You are on page 1of 33

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/342945928

Active Site Engineering in Porous Electrocatalysts

Article  in  Advanced Materials · November 2020


DOI: 10.1002/adma.202002435

CITATIONS READS

164 728

6 authors, including:

Hui Chen Liu Yipu


Jilin University Jilin University
73 PUBLICATIONS   2,373 CITATIONS    54 PUBLICATIONS   3,835 CITATIONS   

SEE PROFILE SEE PROFILE

Xuan Ai Tewodros Asefa


Jilin University Rutgers, The State University of New Jersey
22 PUBLICATIONS   805 CITATIONS    244 PUBLICATIONS   22,195 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Polypyrrole-Derived Nitrogen and Oxygen Co-Doped Mesoporous Carbons as Efficient Metal-Free Electrocatalyst for Hydrazine Oxidation View project

Energy conversion and catalytic materials View project

All content following this page was uploaded by Hui Chen on 02 May 2021.

The user has requested enhancement of the downloaded file.


Review
www.advmat.de

Active Site Engineering in Porous Electrocatalysts


Hui Chen, Xiao Liang, Yipu Liu, Xuan Ai, Tewodros Asefa, and Xiaoxin Zou*

between electrical and chemical energy to


Electrocatalysis is at the center of many sustainable energy conversion store off-peak electricity produced from
technologies that are being developed to reduce the dependence on fossil these resources. The opportunities relate
fuels. The past decade has witnessed significant progresses in the exploi- to the fact that electrocatalytic reactions
tation of advanced electrocatalysts for diverse electrochemical reactions can be employed to convert these excess
and off-peak electricity into chemical
involved in electrolyzers and fuel cells, such as the hydrogen evolution bonds in molecules. A fascinating pros-
reaction (HER), the oxygen reduction reaction (ORR), the CO2 reduction pect is the utilization of renewable elec-
reaction (CO2RR), the nitrogen reduction reaction (NRR), and the oxygen tricity for the conversion of abundant
evolution reaction (OER). Herein, the recent research advances made in resources such as H2O, N2, and CO2
porous electrocatalysts for these five important reactions are reviewed. In into synthetic fuels such as hydrogen,
ammonia, hydrocarbons, and alcohols
the discussions, an attempt is made to highlight the advantages of porous
under ambient conditions (Figure  1).
electrocatalysts in multiobjective optimization of surface active sites Through the reverse processes, clean elec-
including not only their density and accessibility but also their intrinsic tricity can be generated from the products,
activity. First, the current knowledge about electrocatalytic active sites for example, via hydrogen-, ammonia-,
is briefly summarized. Then, the electrocatalytic mechanisms of the five or alcohol-powered fuel cells, ultimately
above-mentioned reactions (HER, ORR, CO2RR, NRR, and OER), the cur- resulting in a closed water cycle, nitrogen
cycle, or carbon cycle.
rent challenges faced by these reactions, and the recent efforts to meet Hydrogen production via electrocata-
these challenges using porous electrocatalysts are examined. Finally, the lytic water splitting, which may enable the
future research directions on porous electrocatalysts including synthetic hydrogen economy, is a notable example to
strategies leading to these materials, insights into their active sites, and these.[2] At present, the annual hydrogen
the standardized tests and the performance requirements involved are production worldwide exceeds more than
65 million tons, mainly for industrial uses
discussed.
such as petroleum refining and ammonia
synthesis. Unfortunately, most of the
hydrogen is produced from fossil fuels
1. Introduction through steam methane reforming and coal gasification and
is accompanied by large emission of the greenhouse gas CO2.
Electrical energy produced from renewable energy resources Producing hydrogen from water using renewable electricity
(e.g., photovoltaic cells, hydropower, and wind turbines) has provides a green and sustainable pathway for future hydrogen
been increasingly contributing to the electricity generation energy cycle. In a similar vein, renewable energy-powered elec-
market.[1] This has also been creating both challenges and trocatalysis of CO2 and N2 reduction produces high-value fuels
opportunities. The challenges arise from the intermittent and or fine chemicals such as formic acid (HCOOH), carbon mon-
unpredictable nature of renewable energy resources (e.g., oxide (CO), methanol (CH3OH), and ammonia (NH3) under
wind and sunlight), which necessitate efficient interconversion ambient conditions.[3] The use of these renewable fuels (e.g.,
hydrogen and alcohols), instead of petroleum-based fuels, for
Dr. H. Chen, X. Liang, Dr. Y. Liu, X. Ai, Prof. X. Zou transportation applications is receiving unprecedented attention
State Key Laboratory of Inorganic Synthesis and Preparative Chemistry because the former are more environmentally friendly and sus-
College of Chemistry tainable. Fuel cell technologies based on these fuels can reliably
Jilin University supply electrical energy and are, thus, highly desired for running
Changchun 130012, P. R. China
E-mail: xxzou@jlu.edu.cn
emerging electric vehicles.[4]
Prof. T. Asefa
To realize the water cycle, carbon cycle, and nitrogen cycle
Department of Chemistry and Chemical Biology & Department of Chemical described above, the central reactions are the hydrogen evolu-
and Biochemical Engineering tion reaction (HER), the CO2 reduction reaction (CO2RR), and
Rutgers the nitrogen reduction reaction (NRR), as well as the oxygen-
The State University of New Jersey related reactions, including the oxygen evolution reaction
Piscataway, NJ 08854, USA
(OER) and the oxygen reduction reaction (ORR), which are
The ORCID identification number(s) for the author(s) of this article
can be found under https://doi.org/10.1002/adma.202002435.
important half reactions in electrolyzers and fuel cells, respec-
tively. In these energy conversion processes, electrocatalysts
DOI: 10.1002/adma.202002435 are indispensable. The desired electrocatalysts should both

Adv. Mater. 2020, 32, 2002435 2002435  (1 of 32) © 2020 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.advmat.de

exhibit high performances (e.g., activity, stability, and product


selectivity) and have minimal or no noble metals such as Pt, Ir, Hui Chen received his Ph.D.
and Ru. This means, such efficient and sustainable catalysts for in materials science from
these reactions should be developed. Till then, the catalytic per- Jilin University in June 2018.
formances of existing noble metal-free electrocatalysts for HER, He is currently a postdoc-
OER, and ORR should be significantly improved to be compa- toral researcher in State
rable with those of the benchmark noble metals (e.g., Pt for Key Laboratory of Inorganic
HER and ORR and IrO2 for OER).[5] In addition, electrocatalysts Synthesis and Preparative
currently available to activate inert CO2 and N2 molecules, for Chemistry in Jilin University.
CO2RR and NRR, respectively, still suffer from unsatisfactory His research interests focus
efficiency and selectivity of products;[6] thus, highly effective, on the synthesis and applica-
stable, and selective catalysts for CO2RR and NRR should be tions of functional inorganic
constructed. Fundamental understanding and elaborate modu- materials in chemical sensing
lation of catalytically active sites are also needed as they are at and catalysis.
the core for guiding and enabling the rational design of effec-
tive electrocatalysts for these reactions. Xiao Liang is currently
The catalytically active sites of desired electrocatalysts should pursuing her Ph.D. degree
satisfy at least three criteria simultaneously (Figure  2): i) they under the supervision of
should have high intrinsic activity, ii) they should be in large Prof. Xiaoxin Zou at State
­density on the surfaces of the catalysts, and iii) they should be Key Laboratory of Inorganic
easily accessible to reactant molecules. The intrinsic activity of Synthesis and Preparative
an active site, which is related to the inherent surface atomic and Chemistry, Jilin University,
electronic structures, can be reflected by its binding energy to P. R. China. Her current
reactive intermediates. Simple descriptors (e.g., hydrogen adsorp- research focus on nano-
tion free energy (ΔGH*) for HER and ΔGO*  −  ΔGOH* for OER) sized perovskites and their
have been established to help evaluate the intrinsic activity of a application in water splitting
given catalytic site. On the one hand, approaches such as heter- electrocatalysis.
oatom incorporation, defect and strain engineering, and single-
atom-site construction, are applied to optimize the adsorption
energy and to increase the intrinsic activity of catalysts.[7] On Xiaoxin Zou has received his
the other hand, attractive approaches to improve the catalytic Ph.D. in inorganic chemistry
activity of a catalyst focus on exposing more active sites and/or from Jilin University (China)
enhancing the accessibility of active sites on the catalyst’s sur- in June 2011; and then
faces to r­eactants.[8] Both approaches are strongly dependent on moved to the University of
the composition and structure of the catalyst at the nanometer California, Riverside, and
scale. Rutgers, The State University
The ease of combining the different aforementioned strate- of New Jersey, as a post-
gies for active site engineering in a porous material endows doctoral scholar from July
such electrocatalysts with appealing catalytic performances 2011 to October 2013. He
toward various reactions including HER, ORR, CO2RR, NRR, is currently a professor at
and OER. Having porous structures in the catalysts can offer the State Key Laboratory
some inherent advantages: large specific surface area, and thus of Inorganic Synthesis and Preparative Chemistry in Jilin
increased density of surface active sites, as well as an ability University. His research interests include electrocatalysis,
to efficiently transfer reactants/products through their struc- nanocatalysis, hydrogen energy, solid state chemistry,
tures and enhanced accessibility of their surface active sites. structural chemistry, perovskites, and intermetallics.
During the formation of porous structures, the process can
also improve the intrinsic activity of catalytic sites or intention-
ally be used to do so, as demonstrated in many reports.[9] This
is possible because the pore-making process can create struc- Template-assisted synthetic processes including hard and
tural defect sites such as edges, corners, vacancies, and grain soft templating methods are the most common ways to finely
boundaries, which can often tune the catalytic activity of the design the pore structures of target materials. In addition,
material. The porous substrates can also serve as good support some versatile strategies such as dealloying, surfactant-assisted
materials for catalytically active nanoclusters and even single- synthesis, and pyrolysis of metal-containing precursors (e.g.,
metal active sites. As a result, a large number of porous mate- metal–organic frameworks (MOFs)) have been developed to
rials, including metals/alloys,[10] metal oxides (e.g., spinel- and synthesize different types of porous materials. Readers who
perovskite-type oxides),[11] nonoxide compounds (e.g., sulfides may seek more detailed information on synthetic strategies
and nitrides),[12] and carbon-based materials,[13] have recently and mechanisms leading to such materials and characteriza-
been studied for electrocatalysis of the reactions involved in tion techniques for various porous materials can check several
the water and carbon cycles mentioned earlier. comprehensive reviews published elsewhere.[14] In this review,

Adv. Mater. 2020, 32, 2002435 2002435  (2 of 32) © 2020 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.advmat.de

Figure 1.  The schematic of future water cycle, nitrogen cycle, and carbon cycle based on electrochemical energy conversion.

