Supporting Information
Materials and methods
10% buffered hydrofluoric acid (BHF, Transene Inc.), 30%(w/w) hydrogen peroxide
(H
2
O
2
), concentrated hydrochloric acid (HCl), tungsten metal powder (W, Sigma
Aldrich), platinum black (Pt
-
black, Sigma
Aldrich), acetonitrile (CH
3
CN, 99.8%, Sigma
Aldrich), tetrabutylammonium hydrogen sulfate (TBA
-
HSO
4
, 97%, Sigma Aldrich),
tetrabutylammonium tetrafluoroborate (TBA
-
BF
4
, 99%, Sigma Aldrich), silver nitrate
(AgNO
3
, >99%, Sigma Aldrich) and tetrahydrofuran (T
HF, Sigma Aldrich) were used as
received. 18 M
Ω
-
cm
resistivity water was obtained from a Barnstead Nanopure
filtration system.
Si microwire growth
Boron
-
doped (0.001
-
0.005
Ω
-
cm resistivity), single
-
side polished, (111)
-
oriented p
+
-
Si
wafers (WRS Material
s, San Jose, CA), obtained with a 300 nm
-
thick thermal oxide
layer, were used as substrates for the growth of Si microwires
1
. Photolithography was
used to produce a pattern of 3 μm diameter holes
on a 7 μm
-
pitch square lattice in a
polymeric photoresist, and the oxide exposed by the holes in the photoresist was etched
by immersion in BHF. 500 nm of high
-
purity Cu (99.9999%,) was then deposited onto
the patterned wafer by thermal evaporation. The
photoresist layer was then removed
with acetone, leaving a patterned, oxide
-
confined, Cu catalyst layer on the wafer. Si
microwires were grown by the vapor
-
liquid
-
solid chemical
-
vapor deposition process at
1000 °C at atmospheric pressure in flowing H
2
(
500 sccm) and SiCl
4
(45 sccm), with
small partial pressures of BCl
3
(1.3 sccm) as a p
-
type dopant. After microwire growth,
the Cu catalyst was removed by immersing the substrates in an RCA2 etch (1:6:1 by
volume HCl:H
2
O:H
2
O
2
) for 20 min at 70 °C. This p
rocess produced patterned 1 μm
diameter p
-
Si microwires on the degenerately doped p
+
-
Si substrate.
To prepare the core
-
shell microwire structures, the substrates containing Si microwire
arrays were etched in BHF for 20 s to remove native oxide on the micro
wire surface. A
125 nm layer of tin
-
doped indium oxide (ITO) was then deposited over the entire
substrate by RF sputtering (90% In
2
O
3
/10% SnO
2
target, Plasmaterials, Inc.). The ITO
layer produced a low
-
resistivity ohmic contact between the p
-
Si core and
the n
-
type WO
3
shell layer, and was sufficiently transparent to allow light of energies below the WO
3
band gap to be absorbed in the Si
2
.
Porous c
ore
-
shell Si/ITO/WO
3
microwire synthesis
Conformal layers of WO
3
were then electrodeposited on the S
i microwire/ITO substrates.
To prepare the electrodeposition bath
2
, a solution of peroxytungstic acid was prepared by
slowly dissolving 4.6
g of W powder in 50
mL of 30%(w/w) H
2
O
2
(aq). After the powder
had dissolved, the remaining H
2
O
2
was decomposed with Pt black until the peroxide
concentration was less than 3
mg mL
-
1
, as measured with peroxide testing strips. The
solution was then filtered to remove the Pt black, and diluted with water and isopropanol
until the final solution was 50
mM peroxytungstic acid in 65% water/35% isopropanol by
volume.
Porous core
-
shell Si/ITO/WO
3
microwire arrays were prepared using electrodeposition of
WO
3
in conjunction with sidewall
-
microsphere lithography. First, a Si microwire
-
array
substrate with a
sputtered ITO layer was connected to a wire by an alligator clip. The
back and sides of the substrate were then painted with nail polish, to prevent deposition
on these surfaces. A conformal base layer of WO
3
was then cathodically electrodeposited
on the
substrate. The electrodeposition was performed potentiostatically at
-
450 mV
versus a Ag/AgCl reference electrode (CH Instruments), in a standard three
-
necked flask
with a Pt wire as the counter electrode. The deposited tungsten oxide was solidified b
y
annealing in air for 20 min at 200
°C. This base layer insulated the conductive ITO layer
from making electrical contact with solution during photoelectrochemical
characterization.
Solutions of polystyrene microspheres of a chosen diameter were prepare
d in Teflon vials
by diluting 2.6%(w/w) stock aqueous solutions of microspheres (PolySciences Inc.) to
0.26% with water, to produce~ 4
mL of diluted microsphere solution for each deposition.
