The sensitivity of Cu for electrochemical carbon dioxide reduction to hydrocarbons as revealed by high throughput experiments

Article abstract of DOI:10.1039/c9ta10111j

Electrochemical CO₂ reduction to valuable products is a centerpiece of future energy technologies that relies on identification of new catalysts. We present accelerated screening of Cu bimetallic alloys, revealing remarkable sensitivity to alloy concentration that indicates the migration of alloying elements to critical sites for hydrocarbon formation.

Full text of DOI:10.1039/c9ta10111j

Journal of
Materials Chemistry A
View Article Online
View Journal | View Issue
COMMUNICATION
The sensitivity of Cu for electrochemical carbon
dioxide reduction to hydrocarbons as revealed by
high throughput experiments†
Cite this: J. Mater. Chem. A, 2019, 7,
6785
2
Received 12th September 2019
Accepted 12th November 2019
a
a
a
a
Yungchieh Lai,
Ryan J. R. Jones,
Yu Wang,
Lan Zhou,
a
ab
Matthias H. Richter
and John Gregoire
*
DOI: 10.1039/c9ta10111j
rsc.li/materials-a
Electrochemical CO
2
reduction to valuable products is a centerpiece
Exploration of CO RR electrocatalysts is hampered by the
2
of future energy technologies that relies on identification of new dearth of accelerated screening systems, which are challenging
catalysts. We present accelerated screening of Cu bimetallic alloys, to realize given the sensitivity of product distribution to the
revealing remarkable sensitivity to alloy concentration that indicates electrochemical environment and the breadth of possible
the migration of alloying elements to critical sites for hydrocarbon reaction products. The ideal characterization of a given catalyst
formation.
is quantitative measurement of the activity and product distri-
bution, as a function of time and for a large range of over-
potentials since product distributions vary with aging of the
operational catalyst as well as the operating potential. This
combination of desired measurement attributes is incompat-
2
The promise of electrochemical CO reduction to valuable
products has prompted substantial research in selective catal-
ysis of the many-electron reduction of CO to products such as
light hydrocarbons. However, most metal electrocatalysts have
been shown to exclusively form the 2 e products CO and
formate (e.g. Au, Ag, In, and Sn) and/or catalyze the competing
hydrogen evolution reaction (HER) (e.g. Fe, Ni, Pt, and Ti). Cu is
2
the only known elemental catalyst capable of reducing CO to
hydrocarbons (HCs) and alcohols with appreciable faradaic
efficiency (FE) but suffers from poor selectivity to any single
product and high overpotentials for generating the highly
reduced products. The uniqueness of Cu has been under-
stood to be due to its moderate binding energy of CO and other
intermediates. Alloying with a secondary metal is a widely-
adopted strategy to alter a catalyst's adsorption energetics for
reactants and intermediates, and some copper-based bimetallic
3
2
ible with traditional methods, perhaps most notably due to the
1
duration of quantitative product distribution measurements via
chromatography or nuclear magnetic resonance, which severely
limits the measurement frequency to observe variations in
product distribution as a function of time and/or overpotential.
Electrochemical reactor with online mass spectroscopy to
detect reaction products has been developed in experimental
À
2
techniques such as differential electrochemical mass spec-
23–25
trometry (DEMS)
time” CO
and has recently has deployed for “real-
2
–4
2
RR product detection with scanning electrochemical
22,26,27
24,28
cells.
dual thin-layer DEMS ow cell that provides excellent charac-
terization of individual CO RR electrocatalysts with quantied
Clark et al.
introduced a modied traditional/
5,6
2
product. We combine the scanning cell concept for combina-
torial measurements with a DEMS-style product detection
system that is decoupled from the electrochemical cell to
mitigate interference of product detection in the electro-
chemical reactor. By additionally introducing selective differ-
ential pumping of water vapor in vacuum, we demonstrate
7
catalysts have been evaluated for altering the CO
While improved selectivity for 2 e products has been achieved,
success in increasing FE for higher order products has been
2
RR selectivity.
À
8
–22
more limited, as discussed further below.