we do not intend to offer a similar review of these topics, but at active sites possess much higher activity than other surface
rather concentrate on the fundamental aspects of active sites atoms to catalyze a specific chemical reaction. Identification of
and active site engineering in porous electrocatalysts. We first catalytic sites is critical to deepen our understanding of electro-
present a brief summary of current knowledge about active catalytic reactions and to guide the design of better electrocata-
sites in different types of electrocatalysts to give a framework lysts. The traditional approach for active sites identification in
of preliminary understanding of the origins of their activity. We heterogeneous catalysts is to prepare single crystalline catalysts
then discuss the electrocatalytic mechanisms of the HER, ORR, of different sizes, on the assumption that the number of corners
CO2RR, NRR, and OER, the current challenges facing these and edges linearly varies with particle sizes.[15] Establishing the
reactions, and the recent efforts being made to meet these chal- definite correlations between the activities of these catalysts and
lenges with porous electrocatalysts. In the discussion, we pro- their corresponding particle sizes can then be used to identify
vide only representative examples of porous electrocatalysts and the active sites. A shortcoming of this method is that it is too
do not give an exhaustive review of them. We also attempt to indirect to provide information about the active sites. In recent
emphasize the following core philosophy: porous electrocata- years, some in situ techniques for monitoring the surfaces and
lysts can achieve multiobjective optimization of surface active solid/liquid interfaces of catalysts, including attenuated total
sites including not only their density and accessibility but also reflectance surface-enhanced infrared absorption spectroscopy
their intrinsic activity. We finally propose and discuss several (ATR-SEIRAS), ambient pressure X-ray photoelectron spectros-
possible future research directions on porous electrocatalysts. copy (APXPS), surface enhanced Raman spectroscopy, surface
interrogation scanning electrochemical microscopy (SI-SECM),
and scanning tunneling microscopy (STM), have been devel-
2. A Brief Summary of Current Knowledge about oped.[16] These advanced techniques can provide important
direct evidences regarding the active sites and adsorbed inter-
Electrocatalytic Active Sites
mediates on catalyst surfaces and give better insights into reac-
Catalytically active sites refer to some specific surface sites tion mechanisms. In addition, density functional theory (DFT)
that contribute to the activity of a catalyst. The surface atoms calculations, which can serve as useful complementary tools to
experiments, can help with identifying the active sites involved
in various electrocatalytic reactions.

2.1. Active Sites of Metal-Containing Electrocatalysts

For metal-containing electrocatalysts, including metals, metal


oxides, phosphides and chalcogenides, their compositions,
crystal structures, and exposed crystal facets generally govern
the properties of their catalytic active sites. Correspondingly,
Figure 2.  The schematic of desired electrocatalysts with large density of optimization of the composition and crystal phase via processes
accessible, highly active catalytic sites. such as alloying is a common strategy to improve the activity

Adv. Mater. 2020, 32, 2002435 2002435  (3 of 32) © 2020 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.advmat.de

Figure 3.  Some typical active sites in different electrocatalysts. a) Unsaturated coordination sites and b) defect sites in metal-containing materials.
c) Doping and defect sites in carbon materials. d) Atomic metal active sites on different support materials.

of the catalytic site and to reduce the amount of noble metals to CO product. Other metals (e.g., Au, Pd, and Cu) have also
needed in the catalysts. Some of these processes can also modify been found to have such geometrical site-dependent catalytic
the electronic structures and surface adsorption at the catalytic activity and selectivity for electrochemical CO2 reduction.[21]
sites promoting electrocatalysis. By taking ORR as a model reac- Materials with high-index planes comprising large numbers of
tion, the Nørskov group investigated a class of Pt- and Pd-based edges, kinks, and steps can have highly active catalytic sites and
binary alloys as electrocatalysts.[17] They showed that the oxygen improved catalytic performances for many reactions. The Lou
adsorption energy on Pt or Pd surfaces could be tuned by incor- group showed the synthesis of concave shaped Pt nanoframes
porating transition metals in them. As a result, the Pt- and Pd- enclosed by {740} high-index facets as ORR electrocatalysts.[22] The
based binary alloys exhibited higher electrocatalytic activity for ORR specific activity on the materials was found to be 2.62 and
ORR than the pure Pt and Pd metals. For example, Pt3Y, which 1.60 times higher than those of a commercial Pt/C catalyst and
has optimal oxygen adsorption energy, presented significantly Pt nanocubes, respectively. This demonstrated the importance
more improved catalytic activity toward ORR compared to Pt, of the exposure of catalytically active high-index surface sites in
as confirmed by theoretical and experimental studies. In addi- electrocatalysts.
tion, by controlling the exposed facets and creating distinct Various surface defects, such as vacancies and grain bound-
atomic arrangements at different facets, the surface adsorption aries in metals and metallic compounds, are often regarded as
ability of active sites can be tailored and the catalytic activity of highly desirable sites for electrocatalysis (Figure  3b). Defect
the materials can be improved. For instance, recent studies on sites have a great ability to modify the coordination environ-
Pd nanosheets showed facet-dependent electrocatalytic activity ment and charge distribution of neighboring atoms, thus
toward HER.[18] Notably, the {100}-exposed surface facets modulating the adsorption energies of reaction intermediates.
showed much better catalytic activity and stability in HER than Defect engineering is widely used to generate these sites and
the {110}- and {111}-exposed facets. create greater catalytic activity in electrocatalysts.[23] Methods,
The unsaturated coordination sites, such as terraces, steps, such as plasma treatment, etching, doping, dealloying, and
kinks, and edge and corner sites of different materials, are chemical reduction, have also been extensively explored to
well-known to be the active sites of metals and solid-state introduce different types of defects and tailor their densities
metallic compounds involved in various electrocatalytic reac- and spatial distributions. Anion vacancies (e.g., O, S, and
tions (Figure  3a). The coordination numbers of surface atoms P vacancies) and cation vacancies are the most common and
are directly related to their adsorption energies.[19] Unsaturated easily controlled defects in catalysts composed of metal oxides,
atoms with lower coordination numbers are more thermody- sulfides, and phosphides. Such sites can be created by various
namically unstable and active to adsorb different reaction inter- methods and lead to effective electrocatalysts. For example,
mediates compared to coordinatively saturated ones. To illustrate P vacancies created in nickel phosphides were reported
this scenario, the edges and corner sites of Ag nanoparticles can to weaken the hybridization between d orbitals of Ni and
be taken as example: the former serve as active sites for CO2RR p orbitals of P, making the neighbouring Ni and P atoms more
leading to CO and the latter serve as active sites for HER, a com- effective catalytic active sites for HER.[24] As a consequence,
petitive reaction with CO2RR.[20] This is because the Ag edge the average turnover frequency (TOF) of each new active site
sites facilitate CO2 adsorption and stabilize the intermediate increased by two orders of magnitude compared with those in
COOH*, favoring the formation of CO. Thus, increasing the the vacancy-free catalyst. When compared with point defects,
ratio between edge and corner sites in Ag nanoparticles, by con- line defects (e.g., grain boundaries) as active sites have often
trolling the sizes of the particles, can lead to superior selectivity been ignored because of their poor controllability and low

Adv. Mater. 2020, 32, 2002435 2002435  (4 of 32) © 2020 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.advmat.de

density. Recently, electrocatalytically active, atomically thin tune the local coordination environment, tailor the electronic
MoS2 with ultrahigh-density grain boundaries (≈1012 cm−2) was transfer from the metal atoms to the support materials, and
grown using a vapor-phase growth method.[25] The material optimize the chemisorption energies of reaction intermedi-
efficiently electrocatalyzed HER, mainly due to the intrinsically ates. The properties of the support materials (e.g., composition,
high catalytic activity of its grain boundaries. nanostructure, and synthetic history) can significantly affect
the coordination environment and mechanisms of various elec-
trocatalytic reactions. In addition, the incorporation of foreign
2.2. Active Sites of Metal-Free Carbon-Based Electrocatalysts atoms (e.g., N, B, S) at desired sites of support materials may
tune the electronic properties of the underlying support mate-
Unlike metal-containing materials, pristine carbons do not rials and enhance activity of the catalysts. This is how some
show much electrocatalytic activity by themselves. However, noble metal-free nanocarbon materials containing M–N–C sites
the introduction of nonmetal dopants (e.g., N, O, S, B, and P) show as high electrocatalytic activity as Pt-based materials for
into these otherwise inert materials has been found to trans- ORR.[33] In some cases, the defect sites (e.g., edge and topo-
form them into active electrocatalysts for HER, OER, ORR, or logical defects) in the support materials impart positive effects
CO2RR.[26] Owing to their differences with carbon atoms in on single-atom catalysts. The defect sites can also facilitate the
terms of electronegativity, bonding states and atomic sizes, the immobilization of metal centers to generate more active sites
dopant atoms create electrocatalytic active sites by modulating and improve the electronic properties and surface reactivities of
the charge distribution and spin density of neighboring carbon the materials.[34]
atoms and by optimizing their adsorption/desorption properties
toward key intermediates. For instance, Guo et al. demonstrated
that the carbon atoms adjacent to pyridinic nitrogen dopants in 3. Porous HER Electrocatalysts
nitrogen-doped graphitic carbons could serve as active sites for
ORR.[27] Recent experimental and theoretical studies also con- 3.1. Reaction Mechanism at HER Active Sites
firmed that dopant-induced and intrinsic defect sites in carbon
materials were instrumental to the materials’ electrocatalytic HER is an important electrocatalysis process that takes place at
activity (Figure 3c).[28] Tang et al. found that the nitrogen-doped the cathode of water splitting electrolyzers and is responsible
graphene mesh could electrocatalyze ORR and OER, and that for the generation of hydrogen (H2) from water. As shown in
not only the nitrogen dopants but also topological defects on Figure 4a, HER is a two-electron transfer and multistep process,
the material were the origins of the catalytic activities.[29] In which involves adsorbed H* (where H* denotes a H atom at an
particular, dopants or defect sites near the edges of the material active site of the catalyst) as the only reaction intermediate. This
(also known as edge effects) exhibited higher activity.[30] Even for intermediate forms under both acidic and alkaline conditions.
undoped carbon materials, defect engineering can be applied The difference between acidic and alkaline HER lies in what the
to activate the carbon π electrons and generate active sites for source of H* is in them. While H* comes from H3O+ in acidic
electrocatalysis. Recently, some dopant-free nanocarbon mate- media, it comes from H2O in alkaline media during the reaction.
rials with intrinsic defect sites have been demonstrated to show There are two different reaction pathways leading to HER,
efficient electrocatalytic activity for ORR.[31] A combined theo- namely, the Volmer–Tafel mechanism and the Volmer–­
retical and experimental study conducted by Yao and co-workers Heyrovsky mechanism. The possible reaction pathway and rate-
showed that the pentagonal defects at edges in a pyrolytic determining step (RDS) at an active site can be determined by
graphite not only were the major catalytic active sites but also experimental Tafel slope analysis. A Tafel slope of ≈30 mV dec−1
had a better activity than pyridinic N sites in a N-doped pyrolytic indicates the Volmer–Tafel mechanism, where the recombina-
graphite for ORR.[31a] Based on these reports, we believe that the tion of two adsorbed H* at two adjacent active sites (i.e., the
collective use of dopants, defects, and edge sites may help with Tafel step: H* + H* → H2) is the rate-determining step.[35]
the design of efficient metal-free electrocatalysts. We also believe Higher Tafel slopes of ≈120 and ≈40 mV dec−1 indicate that the
that designing materials with such structural features will con- Volmer step (the formation of adsorbed H*) and the Heyrovsky
tinue to be an important research topic in the near future. step (the combination of an adsorbed H* with a proton in elec-
trolyte), respectively, are the rate-determining steps.[36] In addi-
tion, comparative transition-state calculations of the activation
2.3. Active Sites of Single-Atom Electrocatalysts energies of the different steps can offer a deeper understanding
of the reaction pathway(s) leading to HER. In alkaline HER, the
Single-atom catalysts (SACs) have emerged as exciting mate- water dissociation barrier is another major factor affecting the
rials for electrocatalysis of different reactions, because the overall reaction rate. Consequently, superior HER electrocata-
atomically isolated atoms, usually metals anchored on sup- lysts in alkaline media possess a moderate H* binding energy
port materials, allow efficient use of metals as active sites. So as well as offer a relatively low dissociation barrier to water.[37]
far, SACs with at least 20 different metals (e.g., Ru, Ir, Mo, Fe) The rate of overall HER reaction is strongly dependent on
have been synthesized for diverse reactions.[32] These metal the binding strength of the adsorbed H* species. According to
atoms can act as highly electrocatalytically active sites when the Sabatier principle, a good HER catalyst should bind the H*
coordinated with support materials (e.g., the nanocarbons and species only moderately to its active sites. If hydrogen strongly
metal oxides shown in Figure 3d). Their high catalytic activities interacts to the surface of the catalyst, the initial adsorption pro-
is often due to the strong metal–support interactions, which cess of hydrogen (the Volmer step) is easy, but the subsequent

