Chapter 4

Downloaded by University of Lancaster on 17/01/2015 21:41:17.
Published on 02 October 2013 on http://pubs.rsc.org | doi:10.1039/9781849737739-00083
CHAPTER 4
Tandem Photoelectrochemical
Cells for Water Splitting
KEVIN SIVULA* AND MICHAEL GRATZEL
Institute of Chemical Sciences and Engineering, Ecole Polytechnique
Federale de Lausanne, 1015-Lausanne, Switzerland
*Email: kevin.sivula@ep.ch
4.1 Introduction and Motivation for Using Multiphoton

Systems
The realization of an energy economy based on solar hydrogen is contingent on
identifying a scalable system for solar H2 production at costs commensurate
with the price of H2 generated from conventional sources (23 US$ kg1 for the
steam reforming of natural gas).1 Thus, a balance between minimizing system
complexity and maximizing energy conversion eciency and device longevity
must be considered for any practical photoelectrochemical (PEC) water splitting system for solar H2 production. The most straightforward approach for
solar H2 production using PEC water splitting employs a single semiconductor
photoelectrode in combination with a metal counter electrode. Here, hydrogen
can be produced from water upon the absorption of two photons (under the
condition that one electron-hole pair is produced for each photon absorbed)
since the water reduction reaction requires two electrons to make one molecule
of H2. This single-absorber, two-photon approach (more simply called S2)
appears quite promising at rst glance. Since only 1.23 V are needed for water
splitting under standard conditions from a thermodynamic standpoint, one
could believe that a semiconductor with a minimum band-gap, Eg, of 1.23 eV
RSC Energy and Environment Series No. 9
Photoelectrochemical Water Splitting: Materials, Processes and Architectures
Edited by Hans-Joachim Lewerenz and Laurence Peter
r The Royal Society of Chemistry 2013
Published by the Royal Society of Chemistry, www.rsc.org
83
View Online

84
Chapter 4
(an absorption wavelength cut-o of 1008 nm) could be eective. Based on the
standard AM1.5 G solar spectrum (1000 W m2), a semiconductor with such a
band-gap would operate at maximum overall solar-to-hydrogen conversion
eciency, ZSTH, of 47.4% assuming a unity quantum conversion eciency
(fconv 1) and no other losses.2
However, signicant loss processes are involved and are unavoidable in any
energy conversion process. Figure 4.1 shows an electron energy scheme of S2
PEC water splitting using a photoanode and highlights these loss processes.
Firstly, there is an entropic energy loss resulting from the amount of work
that must be done to extract the excited states. The chemical potential of the
excited state, Dmex, represents the maximum amount of energy available to do
work,3 and is necessarily signicantly less than Eg. The dierence between these
quantities depends on the band-gap of the semiconductor and the illumination
intensity; it has been calculated by Bolton et al.4 to be about Eg/4 for a typical
semiconductor at terrestrial solar intensities, meaning that DmexE0.75 Eg. In
addition, kinetic overpotentials, Zox and Zred, exist at both electrodes resulting
from the energies of the intermediate species involved in the complex water
reduction and oxidation reaction mechanisms. Even using the best catalysts,5
the overpotential of the water oxidation reaction, Zox, is notoriously large
(0.20.4 eV) for reasonable current densities (ca. 10 mA cm2) as a consequence
of the four-electron nature of the oxygen evolution reaction (OER)
These loss processes imply that a semiconductor with Eg much greater than
1.23 eV must be used to realize S2 PEC water splitting. This is clearly illustrated
in Figure 4.1. Since a semiconductor can only harvest photons with energy
greater than Eg, the requirement to use a material with a larger band-gap
implies a lower possible ZSTH. Indeed, while the ideal (Schottky-Queisser) solar
Figure 4.1
Electron energy scheme of S2 PEC water splitting using a photoanode.

The absorption of a photon (hn) by the semiconductor with a band-gap
(Eg) creates an electron-hole pair with free energy of Dmex. This free energy
must be greater than the energy needed for water splitting (1.23 eV) plus
the overpotential losses at both the anode and the cathode, Zox and Zred,
for the water splitting reaction to occur.
View Online

Tandem Photoelectrochemical Cells for Water Splitting
85
energy conversion limit using one absorber material is 34% for AM1.5G
irradiation, Weber and Dignam2 followed by Bolton et al.6 analyzed the upper
limit for ZSTH using S2 water splitting with realistic losses and reported
maximum values of 11.6% (using Eg 2.2) and 17% (using Eg 2.03),
respectively. The latter report additionally considered photon absorption and
collection losses and estimated that 10% should be entirely feasible with an S2
approach. In reality, the realization of a S2 PEC water splitting device with
ZSTH 10% is clearly a dicult task. A review by Walter et al.5 in 2010 summarizes decades of progress on using either a photocathode or a photoanode
for S2 water splitting. While many promising S2 photocathodes exist (e.g. GaP
and InP), their cost and stability in aqueous environments remain critical
limitations. Semiconductor oxide photoanodes oer unmatched stability for
water splitting, but, due to the large energy loss processes discussed above,
materials with band-gaps greater than 3.0 eV (e.g. SrTiO3 and KTaO3) must be
employed, limiting ZSTH to less than 2%.
It remains possible that a single semiconductor material that possesses the
optimum band-gap, conduction and valence band levels, stability and availability will be identied, and a device exploiting the S2 approach will realize
over 10% ZSTH. However, the failure to do so after decades of work has led
researchers to consider alternative approaches. This chapter introduces an
approach to PEC water splitting that increases system complexity but also
energy conversion eciency. By employing multiple light absorbers in tandem,
PEC systems can both harvest a signicant portion of the solar spectrum and
provide enough Dmex to aord water splitting at high ZSTH. In the next sections,
various approaches using multi-absorber systems will be presented, analyzed,
and discussed using examples from literature. In addition, since the primary
goal of PEC research is to develop a system that balances system complexity,
cost, and eciency, particular attention will be given to these aspects.
4.2 Strategies and Limitations of Multi-Absorber

Systems
4.2.1 Brute Force Strategies
The most obvious approach to accumulate sucient Dmex for water splitting
with semiconductor materials that can also absorb a large fraction of the solar
spectrum is to connect multiple photovoltaic (PV) cells in series. For example, a
traditional pn-junction silicon solar cell generates a voltage of 0.50.6 V at its
maximum power point under standard conditions. Thus three of these cells
connected electrically in series would create sucient voltage to split water.
This brute force PV electrolysis approach is limited by the price and
availability of PV devices and electrolyzers, which, at the time of this writing
makes the price of the H2 produced around 10 US$ kg1.1 Nevertheless, this
path has generated fair amount of interest, and overall ZSTHs up to 9.3% have
been demonstrated using single crystal silicon PV modules and high-pressure
View Online
86
Chapter 4

electrolysis. It has also been suggested that a complete optimization of these

systems could deliver H2 at a price of 4 US$ kg1.8 This price may decrease
further if novel thin-lm PV technologies are employed. For example, Dhere
and Jahagirdar9,10 have reported the use of two high voltage CuIn1xGaxS2
(GIGS) PV cells side-by-side to generate a ZSTH up to 8.8%. A major drawback
with the brute force approach in general comes from the voltage output of a pnjunction solar cell, which is strongly dependent on the illumination intensity.
This necessitates complicated switching mechanisms to ensure that the optimum number of cells is connected in series during variations in light intensity
caused by haze or cloud cover.
4.2.2 The Tandem Cell Concept