Comparison
among reported Cu-alloy systems is limited by the variation in
catalyst syntheses and morphologies in these studies, moti-
vating the systematic study of different alloying elements and
alloying concentrations in the present work.
quantitative faradaic efficiency (FE) measurement of H
2 4
, CH ,
and C during cyclic voltammetry (CV) experiments with
2 4
H
a measurement interval of 12 s, enabling rapid screening of
arrays of catalyst compositions.
a
Through exploration of the ASM binary alloy phase
Joint Center for Articial Photosynthesis, California Institute of Technology,
29
Pasadena, CA 91125, USA. E-mail: gregoire@caltech.edu
b
diagrams we identied 8 non-precious metals that alloy into
ꢀ
Division of Engineering and Applied Science, California Institute of Technology, fcc-Cu by at least 10 at% for some temperature below 700 C.
Pasadena, CA 91125, USA
Binary compositionally-graded thin lms of each Cu–X system
with compositions range from below 3 at% to above 18 at% of
†
Electronic supplementary information (ESI) available: Additional experimental
description and alternate visualizations of data. See DOI: 10.1039/c9ta10111j
This journal is © The Royal Society of Chemistry 2019
J. Mater. Chem. A, 2019, 7, 26785–26790 | 26785

View Article Online
Journal of Materials Chemistry A
Communication
the alloying element were prepared by combinatorial sputtering
Given the excellent partial current density detectability limit,
and measured at 8 alloy compositions, creating a library of 64 the MS signals can be directly used for identifying catalysts and
alloys (Table S1†) that were each evaluated by a series of 3 CVs electrolysis settings that yield HCs. For the present purposes of
using the instrument shown in Fig. 1. The electrolyte ow at identifying whether Cu alloys alter FE for HCs, especially at lower
À1
0
.16 mL s through a catholyte chamber of ca. 0.32 mL creates overpotentials, a CV is the desired electrochemical technique, but
a liquid-only electrochemical reactor except during sustained appreciable potential scanning rates and/or potential-dependent
production of products beyond their solubility limit, e.g. pro- partial current densities result in the inability of the MS signal to
À2
longed partial current density above ca. 5 mA cm for gaseous reach steady state. We established a model for calculating the
products. The electrolysis effluent ows through a pervaporator time-dependent partial current density of each product from the
cell with poly(tetrauoroethylene) (PTFE) membrane, where the respective measured time-dependent MS signal, as described
31
À1
vacuum contact promotes evaporation of volatile dissolved previously. For CVs at 10 mV s between un-compensated
species such as H , CH , and C H . The concomitant evapora- potentials of À0.4 and À1.3 V vs. RHE, we nd that J is well
2
4
2
4
p
tion of water requires substantial differential pumping, tradi- modelled by assuming that J
p
is the same function of applied
tionally by a mechanical pump that also discards the vast potential during both the cathodic and anodic voltage sweeps.
majority of the reaction products. This issue is circumvented in Furthermore, this function can be approximated by a Gaussian,
the present work with an in-line molecular sieve (mol. sieve) which represents a peak in J when the center potential of the
p
desiccant chamber that sorbs water and alcohols to increase the Gaussian is within the CV limits and could also approximate an
partial pressure of HCs in the MS inlet by a factor of approxi- exponential function when the center potential is beyond the CV
31
mately 100 (Fig. S2†). As a result, with 12 s MS measurement limits. Consequently, the series of MS measurements per product
intervals, HCs can be detected at partial current densities (J ) and per CV cycle are used to determine the corresponding 3
p
À2
above 30 mA cm . The MS signals at mass-to-charge (m/z) Gaussian parameters that minimize the loss between the
values of 2, 15, and 26 provide quantitative measurement of the modelled and measured MS signal. Fig. S3–S5† shows the
partial pressure of H , CH , and C H in the post-desiccant stability of the measurement and of this J modelling algorithm
2 4 2 4 p
vacuum effluent, whose signals are calibrated by saturating through repeated measurements on a series of Cu thin lm
the owing electrolyte with calibration gases with the cell held catalysts. The product selectivity and partial current shown
at open circuit. While other strategies for online product agrees well with those of polycrystalline Cu reported in the
3,30
detection have enabled accelerated screening of electro- literature. Fig. 2a illustrates the measurement of 3 MS chan-
22,28,30,32
catalysts,
we note a unique aspect of the present work is nels and working electrode current density over 3 CV cycles with
that the rapid ow and decoupling of the reactor and perva-
porator ensure that products created from the entire working
electrode are detected, as opposed to localized product extrac-
tion where spatial variations in mass transport, etc. can cause
the measured product stream to poorly represent the electro-
chemical measurement, creating artifacts in the inferred FE, etc.
