Journal of The Electrochemical Society
,
163
(5) H261-H264 (2016)
H261
0013-4651/2016/163(5)/H261/4/$33.00
©
The Electrochemical Society
Si/TiO
2
Tandem-Junction Microwire Arrays for Unassisted
Solar-Driven Water Splitting
Matthew R. Shaner,
a,b
Matthew T. McDowell,
a,b
Alex Pien,
b
Harry A. Atwater,
b,c,z
and Nathan S. Lewis
a,b,
∗
,z
a
Division of Chemistry and Chemical Engineering, California Institute of Technology, Pasadena,
California 91125, USA
b
Joint Center for Artificial Photosynthesis, California Institute of Technology, Pasadena, California 91125, USA
c
Thomas J. Watson, Sr. Laboratories of Applied Physics, California Institute of Technology, Pasadena,
California 91125, USA
Tandem-junction microwire array photoelectrodes have been fabricated by coating np
+
-Si radial homojunction microwire arrays
sequentially with fluorine-doped tin oxide (FTO) and titanium dioxide (TiO
2
). These photoelectrodes effected unassisted water
splitting under simulated 1 Sun conditions with an open-circuit potential (
E
oc
)of
−
1.5 V vs the formal potential for oxygen
evolution,
E
0
(OH
−
/O
2
), a current density at
E
=
E
0
(OH
−
/O
2
)of0.78mAcm
−
2
, a fill factor (
ff
)
=
0.51, and a photovoltaic-biased
photoelectrochemical ideal regenerative cell efficiency of 0.6%.
© 2016 The Electrochemical Society. [DOI:
10.1149/2.0141605jes
] All rights reserved.
Manuscript submitted October 23, 2015; revised manuscript received December 17, 2015. Published January 15, 2016.
Arrays of semiconducting nanowires or microwires provide an in-
teresting morphology for solar fuels generation, due to a minimized
ionic transport length between the anode and cathode; the arrange-
ment of individual and complete device units in parallel to provide
protection against catastrophic failure in the event that a single de-
vice unit fails; reduced materials purity and usage requirements due
to orthogonalized directions of light absorption and minority-carrier
collection; the ability to decouple optical absorption and catalytic ac-
tivity of metallic electrocatalysts through strategic placement of the
catalyst in the internal volume of the array; and a high internal surface
area for facile functional support of earth-abundant electrocatalysts.
1–3
Single-junction Si microwires cannot provide sufficient voltage to per-
form unassisted solar-driven water splitting (1.23 V at STP), hence
tandem core-shell structures or dual microwire array structures are be-
ing explored. Theoretical calculations indicate that tandem-junction
devices can produce high efficiencies for integrated solar-driven water
splitting, hence both planar and non-planar tandem-junction systems
have been designed and fabricated.
4–9
Typical demonstration tandem-junction devices have been fabri-
cated using either a homojunction or heterojunction on a Si microwire
core. The core is surrounded by a wide bandgap absorber that forms
the second junction, which is generally a semiconductor-liquid junc-
tion. Si micro- and nano-wire tandem-junction devices have however
exhibited low operating current densities for water splitting, due to
the need for large open-circuit voltages (
V
oc
) to overcome the ther-
modynamic and kinetic requirements for sustained H
2
(g) and O
2
(g)
production under standard-state conditions.
8,9
WO
3
is attractive for a
demonstration system because it is acid-stable, as is Si and Nafion,
enabling the construction of a fully functional membrane-embedded
device for operation in locally acidic conditions.
In such structures, the open-circuit potential (
E
oc
)ofthe
WO
3
/liquid contact ultimately limits the water-splitting performance
of microwire Si/WO
3
core-shell tandem devices.
1
Si/WO
3
devices
fabricated previously also included a tin-doped indium oxide (ITO)
layer to provide efficient charge transfer between the Si and WO
3
,
but the ITO layer was unstable under operating conditions and thus
limited the ultimate device stability.
1
Larger
E
oc
values have been obtained from nanowire Si/TiO
2
core-
shell structures than from microwire Si/WO
3
structures because the
wide bandgap TiO
2
produces a relatively large photovoltage for the
device as a whole. In such structures, relatively low
E
oc
values were
however observed from the radial Si junction,
8
limiting the observed
solar-to-hydrogen (STH) efficiency,
η
STH
, of devices reported to date.
