Research Article

 [Link]

Tuning the C1/C2 Selectivity of Electrochemical CO2

Reduction on Cu–CeO2 Nanorods by Oxidation
State Control
Seungwon Hong, Hafiz Ghulam Abbas, Kyuseon Jang, Kshirodra Kumar Patra,
Beomil Kim, Byeong-Uk Choi, Hakhyeon Song, Kug-Seung Lee, Pyuck-Pa Choi,*
Stefan Ringe,* and Jihun Oh*

drawing attention as a promising tech-

Ceria (CeO2) is one of the most extensively used rare earth oxides. Recently, nology to mitigate excess anthropogenic
it has been used as a support material for metal catalysts for electrochemical CO2 emissions because it can convert
energy conversion. However, to date, the nature of metal/CeO2 interfaces and carbon dioxide into high-value-added
their impact on electrochemical processes remains unclear. Here, a Cu–CeO2 chemicals and fuels in an eco-friendly and
economical way.[2] Among suitable metals,
nanorod electrochemical CO2 reduction catalyst is presented. Using operando Cu (and its compounds) is the only one
analysis and computational techniques, it is found that, on the application capable of producing high value-added
of a reductive electrochemical potential, Cu undergoes an abrupt change in C2+ products, such as C2H4 and C2H5OH,
solubility in the ceria matrix converting from less stable randomly dissolved with high selectivity through CO dimeri-
single atomic Cu2+ ions to (Cu0,Cu1+) nanoclusters. Unlike single atomic Cu, zation.[3] However, the C2+ selectivity of
Cu-based electrocatalysts can be further
which produces C1 products as the main product during electrochemical CO2
improved by tailoring the active sites of
reduction, the coexistence of (Cu0,Cu1+) clusters lowers the energy barrier for Cu through doping,[4] tuning the catalyst
C–C coupling and enables the selective production of C2+ hydrocarbons. As a particle shapes and sizes,[5] and increasing
result, the coexistence of (Cu0,Cu1+) in the clusters at the Cu–ceria interface the number of low coordination sites, for
results in a C2+ partial current density/unit Cu weight 27 times that of a cor- example, the grain boundary density.[6]
responding Cu-carbon catalyst under the same conditions. In particular, the use of mixed oxidation
states (i.e., Cu+ and Cu0) has been sug-
gested as an effective means to achieve
high C2+ product selectivity by lowering
1. Introduction the energy barrier for CO dimerization.[7] However, Cu+ is not
stable and is easily converted to metallic Cu (Cu0) at the nega-
The continuous emission of greenhouse gases, including tive applied potentials used for the CO2RR. As a result, many
carbon dioxide, is causing changes to the climate that threaten strategies have been developed to sustain the mixed oxidation
human life.[1] Therefore, electrochemical carbon dioxide reduc- state of Cu during the CO2RR.[8] For instance, doping Cu with
tion reaction (CO2RR) technology using renewable energy is B atoms can induce the formation of Cu+ under CO2RR condi-
tions, and increase its stability.[4a,9] Nevertheless, it is still chal-
S. Hong, K. Jang, K. K. Patra, B. Kim, B.-U. Choi, H. Song,
lenging to achieve controllable Cu+ coverage at the catalytically
P.-P. Choi, J. Oh active Cu top-surface.
Department of Materials Science and Engineering Ceria (CeO2) is a reducible metal oxide and an excellent sup-
Korea Advanced Institute of Science and Technology (KAIST) port for metal catalysts in many catalytic processes, such as the
Daejeon 34141, Republic of Korea water-gas shift reaction,[10] CO oxidation,[11] and CO2 hydrogena-
E-mail: [Link]@[Link]; [Link]@[Link]
tion reaction.[12] As a catalyst support, CeO2 can disperse metals
H. G. Abbas, S. Ringe
Department of Chemistry at various scales uniformly, from the nanometre in scale to
Korea University single atoms.[13] In addition, CeO2 forms strong metal–support
Seoul 02841, Republic of Korea interactions (SMSIs) with the active metals, which allows the
E-mail: sringe@[Link] formation of a partially charged metal at the metal–CeO2 inter-
K.-S. Lee face.[14] These unique characteristics of CeO2 can also promote
Pohang Accelerator Laboratory (PAL)
Pohang University of Science and Technology (POSTECH)
the electrocatalytic CO2RR activity of metals such as Au and
Pohang 37673, Republic of Korea Cu supported on CeO2. In particular, Cu on ceria (Cu–CeO2)
shows wide tunability in CO2RR product selectivity, and this
The ORCID identification number(s) for the author(s) of this article
can be found under [Link]
is dependent on the sizes and compositions of Cu–CeO2. For
example, CeO2 can stabilize a single-atom (SA) Cu in the Cu2+
DOI: 10.1002/adma.202208996 valence state and generate CH4 with a faradaic efficiency (FE) of

Adv. Mater. 2023, 35, 2208996 2208996 (1 of 11) © 2022 Wiley-VCH GmbH
 15214095, 2023, 8, Downloaded from [Link] by INDIAN INSTITUTE OF TECHNOLOGY (NON -EAL), Wiley Online Library on [12/01/2025]. See the Terms and Conditions ([Link] on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
[Link] [Link]

up to 58%.[15] In addition, enhanced selectivity for ethylene and 2. Results

ethanol production has been reported for a catalyst comprising
interconnected CuO and CeO2 nanocrystals in a Cu/Ce ratio 2.1. Structural Characterization of As-prepared Cu–CeO2
of 0.95: the high C2+ selectivity is attributed to the Cu–CeO2 NR Catalysts
interfacial active sites.[16] However, colloidal Cu–CeO2–x nano-
particles (NPs) having a well-defined Cu–CeO2–x interface A Cu–CeO2 NR catalyst was prepared by loading Cu into
show dominant CH4 formation.[17] This discrepancy indicates CeO2 NRs prepared by hydrothermal synthesis, as reported
that the characteristics of the Cu–CeO2 interface are a key to previously.[19] After loading with Cu, the Cu–CeO2 NRs were
modulating the CO2RR activity and are heavily influenced calcined in air at 400 °C. The Cu loading was varied from 1 to
by catalyst synthesis processes. Although the structural and 5 wt.% with respect to CeO2. X-ray diffractometry (XRD) anal-
chemical properties of the Au- and Pt–CeO2 interfaces have ysis was carried out to investigate the crystal structure of the
been well-established,[18] those of Cu/CeO2 interface for the as-prepared x% Cu–CeO2 NRs (Figure 1a), where x denotes
electrochemical CO2RR require further study.[16,17] the degree of Cu doping in weight percent, as measured by
Here, we present a Cu–CeO2 nanorod (NR) electrochemical inductively coupled plasma optical emission spectroscopy
CO2RR catalyst having a well-controlled Cu/CeO2 interface. (ICP-OES), and the exact value according to the Cu contents
The abrupt transition from Cu SAs to a few atom-thick Cu is shown in Table S1 (Supporting Information). The diffrac-
nanoclusters in/on CeO2 near about 3% Cu doping was iden- tion peaks in all the catalyst corresponds to the fluorite struc-
tified using in operando analysis, as well as computational ture of CeO2, consistent with the standard diffraction pattern of
techniques, including high resolution transmission electron CeO2 (JCPDS No. 034–0394). No Cu peak (2θ = 43°) was found
microscopy (HR-TEM), atom probe tomography (APT), in oper- in the XRD patterns of any sample. A closer look at the CeO2
ando X-ray absorption near-edge structure (XANES), and den- peaks revealed a slight shift in 2θ toward higher angle as the
sity functional theory. In addition, the SMSI of CeO2 results Cu contents increased, followed by a plateau (Figure S1a and
in the formation of a mixed valence state of Cu0/Cu1+ on the Table S2, Supporting Information). Accordingly, the CeO2 lat-
Cu nanoclusters on the Cu–CeO2 surface, which catalyzes the tice constant decreased from its bulk value of 5.4292 Å and then
dimerization of CO and enhances selectivity for C2 products. plateaued at 5.4116 Å at 4% Cu content (Figure 1b). This shift in
Our Cu–CeO2 NRs with SA and NP Cu enable the tunable pro- the peaks is a result of the incorporation of Cu2+ ions into the
duction of CH4 (C1) and C2H4 (C2) products with FEs of 45.5% CeO2 lattice; because these ions have a relatively smaller ionic
and 43.2%, respectively, at an applied potential of –0.9 V versus radius (87 pm) than Ce4+ ions (101 pm), the unit cell shrinks,
the reversible hydrogen electrode (RHE) in a gas-fed flow cell. resulting in a shift to higher scattering angle.[20] In addition,