Fig. 1 Electrolyte (blue) and in-vacuum product stream (green) flows
through the 3 components between the electrolyte reservoir and MS
detector. The planar working electrode is pressed against the o-ring at
the bottom of the electrochemical cell. The inlet flow tube is directed
toward the working electrode (WE), and the outlet flow tube termi-
nates near the membrane at the top of the cell so that any bubbles are Fig. 2 EC-MS demonstration with a thin film Cu electrode. (a) Typical
readily removed. The reference electrode (RE) is horizontal, termi- experimental process for catalyst screening including 3 CV cycles
nating near the center of the chamber, ca. 2.8 mm above the WE, and separated by OCP for enabling MS to return to baseline. After per-
the counter electrode (CE) is separated by the bipolar membrane forming such measurements on a series of catalysts, a daily calibration
(
BPM). The electrolyte is then injected in the center of a cylindrical is performed by flowing through electrolyte bubbled with calibration
pervaporation chamber with volatile species crossing the PTFE gas containing 0.4% H , 0.4% CH , and 0.4% C in CO . (b) Using this
2
4
2
H
4
2
membrane, creating a product stream that is differentially pumped by calibration and pervaporation model, the modeled partial current
the mol. sieve before continuous injection in the MS for product density for each species is calculated and compared to the measured
quantification.
total current.
26786 | J. Mater. Chem. A, 2019, 7, 26785–26790
This journal is © The Royal Society of Chemistry 2019

View Article Online
Communication
Journal of Materials Chemistry A
60 s of open circuit potentiometry (OCP) in between to allow MS differences from the Cu-only catalyst are quite remarkable. The
signals to return to baseline so that each CV can be modelled gure also indicates substantial variation in the total current
independently. Fig. 2b shows the J resulting from the 3-param- density, which is shown for these compositions in Fig. 4a. To
p
eter Gaussian t using a custom tting procedure described further probe the composition-dependent product distribu-
31
previously. The MS signal corresponding to this t is also shown tions, we consider the integral over time of this total current
in Fig. 2a. It is worth noting that the resulting J is quite similar density as well as over each product-specic J , yielding elec-
on the second and third cycles, which can differ from that of the trolysis charge densities for the entire CV. The results shown in
rst cycle. This variability in the rst CV is expected since no Fig. 4b include a dashed line indicating the average over the Cu
electrochemical pretreatment was performed on the sputtered measurements in Fig. S4.†
lm and is likely related to contaminants and/or oxidation from To unpack this wealth of data we consider alloying elements
p
p


air exposure. As a result, the evolving product distribution during that behave qualitatively similar and discuss in the context on
the rst cycle exceeds the scope of the Gaussian-based model the Cu alloy literature, which mostly involves higher concen-
described above and corresponds to substantial uncertainty in trations of the alloying element. The lack of literature on dilute
p
the precise shape of J in Fig. 2b. While additional modelling can alloys of fcc-Cu is emblematic of the intuition that small
be used to explore this rst cycle, for the purposes of the present concentrations of alloying elements will have no substantial
work we note that for all alloy catalysts the results of the second impact on the catalysis, yet Fig. 4b shows that 2–3 at% of Sb, In
and third cycle are consistent, motivating our analysis and or Sn is sufficient to lower total HC production during the CV by
reporting of only the third cycle. The measured current is also a factor of 10 or more, a level of sensitivity to an alloying
smoothed with a 129-point cubic Savitzky–Golay lter and then element that to our knowledge has no precedent in electro-
compensated for resistive losses using the measured 60 U cell catalysis when the contaminant is an alloying element that is
resistance. Uncompensated versions of each such gure are itself an electrocatalyst, as opposed to contaminants such as S
33,34
included in the ESI.†
and Cl that readily poison metal surfaces.
