1
Supporting Information for:
Ordered Silicon Microwire Arrays Grown From
Substrates Patterned Using Imprint Lithography And
Electrodeposition
Heather A. Audesirk
‡
, Emily L. Warren
‡1
, Jessie Ku
2
, and Nathan S. Lewis*
Beckman Institute and Kavli Nanoscience Institute,
210 Noyes Laboratory, 127,72, Division of
Chemistry and Chemical Engineering, California Inst
itute of Technology, Pasadena, CA, 91125,
USA.
Present address:
1
National Renewable Energy Laboratory, 15013 Denver
West Parkway, Golden CO 80401
2
Department of Materials Science and Engineering, N
orthwestern University, 2220 Campus
Drive, Evanston, IL 60208, USA
Corresponding Author
*E,mail:
nslewis@caltech.edu
. Tel: +1 626 395 6335.
‡
These authors contributed equally to this work
2
Supporting Information for:
Ordered Silicon Microwire Arrays Grown From
Substrates Patterned Using Imprint Lithography And
Electrodeposition
EXPERIMENTAL METHODS
Fabrication of Microimprint Stamps:
Microimprint stamps were fabricated by casting two
layers of PDMS, each having different gel strengths
, onto a reusable master that was made from
a Si wafer coated with thermal oxide. The pattern
consisted of a square,packed array of 3 μm
diameter holes spaced 7 μm from center to center.
The oxide thickness, and therefore the height
of the stamp features, was 150 nm.
The high gel,strength x,PDMS was made by combining
vinylmethylsiloxane copolymer,
platinum,divinyltetramethyldisiloxane complex in xy
lene, vinyl,modified Q silica resin (50% in
xylene) and 1,3,5,7,tetravinyl,1,3,5,7,tetramethylc
yclotetrasiloxane (all components from
Gelest, Inc.).
1
These reagents were mixed for 1 min and defoamed
for 1 min (Thinky
Conditioning Mixer, Phoenix Equipment, Inc.). The
mixture was then degassed in a vacuum
chamber for 2 min. Hydride,functional polydimethyl
siloxane (50,55% (methylhydrosiloxane)
45,50% (dimethylsiloxane) copolymer) was added as a
curing agent, and the components were
then mixed and defoamed for 1 min. The mixture was
then degassed for 8 min, and the x,PDMS
was spin,coated onto the silicon master (which had
been treated with trimethylchlorosilane to
prevent adhesion) at 500 rpm for 60 s. The thickne
ss of this layer was ~ 26 μm, as determined
by contact profilometry (Bruker DektakXT). While t
he x,PDMS layer was being pre,cured for
10 min at 55 ̊C, a mixture of 10:1 Sylgard 184 (Dow
Corning) PDMS was made by combining
3
the monomer and curing agent, mixing for 1 min, and
defoaming for 5 min. The 10:1 PDMS
was then poured over the wafer and degassed for an
additional 10 min. The stamp was then
cured overnight at 80 ̊C.
Microimprint Lithography:
A degenerately doped, non,photoactive Si(111) wafer
(with a
resistivity, ρ < 0.003 J,cm, Addison Engineering, I
nc.) was cleaned in buffered HF(aq) for 1
min to remove any native oxide. The wafer was then
thoroughly rinsed in 18 MJ,cm resistivity
deionized H
2
O. A ~ 150 nm thick layer of sol,gel material (Phi
lips or Filmtronics 11F or 400F)
was then spin,coated onto the wafer. The bilayer s
tamp (which is the same size as the wafer to
be patterned) was pressed into the sol,gel, and a g
lass slide was used to force out any air bubbles
trapped between the wafer and the stamp. The stamp
and wafer were dried under ambient
conditions for 1 h, to allow the solvents to diffus
e out through the PDMS stamp and to allow the
sol,gel to fully crosslink. The stamp was then car
efully peeled away from the wafer.
Electrodeposition of the VLS Catalyst
: Before electrodeposition, the sample was then
immersed in a dilute (2% by volume) HF(aq) solution
to remove any residual sol,gel or native
oxide from the patterned regions of the wafer. Cont
rol over the duration of this etch allowed for
exposure of the conductive silicon substrate at the
bottom of the holes, but left a confining SiO
2
layer elsewhere. The wafer was then rinsed with 18
MJ,cm resistivity deionized H
2
O and dried
under a stream of N
2
(g).
A pressed electrochemical cell (
Figure S1
) was used for the
electrodeposition, to avoid epoxy or other methods
of insulating the backside of the wafer during
electrodeposition of the Cu catalyst. An aqueous co
mmercial Cu electrodeposition solution (pH
= 9, Copper Primer, Clean Earth Solutions) was used
, along with a 99.999% Cu rod as the
4
counter electrode. Cyclic voltammetry from ,0.50 V
to ,1.50 V at a scan rate of 50 mV s
,1
was
used to determine the optimal potential for potenti
ostatic deposition (,1.05 V vs. Ag/AgCl). The
amount of charge passed was used to control the thi
ckness of the copper within each hole in the
patterned sol,gel layer. Hence the amount of Cu de
posited ultimately determined the diameter of
the VLS,grown Si microwire. If too little copper w
as deposited, when the wafer was heated to
1000 ̊C, the Cu separated into multiple wire nucleat
ion sites, allowing multiple wires to grow out
of each hole. The hole diameter also limited the f
inal wire diameter, because the maximum wire
diameter is highly correlated with the diameter of
the holes.
