 2811--2824
This journal is
©
The Royal Society of Chemistry 2015
generate power for transportation and, potentially, grid-scale
energy storage.
10–13
Hydrogen can also be used in processes
that reduce CO
2
to liquid fuels.
6
Solar irradiation can be used for thermal and/or electro-
chemical water splitting.
14–17
Electrochemical water splitting
requires the following overall cat
hodic and anodic half-reactions
(in acid):
8,9
2H
+
+2e

-
H
2
(1)
H
2
O
-
2e

+2H
+
+ 1/2O
2
(2)
The free energy change for water splitting to hydrogen (and oxygen)
under standard-state conditions is
D
G
1
= +237 kJ per mol of H
2
or
D
E
1
=

1.23 V. This must be supplied by the energy in sunlight.
To achieve current densities limited by the solar photon flux
(
B
20 mA cm

2
), and considering overpotential requirements of
state-of-the-art electrocatalysts and the trade-off of current and
voltage in light absorbers, a total photovoltage of 1.6–1.7 V
must be generated. However, the open-circuit photovoltages
provided by commercially developed single-junction photo-
voltaic (PV) cells are typically
o
1 V. Therefore, either series
connected cells or wide-bandgap semiconductors must be
employed to drive solar water splitting in the absence of an
external power source. This review article concerns experi-
mental demonstrations of the former type. It begins with a
short historical discussion of the field, which began in the peer-
reviewed literature in the early 1970’s with reports of photo-
driven water splitting.
18,19
It focuses on trends in efficiency and
stability, as well as designs of the photovoltaic and catalytic
elements of the systems.
The solar-to-hydrogen (STH) conversion efficiency,
Z
, for
solar water splitting at standard temperature and pressure of
H
2
and O
2
is given by:
20
Z
¼
ð
1
:
23 V
Þ
J
op

P
in
(3)
where
J
op
is the operational photocurrent density in mA cm

2
,
or the rate of hydrogen production converted to a current
density, and
P
in
is the incident irradiance in mW cm

