S
1
Supporting information for
Photoelectrochemical Behavior of
a Molecular
Ru
-
Based Water
-
Oxidation
Catalyst Bound to TiO
2
-
Protected Si Photoanodes
Roc Matheu,
1,2
Ivan A. Moreno
-
Hernandez,
3
Xavier Sala,
4
Harry B. Gray,
3,5
Bruce
S. Brunschwig,
3
Antoni Llobet,*
1,4
Nathan S. Lewis*
3,5,6
1
Institute of Chemical Research of Catalonia (ICIQ), Avinguda Països Catalans 16, 43007
Tarragona, Spain.
2
Departament de Química Física i Inorgànica, Universitat Rovira i Virgili, Marcel·lí Domingo s/n,
43007 Tarragona, Spain.
3
Division of Chemistry and Chemical Engineering, California Institute of Technology, Pasadena,
CA 91125, USA
4
Departament de Químic
a, Universitat Autònoma de Barcelona, Cerdanyola del Vallès, 08193
Barcelona, Spain
5
Beckman Institute, California Institute of Technology, Pasadena,
CA
91125,
USA
6
Kavli Nanoscience Institute, California Institute of Technology, Pasadena, CA 91125,
USA.
S
2
OUTLINE
Outline
S
2
1.
Materials
and Methods
S3
1.1
General
Materials
S3
1.2 General Methods
S
3
1.3 Preparation of Si/
TiO
2
/
C substrates
S3
1.4 Preparation of Si
/
TiO
2
/C/
CNT
substrates
S
3
1.5 Preparation of Si
/
TiO
2
/
C
/
CNT
/
1
substrates
S4
1.6
Preparation of electrodes
S4
1.7 Preparation of Si
/
TiO
2
/C/
CNT
/[
1
+
1(O)
]
electrodes
S4
1.8 Preparation of Phosphate solutions
S4
1.9 Photoelectrochemical Methods
S5
Average loading and c
urrent density of electrodes
S
8
Picture of hole pattern
S
9
SEM
a
nalysis
of
Si
/
TiO
2
/C/
CNT
/[
1
+
1(O)
]
substrates
S
10
EDX analysis of
Si/
TiO
2
/C/
CNT
/
1
substrates
S11
Fe(CN)
6
3
-
/Fe(CN)
6
4
-
Measurements
S12
Electrochemical data at pH = 7 of
Si/TiO
2
/C and Si/TiO
2
/C/CNT electrodes
S13
Example of coverage estimation
S14
Faradaic efficiency determination
S14
Cyclic voltammetry of Si/TiO
2
/C/CNT/[
1
+
1(O)
]
before and after 1 h of oxygen
evolution
S15
Chronopotentiometry and CV experiments for 3 h of oxygen evolution
S16
References
S17
S
3
1.
MATERIALS
& METHODS
1.1
General
Materials
Na
2
HPO
4
,
Na
H
2
PO
4
,
K
3
Fe(CN)
6
,
K
4
Fe(CN)
6
·3H
2
O
,
NH
4
OH, HCl,
p
oly(methyl methacrylate)
(P
MMA
)
,
t
etrakis
-
dimethylamidotitanium (99.999%)
(
Sigma
-
Aldrich
)
,
anhydrous
methanol
(99.
8
%, Sigma
-
Aldrich)
, tetrahydrofuran (99.9 %, inhibitor free, Sigma
Aldrich)
and
b
uffered HF
(
Transene Company, Inc.
)
w
ere used as
received
.
Silicon wafers
(
Addison Engineering, Inc
.
)
,
multiwalled C nanotubes (
D.D > 50 nm)
(
He
Ji
, Inc
),
In
–
Ga eutectic alloy (99.99%) (Alfa
-
Aesar),
and Ag Paint (SPI, INC) were purchased from commercial suppliers. The
pyrolytic graphite target,
99.999% pure
C
,
was
acquired
from
ACI Alloys and was used as a C target
for sputtering
.
