1
Breaking Scaling Relationships in CO
2
Reduction on Copper Alloys with Organic Additives
Yungchieh Lai,
#
Nicholas B. Watkins,
#
Alonso Rosas
-
Hernández, Arnaud Thevenon, Gavin P. Heim, Lan
Zhou,
Yueshen Wu, Jonas C. Peters,* John M. Gregoire,* Theodor Agapie*
Division of Chemistry and Chemical Engineering, California Institute of Technology, Pasadena, CA,
91125,
USA
.
#
These authors contributed equally to this work. *e
-
mail:
jpeters@caltech.edu
;
gregoire@caltech.edu
;
agapie@caltech.edu
Supporting Information
Table of Contents
Materials Synthesis
Electrochemist
ry
Experimental Uncertainty
Influence of Local pH
Impact of [CO]
Materials Characterization
Synthetic Procedures
Raw Data
Pairwise Product Distribution Analysis
C
2+
Pairwise Product Analysis
SEMs of Catalyst surface before and after catalysis with additive
SEM post
-
catalysis
without
additive
Zoomed in SEM and AFM of postcatalysis surfaces with additive
Time dependence of product distribution
XRD of Cu alloys at different concentrations
Alternate to Figure 3
CO
2
consumption and
product concentrations
References
2
2
-
3
3
3
4
5
5
6
-
7
7
-
8
9
10
1
1
1
2
-
1
3
1
4
1
5
-
1
7
1
8
18
-
19
20
2
Material Synthesis
Cu thin film electrocatalysts were fabricated using DC magnetron sputtering of a 2” Cu metal target at 50
W in 6 mTorr Ar
onto a 100 mm
-
diameter Si wafer with an approximately 170 nm SiO
2
diffusion barrier
and 10 nm Ti adhesion layer, using a previously described sputter system with 10
-
5 Pa base pressure.
1
After deposition, the films were stored in a nitrogen purge box until the day of electrochemical testing,
although no other catalyst treatment was performed prior to electrocatalyst screening. The Cu
-
X (X: Co,
Zn, Mn, In) thin film electrodes were deposi
ted under similar conditions from elemental metal targets
with DC power adjusted to obtain designed composition in the wafer center. All the metal targets were
pre
-
cleaned in the presence of 6 mTorr Ar for 10 min to remove any contaminants from the target
surface.
The non
-
confocal sputtering geometry provided a continuous composition gradient across the Si wafer
with the composition variation within each 5 mm diameter electrode being less than 1% for the most Cu
-
rich catalysts and about 2% for the most Cu
-
p
oor catalysts.
Electrochemistry
ANEC Analytical and Electro
-
chemistry (ANEC) is an analytical electrochemistry system previously
published by our group that can efficiently detects a wide range of CO
2
R product
.
2
This system is applied
in this study to further explore those Cu
-
X catalysts that are representative of the primary conclusions.
Prior to the electrolysis,
the electrolyte 0.1 or 0.2
5
M Potassium bicarbonate (>= 99.95% trace metals
basis) with or without 0.1
mM
1
-
Br
2
was purged with CO
2
(99.999%, Airgas) for at least 30 min. A bipolar
membrane (Fumasep® FBM single film, Fumatech) was used to separate the work
ing and counter
electrodes. Platinum wire (99.9%, Sigma Aldrich) was used as the counter electrode. The surface area of
the counter electrode was about 0.25 cm
2
, while the working electrode surface area was 0.32 cm
2
. The
working electrode chamber has heads
pace volume ~3.3
ml and electrolyte volume ~1.1
ml which is the
optimized ratio to maximize product concentration for detection.
2
Electrolysis was carried out with a
Gamry Reference 600™ potentiostat. The uncompe
nsated solution resistance was measured by
performing electrochemical impedance spectroscopy (EIS) in the frequency range of 100 Hz to 500 kHz
with an amplitude of 10 mV at the open circuit potential of a Pt
-
Pt Working Electrode
-
Counter Electrode
system. T
he uncompensated resistance, Ru, was measured by using a Nyquist plot of the EIS spectra and
was found to be 70 and 32 Ohms for 0.1 and 0.25
M KHCO
3
, respectively. All electrochemical data was
collected vs. a Ag/AgCl reference electrode (LF2, Innovative In
struments) and converted to a reversible
hydrogen electrode (RHE) scale. Prior to electrolysis, a constant potential at
-
1 V vs RHE (without IR
compensation) was conducted as pretreatment for each composition of the library. Since the co
-
sputtered
plate wa
s used as is (unlike Cu foil which would be polished prior to tests), such pre
-
electrolysis was
performed to 1) reduce any impurity oxide on the surface and 2) to pre
-
deposit additives on the surface.