Adv. Mater. 2020, 32, 2002435 2002435  (5 of 32) © 2020 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.advmat.de

Figure 4.  a) The mechanism of HER at metallic catalytic active sites in acidic (green route) and alkaline (blue route) conditions. b) The volcano plot of
HER for MoS2 and some representative metals. b) Reproduced with permission.[38b] Copyright 2017, American Chemical Society. c) Synthetic process
leading to and structural model of ordered mesoporous MoS2 with a double-gyroid morphology. c) Reproduced with permission.[40] Copyright 2012,
Nature Research. d) Synthesis procedure and microstructures of ordered macroporous S-MoS2@C. d) Reproduced with permission.[42] Copyright 2019,
Wiley-VCH.

desorption process (the Tafel or Heyrovsky step) will be slow planes. Thus, to obtain highly efficient MoS2 materials, the
and hinder the reaction rate. On the other hand, if hydrogen fraction of exposed edge sites relative to the basal plane sites
binds to the catalyst surface too weakly, the Volmer step is inhib- needs to be increased. To this end, the Jaramillo group reported
ited and limits the overall reaction. The Gibbs free energy of the synthesis of highly ordered mesoporous MoS2 through elec-
hydrogen adsorption (ΔGH*)—a quantitative description of the trodeposition of Mo around double-gyroid silica templates, then
H* binding strength—is a widely accepted activity descriptor sulfidation of the Mo grown around the templates, and etching
of HER. The ΔGH* values and the experimentally obtained away the templates (Figure 4c).[40] The high surface curvature of
exchange current densities have a volcano relationship for a the mesoporous MoS2 structure inhibited the extended growth
wide range of electrocatalytically active materials (Figure 4b).[38] of basal planes, thereby enabling the preferential formation of
A good HER catalyst should have a near zero ΔGH* value edge sites-rich structures.
(ΔGH*  ≈ 0), just like the benchmark Pt catalyst has. However, In another study, Deng et al. prepared uniform mesoporous
as Pt is extremely scarce, it is desirable to develop alternative MoS2 foam with vertically aligned MoS2 layers and vertical
earth-abundant catalysts or reduce the amount of Pt in a given edge-rich sites on the pore walls of the material.[41] The authors
catalyst without compromising its activity. Over the past several further activated the catalytically inert basal planes of the
years, remarkable advances have been made to develop alter- material by incorporating Co atoms into its structures. Mod-
native non-Pt electrocatalysts as well as those containing low erate Co doping increased the electronic states near the Fermi
amounts of Pt. Many of these advancements have been made level and tuned the Bader charge of in-plane S sites, resulting
by relying on some understanding of active sites and reaction in a moderate ΔGH* value and a high catalytic activity for HER
mechanisms of HER on different existing catalysts. at the active sites of the material. The large number of edge
active sites, combined with additional in-plane active sites,
enabled this catalyst to electrocatalyze the reaction at a current
3.2. Engineering Active Sites in Porous HER Electrocatalysts density of 10 mA cm−2 with a low overpotential of 156 mV.
Besides a low density of catalytic active sites, another factor
3.2.1. Exposing Edge Sites of Porous 2D Layered Catalysts which limits the catalytic activity of MoS2 for HER is its poor
electronic conductivity. In order to facilitate electron transfer as
Early in 2005, Nørskov and co-workers theoretically predicted well as expose more edge sites in MoS2, the Li group developed
that MoS2, a 2D layered material, had Pt-like ΔGH* value at an ordered hybrid macroporous catalyst containing ultrasmall
Mo(1010) edge sites.[39] The ability of the edges of this material MoS2 nanosheets and carbon framework (MoS2@C) by con-
to serve as catalytic active sites for HER was confirmed experi- trolled decomposition of ammonium thiomolybdate-glucose-
mentally later on by Jaramillo et al., who found that the catalytic coated polystyrene nanosphere on polyaniline-modified carbon
activity of MoS2 scaled linearly with its edge length rather than cloth (Figure 4d).[42] The authors found that a subsequent sodia-
basal plane surface area.[38a] However, bulk MoS2 crystals cannot tion/desodiation process could generate S-edge rich surfaces
catalyze HER well due to the dominant exposure of inert basal on MoS2@C, producing a material they called S-MoS2@C. The

Adv. Mater. 2020, 32, 2002435 2002435  (6 of 32) © 2020 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.advmat.de

Figure 5.  a) Possible bimetallic sites on the surface of hierarchical porous Cu–Ti (np-CuTi) and b) their hydrogen-binding energy shown in a volcano
plot. c) The catalytic activities of np-CuTi, Ti-free np-Cu, pure Cu, and Pt/C for HER in 0.1 m KOH electrolyte. a–c) Reproduced with permission.[44]
Copyright 2015, Nature Research. d) The catalytic performances in KOH solutions and e) energy diagrams for NCO, NCS, and NCP. d,e) Reproduced
with permission.[45] Copyright 2018, American Chemical Society.

large density of unsaturated S-edge sites on the resulting mate- alloys that show highly activity for HER.[9b] The above examples
rial played a dominant role in its high catalytic activity for HER, also demonstrated that compared with monometallic catalysts,
as these sites had optimized ΔGH* value as confirmed by DFT multimetallic alloy catalysts would offer more flexibility to
calculations. In addition, the strong MoS2–carbon interaction control composition and active sites. In particular, dealloying
allowed a rapid electron transfer at the catalytic sites while also allows nonessential components to be selectively dissolved and
preventing the aggregation of MoS2 during electrocatalysis. In removed, leaving behind porous materials or alloys with well-
fact, this S-MoS2@C’s catalytic activity for HER is the highest designed catalytic active sites.
reported among all the reported MoS2–carbon materials. Besides metal sulfides and multimetallic alloys, metal phos-
These reports and other related studies have inspired many phides and carbides (which are also known as phosphorus-
researchers to find facile, effective approaches to fabricate other and carbon-bearing alloys) have recently been reported as
hybrid porous electrocatalysts containing 2D layered materials efficient and cost-effective HER electrocatalysts. For instance,
like MoS2.[43] Fang et  al. performed a comparative study on the electrocata-
lytic activities of a class of porous Ni/Co-based nanosheets of
phosphides, selenides, and oxides (labeled as NCP, NCS and
3.2.2. Optimizing Active Sites in Porous Multicomponent Catalysts NCO, respectively) for HER to identify how their electrocata-
lytic properties depend on their anions.[45] They prepared the
Multicomponent catalytic systems, which include various alloys, porous NCO nanosheets by calcining metal ions (Co2+ and
multimetallic compounds, and compounds containing mul- Ni2+)-anchored graphene oxide nanosheets. They then prepared
tiple anions, have been widely studied in catalysis. In the case the porous NCS and NCP nanosheets by selenization and phos-
of electrocatalysis, exploring the ability that the multiple com- phidation of NCO nanosheets, respectively. The activities of
ponents in these materials in creating synergistic effects and the three catalysts for HER were found to be in the order of
highly active catalytic sites for HER has been one major focus. NCP > NCS > NCO (Figure 5d). DFT calculations, together with
For example, Lu et al. found that while HER-inactive Cu and Ti the experimental results, revealed that phosphorus atoms in
surfaces had very weak and very strong hydrogen-binding capa- the catalysts could serve as proton-acceptor sites, modulate the
bilities, respectively, Cu–Ti alloys had unique Cu–Cu–Ti hollow electronic configuration over the catalysts and generate efficient
sites with Pt-like hydrogen binding energy and potentially good active sites for HER (Figure  5e). Control over the active sites
electrocatalytic activity toward HER (Figure  5a,b).[44] With this in porous HER catalysts was also demonstrated for N-doped
in mind, researchers synthesized highly hierarchical porous Mo2C nanosheets by Jia et al.[46] DFT calculations showed that
Cu–Ti alloys by selectively dealloying Al–Cu–Ti, and experi- the incorporation of N into Mo2C created highly active N sites
mentally demonstrated that the resulting material indeed had with a ΔGH* value of 0.07 and activity-enhanced Mo sites with
excellent catalytic activity for HER, even better than that of Pt/C a ΔGH* value of −0.3 eV, which are smaller/better than that of
(Figure 5c). Similarly, Yao et al. designed and prepared porous pure Mo2C (ΔGH* = −0.50 eV). By increasing the amount of N
Pd–Ag–Al alloys with atomically thin Pd–Ag bimetallic surfaces in Mo2C, an optimal ΔGH* value near zero could be achieved
and high density of Pd–Pd–Ag hollow sites with Pt-like cata- at the Mo sites. The N-doping induced a favorable shift in the
lytic activity for HER.[9a] By selectively dealloying ternary Cu– Mo d-band center, which is an important parameter related to
Ru–Mn alloys, Wu et al. synthesized nanoporous binary Cu–Ru the optimization of hydrogen-binding ability of the materials.

