S
1
Supporting Information for
Enhanced Stability and Efficiency
f
or
Photoelectrochemical Iodide Oxidation by
Methyl Termination and Electrochemical Pt
Deposition of
n
-
Si Microwire Arrays
Shane Ardo,
a
Elizabeth A. Santori,
a
Hal S. Emmer,
b
Ronald L. Grimm,
a
Ma
tthew J.
Bierman,
a
Bruce S. Brunschwig,
c
Harry A. Atwater,
bd
and Nathan S. Lewis
acd*
a
Division of Chemistry and Chemical Engineering, California Institute of Technology,
1200 E. California Blvd., Pasadena, California 91125, USA
b
Thomas J. Watson Laborator
ies of Applied Physics, California Institute of Technology,
1200 E. California Blvd., Pasadena, California 91125, USA
c
Beckman Institute, California Institute of Technology, 1200 E. California Blvd.,
Pasadena, California 91125, USA.
E
-
mail: nslewis@calte
ch.edu; Fax: +1 626 395
-
8867; Tel: +1 626 395
-
6335
d
Kavli Nanoscience Institute, California Institute of Technology, 1200 E. California
Blvd., Pasadena, California 91125, USA
*corresponding author
Additional Experimental Details
Chemicals
and Material
s
All chemicals
and materials
were used as received unless noted otherwise. The
hydriodic acid was unstabilized (Sigma Aldrich, 7.6 M, aq, 99.99%, or Alfa Aesar, 7.3
–
7.7 M, aq, ACS grade), except for the evaluation of planar Si samples for which
stabil
ized hydriodic acid was used (Sigma Aldrich, 7.6 M, aq, 99.95%, < 1.5%
hypophosphorous acid as stabilizer). All water was obtained from a Barnsted Nanopure
system and had a resistivity
>18 MΩ
-
cm.
Nafion™ membranes were purchased from Ion
Power (NR
-
212,
50.8 μm thick,
equivalent weight of 1100 g
of dry Nafion per
mol
e
sulfonic acid groups
).
S
2
Materials Synthesis and Processing
Crystalline Si microwire arrays were grown on planar n
-
type S
i(111) wafer substrates
(University Wafer, Si(111) wafers, As
-
doped, 0.001
-
0.004 Ω
-
cm resistivity, 406.6 nm
thermal oxide). Using photolithography, an array of holes (3 μm x 7 μm pitch) was
patterned into the thermal oxide as a hexagonal pattern, and the
holes were then filled
with thermally evaporated Cu (EPSI, 6N). Si microwire arrays were grown on a ~2 cm x
3 cm chip at 1000 °C using flow rates of 500 sccm for H
2
, 50 sccm for SiCl
4
, and 0.6
–
1.0 sccm for PH
3
for 8
–
20 min. These growth conditions ro
utinely produced
microwires ~60
μm in height. After microwire growth, the samples were cooled to ~800
°C under H
2
(g), cooled to ~200 °C under ambient pressure with a He headspace, and then
removed from the reactor.
The samples were then cleaned via a standard procedure, involving a ri
nse with H
2
O,
an etch for 10 s with buffered HF(aq) (Transene, Inc.),
followed immediately by rinsing
with H
2
O,
two consecutive RCA2 cleans consisting of H
2
O with equal parts HCl(12 M,
aq) and H
2
O
2
(9.8 M, aq) (6:1:1 v/v/v) at 70 °C for 20 min. The sample
was then
immersed for 10 s in BHF(aq) followed
immediately by rinsing with H
2
O,
60 s in
KOH(30 wt%, aq), and then 10 s in HF(~6 M, aq), followed immediately by rinsing with
H
2
O, drying under a stream of N
2
(g), and subsequent thermal oxidation for 100 min
at
1100 °C under a 4 L min
-
1
flow of O
2
(g). Oxide “boots” were defined using
polydimethylsiloxane (PDMS). PDMS was cast from a freshly prepared and degassed
solution that contained toluene, PDMS, and an initiator (5 mL: 1 g: 0.1 g). This solution
was sp
in cast on the microwire arrays for 30 s at 150 RPM, then for 30 s at 750 RPM, and
then for 30 s at 1500 RPM. The PDMS was subsequently cured at 60 °C overnight in a
vacuum oven followed by a final cure at 150 °C for 30 min on a hot plate. The oxide was
removed from the exposed regions of the microwires using a 5 min soak in BHF(aq),
followed
immediately by rinsing with H
2
O,
a PDMS etch for ~5 sec that consisted of
N
-
methyl
-
2
-
pyrrolidone with tetra
-
n
-
butylammonium fluoride (2.7 M, aq) (3:1, v/v)
,
and
the
n a PDMS etch that consisted of
N
,
N
-
dimethylformamide with tetra
-
n
-
butylammonium fluoride (1.0 M, tetrahydrofuran) (1:1, v/v) until the PDMS was
completely removed.