Figure 1. Structural characterization of as-prepared Cu-CeO2 NRs. a) XRD patterns of x% Cu–CeO2 NRs; CeO2 lattice planes ■ = (111), ▲ = (200),
◆ = (220), and ● = (311). b) Lattice constants of x% Cu–CeO2 NRs calculated from the XRD peaks. c) HAADF-STEM image of the 2% Cu–CeO2 NR.
d,e) HAADF-STEM image of 2% Cu–CeO2 NRs and corresponding energy-dispersive spectroscopy (EDS) maps. f) HAADF-STEM image of the 4%
Cu–CeO2 NRs. g,h) HAADF-STEM image of 4% Cu–CeO2 NRs and corresponding EDS maps.

Adv. Mater. 2023, 35, 2208996 2208996 (2 of 11) © 2022 Wiley-VCH GmbH
 15214095, 2023, 8, Downloaded from [Link] by INDIAN INSTITUTE OF TECHNOLOGY (NON -EAL), Wiley Online Library on [12/01/2025]. See the Terms and Conditions ([Link] on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
[Link] [Link]

low intensity peaks consistent with CuO (2θ = 35.5° and 38.7°) precipitated CuO in 4% Cu–CeO2 is smaller than the resolution
began to emerge in the XRD patterns of the Cu–CeO2 NRs of the HR-TEM device used in this study.
once the copper doping exceeded 4% (Figure S1b, Supporting The presence of CuO clusters was investigated further
Information). According to the CuO–CeO2 phase diagram, using atom probe tomography (APT). APT provides a quanti-
CuO and CeO2 are immiscible;[21] therefore, the precipitation of tative 3D atomic distribution of nanostructures at near-atomic
CuO in the CeO2 matrix is anticipated even at low CuO concen- resolution.[22] Thus, we analyzed the atomic distribution of as-
tration. Cu is loaded more than the solubility limit, the excess prepared 5% Cu–CeO2 NR, which is expected to contain CuO
Cu no longer incorporates in CeO2 lattice and affects the CeO2 precipitates throughout the CeO2 matrix. Because the NRs
lattice constant as it precipitates out as CuO (Figure 1a,b). In themselves are too small to be fabricated into needle-shaped
our work, the solubility limit for CuO in CeO2 is about 2%, APT specimens, the NRs were embedded in a Co matrix using
and, above this concentration, CuO precipitates out in the CeO2 electrodeposition (see Experimental Section).[23] The mass
NRs. Below the solubility limit, CuO is homogeneously mixed spectrum acquired from the APT analysis shows clear peaks
in CeO2 without segregation. corresponding to Co, Ce, and Cu ions, as expected for the
TEM analyses were performed to investigate the structure Cu–CeO2 NRs (Figure S6, Supporting Information). The
of the Cu–CeO2 NRs in detail. The as-prepared Cu–CeO2 NRs measured atomic ratio of Ce and Cu in the NR is 6.9:1, in agree-
had an average width of 9.8 nm and an aspect ratio ranging ment with the target composition of 5 wt.% Cu–CeO2 NR in
from about 1:1.5 to 1:15 (Figure S2, Supporting Information). atomic percent (Ce:Cu = 7:1).
The high-angle annular dark-field scanning transmission elec- Figure 2a shows a 3D atom map of the as-prepared 5%
tron microscopy (HAADF-STEM) images show that the CeO2 Cu–CeO2 NR embedded in the Co matrix. In the atom map,
support is single crystalline with a (111) crystal plane parallel a single CeO2 NR is included (visualized with isolated Cu
to the long axis of the rod (Figure 1c,f). However, CuO is not atoms and Cu clusters). The isolated Cu atoms, i.e., Cu SAs,
visible in the HAADF-STEM images, even though micro- are assumed to be the dissolved solutes in the CeO2 lattice and
scopic observation was carried out at multiple locations with the Cu clusters appear to originate from the precipitation of
many independent samples. This could be attributed to the CuO. The coexistence of the surface CuO NPs and the Cu SAs
poor crystallinity and the low contrast of CuO compared to in the NR is more clearly shown in the enlarged atom maps
that of CeO2.[16,17] Elemental analysis indicates that Cu is uni- (Figure 2b,c). Such clustering behavior of Cu is also reflected
formly distributed both in the 2% and 4% Cu–CeO2 samples in a Cu nearest neighbor distribution (NND) (Figure S7, Sup-
(Figure 1c–h). However, the segregation of Cu was not observed porting Information). The Cu NND from the atom map shows
in the images of the 4% Cu–CeO2 sample, despite the fact that a clear deviation from a random Cu distribution, indicating the
the precipitation of CuO is expected based on the XRD results. presence of Cu clusters in the NRs.[24] The clustering behavior
The structure and elemental distribution of the Cu–CeO2 NRs of Cu in 5% Cu–CeO2 NR is more clearly confirmed by the χ2-
was maintained even after the CO2RR (Figures S3 and S4, Sup- statistical test and also from the additional APT dataset of 5%
porting Information). In the electron energy loss spectroscopy Cu–CeO2 NRs (Figure S8, Supporting Information). In addi-
(EELS) spectra, small peaks at the Cu L2,3 position were found tion, to confirm if these Cu clusters exist as CuO, we obtained
at/near the CeO2 surface, probably arising from CuO clusters radial distribution functions (RDFs) for Cu and O (Figure S9,
(Figure S5, Supporting Information). However, the presence Supporting Information). The Cu RDF shows that Cu–O, as
of these peaks is not sufficient to prove the presence of CuO well as Cu–Cu, are positioned close together, likely as a result
clusters on the CeO2 surface. Therefore, we conclude that the of the aggregation of Cu and O atoms in the NRs .[24] Likewise,

Figure 2. APT reconstruction of as-prepared 5% Cu–CeO2 NRs. a) 3D atom map of CeO2 NR in a Co matrix. b,c) Enlarged atomic view of Cu clusters
on CeO2 surface. d) Proximity histogram consistent with 25 at.% Ce iso-concentration surface. Error bars represent the standard deviation (σ).