Besides using PV cells side-by-side to obtain sucient voltage for water
splitting, cells can be stacked on top of one another, in tandem, if they can
individually harvest dierent portions of the solar spectrum. The rst (top) cell,
with band-gap Eg1 absorbs photons from the sun with a wavelength smaller
than l1 hc/Eg1. Ignoring scattering events, photons with l4l1 are transmitted through to the next cell with a band-gap Eg2oEg1, and so on (see
Figure 4.2). If the cells are connected electrically in series, the total photocurrent density will be limited by the cell producing the least current, whereas
the total voltage will be the addition of all the cells used.
This tandem cell strategy not only increases the voltage provided by the
overall cell, but also signicantly enhances the upper limit of solar power
conversion (solar-to-electricity) eciency from 34% in the one absorber case to
42% (2 cells), 49% (3 cells), or 68% (for an innite number of cells) for
unconcentrated AM1.5G sunlight.11 Of course, using an innite number of cells
is not possible and in practice the addition of each layer brings technical difculties that greatly increase the complexity and cost of these devices.
Figure 4.2
Electron energy scheme of a generic tri-level tandem cell for solar energy
conversion. Here three semiconductors (Eg13) are employed as pnjunctions and connected by transparent ohmic contacts to give a nal
photopotential of Vcell.
View Online

87
Nevertheless, the precise control of materials used and device architecture

(typically triple-junction epitaxially grown III-V semiconductor devices) has
pushed the solar-to-electricity power conversion eciency, ZPC, past 40%.12
However, the high cost of producing these devices relegates their use to specic
niches (e.g. extra-terrestrial application). The use of these types of PV/PV
tandem cells for PEC water splitting is discussed in Section 4.3 and also other
chapters of this book. First, since the operation requirements and theoretical
maximum ZSTH for a water splitting PEC tandem cell are distinct from its
electricity producing (PV) counterpart, it is worthwhile to examine closely the
benets and limitations of this strategy for solar water splitting.
4.2.3 The D4 Strategy and its Potential

While very impressive photovoltaic eciencies have been demonstrated using
triple junction cells as mentioned above, the maximum possible ZPC does not
change signicantly going from 2 to 3 cells (from 42 to 49%). This brings into
question the return on the device complexity added by the third cell. In addition, gaining the Dmex necessary for water splitting should be feasible with only
two light absorbers. In fact, a two-photosystem solution is precisely what
evolved in plant photosynthesis, which uses sunlight to split water and stores
the energy in carbohydrate vectors. Finally, since the water splitting reaction
necessarily entails two separate half-reactions, it seems natural to use two light
absorbers for PEC water splitting.
In a dual-absorber approach where each absorbed photon creates one excited
electron-hole pair, four photons (two in each absorber) must be absorbed to
create one molecule of H2. Following Boltons convention,6 this is designated a
D4 approach. A simple way to accomplish D4 water splitting is to use a
separately illuminated photoanode and photocathode in a side-by-side (not
tandem) conguration. Weber and Dignam2 evaluated the potential ZSTH of
this approach and reported an upper limit of 16.6% with Eg1 Eg2 1.4 eV a
modest increase over the 11.6% predicted with an S2 approach. However, using
an integrated tandem D4 approach, 22% ZSTH was predicted to be possible
with Eg1 1.8 eV and Eg2 1.15 eV. A more extensive analysis by Bolton
plotted the maximum ZSTH with respect to dierent absorber band-gaps and
reasonable losses.3,6 The formalism presented in that work can be extended to a
generalized expression for the expected tandem D4 solar-to-hydrogen conversion eciency as a function of the chosen semiconductors and the expected
losses:
ZSTH
jmin DG0H2 Zfarad

2qES
4:1
where DG0H2 represents the standard free energy of the hydrogen produced
(lower heating value, 2.46 eV per molecule of H2), Zfarad is the Faradaic eciency for the water splitting reactions, q is the elementary charge
(1.60221019 C 1 eV V1), and ES represents the power ux from the sun
View Online

88
Chapter 4
incident on the tandem cell. For standard conditions (AM1.5G), this latter
value is 1000 W m2. The factor of 2 in the denominator represents the stoichiometric relation between one molecule of H2 and the number of electrons
needed to produce it electrochemically from H2O. Finally, jmin represents the
photocurrent density (A m2) of the D4 tandem device (i.e. the electrical current
that is owing between the two cells), which is zero if there is not sucient Dmex
generated by the tandem cell and is otherwise limited by the individual cell
producing the least amount of photocurrent, as the cells are assumed be connected in series:

jmin
0;
minj1; j2 ;
Dmex 1:23 eV Zox Zred

Dmex 4 1:23 eV Zox Zred
4:2
Here, Dmex is equal to the sum of the free energies produced by each
cell, Dmex1 Dmex2. The individual current densities produced by each cell, j1
and j2, can be expressed with respect to the known spectral irradiance from the
sun13 as a function of photon wavelength (W m2 nm1), ES(l), and the wavelength cut-os for each semiconductor (li hc/Egi, i 1,2, again setting
Eg14Eg2) as
Zli
j
ES ljconv;i l
dl; for i 1; 2
hc=l
4:3
li1
where fconv,i(l) represents the quantum conversion eciency for each semiconductor as a function of wavelength (i.e. the ratio of photocurrent density
and absorbed photon ux), and l0 represents the smallest wavelength for which
there is an appreciable photon ux from the sun (ca. 300 nm for AM1.5G solar
irradiation).
For simplicity, the above formalisms assume that each semiconductor absorbs all photons with energy hn4Eg and transmits all photons with energy
hnoEg. No reection or scattering losses are included for this ideal case. Thus
values for j1 and j2 calculated in this way represent upper bounds for real
systems. In real tandem systems, reection and scattering losses exist, but can
be addressed with device engineering. Also, in real systems, both Dmex and fconv
are complicated functions of the light intensity and the semiconductor properties. Moreover, the overpotentials for the oxidation and reduction reactions,
Zox and Zred, are related to jmin by the Tafel relation. Thus, additional
assumptions are needed to obtain reasonable values for ZSTH. In principle, fconv
can be made very close to unity even in real systems,3 so fconv 1 will be used to
evaluate equation (4.3). If we also assume that no corrosion or undesirable side
reactions occur then Zfarad 1. Bolton selected some values for the remaining
loss processes based on reasonable assumptions regarding the overpotentials
and the free energy.6 The minimum reasonable value for the energy loss,
Uloss Eg1 Eg2 Dmex Zox Zred, was chosen to be 1.2 eV (or 0.6 eV for each
View Online
89

Figure 4.3
(Top) Contour plot (thick grey lines) showing the maximum reasonable
ZSTH with AM 1.5G incident radiation (1000 W m2) and an total loss,
Uloss, set at 1.4 eV as it depends on the chosen semiconductor cut-o
wavelengths, l1 and l2 where (li hc/Egi, i 1,2, with Eg14Eg2). The
maximum ZSTH as it depends on the choice of the wider band-gap
absorber (l1) is also shown (thin black line). See text for full explanation.
(Bottom) The AM 1.5 G solar photon ux as a function of wavelength.
The optimum values of li are indicated and the areas corresponding to the
photons harvested by each absorber layer are shaded.
semiconductor). Figure 4.3 (top) shows a contour plot showing

how a reasonable upper limit for ZSTH depends on the chosen semiconductor
cut-o wavelengths, l1 and l2.The plot was obtained using equations
(4.1)(4.3) and the above assumptions with slightly higher losses
(Uloss 1.4 eV total).
The shapes of the contours (thick grey lines in Figure 4.3), which represent
values of l1 and l2 that result in the same ZSTH, are easily rationalized. First,
the contours remain to the right side of the 451 line because as this is where the
View Online