2
. This
review describes reported STH efficiencies and stabilities
because standard testing by independent research laboratories
does not yet exist. The STH efficiencies are also compared to
theoretical limits, and the review outlines research priorities
for the field.
History of solar-driven photoelectrochemical (PEC) water
splitting
In 1972, Fujishima and Honda published a report of light-
driven PEC water splitting in the absence of applied electrical
bias that gave rise to the modern field of artificial photosynthesis
research.
18,19
Their demonstration used a single-crystal titanium
oxide (TiO
2
,
E
g
E
3.0 eV) photoanode and a platinum (Pt)
cathode. Current–voltage curves were measured under illumina-
tion and oxygen was detected as a product at the photoanode.
In these initial reports, product detection at the cathode and the
pH of the electrolyte solution(s) contacting the electrodes were
not reported. Work by other groups to reproduce the discovery
established the conditions necessary for sustainable, sponta-
neous water splitting.
21–26
Fujishima and Honda’s report ignited considerable interest
in exploring solar water splitting as a practical means to
generate clean fuels and led to efforts to find other semicon-
ductor materials that could yield higher efficiencies. Much of
the subsequent work focused on wide-band gap metal oxides
and oxynitrides, whose valence and conduction band positions
‘‘straddle’’ the water splitting redox potentials. Both powdered
and electrode photocatalysts of this type have been thoroughly
investigated.
27–31
However, very few of these systems achieved
spontaneous (
i.e.
no applied bias) water-splitting using visible
illumination and thus had very low STH conversion efficiencies.
This body of work has been the subject of a number of previous
recent reviews which have focused on particle photocatalyst
systems.
32–40
In 1975, Yoneyama
et al.
experimentally demonstrated that a
p-GaP/n-TiO
2
tandem combination could generate H
2
and O
2
without external bias.
41
Nozik showed in 1976 that this type of
tandem-junction architecture, consisting of a p-type photo-
cathode and an n-type photoanode (Fig. 1), could achieve a
higher STH conversion efficiency than a single photoelectrode.
42
Shortly after, other groups explored related tandem architectures
including n-GaP/p-GaP, p-CdTe/n-TiO
2
,p-CdTe/n-SrTiO
3
,and
p-GaP/n-SrTiO
3
.
43
STH conversion efficiencies in this early work
were low,
o
1%. Also, the stability of the active components,
particularly the photoanode, emerged as a critical challenge that
remains to this day.
41,44,45
Driven by advances in higher efficiency single-junction (1J)
and tandem-junction (2J) solar cells in the mid-1980s, efficien-
cies for solar PEC water splitting also increased. For example,
Bockris and co-workers reported that a p-InN photocathode
wired side-by-side with an n-GaAs photoanode achieved an STH
conversion efficiency of 8% and a lifetime of 10 hours.
46
Monolithic architectures using mu
ltijunction amorphous silicon
(a-Si) were also explored with reported STH conversion efficiencies
in the 2–3% range.
46–49
Starting in the late 1990s high-efficiency approaches based on
all III–V and Si/III–V 2J monolithic architectures were developed.
Fig. 1
Schematic of an idealized tandem-junction photoanode and photo-
cathode device during steady-state operation. The process of solar water
splitting is overlaid on the equilibrium diagram. Proton conduction in the
electrolyte from the anode to the cathode is required for continuous
operation. Adapted from Nozik.
42
Review
Energy & Environmental Science
Published on 24 March 2015. Downloaded on 29/10/2015 17:25:58.
View Article Online
This journal is
©
The Royal Society of Chemistry 2015
Energy Environ. Sci.,
2015,
8
, 2811--2824 |
2813
This work culminated in the early 2000s with demonstrations
of 12 and 18% STH conversion efficiencies by Turner and
co-workers and by Licht
et al.
,respectively.
50–52
Triple-junction
(3J) amorphous silicon (a-Si) cells were also investigated starting
in the late 1980s, as their open-circuit photovoltage can exceed
2V.
50,53–56
Miller and co-workers reported STH conversion
efficiencies as high as 8% with this approach.
50,56
It is also
notable that the longest reported operational stability for solar-
driven PEC water splitting, more than a month, was achieved by
Kelly and Gibson with this architecture.
49
Since 2010, there have been significant efforts to replace
noble-metal electrocatalysts with those made from less expen-
sive elements, to use metal-oxide light absorbers that may
be more stable, and to demonstrate fully integrated devices
(
i.e.
, those with intimate contact between all light absorbing
and catalytic components without w
ires). The first fully integrated
demonstration was in 2011 by Nocera and co-workers where an
STHconversionefficiencyof2.5%wasreportedfora
completely
integrated
3J a-Si cell incorporating h
ydrogen and oxygen evolu-
tion catalysts made from abundant elements on its surfaces.
57
A
number of notable recent reports use metal oxide absorbers such
as WO
3
,Fe
2
O
3
, and BiVO
4
. By coupling with dye sensitized solar
cells and 1J and 2J a-Si solar cells
, STH efficiencies ranging from
2% to over 5% have been achieved.
58–60
While systems of integrated photovoltaic and catalytic com-
ponents may be conceptually attractive, physically separating
the photovoltaic (PV) and electroc
atalyst materials can circumvent
some of the stability issues that are present in the more integrated
PEC water splitting demonstrations. Four recent reports that
demonstrate this approach include: 15% STH conversion effi-
ciency using three side-by-
side 3J III–V/Ge cells with 10

optical
concentration,
61
10%STHconversionefficiencyusingthree
series-connected, side-by-side CuIn
x
Ga
1