Adv. Mater. 2020, 32, 2002435 2002435  (7 of 32) © 2020 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.advmat.de

Figure 6.  a) Schematic illustration of P-doping-induced structural phase transition from c-CoSe2 to o-CoSe2|P. b) ΔGH* values at different sites on
(111) surfaces of o-CoSe2|P and c-CoSe2 for HER. a,b) Reproduced with permission.[47] Copyright 2018, Nature Research. c) Schematic illustration for
the synthesis of H-Co0.85Se|P. d) ΔGH* values for H-Co0.85Se|P, Co0.85Se|P, H-Co0.85Se, and Co0.85Se for HER. c,d) Reproduced with permission.[51]
Copyright 2017, Wiley-VCH.

The N-doped Mo2C catalyzed HER with a current density of confirmed by X-ray photoelectron spectroscopy (XPS) and
10 mA cm−2 at a low overpotential (η) of 99 mV versus a revers- oxygen temperature-programmed oxidation (TPO). The oxygen
ible hydrogen electrode (RHE). vacancies were proposed to act as electron acceptors, besides
Another important class of porous multicomponent cata- promoting the adsorption of water molecules and lowering the
lysts is prepared through phosphorus doping. For example, energy barrier for HER. This mesoporous MoO3−x required
doping phosphorus was found to induce a phase transition an overpotential of 140  mV to catalyze HER with 10  mA cm−2
from cubic phase CoSe2 (c-CoSe2) to orthorhombic phase current density in 0.1 m KOH electrolyte. In another study, the
CoSe2 (o-CoSe2|P).[47] The mechanism leading to phase transi- Zhang group synthesized mesoporous carbon nanowires con-
tion during the phosphorus-doping process was proposed as taining WO2 with a high density of oxygen vacancies using an
shown in Figure 6a. The loss of Se during annealing of c-CoSe2 “oxygen extraction” synthetic strategy.[49] The oxygen vacancies
nanobelts created an abundant vacancy defect sites, which in WO2 led to an increase in the density of states (DOS) near
were then occupied by P atoms. This resulted in the rotation the Fermi level, accelerated the charge transfer, and created
of Se–Se(P) pairs and the subsequent formation of o-CoSe2|P more active catalytic sites, while the mesoporous structures in
nanobelts. DFT calculation confirmed that the new P sites in the carbon nanowires increased the density of accessible active
o-CoSe2|P had a very small ΔGH* value of −0.08 eV (Figure 6b). sites in the materials. These hybrid nanowires catalyzed HER
The DFT calculation also indicated that the Co sites exhibited with a current density of 10  mA cm−2 at an overpotential of
a high affinity toward water molecules and an ability to disso- 58 mV. Their catalytic activity for HER is higher than that of the
ciate them. The synergistic interplay between these new P and aforementioned mesoporous MoO3−x catalyst.
Co sites assisted the catalytic HER over the material. The large Apart from oxygen vacancies, chalcogen vacancies have been
density of highly active and accessible, catalytic sites in this explored in porous chalcogenide-based catalysts for HER. By
porous o-CoSe2|P made it among the most active noble metal- intercalating lithium ions in IrSe2, a porous Se-deficient Li–IrSe2
free catalysts for HER in alkaline solution. with catalytic activity for HER approaching that of Pt was
synthesized by Zheng et  al.[50] DFT calculations showed that
the Se vacancies resulted in lower ΔGH* values at the Ir sites
3.2.3. Creating Defect-Related Active Sites in Porous HER Catalysts (to be −0.09 and −0.02  eV) compared with the ΔGH* values
of the Ir sites in IrSe2 (whose values are higher, 0.13 and
Creating anion vacancies is the most popular approach to pro- −0.16  eV). Moreover, the results obtained with Nyquist plots
duce defect sites in oxides and chalcogenides with improved revealed that the Se vacancies could improve electron mobility
catalytic activity for HER. For example, the Suib group reported and HER kinetics. Due to the copresence of abundant Se
mesoporous MoO3−x containing anion defect sites by a soft tem- vacancy sites, high porosity, and large surface area in it, this
plating synthetic method.[48] The removal of the template via material showed excellent catalytic activity. Similarly, the Feng
calcination not only produced mesoporous structure (with pore group reported a Se-deficient, P-doped Co0.85Se material with
sizes in the range of 20–40  nm and a surface area 52 m2 g−1) efficient catalytic activity for HER (Figure  6c).[51] This material
but also created nonstoichiometric MoO3−x with abundant had a porous structure with a surface area of 74 m2 g−1 and
oxygen vacancy sites. The existence of oxygen vacancies in highly active P-based catalytic sites (Figure 6d). These findings
the material was reflected by its unusual blue color, further and some relevant reports on vacancy–activity relationship can

Adv. Mater. 2020, 32, 2002435 2002435  (8 of 32) © 2020 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.advmat.de

Figure 7.  a) Schematic illustration of the synthesis of Pt confined into N-doped porous carbon matrix. b) Schematic description for isolated Pt atom
with different C/N coordination shells. c) ΔGH* values of Pt site and different C/N coordination shells, in comparison with that of pure graphene.
a–c) Reproduced with permission.[54] Copyright 2018, American Association for the Advancement of Science. d) Schematic illustration of the synthesis
of single Pt atoms in Pt/np-Co0.85Se. e) SEM image and cross-section microstructure (inset) of Pt/np-Co0.85Se. f) Schematic illustration of the experi-
mentally determined HER mechanism of Pt/np-Co0.85Se in neutral media. g) ΔGH* values at different active sites for Co0.85Se (004), Pt/Co0.85Se (004),
and Pt (111). d–g) Reproduced with permission.[57] Copyright 2018, Nature Research.

serve as a guide to rationally design other catalysts with anion atom utilization of Pt, Lou and co-workers produced Pt atoms
vacancies.[52] confined in a N-doped porous carbon matrix by pyrolyzing
The effects of metal cation vacancies on electrocatalytic activ- mesostructured polydopamine (PDA) containing platinum salt
ities have received much less attention compared with those of (Figure 7a).[54] The orbital hybridization between Pt and nearby
anion vacancies. This is because the former entail large forma- nonmetal atoms in the resulting material created new hybrid-
tion energies and are thus difficult to synthesize. A recent work ized electronic states around Pt while the electron transfer from
by Liu et  al. confirmed the effective catalytic centers for HER Pt centers to neighboring C/N atoms increased the electronic
induced by iron cation vacancies in iron-based materials.[53] The states of the latter. The modified density of states and charge
authors synthesized 3D porous structure of ultrathin feroxyhyte density redistribution in the material made both the confined
nanosheets (δ-FeOOH NSs) containing iron cation vacancy-rich Pt atoms and nearby C/N atoms highly catalytic active sites for
sites. The XPS and extended X-ray absorption fine structure HER (Figure  7b,c). As a result, compared with a commercial
(EXAFS) spectra verified the existence of Fe vacancies in the Pt/C catalyst, the mesoporous N-doped carbon with atomically
nanosheets. As revealed by DFT calculations, the second neigh- dispersed Pt atoms displayed 25 times higher mass activity for
boring Fe atom of Fe vacancy constitutes highly active catalytic the reaction. In another work, Wei et  al. successfully synthe-
centers possessing an optimized ΔGH* value. The material sized atomically dispersed Pt species on mesoporous carbon
required a very low overpotential of 108 mV to catalyze OER at via iced-photochemical reduction route, and then showed that
10 mA cm−2, which was smaller than what a pristine δ-FeOOH the material could efficiently electrocatalyze HER.[55] Similarly,
(249 mV) required to do the same. This demonstrated that cre- other noble metal-free single-atom sites (e.g., Co, Ni) anchored
ating metal vacancies could be an effective activation strategy to onto porous carbon-based materials have been successfully
generate new and active catalytic centers in various materials. shown to serve as efficient electrocatalysts for HER.[13,56]
Besides porous carbon materials, porous metal selenides
have recently been demonstrated to serve as support mate-
3.2.4. Constructing Single-Atom Active Sites in Porous rials for single-atom metals. For example, the Tan group syn-
HER Catalysts thesized single Pt atoms on nanoporous cobalt selenide (Pt/
np-Co0.85Se) using an electrochemical selective etching method
Owing to their many remarkable advantages, including good (Figure 7d,e).[57] In the process, a small amount of Co vacancies
stability, high conductivity, and abundant anchoring sites, were formed due to the dissolution of Co from the np-Co0.85Se.
porous carbon materials have been widely used as support These Co vacancies served as anchoring sites for atomically dis-
materials to stabilize single-metal catalytic sites. To maximize persed Pt. In situ and operando X-ray absorption spectroscopy

Adv. Mater. 2020, 32, 2002435 2002435  (9 of 32) © 2020 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.advmat.de

Figure 8.  a) Schematic depiction of ORR mechanism at catalytic active sites. b) The theoretical limiting potential plot for ORR. b) Reproduced with
permission.[65] Copyright 2015, Oxford University Press. c) The volcano plot of catalytic activity of metals for ORR versus their adsorption energies (ΔE)
for oxygen intermediate. c) Reproduced with permission.[64] Copyright 2004, American Chemical Society.

(XAS) analysis revealed that H2 is produced by HER over the atoms (M–N–C), where M represents Fe, Co, Ni, etc.[58] Fe–N–C
Pt/np-Co0.85Se in neutral media through: i) the Volmer step at catalysts are widely considered as the most promising candidates
Co sites and ii) the Tafel step at Pt sites or nearby empty Co to substitute Pt-based catalysts. There is a general consensus
sites (Figure  7f). DFT calculations (Figure  7g) disclosed that that the active sites in M–N–C catalysts are M–Nx moieties. The
the electronic interactions between single-atom Pt and Co0.85Se activity of M–Nx moieties toward ORR were found to decrease
could decrease the water dissociation energy barrier and turn in the order of Fe > Co > Cu > Mn > Ni.[59] At the same time,
the inert Co atoms into more catalytically active sites with metal-free carbon-based catalysts have also attracted prominent
optimized ΔGH* values (−0.083  eV) close to that of Pt sites interests since the discovery in 2009 that N-doped carbon nano-
(−0.079  eV). As a result of its abundant highly active catalytic tubes could serve as highly active catalysts for ORR.[60] Generally,
sites, this porous catalyst showed a high TOF of 3.93 s−1 at defects formed in these metal-free carbon materials due to het-
η = −100 mV and 11 times greater mass activity than the com- eroatoms, carbon vacancies, edges, and local dislocations have
mercial Pt/C catalyst during HER. been believed to be the catalytic active sites for ORR.[61] Despite
the advancements made in the last several years, exactly what
the electrocatalytic active sites in carbon-based materials are
4. Porous ORR Electrocatalysts still remain controversial, and their catalytic stabilities in acidic
media remain a problem. Moreover, most of well-developed cata-
4.1. Reaction Mechanism at ORR Active Sites lysts, even Pt-based catalysts, still need far from-ideal overpoten-
tials to catalyze ORR. This is because there exists a strong linear
The ORR is an essential cathodic half-reaction used in many correlation between adsorption-energies of key reaction interme-
types of fuel cells. Due to the multiple proton–electron transfer diates known as scaling relation. This means, the adsorption of
and inherently sluggish kinetics involved in it, this reaction one intermediate at an active site cannot be optimized individu-
impedes the overall efficiency of fuel cells. Traditionally, Pt ally relative to another based on ORR reaction mechanism.[62]
is used as catalyst for ORR because it is the best one for the Regarding the reaction mechanisms of ORR at a metallic
reaction. However, as Pt is costly and scarce, efforts have been catalytic site (M), two processes are generally accepted. First,
devoted to: i) enhance the catalytic activity of Pt by mixing it with an oxygen molecule adsorbs on the surface of a catalyst and
other inexpensive transition metals and/or by modifying its sur- form M–O2 species. Then, different reaction processes ensue,
faces and ii) develop alternative nonplatinum catalysts, especially depending on the dissociation behaviors of M–O2 or the type
carbon materials containing N-coordinated transition metal of the metal (Figure  8a). By evaluating the oxygen dissociation

