Supporting Information
Photoelectrochemical hydrogen e
volution using Si microwire arrays
Shannon W. Boettcher, Emily L. Warren, Morgan
C. Putnam, Elizabeth A. Santori, Daniel
Turner-Evans, Michael D. Kelzenberg, Michael
G. Walter, James R. McKone, Bruce B.
Brunschwig, Harry A. Atwa
ter*, Nathan S. Lewis*
Supplementary Figures
Figure S1.
Panel (a) shows the
J-E
data collected for the electro
de fabricated with ‘enhanced’
absorption due to light-trappi
ng elements, in 0.5 M aq. H
2
SO
4
under ELH-type W-halogen solar
simulation. Panel (b) shows a cross-sectional
SEM image of the same sample. Panel (c)
compares the spectral respons
e collected for the
sample with light-t
rapping elements
(‘enhanced’) versus the spectral response for th
e normal sample. The red response in the
‘enhanced’ cell is significantly impr
oved. Panel (d) shows the increased
J
sc
with reduced angle
dependence, for the enhanced sample compared
to the normal sample. Panel (e) shows a digital
photograph of a normal Pt/n
+
p-Si wire-array electrode evolving hydrogen under ~ 1 sun
illumination. Small bubbles can be seen nucleati
ng on the wire-array surface. The larger bubbles
are stuck on the epoxy, and are the result of
the coalescence of many small bubbles.
S1
Figure S2.
Spectral irradiance of solar simulator lamp
s compared to the ASTM AM1.5G (global
tilt) reference spectra.
Materials and Methods
Wire-array growth
. Si microwire
arrays were grown using a proc
ess similar to that described
previously.
1, 2
Boron-doped p
+
-Si (111) wafers, having
a resistivity of < 0.001
·cm (Silicon
Quest International), were used as growth subs
trates. The wafers were coated with 450 nm of
thermal oxide that had
been patterned with 4-
μ
m-diameter circular holes arranged on a square
lattice with a 7
μ
m pitch. The holes were defined in th
e oxide by photolithogr
aphic exposure and
development of a photoresist layer (Microchem
S1813), followed by a buffered HF(aq) (BHF)
etch (Transene Inc.). The holes were then f
illed with 600 nm of Cu (ESPI metals, 6N) via
thermal evaporation onto the patterned photoresis
t, followed by liftoff. Patterned substrates
approximately 1.5 cm × 1.5 cm in dimension were th
en annealed in a tube furnace for 20 min at
1000 °C under H
2
flowing at a rate of 500 sccm. Wire
growth was initiated by flowing SiCl
4
(Strem, 99.9999+%), BCl
3
(Matheson, 0.25% in H
2
), and H
2
(Matheson, research grade) at rates
of 10, 1.0, and 500 sccm, respectivel
y, for 30 min. After growth,
the tube was purged with N
2
at
200 sccm and was allowed cool to ~ 650 °C over the course of ~30 min. The resulting wires
were typically between 40-60
m in length and ~2.8
m in diameter. These growth conditions
have been shown to yield p-Si wires
with active doping concentrations of ~ 10
17
cm
-3
.
1, 3
S2
Diffusion of the radial n
+
emitter.
The Cu catalyst was removed from the as-grown wire arrays
by etching in 10 % aq. HF for 10 s, 6:1:1 by volume H
2
O:H
2
O
2
(30 % in H
2
O):conc. aq. HCl at
75 °C for 15 min, 10% aq. HF for 10 s, and 20 wt % aq. KOH at 20 °C for 60 s. A conformal
SiO
2
diffusion-barrier that was ~ 200 nm in thickness was grown via dry thermal oxidation for
2 h at 1100 °C under a pure O
2
ambient. The wire-array samples were then coated with a solution
that contained 4.4 g hexamethycyclotrisiloxane (Sigma-Aldrich), 1 g polydimethylsiloxane
PDMS (Sylgard 184, Dow Corning), and 0.10
g of PDMS curing agent in 5 ml of
dicholoromethane. These samples were then s
pun at 1000 RPM for 30 s and cured at 150 °C for
30 min to produce a 10–20
μ
m thick PDMS layer selectively at the base of the wire array.
6
After
a ~ 2 s etch in a 1:1 mixture of 1.0 M tetrabut
ylammonium fluoride in tetrahydrofuran (Sigma-
Aldrich) and dimethylformamide (re
ferred to as ‘PDMS etch’) and a H
2
O rinse, these partially
in-filled arrays were immersed
for 5 min in BHF to remove th
e exposed diffusion-barrier oxide.