To be consistent, 15 min CA was applied to all librarie
s tested. Electrolyses were then performed at
constant potentials (chronoamperometry) mostly between
-
0.9 to
-
1.3 V vs RHE (without IR
compensation). While electrolysis, the electrolyte was recirculated to quickly accumulate reaction
product for detection
at a flow rate ~150
uL/s. This high flow rate, compared to other flow cells reported
for CO2R (typically with 1
-
2 uL/s), creates an environment for less mass transport limitations.
3
It is noted
that any electrolysis tests associated with additive
electrolytes used is in the presence of 0.1
mM
1
-
Br
2
.
The duration for electrolysis typically ranging from 5 to 15 mins depends on t
he total current of each test
to maximize the concentration of reaction product while maintaining high throughput experimentation
and avoiding pH hikes in the case >5% of CO
2
in the headspace is consumed.
2
To ass
ure this varying
reaction durations will not change the product ratios, we do multiple tests on one Cu sample with different
3
CA durations and range it from 3 to 15min. The results show CO2R products grow linearly with time, for
example, Figure S11 shows [C
H
4
]/[C
2+
] remains constant at this time range. At the end of each electrolysis,
gaseous and liquid products were sampled by the robotic sample handling system (RSHS) and analyzed by
GC (Thermo Scientific™ TRACE™ 1300) and HPLC (Thermo Scientific UltiMate 3
000). Detailed product
detection (method) can be found at the previous publication.
2
The cell and all solution handling lines are
purged with fresh electrolyte and CO
2
between electrolysis to avoid cross
-
contamination. The actual
(compensated) potential shown in this manuscript was corrected with the uncompensated resistance Ru
measured above prior to further data analysis.
Experimental Uncertainty
Variation in the cur
rent during an electrolysis experiment leads to a variation in the compensated
resistance, and the standard deviation thereof is illustrated as horizontal error bars in Figure 2A. For
correlation analysis of partial current densities, since each pair of pa
rtial current densities result from the
same electrolysis, this variation in potential is negligible under the assumption that it does not span
multiple kinetic regimes. The uncertainty in partial current density has one contribution from the aliquot
and a
nalytic chemistry processes, which we characterize during chromatography calibration for each
product. The uncertainty is well modelled as a relative error in each measured concentration, which
corresponds to the same relative error in partial current dens
ities. For example, the relative error in CH4
quantification is 2.7%. The relative error in C
2+
quantification varies depending on the specific combination
of products and is between 2% and 7% for all electrolyses reported herein. For every partial current
density data point in Figures 2a, 2b, 2c, and 2f, the corresponding error bars are smaller than the marker
size. While this sampling error is negligible, more substantial sources of variability in measured partial
current densities may result from the imp
acts of turbulent flow, bubble occlusion of part of the working
electrode, etc., which are unquantified in the present work. Rather than perform many repetitions of a
single experiment to quantify this variability, we perform a breadth of experiments to be
tter characterize
the universality of the relationships, where a large unquantified uncertainty would obscure the
observation of correlations or other relationships; fortunately, this is not the case.
Influence of Local pH
Local pH is important for CO
2
R p
roduct distribution, however, our current cell is not capable of measuring
the pH at the electrode surface. It is particularly difficult to accurately measure the pH of the
microenvironment at the catalyst surface and would be a significant challenge to ta
ke these
measurements. However, given the rapid flow condition, a substantial pH change will be limited to the
diffusion layer and will be driven by the total current density. Therefore, an indirect study of any influence
of pH shift can be made by evaluat
ing whether the current density is related to the
observed product
ratio. The
Figure S
18
shows that the current density is not a primary determinant of the product ratio.
4
Impact of [CO]
on
CH
4
and C
2+
formation
S
ince *CO is understood to be the
precursor for both CH
4
and C
2+
products, as opposed to formic acid per
various mechanisms reported in the literature
,
we specifically investigate whether there is evidence of
molecular CO being a reactant for
CH
4
and C
2+
formation.
4,5
Figure S19
shows the corresponding partial
current densities as a function of the [CO], showing no systematic relationship and especially not a strong
positive correlation that would result from CO reduction
being a significant source of these products. In
the presence of the additive, the negative correlation between CO and both
CH
4
and C
2+
shows that the
competition for the common *CO intermediate is more prominent than CO reduction. In the time
-
dependent m
easurements
(Figure S11
)
, the proportionality of both
CH
4
and C
2+
with electrolysis time for
Cu further corroborates that CO reduction is insignificant
.