Adv. Mater. 2020, 32, 2002435 2002435  (10 of 32) © 2020 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.advmat.de

barrier of a catalyst, the reaction mechanism of ORR can be activity of catalysts for ORR versus their oxygen binding energy
grouped into either dissociative pathway or associative pathway: (ΔEO). The volcano shaped graph shows the state-of-the-art
catalyst Pt at the top.[64] Similar scaling relations also exist for
a) Dissociative pathway
carbon-based ORR electrocatalysts.[66]
M + M − O2 → 2M − O (1) Some descriptors derived from adsorption-energy scaling
relations (such as ΔG(O*) −  ΔG(OH*), d-band center, and eg
occupation) have also been developed to predict and ration-
2M − O + 2H+ + 2e − → 2M − OH (2) ally optimize the performances of ORR catalysts.[67] Alloying
Pt with other transition metals is an efficient way to modulate
2M − OH + 2H+ + 2e − → 2H2 O + 2M (3) the adsorption energy of oxygen intermediates on the metals
and further improve their electrocatalytic activity for ORR.[68]
b) Associative pathway It is generally accepted that the surface adsorption properties
of Pt-based alloy catalysts are determined by their electronic
M − O2 + H+ + e − → M − OOH (4) structure (e.g., d-band center). The incorporation of other metal
atoms into Pt can generate ligand effect (e.g., orbital hybridiza-
tion between Pt and heteroatom) and strain effect (e.g., PtPt
M − OOH + H+ + e − → H2 O2 + M (5) bond length alternation) that modifies the d-band property of Pt
and thus the binding strength of Pt–O. Meanwhile, the scaling
M − OOH + H+ + e − → M − O + H2 O (6) relation describes the limitation in the activity of known cata-
lysts. Even for the optimal catalysts sitting at the top of the vol-
cano curve, their minimum theoretical overpotential for ORR
M − O + H+ + e − → M − OH (7) can be as high as 0.3–0.4 V. Developing effective active sites that
favor specific intermediates or lower the energy barrier for O2
M − OH + H+ + e − → H2 O + M (8) dissociation are highly likely to break the limitation associated
with the current scaling relations.[69]
In a dissociative pathway, M–O2 first dissociates to generate
two M–O moieties by using a second nearby M atom, and
each of the M–O species then reacts with protons/electrons to 4.2. Engineering Active Sites in Porous ORR Electrocatalysts
form M–OH and H2O. In associative pathway, the M–O2 first
reacts with a proton and an electron to produce M–OOH. Then, 4.2.1. Engineering Active Sites in Porous Pt-Based Electrocatalysts
M–OOH leads to either H2O2 in a direct two-electron pathway
or H2O in a four-electron pathway, successively generating Pt’s ability to effectively catalyze ORR has been vital in the
M–O/M–OH at the active site. The latter process is more attrac- realization of many commercial fuel cells. However, Pt is
tive for fuel cells and metal–air batteries because it leads to a scarce and expensive; as a result, scaling-up fuel cells is dif-
higher energy and power density. ficult. Reducing the amount of this precious metal needed
Based on the adsorption modes of O2 at the active site, dif- in each fuel cell requires improving its mass activity, or the
ferent mechanisms of ORR have been proposed. If O2 binds catalytic activity per unit mass of catalyst. The widely accepted
end-on to an active site of a catalyst, the mechanism follows solution for this is simultaneously enhancing its intrinsic
an associative pathway. If O2 adsorbs side-on to the active site activity and increasing the number of catalytic sites available
or bridges two adjacent active sites, the mechanism follows on it for the reaction. Some important progress in recent
a dissociative pathway due to the weakening of OO bond years has been made in the development of porous Pt-based
originated from the occupation of the π* orbital of O2 during ORR electrocatalysts, in which the intrinsic activity of the
adsorption. However, recent studies suggested that ORR could catalytic sites is optimized by tuning the local coordination
readily follow the associative mechanism due to the high environment around the Pt atoms.[70] For example, Cheng
oxygen dissociation barrier.[63] These also mean that the gen- et al. reported the synthesis of grain boundary-rich, porous Pt
eral mechanism of ORR on catalysts could involve the forma- electrocatalyst consisting of 3  nm nanocrystals by a confined
tion and transformation of OOH, O, and OH intermediates at space pyrolysis strategy (Figure  9a).[70a] Their DFT calculation
the active site. demonstrated that the Pt sites at the grain boundaries had a
Theoretical analysis of ORR can be conducted using the higher coordination number than those on isolated Pt grains
model developed by Nørskov et  al.[64] The universal scaling (Figure  9b), and this was supported by X-ray absorption fine
relation among different intermediates exist for various cata- structure (XAFS) analysis. This allowed the former to exhibit
lysts (Figure  8b) due to the similar binding patterns of the more optimal adsorption energies toward oxygenated species
oxygen-containing intermediates to their surfaces.[65] Under during ORR. This, plus its porosity and rich grain boundaries,
this theoretical framework, different catalysts follow a volcano made the material exhibit a sevenfold higher electrocatalytic
trend based on their different surface adsorption features: activity for ORR in comparison with the commercial Pt cata-
While the active site that binds oxygen too strongly is slowed lyst. A fuel cell containing this porous ORR electrocatalyst also
by the proton–electron transfer on O* or OH*, the one that showed a high power density and stability in many cycles, even
binds oxygen too weakly is limited by the formation of OOH*. surpassing the required target set by the U.S. Department of
Figure  8c presents the relationship between the theoretical Energy (DOE) for 2020.

Adv. Mater. 2020, 32, 2002435 2002435  (11 of 32) © 2020 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.advmat.de

Figure 9.  a) Schematic illustration of the synthesis of grain boundary-rich, porous Pt (GBP-Pt). b) The ΔG values of two possible rate-determining
steps on different Pt materials during ORR. a,b) Reproduced with permission.[70a] Copyright 2019, Royal Society of Chemistry. c) Schematic illustration
of the synthesis of PtNi alloy and d) synergistic lattice strain and ligand effects in the catalysts. c,d) Reproduced with permission.[71] Copyright 2019,
American Association for the Advancement of Science.

In another important study, Xia and co-workers reported a (Figure  10b). As a result, this material showed a better catalytic
highly effective porous Pt-based electrocatalyst for ORR, in performance for ORR than both a commercial Pt/C catalyst and
which the coordination and electronic structures of Pt-based most previously reported M–N–C catalysts. The superior catalytic
catalytic sites were optimized by Ni alloying and dealloying.[71] activity of the edge-located Fe–N4 sites for ORR compared with
Specifically, the authors made 1D Pt–Ni alloy strings with a one- the intact Fe–N4 sites was corroborated by Wang et  al.[9c] These
pot solvothermal synthetic method and then selectively etching results demonstrated that introducing porosity in such catalysts
Ni from them (Figure  9c). The resulting PtNi electrocatalyst could endow single-atom catalytic groups around edge-sites with
exhibited about 17 and 14 times higher mass and specific activi- optimized geometry and electronic structure for electrocatalysis.
ties for ORR, respectively, than a commercial Pt/C catalyst. The The incorporation of other heteroatoms into Fe–N–C has
material also remained stable, keeping its catalytic activity even also been successfully applied as a good strategy to enhance
after 50 000 cycles. The PtNi alloy showed a superior intrinsic the intrinsic catalytic activity of these species for ORR.[73]
activity and a better performance for ORR because: i) The com- For example, Yuan et  al. derived N and P dual-coordinated
bined effects of lattice strains and interatomic orbital hybridi- single-atom Fe sites anchored onto porous carbon nanosheets
zation in it weakened the Pt–O binding strength (Figure  9d) (Fe–N/P–C) by preparing polypyrrole hydrogel containing
and ii) the unique porous hollow structure in it enhanced mass phytic acid and Fe ions and then pyrolyzing it.[73a] The polypyr-
transfer and exposed more accessible catalytic active sites. role served as a source of N while the phytic acid served as a
source of P. The resulting pyrolyzed product, Fe–N/P–C, had
a large density of N and P dual-coordinated Fe sites (Fe–N3P),
4.2.2. Constructing Single-Atom Active Sites in Porous a large specific surface area, and a conductive structure. It also
Carbon Electrocatalysts displayed a decent activity and superb durability during elec-
trocatalysis of ORR. DFT calculations indicated that, compared
Single-atom catalysts (particularly Fe–N–C) have emerged as with Fe–N4 sites, the Fe–N3P sites exhibited favorable adsorp-
among the most promising alternative nonplatinum catalysts for tion/desorption properties toward oxygen-containing interme-
ORR. To increase the number of exposed atomic active sites, some diates. The asymmetric coupling between N/P and Fe sites is
synthetic approaches that can anchor isolated single Fe atoms believed to be responsible for the higher catalytic performances
on porous N-doped carbons have recently been proposed.[72] exhibited by the Fe–N3P sites in ORR. Thus, further studies to
For instance, Jiang et  al. developed a pore engineering strategy design other related dual-coordinated metallic sites are worth
to tune the local coordination of pyridine-like N groups to opti- pursuing for electrocatalysis in the future.
mize the electronic structure of Fe–N4 moieties.[72a] They did this
by selectively cleaving CN bonds next to single-atomic Fe sites
using acidic solution as illustrated in Figure  10a. This process 4.2.3. Creating Nonmetallic Active Sites in Porous
produced hierarchically porous carbon-supported Fe–N4 moieties Carbon Electrocatalysts
located at pore edges. The edge-hosted Fe–N4 configuration (FeN4-
6r-c2) helped with adsorption of intermediates of ORR and low- As discussed earlier, it is of considerable interest to completely
ered the energy barriers associated with the overall ORR process substitute Pt-based ORR catalysts with alternative metal-free

Adv. Mater. 2020, 32, 2002435 2002435  (12 of 32) © 2020 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.advmat.de

Figure 10.  a) Porosity-induced edge-site engineering in Fe–N–C catalysts. b) Free-energy diagrams for ORR of pristine Fe–N4 and edge-hosted Fe–N4
sites. a,b) Reproduced with permission.[72a] Copyright 2018, American Chemical Society. c) Schematic illustration of the synthesis of N-modified
S-defect-containing carbon aerogel. d) LSV curves for ORR over N-modified S-defect carbon aerogel and Pt/C in 0.1 m HClO4 solution. e) Free-energy
diagrams for ORR on various catalytic sites. c–e) Reproduced with permission.[78] Copyright 2018, Elsevier B.V.