Adv. Mater. 2023, 35, 2208996 2208996 (3 of 11) © 2022 Wiley-VCH GmbH
 15214095, 2023, 8, Downloaded from [Link] by INDIAN INSTITUTE OF TECHNOLOGY (NON -EAL), Wiley Online Library on [12/01/2025]. See the Terms and Conditions ([Link] on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
[Link] [Link]

the O RDF shows the strong correlation between O–O and X-ray photoelectron spectroscopy (XPS) was also car-
O–Cu, suggesting that Cu is present as CuO clusters within the ried out to compare the valence states change of the Cu and
CeO2 bulk, as well as on the surface; this is consistent with the Ce atoms on the surface of the Cu–CeO2 NRs before and
XRD pattern of the 5% Cu–CeO2 NR (Figure S1b, Supporting after the CO2RR (Figures S12 and S13, Supporting Informa-
Information). To confirm the absence of Cu clusters in lower- tion). The XPS spectra of the as-prepared 4% Cu–CeO2 NR
doped CeO2 NRs, we also conducted the APT analyses on the show the mixed oxidation states of Ce and Cu (Figure 3b,c).
as-prepared 1% Cu–CeO2 NR (Figure S10, Supporting Informa- Although the as-prepared 4% Cu–CeO2 NR was calcined in air,
tion). The resulting data shows well-matched composition with the partial reduction of CuO and CeO2 was observed before
the 1 wt.% Cu–CeO2 NR (i.e., Ce:Cu = 37:1 in atomic ratio), and the CO2RR. Figure 3d,e shows the change of surface atomic
no remarkable Cu clusters are detected from the 3D atom map ratio of Ce3+ and (Cu0, Cu+), respectively, by calculating the
(Figure S10a–c, Supporting Information). In addition, the rea- area ratio shown in the XPS spectrum of elements having dif-
sonable fit between random and experimental Cu distribution ferent valences before and after CO2RR. Relative amount of
from the Cu NND, (Figure S10d, Supporting Information) and surface atom of each catalysts in Figure 3d,e is determined by
the χ2-statistical test further supports that the Cu clusters are dividing the amount of surface atom of the reference samples
unlikely to exist in the lower doped CeO2 nanorods as expected that are undoped CeO2 and 2% Cu–CeO2 NR, respectively
from the XRD pattern and solubility limit of CuO in CeO2 (Tables S3 and S4, Supporting Information). The amount of
(≈2 wt.%). partially reduced CuO and CeO2 increased with increase in Cu
The presence of surface CuO clusters was also identi- concentration, although it decreased at higher Cu doping con-
fied from a proximity histogram, which was obtained for an centration (Figure 3d,e). Therefore, the Cu and Ce atoms on the
iso-concentration surface of 25 at.% Ce, which is half the Cu–CeO2 surface interact closely, thus affecting their valence
maximum concentration of Ce in the NRs (Figure 2d).[25] states.[15,19b,28]
The proximity histogram shows the interface between CeO2 The changes in the valences of Cu and Ce were strongly
and the Co matrix, and the Cu concentration increased by dependent on the Cu doping concentration and showed dra-
≈2 at.% ≈2 nm from the interface before it dropped. Because matic differences after the CO2RR. When the Cu doping con-
the proximity histogram provides the average atomic con- centration was 3% or more, the ratio of Ce3+ and (Cu0,Cu1+)
centration with respect to distance from the interface,[26] this increased after the CO2RR (Figure 3d,e). This can be attrib-
result indicates the presence of CuO clusters at the CeO2 uted to the fact that the CuO clusters segregated on the sur-
interface. face of the catalyst and were reduced to metallic Cu clusters
as the CO2 reduction reaction proceeded. At the same time,
electrons from the reduced Cu clusters are transferred to form
2.2. Valence State Characterization of the Cu–CeO2 Interface Ce3+ and Cu1+ at the Cu/CeO2 interface, allowing them to exist
in a stable state on the CeO2 surface. The adsorbed Cu atoms
The abrupt transition in the structure of Cu from SAs to NPs on the CeO2 surface can be stabilized via bonding with the
on the CeO2 surface, which is a result of the low solubility of oxygen vacancies (VO) of CeO2 in a form Cu+–VO–Ce3+ at the
CuO in CeO2, can produce dramatic differences in the valence Cu/CeO2 interface.[14,28a] These results are consistent with
state of Cu, and this affects the CO2RR activity. Accordingly, we the in operando XANES results. Notably, the valence state of Cu
attempted to determine the valence state of the Cu atoms con- varies more significantly than that of Ce: the relative amount
stituting the x% Cu–CeO2 catalyst using in operando XANES of surface (Cu0, Cu+) increases ≈2.8-times, whereas that of sur-
analysis. The analysis was performed with a constant current in face Ce3+ increases only ≈25% in 4% Cu–CeO2 (Table S4, Sup-
a gas-diffusion electrode (GDE) flow cell (Figure S11, Supporting porting Information). Because the formation of Cu+ and Ce3+ is
Information). The normalized Cu K-edge XANES spectra of coupled at the Cu/CeO2 interface and CeO2 itself is inert during
the 2% and 4% Cu–CeO2 NR samples were measured under CO2RR, this indicates that the Cu clusters consist of more than
open-circuit conditions (dotted lines) and during the CO2RR at a few atomic layers, as revealed in the APT analysis (Figure 2c).
−100 mA cm−2 (solid lines) to allow the valence state changes to This also means that a layer of stacked Cu+–Cu0 species would
be compared (Figure 3a). Under open-circuit conditions, both form along the perimeter of the interface, and, thus, the struc-
the 2% and 4% Cu–CeO2 catalysts have characteristic peaks ture of the CuO–CeO2 interface can lower the energy barrier for
(8986.0 and 8997.7 eV) of Cu2+,[27] consistent with the reference C–C coupling.
CuO spectrum shown in lower panel of Figure 3a. When a con-
stant current was applied to the Cu–CeO2 NR catalysts, these
catalysts showed different changes in their spectra. In the case 2.3. Insights from Density Functional Theory Calculations
of 2% Cu–CeO2 NR, there was no significant change in the
spectra during the reduction process, whereas, in the case of To gain further insights into the formation of different oxidation
4% Cu–CeO2 NR, the peak at 8997.7 eV slightly decreased, and states and clusters under operating conditions, we performed
the edge also moved to lower energy during reduction. This extensive spin-polarized density functional theory (DFT) cal-
indicates that the initial Cu2+ state changed to Cu0, Cu1+ as the culations using the RPBE+U method.[29] All calculated forma-
reaction proceeded. This is because CuO clusters on/near the tion energies are summarized in the Tables S5–S8 (Supporting
CeO2 surface in 4% Cu–CeO2 NR were reduced to Cu1+and Cu0 Information). We first considered the pristine CeO2 catalyst and
during the CO2RR. As a result, both Cu0 and Cu+ co-exist in the investigated the probability of oxygen vacancy (VO) formation
4% Cu–CeO2 NRs. by searching for the most stable configuration. For this, we

Adv. Mater. 2023, 35, 2208996 2208996 (4 of 11) © 2022 Wiley-VCH GmbH
 15214095, 2023, 8, Downloaded from [Link] by INDIAN INSTITUTE OF TECHNOLOGY (NON -EAL), Wiley Online Library on [12/01/2025]. See the Terms and Conditions ([Link] on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
[Link] [Link]

Figure 3. Analysis of valence state changes of atoms at the surface of Cu–CeO2 NRs. a) Normalized Cu K-edge XANES spectra of 2% and 4% Cu–CeO2
NR catalysts recorded under open-circuit conditions and at -100 mA cm−2 for 10 min. Spectra of a Cu foil and Cu2O and CuO powder samples are
shown at the bottom as references. b,c) XPS spectra of the Ce 3d5/2 and Cu 2p3/2 in as-prepared 4% Cu–CeO2. Note that v0,v’ indicate the peaks cor-
responding to Ce3+, and v,v″″ indicate peaks corresponding to Ce4+. d) Relative amount of Ce3+ in x% Cu–CeO2 NR catalysts determined from Ce 3d5/2
XPS compared to pure CeO2 NR. e) Relative amount of (Cu0,Cu1+) in x% Cu–CeO2 NR catalysts from Cu 2p3/2 XPS compared to 2% Cu-NR. The relative
amount of surface atoms is the ratio at which the amount of surface atoms increases based on the reference sample having the lowest Cu content. In
the Ce 3d5/2 and Cu 2p3/2 XPS data, pure CeO2 NR (0% NR) and 2% NR were used as references, respectively.

