of 9
1
Molecular coatings improve the selectivity and durability of CO
2
reduction chalcogenide
photocathodes
Yungchieh Lai
1
, Nicholas B. Watkins
2
, Christopher Muzzillo
3
, Matthias Richter
1
, Kevin Kan
1
, Lan Zhou
1
, Joel
A. Haber
1
, Andriy Zakutayev
3
, Jonas C.
Peters
2,*
, Theodor Agapie
2,*
, John M. Gregoire
1,*
1
Division of Engineering and Applied Science, California Institute of Technology, Pasadena, CA, USA
2
Division of Chemistry and Chemical Engineering, California Institute of Technology, Pasadena, CA, USA
3
Materials Science Center, National Renewable Energy Laboratory, Golden, CO, USA
*
E
-
mail:
gregoire@caltech.edu
(J.M.G.)
,
agapie@caltech.edu
(
T.A
.)
,
jpeters@caltech.edu
(J.
C.P
.)
Supporting Information
Experimental Methods
Photocathode synthesis
CuGa
3
Se
5
absorbers were deposited on soda
-
lime glass substrates with Mo back contact deposited by
sputtering. The deposition was performed at 600 °C using a three
-
stage co
-
evaporation process, with Ga
-
Se sourced in the 1st stage, Cu
-
Se sourced in the 2nd stage, an
d Ga
-
Se sourced in the 3rd stage. The
solution for CdS deposition contained 1.2 mM of cadmium sulfate (CdSO
4
) and 59 mM of thiourea (CH
4
N
2
S)
in NH
4
OH and DI water, and was heated to 65°C during the coating process. More details about these
experiments can
be found in prior publications.
24,25
Additive synthesis and characteriz
ation
Additives were synthesized according to previously published procedures using chemicals as received
from commercial suppliers without further purification.
32,33
High throughput electrochemical testing
Prior to the electrolysis, the e
lectrolyte (0.1 M KHCO
3
, ≥99.95% trace metals basis, Sigma Aldrich) with or
without additives (
Add
.
1
or
Add
.
2
) was purged with CO
2
(99.999%, Airgas) for at least 30 min. A bipolar
membrane (BPM, Fumasep® FBM single film, Fumatech) was used to separate th
e working and counter
electrodes. Platinum wire (99.9%, Sigma Aldrich) was used as the counter electrode. Electrolysis was
carried out with a Gamry Reference 600TM potentiostat. All electrochemical data were collected using a
Ag/AgCl reference electrode (L
F2, Innovative Instruments) and converted to a reversible hydrogen
electrode (RHE) scale using the measured solution pH of 6.8. All cells and all solution handling lines were
purged with fresh electrolyte and CO
2
between electrolysis to avoid cross
-
contami
nation. These systems
were coupled with front
-
side electrode illumination using fiber
-
coupled LEDs. The surface area of the
counter electrodes were about 0.25 cm
2
, while the working electrode surface areas were 0.32 cm
2
. The
flow rate of electrolyte was 16
0 μL/s throughout the tests.
Two cell designs were used for the bulk testing of photocathodes.
(1) An electrochemical mass spectrometry (ECMS) system was previously published for rapid CO2R
electrocatalyst screening.
43
Mass spectra were acquired on a Hiden
HPR20 mass spectrometer connected
to the outlet of the desiccant chamber. An electron energy of 70 eV was used for ionization of all species,
2
with an emission current of 500 μA to maximize detection sensitivity. Hydrogen (m/z = 2), methane (m/z
= 15), and
ethylene (m/z = 26) ions were accelerated with a voltage of 3 V. All mass
-
selected product
cations were detected by a secondary electron multiplier with a detector voltage of 1050 V.
(2) Analytical and Electro
-
chemistry (HT
-
ANEC) is an analytical electro
chemistry system previously
published by our group to efficiently detect a wide range of CO
2
R products.