catalysts. Many recent studies have demonstrated that heter- notable findings in the report: i) increasing the proportions of
oatom-doped porous carbon materials are promising nonplat- quaternary N centers in the carbon materials favored the four-
inum catalysts for ORR.[74] For instance, Jiang et  al. reported electron reduction process over the two-electron reduction pro-
porous, nitrogen-doped carbon nanosheets with an ultrahigh cess during ORR; ii) the oxygen dopants further enhanced the
specific surface area of 1793 m2 g−1 and an abundant amount catalytic activity of the materials for ORR; and iii) the metal-
of edge defect sites that show excellent electrocatalytic activity free materials exhibited higher activity for ORR than their
for ORR.[75] The catalytic activity of these metal-free catalysts for metal-containing counterparts. These exciting findings on
ORR in acidic media were found to be comparable to that of heteroatom-codoped metal-free carbon materials inspired the
Pt/C. The materials also remained stable during the catalytic same group to synthesize N-, O-, and S-tridoped nanoporous
reaction. DFT calculations were applied to investigate the N carbons for electrocatalysis of ORR.[77] The synergistic effects by
dopant species (e.g., pyrrolic, pyridinic and graphitic N) as well the multiple types of dopant species resulted in not only an effi-
as their positions in the nanosheets that are responsible for cient catalytic activity toward ORR but also a better selectivity
catalyzing the ORR. The results revealed that the carbon atoms to the desirable ORR product (i.e., H2O over H2O2). In another
next to graphitic N dopants and at armchair nanoribbon edge study, the Yao group synthesized hierarchically porous carbon
were the most effective active sites for the reaction. aerogels doped with N and S atoms via pyrolysis of carra-
It was found that combining N dopants with other geenan-urea hydrogel, as shown in Figure 10c.[78] The resulting
heteroatoms (e.g., O, S, B, P) in porous carbon-based materials carbon-based material contained an abundant amount of inter-
could form synergistic effects that result in very good electrocat- connected macropores and mesopores and had a large specific
alytic activities for ORR. The Asefa group synthesized oxygen- surface area of 1307 m2 g−1, which facilitated the mass trans-
and nitrogen-codoped mesoporous carbons via pyrolysis of pol- port and the exposure of more active sites in the material. As a
yaniline within mesoporous silica (SBA-15) template and then result, the material showed excellent electrocatalytic activity for
etching away the silica template.[76] The group then showed ORR under acidic condition, with a performance comparable
that these materials had excellent electrocatalytic activity for to that of a commercial Pt/C catalyst (Figure  10d). Theoretical
ORR. XPS survey spectra of the materials confirmed that the study showed that the material had N-S-C defect sites that could
materials had N and O, mainly in forms of pyrrolic/pyridone serve as active sites for the reaction. In particular, the N-S-D-G-6
and C–O/N–O species. The authors also made the following and N-S-D-G-4 sites in the material were found to be the most

Adv. Mater. 2020, 32, 2002435 2002435  (13 of 32) © 2020 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.advmat.de

Figure 11.  a) Schematic illustration of the synthesis of NGM. b) TEM image of NGM. c) Different kinds of N-doping or topological defect sites on
graphene nanoribbon, and d) their locations on the ORR volcano plot of overpotential versus ΔGOH*. G, PR, PN, QN, Q, C5, C7, and C5+7 represent
pristine graphene, pyrrolic N, pyridinic N, quaternary N, quaternary N on edge, five-C ring, seven-C ring, and five-C ring adjacent to seven-C ring,
respectively. a–d) Reproduced with permission.[29] Copyright 2016, Wiley-VCH. e) Schematic structural characters and f) TEM image of porous carbon
nanocages. g) Free-energy diagrams for ORR of different defect sites. e–g) Reproduced with permission.[80] Copyright 2015, American Chemical Society.

effective electrocatalysts and to have the lowest overpotentials sites in ORR electrocatalysis was also confirmed by Hu and
for ORR among the theoretically constructed and studied sites co-workers.[80] These researchers designed dopant-free carbon
(Figure  10e). This demonstrated the critical roles of pentagon nanocages with hollow structure that possess micropores and
S defect and graphitic-type N dopant sites play in the catalytic mesopores on the walls. The nanocages possessed three types
activity of such materials toward ORR. of defect sites: I) pentagon defects, II) edge defects, and III)
The pivotal contribution of intrinsic defects in metal-free hole defects (Figure  11e,f). To identify their respective contri-
carbon materials to the electrocatalytic activity of the mate- bution to electrocatalysis of ORR, the authors performed DFT
rials for ORR has been appreciated in other works as well.[79] calculations and controlled experiments. They found that the
Tang et  al. prepared a porous nitrogen-doped graphene mate- pentagonal and zigzag edge defect sites were responsible for
rial (NGM) with lots of microholes and edge sites that can the high catalytic activities or performances of these dopant-
catalyze ORR (Figure  11a,b).[29] They synthesized the material free carbon nanocages toward ORR (Figure 11g).
by pyrolyzing a ternary hybrid material containing sticky rice The above studies also highlight the fundamental progresses
(as a carbon source), melamine (as a nitrogen source) and being made toward our understanding of the sites respon-
Mg(OH)2 (as a template). The disk current density given by this sible for the catalytic activity of metal-free porous carbon elec-
material during ORR, as measured with a rotating ring-disk trocatalysts for ORR. It is believed that the combination of
electrode, was higher than that of Pt/C. Moreover, the authors heteroatoms, defect sites and porous structures can provide
­
found that the undoped porous graphene, prepared from a pre- metal-free ORR catalysts with an abundant amount of highly
cursor containing no melamine, exhibited a better activity for active sites. But the exact roles of each type of active site (e.g.,
ORR than the nitrogen-doped one. Consequently, they claimed intrinsic carbon defect sites, dopants, and pore edges) in these
that the high catalytic property exhibited by this material was metal-free catalysts still remain unknown.[81] Advanced charac-
mainly due to the edge-induced topological defects. Theoretical terization methods, combined with accurate synthesis and DFT
study indicated that adjacent pentagon and heptagon carbon calculations, are needed to obtain a clearer picture of the exact
ring defects (C5+7) were particularly the most active sites for contributions of the different types of sites to the overall cata-
ORR, with optimal adsorption ability to oxygen intermediates lytic activity of the materials. In addition, the potential contami-
(Figure  11c,d). The important roles played by carbon defect nation of such catalysts by metals stemming from reactants,

Adv. Mater. 2020, 32, 2002435 2002435  (14 of 32) © 2020 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.advmat.de

Figure 12.  a) Schematic depiction of CO2RR mechanism at catalytic active sites. b) The binding energies of CO* and H* intermediates on several metal
surfaces. b) Reproduced with permission.[82] Copyright 2017, Wiley-VCH. c) Kinetic volcano reflecting the dependence of catalytic activity on binding
energy of CO* intermediate. c) Reproduced with permission.[84] Copyright 2017, Nature Research.

precursors, or reaction vessels during their synthesis should active sites govern what the ultimate reaction product(s) will
be avoided to exclude the possible contributions of metallic be.[82] According to the binding energy of various intermedi-
impurities to the actual activity of the catalysts. ates and reduction products, metal-based electrocatalysts for
CO2RR can be roughly divided into the following four groups.
i) Metals such as Pb, Hg, Tl, In, Sn, Cd, and Bi, which selec-
5. Porous CO2RR Electrocatalysts tively produce formate or formic acid because COOH* inter-
mediate hardly binds to these metals. ii) Metals such as Au, Ag,
5.1. Reaction Mechanism at CO2RR Active Sites and Zn, which can bind COOH* intermediate tightly enough
but too weakly to CO* intermediate, produce CO as the main
Compared to HER and ORR, the CO2RR involves a much more product. iii) Metals such as Ni, Fe, Pt, and Ti, which have strong
complicated and multistep proton-coupled electron transfer adsorption to CO, inhibit further reduction reaction; these
process. The CO2RR can also involve various reaction pathways metals instead favor the competitive HER and produce H2 as
and produce different reduction products, such as CO, formate, a product. iv) Metallic Cu is the only one that can convert CO*
hydrocarbons, and alcohols. Figure 12a shows the mechanistic intermediate into CO2+ products (e.g., hydrocarbons and
pathways of CO2RR at catalytic active sites. As shown in the alcohols), because Cu can bind CO* intermediate well and has
scheme, first CO2 is adsorbed at the active site of a catalyst, a relatively weak binding energy to H* (Figure 12b).[82]
then reacts with protons and electrons to form key intermedi- The adsorption energies of the intermediates of CO2RR (e.g.,
ates, and eventually transforms into different products. Due CO*, COOH*, and COH*) forming on a metal surface have
to the thermodynamic stability of a CO2 molecule, the first step been found to follow scaling relations; however, the diversity of
in the reaction—the activation of CO2 by one-electron transfer products makes it difficult to obtain a fixed descriptor associated
to generate key intermediate (CO2−)—requires a considerable with the catalytic activity of the metal. Despite this fact, some
potential. Due to the high reactivity of the CO2− intermediate, classic theoretical models such as activity volcano schemes and
the subsequent steps can proceed through two, four, six, eight, d-band center theory, have been applied for CO2RR catalysts.[83]
and twelve electron transfer processes at the same time and Figure 12c demonstrates the kinetic volcano plot for CO reduc-
thus produce multiple reduction products. This is the basis of tion to CH4 over several transition metal catalysts.[84] The mul-
the major issue associated with the selectivity of products in tielectron reduction of CO2RR can usually be achieved via two
CO2RR. In addition, the CO2RR in H2O electrolyte often faces processes: CO2 reduction to CO, and then CO reduction to some
severe competition from the HER. final products. This also means CO* is the key intermediate
During CO2RR, H*, COOH*, CO*, and CH3O* are the four during deep reduction of CO2. Combining it with experimental
key intermediates that form, and their binding energies to the results, the activity volcano plot shows that Cu is situated close

Adv. Mater. 2020, 32, 2002435 2002435  (15 of 32) © 2020 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.advmat.de

to the top of the plot. This indicates that Cu is the best metal efficiency for HCOOH (78%) than the latter (45%). These results
catalyst with a moderate binding energy for CO*. The metals on reiterated how having nanoporosity with a large density of grain
the left side of Cu in the plot have stronger binding affinity to boundaries in catalysts can enhance their activity for CO2RR.
*CO and their surfaces thus get poisoned by CO, whereas the In recent studies, bismuth-based catalysts have attracted a lot
metals on the right side of the plot have weaker binding affinity of attention due to their selective production of formate.[89]
to *CO and thus form less coverage of *CO. For example, the Li group synthesized Bi2O3 nanotubes con-
Based on the research done so far, it is not difficult to con- taining defect sites (Figure  13a).[89a] The nanotube-derived Bi
clude that the adsorption strength of intermediates on catalyst catalyst exhibited a very good catalytic activity for CO2RR (with
surface directly affects the activity and selectivity of CO2RR. 60 mA cm−2 at −1.05 V vs RHE) and excellent selectivity (close
The adsorption strength of intermediates is closely related to to 100%) to a formate product (Figure 13b), outperforming some
the electronic structure of the catalyst. To control the electronic recently reported formate-producing electrocatalysts.[89b] DFT
structure of a catalyst and thereby improve its catalytic perfor- calculations revealed that the Bi defect sites on nanotube walls,
mance for CO2RR, especially the selectivity of the reaction prod- which can stabilize OCHO* intermediate (Figure  13c), render
ucts, various strategies such as defect engineering, heteroatom the nanotubes good and selective electrocatalysts for CO2RR.
doping and alloying, have been proposed.[85] In addition, theo- These results, once again, demonstrate that creating porous
retical calculations and advanced characterizations are pursued structures in metal-based CO2RR electrocatalysts is an effective
to help establish the relationship between the structure and the strategy to improve their catalytic performances. This is not sur-
property of the catalysts at the active-site level. The theoretical prising because the porous structure results in a high surface
and characterization results can be used to help optimize the area and a large density of accessible coordinatively unsatu-
performances of electrocatalysts as well as to discover high- rated sites (e.g., edge and step sites) and structural defects (e.g.,
performance electrocatalysts for CO2RR. vacancy, grain boundary) that can serve as active catalytic sites
for CO2 electroreduction in the catalysts.