considered both bulk CeO2 and the (111) facet, as well as two Starting from the pristine CeO2 crystal, we prepared a
different environmental conditions: ambient gas phase condi- 1.6 mass% Cu-doped CeO2 crystal containing both substitu-
tions with atmospheric O2 playing the role of the oxidant and tional (CuS) and interstitial (CuI) defects, as reported in ear-
reducing electrochemical conditions, for which H2O facilitated lier works.[31] We first evaluated the formation energies of the
by an applied bias potential results in oxidation (the computa- Cu defects (Figure S17a,b, Supporting Information). Under
tional hydrogen electrode, CHE approach was used for this).[30] ambient conditions, we found CuI to be more favorable than
Under ambient air conditions, we found that vacancy formation CuS and preferably located on the surface rather than the bulk.
was not thermodynamically feasible at the considered vacancy We further found vacancy formation in the case of the CuI
concentration, neither on the CeO2 crystal nor on the surface defects to occur randomly throughout CeO2 without prefer-
(positive formation energy relative to the pristine materials). ence for the CuI site (Figures S15 and S19, Supporting Infor-
However, at smaller vacancy concentrations, the configurational mation) indicating decoupled vacancy and defect formation.
entropy reduces the formation energy so that we can expect the Therefore, the CuI site maintains its oxidation state, which we
formation of vacancies at low concentrations. Under reducing estimated from Bader charge analysis,[32] of between +1 and +2
conditions, we saw the expected strongly favourable formation on the surface and +1 in the bulk, and the oxidation state of
of oxygen vacancies.[31] The vacancies were most stable in the the neighbouring Ce is reduced to stabilize the lower oxidation
bulk of CeO2 rather than on the (111) surface, Figure S16a,b state of the cationic defects (Figure S19, Supporting Informa-
(Supporting Information) show all plotted vacancy forma- tion). In contrast to the CuI defect, we found that CuS had a
tion energies and the first row of Figure 4a summarizes these sufficiently high energetic driving force to induce the formation
findings. of a nearby oxygen vacancy (Figures S14 and S18, Supporting

Adv. Mater. 2023, 35, 2208996 2208996 (5 of 11) © 2022 Wiley-VCH GmbH
 15214095, 2023, 8, Downloaded from [Link] by INDIAN INSTITUTE OF TECHNOLOGY (NON -EAL), Wiley Online Library on [12/01/2025]. See the Terms and Conditions ([Link] on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
[Link] [Link]

Figure 4. Computational insights into the thermodynamic stability of Cu/CeO2 systems. a) Schematic of the combined theoretical and experimental
findings concerning the thermodynamic stability of Cu/CeO2 systems under different environmental and doping conditions. b) Energetic cost of
forming substitutional and interstitial Cu single atom defects from metallic Cu, as well as oxygen vacancies, in a pristine CeO2(111) surface. Formation
energies under ambient conditions/air are shown in black and those under electrochemical conditions are shown in blue. c) CeO2(111)/Cu(111) inter-
face: xy-plane averaged electronic charge density difference Δρ(Z) relative to the isolated components. Inset: Isosurface of Δρ at 0.003 e a0−3. Green,
blue, and red balls represent the Ce, Cu, and O atoms.

Information), possibly because of the smaller possible oxida- stability here. Consequently, interstitial Cu defects are mostly
tion state (+2) of Cu compared to that of Ce (+4), which can be expected close to the surface, whereas oxygen vacancies are
stabilized by introducing an oxygen vacancy. Indeed, as shown expected randomly in the bulk of CeO2, as summarized in the
in Figure 4b and Figure S17c (Supporting Information), the for- second row of Figure 4a.
mation of a CuS–VO pair in the bulk of CeO2 is energetically At higher Cu doping concentrations, Cu cluster formation
comparable to the formation of CuI in the bulk, and both could, will unavoidably occur because of the higher stability of the
thus, be present. However, we expect surface CuI formation to phase-separated system compared to the SAs.[21] As shown in
be dominant; therefore, Cu defects should accumulate in the Figure 4b, all Cu formation energies are positive relative to the
surface region. Under reducing conditions, the energetic cost reference system, i.e., bulk metallic Cu. Note that copper oxide
of forming a CuS defect increases. For demonstration, if a Cu phases are more stable compared to Cu under ambient condi-
defect originates from metallic Cu and is inserted into the pris- tions (Figure S17d, Supporting Information). We also evaluated
tine CeO2 structure, four electrons must be released during the the energetic costs of forming solid–solid copper oxide–CeO2
oxidation reaction: interfaces. Tables S7 and S8 (Supporting Information) show
that, although energy is required to form a free-standing Cu,
Cu(s) + ( CeO2 )x(s) + 2H2O( l) → copper oxide, or CeO2 surface, energy is released when forming
(1)
(CuO2 )(CeO2 )x −1(s) + CeO2(s) + 4e − + 4H(aq)
+
a CeO2/Cu or CeO2/Cu oxide interface as a result of attractive
interactions. Following standard nucleation theory, this result
Here, x depicts the formula units of CeO2 per supercell and, indicates that there is practically no nucleation barrier, and NPs
thus, represents the doping concentration. Reducing conditions can grow freely in the bulk. Because of the large size of the NPs,
will shift this equilibrium to the left so that the formation of we also expect the formation of metastable NPs on the surface.
CuS becomes increasingly energetically unfavorable. This ener- In general, a reduction in the radii of NPs exposes more surface
getic penalty can be reduced by increasing the vacancy con- facets to the environment (CeO2 or gas/water), which stabilizes
centration because the reference system converges to a less Cu NP formation.[33] For all considered radii, the formation
oxidized state and the number of electrons released during the energy of Cu in NPs is much lower than the formation energy
oxidation reaction decreases. However, as shown in Figure 4b of a CuI SA (Figure S17a,b,d, Supporting Information). Under
and Figure S17c (Supporting Information), CuI is expected to be ambient conditions, CuO NPs with Cu in the oxidation state +2
energetically preferable in all cases based on our calculations, will form as the most stable phase (Figures S17d and S21, Sup-
although the formation energies on the surface suggest greater porting Information). Under reducing conditions, these NPs are

Adv. Mater. 2023, 35, 2208996 2208996 (6 of 11) © 2022 Wiley-VCH GmbH
 15214095, 2023, 8, Downloaded from [Link] by INDIAN INSTITUTE OF TECHNOLOGY (NON -EAL), Wiley Online Library on [12/01/2025]. See the Terms and Conditions ([Link] on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
[Link] [Link]

readily reduced to Cu(0) (Figure S17d, Supporting Information) CuI is most stable and its stability is independent of vacancy
and vacancies form in the bulk of CeO2, as discussed above. We formation. At higher doping concentrations, Cu is thermody-
also considered the possible formation of oxygen vacancies at namically prone to form NPs via phase separation. In contrast,
the Cu (oxide)/CeO2 interface. Although the surface tension of under ambient conditions, Cu will form Cu2+O clusters, which
Cu(111)/CeO2(111) in the presence of oxygen vacancies is less can be reduced to Cu0/+ under electrochemical conditions; this
negative, indicating still formation of vacancies in the bulk, the is an oxidation state change that was confirmed experimentally
copper oxide interface could be subject to some vacancy forma- for both the bulk (XANES) and surface (XPS), showing that the
tion (Table S8, Supporting Information). Next, we analyzed the clusters are present both at the surface and the bulk.
oxidation state of metallic Cu NPs adsorbed on the CeO2(111)
surface by considering an overlayer structure (Figure 4c). We
found a slightly increased oxidation state of Cu close to the 2.4. Performance of Cu–CeO2 NRs for CO2RR
interface, in between Cu(0) and Cu+, as also evidenced by
the charge distribution. The bottom layer of the Cu cluster con- The electrocatalytic performance of all Cu–CeO2 NR cata-
tains slightly oxidized Cu atoms that are bonded to O sites on lysts was evaluated in a GDE based flow electrolyzer.[35] As
reduced CeO2(111); this is consistent with literature.[28a,34] reference, the performance of the pure CeO2 NRs was evalu-
Overall, our computational findings are consistent with the ated, and they were found to be inactive for the CO2RR: that
XPS and XANES data. At low doping concentrations, there is no is, hydrogen production was observed at all current densities
change in the Cu oxidation state, neither in the bulk (XANES) (Figure S23, Supporting Information). Figure 5a shows the
nor the surface (XPS), even under reducing conditions because CO2RR partial current densities of Cu–CeO2 NRs at various

Figure 5. Electrochemical CO2 reduction performances of x% Cu–CeO2 NR. a) Partial current density for CO2 reduction of the x% Cu–CeO2 NR cata-
lysts. b) Faradaic efficiency of x% Cu–CeO2 NR catalysts at –0.9 V versus RHE. c,d) Partial current density for CH4 and C2H4 of the x% Cu–CeO2 NR
catalysts. e) Stability test results of 2% and 4% Cu–CeO2 NR at −0.9 V versus RHE.