37
At the end of each
(photo)electrolysis, gaseous and liquid products were sampled by the robotic sample handling system
(RSHS) and analyzed by GC (Thermo Scientific™ TRACE™ 1
300) and HPLC (Thermo Scientific UltiMate
3000). Detailed product detection (method) can be found at the previous publication.
37
H
-
cell
electrochemical testing
For photoelectrochemical CO
2
reduction durability experiments in an H
-
Cell a custom peek cell was
utilized.
44
The anode and membrane had an area of 1 cm
2
whereas the a
rea of the cathode was further
reduced by a Viton mask to 0.5 cm
2
. CO
2
saturated 0.10 M potassium bicarbonate with and without
additive was used as the electrolyte. A Pt foil anode was used behind a bipolar membrane (Fumasep®
FBM single film, Fumatech) mem
brane. A leakless Ag/AgCl electrode was used as a reference electrode.
A monochromatic LED illumination with a wavelength of 450 nm and an intensity of 65 mW/cm
2
was used.
Carbon dioxide was provided to the electrochemical cell at a flow rate of 5 sccm as
controlled by an
Alicat flow controller. The gas stream was humidified by a gas bubbler connected in series between the
electrochemical cell and flow controller. The gas exhaust stream of the electrochemical cell was passed
through a liquid trap before flo
wing through the gas sampling loop of an Agilent 7820a GCMS/TCD with
an Alicat flow meter connected to its exhaust. Quantitative analysis of gaseous products was based on
calibrations with several gas standards over many orders of magnitude in concentratio
n. With the help
of the calibration, the Faradaic efficiency towards CO2R and hydrogen evolution products could be
calculated from the measured current density.
For isotope labeling experiments the same experimental
configurations as described
above were e
mployed except
13
CO
2
was used as
the
CO
2
source.
Pre
-
and post
-
(photo)electrochemistry sample characterization
TEM/EDS
Cd(S) layer in CdS/CuGa
3
Se
5
at different conditions was characterized by cross
-
sectional Transmission
electron microscopy (TEM) and energy
-
dispersive X
-
ray spectroscopy (EDS). To prepare a cross
-
sectional
TEM specimen, a FEI DualBeam Focused Ion Beam/scanning electron microscope (FI
B/SEM) was used and
the sample was capped carbon/I
-
C prior to milling. TEM experiments were carried out in a FEI Tecnai Osiris
FEG/TEM operated at 200 kV in bright
-
field (BF) and high
-
resolution (HR) TEM mode. The EDX elemental
mapping was acquired using B
ruker Quantax. This characterization was performed by Eurofins EAG
Precision TEM in Santa Clara, California.
XRF and ICP
-
MS
The Cd in CdS/CuGa
3
Se
5
sample was characterized by X
-
ray fluorescence (XRF) using an EDAX Orbis Micro
-
XRF system to identify if any
was leached out during the photo(electrochemistry) measurements.
Inductively coupled plasma mass spectrometry (ICP
-
MS) by
Thermo Fisher Scientific iCAP™ RQ instrument
was used to determine the concentration of dissolved metals in electrolytes used for ele
ctrochemistry.
SEM
3
Morphology of the Cd(S) layer and additives were characterized by Cross
-
sectional and top
-
down view
Scanning
-
electron microscopy images and were obtained with a FEI Nova NanoSEM 450 microscope.
XPS
X
-
ray photoelectron spectroscopic (XPS
) data were collected using a Kratos Axis Nova system with a
base pressure of 1×10
-
9
Torr. The X
-
ray source was a monochromatic Al Kα line at 1486.6 eV.
Photoelectrons were collected at 0° from the surface normal with a retarding pass energy of 160 eV for
survey XPS scans with a step size of 0.5 eV, and a pass energy of 20 eV for high
-
resolution core level
scans with a step size of 0.05 eV. No charge neutralization was used. The XPS was calibrated using the
Au 4f
7/2
line (84 eV) of a sputtered gold foil. Da
ta was analyzed using CasaXPS. To calculate the
composition (atomic ratio) a Shirley background was subtracted. The core level intensities were
corrected by the analyzer transmission function and relative sensitivity factors to obtain corrected peak
intens
ities which were used to calculate the atomic ratios.