∗
Electrochemical Society Active Member.
z
E-mail:
haa@caltech.edu
; nslewis@caltech.edu
We describe herein the fabrication and properties of a device in which
the WO
3
has been replaced by TiO
2
, to generate a larger
E
oc
,andin
which the ITO has been replaced by fluorine-doped tin oxide (FTO),
to improve the long-term stability of the demonstration device. In
conjunction with the high bandgap of TiO
2
, the presence of a high-
quality np
+
-Si homojunction is expected to result in a larger complete
device
E
oc
that should enable higher operating current densities and
thus produce improved STH efficiencies in a demonstration-type of
device.
np
+
-Si/FTO/TiO
2
devices were fabricated through formation of
a radial np
+
-Si microwire homojunction, followed by sequential de-
position of conformal FTO and TiO
2
layers (see the Supplementary
Information for complete details). Si microwires were first grown
on a photolithographically patterned wafer in which 3
μ
m diame-
ter circular Cu catalysts were deposited in a 7
μ
m
×
7
μ
m square
pattern, with a SiO
2
mask layer on the remainder of the Si surface
(Figure
1a). The Si microwires were grown using a vapor-liquid-solid
(VLS) chemical-vapor deposition (CVD) process in which SiCl
4
at
1000
◦
CinH
2
(g) was the Si source, and PH
3
(g) was the n-type dopant
source (Figure
1b). The Si microwire arrays were then cleaned using
RCA I and II processes to remove organics and metals, respectively,
including the remaining Cu catalyst. A SiO
2
diffusion barrier was
formed at the microwire bases by producing a SiO
2
layer using a
dry oxidation, masking the bottom of the microwire array with poly-
dimethylsilicon (PDMS), etching the exposed SiO
2
in HF(aq), and
then removing the PDMS layer (Figure
1c). Radial p
+
-Si emitters were
formed through gas-phase diffusional doping using BCl
3
at 950
◦
Cin
H
2
(g) (Figure
1d). A conformal FTO layer was then deposited on a
Si microwire array via spray pyrolysis of 0.015 M ammonium flu-
oride and 0.49 M butyltin trichloride in an ethanol/water solution
(Figure
1e), while the sample was placed on a hotplate and main-
tained at 500
◦
C. Finally, a conformal nano-structured TiO
2
layer
was deposited at 150
◦
C for 6 h on the FTO in a hydrothermal pro-
cess that used an aqueous solution of 0.05 M titanium n-butoxide
and 6 M HCl.
10
Subsequent annealing of the array at 450
◦
C for 30
min resulted in the formation of a nano-structured, conformal rutile
TiO
2
(Figure
1f) coating. Figure
2 shows scanning-electron micro-
graph (SEM) images of the structures obtained after the process steps
depicted schematically in Figures
1d–1f.
In contact with a non-aqueous ferrocenium/ferrocene redox cou-
ple, the np
+
-Si microwire arrays demonstrated similar performance
to arrays reported previously, exhibiting
E
oc
=−
0.483 V vs the Nern-
stian potential of the solution
,
a short-circuit current density (
J
sc
)of
13 mA cm
−
2
, and a fill factor (
ff
) of 0.48, resulting in an ideal regener-
ative cell efficiency,
η
IRC
, of 3.0% (Figure S2).
1,11
Contact resistance
measurements between p
+
-Si and FTO performed on planar substrates
using the circular transmission line measurement technique (see
) unless CC License in place (see abstract).
ecsdl.org/site/terms_use
address. Redistribution subject to ECS terms of use (see
131.215.70.231
Downloaded on 2016-03-01 to IP
H262
Journal of The Electrochemical Society
,
163
(5) H261-H264 (2016)
Figure 1.