Once the Cu electrodeposition was
complete, the electrodeposition solution was pipett
ed out of the cell, and the wafer was gently
rinsed and dried under a stream of N
2
(g).
Figure S1
: Schematic of the copper electrodeposition cell.
Microwire Growth:
After electrodeposition of Cu, the Si substrate waf
er was cleaved into
chips ~ 1.5 cm x 3 cm in size. The chips were rinse
d with isopropanol and then thoroughly dried
in a stream of N
2
(g). Each chip was placed in a quartz tube in a ch
emical vapor deposition
system, and exposed to vacuum for 30 min to remove
oxygen and any adsorbed gases. The
sample was then heated to 1000 ̊C under He and anne
aled for 20 min at ~750 torr under a 500
sccm flow of H
2
(g). Si MWs were grown using 450 sccm of H
2
, 50 sccm of SiCl
4
and 1.3 sccm
of BCl
3
(to create p,type microwires), with a growth time
of 8 – 20 min, depending on the
5
desired microwire length. The sample was cooled und
er H
2
to 750 ̊C over the course of 5 min,
and was then cooled to room temperature (under ~750
torr of He) over the course of 20 min.
Array Processing:
Prior to electrochemical testing, the wire arrays w
ere cleaned, processed
and made into electrodes using an RCA2 (5:1:1 H
2
O:HCl:30% H
2
O
2
at 70 ̊C) cleaning
procedure that has been reported elsewhere.
2,3
To create a protective boot at the base of the
wires, a dry thermal ~150 nm thick oxide was grown a
t 1100 ̊C over the entire array (
Figure
S2
). The oxide,coated arrays were then infilled with
a mixture of 10:1 Sylgard 184 PDMS
mixed 1:3 (v:v) with toluene (Sigma), to protect th
e oxide at the base of the microwires. This
mixture had a sufficiently low viscosity to spin in
to the Si MW arrays and thereby produce a ~
20 μm thick infill layer. The arrays were then bri
efly etched in a 3:1 (v:v) solution of 1,methyl,
2,pyrrolidinone (NMP, Sigma,Aldrich, 99.5%):
tert
,butyl ammonium fluoride (Sigma,Aldrich,
75 wt% in water) to remove any PDMS from the wire t
ops, and were then etched for ~3 min in
buffered HF(aq) (Transene, Inc.) to remove the ther
mal oxide from the tops of the wires. To
remove the PDMS infill, the samples were etched for
>30 min in the same 3:1 solution of 1,
methyl,2pyrrolidinone:tert,butyl ammonium fluoride.
4
Figure S2
: Scheme for creating a protective oxide “boot” on
the microwires to prevent shunting
through the wire bases and/or the degenerately dope
d substrate wafer.
6
Characterization:
The fidelity of the patterned template was confirme
d by optical microscopy,
scanning electron microscopy (SEM), profilometry an
d atomic force microscopy (AFM). The
fidelity of the microwire arrays was confirmed by S
EM and the electrochemical performance
was measured using a Princeton Applied Research Mod
el 273 potentiostat.
Electrode Fabrication:
Electrodes were made by breaking the substrates coa
ted with Si wire
arrays into chips that had areas between 0.01 cm
2
and 0.1 cm
2
. Epoxy (Loctite 9460) was used
to define the active area of the electrodes. A Ga,I
n eutectic was scratched into the back of the
chips, to create an ohmic contact to the p
+
silicon substrate. Each chip was then attached wit
h Ag
paint to a coil of tinned Cu wire. The electrodes w
ere sealed into glass tubes ~18 cm in length,
and the back, sides and any exposed wire were cover
ed in epoxy (Loctite 9460, Hysol 1C) to
ensure that the only path for the photogenerated ca
rriers was from the microwire sample through
the insulated wire to the potentiostat. Electrode a
reas were determined using a scanner and image
processing software (ImageJ).
5
Photoelectrochemical Testing:
To evaluate the electrical performance of the micro
wire arrays,
the electrodes were tested in an aqueous solution o
f 50 mM MV
2+
, from which the reduced MV
+
species is generated
in situ,
in a solution of 0.10 M phthalate buffer and 0.40 M
potassium sulfate
(adjusted to pH = 3.0). The electrochemical cell w
as illuminated from the bottom using an 808
nm diode laser at 60 mW cm
,2
. A Pt counter electrode and the SCE were used to d
etermine the
limiting current densities in the oxidized methyl v
iologen solution, and the carbon cloth counter
electrode and reference electrode were used during
the collection of the current density vs
potential (
J
,
E
) data in the solution that contained the reduced f
orm of the redox species, MV
+
.