The
Ru(tda)(py
-
py
r)
2
c
omplex
was synthe
s
ized as reported
previously
.
1
1.2
General methods
A
Fuji F200
Ultratech
was used for
atomic
-
layer deposition (
ALD
)
and
a
n AJA Orion was used for
sputtering.
An e
nvironmental
s
canning
-
e
lectron
m
icroscope from FEI (Quanta 600) with a
n
energy
-
dispersive X
-
ray (
EDX
)
detector
(
Oxford Instruments
)
was use
d for EDX and SEM
measurements, and an
EPSON Perfection v39 was used as an optical scanner. The pH of the
solutions was determined by a pH
meter (CRISON, Basic 20
+
)
that was
calibrated bef
ore
measurements
by use of
standard solutions at pH= 4.01, 7.00
or
9.21. Oxygen evolution was
analyzed
with a gas
-
phase Clark
-
type oxygen electrode (Unisense Ox
-
N needle microsensor)
that
was
calibrate
d
by addition of small quantities of
O
2
(g)
(99%).
A D
e
ktak XT stylus was used for
profilometry.
1.3
Preparation of Si/
TiO
2
/
C
substrates
The preparation was adapted from the literature.
2
n
-
Si(100) wafers (P
-
doped with
a resistivity
ρ
= 0.1
-
0.3Ω∙cm, 525 ± 25
μ
m thick) or
p
+
-
Si(100) wafers (B
-
doped with
ρ
< 0.005 Ω∙cm, 381 ± 25
μ
m thick) were cleaned using an RCA etch process
that
consist
ed
of (1) etching
the
wafer with
buffered HF
(aq);
(2) soaking
the wafer
in
a 5:1:1
(by volume)
H
2
O/H
2
O
2
/NH
4
OH solution at 75 °C
for 10 min; (3) etching the wafers
again
with buffered HF
(aq)
; and then (4) soaking
the wafers
in
a 5:1:1 H
2
O/H
2
O
2
/HCl solution at 75 °C for 10 min. TiO
2
was deposited on the films by ALD
using tetrakis
-
dimethylamidotitanium (TDMAT) and
H
2
O
as reagents. Each ALD cycle consisted
of a 0.06
0
s pulse
of distilled, deionized H
2
O (18.2 M
Ω
·cm resistivity, Millipore) followed by a
0.25 s TDMAT pulse. After each pulse, N
2
(g) was purged
through
the chamber for 15 s at a flow
rate of 20 sccm. A total of 1250 cycles were performed. The substrate was maintaine
d at 150
°C
during the deposition, and the TDMAT precursor was heated to 75
°C with a heating jacket. The
H
2
O was maintained at room temperature. C
was sputtered using a pyrolytic g
raphite target in
an Ar plasma. A RF power source of 150 W was maintained f
or 2
h
and the gas flow rate was 20
psi of Ar
at
a
total pressure
of
5 mTorr. No
intentional
heating was provided to the samples
during deposition. The thick
n
ess of the layer was determined by profilometry
.
The Si/TiO
2
electrode
as
-
prepared is not
conductive due the resistance of the top
-
most TiO
2
layer
.
2
Consequently, the molecular catalysts was not deposited directly on Si/TiO
2
substrates
as was done on electrodes with conductive
TiO
2
/SnO
2
layers
.
3
,
4
The layer of sputtered C makes
S
4
the electrodes conductive, while the pyrene
pi
-
stacking provided the anchor for the molecular
catalysts on the drop
-
casted CNT materials.
1.4
Preparation of Si/
TiO
2
/C/
CNT
substrates
The Si/
TiO
2
/
C substrates were cleaved into pieces
~
0.25 cm
2
in area, with the actual area
measured using an
optical scanner
and ImageJ software.