The PDMS was then completely removed by et
ching for 30 min in PDMS etch, which was
followed by a 10 min piranha etch (3:1 aq. conc. H
2
SO
4
:H
2
O
2
) to remove residual organic
contamination. After 5 s in 10 % aq. HF, ther
mal P diffusion was performed using solid-source
CeP
5
O
14
wafers (Saint-Gobain, PH-900 PDS)
at 850 °C for 15 min under an N
2
ambient, to yield
a radial n
+
emitter region in the regions of the wires unprotected by the thermal oxide. A SEM
image of these wires (after a 10 s etch in BHF
to removed the thin dopant oxide) is shown in
Figure 4d. Based on spreading-resi
stance measurements on planar control wafers, we estimate
the dopant concentration in the n
+
layer at ~10
19
cm
-3
with a junction depth of ~ 200 nm.
The wire array chip was then heated
to 150 °C on a hot plate, and mounting wax
(Quickstick 135, South Bay Tech.) was melted into
the array. Excess wax was removed from the
array by applying gentle, even pressure with th
e flat surface of a glass cover slip to a tissue
(KimWipe) that was draped over the sample on th
e hotplate. The mounting wax was then etched
in an O
2
plasma (400 W, 300 mTorr) until ~ 10 - 20
m of the wire tips were exposed (~60 min)
as shown in Figure 3d.
Fabrication of planar n
+
p-Si.
The process for pn-junction fabrication was adapted from
Fahrenbruch and Bube.
4
Boron-doped
p-type (100)-oriented Si wafe
rs (Silicon Inc.) with a
resistivity of ~ 0.7
cm were unpacked in a clean-room and cleaved into chips ~ 3 cm x 3 cm.
S3
These chips
were cleaned for 15 min in 6:1:1 by volume H
2
O:H
2
O
2
(30 % in H
2
O):conc. aq.
NH
3
OH at 75 °C (RCA 1) followed
by 15 min in 6:1:1 by volume H
2
O:H
2
O
2
(30 % in H
2
O):conc.
aq. HCl at 75 °C (RCA 2). The chips were th
en etched for 30 s in BHF, rinsed in H
2
O, and dried.
These clean p-Si chips were then
stacked in-between solid source CeP
5
O
14
diffusion
wafers (Saint-Gobain, PH-900 PDS) and heated at 850 °C for 15 min under an N
2
ambient in a
tube furnace. The n
+
p-Si wafers were then etched for 30 s in BHF to remove the dopant oxide.
500 – 1000 nm of Al was then thermally evaporated onto the unpolished back surface of the
wafers. The wafers were annealed for 10 min at
800 °C to drive the ev
aporated Al through the
backside n
+
layer to make ohmic contacts. The edges of
the chips were cleaved off and discarded,
to eliminate shunts from the backside Al oh
mic contact around the edge to the diffused n
+
emitter.
Electrode fabrication.
Four types of electrodes were fabricat
ed, consisting of p-Si planar, p-Si
wire-array, n
+
p-Si planar, and n
+
p-Si wire-array samples. In each case, the Si samples were
cleaved into square ~ 0.5 cm × 0.5 cm pieces. Ohmi
c contact to the p-Si wire-array, p-Si planar,
and n
+
p-Si wire-array chips were made
by rubbing Ga-In eut
ectic on the back
side of the chip.
For the n
+
p-Si samples, the annealed Al back-contac
t served as an ohmic contact. Electrical
connections were made to the samples by attach
ment of coiled tin-copper wire using conductive
silver paint. The wire arrays
were then sealed, using Hysol 1C
epoxy, at the end of glass tubing
through which the wire had been directed such th
at the surface-normal direction to the chip was
perpendicular to the glass tube. A second type
of epoxy (Hysol 9460) was used to define the
active area of the electrode (~0.02 to 0.1 cm
2
), because Hysol 1C was found to wick in-between
the Si microwires on the face of the electrode.
Prior to Pt deposition and photoelectrochem
ical measurements, the p-Si wire-arrays
samples (but not the others) were etched as
follows: 10 s 10 % aq. HF, 30 min 30 wt. % aq.
FeCl
3
, 10 s 10 % aq. HF, 1 min 20 wt. %
aq. KOH, and 10 s 10 % aq. HF.
After each step, the
wires were rinsed with 18.3 M
Ω
cm resistivity H
2
O, and dried under a stream of N
2
(g). This
etching sequence removed the copper catalyst,
the outer ~50 nm of Si, and the native oxide,
leaving a clean surface for Pt deposition.
Platinum deposition.
Pt deposition on the p-Si electrode
s was accomplished via a galvanic
displacement reaction whereby Si is
oxidized (and then etched by
HF) and Pt is reduced onto the
S4
electrode su
rface.