Adv. Mater. 2023, 35, 2208996 2208996 (7 of 11) © 2022 Wiley-VCH GmbH
 15214095, 2023, 8, Downloaded from [Link] by INDIAN INSTITUTE OF TECHNOLOGY (NON -EAL), Wiley Online Library on [12/01/2025]. See the Terms and Conditions ([Link] on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
[Link] [Link]

Cu doping concentrations (Table S9, Supporting Information). Cu-based catalysts reported so far in terms of properties and
The CO2RR activity at a given potential increased with increase performance (Table S12, Supporting Information).
in the Cu doping concentration, and the highest current den- Stability is also a very important factor in the evaluation
sity of −206.8 mA cm−2 was achieved at −0.81 V versus the of catalyst performance. Therefore, stability tests were also
RHE for 5% Cu–CeO2 NR. This is because of the increasing performed at −0.9 V versus RHE for 2% and 4% Cu–CeO2,
number of active sites for the CO2RR with increase in the Cu which produce CH4 and C2H4 as main products, respectively
concentration. (Figure 5e). Both catalysts showed stable catalytic activity, and
Figure 5b shows the product distribution of all Cu-CeO2 NRs the FEs of CH4 and C2H4 were maintained above 40% even
at a constant potential of −0.9 V versus RHE (Table S10, Sup- after 250 min of CO2RR.
porting Information). Interestingly, in the case of 2% Cu–CeO2,
CH4 was the main product, having an FE of ≈45.5%. However,
as the doping concentration of Cu gradually increased, the 3. Conclusions
amount of CH4 diminished, and the FE decreased below 5%.
Further, C2H4 became the main product, having an FE of 43.2%. In summary, we have reported the tunable CH4/C2H4 selec-
Similar transition from CH4 to C2H4 with Cu doping concen- tivity of Cu–CeO2 NRs during the electrochemical CO2RR
tration is apparent in other applied potentials (Figure S24 and by precisely controlling the structure and valence state of
Table S11, Supporting Information). Figure 5c,d show the par- the Cu active sites on CeO2 NRs; the unique tunability is a
tial current densities for CH4 and C2H4 in x% Cu–CeO2 as a result of the immiscible nature of the CuO and CeO2 binary
function of applied potential. The 2% Cu–CeO2 NRs yielded oxide system. Briefly, when the CuO content is less than the
CH4 with current density for CH4 (JCH4) of −123.4 mA cm−2 solubility limit in CeO2, CuO is present as a single atomic,
at −0.99 V versus RHE, whereas 5% Cu–CeO2 NR yielded a random solid solution in CeO2 and is maintained as a stable
JC2H4 of −117.1 mA cm−2 at −0.81 V versus RHE. Thus, both the state even at the large negative potentials used for the CO2RR.
abrupt suppression of CH4 production and increase in C2H4 In contrast, when the concentration of CuO exceeds the solu-
production at Cu concentrations larger than 4% are evident. bility limit of CeO2, excess CuO precipitates out as CuO clus-
For example, the ratio of the partial current density for C2+ pro- ters throughout the CeO2 NRs, and these can be observed by
duction to that of C1 is ≈0.09 for 2% Cu–CeO2 NR, but it rap- APT measurements. These results were obtained from both
idly increased to 2.85 for 4% Cu–CeO2 NR at a current density experimental analysis and extensive quantum chemical cal-
of −200 mA cm−2: that is, the selectivity toward C2+ products culations. As the CO2RR proceeds, the CuO clusters undergo
increased by 34-times when the Cu doping concentration was phase transformation to metallic Cu clusters with a partially
doubled (Figure S25, Supporting Information). This sudden oxidized Cu+ layer at the Cu/CeO2 interface, leading to the
change in product selectivity is attributed to the different struc- formation of mixed Cu+/Cu0 active sites along the perim-
ture and valence state of Cu in CeO2 associated with the phase eter of the Cu/CeO2 interface. The oxidized form of Cu at
separation of CuO or Cu in CeO2. Specifically, when Cu SAs the interface was also confirmed using quantum chemical
exist as a random solid solution in the CeO2 NRs, CO–CO calculations. Further, electrochemical CO2RR evaluation in a
dimerization is suppressed and CH4 is produced dominantly. gas-fed flow cell revealed that single atomic CuO in CeO2 is
In contrast, when the amount of Cu is above the solubility limit unfavorable for C–C coupling and produces CH4 as the main
in CeO2, the CeO2 support stabilizes the (Cu0,Cu1+) cluster product. However, the coexistence of (Cu0,Cu1+) in the clus-
structure, thus resulting in a favored pathway to C2 products, ters lowers the energy barrier for C–C coupling and enables
such as C2H4. Notably, the electrochemically active surface the selective production of C2+ hydrocarbons. This work pro-
area (ECSA)-normalized CO2RR activity shows a similar trend vides a rational and fundamental strategy for the fabrication
(Figures S26 and S27, Supporting Information). of high-performance, highly efficient electrocatalysts through
To confirm the stabilizing effect of the ceria NR support on the precise control of the active sites and their structure on
the mixed-valent Cu clusters at the Cu–ceria interface, compar- metal oxide supports.
ison with a 4% Cu–carbon catalyst was carried out (Figure S28,
Supporting Information). The CuO existing in the Cu-carbon
structure is reduced to metallic Cu0 during the electrochemical 4. Experimental Section
reduction reaction, and these act as active sites. Therefore, by
comparing the Cu-carbon catalyst containing the same amount Catalyst Preparation: CeO2 nanorods (NRs) was prepared by a
of Cu as a reference, the effect of mixed-valent Cu clusters at hydrothermal synthesis method. After dissolving 16.8 g of NaOH
(JUNSEI) in 70 mL of deionized (DI) H2O for 10 m, 10 mL of DI H2O
the Cu–ceria interface could be identified. These experiments in which 1.736 g of cerium(III) nitrate hexahydrate (99.99%, Sigma
were conducted at constant potential (−0.9 V versus RHE), Aldrich) was dissolved with the dropwise addition and stirred for 30 m
and, on comparing the FE of the C2+ product, the performance with 500 RPM. A total of 80 mL of the mixed solution was sealed in
was found to be ≈5-times better when the ceria support was a hydrothermal reactor with a capacity of 100 mL and maintained at
used. Further, comparing the partial current density of the C2+ 100 °C for 24 h. Then, Cu was deposited on CeO2: a desired amount
product per unit mass of Cu, the Cu–ceria NR showed 27-times of Cu(II) nitrate trihydrate (Sigma Aldrich) was added in the mixed
solution, followed by stirring at 500 RPM at room temperature for 12 h.
higher performance than the Cu-carbon catalyst. This result Finally, the catalysts were dried at 60 °C overnight and then calcined
underlines the notable (and positive) effect of mixed valent in a tube furnace in an air atmosphere at 400 °C for 2 h to obtain x%
Cu at the Cu–ceria interface. These results indicate superior Cu–CeO2 NR catalyst powder. For the Cu-CB sample, the same wt.% of
catalytic activity of the Cu–ceria NRs compared to Cu–ceria or Cu and carbon black (Sigma Aldrich) as that of the Cu–CeO2 sample