90
Chapter 4
set condition Eg14Eg2 is satised. The upper-right region, where ZSTH is

undened, represents semiconductor combinations that do not possess
sucient Dmex for water splitting in a D4 conguration given Uloss. Calculating
the upper limit of ZSTH and optimum l1, l2 pairing for an arbitrary system can
be simply done with the help of Figure 4.3 by rst choosing l1. For example,
take l1 477 nm (Eg1 2.6 eV), which corresponds to WO3, a material
commonly used as a semiconductor photoelectrode for water splitting.14 A
horizontal line drawn at l1 477 nm would intercept rst with the solid black
maximum-ZSTH line. Reading the value on the top scale gives the maximum
value of ZSTH for WO3 as 6.3%. Continuing back on the imagined horizontal
line at l1 477 nm further, it intercepts with a dashed black line running
against the contour lines. This line represents the values of l1 and l2 that meet
the current matching condition. Reading the value of l2 at that point gives
l2 561 nm and thus the minimum value for Eg2 hc/l2 2.2 eV. Choosing
larger l2 (smaller Eg2) will not result in a higher ZSTH as the photocurrent will
be limited by the top absorber, l1. For this reason the grey contour lines are all
horizontal to the right of the dashed black current matching line.
To identify the optimum values for both l1 and l2 with Uloss 1.4 eV, the
current matching line can be followed for increasing l1 and l2 (and ZSTH) until
the maximum ZSTH of 29.9 % with l1 755 and l2 1253 nm. The distribution
of the photons harvested by each absorber in this optimum case can be visualized in the bottom of Figure 4.3,where the AM 1.5G solar photon ux is
plotted as a function of the wavelength. The optimum values for l1 and l2 are
indicated, and shaded areas indicate photons harvested (lighter shading for l1).
The areas of the two shaded regions in this plot are identical, as the optimized
circumstances exactly satisfy the current matching condition. By coincidence,
the cut o for absorber 1, l1, coincides with the sharp dip in the solar spectrum
at 762 nm, which is due to the weak spin-forbidden electronic absorption of
molecular oxygen.
Higher assumed values for Uloss result in lower values for the maximum ZSTH
at the optimum conditions. For example, with Uloss (total) 1.6 eV, the optimum conditions give ZSTH 27.1 % using l1 720 and l2 1120 nm, and with
Uloss 2.0 eV, ZSTH 21.6 % using l1 655 and l2 925 nm.6 The maximum
predicted ZSTH for arbitrarily chosen l1, l2 does not change when assuming
higher values for Uloss, as long as sucient Dmex can still be generated. For
example, with Uloss 2.0, the maximum ZSTH for WO3 remains 6.3%, with the
current matching l2 remaining as 561 nm. However, in this case, l2 must also be
greater than 1914 nm to ensure that sucient Dmex is generated.
Overall the model presented in this section only gives reasonable expected
maximum values for ZSTH, which can be of use in describing the basic capability
(or limitations) of specic materials or materials combinations. This is certainly
of use in determining the interest in developing specic systems. However, the
model does not include any additional constraints on selecting the materials
with respect to the conduction and valence band energy levels, or work function; factors that add to the diculty of nding ideal and complementary
materials for a water-splitting tandem cell.
View Online
91

4.2.4 D4 Device Architectures

Beyond band-gap selection, an important practical consideration for
tandem cell construction is the types of junctions used. While photovoltaic
(pn-junction) technology is well developed for solar electricity production, it
has been reasoned that the direct photoelectrolysis of water at a semiconductorliquid junction can potentially produce more ecient solar-to-hydrogen
conversion compared PV-electrolysis device.15 For this reason, many groups
have investigated tandem cells where one or both of the absorbers are in direct
contact with the aqueous electrolyte. In addition, alternative photoelectrochemical devices, like dye sensitized solar cells (DSSCs) can also be
employed as one of absorbers in a tandem scheme. Thus, many D4 congurations are possible. Figure 4.4 shows, from an energy perspective, two classic
examples in addition to the PV/PV combination previously mentioned.
The p-type photocathode/n-type photoanode device (Figure 4.4a) is the
simplest in terms of junctions with only one interconnect between the two
semiconductors together with two semiconductor liquid junctions (SCLJs).
Here, materials must be chosen such that when the device is in contact with
the electrolyte, CB2oqE0(H /H2)oCB1oVB2oqE0(H2O/O2)oVB1, where
VBi and CBi represent the valence band and conduction band energy levels of
semiconductor i, respectively. A similar situation is necessary for a photoelectrode/PV conguration (Figure 4.4b) with the added complication of
aligning an additional n-type energy level and the work function of the metal
electrode used to perform the reduction reaction (this is usually necessary as the
semiconductors employed for PV cells are not stable when performing the
water splitting reaction).
Despite the constraints on both the materials band-gap and energy levels,
various material combinations have been demonstrated in D4 congurations.
Specic examples can be categorized as photocathode/photoanode, photoelectrode/PV, or PV/PV. Specic examples of these congurations are explored
in the next sections.
4.3 PV/PV Strategies

Devices employing two or more pn-junctions in tandem have attained the
highest reported ZSTH.16,17 Champion devices use the well-known band-gap
engineering of III-V semiconductor systems to optimize light harvesting and
Dmex. In particular, Licht et al. have employed an AlGaAs/Si (pn-pn) structure
using Pt-black and RuO2, as reduction and oxidation catalysts, respectively,
and obtained solar-to-hydrogen conversion eciencies as high as 18.3% under
simulated AM0 sunlight (135 mW cm2).18,19 Turner and co-workers have investigated monolithic GaAs/GaInP2 (pn-p, pn-pn, or pn-pn-p) systems and
attained ZSTH up to 12.5%.2022 While the conversion eciencies reported with
these systems are quite impressive, major concerns exist about the price (due to
the requirements of high material purity and costly fabrication methods) and
stability of these devices when they are used in contact with aqueous
View Online
Chapter 4

92
Figure 4.4
Electron energy schemes of two possible D4 tandem water splitting device

congurations: (a) shows an n-type photoanode connected to p-type
photocathode and (b) shows a n-type photoanode with a pn-junction.
electrolyte.23 A comprehensive overview of these III-V systems will not be

presented here as other sections of this book are dedicated to that purpose.
Of course, it is not necessary to construct an all-PV tandem system from
III-V semiconductors; many other systems have also been reported. One system
of particular interest is based on amorphous silicon (a-Si). Bockris and coworkers rst showed that a triple stack of n-type/intrinsic/p-type (nip) a-Si (not
a D4 system, as 6 photons need to be absorbed for one H2) on a Ti substrate
could give up to ZSTH 7.5% under AM1 (100 mW cm2) illumination when
View Online

93
islands of Pt and RuO2 were used as a reduction and oxidation catalysts, and
the device was directly submerged into aqueous electrolyte.24 Since the bandgap of the a-Si was the same for each nip-junction, the layer thicknesses were
optimized so each layer produced the same photocurrent (i.e. the top two layers
were thin enough to transmit some photons with energy hn4Eg). This concept
was later extended to increase the eciency for electricity production by mixing
some Ge in with the Si to modify the Eg of the bottom two layers.25 Using these
devices for water splitting, an ZSTH of 7.8% under AM1.5 (100 mW cm2) illumination was achieved in 1 M KOH electrolyte with CoMo (reduction) and
NiFeyOx (oxidation) catalysts.26 The stability of these devices was improved to
greater than 31 days while maintaining 56% ZSTH using a uorine-doped SnO2
protection layer on the cathode.27 More recent eorts with these devices have
also shown that operation at near neutral pH is possible with ZSTH 4.7%.28
While these a-Si based devices are presumably less expensive than III-V
semiconductor based tandem cells, their fabrication still requires relatively
expensive vacuum processing. Despite this, at the time of this writing, the mass
production by several companies of triple junction a-Si device modules for
photovoltaic energy conversion is beginning. It is not yet clear what the price
per kg H2 would be if these cells were used for water splitting. Furthermore, the
open-circuit voltages produced by such triple junction a-Si cells are around
2.4 V much higher than is needed for water electrolysis. The optimization of
this class of devices specically for solar water splitting has yet to be
accomplished.
4.4 Photoelectrode/PV Systems

The most widely recognized photoelectrode/PV device is a monolithic, epitaxially grown III-V semiconductor system developed by Turner with a GaAs
pn-junction coupled to a GaInP2 photocathode.22 While the performance of
this system was impressive at ZSTH 12.5%, the GaInP2 photocathode quickly
corrodes when in contact with the aqueous electrolyte.23 In general, this is a
major drawback of using direct semiconductor-liquid junctions. However, with
certain semiconductors the photoelectrode/PV system has many advantages.
4.4.1 Advantages and Operation of a Photoanode/PV Tandem