This finding is also intuitive in the context of the
concentration data shown
in
F
igure S19
, where a ~2
% maximum CO concentration corresponds to a partial
pressure of 0.02 atm of CO that equilibrates to 0.
02 atm * 9.5E
-
4 mol/L/atm = 19
μ
M, more than 1000×
less than the concentration of dissolved CO
2
(~32
mM)
.
With a given population of *CO that may or may
not be equilibrated with a local aqueous [CO], Figure 3A
illustrates that the reactions pathways fall into 3 important categories labelled by the resulting (measured)
products: CO, CH
4
and C
2+
. For a given catalyst and potential there is some branching rat
io for each of
these paths.
There is limit
ed
literature
regarding
whether the relative free energy barriers for these
products could be changed independently.
Since
CO production can vary drastically without a systematic
change in the o
ther
two
types of products
, there is an implicatio
n
that no free energy scaling relationship
exists between CO and either CH
4
or C
2+
products.
Those null results are equally important
observations
as the scaling relationship that we did identify, although each null result demonstrates that
th
ere is
not a
necessary
mechanism to break that (nonexistant) scaling relationships. Hence the focus
in this report
on
the CH
4
-
C
2+
relationship and its disruption via addit
ion of 1
-
Br
2
. Although we find no evidence of a CO
-
CH
4
or CO
-
C
2+
power law relationship, these products are linked through their common intermediates, which
is most evident in the presence of the additive where a negative correlation coefficient indicates the
kinetic competition for the intermediate.
5
Material Characterization
The bulk compositions of the Cu
-
X alloys were characterized via x
-
ray fluorescence (XRF, EDAX Orbis
MicroXRF). The composition of all the alloys screened is shown in table S1. Additional XRD characterization
shown in Figures S12
-
16.
Table S1:
Alloy composi
tions tested in ANEC cell for performance in CO
2
reduction.
Cu
CuMn
CuIn
CuZn
CuCo
–
26.5 : 73.5
–
29.5 : 70.5
–
–
48.5 : 51.5
–
48.7 : 51.3
–
–
–
–
79 : 21
–
–
84 : 16
83.5 : 16.5
87 : 13
83.5 : 16.5
–
97 : 3
–
95.8 : 4.2
96.7 : 3.3
–
–
–
97.0 : 3.0
–
100: 0
98 : 2
97.8 : 2.2
97.4 : 2.6
97.8 : 2.2
Synthetic Procedures
Synthesis of N,N’
-
ethylene
-
phenanthrolinium dibromide (
1
-
Br
2
) In a round bottom flask charged with a
magnetic stir bar, phenanthroline (500 mg, 2.8 mmol, 1 equiv.) was dissolved
in dibromoethane (5 mL,
67.4 mmol, > 24 equiv.) and the final mixture was heated to 110 °C for 18 h. The precipitate formed was
collected by filtration and washed with hexane (3 x 10 mL) and acetone (3 x 10 mL) to afford the final
product. Yield: 970 mg (9
4 %, 2.6 mmol). 1H and 13C NMR spectra were in accordance with reported
values
.
6
6
Raw Data
Figure S1
: Cu alloy product distribution i
n the absence of additives in 0.1M KHCO
3
. The error bars shown in the figure include
sampling/leak as well as the analytical instrument calibration errors.
Figure S2:
Cu alloy product distribution with 0.1 mM
1
-
Br
2
in 0.1
M KHCO
3
.
The error bars shown in
the figure include
sampling/leak as well as the analytical instrument calibration errors.
7
Figure S3:
Cu alloy product distribution in the absence of additives in 0.25
M KHCO
3
(pH=7.15). The error bars shown in the figure
include sampling/leak as well as
the analytical instrument calibration errors.
Figure S4:
a)
The data underlying the correlation analysis from 0.25
M KHCO
3
.
b)
C
ompar
ison for
the trend
s
between 0.1 and 0.25
M KHCO
3
. The latter shows a relatively gradual slope suggests an increased
selectivity toward methane at higher bicarbonate
concentrations.
8
Figure S5:
Visualization of the pairwise relationships in the current density (bottom
-
left) and FE (upper
-
right). Each data point
corresponds to a single catalyst composition and potential.
The pairwise relationships are shown for representative reaction
products, and in the current density plots the total cathodic current density is also shown.
9
Figure S6:
For the 5 prominent C
2
and C
3
products, the 10 pairwise relationships of the partial
current densities (mA cm
-
2
) are
shown, illustrating a high degree of correlation among these products, both in the presence and absence of the additive, whic
h
is expected given common initial pathways for formation of each product.