The microwire substrates were placed along the inside of the via
l, and were oriented
vertically with the full deposition area submerged in the microsphere solution. The vials
were then placed in a temperature
-
controlled sand bath and covered to control the
surrounding airflow. The vials were submerged in the sand to
the initial level of the
microsphere solution and were held at ~ 55
°C.
After the solvent had completely evaporated (~24 hr), the film of microspheres was
observed to have infiltrated the microwire substrate. Slight variations in the optical
density we
re observed along the substrate in locations where temperature changes
affected the evaporation rate of the solvent. Hence, the temperature was controlled (55 ±
3 °C) for all of the samples described herein. The substrates were removed from the vials
and
excess polystyrene material was scraped from the sides and the back of the substrate.
The infilling polystyrene colloidal assembly was stabilized by sintering for 1 h at 98
°C,
slightly below the glass
-
transition temperature of polystyrene
3
.
To prepare the porous shell layer, the microsphere
-
patterned substrate was electrically
contacted to a wire by an alligator clip. The back and sides of the substrates were again
painted with nail polish, to prevent deposit
ion on these surfaces. WO
3
was
potentiostatically electrodeposited, as described above, on the templated substrate until
the desired charge density had been passed. The substrates were then annealed at 200
°C
in air for 20 min to solidify the electrodepo
sited film. After cooling to room
temperature, the substrate was submerged overnight in THF to dissolve the polystyrene
template. Finally, the substrates were annealed at 500
°C for 2 h in air (ramp rate = 1
°C
min
-
1
) to crystallize the WO
3
in the monoc
linic phase. Plain core
-
shell Si/ITO/WO
3
microwire arrays were prepared using nominally the same procedure, except that the
production of the microsphere template was omitted from the process steps.
The total amount of electrodeposited WO
3
was controll
ed by limiting the total projected
-
areal charge passed during electrodepositon (
-
1.0 C cm
-
2
and
-
2.0 C cm
-
2
), not accounting
for any additional surface area due to the microwires. For photoelectrochemical
experiments, control electrodes of plain core
-
shel
l microwire arrays were prepared on
substrates that had arrays of 15 μm
-
long p
-
Si microwires. WO
3
was electrodeposited
onto these electrodes deposited in two steps, first with a base layer of
-
0.25 C cm
-
2
, and
then with a second layer deposited until the
desired total charge density (
-
0.75 C cm
-
2
or
-
1.75 C cm
-
2
) was reached. Scanning electron microscopy (SEM) images (Zeiss LEO
1550VP field
-
emission SEM) indicated that the diameters for these core
-
shell microwire
structures were 2.2(±
0.1)
μm for
-
1.0 C c
m
-
2
of total deposition charge density and
3.8(±0.1)
μm for
-
2.0 C cm
-
2
of total deposition charge density. To study the effect of
patterning the WO
3
shell using sidewall microsphere lithography, core
-
shell microwire
electrodes were prepared on 15 μm
-
long
p
-
Si microwires with
-
1.0 C cm
-
2
of total charge
density passed (
-
0.25 C cm
-
2
in the base layer,
-
0.75 C cm
-
2
in the porous layer), using
microspheres having diameters of 350 nm, 500 nm, 750 nm, or 1000 nm. To minimize
growth
-
to
-
growth variations in the
Si microwires, all of the electrodes described herein
were prepared from a single growth of Si microwire arrays.
Electrodes were prepared from microwire
-
array substrates by scratching Ga
-
In eutectic
into the back of the substrate, and affixing the substrat
es to a tinned Cu wire with
conductive Ag paint (SPI Supplies, Inc.). After the paint had dried, the wires were
threaded through a glass tube and the substrate was sealed with epoxy (Hysol 9460),
leaving only the microwire array exposed. The epoxy was cu
red overnight at room
temperature and dried at 60
°C for 2 h. The area of each electrode was determined by
scanning the electrode surface with a high
-
resolution (600dpi) scanner, followed by
measurement of the active area with the ImageJ software package.
Photoelectrochemical characterization of core
-
shell Si/ITO/WO
3
microwire electrodes
Photoelectrochemical current density vs
.
potential (
J
-
E) data were obtained in CH
3
CN
solutions that had 100
mM TBA
-
HSO
4
as the supporting electrolyte. The
electrochemical cell was a standard three
-
necked flask that had been modified to include
a quartz window in the path of illumination. Under these solution conditions, HSO
4
-
anions are photoanodically oxidized to S
2
O
8
2
-
at the WO
3
surface
4
. A platinum mesh was
the counter electrode, and a Ag/Ag
+
reference electrode was prepared from a Ag wire in a
solution of 10 mM AgNO
3
/10 mM TBA
-
BF
4
in CH
3
CN. The light source was a 150
W
Xe arc lamp (Oriel), mounted with a collimator and an Air
-
Mass 1.5
G (AM1.5G) filter.