A suspension of
multiwalled
c
arbon
n
anotubes (
CNT
)
was prepared by sonicating
the CNT
for
1
h
in tetrahydrofuran (THF) (1 mg / 1
mL)
. 600
L
cm
-
2
of the s
uspension
was then
deposited on the Si/
TiO
2
/
C substrates, using
several
volumes
(10 times 60
L cm
-
2
) with the use of an Eppendorf pipette,
to a
void overflow
. The
samples were
air
-
dried for 10
min
and
60
L
cm
-
2
of a
PMMA
solution
in
dichloromethane
(
0.25
mg
mL
-
1
)
was
added on
to
the electrodes.
This
process resulted in
a nominal coverage of 0.6 mg
cm
-
2
of CNT.
The samples were dried
for 2
h
at room temperature.
H
oles
were
carefully
scratched on top of the Si/TiO
2
/C/CNT substrates
using a
0.64 ± 0.01 mm
diameter
mask with a
pitch of
1.02 ± 0.05 mm
(
Figure S
1
)
and a pointer of
~
0.2
m
m diameter
.
The surface area of the
CNT is
4
0
–
6
00 cm
2
mg
-
1
, as calculated from
the ratio of the surface area of the
CNT to the
geometric area of
24
–
36
0
.
5
This
value
is in
good agreement
with the ratio
estimated from
electrochemical measurements
described in more detail below
.
1.5
Preparation of
Si
/
TiO
2
/C/
CNT
/
1
substrate
Si
/
TiO
2
/
C
/
CNT
substrates were soaked in a solution of complex
1
in methanol
(
0.3
0
mM
)
for 12
h
,
rinsed with
a fresh solution of m
ethanol
,
air
-
dried
,
and analyzed by
EDX
and electrochemical
techniques. The EDX of the modified electrodes clearly show
ed
the incorporation
of the Ru on
the substrates (
Figure S3
)
.
1.6
Preparation of
Si
/
TiO
2
/
C,
Si/
TiO
2
/C/
CNT
and
Si/
TiO
2
/C/
CNT
/
1
electrodes
The preparation of elect
rodes
based on
Si/
TiO
2
/C, Si/
TiO
2
/C/
CNT
and Si/
TiO
2
/C/
CNT
/
1
substrates was adapted from the literature
.
2
In
–
Ga eutectic alloy
was used to
sc
r
ibe
the
back
side of the samples
to make an Ohmi
c contact
. A
Sn
-
coated Cu wire was passed through a glass
tube
and affixed to the In
–
Ga by Ag p
aint. Once the Ag paint
had
dried
, epoxy was used to seal
the samples
to
the glass tube. The resulting exposed active area of each electrode was
measured
with
an
optical scanner
and ImageJ software. The area of the final samples was between 0.1
cm
2
-
0.2 cm
2
.
1.7
Pr
eparation of Si/
TiO
2
/C/
CNT
/[
1
+
1(O)
]
electrodes
Electrodes of
Si/TiO
2
/C/
CNT
/
1
were
used as working electrodes in
a pH = 12 phosphate solution
in which a
Pt wire and Hg/Hg
2
SO
4
were used as counter and reference electrode
s, respectively
.
A potential of 1.
30
vs NHE was applied to the
p
+
-
Si/TiO
2
/C/
CNT
/
1
electrodes in the absence of
illumination
whereas
a potential of 1.
10
V
vs NHE
was applied to
the
n
-
Si/TiO
2
/C/
CNT
/
1
electrodes
under 3
S
un illumination. The electrodes were then rinsed with fresh water and dried
in air
.
S
5
An
alogous systems have been studied in solution and immobilized on glassy carbon
electrodes.
1
,
6
In both cases, the amount of Ru=O generated in situ
was
not sufficient for
extended catalysis and thus
more time at oxidizing potentials was required to produce sufficient
Ru=O, as
was
done
in the present work.
Molecular Ru catalysts that form RuO
2
upon
decomposition show enhanced performance
.