It is worth noting
Fig. 3 shows the MS measurements and subsequent model-
ling for the 8 alloy composition systems along with that of
a pure Cu reference. Catalysts exhibiting current densities in
À2
excess of 5 mA cm for a substantial fraction of each CV cycle
typically causing bubble formation that gives rise to the large
positive outlier values in the MS measurements. While such
data points cause some uncertainty in the resulting J , the
p
modelling procedure mitigates sensitivity to the outlier and
preferably weights the data points at the onset of product
31
detection.
Complete datasets for 8 alloy systems elements are provided
as Fig. S6–S13,† and 3 includes the compilation of the
Cu0.95X0.05 compositions. A range of potential-dependent FE
signals for each product are observed, and the observed
Fig. 4 For each of 8 alloy elements, the (a) total current density and (b)
charge density (integral of J over time(s)) for the third CV cycle is
p
(left column) and FE (right column) for the third CV cycle shown as a function of alloy concentration. The dashed black line
p
Fig. 3 The J
is shown for each of the 8 Cu-based bimetallic catalysts with ca. 5% of indicates the average charge density for pure Cu. The bottom 3 panels
the alloying metal. The potentials include resistance compensation include a broken charge density axis corresponding to product
and legends in bottom-right panel apply to all 6 panels.
generation near or below the detectability limit.
This journal is © The Royal Society of Chemistry 2019
J. Mater. Chem. A, 2019, 7, 26785–26790 | 26787

View Article Online
Journal of Materials Chemistry A
Communication
that in the CV experiments, high partial current density for non- motivates further inspection of whether Zn directly plays
HC products mitigates the ability to form HCs, not only due to a catalytic role in increasing the FE of C H or if it instead
2
4
the competitive concentrations of reaction intermediates, but stabilizes the hypothesized unique Cu sites generated through
38
also due to resistive losses as reected in the range of applied electrochemical reduction from the metal oxide.
potentials aer resistance compensation shown in Fig. 4a. While the 2 at% Co alloy enhances HC formation by a factor
Sb, In, and Sn alloys exhibit a substantial increase in the of about 2, Co is otherwise the most innocuous alloying element
current density with the FE for H being relatively unchanged, with no substantial change in the overall current and slightly
indicating that the suppression in HC formation is commen- increasing H and decreasing HC production with increasing
2
2
surate with a substantially increased activity for other products, alloying concentration. Similar trends of HC formation over
À
which were conrmed to be 2 e products CO and formate via CuCo was observed by Grote and his coworkers where max
22
detailed experiments on select catalysts (see ESI†). These results production was found at Co concentration <10%. Al is also
9–11
14
are in agreement with previous reports on Cu–In and Cu–Sn
somewhat innocuous with the primary effect being increase H2
catalysts where the weak H adsorption and strong O binding of formation, which is markedly different from the heavier main
In and Sn enhances selectivity toward formate and CO. For group elements discussed above.