Spectral response measurements were performed in the same electrolyte and cell
conditions, but illumination was provided from a 150
W Xe arc lamp (Oriel) that was
passed through a monochromator (Oriel Cornerstone 260, Newport Corp.) b
efore striking
the electrochemical cell. The electrode was held at a fixed potential (1.0 V vs. a Pt wire
poised at the Nernstian potential of the solution) and the photocurrent was measured with
a Gamry Series G300 potentiostat. The illumination was foc
used to an area smaller than
the active area of the photoelectrode (<1
mm
2
) and was chopped at a frequency of 1.0
Hz.
A lock
-
in amplifier (SR830, Stanford Research Systems) was used to discriminate the
chopped photocurrent from the dark current. The exte
rnal quantum yield was calculated
based on the wavelength
-
dependent photon flux as measured with a calibrated Si
photodiode (UV
-
50, OSI Optoelectronics) at the sample position.
Simulations of Si/WO
3
core
-
shell absorption
Simulations of the light
-
absorpti
on profiles in representative porous and plain core
-
shell
structures were performed by finite
-
difference time
-
domain (FDTD) methods
5
using the
freely available software package MEEP
6
. Monochromatic simulations were performed
by initially running the FDTD simulation until steady state was reached. The time
-
averaged absorption profile was then calculated
from a single illumination period
(1/frequency) of further simulation. The core and shell structures were assigned
wavelength
-
dependent complex index of refraction values for Si
7
and WO
3
8
, respectively,
which were interpolated from published, spectroscopically determined data. Finite values
for the extinction coefficient for wavelengths corresponding to energies near the band gap
are not reported in the available data. However, spectroscopic measuremen
ts
indicted
that photocurrent was produced at these wavelengths (c.f., Figure 4e). To account for
such behavior, the absorption values at these wavelengths were interpolated in the
simulation
between the last finite entry and a value of zero for wavelengt
hs larger than
450
nm. The simulations were performed as a single core
-
shell microwire structure in a 7
μm x 7 μm box (voxel size = 20 nm), with periodic boundary conditions in the x and y
directions to represent an array of structures matching the geomet
ry of the actual
substrates used in the experiments. Perfectly matched absorber
-
layer (PML) boundaries
were defined on the ends of the simulation box along the z
-
direction.
The core structure of the core
-
shell microwire was modeled as a 10
μm long
Si mi
crowire
(radius = 0.5
μm) with a 0.1
μm thick WO
3
base layer. The shell structure was modeled
as a cylinder with spherical voids of a single diameter to represent the pores generated by
the templating microspheres. Because SEM images indicated that the m
icrosphere
assemblies were close
-
packed and highly periodic, the spherical voids were arranged in a
face
-
centered cubic lattice, in accord with the behavior of artificial opals grown by
evaporative self
-
assembly
9
. Spheres that overlapped with the c
ore structure were
omitted from the lattice, so the core structure was unchanged by the presence of voids.
The radius of the cylinder that represented the shell structure was chosen so that the
volume of the porous structure was equal to the volume of a p
lain core
-
shell microwire
having a radius of 1.3
μm. This radius was nearly 2
μm for all chosen diameters of the
microspheres, except for the 750
nm porous core
-
shell microwire. Compared to spheres
of other diameters, the omission condition of the defini
tion of the lattice of void spheres
in the simulation resulted in a longer distance of closest approach to the microwire core
for 750
nm diameter spheres. This condition resulted in a larger volume of untemplated
WO
3
near the center of the 750 nm case com
pared to the other porous structures, and an
overall smaller diameter for the cylinder representing the shell structure required to
satisfy the equal volume constrain
t of the simulation geometry.
Table SI.1 contains a
table of the chosen radii and volumes
for each geometry.
Template sphere diameter
radius(um)
volume(um
3
)
None (equal volume case)
1.3
61.58
None (equal radius case)
2.065
175.7
350nm
2.079
61.88
500nm
2.063
61.58
750nm
1.837
61.13
1000nm
2.064
61.63
Table SI.1
-‐
Chosen radius and
calculated volume for plain and porous core
-‐
shell structures
used in FDTD simulations.
Figure SI.1
-‐
SEM image of a Si microwire array patterned with a continuous assembly of
polystyrene microspheres. The regularly ordered white dots in the image are
the tips of the
microwires near the top of the template. Microwire substrates can be patterned with this
uniformity over all of the substrates sizes tested here, up to 2cm
2
.