S
3
Surface Methylation
Immediately after etching samples with HF(~6 M, aq), samples were ri
nsed with H
2
O,
dried under a stream of N
2
(g), and transferred to a
n
N
2
(g)
-
purged glovebox (<10 ppm
O
2
(g)).
Samples were then immersed in a test tube containing chlorobenzene with
saturated
PCl
5
and
freshly
added benzoyl peroxide radical initiator, and
hea
ted at
90 °C
for ~50
min
to chlorinate the Si surface.
T
he samples were rinsed ten times with
neat
chlorobenzene
followed by ten times with neat
tetrahydrofuran
, dried briefly on a hot
plate,
rinsed three more times with neat tetrahydrofuran,
and then tra
nsferred to another
test tube containing a Grignard solution consisting of methy
l
magnesium chloride (
~
1
M)
in tetrahydrofuran
, which was used to
m
ethylate the Si surface via a
reaction
at
~
5
5
°C
for
~4
hr
.
The samples were then rinsed ten times
with
neat
tetrahydrofuran followed by
ten times with
neat
methanol
, removed from the glovebox,
rinsed
three more times with
neat methanol followed by three times with H
2
O,
dried under a stream of N
2
(g)
,
and
stored under N
2
(g)
until use
.
S
uccess of the two
-
step chlo
rination/alkylation reaction
sequence in methyl
ating atop s
ites on the silicon microwires i
s supported by large
increases in water contact angle, X
-
ray photoelectron spectroscopy data, and electrode
stability measurements (Figure S
5
).
Photoelectrode Fabri
cation and Photoelectrochemical Evaluation
The wafer was diced into several ~2.0 mm
2
pieces, and each piece was fashioned into
an electrode
of approximately the same area
by lightly scratching the back side with In
–
Ga eutectic and using Ag paint to affix
the eutectic
-
coated electrode onto a coiled, tinned
Cu wire. The other end of the wire was then inserted into a glass tube and sealed using
epoxy (Hysol 9460). The epoxy was left to cure overnight in a 60 °C oven and, if
desired, Pt was then deposited on
the electrode either electrochemically or by electron
-
beam evaporation (vide infra). Prior to electrochemical evaluation, each electrode was
briefly etched in HF
(~6 M,
aq
), and the current
versus
potential data were measured
using a three
-
electrode conf
iguration with a Si MW array working electrode, a Pt wire or
small carbon
-
cloth quasi
-
reference electrode, and a Pt mesh or large carbon
-
cloth counter
electrode. The electrolyte consisted of ~7.6 M HI(aq) that contained adventitious I
3
–
, and
air was conti
nuously blown onto the cell for temperature control. The electrolyte was
S
4
rapidly stirred and was continuously purged with Ar(g). 1 Sun of simulated solar
illumination at an intensity of 100 mW cm
-
2
was obtained from an ELH
-
type W
–
halogen
lamp. The rever
sible
formal
potential for the aqueous I
3
–
/I
–
redox couple (
E
(I
3
–
/I
–
)) was
determined using a Pt disk/button working electrode and a standard calomel electrode
(KCl saturated) (SCE) as the reference electrode. The light intensity was determined
using a calibrate
d Si photodiode (ThorLabs, Inc., FDS100) positioned at the location of
the electrode, with the incident light passing through the borosilicate glass window of the
electrochemical cell as well as through a thin path length of electrolyte. All reported
curr
ent densities were referenced to the projected geometric area of the electrolyte
contact to the electrode. Ultraviolet
–
visible electronic absorption spectra of the dissolved
I
3
–
were obtained using an HP 8452A diode
-
array spectrophotometer.