Cell
Identifying semiconductor materials that form stable junctions with water
under illumination has been a major challenge in the eld of PEC water
splitting. Transition metal oxide semiconductors have, however, excelled in this
aspect, with the prototype being TiO2.29,30 Unfortunately, a conundrum exists
for transition metal oxides with respect to their band-edge energy levels and
band-gap. The conduction band-edge energy is strongly inuenced by the metal
d-orbitals and varies from metal to metal. However, the valence band edge
energy of a transition metal oxide is most strongly inuenced by oxygen 2p
View Online

94
Chapter 4
orbitals, thus it does not change appreciably from a potential around 2.53.0 V
vs. NHE when changing the metal.31 In addition, the band-edge energies
typically exhibit a Nertsian pH response. Thus, for a transition metal oxide to
have a small enough Eg for adequate solar light harvesting its CB energy must
be well below the water reduction potential. This excludes them from being
used in a S2 water splitting scheme but essentially makes them ideal for use as
photoanodes in a D4 tandem cell. Moreover, the diculty in identifying a
stable p-type cathode makes a photoanode/PV tandem device a good compromise, with device complexity and stability.
Since there are several examples of this type of tandem cell, it is useful to illustrate general device function. The operation conditions of a photoanode/PV
tandem device can be predicted by comparing the current density-voltage characteristics of each individual cell under the actual illumination conditions of that
separate part (given its position in the particular tandem conguration).2
Figure 4.5 shows the current density with respect to the applied potential for a
realistic but arbitrary photovoltaic cell (solid black curve) and a photoanode
(solid grey curve) under the appropriate illumination conditions. The operating
potential Uop is dened as the potential measured at a common node between the
PV and the PEC cell (i.e. where the curves overlap). The current density at this
point, jop, can be used directly in equation (4.1) as jmin to predict the ZSTH of the
tandem cell. In this way the properties of the individual cell can be independently
varied to maximize jop and, accordingly, the ZSTH of the device.
Figure 4.5
The operation conditions of a photoanode/PV tandem device as shown by

the idealized current-voltage curves for a PV cell (black) on top of a n-type
photoanode (grey) in the dark (broken lines) and under illumination
(solid lines). Relevant parameters of the curves are indicated and described
in the text.
Reprinted with permission from J. Brillet, et al., J. Mater. Res., 2010, 25,
1724. Copyright 2010 Materials Research Society.
View Online

95
It should be noted that, in a real device, the maximum jop is not necessarily
jmin as expressed in equation (4.2). That is to say that the maximum jop is not
found by merely maximizing the current output from each cell (denoted as the
short circuit current density, jsc, for the PV cell and the plateau photocurrent
density, jpl, for the photoanode). Uloss (as dened in section 4.2.3) and other
device non-idealities (such as high series and low shunt resistances) necessitate
that the characteristic cell potentials (denoted as the open circuit photopotential, Uoc, and the photocurrent onset potential, Uon, for the PV and the
photoanode, respectively) be also considered in device optimization.
4.4.2 Photoanode/PV Examples

The possibility of a photoanode/PV tandem device to employ a stable transition metal oxide photoanode and a complimentary PV device has attracted
many research groups to investigate various promising systems. For example, a
TiO2 photoanode in tandem with a thin lm PV device based on Cu(In,Ga)Se2/
CdS produced hydrogen at a rate of 0.052 mL cm2 s1 during unassisted solar
water splitting (corresponding to an external quantum eciency of 1.02%).32
While less interest has been paid to using the prototypical TiO2 as a photoanode due to its large band-gap (Eg 3.2 eV, max ZSTH 2%), this work
demonstrated the importance of using optimized protective layers
(Nb0.03Ti0.97O1.84 in this case) to eliminate corrosion of the PV cell in the
aqueous conditions.
Further research eorts have focused on using more promising transition
metal oxides such as WO3, and Fe2O3. As mentioned in section 4.2, WO3 can
potentially convert up to 6.3% of the energy of AM1.5G sunlight into
hydrogen. Hematite (a-Fe2O3) is a more promising transition metal oxide with
a band gap of 2.1 eV. This corresponds to a maximum ZSTH of 15.4% according
to Figure 4.3, if a second absorber with l24774 nm (Eg2o1.6 eV) is used. Besides its relatively small band-gap, a-Fe2O3 is the most stable form of iron oxide
at ambient conditions, and both iron and oxygen are ubiquitous atoms in the
earths upper crust, making hematite an outstanding candidate for solar energy
conversion on a scale commensurate with the global energy demand. Furthermore, both WO3 and Fe2O3 photoanodes can be prepared by inexpensive
solution-based techniques.33
Miller and co-workers have investigated combining tungsten34 or iron35
oxide photoanodes in tandem with a-Si:Ge PV. As a drawback, the large
overpotential for water oxidation (Zox) and the relatively low Uoc of a a-Si
device requires a double junction PV in tandem to provide sucient potential
to drive the overall water splitting reaction (similar to the PV only case where 3
pn-junctions were needed with a-Si). Despite this, 3% ZSTH was obtained with
the WO3/a-Si:Ge/a-Si:Ge device.34 A similar device for iron oxide with only one
PV (Fe2O3/a-Si:Ge) did not split water without an external bias, but it was
shown that a 0.65 V bias savings was earned under AM1.5G illumination.35
This result could reasonably be improved with the recent advances in a-Fe2O3
photoanode performance.36
View Online
96
Chapter 4

4.4.3 The Photoanode/DSSC Tandem Cell

From a practical perspective, the attractive aspects of using a widely available,
highly stable and inexpensively produced photoanode are diminished when
using a tandem component that requires relatively expensive processing techniques (e.g. the chemical vapour deposition of a-Si). Thus, more recently, investigations have focused on using next-generation photovoltaics that can also
be fabricated with inexpensive, solution processed methods. The dye sensitized
solar cell (DSSC) is the prototype example37 of this class of photovoltaic device
and thus has attracted signicant attention for use in solar water splitting
tandem cells with a stable photoanodes.38 The photoanode/DSSC combination
was rst suggested by Augustynski and Gratzel39 with WO3 as the photoanode
suggesting that device would be capable of 4.5% ZSTH given the performance
of the two devices.40 In practice, tandem devices with WO3 have been constructed by Park and Bard41 and Arakawa et al.42 giving ZSTHs up to 2.8% at
AM1.5G (100 mW cm2) using Pt as the cathode. However, similar to the a-Si
based devices, two DSSCs connected in series to the photoanode were necessary
to achieve overall water splitting. This was accomplished by positioning these
two DSSCs side-by-side behind the WO3 photoanode.
This photoanode/2DSSC architecture does not fundamentally provide a
limitation to the possible solar-to-hydrogen conversion eciency for WO3 or
even the more promising Fe2O3, as less than one third of the available solar
photons have wavelengths shorter than 600 nm, and pan-chromatic dyes with
high quantum eciency extending beyond 900 nm are being developed.43
However, the architecture does present a challenge to device construction as
the two DSCs need to be constructed each with half of the active area of the
photoanode in order to normalize the total area of the device.
The development of new dyes for the DSSC during the past decade has
initiated work to explore alternative device architectures for optimizing light
harvesting in water splitting tandem cells. For example, all-organic dyes, such
as the squaraine dyes, which have a narrow absorption bandwidth extending
into the far red region of the visible, have demonstrated solar power conversion
eciencies over 6% under AM 1.5G illumination.44,45 Specically, the dye
coded SQ1 only absorbs light with wavelengths between 550 and 700 nm. In
addition to reducing costs considerably compared to a ruthenium-based dye,
Brillet et al. pointed out that two new distinct tandem cell congurations become accessible when using this type of dye.46 The possible congurations are
shown in Figure 4.6. Besides the benchmark photoanode/2DSSC design
(Figure 4.6a), a true tri-level device (photoanode/DSSC-squaraine dye/DSSCpanchomatic dye) would also be eective (Figure 4.6b). This conguration
would eliminate the need to construct two DSSCs side-by-side. A second
possibility is to arrange two side-by-side DSSCs in front of a hematite
photoanode (Figure 4.6c). This front DSSC conguration is particularly
attractive in view of light harvesting, as transition metal oxides (in particular
hematite) have high indexes of refraction, which increase reection. Moreover,
the front DSSC approach eliminates the need to deposit the photoanode on a
View Online
97