x
Se
2
(CIGS) solar cells,
62
12% STH conversion efficiency u
sing two organic–inorganic
halide perovskite solar cells,
63
and 10% STH conversion efficiency
using 4 side-by-si
de Si minimodules.
64
Very recently, since 2013, efforts to use non-planar semi-
conductor geometries and advanced photon management
strategies and concepts have received interest. These approaches
have a number of potential advantages. The directions of light
absorption and charge separation can be orthogonalized, allow-
ingtheuseoflesspurematerials,
65
and properly designed arrays
can use light trapping to reduce the amount of required absorber
material.
66
There are a few reports of achieving spontaneous
solar-driven water splitting using this type of approach but, so
far, the reported STH efficiencies have remained low (
o
1%).
67–69
Nomenclature, device description, and data presentation
Device description.
The nomenclature used herein is adopted
from a recent photoelectrochemical taxonomy, which is summarized
in Table 1.
70
All electrical architectures covered in this review consist
of two or three photovoltaic junctions connected electrically in series.
Unless otherwise noted, the opti
cal architecture is assumed to be a
stacked arrangement, with the h
igher bandgap absorber on top,
facing the light source. Side-by-s
ide arrangements are also reported
and are designated as such. We also d
istinguish between integrated
cells and those in which wires connect the PV cells.
We make a distinction between cells that use semiconductor–
liquid junctions to separate photoinduced charge carriers, as
shown in Fig. 1, with those that use ‘‘buried’’ solid–solid junc-
tions (
e.g.
pn) to perform the charge separation. Devices that use
at least one semiconductor–liquid junction are called ‘‘photo-
electrosynthetic’’ and those that employ buried junctions are
called ‘‘photovoltaic-biased electrosynthetic.’’ We also denote
the method used to electrically connect the PV junction to the
HER and OER catalysts, if these are employed in the design. In
integrated devices, the catalysts are directly deposited on the PV
element, often as a thin film or as nanoparticles. In other
approaches, the catalyst is wired t
o the PV element(s). Approaches
that wire both the HER and OER catalysts are often called ‘‘PV +
electrolyzer.’’ Finally, we note whether or not the demonstration
attempted to separate the chemic
al reaction products to yield a
Table 1
Device nomenclature
SLJ
Semiconductor–liquid junction
Photoelectrosynthetic cell
A cell whose photo-voltage producing junctions are all semiconductor–liquid in character
Photovoltaic-biased photoelectrosynthetic cell
A cell whose photo-voltage producing junctions consist of at least one
semiconductor–liquid junction and one solid-state junction
Photovoltaic-biased electrosynthetic cell
A cell whose photo-voltage producing junctions are all solid-state in character
Tandem junction (2J)
A device containing two photo-voltage producing junctions
Triple junction (3J)
A device containing three photo-voltage producing junctions
a
Amorphous
c
Crystalline
Pin
Buried junctions in series as p-type, intrinsic, and then n-type
DSSC
Dye-sensitized solar cell
CIGS
CuIn
x
Ga
1