7
Extremely positive potentials were avoided in this
work to minimize such effects.
1.8
Preparation of Phosphate solutions
pH = 7.0 buffered solution (
I
= 0.1 M): p
owders of NaH
2
PO
4
(2.31 g, 0.0193 M) and Na
2
HPO
4
(3.77g, 0.0266 M) were dissolve
d
with
sufficient
deioni
z
ed
H
2
O to make
up 1 L
of
solution.
pH = 12.0 buffered solution (
I
= 0.1 M): p
owders of Na
2
HPO
4
(10.293g, 0.0073 M) and Na
3
PO
4
(2.06g, 0.0126 M) were dissolved with
sufficient deionized H
2
O to make up 1 L of solution
.
1.9
Photoelectrochemical
Methods
1.9.1
Instrument
s
A Bio
-
Logic model SP
-
200 potentiostat or a
CH Instruments
CHI
660d
potentiostat
were
used in a 3
-
electrode
configuration
.
1.9.2
Fe(CN)
6
3
-
/Fe
II
(CN)
6
4
-
, pH =7 and pH = 12 measurements
Techniques
Cyclic v
oltammetry
(CV)
was performed at 40 mV
s
-
1
unless
otherwise specified
.
Chronoamperometry
for 150 s
was used for the generation of
1(O)
by apply
ing
a potential of 1.05 V
to
n
-
Si electrodes and 1.30 V
to
p
+
-
Si electrodes.
The s
urface roughness
of
Si/TiO
2
/C/CNT electrode
s
was
estimating
using the
capacitance of the electrode
obtained
from the CV
(
Figure S
5
)
made using the
CHI660d potentiostat
.
The
CHI660d
digital
potentiostat
reports the average
current at each potential step by
taking the mean of
500 current measurements
made on
the
step.
This current is equivalent to
the current that an
analog
potentiostat measures.
The figure shows a difference
between the cathodic and
anodic scans of ~0.2 mA cm
-
2
at a scan rate of 40 mV s
-
2
. This
difference yields
a
nominal capacitance of ~2
mF
cm
-
2
based on the
geometric area
of the
electrode.
A
specific
capacitance of ~
0.
0
4
mF cm
-
2
is
observed based on the
electrochemically active area.
8
These values allow estimation of
an approximate
ratio
of 50 between
the electrochemically active surface area
and
the geometric
area
of the electrode
.
Chronopotentiometry was performed at 1 mA cm
-
2
. For all measurements at p
H
= 7, the electrochemical data were corrected for uncompensated resistance (90
% corrected).
S
6
Chronopotentiometry (CP) was used to evaluate the OER electrode stability
.
8
The CP technique fixed the current density at 1 mA cm
-
2
. The potentials observed
in the CVs at 1 mA cm
-
2
(p
+
, 1.33 V; 1 sun, 1.28 V; 2 sun, 1.18 V; and 3 sun: 1.08
V) are slightly lower than
those observed after 0.05 h in the CP experiment (p
+
,
1.34 V; 1 sun, 1.38 V; 2 sun, 1.20 V; and 3 sun: 1.08 V) due to the contribution of
the anodic non
-
faradaic current in the CV (0.1 mA·cm
-
2
) not present in the CP
technique. However, these observations
are limited by the experimental
variability among the samples (Table S1 and S2)
Electrodes
An
n
-
Si
/
TiO
2
/C/
CNT
/[
1+
1(O)
]
sample
or
p
+
-
Si/
TiO
2
/C/
CNT
/[
1
+
1(O)
]
sample
was
used as a working electrode.
For
measurements in contact with
Fe(CN)
6
3
-
/Fe
II
(CN)
6
4
-
(aq)
,
two distinct Pt disks were used as
the
counter and reference
electrode
s
, respectively
. For
measurements at pH =7 and at pH = 12, a
Pt
wire
was used as a counter electrode and Hg/Hg
2
S
2
O
4
was used as a reference
electrode.