Electrochemic
al Deposition of Platinum
Electrodes were wetted with H
2
O, etched for 15 s in HF(~6 M,
aq
),
followed
immediately by
rins
ing
with H
2
O
,
and then immersed in an aqueous solution of 5 mM
K
2
PtCl
4
(99.9%, Alfa Aesar) and 200 mM LiCl(aq). Using a three
-
electrod
e
configuration with an SCE reference electrode and a Pt mesh counter electrode, the
working electrode was held potentiostatically at
–
1.0 V
vs
SCE. The deposition was
performed until at least 100 mC cm
-
2
of cathodic charge
density
had passed. The sample
s
were then rinsed with H
2
O and dried under a stream of N
2
(g).
Electron
-
beam Evaporation of Platinum
Immediately after etching samples with HF(~6 M, aq),
samples were rinsed with H
2
O,
dried under a stream of N
2
(g),
mounted onto a sample holder using Kapt
on tape affixed at
the corners, placed into an evaporator (Denton Explorer) and positioned at an angle, set
on planetary rotation, and pumped down to < 1 x 10
-
5
Torr. Pt was then deposited until
the desired planar equivalent value was obtained on a calibr
ated quartz crystal
microbalance.
Scanning Electron Microscopy with Energy Dispersive Spectroscopy
Imaging of Si microwire arrays during and after fabrication and processing, and when
S
5
affixed as
electrodes, was performed on field
-
emission scanning
-
electron microscopes
(Zeiss 1550VP). Validation that Pt was present on samples was determined using an
energy
-
dispersive spectrometer with a silicon drift detector (Oxford X
-
Max
N
).
Stability
Analysis
Based on the data in Figure
5
, th
e turnover number for net h
+
passed per Si atom was
calculated to be ~900 as follows
.
A net integrated charge density of 837 C cm
-
2
was
passed through the microwire array and b
ecause each microwire array
contain
s
one
microwire
per 7 x 7 μm unit cell
, the
average net charge passed per microwire was 410
μC = 4.25 nmol e
–
and h
+
. A ~50 μm tall microwire that is ~1.2 μm in diameter,
representative of the microwires used in this study, has a volume of ~57 μm
3
, which
b
ased on
the density of crystalline Si
(
2.33
g mL
-
1
= 83 mmol Si
cm
-
3
)
,
yields ~4.7 pmol
Si per microwire.
Division of 4.25 nmol h
+
by ~4.7 pmol Si, yields the turnover number
of h
+
per microwire. Since complete
oxid
ation of
each Si atom requires four holes
,
the
observed photoanodic current cannot
predominantly be ascribed to the oxidation of Si.
S
6
Additional Figures
0
1
2
3
4
0
1x10
17
2x10
17
3x10
17
4x10
17
5x10
17
Dopant Density (cm
-3
)
Phosphine Flow Rate (sccm)
1.8x10
18
9.45x10
17
7.3x10
17
Figure S1
.
Mean (± standard deviation) d
opant density
data
as a function of PH
3
flow
rate for
single
n
-
type Si microwires
taken from arrays
grown at 1000 °C.
Each point was
obtain
ed from a different microwire growth and error bars
are
shown
when
measurements
were performed on
multiple single microwires
for a
sample.