Figure S1: Initial experiments in an electrochemical flow cell where for each of 2 electrodes without
(left) and with (right) CdS coating were operated as a sequence of applied bias, chronologically f
rom left
to right. At each potential, after 120 s operation in the dark illumination from a 455 nm LED was used to
observe any photocurrent and change in H
2
, CH
4
, and C
2
H
4
production rate, which were quantified using
mass spectrometry via a pervaporation cell in the effluent of the electrolyte from the electrochemical
reactor. Both samples show negligible dark current and appreciable photocurrent from 0 to
-
0.4 V vs
RHE wit
h undesirably large dark current at more negative bias. While no hydrocarbons are observed, the
observed H
2
signal is used to estimate the Faradaic efficiency, which is near 100% without the CdS
coating and is lower with CdS, indicating formation of a non
-
detected product such as CO or another
electrochemical reaction occurring.
4
Figure S2: Evaluation of the product distribution with illumination ranging from 617 nm to 385 nm at 0V
vs RHE. The testing sequence is 617, 530, 455, and then 385nm. The FE, espe
cially for CO, is similar for
the first 3 illumination sources and increases at 385 nm illumination. The 455 nm illumination was
chosen for further experimentation to represent visible illumination. Note that xrf was used to monitor
the Cd corrosion with 8
1 (counts) prior to test and 14.5 (counts) after photoelectrolysis with the four
wavelengths.
Figure S3: a) The data from Figure 2 is shown using the partial current density for CO as opposed to the
total current density, and multiple experiments on a gi
ven electrode are shown with arrows indicating
the sequence of measurements. b) Aggregation of the experiments in a) for the
-
0.4 and 0V, indicating
that the primary mechanism of selectivity enhancement with the additives is suppression of HER. The
error b
ars for both J_eche and J_CO at
-
0.4 V and 0 V vs RHE represent the standard deviation of the
respective measurements over the several photoelectrocatalysis experiments shown in a).
5
Figure S4: GC
-
MS analyses
w
ith isotop
ically
-
labelled
13
CO
2
with aliquots ac
quired
before (“pre”), at
3
time
intervals during photoelectrolysis at 0
V vs RHE with 10 mM
Add. 2
, and after photoelectrolysis (“post”).
For each of these 5 headspace samples, mass spectra were
acquired at
a series of GC retention times
. The
2 m/z values of interest are 28 and 29, which correspond to
1
2
CO
and
13
CO
, respectively
, the la
t
ter being
th
e target product of ph
otoelectrochemical reduction of
13
CO
2
. The
N
2
from air contamination
has similar
retention time as that of CO and contributes to both
a)
m/z= 28 and
b)
m/z=29 signals
as shown in the
“pre” and “post” measurements. The mass spectrum acquired at 1.68 min retention time
,
marked
with a
vertical gray line in a
) and b),
best characterizes CO with some contribution
from
N
2
. For
each of the 5
conditions, the relevant portion of the mass spectrum is shown in c) where the colors match the legend
in a). To compare the signals during photoelectrolysis to th
ose
of the “pre” baseline, d) contains the
relative intensity for each m/z val
ue with
detected
intensity.
The approximately 5
-
fold enhancement in
m/z=29 signal in each of the measurements during photoelectrolysis is marked with an asteri
sk
. As shown
in a), t
he
m/z=28 “pre”
signal, which is from
14
N
2
,
is similar to the m/z=28 signals
in subsequent
measurements. The concomitant m/z=29
signal
from
15
N
14
N
is thus also
similar in each measurement,
demonstrating that the
~
5
-
fold increase in m/z=29 signal for the measurements during
photoelectrolysis
arises from
photoelectrochemical generation
of
13
CO
from
13
CO
2
.
6
Figure S
5
: a) Plan
-
view SEM images of electrodes including as
-
synthesized, each condition from the HT
-
ANEC measurements of Figure 2, and the H
-
cell measurement of Figures 3b
-
3c. b)
C
ross
-
sectional
SEM
for H
-
cell
with
Add
.