x
Se
2
OER
Oxygen-evolution reaction
HER
Hydrogen-evolution reaction
PEM
Proton-exchange membrane
MEA
Membrane-electrode assembly
Photocatalyst
A single material that simultaneously acts as semiconductor light absorber and as catalyst
Co-evolved products
H
2
and O
2
evolve without a physical barrier such as a membrane or
separator to prevent chemical cross-over
Integrated
Intimate contact between the catalyst and semiconductor surface
Wired
Physically separated catalyst and semiconductor surfaces connected
through a wire or the equivalent
Energy & Environmental Science
Review
Published on 24 March 2015. Downloaded on 29/10/2015 17:25:58.
View Article Online
2814
|
Energy Environ. Sci.,
2015,
8
, 2811--2824
This journal is
©
The Royal Society of Chemistry 2015
pure H
2
fuel stream. Fig. 2 depicts some of the more commonly
employed geometries in the form of circuit diagrams.
Device performance.
The data summarized in the tables and
figures that follow are reported directly from the original
references. The only change which has been made is correction
of efficiencies reported using the higher heating value of H
2
(1.48 eV per electron); in these cases, the efficiencies were
adjusted to use the free energy of the water splitting reaction
(1.23 eV per electron).
71
There are a number of published recommendations for
standardized photoelectrochemical testing of half cells and full
cells.
71–73
Ideally, analogous with the well-established testing
protocols for solar cells,
74
the STH efficiency of each demonstra-
tion would be confirmed by an independent testing laboratory
using incident light that corresponds to the solar spectrum,
together with direct and accurate measurement of H
2
and O
2
products. However, independent testing labs of this type do not
currently exist for solar PEC water splitting or for any other solar-
to-fuel conversion technology.
Thus, most of the demonstrations used solar simulators
optimized for testing Si solar cells and calculated the STH
conversion efficiency
via
a current density measurement
assuming 100% Faradaic efficiency for H
2
production. Accurate
testing of tandem solar cells, the type of architecture used by
most of the demonstrations in this review, actually requires
careful control of temperature, solar simulator spectrum, and a
number of other factors.
75
Also, most of the studies measured
the current only; quantification of the amount of H
2
and O
2
generated, and confirmation of their 2 : 1 ratio expected from
reactions (1) and (2), was less common.
A consensus definition of device stability that is evaluated by
most researchers does not yet exist in the solar PEC water
splitting community. In this review, we tabulate, if available,
the duration and results of long-term testing performed on the
devices. We also note, briefly, the criterion used by the authors
to evaluate or terminate their stability test. Most often, the
authors either establish a period of time over which the
photocurrent is reasonably stable or, alternatively, drops by
ca.
10–20%. Less common is the monitoring of H
2
(and even
less commonly, O
2
) over time. We also observe that, in the vast
majority of cases, stability data from a single device is pre-
sented. This contrasts with the parallel testing, often under
accelerated conditions, which is used in the evaluation of PV
device lifetimes.
In the absence of accepted standards and independent
testing, it is not valid to directly compare the claimed STH
efficiencies and stabilities reported herein or to declare a
‘‘world record.’’ Nevertheless, the two tabulated metrics (STH
conversion efficiency and device stability) currently provide a
means of tracking progress and identifying bottlenecks in the
field. Finally, it is important to acknowledge that achieving
ultimate efficiency or stability was not necessarily the primary
objective for many of the reports of solar water splitting. Instead,
much of the work was dedicated to exploring new approaches or
concepts in photoelectrochemical energy-conversion research.
Data presentation and guide to tables.
Experimental reports
of spontaneous solar water splitting are summarized in Fig. 3,
with the reported STH conversion efficiency graphed
versus
the
year of the report. Tables 2–5 contain short descriptions of the
demonstrations presented in reverse chronological order. Fig. 3
is analogous to the plot of solar PV efficiency
versus
time
maintained by the National Renewable Energy Laboratory
76
and the tables are modelled after a semi-annual report of ‘‘world
records’’ and ‘‘notable exceptions’’ for PV solar cells.
74
The tables
are grouped by the number and type of PV junction(s) as follows:
Table 2: 2J PEC cells with at least one SLJ,
Table 3: 2J PV-biased electrosynthetic cells, including
PV + electrolyzer approaches.
Table 4: 3J PEC cells with at least one SLJ, and
Table 5: 3J PV-biased electrosynthetic cells including
PV + electrolyzers.
The format used for each entry is as follows.
Photocathode//photoanode
Architecture and/or configuration
Configuration and type of HER catalyst
Configuration and type of OER catalyst
For example, the following description,