The Pt wire was separated from the
solution by a membrane
(
Sintered Glass Filter Discs, Porosity D, aceglass®)
Cells
A
100
mL home
-
made
cell
consisted of 5 openings at the top
into which
the
working, counter and reference electrodes were placed
, with the cell also
containing a
12 x 3 mm
stir bar.
Solutions
A
phosphate buffered solution was used
for measurements at pH= 7 and at pH
= 12
.
For
measurements in contact with
Fe(CN)
6
3
-
/Fe
II
(CN)
6
4
-
,
a solution
containing 350 mM of
Fe(CN)
6
3
-
,
50 mM K
-
/Fe
(
CN)
6
4
-
and 1
.0
M
KCl(aq)
was used.
1.9.3
Light source and calibration
Illumination during cyclic voltammetry
and other
measurements
under light
was
provided by
a
Xe lamp (300 W, USHIO) with a quartz filter (cut off at 400 nm).
Before
addition of solution to the cell, a calibrated silicon photodio
de was used to adjust the
light intensity of the solar illuminator on the working electrode to the same reading that
would be observed under AM 1.5G of solar radiation with a total light intensity of 100
mW/cm
2
, 1 Sun
.
The
distance of the solar illuminator from the cell could be
adjusted
to
achieve an intensity of
1, 2 or 3
Sun
of simulated AM 1.5G light.
1.9.4
O
2
Evolution
Bulk electrolysis was performed using a 10 mL two
-
compartment cell with a sintered
glass filter disc
(aceglass®, Porosity D)
separator between the two compartments.
Both
compartments were filled with 5 mL of pH = 7 solution and both compartments were
equipped with a stir bar.
A Ag/AgCl (KCl sat.)
electrode
was used as the r
eference
.
The
counter electrode
was placed in one compartment and the
working electrode, reference
S
7
electrode and a Clark electrode to measure oxygen evolution and to calculate the
Faradaic efficiency
,
were in the other compartment
.
At the end of the bulk electrolysis,
t
he
Clark electro
de was calibrated by
addition of
known
volumes of 99%
purified
O
2
(g)
.
1.9.5
Surface coverage Estimation (
훤
)
The surface coverage (
훤
) of complex
es
1
and
1(O)
on the electrodes was estimated by
applying the formula
훤
(
mol·cm
-
2
) =
Q
/ (
n
*
S
*
F
),
where
Q
is the charge under the Ru
III
/Ru
II
cathodic
wave
of the
1
and
1(O)
species,
n
is the number of electrons involved in the
electron transfer (1 e
–
in all the cases),
S
is the
geometric
surface
area
of the electrode,
and
F
is
Faraday’s constant
.
The
integration of the
cathodic wave for the two species
requires
the use of a
baseline
for each peak.
The drawing of a
n
accurate baseline is
difficult, particular
ly
for the
1(O)
peak,
because
the two peaks are close together
(
Figure
S
6
)
.
The ratio
for the concentrations of
1
and
1(O)
was calculated
by
dividing the
훤
of
the former by the
훤
of the latter.
Figure
S
6
provides
an example of the estimation of the
coverage
,
whereas
Table S1
contains
the
resulting coverage values
.
The average coverage was
estimated from 5 independent experiments and the standard
devi
ation between samples was used to estimate the standard
error
; however, the
actual error maybe
substantially
large
r
due to systematic error in drawing the baseline
.
For individual experiments,
the
value for the
coverage of the sample is pro
vi
ded in the
captions of the
related f
igures.
1.9
.
6 Attempts to analyze the samples by XPS and ICP
-
MS/OES
The XPS C 1s and the Ru 3d
3/2,
Ru 3d
5/2
signals overlap because of the high content of C
relative to Ru and the close proximity of the binding energies. The overlap prevented
detection of RuO
2
.
ICP
-
MS/OES was not performed because a submonolayer of Ru is
expected to yield a concentration of Ru
in solution that is below the detection limit of
the technique.