Adv. Mater. 2023, 35, 2208996 2208996 (8 of 11) © 2022 Wiley-VCH GmbH
 15214095, 2023, 8, Downloaded from [Link] by INDIAN INSTITUTE OF TECHNOLOGY (NON -EAL), Wiley Online Library on [12/01/2025]. See the Terms and Conditions ([Link] on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
[Link] [Link]

was used, and the rest of the process was prepared in the same way where V°Ag/AgCl is 0.197 V in saturated KCl solution at room temperature.
except for hydrothermal synthesis. To make a catalyst ink, 50 mg of The calculation method of Faradaic efficiency (FE) used in this study is
catalyst powder was mixed with 20 mL of isopropanol and 20 µL of as follows.
nafion (Sigma Aldrich). This catalyst ink was evenly sprayed on a carbon
z ⋅n ⋅F
paper (Sigracet, 39BC) using a spray gun and the loading amount is FE ( % ) = (3)
≈1.5 mg cm−2 for all samples. Q
Materials Characterizations: The powder X-ray diffraction (XRD) where z is the number of electrons transferred, n is the number of moles
patterns were acquired with X’Pert-PRO MRD diffractometer of products, F is Faraday constant, and Q is the amount of charge.
(PANalytical) equipped with Cu K-alpha light source. X-ray photoelectron ECSA values were analyzed by the Pb underpotential deposition
spectra (XPS) were recorded with a K-alpha system provided by (UPD) method in 0.1 m HClO4 and 1 mm PbCl2 at a scan rate of
VG scientific and the peak deconvolution was done in an Avantage 10 mV s−1. During the measurement, the solution was continuously
software. The HAADF-STEM and EELS analysis were performed at an purged with N2 gas and charge was calculated by integrating the
acceleration voltage of 300 kV with Titan G2 60–300 which was equipped oxidation peak area of the Pb monolayer. ECSA values for all samples
with a monochromator, aberration corrector and Gatan Quantum were calculated by dividing the stripping charge by the reported
965 spectrometer. ICP-OES 720 (Agilent) was used to measure the conversion factor (Figure S26, Supporting Information).[3d]
Cu concentration in Cu–CeO2 catalysts. The operando XANES was Computational Framework: Structure relaxations were performed using
conducted using 10C-Wide XAFS line using at the Pohang light source the Vienna ab-initio simulation package (VASP).[29a,b] The electron-ion
(PLS-II) in Republic of Korea. All experimental data were measured interactions were described using the projector-augmented wave (PAW)
in the fluorescence mode and the references (CuO powder, Cu2O method, primarily a frozen-core all-electron calculation.[29c] Attractive
powder, and Cu metal) were measured in transmission mode at Cu van der Waals interactions included using Grimme’s correction for the
K-edges. Commercial products (CuO, Cu2O : Sigma Aldrich) were used revised Perdew-Burke-Ernzerhof (RPBE) with the Pade method.[29d,37]
as references. During the operando XANES measurements, 1 m KOH For structure optimization, the atoms were relaxed in the direction of
solution (Liquid pump : Masterflex L/S) and 20 sccm of CO2 gas (Mass the Hellmann-Feynman force using the conjugate gradient method with
flow controller : MFC KOREA) were continuously fed into the cathode an energy cut-off of 520 eV until a stringent convergence criterion (of
part of the GDE cell. The operando test was performed under a constant 0.001 eV Å−1) was satisfied. The electrons in Ce and Cu metals were
current of −100 mA cm−2. Data fitting was done using ATHENA and divided into two groups: strongly correlated ones usually localized on
ARTEMIS software. atomic d or f orbitals and subject to the Hubbard treatment.[38] The
Atom Probe Tomography: As-prepared 5% Cu–CeO2 NRs were spray remaining electrons were treated using standard density functional
coated on porous carbon substrate and electrodeposited by Co using theory. Lattice constants and accurate electronic structure calculations
the process reported by Kim et al. for enabling APT analysis.[23] Co were performed using the RPBE exchange-correlation functional by
electrodeposition bath was prepared by dissolving 15 g of cobalt sulfate employing Hubbard U corrections for Ce f and Cu d orbitals with
heptahydrate (CoSO4·7H2O, Sigma Aldrich), and 1.75 g of boric acid UCe = 5 eV and UCu = 4 eV, respectively.[39] The U parameter was chosen
(H3BO3, Sigma Aldrich) in 50 mL of DI water. Electrodeposition was for Ce f orbitals because spin saturates in the Ce+3 oxidation state and
performed at a constant current of 2 mA for 1600 s. APT measurements accurately models chemical expansion in ceria and adsorption properties
were conducted using a CAMECA LEAP 4000X HR instrument with a of reduced ceria.[39a,40] However, the U parameter correctly predicted
pulsed-laser mode at evaporation rate of 0.003 atoms per pulse, base bandgap and formation enthalpy of CuO and Cu2O.[39b] Additional
temperature of 57 K, pulse frequency of 125 kHz, and laser pulse energy details are reported in Supporting Information.
of 100 pJ. Needle-shaped APT specimens were prepared by ultra-high
resolution focused ion beam (Helios G4, FEI) according to the standard
lift-out method.[36] APT data were analyzed using IVAS 3.8.4 software Supporting Information
developed by CAMECA instruments.
Electrochemical CO2RR in a Flow Cell: All the electrochemical Supporting Information is available from the Wiley Online Library or
measurement were taken on an (SP-150, Biologic) in a three- from the author.
electrode system. The three electrode system consists of a counter
electrode (Ni–Fe–Mo alloy supplied from Sigma Aldrich), a references
electrode (Ag/AgCl saturated with KCl), and a working electrode Acknowledgements
(catalyst sprayed on a GDE). The anodic and cathodic compartment
of the flow cell were separated by a sustainion membrane (supplied S.H., H.G.A., and K.J. contributed equally to this work. The authors
by Dioxide material). The cathodic compartment filled with 7 mL gratefully acknowledge the financial support provided by the National
of 1 m KOH was connected to an electrolyte reservoir of 93 mL. The Research Foundation of Korea (NRF) grant funded by the Korean
whole electrolyte was circulated via a pump (Masterflex L/S) between government (MSIT) (Nos. 2021R1A2C3007280, 2021R1C1C1008776,
the cathodic compartment and the reservoir at a rate of 10 mL min−1. and 2021R1A2C1011415) and the Carbon to X Project (No. NRF-
The cathodic gas compartment was continuously fed with 20 sccm of 2020M3H7A1096388). This paper is based on a research conducted as
CO2 flow controlled via a mass flow controller (MFC KOREA). The flow part of the KAIST-funded Global Singularity Research Program for 2021
rate at the outlet of the GDE was monitor during each experiment and (award number: 1711100689) and supported by Samsung Electronics Co.,
used to estimate the CO2RR performances. The gas products such Ltd. (IO201211-08058-01). The authors also acknowledge the support of
as hydrogen, CO, methane, and ethylene were quantified by using a the National Supercomputing Center for supercomputing resources,
gas chromatography (YL instrument, 6500 GC system) equipped with including technical support (No. KSC-2021-CRE-0410).
thermal conductivity (TCD) and flame ionization (FID) detectors.
Liquid products collected from the cathode part were quantified by
using a high performance liquid chromatography (HPLC, YL9100, Conflict of Interest
Youngin Chroma). All measurement results were recorded after
The authors declare no conflict of interest.
20 m of reaction under each condition and reproducibility test was
performed on three independent samples under the same conditions.
All potentials used in this experiment were based on the RHE electrode
after 90% iR compensation as shown below. Data Availability Statement
The data that support the findings of this study are available in the
VRHE = VAg/AgCl + 0.059 pH + VAg/AgCl
o (2) supplementary material of this article.