Fig. 2
Depiction of commonly employed solar photoelectrochemical
water splitting architectures in circuit diagram form. (a) Key for symbols
used; see Table 1 for abbreviations. Wires are indicated by solid lines. If
elements are touching without a wire (
e.g.
the PVs) they are monolithically
integrated. (b) The photoelectrosynthetic geometry shown in Fig. 1 with
integrated PV elements and with both catalysts integrated. (c) A PV-biased
photoelectrosynthetic device with one buried junction (the photoanode
wired to the OER catalyst) and one SLJ. (d) A PV-biased photoelectro-
synthetic device with series-connected PV elements where the OER
catalyst is integrated and the HER catalyst is wired. (e) A PV + electrolyzer
approach with 3 PV cells wired in series and a membrane is used to
separate chemical reaction products.
Review
Energy & Environmental Science
Published on 24 March 2015. Downloaded on 29/10/2015 17:25:58.
View Article Online
This journal is
©
The Royal Society of Chemistry 2015
Energy Environ. Sci.,
2015,
8
, 2811--2824 |
2815
GaInP
2
(pn)//GaAs(pn)
monolithic PV
wired Pt cathode
integrated Pt OER catalyst,
is for a GaInP–GaAs monolithic tandem solar cell with
buried pn junctions for both the 1.8 eV bandgap top cell and
1.4 eV bandgap bottom cell.
50
H
2
production is at a remote Pt
cathode wired to the GaInP cathode. O
2
production occurs at
the surface of the GaAs, which is coated by Pt.
Discussion
Solar-to-hydrogen conversion efficiency
It is interesting to compare the reported efficiencies to calcula-
tions of the theoretical limits for tandem solar to hydrogen
conversion.
94–100
While the assumptions regarding catalyst
overpotentials and device architectures vary, the consensus of
these studies is that a STH conversion efficiency of
4
25% is
possible with a 2J approach for integrated systems in which the
catalyst and absorber areas are equivalent. Both 1J and 3J
approaches have lower efficiency limits. For 1J devices, the
absorber bandgap necessary to generate the required voltage
(1.6–1.7 V) at the point of maximum power generation signifi-
cantly limits the usable solar photon energies and thus results
in current densities below those for 2J devices. 3J devices have
the highest demonstrated efficiencies for PV power generation
(for both 1 sun and optical concentration conditions), but this
is the result of a relatively high photovoltage and low photo-
current density at the point of maximum power generation.
However, if the absorber junction area and catalyst surface area
can be independently varied to optimize the photovoltaic power
curve to the catalyst load curve, as in the PV + electrolyzer
approaches, higher efficiencies are possible with three or more
junctions.
61,101
It is clear from Fig. 3 that the experimentally
demonstrated STH efficiencies to date (
o
19%) are far from the
theoretical limit. This contrasts somewhat with the situation for
solar photovoltaics, where recent
work has produced single-junction
cells close to the theoretical limit (
e.g.
GaAs with near 30% efficiency
compared to the thermodynamic limit of
B
31%).
74,102,103
Architectures and semiconductor–liquid
vs.
solid state
junctions
Subject to the constraints discussed above regarding direct
comparison of STH efficiency values, it is nevertheless interest-
ing to compare the approaches used to achieve relatively high
STH conversion efficiencies. There are 8 reports of
4
10%
efficiency depicted in Fig. 3. Six of these can be categorized as
photovoltaic-biased electrosynthetic, or ‘‘PV + electrolyzer’’,
approaches with essentially decoupled PV and catalytic func-
tions.
52,61–64,84
The remaining two demonstrations are from
Turner and co-workers.
50,51
One of these employed two buried
PV junctions in GaInP
2
and GaAs with a wired Pt cathode
and an integrated Pt anode. The other device, the so-called
Fig. 3
Reported solar to hydrogen (STH) conversion efficiencies as a function of year and sorted by the number of tandem photovoltaic junctions used
(2 or 3). The degree of integration of photovoltaic and catalyst elements is also distinguished, see Fig. 2. The fill colour represents the semiconduc
tor
materials used in the photovoltaic portion of the device. All STH conversion efficiencies are as reported in the original publications (see Tables 2–5)
.
Energy & Environmental Science
Review
Published on 24 March 2015. Downloaded on 29/10/2015 17:25:58.
View Article Online
2816
|
Energy Environ. Sci.,
2015,
8
, 2811--2824
This journal is
©
The Royal Society of Chemistry 2015
Table 2
Two junction (2J) tandem PEC cell water-splitting demonstrations with at least one semiconductor–liquid junction (SLJ) listed in reverse chronolo
gical order. The best-in-class efficiency is
displayed in bold typeface and the best-in-class stability is italicized
Publication
institute(s)
Device structure
(cathode//anode)
Electrolyte and illumination
conditions
STH
efficiency
Stability, notes
Bornoz
et al.
77
(2014)
EPFL, Institute for Solar Fuels,
Delft University
SLJ Cu
2
O(p)//BiVO
4
(n) SLJ
wired PV
integrated RuO
x
HER catalyst
integrated CoP
i
OER catalyst
K
3