example, FE of 90% for formate with Cu–Sn and for CO with Cu–
Other than the overall sensitivity of Cu to small alloying
In were reported at low potentials, with the present work being concentrations, perhaps the most intriguing observation is the
the rst to explore more dilute alloys of these elements. While select cases where the alloying element changes the relative
further study of the highly active Cu–(Sb, In, Sn) catalysts may production of CH and C H , which is shown in Fig. S14.† The
4
2
4
be of interest for formation of products other than HCs, their most Cu-rich alloys with Zn, Al, and Co slightly increase selec-
primary scientic importance for CO RR is understanding how tivity for CH . The 2–5 at% Sb catalysts are interesting in that
2
4
such a small concentration can so strongly inuence product they eliminate CH
distribution, with Cu–Sb affecting CH and C differently (see possibility that improved catalysts could be obtained if this
below). In the concerted community effort to understand the selectivity control could be harnessed without the increased FE
uniqueness of Cu, these alloys provide a new line of inquiry. in other products. This intriguing result was validated with
4 2 4
from detectability but not C H , posing the
4
2 4
H
The Cu–Zn alloys share some important similarities and more detailed experiments on the 2% Sb sample (see ESI†),
differences compared to the Sb, In, and Sn systems. The HC which validated the observation of major CH suppression with
4
formation is less sensitive to alloy concentration in the 2–6 at% less C H suppression and additionally revealed the CO is the
2
4
range. Catalysts with higher Zn concentrations exhibit the primary contributor to the “Other charge” in Fig. 4. A similar
highest H production among all the alloys, which not only but more gradual trend is observed with Mn (and to a lesser
hampers formation of HCs but also other CO RR products. extent Ni) as relative C to CH production increases with
These results are somewhat surprising given literature prece- increasing alloy concentration, which was conrmed by addi-
2
2
2
H
4
4
35
dent of Cu–Zn catalysts. Zn-rich alloy lms and nanoporous tional experiments (see ESI†). While the total HC production of
18
Zn-rich structures have been studied for formate and CO Ni also declines with increasing alloy concentration, the Mn
production at low overpotential, which are not directly compa- alloys up to 5 at% show slightly higher total HC production,
rable to the Zn-poor catalysts in Fig. 4 and collectively indicate with the 3–4 at% Mn increasing the fraction of C2 HCs
that product distributions of Cu–Zn catalysts do not vary (compared to total HCs) from the ca. 40 at% value observed with
monotonically with composition. Zn coatings on Cu have been pure Cu to ca. 60 at%.
36
shown to suppress HER and increase CH
4
production, in stark
The alloys with 5 at% Cu–Sn and Cu–Mn were measured with
contrast to the results of Fig. 4, indicating that the surface and XPS (Fig. S19†) to assess whether the near-surface concentra-
near-surface Zn in the alloy catalysts of the present work are tions of the alloying elements far exceeded the bulk values, and
critically different from the previously reported application of while enhancements of 1.6Â and 2.8Â, respectively, were
37
Zn coatings. Ren et al. studied electrodeposited Cu–Zn alloys observed, the sensitivity of product distribution to alloy
with Zn concentrations of 9–30 at% and focused on the concentration remains remarkable and largely unexplained by
increased ethanol production at high Zn concentrations. They present understanding of CO RR on Cu-based catalysts. Theo-
2
also observed increased total current with Cu–Zn alloys retical characterization of absorption energies for bimetallic
39
compared to Cu at potentials above À0.95 V vs. RHE, in agree- alloys will help elucidate the composition-selectivity trends,
ment with the present work, although the most Cu-rich alloy in although such data is not presently available in the dilute alloy
2 5
that work (9 at% Zn) exhibited an increase in FE of C H OH, concentration range and will additionally need to consider
17
which is not measured by our study. Feng et al. observed elemental surface segregation and reorganization processes.
enhanced FE for C H on oxide-derived Cu–Zn nanoparticles Since it is unlikely that such dilute alloys substantially alter the
2
4
where Zn concentrations from 12 to 50 at% were studied, electronic structure of Cu, the most likely underlying
resulting in identication of 20 at% Zn as optimal for C H phenomenon is that the HC-forming Cu sites have a relatively
2
4
formation. In this context, the results of Fig. 4 indicate that the high surface energy that concomitantly drives surface alloying
morphology and exposed catalyst sites from the oxide-derived elements to these same sites, enabling a dilute concentration of
synthesis technique is more critical than the presence of Zn, these active sites to be contaminated by dilute alloy elements.
38
which is commensurate with recent studies on product- Validating this hypothesis will require further experimentation
specic active sites in oxide-derived Cu catalysts and and/or theoretical investigation, and the notable exceptions to
26788 | J. Mater. Chem. A, 2019, 7, 26785–26790
This journal is © The Royal Society of Chemistry 2019

View Article Online
Communication
Journal of Materials Chemistry A
this trend, such as unique product distributions of Sb and Mn-
based alloys, merit further investigation. The validation of the
accelerated screening results with more detailed and traditional
methods demonstrates the utility of the screening platform for
catalyst discovery. These results also represent a cautionary tale
7 A. Vasileff, C. Xu, Y. Jiao, Y. Zheng and S.-Z. Qiao, Chem Rev.,
2018, 1809–1831.
8 J. Christophe and T. Doneux, Electrocatalysis, 2012, 3, 139–
146.