5.2. Engineering Active Sites in Porous CO2RR Electrocatalysts


5.2.2. Constructing Single-Atom Active Sites
5.2.1. Optimizing Active Sites in Porous Metal-Based in Porous Electrocatalysts
Electrocatalysts
Many transition metal atoms (e.g., Cu, Fe, Co, Ni, Ru) anchored
Metal-based electrocatalysts are the largest class of materials on porous carbon or metal oxide matrices have been demon-
that have recently been studied for CO2RR. Silver is a represent- strated to have good catalytic performances for CO2RR.[90] For
ative example of metal-based electrocatalysts with the ability to instance, Yang et al. synthesized Cu/ZIF-8 embedded polymer
reduce CO2 into CO with good selectivity. In order to enhance nanofibers and thermally treated it produce porous carbon
its catalytic performance in the reaction, silver was made with nanofibers with supported single Cu atoms (Figure 13d).[91] The
nanostructures and also used with ionic liquid electrolyte materials had pore sizes of mainly ≈100  nm and a large spe-
(EMIM-BF4), instead of the commonly used aqueous electro- cific surface area of 618 m2 g−1. The isolated Cu atoms were
lytes.[86] However, the catalytic performances of nanosized silver coordinated with four N atoms on the carbon substrates in the
catalysts, such as their Faradaic efficiency for CO and current form of Cu–N4 groups, as confirmed by XPS and XAFS anal-
density of the reaction on them, were still unsatisfactory. An yses. These nanofibers electrocatalyzed CO2RR with a Faradaic
important breakthrough work addressing the limited catalytic efficiency of 44% to methanol as a product and a high partial
activity of silver was reported by the Jiao group.[87] In their current density of −93  mA cm−2 at −0.9  V (vs RHE). To get
work, the authors synthesized nanoporous silver catalyst by some insights as to why the materials showed high selectivity
making Ag–Al and then chemically dealloying it in acidic solu- toward methanol rather than CO, the authors performed DFT
tion. The resulting nanoporous silver catalyzed the reduction of calculations. The Cu–N4 sites were found to have slightly posi-
CO2, producing CO with a Faradaic efficiency of up to 92% and tive free energy for *CO desorption, inhibiting the evolution
with 3000 times larger high current density than polycrystal- of CO. By contrast, Ni–N4 with pyridinic N sites possessed a
line silver catalyst at an overpotential of 0.49 V. The higher elec- more negative free energy for *CO desorption (Figure 13e). In
trocatalytic activity for CO2RR exhibited by nanoporous silver another work, the Zheng group found that the introduction of
compared with polycrystalline silver was attributed to: i) its single Cu atoms substituting Ce atoms in CeO2 would result
150 times higher electrochemically active surface area and ii) its in a Cu site surrounded by three oxygen vacancies and one
more abundant step sites, which have 20 times higher intrinsic oxygen atom in the material (Figure  13f).[92] CO2 molecules
activity than the catalytic sites on polycrystalline silver. adsorbed on such Cu sites were found to have bent structures
Similar increased electrocatalytic activity toward CO2RR has (Figure  13g), indicating the strong ability of these Cu sites to
been achieved by introducing porosity into other metal catalysts activate CO2. Based on this theoretical result, the authors pre-
(e.g., Sn, Cu, and Ag).[88] For example, Kumar et al. reported a pared mesoporous Cu-doped CeO2 nanorods with a high sur-
porous Sn catalyst with a high density of grain boundaries that face area of 87 m2 g−1. XPS and EXAFS spectra confirmed that
can electrocatalyze CO2RR efficiently.[88a] The porous Sn catalyst the substituted Cu sites were single-atoms dispersed adjacent to
showed not only a much higher current density (10 mA cm−2 at multiple oxygen vacancies. The catalyst was shown to selectively
−1.0 V vs RHE) than Sn nanowires containing no grain bounda- reduce CO2 and produce CH4 with a high Faradaic efficiency of
ries (0.82 mA cm−2 at −1.0 V vs RHE), but also a higher Faradaic 58% (Figure 13h), which is the highest efficiency reported so far

Adv. Mater. 2020, 32, 2002435 2002435  (16 of 32) © 2020 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.advmat.de

Figure 13.  a) SEM image and structure of Bi2O3 nanotubes. b) Faradaic efficiencies of CO2RR by the nanotube-derived Bi catalyst at different overpo-
tentials. c) Free-energy profiles for *OCHO formation on defective and ideal surfaces. a–c) Reproduced with permission.[89a] Copyright 2019, Nature
Research. d) Synthesis procedure of Cu single atoms anchored carbon nanofibers. e) Free-energy diagrams for transformation of CO2 to CO on
Cu–N4, Ni–N4, and pyridinic N structures. d,e) Reproduced with permission.[91] Copyright 2019, American Chemical Society. f) Structural model and
g) Adsorption model of CO2 on single Cu site surrounded by three oxygen vacancies on CeO2 (110) surface. h) Faradaic efficiencies and current density
of Cu-doped CeO2 at different overpotentials. f–h) Reproduced with permission.[92] Copyright 2018, American Chemical Society.

for CH4 product from electrocatalytic CO2RR. In view of their the potential ­limiting steps of the reaction, which take place on
high utilization efficiency and accessibility, such electrocatalysts pyridinic N and graphitic N sites, had lower energy barrier than
based on single-atom active sites in porous materials should those on pyrrolic N and oxidized N sites; the result suggested
be fostered. They can also serve as ideal systems to understand that the former two were active sites for CO2 reduction to CO.
many fundamental properties of active sites, and thereby to fur- The Zhang group also found that porous N-doped carbons were
ther develop even better electrocatalysts for CO2RR. active catalysts for the electroreduction of CO2 into CO, with
a Faradaic efficiency of 94.5% at −0.6  V versus RHE.[94] Addi-
tionally, the authors found that there was a positive correlation
5.2.3. Creating Nonmetallic Active Sites in Porous between the density of defect sites in these nanocarbon mate-
Carbon Electrocatalysts rials and their catalytic activity for CO2RR, suggesting that the
intrinsic defect sites on the material might be the catalytic sites.
Application of metal-free heteroatom-doped carbon materials as This hypothesis was supported by DFT calculations, which
CO2RR electrocatalysts have recently become a hot topic. The showed that the sp2 defect sites (octagonal and pentagonal) at
incorporation of dopants such as N into carbon materials can the carbon atoms had a small free energy for the reduction pro-
induce charge transfer from C atoms to the more electronega- cess leading to COOH* intermediate (Figure 14a,b). Obviously,
tive N atoms, generating active sites for CO2RR. Thus, the intro- in the above two reports, the insights obtained regarding the
duction of heteroatom dopants into porous carbon s­tructures CO2RR active sites are very different. This is further compli-
may lead to metal-free carbon materials with prominent cata- cated by the results obtained by another study by Song et al. on
lytic activities for CO2RR.[93] For example, ­electrocatalytically mesoporous N-doped carbons (c-NC) with ordered cylindrical
active N-doped porous carbon possessing large pore sizes channel pores (Figure  14c).[95] Instead of producing CO, the
(≈10  nm) and a high amount of catalytically active N species c-NC materials were highly selective to ethanol product with a
was synthesized by the pyrolysis of MOF-74.[93a] The researchers high Faradaic efficiency of 77% during CO2RR electrocatalysis
found that optimization of the calcination temperature and at an applied potential of −0.56  V versus RHE (Figure  14d).
time could dictate the amount of catalytically active N species The pyridinic/pyrrolic N sites on cylindrical pore structures
making it into the carbon structure. The porous carbon with were proposed to be the catalytic active sites due to their high
the highest percentage of active N species (68.31%) concomi- electron density and the favorable dimerization of CO* inter-
tantly showed the highest Faradaic efficiency of 98.4% during mediates there. The authors also made an inverse mesoporous
CO2RR favoring a CO product. DFT calculations revealed that N-doped carbon (i-NC) containing nitrogen content similar to