5
The planar samples were immersed
for 3-4 ~2.5 min intervals in an aq.
solution that contained 0.5 M HF and 1 mM K
2
PtCl
6
. The electrode was tested after each
deposition interval (see below). The electrod
e performance typically increased after every
deposition cycle up to the 3
rd
or 4
th
cycle, after which additional
deposition of Pt did not improve
the performance. Slightly improved performan
ce could be achieved by sonicating the electrode
to remove excess Pt and then re-platinizing fo
r an additional 2.5 min. The data shown in Figure
1a are representative of the best performan
ce that was achieved using this procedure.
The wire-array samples were similarly immers
ed in the platinization solution for 2.5 min
intervals, with the optimum time around 7.5 min.
Depositions for longer times were attempted,
but led to lower photocurrents
and hence worse performance.
For the n
+
p-Si samples, Pt was deposited by electr
on beam evaporation at a base pressure
of ~ 6 x 10
-6
torr onto the completed epoxy-sealed electrode
s as the final step before testing. The
thickness of Pt was monitored by
a quartz crystal microbalance. 1
nm of Pt was deposited on the
planar electrodes. For
the wire arrays, 1.5 nm was deposited,
and the wires were tilted along one
axis during deposition to improve the sidewa
ll coverage of the exposed wire tips.
Photoelectrochemical characterization.
All photoelectrochemical measurements were
performed in a flat-sided Pyrex glass electrochem
ical cell. The solution was stirred rapidly to
minimize bubble nucleation on the el
ectrode and to reduce the as
sociated variation of the
measured
J-E
data. The measurements were performe
d using a three-electrode configuration
with a saturated calomel refere
nce electrode (SCE) and Pt-coil
counter electrode, each separated
from the main compartment by a medium porosity gl
ass frit. After data collection, the data were
shifted on the potential axis such that the pot
ential of the reversible hydrogen electrode (RHE)
was zero, but no other corrections were performed (i.e. the data were not iR corrected or
corrected for any other extrinsic losses). For al
l measurements reported herein, research-grade H
2
was continuously bubbled through the solution in order to maintain a fixed Nernst potential for
the H
+
/H
2
redox couple and to remove dissolved oxygen.
The Pt/p-Si electrodes were tested in 0.5 M aq. K
2
SO
4
adjusted to pH ~ 2 using H
2
SO
4
.
To remove detrimental dissolved metal impurities from these standard-grade lab reagents, the
electrolyte solution was pre-electrolyzed for >
12 h with ~3 V applied across two large carbon-
cloth electrodes under stirring. The electrodes were te
sted at pH ~ 2, because this pH appeared to
S5
be a good com
promise between, (1) the reduced Pt
activity and Si stability
at higher pH’s and,
(2) reduced
V
oc
values at lower pH’s. Platinized p-Si wire-array electrodes w
ith the base masked
off with SiO
2
(in an identical fashion to the n
+
p-Si electrodes), did not show performance
significantly different than p-Si wi
re-array electrodes without the SiO
2
mask.
The Pt/n
+
p-Si electrodes were test
ed in ultrapure 0.5 M H
2
SO
4
(Aristar Ultra) in 18.2 M
resistivity water. The lower pH was selected due to
the higher activity of the Pt catalyst in acidic
media.
During measurements, the cell was illumina
ted using either a
300 W tungsten-halogen
ELH lamp (OSRAM) or a Newport Oriel Xe lamp
with an AM1.5G filter set. The light intensity
was calibrated using a Si photodiode to produce a
photocurrent equivalent to that obtained under
100 mW cm
-2
of AM1.5 illumination at the working electr
ode. The spectral irradiance curves for
both lamps are given in Figure S2.
The quality factor (
n
) for the electrodes was extracted by
a linear fit of the dependence of
the photovoltage (
V
oc
) on the logarithm of the light-i
ntensity (and hence photocurrent,
J
ph
). The
analytical expression for the pho
tovoltage is given by the ideal di
ode equation solved
for zero net
current:
V
oc
=
(nk
B
T/q) ln (J
ph
/
J
s
)
,
(1)
where
n
is the diode quality factor,
k
B
(m
2
kg s
-2
K
-1
) is Boltzmann’s constant,
T
(in K) is the
temperature,
q
(C) is the charge on an electron,
J
ph
(A m
-2
) is the photocurrent density,
J
s
is the
saturation current density (related to recombination pathways), and
is the ratio of the actual
junction area to the geometric surface area
of the electrode (i.e
. the roughness factor).
1
Angle-resolved photocurrent measurements.
The short-circuit (
E
= 0 vs. RHE) photocurrent
was measured under broad-area 100 mW cm
-2
of ELH-type illumination while the electrode was
rotated along one axis. Normal incidence was take
n to be the angle at which the photocurrent
was minimized.
Spectral response.