Adv. Mater. 2023, 35, 2208996 2208996 (9 of 11) © 2022 Wiley-VCH GmbH
 15214095, 2023, 8, Downloaded from [Link] by INDIAN INSTITUTE OF TECHNOLOGY (NON -EAL), Wiley Online Library on [12/01/2025]. See the Terms and Conditions ([Link] on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
[Link] [Link]

Keywords [12] a) S. Bai, Q. Shao, P. Wang, Q. Dai, X. Wang, X. Huang, J. Am.

Chem. Soc. 2017, 139, 6827; b) S. Kattel, P. Liu, J. G. Chen, J. Am.
atom probe tomography, copper-ceria, electrochemical CO2 reduction, Chem. Soc. 2017, 139, 9739; c) J. Graciani, K. Mudiyanselage, F. Xu,
gas diffusion electrodes, interface A. E. Baber, J. Evans, S. D. Senanayake, D. J. Stacchiola, P. Liu,
J. Hrbek, J. F. Sanz, Science 2014, 345, 546; d) J. A. Rodriguez, P. Liu,
Received: September 29, 2022 D. J. Stacchiola, S. D. Senanayake, M. G. White, J. G. Chen, ACS
Revised: November 19, 2022 Catal. 2015, 5, 6696.
Published online: December 20, 2022 [13] a) J. A. Farmer, C. T. Campbell, Science 2010, 329, 933; b) H. Xiong,
S. Lin, J. Goetze, P. Pletcher, H. Guo, L. Kovarik, K. Artyushkova,
B. M. Weckhuysen, A. K. Datye, Angew. Chem. 2017, 129, 9114.
[14] M. Nolan, J. Chem. Phys. 2012, 136, 134703.
[1] C. Mora, D. Spirandelli, E. C. Franklin, J. Lynham, M. B. Kantar, [15] Y. Wang, Z. Chen, P. Han, Y. Du, Z. Gu, X. Xu, G. Zheng, ACS Catal.
W. Miles, C. Z. Smith, K. Freel, J. Moy, L. V. Louis, Nat. Clim. Change 2018, 8, 7113.
2018, 8, 1062. [16] C. W. Lee, S.-J. Shin, H. Jung, D. L. T. Nguyen, S. Y. Lee, W. H. Lee,
[2] a) S. Nitopi, E. Bertheussen, S. B. Scott, X. Liu, A. K. Engstfeld, D. H. Won, M. G. Kim, H.-S. Oh, T. Jang, ACS Energy Lett. 2019, 4,
S. Horch, B. Seger, I. E. Stephens, K. Chan, C. Hahn, Chem. Rev. 2241.
2019, 119, 7610; b) Y. Y. Birdja, E. Pérez-Gallent, M. C. Figueiredo, [17] S. B. Varandili, J. Huang, E. Oveisi, G. L. De Gregorio, M. Mensi,
A. J. Göttle, F. Calle-Vallejo, M. T. Koper, Nat. Energy 2019, 4, 732; M. Strach, J. Vavra, C. Gadiyar, A. Bhowmik, R. Buonsanti, ACS
c) P. De Luna, C. Hahn, D. Higgins, S. A. Jaffer, T. F. Jaramillo, Catal. 2019, 9, 5035.
E. H. Sargent, Science 2019, 364, eaav3506; d) M. Jouny, W. Luc, [18] a) A. M. Gänzler, M. Casapu, F. Maurer, H. Störmer, D. Gerthsen,
F. Jiao, Ind. Eng. Chem. Res. 2018, 57, 2165. G. Ferre, P. Vernoux, B. Bornmann, R. Frahm, V. Murzin, ACS
[3] a) M. B. Ross, P. De Luna, Y. Li, C.-T. Dinh, D. Kim, P. Yang, Catal. 2018, 8, 4800; b) Y. Lin, Z. Wu, J. Wen, K. Ding, X. Yang,
E. H. Sargent, Nat. Catal. 2019, 2, 648; b) D. Gao, R. M. Arán-Ais, K. R. Poeppelmeier, L. D. Marks, Nano Lett. 2015, 15, 5375;
H. S. Jeon, B. R. Cuenya, Nat. Catal. 2019, 2, 198; c) I. Merino- c) L. Soler, A. Casanovas, A. Urrich, I. Angurell, J. Llorca, Appl.
Garcia, J. Albo, J. Solla-Gullon, V. Montiel, A. Irabien, J. CO2 Catal., B 2016, 197, 47; d) N. Ta, J. Liu, S. Chenna, P. A. Crozier, Y. Li,
Util. 2019, 31, 135; d) S. D. Giri, A. Sarkar, Adv. Powder Technol. A. Chen, W. Shen, J. Am. Chem. Soc. 2012, 134, 20585.
2018, 29, 3520; e) I. Merino-Garcia, J. Albo, P. Krzywda, G. Mul, [19] a) S. Yao, W. Xu, A. C. Johnston-Peck, F. Zhao, Z. Liu, S. Luo,
A. Irabien, Catal. Today 2020, 346, 34; f) J. Santos-Lorenzo, R. San S. Senanayake, A. Martínez-Arias, W. Liu, J. Rodriguez, Phys.
José-Velado, J. Albo, G. Beobide, P. Castano, O. Castillo, A. Luque, Chem. Chem. Phys. 2014, 16, 17183; b) L. Lin, S. Yao, Z. Liu,
S. Pérez-Yáñez, Microporous Mesoporous Mater. 2019, 284, 128; F. Zhang, N. Li, D. Vovchok, A. Martinez-Arias, R. Castañeda, J. Lin,
g) J. Albo, M. Perfecto-Irigaray, G. Beobide, A. Irabien, J. CO2 Util. S. D. Senanayake, J. Phys. Chem. C 2018, 122, 12934.
2019, 33, 157; h) I. Merino-Garcia, J. Albo, A. Irabien, Nanotech- [20] a) D. Wu, C. Dong, D. Wu, J. Fu, H. Liu, S. Hu, Z. Jiang, S. Z. Qiao,
nology 2017, 29, 014001; i) M. Perfecto-Irigaray, J. Albo, G. Beobide, X.-W. Du, J. Mater. Chem. A 2018, 6, 9373; b) R. Si, J. Raitano, N. Yi,
O. Castillo, A. Irabien, S. Pérez-Yáñez, RSC Adv. 2018, 8, 21092; L. Zhang, S.-W. Chan, M. Flytzani-Stephanopoulos, Catal. Today
j) J. Albo, A. Sáez, J. Solla-Gullón, V. Montiel, A. Irabien, Appl. 2012, 180, 68; c) L. Li, L. Song, C. Chen, Y. Zhang, Y. Zhan, X. Lin,
Catal., B 2015, 176, 709. Q. Zheng, H. Wang, H. Ma, L. Ding, Int. J. Hydrogen Energy 2014,
[4] a) Y. Zhou, F. Che, M. Liu, C. Zou, Z. Liang, P. De Luna, H. Yuan, 39, 19570.
J. Li, Z. Wang, H. Xie, Nat. Chem. 2018, 10, 974; b) Q. Wan, J. Zhang, [21] J. Jorda, M. T. S. Cohen-Adad, J. Less-Common Met. 1991, 171, 127.
B. Zhang, D. Tan, L. Yao, L. Zheng, F. Zhang, L. Liu, X. Cheng, [22] A. Devaraj, M. Gu, R. Colby, P. Yan, C. M. Wang, J. Zheng, J. Xiao,
B. Han, Green Chem. 2020, 22, 2750. A. Genc, J. Zhang, I. Belharouak, Nat. Commun. 2015, 6, 1.
[5] a) R. Reske, H. Mistry, F. Behafarid, B. Roldan Cuenya, P. Strasser, J. Am. [23] S.-H. Kim, P. W. Kang, O. O. Park, J.-B. Seol, J.-P. Ahn, J. Y. Lee,
Chem. Soc. 2014, 136, 6978; b) G. L. De Gregorio, T. Burdyny, A. Loiudice, P.-P. Choi, Ultramicroscopy 2018, 190, 30.
P. Iyengar, W. A. Smith, R. Buonsanti, ACS Catal. 2020, 10, 4854. [24] J. E. Schmidt, R. Oord, W. Guo, J. D. Poplawsky, B. M. Weckhuysen,
[6] a) X. Feng, K. Jiang, S. Fan, M. W. Kanan, J. Am. Chem. Soc. 2015, Nat. Commun. 2017, 8, 1.
137, 4606; b) C. W. Li, J. Ciston, M. W. Kanan, Nature 2014, 508, 504; [25] D. Isheim, D. N. Seidman, Surface and Interface Analysis: An Inter-
c) R. G. Mariano, K. McKelvey, H. S. White, M. W. Kanan, Science national Journal devoted to the development and application of tech-
2017, 358, 1187. niques for the analysis of surfaces, interfaces and thin films 2004, 36,
[7] H. Xiao, W. A. Goddard, T. Cheng, Y. Liu, Proc. Natl. Acad. Sci. USA 569.
2017, 114, 6685. [26] O. C. Hellman, J. A. Vandenbroucke, J. Rüsing, D. Isheim,
[8] T.-C. Chou, C.-C. Chang, H.-L. Yu, W.-Y. Yu, C.-L. Dong, D. N. Seidman, Microsc. Microanal. 2000, 6, 437.
J.-J. s. Velasco-Vélez, C.-H. Chuang, L.-C. Chen, J.-F. Lee, J.-M. Chen, [27] N. Kosugi, H. Kondoh, H. Tajima, H. Kuroda, Chem. Phys. 1989,
J. Am. Chem. Soc. 2020, 142, 2857. 135, 149.
[9] K. K. Patra, S. Park, H. Song, B. Kim, W. Kim, J. Oh, ACS Appl. [28] a) A. Chen, X. Yu, Y. Zhou, S. Miao, Y. Li, S. Kuld, J. Sehested, J. Liu,
Energy Mater. 2020, 3, 11343. T. Aoki, S. Hong, Nat. Catal. 2019, 2, 334; b) K. Yu, L. L. Lou, S. Liu,
[10] a) S. Hilaire, X. Wang, T. Luo, R. Gorte, J. Wagner, Appl. Catal., A W. Zhou, Adv. Sci. 2020, 7, 1901970.
2001, 215, 271; b) W. Deng, M. Flytzani-Stephanopoulos, Angew. [29] a) G. Kresse, J. Hafner, Phys. Rev. B 1993, 47, 558; b) G. Kresse,
Chem., Int. Ed. 2006, 45, 2285; c) Q. Fu, H. Saltsburg, M. Flytzani- J. Furthmüller, Phys. Rev. B 1996, 54, 11169; c) G. Kresse, D. Joubert,
Stephanopoulos, Science 2003, 301, 935. Phys. Rev. B 1999, 59, 1758; d) B. Hammer, L. B. Hansen,
[11] a) J. S. Elias, K. A. Stoerzinger, W. T. Hong, M. Risch, L. Giordano, J. K. Nørskov, Phys. Rev. B 1999, 59, 7413.
A. N. Mansour, Y. Shao-Horn, ACS Catal. 2017, 7, 6843; [30] J. K. Nørskov, J. Rossmeisl, A. Logadottir, L. Lindqvist, J. R. Kitchin,
b) M. Cargnello, V. V. Doan-Nguyen, T. R. Gordon, R. E. Diaz, T. Bligaard, H. Jonsson, J. Phys. Chem. B 2004, 108, 17886.
E. A. Stach, R. J. Gorte, P. Fornasiero, C. B. Murray, Science 2013, [31] L. Sun, B. Yildiz, J. Phys. Chem. C 2018, 123, 399.
341, 771; c) M. Piumetti, T. Andana, S. Bensaid, N. Russo, D. Fino, [32] a) M. Yu, D. R. Trinkle, J. Chem. Phys. 2011, 134, 064111; b) E. Sanville,
R. Pirone, Nanoscale Res. Lett. 2016, 11, 1; d) H. Zhu, Y. Chen, S. D. Kenny, R. Smith, G. Henkelman, J. Comput. Chem. 2007, 28,
Z. Wang, W. Liu, L. Wang, RSC Adv. 2018, 8, 14888. 899.