Figure S2
. Three
-
electrode current density
versus
potential data recorded in the dar
k
(dashed) or under 100 mW cm
-
2
of simulated AM1.5 G solar illumination
(solid lines)
provided by an ELH
-
type W
-
halogen bulb with a dichroic rear reflector for an n
-
type Si
microwire array electrode immersed in Ar
-
purged 10 mM 1,1’
-
Me
2
Fc
+
,
10
0
m
M Me
2
Fc
(wh
ere Me
2
Fc is 1,1’
-
dimethylferrocene)
in
1.0 M LiClO
4
–
CH
3
OH. Also shown is the
response after correcting for concentration overpotential (solid, purple
; CORR
)
,
45
and the
calculated ideal regenerative
-
cell energy
-
conversio
n efficiency, η
,
for each.
-0.4
-0.2
0.0
0
5
10
15
= 2.9%
CORR
= 3.4%
1sun of ELH
Current Density (mA/cm
2
)
Potential (V vs.
E
(Fc
+/o
))
S
7
400
600
800
1000
0.00
0.05
0.10
0.15
0.20
0.25
External Quantum Yield
Wavelength (nm)
Figure S3
.
Mean
(± standard deviat
ion) normal
-
incidence spectral response data at 0 V
vs.
E
(Me
2
Fc
+/0
) for two n
-
type Si
microwire
array electrode
measurement
s in Ar
-
purged,
1 mM Me
2
Fc
+
, 10
0
m
M Me
2
Fc
in
1.0 M LiClO
4
–
CH
3
OH.
The dip in the data at
wavelengths less than 600 nm is due to solution absorption by Me
2
Fc
+
.
52
Figure S4
. Three
-
electrode current density
versus
potential data recorded in the dark
(dashed) or under 100 mW cm
-
2
of simulate
d AM1.5 G solar illumination
(solid lines) for
methylated n
-
type Si
microwire
array electrodes immersed in Ar
-
purged ~7.6 M HI(aq)
containing adventitious I
3
–
, with (purple) or without (orange) electrochemically deposited
Pt. The purple data depict the be
havior of the best
-
performing sample containing Pt,
whereas the orange data represent a typical response in the absence of Pt
, along with the
calculated ideal regenerative
-
cell energy
-
conver
sion efficiency, η
, for each
. The
reference electrode was a Pt wire poised at the Nernstian potential of the I
3
–
/I
–
(aq) redox
couple.
-0.4
-0.2
0.0
0
5
10
15
20
= 0.38%
= 3.6%
Current Density (mA/cm
2
)
Potential (V vs.
E
(I
-
3
/I
-
))
1sun of ELH
HI(7.6M)/I
3
-
–
Ar
(1
atm
)
n
-
Si
–
CH
3
n
-
Si
–
CH
3
–
Pt
1sun ELH illumination
HI(7.6M)/I
3
-
–
Ar
(1atm)
Pt
pseudo
-
RE
S
8
-0.4
-0.2
0.0
0
5
10
Initial
After 50 hours
Current Density (mA/cm
2
)
Potential (V vs.
E
(I
3
-
/I
-
))
Figure S5
. Three
-
electrode current density
versus
potential data recorded under 100 mW
cm
-
2
of simulated AM1.5 G
solar illumination
for a methylated n
-
type Si
microwire
array
electrode immersed in Ar
-
purged ~7.6 M HI(aq) containing adventitious I
3
–
, initially
(black) and after 50 h of continuous cyclic voltammetric operation (black) that pass
ed
a
net
total
anodic
ch
arge
density o
f 529 C cm
-
2
.
Figure S6
.
Representative s
canning
-
electron micrograph images at 70,000 magnification
(in lens detector) of
nominally identical
Si microwires
obtained from the sample used in
Figure
5
showing P
t nanoparticles (white dots)
(a, b)
before and
(c)
after
photoelectrochemical analysis for 21 h, per the operation shown in Figure
5
.
Within
variation
between microwires
no substantial change
was observed
in
the
microwire
dimensions or in the shape or cov
erage
of Pt
, indicating that etching of Si and Pt was
slow compared to oxidation of I
–
(aq). All three scale bars are 500 nm.
(a)
(b)
(c)