Spectral response measurements were
made using illumination from a 75 W
Xe lamp that was passed through an Oriel mo
nochromator (0.5 mm slits), chopped at ~13 Hz,
and focused to a beam spot that was adjusted in
size to slightly
under-fill the electrode area. A
S6
calibrated Si diode (UDT UV-050) was used as a
standard for the sample channel. Another Si
photodiode was used to m
easure a beam-split portion of the illumination and hence serve as a
continuous calibration of the output intensity. A potentiostat (Gamry Series G 300) was used to
hold the potential of the Si workin
g electrode at short circuit (
E
= 0 vs. RHE) and to record the
sample current. The chopped components of the si
gnals were measured with independent lock-in
detection of the sample channel (potentiostat
analog output) and of the
calibration channel.
Fabrication of electrodes with light-trapping features.
Electrodes were also fabricated with
light trapping features that included: (1) an
a
-SiN
x
:H antireflective coating
(which also serves to
enhance carrier collection from
the non-pn-junction portion of the
wire initially protected with
SiO
2
)
6
, (2) a Ag back reflector, and (3) Al
2
O
3
light-scattering particles (Figure S1e). We have
recently shown that these elements
enhance the optical absorption
7
and hence the efficiency of
solid-state wire-array photovoltaics.
8
After pn-junction fabrication, th
e wire arrays were etched
for 5 min in BHF, to completely remove the re
maining oxide diffusion barrier. The wires were
then cleaned for 15 min in 6:1:1 by volume H
2
O:H
2
O
2
(30 % in H
2
O):conc. aq. HCl at 75 °C and
30 s in BHF, prior to deposition of an
a
-SiN
x
:H layer (~ 140 nm thick at
the wire tip and ~ 60 nm
thick at the wire base) using plasma-enhan
ced chemical vapor deposition, as described
previously.
7
1-
m planar-equivalent thickness of Ag was th
en deposited in successive 500-nm-thick
thermal evaporations at two differe
nt shallow angles (± ~5°) while
the sample was slowly rotated.
The array was then infilled with ~5
μ
m of PDMS using a process similar to the one described
above. This PDMS etch barrier a
llowed the Ag at the wire tips an
d sidewalls to be selectively
removed by etching for 6.5 min in
8:1:1 methanol: conc. aq. NH
4
OH: 30 wt.% aq. H
2
O
2
.
Al
2
O
3
light-scattering particles w
ith an 80-nm nominal diameter (South Bay Technology)
were then added to the wire array. The wire-a
rray was placed face-up in a flat-bottomed glass
centrifuge tube and ~ 3 ml of an ethanolic disp
ersion of the particles
(~0.3 mg/ml) was added.
Centrifugation (~3000 RPM) for 5 min was used to
drive the particles to
the base of the wire-
array. The array was then infilled with wax and
processed similarly to the devices without the
added light trapping features. Prior to Pt deposition the
a
-SiN
x
:H layer at the exposed tips of the
wires was removed with a ~2 min BHF etch.
S7
S8
Supplementary References
(1) Boettcher, S. W.; Spurgeon, J. M.; Putnam
, M. C.; Warren, E. L.; Turner-Evans, D. B.;
Kelzenberg, M. D.; Maiolo, J. R.; Atwater, H. A.; Lewis, N. S.,
Science
2010,
327, (5962),
185-187.
(2) Kayes, B. M.; Filler, M. A.; Putnam, M. C.;
Kelzenberg, M. D.; Lewis, N. S.; Atwater, H.
A.,
Appl. Phys. Lett.
2007,
91, (10).
(3) Putnam, M. C.; Turner-Evans, D. B.; Kelzenberg, M. D.; Boettcher, S. W.; Lewis, N. S.;
Atwater, H. A.,
Appl. Phys. Lett.
2009,
95, (16).
(4) Fahrenbruch, A. L.; Bube, R. H.,
Fundamentals of Solar Cells
. Academic Press: Pg. 272,
1983.
(5) Lombardi, I.; Marchionna, S.; Zangari, G.; Pizzini, S.,
Langmuir
2007,
23, (24), 12413-
12420.
(6) Kelzenberg, M. D.; Turner-Evans, D. B.; Putnam, M. C.; Boettcher, S. W.; Briggs, R. M.;
Baek, J. Y.; Lewis, N. S.; Atwater, H. A.,
Energy Environ. Sci.
2010,
Submitted.
(7) Kelzenberg, M. D.; Boettcher, S. W.; Petyki
ewicz, J. A.; Turner-Evans, D. B.; Putnam, M.
C.; Warren, E. L.; Spurgeon, J. M.; Briggs, R. M.; Lewis, N. S.; Atwater, H. A.,
Nat. Mater.
2010,
9, (3), 239-244.
(8) Putnam, M. C.; Boettcher, S. W.; Kelzenberg, M. D.; Turner-Evans, D. B.; Spurgeon, J.
M.; Warren, E. L.; Briggs, R. M.; Lewis, N. S.; Atwater, H. A.,
Energy Environ. Sci.
2010
,
3
, 1037.