S
8
Table S1
: Loading
of the molecular complexes
1
and
1(O)
on
n
-
Si/
TiO
2
/C/
CNT
/
1,
n
-
Si/
TiO
2
/C/
CNT
/[
1
+
1(O)
]
,
p
+
-
Si/
TiO
2
/C/
CNT
/
1
and p
+
-
Si/
TiO
2
/C/
CNT
/[
1
+
1(O)
]
electrodes.
(nmol
cm
-
2
)
on
Si
/
TiO
2
/C/
CNT
/
1
/
(nmol
cm
-
2
)
on
Si/
TiO
2
/C/
CNT
/
[
1
+
1(O)
]
p
+
-
Si
14.6 ± 2.4
3.0 ± 0.8 / 1.2 ± 0.6
(ratio
1/1
(0)
= 2.5 ± 1.0)
n
-
Si
13.
0
± 2.2
2.9 ± 0.
8
/ 0.7 ± 0.3
(ratio
1/1
(0)
= 4.2 ± 1.3)
Table S2
:
C
urrent density
(
J
,
mA
cm
-
2
)
measured at 1.3
0
V
for
n
-
Si/
TiO
2
/C/
CNT
/[
1
+
1(O)
]
electrodes at pH = 7 at different light
intensities
.
1 Sun
2 Sun
3
S
un
J
= 0.9 ± 0.1 mA
cm
-
2
J
=
1.4 ± 0.1
mA
cm
-
2
J
=
1.8± 0.2
mA
cm
-
2
S
9
Figure S1
:
Alumin
um foil with patterned
0.32
±
0.0
1
mm
2
holes
(
diameter = 0.64 ±
0.01 mm)
)
with a
pitch
of
1.02
± 0.
05
mm
that was
used
to prepare Si/
TiO
2
/C/
CNT
electrodes
.
A cm ruler
is shown for comparison.
S
10
Figure S
2
:
Representative SEMs of
Si/TiO
2
/C/
CNT
/
1
electrodes:
Top, cross
-
secti
on showing
the
14 ± 1
m
thick layer of
CNT
; Bottom
, front
view of
Si/TiO
2
/C/
CNT
/
1
showing the 0.2
1
±
0.
01
mm
2
holes
(diameter = 0.52 ±
0.02 mm)
separated by
1.10 ± 0.06
mm
.
The measured
size of the
hole
s
lead
to
an
estimated
18% of area
of the
Si/TiO
2
/C/
CNT
/
1
substrates
exposed
to
the
illumination.
S
11
Figure S3
: EDX profiles of a Si/TiO
2
/C/CNT sample before (left) and after (right) soaking in a
solution that contained
1
: left, EDX spectra of substrates after soaking (Si/TiO
2
/C/CNT)
;
and
right EDX spectra
of Si/TiO
2
/C/CNT/
1
. The measured
amount
in the Si/TiO
2
/C/CNT/
1
substrate
was 0.8 % by weight.
The traces of Ni content in the CNT are due to the synthesis as indicated
by the
CNT
supplier.
5
S
12
Fe(CN)
6
3
-
/Fe(CN)
6
4
-
Measurements
The electrical properties of the p
+
-
Si electrodes without an attached catalyst were analyzed by
measuring the current density vs potential
(J
-
E
) response in a 350 mM
[Fe(CN)
6
]
3
-
-
50 mM
[Fe(CN)
6
)]
4
-
(aq) solution (Figure S4). The p
+
-
Si/TiO
2
/C electrodes exhibited small (~60 mV) kinetic
overpontential at anodic current densities of 10 mA cm
-
2
, in accord with previous reports for p
+
-
Si/TiO
2
electrodes.
2
The overpotential of the p
+
-
Si/TiO
2
/C/CNT electrodes (~30 mV) was less than
that of p
+
-
Si/TiO
2
/C electrodes, presumably due to the roughness of the CNT layer.