Figure 4.6
The layouts of three architectures for water splitting tandem cells using a
n-type photoanode and two dye sensitized solar cells in series. (a) the
conventional back DSSC conguration, (b) a trilevel conguration,
and (c) the front DSSC conguration.
View Online

98
Chapter 4
transparent conducting glass and an inexpensive metal foil support could instead be used.
Brillet et al. determined how these two new photoanode/DSC tandem concepts performed compared to standard architecture using state-of-the-art
nanostructured hematite47 photoanodes. The tri-level tandem architecture
(photoanode/DSSC-squaraine dye/DSSC-panchromatic dye) produced the
highest operating current density and thus the highest expected solarto-hydrogen eciency of 1.36%. The conventional photoanode/2DSSC
architecture gave an eciency of 1.16% and the front DSSC conguration
0.76%. It should be noted that these values were far below the expected 3.3%
that should have been possible with the nanostructured hematite photoanodes
used. Evidently, reduced light harvesting caused by scattering and reection
were limiting the overall conversion eciency. This is illustrated by the optical
and electronic characteristics of the tri-level tandem cell presented in Figure 4.7.
Here it is shown that while the hematite photoanode only produces photocurrent for l4600 nm, the light transmitted to the middle and back cells is
severely attenuated by scattering and reection losses. This causes the DSSC
photocurrent to limit the overall performance of the tandem cell. While device
engineering should improve the operating current density by reducing reection, scattering and resistive losses, DSSCs with higher Uocs would enable
construction of a true D4 photoanode/DSSC water splitting tandem cell.
Indeed, in 2012, a specically designed cobalt redox couple combined with an
all-organic dye (coded Y123) gave DSSCs with Uoc41.0 V at 1 sun conditions.48 This breakthrough allowed the rst demonstration of a D4 photoanode/DSSC water splitting tandem cell.49 Devices were assembled with both
nanostructured WO3 and Fe2O3 photoanodes, and jop was measured to give
ZSTH values of 3.10% and 1.17% with the WO3/DSSC and Fe2O3/DSSC
combinations, respectively. An optical analysis also compared the predicted
photocurrent from the integration of IPCE measurements and the actual j-U
behavior of the device measured in situ. This optical analysis and the j-U curves
for each device are shown in Figure 4.8. In the case of the Fe2O3/DSSC tandem
cell, the jop is far from the plateau region of the hematite electrode photocurrent. This results in a performance far from the maximum obtainable. The
limitation of this system is clearly the late onset of the photocurrent in the
photoelectrode, despite the use of state-of-the-art strategies to shift the onset of
the photocurrent to less positive potentials by means of surface catalysis and
passivation. However, in the case of the WO3/DSSC tandem cell, the photocurrent onset is not a limitation and the jop is very close to the plateau region of
the photoanode (approx. 1000 mV). The limiting factor in this case is the low
photocurrent obtainable by the photoanode due to less than ideal absorption
capability of tungsten trioxide in the visible region of the solar spectrum.
Despite this, the Fe2O3/DSSC device still exhibited a near unity faradaic eciency and good stability over an eight-hour testing period. Overall this work
suggests that the low ZSTH of the D4 hematite/DSSC tandem cell oers the
larger room for improvement, in particular, further reduction of the overpotential for water oxidation.
View Online
99

Figure 4.7
The optical and electronic characteristics of the trilevel tandem cell.

Top panel: The solid line shows the incident solar ux. The IPCE of the
hematite photoanode employed (dashed line) is convoluted with the solar
ux to give the electron ux in the photoanode (shaded area). Middle
panel: The solar ux is convoluted with the transmittance of the hematite
(dotted line) to give the incident irradiance on the squaraine DSC (solid
bold line). The irradiance is convoluted with the IPCE of the squaraine
DSC (dashed line) to give the electron ux in the intermediate photovoltaic device (shaded area). Bottom panel: The solar ux is convoluted
with the transmittance of the squaraine dye plus hematite (dotted line) to
give the incident irradiance on the black DSC (solid bold line). The
irradiance is convoluted with the IPCE of the black DSC (dashed line)
to give the electron ux in the bottom photovoltaic device (shaded area).
View Online

100
Figure 4.8
Chapter 4
Spectral response study of the (a) WO3 (light grey)/DSC (dark grey) and
(c) Fe2O3 (light grey)/DSC (dark grey) D4 tandem cells. The transmittance
of the photoanode (dashed lines) convoluted to the photon ux on the
photoanode (thin black line) allows the calculation of the photon ux on
the DSC (thin grey line). The shaded areas represent the electron density
calculated from the IPCE (solid lines) convoluted to the photon ux on
each device. j-U curves of (b) WO3 (light grey)/DSC (dark grey) and (d)
Fe2O3 (light grey)/DSC (dark grey) tandem cells under AM 1.5 G irradiation. The solid lines refer to the linear voltage sweep predicted from the
IPCE integration and the triangles refer to in situ measurements.
Reprinted from reference 49.
4.5 Photoanode/Photocathode Systems

As mentioned in Section 4.2, the most straightforward and simple way to
construct a D4 tandem water splitting cell is to use an n-type semiconductor
photoanode and a p-type semiconductor cathode. However, this approach is
the least developed of the water splitting systems found in the literature. The
reason for this is clearly due to a lack of suitable photocathode materials for
water reduction. Many materials, for example GaInP2,50 Si,51,52 SiC,53 WS2,54
Cu(In,Ga)Se2,55Cu2O,56 CuYO2,57 CaFe2O4,58 and even p-type (Mg21 doped)
Fe2O359 have been investigated as water reducing p-type electrodes, but the
magnitude of the photocurrent or the stability in aqueous solutions have
remained limiting factors. Despite this a few eorts have been made to
demonstrate D4 photoanode/photocathode tandem cells.
Early work by Nozik60 established the general theory for combining
photoanodes and photocathodes into tandem cells and introduced the n-TiO2/
p-GaP system. Ohmic contacts between single crystals of n-TiO2 and p-GaP
View Online

101
(Eg 2.26 eV) gave a tandem cell that was found to evolve both hydrogen and
oxygen without an applied potential. A high internal resistance limited the
conversion eciency of the cell for H2 evolution, which was calculated to be
0.25% at zero bias (based on a total electrode area of 1.9 cm2 and 85 mW cm2
of net incident simulated sunlight). An oxide layer forming on the surface of the
p-GaP was the likely cause of the device instability.61
Very little research attention was given to constructing photoanode/photocathode tandem devices for many years after this seminal demonstration. More
recently, following the advances in oxide photoanode performance, additional
demonstrations of photoanode/photocathode tandem devices have appeared.
For example, while also known to be unstable in aqueous systems,23 GaInP2
(Eg 1.83 eV) photocathodes have been combined with either WO350 or
Fe2O362 photoanodes by Wang and Turner. Under intense white light
illumination (4200 mW cm2) the WO3/GaInP2 combination produced a
detectable photocurrent that rose linearly with light intensity to reach
jop 20 mA cm2 at 1000 mW cm2. Due to the insucient potential dierence,
the system did not function at illumination intensities below 200 mW/cm2 (recall that the minority carrier quasi Fermi level and thus the Dmex of each
electrode should change proportionally with the logarithm of the illumination
intensity). For the case of Fe2O3/GaInP2, negligible photocurrent was observed
even at 10 sun illumination due to the mismatch of the conduction band
minimum of the spray-pyrolyzed Fe2O3 thin lm and the valence band maximum in the GaInP2. Employing surface dipoles to raise the conduction band63
of Fe2O3 may be useful for this combination of materials, which is clearly not
ideal. Moreover the limited availability of indium in the earths crust prevents
the application of this material on a global scale.
A more novel p-type material made from abundant elements, CaFe2O4
(Eg 1.9 eV), has been paired with n-TiO2 (in a side-by-side conguration) to
give a device operating at jop 110 mA cm2 in 0.1 M NaOH with the light from
a 500 W Xe lamp.64 However, the Faradaic eciency for water splitting was
found to be only 12%, and Fe and Ca were detected in the electrolyte after the
device test.
Nanostructuring techniques have also been employed to enhance the performance of inexpensive electrode materials for photoanode/photocathode
tandem cells. Grimes and co-workers used an anodization technique to oxidize
Cu-Ti lms to obtain p-type nanotubular Cu-Ti-O lms (1000 nm length, 65 nm
pore diameter, 35 nm wall thickness).65 The lms were primarily CuO
(Eg 1.4 eV) with a Cu4Ti3 impurity phase detected. These lms (on transparent F:SnO2 substrates) were placed in tandem with nanotubular TiO2 to
give a working tandem device (see Figure 4.9). A jop 0.25 mA cm2 (ZSTH
around 0.30%) under standard illumination conditions and a reasonable
stability in the minutes time scale were observed (however, photocurrent was
negligible after 5 hours). It should be noted that the TiO2 was exposed to 1 M
KOH while the Cu-Ti-O was exposed to 0.1 M Na2HPO4 and the electrolyte
compartments were connected by a salt bridge. The dissimilar pH values lead to
a favourable chemical bias equivalent to about 0.4 V in this case. Exposing the
View Online