x
H
x
PO
4
buffer, pH 6
co-evolved products
100 mW cm

2
0.5%
2 minutes
20% current loss
Wang
et al.
78
(2014)
Beijing Normal University,
Chinese Academy of Sciences,
Soochow University
c-Si(pn)//Fe
2
O
3
(n) SLJ core–shell nanowire
monolithic PV
wired Pt HER catalyst
integrated Au OER catalyst
1MNa
3
PO
4
co-evolved products
60 mW cm

2
6.0%
40 minutes
measurement of O
2
Shaner
et al.
68
(2014)
Caltech, JCAP
c-Si(pn)//WO
3
(n) SLJ core–shell nanowire
monolithic PV
wired Pt cathode
1MH
2
SO
4
co-evolved products
1080 mW cm

2
(
B
11 suns)
0.0068%
10 min
stable current
Liu
et al.
67
(2013)
UC Berkeley
c-Si(pn)//TiO
2
(n) SLJ core–shell nanowire
monolithic PV
integrated Pt HER catalyst
integrated IrO
x
OER catalyst
0.5 M H
2
SO
4
co-evolved products
150 mW cm

2
(1.5 suns)
0.12%
4.5 hours
H
2
and O
2
generation rate
measured
Abdi
et al.
59
(2013)
Delft University, HZB Berlin
a-Si:H(pin)//BiVO
4
(n) SLJ
wired Pt HER catalyst
integrated CoP
i
OER catalyst
pH 7.3
co-evolved products
100 mW cm

2
3.6%
1 hour
stable current
Brillet
et al.
58
(2012)
EPFL
DSSC//WO
3
(n) SLJ
DSSC//Fe
2
O
3
(n) SLJ
wired PV
wired Pt HER catalyst
integrated Al
2
O
3
,Co OER catalyst
1 M HClO
4
,pH0,
1 M NaOH, pH 13.6,
respectively
co-evolved products
100 mW cm

2
3.1%,
1.17%,
respectively
8 hours
30% current drop
Lin
et al.
79
(2012)
University of Cambridge
SLJ Cu
2
O//WO
3
(n) SLJ
wired PV
integrated NiO
x
HER catalyst
0.1 M Na
2
SO
4
,pH6
100 mW cm

2
0.04%
Not reported
Park and Bard
80
(2006)
UT Austin
DSSC//WO
3
(n) SLJ
wired PV
wired Pt HER catalyst
0.25 M Na
2
SO
4
,pH4
200 mW cm

2
1.9%
30 min
measurement of H
2
Khaselev and Turner
51
(1998)
NREL
SLJ GaInP
2
(p)//GaAs(pn)
monolithic PV
integrated Pt HER catalyst
wired Pt OER catalyst
3MH
2
SO
4
,
0.01 M triton X-100
co-evolved products
11 suns
12.4%
20 hours
20% drop in current
Kainthla
et al.
46
(1987)
Texas A&M
SLJ InP(p)//GaAs(n) SLJ
cells wired side-by-side
integrated Pt HER catalyst
integrated MnO OER catalyst
6 M KOH
8.2%
10 hours
initial 10% current drop
Review
Energy & Environmental Science
Published on 24 March 2015. Downloaded on 29/10/2015 17:25:58.
View Article Online
This journal is
©
The Royal Society of Chemistry 2015
Energy Environ. Sci.,
2015,
8
, 2811--2824 |
2817
‘‘Turner cell’’, uses a semiconductor–liquid junction for the
photocathode. To date, no other semiconductor–liquid junc-
tion devices have been able to approach the Turner cell’s
efficiency of 12.4%.
51
The challenges responsible for the low
SLJ device efficiencies are the availability of combinations of
stable photocathode and photoanode materials with bandgaps
commensurate with the solar spectrum and optimized band-
edge positions for the hydrogen and oxygen evolution reactions.
Solutions to these challenges, including new material dis-
coveries, will be required for SLJ devices to rival non-SLJ device
efficiencies.
Semiconductor materials
Traditionally, semiconductor materials used in the high efficiency
PEC devices have been first developed by the solid-state photo-
voltaics community and adapted for use in PEC cells. The first
demonstrations to claim
4
10% STH conversion efficiency
utilized Si and compound III–V and II–VI materials (purple
and blue points in Fig. 3).
50–52,84
More recently, materials such
as CIGS and halide perovskite-based cells have been adapted
into PEC cells that exceeded 10% STH efficiency.
62–64
Over the
last decade, materials such as metal oxides, which have been
developed specifically for PEC applications, have seen substantial
research interest and progress. Reported STH efficiencies for
devices containing metal-oxide-based active components now
exceed 5%.
60
Optical concentration
Optical concentration has also been used in some of the
4
10%
efficient devices depicted in Fig. 3 because it can enhance
photovoltaic efficiencies and utilize smaller areas of semiconductor
material.
51,84,101
Furthermore, concentrator configurations have the
potential to reduce the volume of electrolyte and the balance of
systems burdens associated with liquid handling. However,
additional engineering challenges arise from optical concen-
tration in integrated PEC devices because the increased current
density may increase the load on the catalyst and introduce ionic
conduction limitations in solution, depending on the specific
design. In addition, optical concentration results in increased
photovoltage, which is desired, and increased temperature,
which is detrimental for photovoltaic performance but beneficial
for increasing catalytic activity. The complex trade-offs between
these phenomena have been subject of recent investigations,
97
and deserve more attention toward development of efficient
designs.
Stability
Device stability is a critical challenge for PEC devices to be
commercially deployable. Renewable energy technologies must
provide a positive monetary and net energy balance over their
lifetimes to be viable for large scale deployment. Studies which
have considered the techno-economic
104,105
and energy balance
106,107
considerations of practical PEC solar to hydrogen conversion
have recommended minimum operational lifetimes of at least
several years as well as efficiencies exceeding 10%.
Table 2 (
continued
)
Publication
institute(s)
Device structure
(cathode//anode)
Electrolyte and illumination
conditions
STH
efficiency
Stability, notes
Nakato
et al.
81
(1982)
Osaka University
CdS(n)//TiO
2
(n) SLJ
wired PVs
wired Pt cathode
1.0 M NaOH
co-evolved products
250 W Hg lamp
Not reported
1 hour
steady photocurrent
Mettee
et al.
82
(1981)
UC Berkeley
SLJ GaP(p)//Fe
2
O
3
(n) SLJ
wired PV
integrated Pt HER catalyst
integrated RuO
2
OER catalyst
1MNa
2
SO
4
fritted compartments
sunlight
0.02–0.1%
Not reported
Ohashi
et al.
43
(1977)
Flinders University, Adelaide,
South Australia
SLJ CdTe(p)//TiO
2
(n) SLJ
SLJ GaP(p)//TiO
2
(n) SLJ
SLJ CdTe(p)//SrTiO3(n) SLJ
SLJ GaP(p)//SrTiO
3
(n) SLJ
cells wired side-by-side
1 M NaOH
co-evolved products
100 mW cm