Figure S5
shows the CVs for the n
-
Si electrode with and without the CNT layer at pH = 7.0. The capacitive
charging currents were much larger for the electrodes with the CNT layer, being ~0.2 m
Α
cm
-
2
at a scan rate of 40 mV s
-
1
, suggesting a 50
-
fold increas
e in the electrochemically active surface
area of the electrode. The electrochemical response of the electrodes was identical with or
without the PMMA, confirming the electrochemical inertness of the PMMA under the test
conditions and indicating that the
PMMA layer did not block solution contact with the electrode.
For n
-
type electrodes, measurements under illumination allowed estimation of the
photovoltage (
V
oc
) and the photogenerated current density (
J
l
). Under simulated 1 Sun
illumination, the n
-
Si/TiO
2
/C electrode showed
V
oc
=
-
295 ± 20 mV and
J
l
= 10.7 ± 1.7 mA cm
-
2
,
similar to
related photoanodes in 350 mM [Fe(CN)
6
]
3
-
-
50 mM [Fe(CN)
6
)]
4
-
-
1.0 M KCl(aq) (Figure
S4B).
2
,
9
The n
-
Si/TiO
2
/C/CNT electrodes with no holes exhibited virtually no light
-
induced
current density (
J
=0.08 ± 0.01 mA cm
-
2
), whereas with holes present, the value of
J
l
= 2.0 ± 0.3
mA cm
-
2
w
a
consistent with the 18% of exposed area arising from the hole patterning step.
Figure S4.
Determination of the photoelectrochemical properties of p
+
and n
-
Si/TiO
2
/C/CNT
electrodes by electrochemical measurements in Fe(CN)
6
3
-
/4
-
(aq).
A
Cyclic
voltammetry for p
+
-
Si/TiO
2
/C electrodes in Fe(CN)
6
3
-
/4
-
(350 mM, 50 mM, 1.0 M KCl(aq), scan rate 40 mV s
-
1
)
compared to a Pt disk (dark line). Inset: Expansion of the
-
10 mA cm
-
2
to 10 mA cm
-
2
current
S
13
density region. Relative to the Pt disk, at 10 mA cm
-
2
on p
+
-
Si/TiO
2
/C (blue solid line) and p
+
-
Si/TiO
2
/C/CNT (pink solid line) electrodes, the potential loss was ~ 60 mV and 30 mV,
respectively. The electrochemical response of p
+
-
Si/TiO
2
/C/CNT was essentially identical with
and without a PMMA layer (pink and
green lines respectively).
B
Representative cyclic
voltammetry for n
-
Si electrodes in a Fe(CN)
6
3
-
/4
-
solution (350 mM, 50 mM, 1.0 M KCl(aq)) under
1 Sun illumination. The
V
oc
and the light
-
limited current density (
J
l
) w
e
re estimated from the
average
behavior of 3 independent electrodes: Si/TiO
2
/C ,blue solid line,
V
oc
=
-
295 ± 20 mV,
J
l
=
10.7 ± 1.7 mA cm
-
2
; Si/TiO
2
/C/CNT, pink solid line,
V
oc
=
-
220 ± 20 mV mV,
J
l
= 2.0 ± 0.3 mA cm
-
2
;
and Si/TiO
2
/C/CNT without holes, orange solid line,
V
oc
=
-
70 ± 3
0 mV,
J
l
= 0.08 ± 0.01 mA cm
-
2
.
Figure S5.
Cyclic voltammetry of bare n
-
Si/TiO
2
/C/CNT (yellow line) and
n
-
Si/TiO
2
/C electrodes
(pink line)
under 3 Sun illumination at pH = 7
together with CVs of p
+
-
Si/TiO
2
/C/CNT (grey line)
and p
+
-
Si/TiO
2
/C
(black
line) electrodes in the dark
at pH = 7.
All scans at 40 mV s
-
1
.