Schematic of the fabrication process. a) 3
μ
m diameter circular Cu growth catalysts photolithographically patterned in a 7
μ
m
×
7
μ
m matrix in a 500
nm thick SiO
2
layer on an n
+
-Si (111)-oriented wafer. b) n-Si microwire array growth at 1000
◦
C in a flowing gas consisting of SiCl
4
and PH
3
in H
2
. c) Metal
and organic cleaned Si microwire array that has a SiO
2
diffusion barrier around each wire’s base, formed by a dry oxidation of the entire wire array followed
by a polymetyldisiloxane mask infill and subsequent etch in HF(aq) to remove the exposed SiO
2
.d)p
+
-Si emitter formation by gas phase doping with BCl
3
.
e) Conformal deposition of FTO by spray pyrolysis at 500
◦
C in air. f) TiO
2
deposition by an aqueous hydrothermal process followed by a 30 min anneal at 450
◦
C in air.
Supplementary Information for complete details)
12
yielded a spe-
cific contact resistance value of 16.6
±
9.3
cm
2
. This resistance
corresponds to a
∼
3 mV voltage loss due to the p
+
-Si/FTO contact,
assuming light-limited current densities of
∼
2mAcm
−
2
for rutile
TiO
2
under Air Mass (AM) 1.5 G illumination conditions and the
Figure 2.
Scanning-electron microscopy (SEM) images showing samples at
various stages in the fabrication process. a) A radial junction np
+
-Si microwire
array (Figure
1d). Inset shows a zoomed-in image of the SiO
2
diffusion barrier
at the bases of the microwires. b) FTO-coated np
+
-Si microwire array (Figure
1e). c) TiO
2
-coated np
+
-Si/FTO microwire array (Figure
1f). d) Cross section
of a single np
+
-Si/FTO/TiO
2
microwire with each layer labeled.
∼
10x surface area increase of Si microwires relative to the area of
an analogous planar device. At geometric current densities of
∼
10
mA cm
−
2
, the voltage losses due to this contact resistance will be
∼
16 mV and
∼
160 mV for the Si microwire and planar architectures,
respectively. Hence the contact resistance of the p
+
-Si/FTO interface
is sufficiently low to allow use of Si microwire arrays under 1 Sun
illumination conditions. However, further contact development is re-
quired for planar architectures because a 160 mV contact resistance
would significantly affect the overall device performance.
Figure
3 shows the current density versus potential (
J
-
E
)behav-
ior of p
+
-Si/FTO/TiO
2
microwire arrays in 1.0 M H
2
SO
4
(aq) and
Figure 3.
Current density versus potential behavior of p
+
-Si/FTO/TiO
2
mi-
crowire array photoelectrodes under dark (1.0 M KOH) and simulated 1 Sun
conditions in 1.0 M KOH(aq) and in 1.0 M H
2
SO
4
(aq). This behavior demon-
strates the superior photoanode performance in 1.0 M KOH, consistent with
higher catalytic activity for oxygen evolution on the TiO
2
surface in alkaline
media relative to under acidic conditions.
) unless CC License in place (see abstract).
ecsdl.org/site/terms_use
address. Redistribution subject to ECS terms of use (see
131.215.70.231
Downloaded on 2016-03-01 to IP
Journal of The Electrochemical Society
,
163
(5) H261-H264 (2016)
H263
Figure 4.
Current density versus potential behavior of p
+
-Si/FTO/TiO
2
and
np
+
-Si/FTO/TiO
2
microwire array photoelectrodes in the dark and under sim-
ulated 1 Sun conditions. This demonstrates the additive voltage of the tandem
junction device and its high
E
OC
,
−
1.5 V vs
E
0
(OH
−
/O
2
), allowing for a rela-
tively high current density of 0.32 mA cm
−
2
at
E
=−
1.23 V vs
E
0
(OH
−
/O
2
)
(0 V vs RHE).