2
0.044%,
0.098%, 0.18%,
0.67%,
respectively
1 hour
stable photovoltage
some cells tested to
50 hours
Morisaki
et al.
83
(1976)
University of Electro-Communications
c-Si(pn)//TiO
2
(n) SLJ
monolithic PV
wired Pt cathode
0.1 M NaOH
co-evolved products
sunlight
0.1%
Not reported
Nozik
42
(1976)
Materials Research Center,
Allied Chemicals Corp.
SLJ GaP(p)//TiO
2
(n) SLJ
wired PVs
0.2 N H
2
SO
4
co-evolved products
85 mW cm

2
0.25%
Not reported
Energy & Environmental Science
Review
Published on 24 March 2015. Downloaded on 29/10/2015 17:25:58.
View Article Online
2818
|
Energy Environ. Sci.,
2015,
8
, 2811--2824
This journal is
©
The Royal Society of Chemistry 2015
Table 3
Two junction (2J) photovoltaic-biased electrosynthetic cell (including PV + electrolyzer) demonstrations of solar water splitting in reverse chr
onological order. The best-in-class reported
efficiency is displayed in bold typeface and the best-in-class reported stability is italicized
Publication
institute(s)
Device structure
Electrolyte and illumination
conditions
STH
efficiency
Stability
Luo
et al.
63
(2014)
EPFL
Halide perovskite (CH
3
NH
3
PbI
3
)
2 cells wired side-by-side
wired NiFe HER catalyst
wired NiFe OER catalyst
PV + electrolyzer
1 M NaOH
Co-evolved products
100 mW cm

2
12.3%
2 hours
stable photocurrent and H
2
/O
2
production
10 hours small degradation
Peharz
et al.
84
(2007)
ISE Fraunhofer, Freiburg, Germany
Ga
0.83
In
0.17
As(pn)//Ga
0.35
In
0.65
P(pn)
monolithic PV
wired Pt HER catalyst
wired IrO
2
OER catalyst
Nafion MEA
PV + electrolyzer
pH 7
membrane separated
evolution of products
500

optical concentration
15%
a
2.3 hours
stable current in
outdoor test
Khaselev
et al.
50
(2001)
NREL
GaInP
2
(pn)//GaAs(pn)
monolithic PV
wired Pt HER catalyst
integrated Pt OER catalyst
2 M KOH
co-evolved products
100 mW cm