.
7
The light intensity was monitored by placement of a
calibrated Si photodiode next to the working
electrode. To minimize mass transport effects, rap
id stirring was used during all electrochemical
measurements of the properties of Si MW arrays.
The spectral response system consisted of a 150 W X
e lamp and a monochromator (Oriel),
along with a potentiostat (Gamry Series G 300) and
a 30 Hz chopper.
5
The electrodes were
immersed in the MV
2+/+
redox couple (50 mM), and the working electrode wa
s poised at ,0.50 V
vs. SCE. A beam splitter and reference photodiode
provided a continuous measurement of the
monochromator output light intensity. The potentio
stat measured both the current from the Si
MW working electrode and from the reference photodi
ode. The data were then analyzed to
compute an external quantum yield for each waveleng
th, and thereby to determine the spectral
response characteristics of the microwire arrays.
Analysis of J-E Data in Methyl Viologen Redox Solut
ions
The potential data were corrected for iR losses us
ing:
5,6
= −
−
(1)
where the correction for the concentration overpot
ential (η
conc
) was performed using:
5,6
=
ln
,
,
− ln
,
,
(2)
In
Equation (2)
,
k
B
is Boltzmann’s constant,
T
is the absolute temperature,
n
is the number of
moles of electrons transferred,
q
is the unsigned charge on an electron,
J
l,a
is the limiting anodic
current density, and
J
l,c
is the limiting cathodic current density. To dete
rmine the limiting anodic
and cathodic current densities, a glassy carbon wor
king electrode was used, and
J
,
E
data (with a
carbon cloth reference electrode and a carbon cloth
counter electrode) were taken from 0.00 V to
,0.40 V to +0.20 V vs. the carbon cloth reference e
lectrode.
5
The data were corrected according
8
to Equation (2), with the slope of the corrected vo
ltage data yielding a value for
R
cell
. To ensure
that the data were not overcorrected, data were als
o collected from a planar p,Si wafer sample,
and the corrections were first applied to the plana
r sample. The
J
,
E
data from the wire array
electrodes were then corrected using
Equation (1)
, and the figures of merit for each electrode
were extracted from the corrected data.
Planar p-Si Controls:
To ensure the validity of the Si MW data, a planar
p,Si (R = 0.6,0.8 J,
cm) electrode was also tested in contact with the m
ethyl viologen redox species for both
J-E
(Figure S3)
and spectral response measurements. The higher ove
rall current density of the
planar sample makes it harder to get the highly abs
orbing MV
+
away from the electrode. Thus,
the jaggedness of the planar curves is due to stirr
ing and mass transport effects. The figures of
merit for this electrode were
V
oc
= 540 ± 10 mV,
J
sc
= 20 ± 3 mA cm
,2
, Φ
ext,sc
= 0.51 ± 0.00,
ff
=
0.74 ± 0.04 and η
808
= 13 ± 2%.
Figure S3
:
J-E
data and external quantum yield for planar p,Si mea
sured in aqueous methyl
viologen electrolyte.
9
REFERENCES
1. Verschuuren, M. A. Substrate Conformal Imprint L
ithography for Nanophotonics. Ph. D.
Thesis, Utrecht University, Utrecht, Netherlands,
2010
.
2. Putnam, M. C.; Filler, M. A.; Kayes, B. M.; Kelz
enberg, M. D.; Guan, Y. B.; Lewis, N.
S.; Eiler, J. M.; Atwater, H. A. Secondary Ion Mass
Spectrometry of Vapor,Liquid,Solid Grown,
Au,Catalyzed, Si Wires.
Nano Lett.
2008,
8
, 3109,3113.
3. Warren, E. L.; McKone, J. R.; Atwater, H. A.; Gr
ay, H. B.; Lewis, N. S. Hydrogen,
Evolution Characteristics of Ni,Mo,Coated, Radial J
unction, n
+
p,Silicon Microwire Array
Photocathodes.
Energy Environ. Sci.
2012,
5
, 9653,9661.
4. Takayama, S.; Ostuni, E.; Qian, X. P.; McDonald,
J. C.; Jiang, X. Y.; LeDuc, P.; Wu, M.
H.; Ingber, D. E.; Whitesides, G. M. Topographical
Micropatterning of Poly(dimethylsiloxane)
using Laminar Flows of Liquids in Capillaries.
Adv. Mater.
2001,
13
, 570,574.
5. Warren, E. L. Silicon Microwire Arrays for Photo
electrochemical and Photovoltaic
Applications. California Institute of Technology, P
asadena, CA,
2013
.
6. Fajardo, A. M.; Lewis, N. S. Free,Energy Depende
nce of Electron,Transfer Rate
Constants at Si/Liquid Interfaces.
J. Phys. Chem. B.
1997,
101
, 11136,11151.