2
,
which shows variations in the film thick
ness.
c)
C
ross
-
section TEM image
of an
electrode from
HT
-
ANEC without additive. With
no
additive, the surface restructures into nanocubes.
The images from experiments with a molecular additive exhibit a morphology that is more similar to the
as
-
synthesized
electrode with some apparent restructuring and/or changes in morphology due to
heterogeneities in organic coating.
c)
Figure S
6
: a) XPS survey scans for the 5 electrode conditions of Figures 4a
-
4b. The analysis of the
detected species from each elect
rode are shown in b) and
F
igure 4c. All samples that underwent
electrochemistry have a signal near a binding energy of 293 eV that is likely from precipitates of KCO
3
from the electrolyte. The samples with
Add
.
2
also show F and C signals corresponding to CF
3
, which is
the counterion in the molecular additive (the CPS of the corresponding C 1s peak is shown in b), which
may also result from precipitates of the electrolyte. Adventitious carbon and the molecular add
itive
7
both contribute to C1, C2 can be assigned to carbonate species (CO
3
), C3 and C4 are satellite and shake
-
up peaks.
c)
The summary of all quantified XPS peaks that provide the basis for Figure 4c and b).
The
N:Cd
is
7.3 after the 10 min HT
-
ANEC operati
on with
Add. 2
. After the
H
-
cell measurement included 3.3
h without additive, the N signal is slightly higher and the Cd signal is below the detectability limit,
indicating that the N:Cd ratio is above 100.
These results are commensurate with the microscop
y images
of Figure 4 that show a conformal and thick additive coating after the H
-
cell
measurements.
Figure S
7
: Additional TEM data supporting Figure 4b. The depth profile of elemental composition is
obtained by horizontal averaging of the EDS mapping i
mages for each sample.
8
Figure S
8
: High resolution core level spectra for the energy regions containing C
1s and K
2p (a, d, g, j, m),
S
2p and Se
3p (b,e, h, k, n) and Cd 3d and N 1s (c, f,i,l,o) core levels. The N originating from the
additive
molecules is also apparent in the respective samples. XPS signals from the as
-
synthesized and no
-
additive
samples include an S 2p
3/2
binding energy near 161.5 eV (S
sulfide
in Figure 4c), as expected for a CdS.
45
The
electrodes with additive show an additional S signal with 2p
3/2
binding energy of 168
-
169 eV (S
sulfate
in
Figure 4c), which is more characteristic of sulfate
species.
46
48
In the
Add
.
1
sample, this signal may arise
from CdSO
4
from oxidation of the CdS layer, although the low Cd signal may sugg
est some sulfate
complexation in the molecular coating, which initially contains no sulfur. In the
Add
.
2
samples, this signal
may arise from the triflate counterion to the molecular additive. The C 1s signal involves multiple
components whose intensities
vary with condition (Figure S5), although the multiple sources of carbon
from the electrolyte, additive, and adventitious sources preclude detailed interpretation at this time. The
XPS signals collectively corroborate the findings from electron microscopy
and provide guidance for future
detailed exploration of the chemistry.
9
Table S1: The photocurrent and product distribution, as well as XRF and ICP
-
MS characterization of Cd
corrosion, from each electrolysis experiment tested with 455 nm LED (450nm LED for H
-
cell tests). The
measured Cd concentration prior to (photo)electrolys
is by ICP
-
MS ranges between 0
-
4 ppb, which is
within the
detectability
limit
. The Cd characterized by XRF prior to
photo
electrolysis is 80 ± 6 (counts) and
for post
-
run characterization it is only conducted after all photo
-
electrolysis on a sample are completed.
Sample
spot
Add.
type
Add.
conc. (mM)
Poten.