102
Figure 4.9
Chapter 4
(Top) Illustration of a D4 tandem cell comprised of n-type TiO2 and

p-type Cu-Ti-O nanotube array lms, with their substrates connected
through an ohmic contact. (Bottom) Photocurrent from the D4 tandem
cell under global AM 1.5 illumination. Light is incident upon the oxygen
evolving TiO2 side of the diode, with the visible portion of the spectrum
passing to the Cu-Ti-O side. The n-TiO2 side of the device is kept
immersed in a 1 M KOH aqueous solution, the p-Cu-Ti-O side is kept
in 0.1 M Na2HPO4 with a salt bridge linking the two sides solution.
Reprinted with permission from G. K. Mor, et al., Nano Lett., 2008, 8,
19061911. Copyright 2008 American Chemical Society.
Cu-Ti-O to the KOH caused the rapid decay of photocurrent as the CuO was
reduced to copper. Recent eorts to stabilize p-type photocathodes using
overlayers deposited via atomic layer deposition may improve the performance
of this and many other photocathode materials.66
4.6 Practical Device Design Considerations

Regardless of the nature of the combination of devices chosen to make a
tandem cell for solar water splitting, consideration must be given to how the
device will be illuminated and how the evolved gases will be collected. This
View Online

103
latter point is especially important, given that the stoichiometric combination

of H2 and O2 produced during water splitting is highly explosive when mixed.
Thus, eective gas separation is necessary. Fortunately, the intrinsic conguration of water splitting tandem cell one side for water oxidation and
another side for water reduction facilitates the separate collection of gases.
A simple device has been described which uses 2 electrolyte compartments with
identical compositions separated by a glass frit or membrane for ionic equilibration and to prevent reverse water splitting reaction from occurring.67 An
idealized variation of this device is shown in Figure 4.10a. Here, gravity is the
driving force for separating the bubbles of gas forming on the electrodes
surface and the remaining electrolyte. However, this force is maximized when
the photoelectrodes are vertical, which requires the illumination direction to be
perpendicular to the gravitational force. Without using mirrors, this illumination condition is only satised at sunrise or sunset. Thus, additional device
considerations are necessary. For example, the device could simply be placed
with the photoanode facing in the direction of the Earths equator and at an
angle optimized to maximize insolation while still providing sucient buoyancy
Figure 4.10
(Top) A simple schematic of a photoanode/photocathode tandem cell

with gas collection. (Bottom) Conceptual design of a large-scale reactor
with integrated tandem cells and owing electrolyte to facilitate gas
collection.
The bottom gure is adapted with permission from E. L. Miller, et al.,
Int. J. Hydrogen Energy, 2003, 28, 615623. Copyright 2003 Elsevier.
View Online

104
Chapter 4
force. At large latitudes (north or south), this would allow for near optimal
illumination during the whole day while also permitting the gravity collection
of the gases. Since the illumination direction near the equator is parallel to the
gravitational force, the cells should be oriented at with respect to the earths
surface to maximize insolation. This leads to diculties with bubble accumulation at the surface of the bottom cell which would reduce the performance of
the tandem cell by lowering the surface area of the electrode exposed to the
electrolyte.
Relying on gravity to separate the evolved gases from the electrolyte also
requires the bubbles to reach a certain size before buoyancy forces become
larger than surface forces and they detach from the photoelectrode surface in
any orientation. Thus to overcome any potential losses in the large scale implementation of the tandem cell, a trough-type ow design has been proposed
(Figure 4.10b). Here two glass frits (or membranes) are used to prevent mixing
of the two electrolyte compartments, and the electrolyte is pumped though the
two compartments.68 The owing electrolyte adds an extra force to shear the
bubbles o the surface, which are then carried along with the electrolyte to
separating tanks where the gases are collected at ambient pressure. This type of
system would allow the tandem cell to be at any angle, enabling the optimum
insolation at any latitude or by using solar tracking without any concern of
bubbles masking the active area of either electrode.
In addition to these relatively simple methods of tandem cell device implementation, many other systems have been proposed for dierent operation
scales, and even with concentrated sunlight.68,69 A common issue addressed in
tandem cell device design is implementing separate functional layers for catalysis or protection. Of course these layers increase the cost of the system, but
they are necessary when employing unstable materials or those with large
overpotentials. The potential cost of producing the tandem device and its
longevity are equally important as its overall solar-to-hydrogen eciency. Is
has been estimated that a device lifetime of 10 years, a cost of 100 US$ m2, and
a ZSTH 10% are necessary to produce hydrogen by PEC at a cost comparable
to that of steam reforming of methanol (H2 at a price of about 5 US$ kg1).70
Thus reducing device complexity and employing inexpensive fabrication
methods while employing stable materials are the main issues for the practical
implementation of the solar water splitting tandem cell.
4.7 Summary and Outlook

In this chapter, the motivation and theoretical framework for the PEC water
splitting tandem cell were examined. While a single perfect material (in an S2
scheme) could reasonably convert 10% of the incident solar irradiation to
chemical energy stored in hydrogen, that single perfect material has not been
found. An integrated tandem approach (D4 scheme) could reasonably convert
over 20% of the suns energy (even with large assumed losses) and is more
exible regarding material choice. Many dierent systems have been investigated using various combinations of photovoltaic cells and photoelectrodes.
View Online