Adv. Mater. 2023, 35, 2208996 2208996 (10 of 11) © 2022 Wiley-VCH GmbH
 15214095, 2023, 8, Downloaded from [Link] by INDIAN INSTITUTE OF TECHNOLOGY (NON -EAL), Wiley Online Library on [12/01/2025]. See the Terms and Conditions ([Link] on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
[Link] [Link]

[33] W. Sun, D. A. Kitchaev, D. Kramer, G. Ceder, Nat. Commun. 2019, [37] a) J. Klimeš, D. R. Bowler, A. Michaelides, Phys. Rev. B 2011, 83,
10, 1. 195131; b) S. Grimme, J. Comput. Chem. 2006, 27, 1787; c) J. Klimeš,
[34] a) L. Kang, B. Wang, Q. Bing, M. Zalibera, R. Büchel, R. Xu, D. R. Bowler, A. Michaelides, J. Phys.: Condens. Matter 2009, 22, 022201.
Q. Wang, Y. Liu, D. Gianolio, C. C. Tang, Nat. Commun. 2020, 11, [38] a) V. I. Anisimov, F. Aryasetiawan, A. Lichtenstein, J. Phys.: Condens.
1; b) S. Chu, X. Yan, C. Choi, S. Hong, A. W. Robertson, J. Masa, Matter 1997, 9, 767; b) V. I. Anisimov, J. Zaanen, O. K. Andersen,
B. Han, Y. Jung, Z. Sun, Green Chem. 2020, 22, 6540. Phys. Rev. B 1991, 44, 943; c) S. Dudarev, G. Botton, S. Y. Savrasov,
[35] a) S. Verma, Y. Hamasaki, C. Kim, W. Huang, S. Lu, H.-R. M. Jhong, Z. Szotek, W. Temmerman, A. Sutton, Phys. Status Solidi A 1998,
A. A. Gewirth, T. Fujigaya, N. Nakashima, P. J. Kenis, ACS Energy 166, 429.
Lett. 2017, 3, 193; b) B. Kim, H. Seong, J. T. Song, K. Kwak, H. Song, [39] a) D. Marrocchelli, S. R. Bishop, H. L. Tuller, G. W. Watson, B. Yildiz,
Y. C. Tan, G. Park, D. Lee, J. Oh, ACS Energy Lett. 2019, 5, 749. Phys. Chem. Chem. Phys. 2012, 14, 12070; b) L. Wang, T. Maxisch,
[36] K. Thompson, D. Lawrence, D. Larson, J. Olson, T. Kelly, G. Ceder, Phys. Rev. B 2006, 73, 195107.
B. Gorman, Ultramicroscopy 2007, 107, 131. [40] M. Nolan, G. W. Watson, J. Phys. Chem. B 2006, 110, 16600.

Adv. Mater. 2023, 35, 2208996 2208996 (11 of 11) © 2022 Wiley-VCH GmbH