1.0 M KOH(aq), respectively, in a three-electrode electrochemical
cell configuration. In 1.0 M KOH, p
+
-Si/FTO/TiO
2
microwire arrays,
in which the p
+
-Si microwires acted as a physical scaffold that fa-
cilitated evaluation of the performance of the TiO
2
in the microwire
geometry, exhibited
E
oc
=−
1.0 V vs
E
0
(OH
−
/O
2
), a current density
at
E
0
(OH
−
/O
2
)
=
0.89 mA cm
−
2
,and
ff
=
0.39, for a solar-to-oxygen
ideal regenerative cell efficiency of
η
IRC
=
0.35% (Figure
4). In con-
trast, complete np
+
-Si/FTO/TiO
2
core-shell microwire array devices
exhibited voltage addition across the series-connected np
+
-Si and
TiO
2
-liquid junctions, with
E
oc
=−
1.41
±
0.11 V vs
E
0
(OH
−
/O
2
),
as expected for this tandem junction. The devices exhibited
J
=
0.62
±
0.20 mA cm
−
2
at
E
=
E
0
(OH
−
/O
2
)
ff
=
0.52
±
0.04, and
η
IRC
=
0.47
±
0.18% over the 9 photoelectrodes tested. Figure
4 presents
the behavior for one of the best-performing devices tested, which
exhibited
E
oc
=−
1.50 V vs
E
0
(OH
−
/O
2
),
J
=
0.78 mA cm
−
2
at
E
=
E
0
(OH
−
/O
2
),
ff
=
0.51, and
η
IRC
=
0.60%. Due to the low
current densities, this device is expected to operate near 0 V vs the
reversible hydrogen electrode (RHE) for unassisted water splitting,
and thus ought to exhibit
J
=
0.32 mA cm
−
2
and
η
STH
=
0.39% in a
two-electrode full cell configuration.
The efficiency of complete core-shell np
+
-Si/FTO/TiO
2
microwire-based devices was higher in 1.0 M KOH(aq) than in 1.0 M
H
2
SO
4
(aq), consistent with the reduced catalytic activity of TiO
2
for
the oxygen-evolution reaction in acidic media relative to the catalytic
activity under alkaline conditions. In contact with 1.0 M H
2
SO
4
(aq),
addition of an IrO
2
electrocatalyst to the TiO
2
surface only slightly
enhanced the fill factor, in accord with prior work.
8
Under 1 Sun of simulated sunlight, the current density decreased
by
∼
15% from its initial value during 24 h of continuous operation
under potential control at
−
1.23 V vs
E
0
(OH
−
/O
2
)(0VvsRHE)
(Figure
5). However, when the illumination was discontinued for 1 h
and then reestablished, the electrode returned to near its initial per-
formance and subsequently exhibited a similar, albeit slightly lower,
current density vs time behavior as that observed in the initial stability
measurement. Cyclic voltammograms taken throughout the stability
measurement demonstrate a decreased current density near
E
oc
as the
measurement progressed (Figure S4). However, at more positive po-
tentials the current density was essentially constant vs time, suggesting
that a potential dependent mechanism affects the current density at
the most negative potentials. Two possible physical mechanisms con-
sistent with a long time constant and potential dependent behavior
include a surface reaction at the TiO
2
/liquid interface that interferes
with catalysis and/or carrier separation and activation of trap states
within the TiO
2
layer.
Figure 5.
Current density versus time behavior of np
+
-Si/FTO/TiO
2
mi-
crowire array photoelectrodes under simulated 1 Sun conditions and held
potentiostatically at 0 V vs RHE. The first run lasted 24 h and was followed
by 1 h without illumination at open circuit, in addition to a subsequent 4 hour
run under simulated 1 Sun conditions while being held potentiostatically at
E
=−
1.23 V vs
E
0
(OH
−
/O
2
)(0VvsRHE).
Although the results herein demonstrate an improvement in open-
circuit voltage and stability for microwire array devices when used to
effect unassisted solar-driven water splitting relative to the behavior
of np
+
-Si microwire/WO
3
devices, the bandgap of the TiO
2
severely
limits the current density achievable under 1 Sun illumination condi-
tions. Accordingly, maintaining the high
E
oc
from the wide-bandgap
material while narrowing the bandgap of the top absorber layer is
crucial to obtain further significant efficiency improvements, as has
been demonstrated in many planar designs.
6,7
To date, incorporation
of lower bandgap, high quality materials has been challenging on Si
microwire arrays and other similar semiconductor architectures.
13,14
The combination of material stability and sufficient voltage
production has limited and continues to limit microwire and similar
three dimensionally structured water splitting device efficiencies.
Furthermore, the operational conditions under which the highest
performance is achieved further limit options for freestanding device
fabrication, such as membrane compatibility. For example, the TiO
2
used herein limits operation to aqueous alkaline conditions, which
requires an anion exchange membrane to provide ionic conductivity
and gas separation capabilities. However, anion exchange membranes
remain at the fundamental research level and are not generally
designed to be recast in Si microwire arrays while also providing
suitable mechanical support when removed from the growth substrate.