9 A. Jedidi, S. Rasul, D. Masih, L. Cavallo and K. Takanabe, J.
Mater. Chem. A, 2015, 3, 19085–19092.
2
for development of CO RR electrolysis devices as any plating of
metal contaminants in the electrolyte can substantially alter 10 Z. B. Ho, T. S. Gray, K. B. Moraveck, T. B. Gunnoe and
product distributions, creating demanding requirements for G. Zangari, ACS Catal., 2017, 7, 5381–5390.
elimination of trace metal contaminants. A notable precedence 11 G. O. Larrazabal, A. J. Martin, S. Mitchell, R. Hauert and
for this level of sensitivity to trace metals exists in alkaline J. Perez-Ramirez, ACS Catal., 2016, 6265–6274.
oxygen evolution electrocatalysis where trace adventitious Fe 12 S. Rasul, D. H. Anjum, A. Jedidi, Y. Minenkov, L. Cavallo and
from the electrolyte substantially inuenced catalytic activity for K. Takanabe, Angew. Chem., Int. Ed., 2015, 54, 2146–2150.
initially Fe-free catalysts. While the activity boost in that 13 J. He, K. E. Dettelbach, D. A. Salvatore, T. Li and
́
40
situation posed problems for scientic studies but not electrode
C. P. Berlinguette, Angew. Chem., Int. Ed., 2017, 56, 6068–
development, the sensitivity of Cu to metal impurities poses
6072.
more substantial challenges for electrocatalyst research and 14 M. Morimoto, Y. Takatsuji, R. Yamasaki, H. Hashimoto,
development.
I. Nakata, T. Sakakura and T. Haruyama, Electrocatalysis,
018, 9, 323–332.
2
1
5 M. Li, J. Wang, P. Li, K. Chang, C. Li, T. Wang and B. Jiang, J.
Mater. Chem. A, 2016, 4, 4776–4782.
Conclusions
We address challenges in CO2 reduction electrocatalyst 16 D. Ren, B. S.-H. Ang and B. S. Yeo, ACS Catal., 2016, 6, 8239–
discovery through the design of an electrochemical ow cell 8247.
with online mass spectroscopy-based quantication of the 17 Y. Feng, Z. Li, H. Liu, C. Dong, J. Wang, S. A. Kulinich and
faradaic efficiency for H , CH , and C with 12 s measure- X. Du, Langmuir, 2018, 34, 13544–13549.
2
4
2 4
H
ment intervals and partial current detectability of ca. 0.03 mA 18 H. Hu, Y. Tang, Q. Hu, P. Wan, L. Dai and X. J. Yang, Appl.
À2
cm , enabling high throughput screening of Cu bimetallic
Surf. Sci., 2018, 445, 281–286.
alloys with Al, Mn, Co, Ni, Zn, In, Sn, and Sb. The product 19 G. Yin, H. Abe, R. Kodiyath, S. Ueda, N. Srinivasan,
distribution of Cu-rich catalysts is remarkably sensitive to alloy
concentration with 2 at% of some elements causing tenfold
A. Yamaguchi and M. Miyauchi, J. Mater. Chem. A, 2017, 5,
12113–12119.
decrease of hydrocarbon formation, indicating that alloying 20 G. Keerthiga and R. Chetty, J. Electrochem. Soc., 2017, 164,
elements segregate to critical sites for hydrocarbon formation, H164–H169.
which poses substantial challenges for development and 21 E. L. Clark, C. Hahn, T. F. Jaramillo and A. T. Bell, J. Am.
sustainable operation of Cu-based catalysts. Specic alloy
concentrations of Sb and Mn increase C relative to CH
providing a new direction for catalyst development.
Chem. Soc., 2017, 139, 15848–15857.
22 J. P. Grote, A. R. Zeradjanin, S. Cherevko, A. Savan,
B. Breitbach, A. Ludwig and K. J. J. Mayrhofer, J. Catal.,
2
H
4
4
,
2
016, 343, 248–256.
3 H. Baltruschat, J. Am. Soc. Mass Spectrom., 2004, 15, 1693–
706.