Figure SI.2
-‐
An SEM image of microsphere templated porosity on a
Si/ITO/WO3 microwire
array. This image represents the maximum field of view of the SEM, but patterning with
this fidelity was observed throughout the entire sample (1cm
2
).
Figure SI.3
-‐
Integrated depth profiles of fractional absorption
in the WO
3
shell of a Si/WO
3
core
-‐
shell microwire as
calculated from FDTD simulations. For plain core
-‐
shell equal
volume and equal radius electrodes, a significant fraction of illumination is absorbed at
depths much larger than the minority carrier diffusion length
(L
p
=500nm, dashed black
line).
In the porous core
-‐
shell structures, almost all of the absorption occurs at depths
much shallower than L
p
due to the small feature size of the inverse opal structure.
0
0.5
1
1.5
0.05
0.1
0.15
0.2
0.25
absorption depth (um)
fractional absorption
plain (equal volume)
320nm
330nm
350nm
370nm
380nm
390nm
400nm
410nm
420nm
430nm
440nm
0
0.5
1
1.5
2
2.5
0.05
0.1
0.15
0.2
0.25
absorption depth (um)
fractional absorption
plain (equal radius)
320nm
330nm
350nm
370nm
380nm
390nm
400nm
410nm
420nm
430nm
440nm
0
0.2
0.4
0.6
0.2
0.4
0.6
0.8
1
absorption depth (um)
fractional absorption
350nm pores
320nm
330nm
350nm
370nm
380nm
390nm
400nm
410nm
420nm
430nm
440nm
0
0.2
0.4
0.6
0.2
0.4
0.6
0.8
1
absorption depth (um)
fractional absorption
500nm pores
320nm
330nm
350nm
370nm
380nm
390nm
400nm
410nm
420nm
430nm
440nm
0
0.2
0.4
0.6
0.2
0.4
0.6
0.8
1
absorption depth (um)
fractional absorption
750nm pores
320nm
330nm
350nm
370nm
380nm
390nm
400nm
410nm
420nm
430nm
440nm
0
0.2
0.4
0.6
0.2
0.4
0.6
0.8
1
absorption depth (um)
fractional absorption
1000nm pores
320nm
330nm
350nm
370nm
380nm
390nm
400nm
410nm
420nm
430nm
440nm
Figure SI.4
-‐
External quantum yield measurem
ent
s for independently prepared, p
-‐
Si/ITO/WO
3
core
-‐
shell
electrodes for the (red)
-‐
0.25/
-‐
0.75 C cm
-‐
2
, 750nm porous core
-‐
shell
electrodes and (blue)
-‐
0.25/
-‐
1.75 C cm
-‐
2
plain core
-‐
shell electrodes
.
340
360
380
400
420
440
460
480
0
0.01
0.02
0.03
0.04
0.05
0.06
h
(nm)
External quantum yield
750nm porous, #1
750nm porous, #2
plain (2X deposition), #1
plain (2X deposition), #2
References:
1.
Kelzenberg, M. D.; Boettcher, S. W.; Petykiewicz, 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.
2.
Coridan, R. H.; Shaner, M.; Wiggenhorn, C.; Brunsch
wig, B. S.; Lewis, N. S.
J. Phys.
Chem. C
2013, 117, (14), 6949
-
6957.
3.
Wunderlich, B.; Bodily, D. M.; Kaplan, M. H.
Journal of Applied Physics
1964, 35, (1),
95
-
102.
4.
Mi, Q.; Coridan, R. H.; Brunschwig, B. S.; Gray, H. B.; Lewis, N. S.
Energy Environ.
Sci.
2013, 6, (9), 2646
-
2653.
5.
Taflove, A.; Hagness, S. C.,
Computational Electrodynamics: The Finite
-
Difference
Time
-
Domain Method
. 3rd ed.; Artech House: 2005.
6.
Oskooi, A. F.; Roundy, D.; Ibanescu, M.; Bermel, P.; Joannopoulos, J. D.; Johnson, S. G.
Computer Physics Communications
2010, 181, (3), 687
-
702.
7.
Aspnes, D. E.; Studna, A. A.
Phys. Rev. B
1983, 27, (2), 985
-
1009.
8.
von Rottkay, K.; Rubin, M.; Wen, S. J.
Thin Solid Films
1997, 306, (1), 10
-
16.
9.
Míguez, H.; López, C.; Meseguer, F.; Blanco,
A.; Vázquez, L.; Mayoral, R.; Ocaña, M.;
Fornés, V.; Mifsud, A.
Applied Physics Letters
1997, 71, (9).