105
In order to be economically competitive with simple brute force strategies or

the production of H2 from fossil fuels, a practical water splitting tandem cell
must optimize cost, longevity and performance. Due to the lack of stable
photocathode materials, a promising approach to meet the cost and performance targets is to use a stable photoanode material in tandem with an inexpensive PV cell like the dye sensitized solar cell (DSSC). Promising
photoanode materials are stable transition metal oxides exemplied by
hematite (a-Fe2O3), which has vast potential given its band-gap, energy levels,
abundance, and stability. The successful demonstration of a D4 Fe2O3 DSSC
tandem cell with an solar-to-hydrogen eciency of 1.17%, demonstrates the
convincing promise of this device, but much more research is needed to realize a
D4 tandem device with an ZSTH410%. Further reducing the overpotential for
the oxygen evolution reaction (by passivating surface traps and adding catalysts)71,72 and increasing the photocurrent (by nanostructuring)7375 are, at the
time of writing, ongoing research topics for hematite and other encouraging
photoanode materials. Device complexity and cost would also drop dramatically if stable photocathode materials with suitable band-gap energies for
employment with hematite were developed. Other chapters of this book are
dedicated to describing approaches to improving the performance of materials
like hematite and to nding new materials that will someday be employed in an
ecient, inexpensive and stable tandem device for the overall conversion and
storage of the suns energy into molecular hydrogen.
Acknowledgement
The authors acknowledge the Swiss Federal Oce of Energy (Project number
102326, PECHouse) for supporting our research and development of the PEC
tandem cell.
References
1. N. R. C. Committee on Alternatives and Strategies for Future Hydrogen
Production and Use, National Academy of Engineering, The Hydrogen
Economy:Opportunities, Costs, Barriers, and R&D Needs, National Academies Press, Washington, 2004.
2. M. F. Weber and M. J. Dignam, J. Electrochem. Soc., 1984, 131, 1258
1265.
3. J. R. Bolton, Solar Energy, 1996, 57, 3750.
4. J. R. Bolton, A. F. Haught and R. T. Ross, in Photoelectrochemical
Conversion and Storage of Solar Energy, J S Connolly, Academic New York,
1981, pp. 297339.
5. M. G. Walter, E. L. Warren, J. R. McKone, S. W. Boettcher, Q. Mi, E. A.
Santori and N. S. Lewis, Chem. Rev. (Washington, DC, U. S.), 2010, 110,
64466473.
6. J. R. Bolton, S. J. Strickler and J. S. Connolly, Nature, 1985, 316, 495500.
View Online

106
Chapter 4
7. N. Kelly, T. Gibson and D. Ouwerkerk, Int. J. Hydrogen Energy, 2008, 33,

27472764.
8. T. L. Gibson and N. A. Kelly, Int. J. Hydrogen Energy, 2008, 33, 5931
5940.
9. N. G. Dhere and A. H. Jahagirdar, Thin Solid Films, 2005, 480481, 462
465.
10. A. H. Jahagirdar and N. G. Dhere, Sol. Energy Mater. Sol. Cells, 2007, 91,
14881491.
11. A. D. Vos, J. Phys. D: Appl. Phys, 1980, 13, 839.
12. H. Cotal, C. Fetzer, J. Boisvert, G. Kinsey, R. King, P. Hebert, H. Yoon
and N. Karam, Energy Environ. Sci, 2009, 2, 174.
13. ASTM-International, West Conshohocken, PA, 2008.
14. C. Santato, M. Ulmann and J. Augustynski, J. Phys. Chem. B, 2001, 105,
936940.
15. A. J. Bard and M. A. Fox, Acc. Chem. Res., 1995, 28, 141145.
16. S. Licht, J. Phys. Chem. B, 2001, 105, 62816294.
17. O. Khaselev, A. Bansal and J. A. Turner, Int. J. Hydrogen Energy, 2001,
26, 127132.
18. S. Licht, B. Wang, S. Mukerji, T. Soga, M. Umeno and H. Tributsch,
J. Phys. Chem. B, 2000, 104, 89208924.
19. S. Licht, B. Wang, S. Mukerji, T. Soga, M. Umeno and H. Tributsch, Int.
J. Hydrogen Energy, 2001, 26, 653659.
20. X. Gao, S. Kocha, A. J. Frank and J. A. Turner, Int. J. Hydrogen Energy,
1999, 24, 319325.
21. S. S. Kocha, D. Montgomery, M. W. Peterson and J. A. Turner, Sol.
Energy Mater. Sol. Cells, 1998, 52, 389397.
22. O. Khaselev and J. A. Turner, Science, 1998, 280, 425427.
23. O. Khaselev and J. A. Turner, J. Electrochem. Soc., 1998, 145, 33353339.
24. G. H. Lin, M. Kapur, R. C. Kainthla and J. O. M. Bockris, Appl. Phys.
Lett., 1989, 55, 386387.
25. X. Deng, X. Liao, S. Han, H. Povolny and P. Agarwal, Sol. Energy Mater.
Sol. Cells, 2000, 62, 8995.
26. R. E. Rocheleau, E. L. Miller and A. Misra, Energy Fuels, 1998, 12, 310.
27. N. A. Kelly and T. L. Gibson, Int. J. Hydrogen Energy, 2006, 31, 1658
1673.
28. S. Y. Reece, J. A. Hamel, K. Sung, T. D. Jarvi, A. J. Esswein,
J. J. H. Pijpers and D. G. Nocera, Science, 2011, 334, 563.
29. A. Fujishima and K. Honda, Nature, 1972, 238, 3738.
30. A. Fujishima, X. T. Zhang and D. A. Tryk, Surf. Sci. Rep, 2008, 63,
515582.
31. Y. Xu and M. A. A. Schoonen, Am. Mineral., 2000, 85, 543556.
32. B. Neumann, P. Bogdano and H. Tributsch, J. Phys. Chem. C, 2009, 113,
2098020989.
33. B. D. Alexander, P. J. Kulesza, L. Rutkowska, R. Solarska and
J. Augustynski, J. Mater. Chem., 2008, 18, 22982303.
34. A. Stavrides, Proc. SPIE, 2006, 6340 , 63400K.
View Online

107
35. E. Miller, R. E. Rocheleau and S. U. M. Khan, Int. J. Hydrogen Energy,

2004, 29, 907914.
36. K. Sivula, F. Le Formal and M. Gratzel, Chemsuschem, 2011, 4, 432449.
37. B. ORegan and M. Gratzel, Nature, 1991, 353, 737740.
38. M. Gratzel, Nature, 2001, 414, 338344.
39. J. Augustynski, G. Calzaferri, J. C. Courvoisier and M. Gratzel, Hydrogen
energy progress XI : proceedings of the 11th World Hydrogen Energy
Conference, Stuttgart, Germany, 1996.
40. M. Gratzel, Cattech, 1999, 3, 317.
41. J. H. Park and A. J. Bard, Electrochem. Solid-State Lett, 2006, 9, E5E8.
42. H. Arakawa, C. Shiraishi, M. Tatemoto, H. Kishida, D. Usui, A. Suma,
A. Takamisawa and T. Yamaguchi, Solar Hydrogen and Nanotechnology
II, 2007, 6650, 6500365003.
43. M. K. Nazeeruddin, P. Pechy, T. Renouard, S. M. Zakeeruddin,
R. Humphry-Baker, P. Cointe, P. Liska, L. Cevey, E. Costa, V. Shklover,
L. Spiccia, G. B. Deacon, C. A. Bignozzi and M. Gratzel, J. Am. Chem.
Soc., 2001, 123, 16131624.
44. J. H. Yum, P. Walter, S. Huber, D. Rentsch, T. Geiger, F. Nuesch, F. De
Angelis, M. Gratzel and M. K. Nazeeruddin, J. Am. Chem. Soc., 2007,
129, 10320.
45. Y. Shi, R. B. M. Hill, J.-H Yum, A. Dualeh, S. Barlow, M. Gratzel,
S. R. Marder and M. K. Nazeeruddin, Angew. Chem., 2011, 123, 6749
6751.
46. J. Brillet, M. Cornuz, F. Le Formal, J. H. Yum, M. Graetzel and K. Sivula,
J. Mater. Res., 2010, 25, 1724.
47. I. Cesar, K. Sivula, A. Kay, R. Zboril and M. Graetzel, J. Phys. Chem. C,
2009, 113, 772782.
48. J.-H Yum, E. Barano, F. Kessler, T. Moehl, S. Ahmad, T. Bessho,
A. Marchioro, E. Ghadiri, J-E Moser, C. Yi, M. K. Nazeeruddin and
M. Gratzel, Nat. Commun, 2012, 3, 631.
49. J. Brillet, J. H. Yum, M. Cornuz, T. Hisatomi, R. Solarska, J. Augustynski,
M. Graetzel and K. Sivula, Nat. Photonics, 2012, 6, 824.
50. H. Wang, T. Deutsch and J. A. Turner, J. Electrochem. Soc., 2008, 155, F91.
51. R. N. Dominey, N. S. Lewis, J. A. Bruce, D. C. Bookbinder and M.
S. Wrighton, J. Am. Chem. Soc., 1982, 104, 467482.
52. Y. Hou, B. L. Abrams, P. C. K. Vesborg, M. E. Bjorketun, K. Herbst,
L. Bech, A. M. Setti, C. D. Damsgaard, T. Pedersen, O. Hansen,
J. Rossmeisl, S. Dahl, J. K. Nrskov and I. Chorkendor, Nat. Mater.,
2011, 10, 434438.
53. D. H. van Dorp, N. Hijnen, M. Di Vece and J. J. Kelly, Angew. Chem. Int.
Ed., 2009, 48, 60856088.
54. J. A. Baglio, G. S. Calabrese, D. J. Harrison, E. Kamieniecki, A. J. Ricco,
M. S. Wrighton and G. D. Zoski, J. Am. Chem. Soc., 1983, 105, 22462256.
55. D. Yokoyama, T. Minegishi, K. Maeda, M. Katayama, J. Kubota,
A. Yamada, M. Konagai and K. Domen, Electrochem. Commun., 2010, 12,
851853.
View Online