2
16.5%
9 hours
stable current in
outdoor test
Licht et al.
52
(2000)
Technion, Israel
Al
0.15
Ga
0.85
As(pn)//Si(pn)
monolithic PV
integrated Pt HER catalyst
integrated RuO
2
OER catalyst
PV is not in contact with electrolyte
1 M HClO
4
co-evolved products
135 mW cm

2
18.3%
14 hours
stable photocurrent
Sakai
et al.
48
(1988)
Osaka University
a-Si(pin)//a-Si(pin)
monolithic PV
integrated Pt HER catalyst
wired RuO
2
OER catalyst
0.5 M H
2
SO
4
co-evolved products
100 mW cm

2
2.93% a-Si in solution
3.23% a-Si out of solution
Not reported
a
Value adjusted from that reported using higher heating value of H
2
. Reported value is that obtained after multiplication by (1.23/1.48).
Review
Energy & Environmental Science
Published on 24 March 2015. Downloaded on 29/10/2015 17:25:58.
View Article Online
This journal is
©
The Royal Society of Chemistry 2015
Energy Environ. Sci.,
2015,
8
, 2811--2824 |
2819
Table 4
Triple junction (3J) tandem PEC cell water-splitting demonstrations with at least one semiconductor–liquid junction (SLJ) listed in reverse chron
ological order. The best-in-class reported
efficiency is displayed in bold typeface and the best-in-class reported stability is italicized
Publication
institute(s)
Device structure
Electrolyte and illuminationconditions
STH
efficiency (%)
Stability, notes
Walczak et al.
85
(2015)
LBNL, Caltech, JCAP
c-Si(pn
+
)//c-Si(p
+
n)//WO
3
SLJ
monolithic PV
Integrated Pt HER catalyst with TiO
2
passivation layer
1 M HClO
4
Membrane Nafion XL
200 mW cm

2
0.24
20 hours
stable H
2
generation
tested to 48 hours
Han
et al.
60
(2014)
Delft University, HZB Berlin
a-Si:H(pin)//nc-Si:H(pin)//BiVO
4
(n) SLJ
micromorph Si monolithic
wired Pt HER catalyst
integrated CoPi OER catalyst
pH 7
co-evolved products
100 mW cm
5.2
1hour
o
5% current loss
Abdi
et al.
59
(2013)
Delft University, HZB Berlin
a-Si:H(pin)//a-Si:H(pin)//BiVO
4
(n) SLJ
a-Si monolithic
wired Pt HER catalyst
integrated CoPi OER catalyst
pH 7.3
co-evolved products
100 mW cm

2
4.9
1 hour
stable photocurrent
Gaillard
et al.
86
(2010)
U. Hawaii
a-Si(pin)//a-Si(pin)//WO
3
(n) SLJ
monolithic PV
wired Pt HER catalyst
0.33 M H
3
PO
4
co-evolved products
100 mW cm

2
3.0
Not reported
Zhu
et al.
87
(2010)
U. Hawaii, MV Systems, NREL
SLJ a-SiC:H(ip)//a-Si(pin)//a-Si(pin)
monolithic PV
wired RuO
2
OER catalyst
pH 2
sulphamic acid solution with added
potassium biphthalate
co-evolved products
1.6
Not reported for
integrated device
Brillet
et al.
88
(2010)
EPFL
DSSC//DSSC//Fe
2
O
3
(n) SLJ
wired PV
wired Pt HER catalyst
1 M NaOH, pH 13.6
co-evolved products
100 mW cm

2
1.36
Not reported
Miller
et al.
89
(2005)
U. Hawaii
a-Si(pin)//a-Si(pin)//WO
3
(n)
SLJ monolithic PV
wired Pt HER catalyst
1NH
3
PO
4
co-evolved products
100 mW cm

2
0.7
10 hours
stable H
2
production
Energy & Environmental Science
Review
Published on 24 March 2015. Downloaded on 29/10/2015 17:25:58.
View Article Online