10
Figure S7:
SE
M of catalysts before (left column) and after (right column) catalysis with molecular additives. a) CuZn samples b)
CuMn samples c) CuIn samples d) CuCo samples. Due to being deposited on SiO
2
disks, charging of the surface with SEM was
notable. In the sec
ond column with molecular additives, the dark charging regions correspond to additive on the surface. No
significant surface restructuring was observed for any catalyst tested. Any texture observed in the right column corresponds
to
thicker regions of the
film on the surface of the catalyst.
11
Figure S8:
SEM of Cu catalyst after catalysis without molecular additives
-
no notable nanostructuring is observed.
12
Figure S9:
a) Precatalysis SEM of CuMn shows a grain size of 34.2 ± 18.5 nm. b) Zoomed in
postcatalysis SEM of CuMn with
additive shows there is a film on the surface and the surface structure underneath remains unchanged. c) AFM of precatalysis
CuMn with grain size of approximately 40.5 nm. d) Postcatalysis AFM of CuMn with additive shows aggl
omerated film on surface,
as shown by the increase in magnitude of the scale bar. e) AFM of CuMn postcatalysis surface with additive after washing off
film
shows consistent grain size with the precatalysis surface of approximately 38.7 ± 14.1 nm.
13
Figure S10:
a) Precatalysis SEM of CuIn shows a grain size of 50.2 ± 17.3 nm. b) Zoomed in postcatalysis SEM of CuIn with additive
shows there is a film on the surface and the surface structure underneath remains unchanged. c) AFM of precatalysis CuIn ha
s a
grain size of approximately 35 nm. d) Postcatalysis AFM of CuIn with additive shows agglomerated film on surface, as shown by
the increase in magnitude of the scale bar. e) AFM of CuIn postcatalysis surface with additive after washing off film shows
co
nsistent grain size with the precatalysis surface of approximately 37.7 ± 16.0 nm.
14
Figure S11:
Resulting CH
4
vs C
2+
from CA at
-
1.04
V vs RHE with different durations on Cu in 0.1
M KHCO
3
. The testing sequence
was 5, 3, 10, 15, and then 5 mins. The repea
ted 5 min CA experiment was conducted to check for variation after multiple
experimental runs at different duration. The lack of variation at 5 min suggests that the electrode performs consistently and
reproducibly over time
15
Figure S12:
XRD for the as
-
s
ynthesized alloys investigated. All compositions shown are alloyed cubic Cu structure and are with
space group of Fm
-
3m. For In, Mn, Zn alloys, peaks shift to smaller 2
-
theta (larger d
-
spacing) with increasing alloy content. For
Co, peaks shift to larger 2
-
theta (smaller d
-
spacing) with increasing Co.
16
Figure S13:
XRD for alloy CuCo before and after electrolysis.
Blue: pristine, red: absence of additiv
e
, and black: presence of
additive. The slight shift of peak position is due to slight sample composition
variation from sample to sample and is estimated
to be < 1% for the Cu
-
rich catalysts and < 2% for the Cu
-
poor catalysts.
Figure S14:
XRD for alloy CuZn before and after electrolysis.
Blue: pristine, red: absence of additiv
e
, and black: presence of
additive. The slight shift of peak position is due to slight sample composition variation from sample to sample and is estim
ated
to be < 1% for the Cu
-
rich catalysts and < 2% for the Cu
-
poor catalysts.
17
Figure S15:
XRD for alloy CuMn before and after ele
ctrolysis.
Blue: pristine, red: absence of additiv
e
, and black: presence of
additive. The slight shift of peak position is due to slight sample composition variation from sample to sample and is estim
ated
to be < 1% for the Cu
-
rich catalysts and < 2% for
the Cu
-
poor catalysts.
Figure S16:
XRD for alloy CuIn before and after electrolysis.
Blue: pristine, red: absence of additiv
e
, and black: presence of
additive. The slight shift of peak position is due to slight sample composition variation from sample
to sample and is estimated
to be < 1% for the Cu
-
rich catalysts and <2% for the Cu
-
poor catalysts.
18
Figure S17:
A complementary figure for Figure 3 in the main text.
Figure S18:
E
lectrochemical current density, CO
2
consumption (in the headspace) at the end of electrolysis, gas product
concentration, and liquid product concentration vs log (C
2+
/CH
4
). Solid symbol: electrolyses with additive
;
hollow symbol:
electrolyses without additive in 0.1
M KHCO
3
.