24 E. L. Clark, M. R. Singh, Y. Kwon and A. T. Bell, Anal. Chem.,
015, 87, 8013–8020.
2
Conflicts of interest
1
There are no conicts to declare.
2
2
5 S. J. Ashton, Design, construction and research application of
a Differential Electrochemical Mass Spectrometer (DEMS),
Springer Science & Business Media, 2012.
Notes and references
1
2
3
4
P. D. Luna, C. Hahn, D. Higgins, S. A. Jaffer, T. F. Jaramillo
and E. H. Sargent, Science, 2019, 364, eaav3506.
Y. Hori, H. Wakebe, T. Tsukamoto and O. Koga, Electrochim.
Acta, 1994, 39, 1833–1839.
K. P. Kuhl, E. R. Cave, D. N. Abram and T. F. Jaramillo, Energy
Environ. Sci., 2012, 5, 7050–7059.
26 J. P. Grote, A. R. Zeradjanin, S. Cherevko and
K. J. J. Mayrhofer, Rev. Sci. Instrum., 2014, 85, 104101.
27 P. Khanipour, M. Lçffler, A. M. Reichert, F. T. Haase,
K. J. J. Mayrhofer and I. Katsounaros, Angew. Chem., Int.
Ed., 2019, 131, 7273–7277.
S. Nitopi, E. Bertheussen, S. B. Scott, X. Liu, A. K. Engstfeld, 28 E. L. Clark and A. T. Bell, J. Am. Chem. Soc., 2018, 140, 7012–
S. Horch, B. Seger, I. E. L. Stephens, K. Chan, C. Hahn,
7020.
J. K. Nørskov, T. F. Jaramillo and I. Chorkendorff, Chem. 29 ASM Alloy Phase Diagram Database™ – ASM International,
Rev., 2019, 119, 7610–7672.
K. P. Kuhl, T. Hatsukade, E. R. Cave, D. N. Abram,
https://www.asminternational.org/home/-/journal_content/
56/10192/15469013/DATABASE, (accessed June 25, 2019).
J. Kibsgaard and T. F. Jaramillo, J. Am. Chem. Soc., 2014, 30 E. L. Clark, M. R. Singh, Y. Kwon and A. T. Bell, Anal. Chem.,
36, 14107–14113. 2015, 87, 8013–8020.
A. A. Peterson and J. K. Nørskov, J. Phys. Chem. Lett., 2012, 3, 31 Y. Lai, R. J. R. Jones, Y. Wang, L. Zhou and J. M. Gregoire,
51–258. ACS Comb. Sci., 2019, 21, 692–704.
5
6
1
2
This journal is © The Royal Society of Chemistry 2019
J. Mater. Chem. A, 2019, 7, 26785–26790 | 26789

View Article Online
Journal of Materials Chemistry A
Communication
3
3
3
3
2 K. Jambunathan, S. Jayaraman and A. C. Hillier, Langmuir, 36 G. Keerthiga and R. Chetty, J. Electrochem. Soc., 2017, 164,
004, 20, 1856–1863. H164–H169.
3 A. F. Ghenciu, Curr. Opin. Solid State Mater. Sci., 2002, 6, 37 D. Ren, B. S.-H. Ang and B. S. Yeo, ACS Catal., 2016, 6, 8239–
89–399. 8247.
4 N. Job, M. Chatenet, S. Berthon-Fabry, S. Hermans and 38 Y. Lum and J. W. Ager, Nat. Catal., 2019, 2, 86.
2
3
F. Maillard, J. Power Sources, 2013, 240, 294–305.
39 O. Mamun, K. T. Winther, J. R. Boes and T. Bligaard, Sci.
Data, 2019, 6, 76.
5 G. Yin, H. Abe, R. Kodiyath, S. Ueda, N. Srinivasan,
A. Yamaguchi and M. Miyauchi, J. Mater. Chem. A, 2017, 5, 40 L. Trotochaud, S. L. Young, J. K. Ranney and S. W. Boettcher,
2113–12119. J. Am. Chem. Soc., 2014, 136, 6744–6753.
1
26790 | J. Mater. Chem. A, 2019, 7, 26785–26790
This journal is © The Royal Society of Chemistry 2019