(V vs RHE)
J_eche
(mA/cm
2
)
d
FE_H
2
(%)
FE_CO
(%)
Cd, pre
(ppb)
a
Cd, post
(ppb)
b
Cd, XRF
(counts)
c
1
N/A
N/A
0
-
0.32
41.1
14.3
N/M
N/M
N/M
1
N/A
N/A
-
0.4
-
0.47
41.7
20.4
N/M
N/M
31
2
N/A
N/A
0
-
0.82
43.8
23.6
N/M
N/M
9.2
3
N/A
N/A
0
-
0.47
45.9
14.9
N/M
N/M
15
4
N/A
N/A
0
-
0.5
51.3
16.0
N/M
N/M
N/M
4
N/A
N/A
-
0.4
-
1.96
58.5
22.7
N/M
N/M
5
5
N/A
N/A
0
-
0.57
49.2
20.7
4.2
45.53
N/M
5
N/A
N/A
-
0.2
-
0.98
55.4
24.4
4.2
38.06
N/M
5
N/A
N/A
-
0.4
-
1.01
60.9
23.5
4.2
40.94
19
6
N/A
N/A
0
-
0.38
42.0
11.9
2.9
39.05
68
7
N/A
N/A
0
-
0.76
56.7
14.9
0
30.59
69
8
Add. 1
0.1
0
-
0.38
13.8
42.4
N/M
N/M
N/M
8
Add. 1
0.1
-
0.4
-
0.54
4.8
69.5
N/M
N/M
N/M
8
Add. 1
0.1
-
0.4
-
0.5
7.6
61.7
N/M
N/M
22
9
Add. 1
0.1
0
-
0.38
17.0
39.8
5.3
3.88
N/M
9
Add. 1
0.1
-
0.4
-
0.82
14.1
59.2
5.3
23.13
N/M
10
Add. 1
0.1
0
-
0.44
20.8
34.9
6.8
28.20
67
11
Add. 1
0.1
0
-
0.41
19.3
36.2
2.9
4.27
74
12
Add. 1
0.3
0
-
0.35
17.3
30.4
4.4
32.50
60
13
Add. 1
0.3
0
-
0.38
30.4
22.7
1.5
48.70
52
14
Add. 1
10
0
-
0.32
3.0
24.8
3.6
37.13
64
15
Add. 1
10
-
0.4
-
0.47
8.2
59.9
3.6
0.58
54
16
Add.
2
10
0
-
0.28
3.0
67.4
6.59
4.08
82
17
Add.
2
10
0
-
0.19
2.2
53.2
1.32
0.76
N/M
17
Add.
2
10
0
-
0.25
2.0
64.6
1.32
0.57
N/M
17
Add.
2
10
0
-
0.22
2.0
67.5
1.32
0.46
N/M
17
Add.
2
10
0
-
0.16
3.6
70.9
1.32
0.36
N/M
17
Add.
2
10
-
0.4
-
0.35
2.0
74.0
1.32
0.31
83
18
Add.
2
10
0
-
0.32
3.3
55.4
0.73
0.46
N/M
18
Add.
2
10
0.2
-
0.06
4.9
34.0
0.73
0.59
N/M
18
Add.
2
10
-
0.2
-
0.44
2.7
75.9
0.73
0.17
N/M
18
Add.
2
10
-
0.4
-
0.28
3.2
83.3
0.73
0.32
82
19
e
Add.
2
10
0
-
0.13
8.7
53.9
N/M
N/M
N/M
19
f
N/A
N/A
0
-
0.13
14.0
51.0
N/M
0.23
79
N/A: not applicable
;
N/M: not measured
;
a)
Pre
-
PEC electrolyte measured for Cd by ICP
-
MS
and it is only
measured once prior to 1
st
PEC of each sample spot
;
b)
Post
-
PEC electrolyte measured for Cd by ICP
-
MS
;
c)
Post
-
PEC sample measured for Cd by X
RF
;
d)
Averaged J_eche current during the CA period
;
e)
First 1.2 hour of stability
test by H
-
cell from figure 3b
;
f)
Continued 3.3 hour of stability test by H
-
cell from
F
igure 3c