To-date, attempts to incorporate anion exchange membranes into
these devices and thereby obtain a freestanding device failed due
to the inability of the membrane casting process to infiltrate the Si
microwire array and/or membrane brittleness thwarting attempts to
remove the arrays intact from the substrate.
The tandem-junction np
+
-Si/FTO/TiO
2
microwire device reported
herein demonstrates an efficiency and stability improvement over pre-
viously reported devices. This improvement was achieved by combin-
ing two high output-voltage junctions with complementary absorption
regimes that simultaneously provide intrinsic stability under the des-
ignated operational conditions. Incorporation of an intermediate FTO
layer provided a sufficiently low resistance contact between the p
+
-
Si and TiO
2
. Further efforts to develop compatible anion exchange
membranes or to improve the performance in acidic media should
provide additional improvements that are necessary for application of
freestanding devices in scalable solar-driven water-splitting systems.
Acknowledgments
The authors acknowledge Stefan Omelchenko for assistance with
the XRD measurements, John Lloyd for discussions about the cTLM
) unless CC License in place (see abstract).
ecsdl.org/site/terms_use
address. Redistribution subject to ECS terms of use (see
131.215.70.231
Downloaded on 2016-03-01 to IP
H264
Journal of The Electrochemical Society
,
163
(5) H261-H264 (2016)
measurements and Dr. Shawn Chatman for providing the Xe arc lamp
spectral irradiance data. This material is based upon work performed
by the Joint Center for Artificial Photosynthesis, a DOE Energy In-
novation Hub, supported through the Office of Science of the U.S.
Department of Energy under Award Number DE-SC0004993. M.R.S.
acknowledges the Resnick Sustainability Institute for a graduate fel-
lowship.
References
1. M. R. Shaner, K. T. Fountaine, S. Ardo, R. H. Coridan, H. A. Atwater, and
N. S. Lewis,
Energy & Environmental Science
,
7
, 779 (2014).
2. B. M. Kayes, H. A. Atwater, and N. S. Lewis,
Journal of Applied Physics
,
97
, 114302
(2005).
3. S. Haussener, C. Xiang, J. M. Spurgeon, S. Ardo, N. S. Lewis, and A. Z. Weber,
Energy & Environmental Science
,
5
, 9922 (2012).
4. S. Hu, C. Xiang, S. Haussener, A. D. Berger, and N. S. Lewis,
Energy & Environ-
mental Science
,
6
, 2984 (2013).
5. J. R. Bolton, S. J. Strickler, and J. S. Connolly,
Nature
,
316
, 495 (1985).
6. O. Khaselev and J. A. Turner,
Science
,
280
, 425 (1998).
7. S. Licht, B. Wang, S. Mukerji, T. Soga, M. Umeno, and H. Tributsch,
Journal of
Physical Chemistry B
,
104
, 8920 (2000).
8. C. Liu, J. Tang, H. M. Chen, B. Liu, and P. Yang,
Nano Lett.
,
13
, 2989
(2013).
9. M. T. Mayer, C. Du, and D. Wang,
J. Am. Chem. Soc.
,
134
, 12406 (2012).
10. B. Liu and E. S. Aydil,
J. Am. Chem. Soc.
,
131
, 3985 (2009).
11. M. R. Shaner, S. Hu, K. Sun, and N. S. Lewis,
Energy & Environmental Science
,
8
,
203 (2014).
12. J. H. Klootwijk and C. E. Timmering,
Proc IEEE (ICMTS)
,
17
, 247
(2004).
13. N. C. Strandwitz, D. B. Turner-Evans, A. C. Tamboli, C. T. Chen, H. A. Atwater, and
N. S. Lewis,
Adv. Energy Mater.
,
2
, 1109 (2012).
14. C. C. Chen, D. B. Turner-Evans, H. Emmer, S. Aloni, and H. A. Atwater,
39th IEEE
Photovoltaic Specialists Conference
, 1 (2013).
) unless CC License in place (see abstract).
ecsdl.org/site/terms_use
address. Redistribution subject to ECS terms of use (see
131.215.70.231
Downloaded on 2016-03-01 to IP