108
Chapter 4
56. W. Siripala, A. Ivanovskaya, T. F. Jaramillo, S-H Baeck and E.

W. McFarland, Sol. Energy Mater. Sol. Cells, 2003, 77, 229237.
57. M. Trari, A. Bouguelia and Y. Bessekhouad, Sol. Energy Mater. Sol. Cells,
2006, 90, 190202.
58. Y. Matsumoto, M. Omae, K. Sugiyama and E. Sato, The Journal of
Physical Chemistry, 1987, 91, 577581.
59. C. Leygraf, M. Hendewerk and G. Somorjai, J. Solid State Chem., 1983,
48, 357367.
60. A. J. Nozik, Appl. Phys. Lett., 1976, 29, 150.
61. D. S. Ginley and M. B. Chamberlain, J. Electrochem. Soc., 1982, 129,
21412145.
62. H. Wang and J. A. Turner, J. Electrochem. Soc., 2010, 157, F173F178.
63. Y. S. Hu, A. Kleiman-Shwarsctein, G. D. Stucky and E. W. McFarland,
Chem. Commun. (Cambridge, U. K.), 2009, 26522654.
64. S. Ida, K. Yamada, T. Matsunaga, H. Hagiwara, Y. Matsumoto and
T. Ishihara, J. Am. Chem. Soc., 2010, 132, 1734317345.
65. G. K. Mor, O. K. Varghese, R. H. T. Wilke, S. Sharma, K. Shankar, T.
J. Latempa, K.-S. Choi and C. A. Grimes, Nano Lett., 2008, 8, 19061911.
66. A. Paracchino, V. Laporte, K. Sivula, M. Gratzel and E. Thimsen, Nat.
Mater., 2011, 10, 456461.
67. M. Gratzel and J Augustynski, International Patent WO 2001002624,
2000.
68. E. L. Miller, R. E. Rocheleau and X. M. Deng, Int. J. Hydrogen Energy,
2003, 28, 615623.
69. L. J. Minggu, W. R. Wan Daud and M. B. Kassim, Int. J. Hydrogen
Energy, 2010, 35, 52335244.
70. B. D. James, G. N. Baum, J. Perez and K. N. Baum, Technoeconomic
Analysis of Photoelectrochemical (PEC) Hydrogen Production DOE Report
2009. Contract # GS-10F-009J.
71. F. Le Formal, N. Tetreault, M. Cornuz, T. Moehl, M. Gratzel and
K. Sivula, Chem. Sci, 2011, 2, 737743.
72. D. K. Zhong, M. Cornuz, K. Sivula, M. Gratzel and D. R. Gamelin,
Energy Environ. Sci, 2011, 4, 17591764.
73. K. Sivula, F. Le Formal and M. Gratzel, Chem. Mater., 2009, 21, 2862
2867.
74. J. Brillet, M. Gratzel and K. Sivula, Nano Lett.,, 2010, 10, 41554160.
75. K. Sivula, R. Zboril, F. Le Formal, R. Robert, A. Weidenka, J. Tucek,
J. Frydrych and M. Gratzel, J. Am. Chem. Soc., 2010, 132, 74367444.

Chapter 4

Uploaded by

Copyright:

Available Formats

Chapter 4

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Chapter 4

Uploaded by

Copyright:

Available Formats

Downloaded by University of Lancaster on 17/01/2015 21:41:17.

Published on 02 October 2013 on http://pubs.rsc.org | doi:10.1039/9781849737739-00083

4.1 Introduction and Motivation for Using Multiphoton

Downloaded by University of Lancaster on 17/01/2015 21:41:17.

Electron energy scheme of S2 PEC water splitting using a photoanode.

Downloaded by University of Lancaster on 17/01/2015 21:41:17.

Tandem Photoelectrochemical Cells for Water Splitting

4.2 Strategies and Limitations of Multi-Absorber

Downloaded by University of Lancaster on 17/01/2015 21:41:17.

electrolysis. It has also been suggested that a complete optimization of these

4.2.2 The Tandem Cell Concept

Downloaded by University of Lancaster on 17/01/2015 21:41:17.

Tandem Photoelectrochemical Cells for Water Splitting

Nevertheless, the precise control of materials used and device architecture

4.2.3 The D4 Strategy and its Potential

jmin DG0H2 Zfarad

Downloaded by University of Lancaster on 17/01/2015 21:41:17.

Dmex 1:23 eV Zox Zred

Downloaded by University of Lancaster on 17/01/2015 21:41:17.

Tandem Photoelectrochemical Cells for Water Splitting

semiconductor). Figure 4.3 (top) shows a contour plot showing

Downloaded by University of Lancaster on 17/01/2015 21:41:17.

set condition Eg14Eg2 is satised. The upper-right region, where ZSTH is

Tandem Photoelectrochemical Cells for Water Splitting

Downloaded by University of Lancaster on 17/01/2015 21:41:17.

4.2.4 D4 Device Architectures

4.3 PV/PV Strategies

Downloaded by University of Lancaster on 17/01/2015 21:41:17.

Electron energy schemes of two possible D4 tandem water splitting device

electrolyte.23 A comprehensive overview of these III-V systems will not be

Downloaded by University of Lancaster on 17/01/2015 21:41:17.

Tandem Photoelectrochemical Cells for Water Splitting

4.4 Photoelectrode/PV Systems

4.4.1 Advantages and Operation of a Photoanode/PV Tandem

Downloaded by University of Lancaster on 17/01/2015 21:41:17.

The operation conditions of a photoanode/PV tandem device as shown by

Downloaded by University of Lancaster on 17/01/2015 21:41:17.

Tandem Photoelectrochemical Cells for Water Splitting

4.4.2 Photoanode/PV Examples

Downloaded by University of Lancaster on 17/01/2015 21:41:17.

4.4.3 The Photoanode/DSSC Tandem Cell

Downloaded by University of Lancaster on 17/01/2015 21:41:17.

Tandem Photoelectrochemical Cells for Water Splitting

Downloaded by University of Lancaster on 17/01/2015 21:41:17.

Downloaded by University of Lancaster on 17/01/2015 21:41:17.

Tandem Photoelectrochemical Cells for Water Splitting

The optical and electronic characteristics of the trilevel tandem cell.

Downloaded by University of Lancaster on 17/01/2015 21:41:17.

4.5 Photoanode/Photocathode Systems

Downloaded by University of Lancaster on 17/01/2015 21:41:17.

Tandem Photoelectrochemical Cells for Water Splitting

Downloaded by University of Lancaster on 17/01/2015 21:41:17.

(Top) Illustration of a D4 tandem cell comprised of n-type TiO2 and

4.6 Practical Device Design Considerations

Downloaded by University of Lancaster on 17/01/2015 21:41:17.

Tandem Photoelectrochemical Cells for Water Splitting

latter point is especially important, given that the stoichiometric combination

(Top) A simple schematic of a photoanode/photocathode tandem cell

Downloaded by University of Lancaster on 17/01/2015 21:41:17.

4.7 Summary and Outlook