Adv. Mater. 2020, 32, 2002435 2002435  (17 of 32) © 2020 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.advmat.de

and high pressure (>10 MPa). Consequently, it results in a huge


consumption of fossil fuels, thus causing heavy CO2 emis-
sions. Thus, over the last few years, electrochemical NRR under
ambient conditions has received a growing attention, due to its
high potential to realize eco-friendly and sustainable ammonia
production. However, because of the chemical inertness of a N2
molecule, due to its low polarizability and strong NN triple
bond, NRR typically has an extremely sluggish kinetics. It is,
therefore, essential to design efficient electrocatalysts that can
accelerate NRR.
Electrochemical NRR at catalytic active sites involves mul-
tiple nitrogen-containing intermediates and multistep proton–
electron transfer processes, based on two main mechanisms
currently proposed (Figure 15a).[96] In the first mechanism, also
called the dissociative pathway, NN triple bond first cleaves
to generate two adsorbed N atoms on the catalyst. Each of the
isolated N atom then undergoes consecutive hydrogenation to
produce NH3 molecule. In the second mechanism, also called
the associative pathway, the adsorbed N2 molecule first under-
goes hydrogenation process while keeping its NN triple bond
Figure 14.  a) Various carbon defect models and b) the reaction energy until the release of first NH3 molecule. An associative pathway
diagrams for CO2RR on them. a,b) Reproduced with permission.[94] can comprise two different hydrogenation sequences: alter-
Copyright 2019, Wiley-VCH. c) Illustration of cylindrical mesoporous nating pathway and distal pathway. While two N atoms are
N-doped carbon (c-NC) and inverse mesoporous N-doped carbon (i-NC) alternately hydrogenated in alternating pathway, the remote
for CO2 electroreduction. d) Faradaic efficiency of CO2RR products at
various potentials over c-NC and i-NC catalysts. c,d) Reproduced with
N atom is preferentially hydrogenated and released as NH3 in
permission.[95] Copyright 2017, Wiley-VCH. distal pathway.
Just like the oxygen-containing intermediates in ORR and
those in c-NC using mesoporous silica as template. The mate- the carbon-containing intermediates in CO2RR, the nitrogen-­
rial, i-NC, showed much lower Faradaic efficiency to ethanol containing intermediates (i.e., N2Hx and NHx species, x  =
than c-NC. This demonstrates the crucial role of the porous 0, 1, 2) in NRR follow linear scaling relations with respect to
structure of N-doped carbon catalysts play in terms of catalytic adsorption energies.[97] This scaling relation correlates the NRR
activity and selectivity during CO2RR. catalytic activity of metal surfaces with the binding energy
Although the exact roles played by the active sites in product of nitrogen intermediates. As depicted in the volcano plot
selectivity during CO2RR is still not fully understood, the above (Figure  15b), the catalytic activity (reflected by theoretical lim-
results clearly demonstrate that the introduction of heteroatom iting potential) for NRR of transition metals follows the Sabatier
dopants and porous structures into carbon materials endow the principle,[97] which suggests that the ideal catalyst should bind
materials with high catalytic performances for CO2RR. In terms nitrogen intermediates neither too strongly nor too weakly. For
of selectivity of products and product distributions of CO2RR, metals on the left side of the volcano plot (e.g., Re, W), the lim-
the heteroatom-doped porous carbon catalysts have been iting potential could be attributed to *NH2  → *NH3(g) step or
reported to lead to different outcomes.[94,95] This phenomenon *NH → *NH2 step for both dissociative and associative mecha-
implies that the electrocatalytic performances of these kinds of nisms. For metals on the right side of the plot (e.g., Pd, Pt), the
materials are highly sensitive to a wide range of parameters limiting potential could be attributed to N2(g) → *N2H step for
including the type, amount and location of dopants; the ways associative mechanism and N2(g)  → *2N step for dissociative
by which the materials are synthesized; and even the manner mechanism. The ability of nitrogen intermediate to bind to the
by which their electrodes are fabricated. However, there still surface can be used as a descriptor for NRR catalyst screening.
lacks a full understanding of the exact locations of the catalytic Despite these encouraging progress on theoretical studies
active sites on the surfaces of the heteroatom-doped porous and findings, the catalytic activities of known catalysts for NRR
carbon materials and how these sites mediate the CO2RR. are strictly limited by the following factors. i) The scaling rela-
tions indicates that the adsorption energies of key adsorbates
(e.g., *N2H and *NH2) cannot be optimized independently at
6. Porous NRR Electrocatalysts a single type of active site, demanding a minimum theoretical
overpotential for NRR of ≈0.4 V.[98] ii) The nitrogen adsorption
6.1. Reaction Mechanism at NRR Active Sites energy and N2 dissociation transition-state energy also exhibit
a linear scaling relation (Figure  15c).[97b,99] This suggests that
Ammonia is a key component in industrial nitrogen chemistry the catalyst with an optimal nitrogen adsorption energy (at the
to manufacture fertilizers and other chemicals, and is increas- top of “volcano” curve) still possesses a high activation barrier
ingly recognized as a green chemical energy carrier. Industrial for N2 dissociation. iii) HER is a major competing reaction
ammonia synthesis, which involves the well-established Haber– during electrochemical NRR generating NH3 in aqueous elec-
Bosch method, relies on high reaction temperature (>300  °C) trolyte. As shown in Figure  15b, although some metals (e.g.,

Adv. Mater. 2020, 32, 2002435 2002435  (18 of 32) © 2020 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.advmat.de

Figure 15.  a) Schematic depiction of NRR mechanism at catalytic active sites. b) NRR volcano plots of theoretical limiting potential for some transition
metals versus their nitrogen adsorption energy. b) Reproduced with permission.[97a] Copyright 2012, Royal Society of Chemistry. c) Scaling relationship
between the N2 transition-state energy for some transition metals versus their nitrogen adsorption energy. c) Reproduced with permission.[99] Copyright
2015, Elsevier B.V.

Rh and Ru) are considered to be most promising catalysts for active sites is highly needed to boost NRR catalysis. In addition,
NRR, their surfaces tend to be covered by adsorbed hydrogen strategies to break the restrictive effects of scaling relations may
instead of nitrogen. This severely hampers their activities for open a new avenue for achieving breakthrough results.
NRR, and they instead favor the competitive HER. Due to
these fundamental limitations, most reported catalysts exhibit
poor activity (with a current density of <0.1 mA cm−2, a yield of 6.2. Engineering Active Sites in Porous NRR Electrocatalysts
NH3 of <3 × 10−10 mol s−1 cm−2), and a low selectivity (a Fara-
daic efficiency of <10%) during NRR. 6.2.1. Optimizing Active Sites in Porous Metal-Based
The low rate of production of NH3 by NRR itself brings about Electrocatalysts
the issue of how to accurately determine the amount of NH3
coming from electrochemical N2 reduction. This is true espe- As potential catalysts for electrochemical NRR, noble metal
cially when labile nitrogen-containing species exist in the electro- materials such as Ir, Rh, Pd, and Ru have been mainly inves-
lyte or catalyst, which may also contribute to NH3. Recent retests tigated.[102] For Pd catalyst, the first protonation step of dini-
of some reported catalysts have demonstrated that the NH3 are trogen (N2(g) → *N2H) is the rate-determining step that results
originated from contaminants or decomposition of catalysts in a low activity and selectivity in NRR. To decrease the energy
rather than from NRR.[100] To avoid these false positive results barrier for the *N2H formation step, Xu et  al. synthesized
during NRR, establishing benchmarking protocols for control nanoporous palladium hydride (np-PdH0.43) by chemical deal-
experiments and ammonia measurements have been proposed loying of Pd–Al alloys and subsequent hydrogen injection and
by several groups.[101] In addition to the difficulty of accurately then investigated its potential as NRR catalyst (Figure 16a).[102a]
detecting NRR-generated NH3, the current most important The hybridization between Pd and H in np-PdH0.43 was
issues in NRR electrocatalysis involve acquiring desirable found to cause a higher d-band center (relative to the Fermi
activity and selectivity. Rational design of materials to optimize level) as compared to pristine Pd, strengthening the interac-
nitrogen adsorption as well as suppress hydrogen adsorption at tion between the active sites and reaction intermediates. DFT

Adv. Mater. 2020, 32, 2002435 2002435  (19 of 32) © 2020 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.advmat.de

Figure 16.  a) Schematic illustration of the synthesis of nanoporous palladium hydride. a) Reproduced with permission.[102a] Copyright 2020, Wiley-VCH.
b) TEM image of porous IrTe4 nanorods. c) Average rate of production of NH3 and Faradaic efficiency (FE) by NRR over MTe4 (M = Ru, Rh, Ir) and Te
at −0.2 V versus RHE. d) Adsorption energy of N2 and H2O over MTe4. e) Energy diagram for the pathway of HER over MTe4. b–e) Reproduced with
permission.[102b] Copyright 2020, Wiley-VCH. f) An illustration of SA-Mo/NPC and its atomic structure. f) Reproduced with permission.[104b] Copyright
2019, Wiley-VCH.

calculations confirmed that the Pd sites on np-PdH0.43 surface Unlike the above-mentioned noble metal-based catalysts,
possessed a much lower free-energy than those on pristine Pd semimetal bismuth (Bi) has an intrinsically poor binding ability
surface for the rate-limiting step involving *N2H formation. to H adatoms and is recently considered as a promising cata-
Its optimized active sites, combined with its porous structure, lyst for NRR. However, N2 adsorption and activation on pristine
made np-PdH0.43 catalyst to give a relatively high Faradaic effi- Bi surface is difficult due to its stable valence-electronic struc-
ciency (43.6%) and high rate yielding NH3 (20.4 mg h−1 mg−1cat) ture. Aiming at tackling this problem, Wang et al. synthesized
in NRR at −0.15  V versus RHE under ambient conditions, porous defect-rich Bi nanoplates through a low-temperature
which is almost threefold the value obtained for a nanoporous plasma bombardment method.[103] The defect-rich Bi sites
Pd. It is worth noting here that the lattice hydrogen atoms in presented a lower free energy to form adsorbed *N2H species
np-PdH0.43 would serve as the sources of hydrogen ending up from N2 as compared to original Bi sites, while maintaining a
in the formation of NH3 during the reaction, as revealed by poor binding affinity to H adatoms, as revealed by DFT calcu-
isotopic hydrogen labeling experiments. lation. As a result, the material exhibited a high NH3 produc-
A major challenge for most transition metal electrocatalysts tion rate (5.45 µg h−1 mg−1cat) and Faradaic efficiency (11.68%) at
during NRR is the competing HER in an aqueous environ- −0.6 V versus RHE. This demonstrates the significance in cre-
ment, in which a hydrogen atom can adsorb and occupy the ating defect sites in porous NRR catalysts to tune their surface
catalyst surface more easily than a nitrogen atom does. In order adsorption to the reaction intermediates involved, and thereby
to weaken the binding of hydrogen on the surface to suppress improve their performances in NRR.
HER and to catalyze NRR better, the Huang group reported sev-
eral porous MTe4 (M = Ru, Rh, Ir) nanorods with such an ability
(Figure 16b).[102b] The porous IrTe4 nanorods catalyzed the reac- 6.2.2. Constructing Single-Atom Active Sites in Porous
tion with a much higher rate giving NH3 (51.1  µg h−1 mg−1cat) Electrocatalysts
and Faradaic efficiency (15.3%) at −0.2 V versus RHE than the
RhTe4 and RuTe4 nanorods (Figure  16c). DFT calculations and Isolated metal atoms (e.g., Ru, Mo, Au) stabilized on porous
experimental characterizations indicated that IrTe4 had the support materials, with maximum atom utilization efficiency,
lowest adsorption energies for N2 and H2O favoring NRR pro- have been studied recently for efficient electrochemical
cesses, but the highest energy barrier for HER pathway, among NRR.[104] For instance, Tao et  al. fixed Ru ions in zirconium-
the three materials (Figure 16d,e). Further calculation attributed based MOF (UiO-66) and subsequently annealed it to obtain
the optimal adsorption strength to the synergistic interaction single Ru sites in N-doped porous carbon (Ru@ZrO2/NC)
between electroactive Te atoms and electron-rich Ir atoms in and then Ru/NC by removal of ZrO2 from Ru@ZrO2/NC.[104a]
IrTe4. Ru/NC catalyzed NRR with a large NH3 production rate of

Adv. Mater. 2020, 32, 2002435 2002435  (20 of 32) © 2020 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.advmat.de

3.7  mg h−1 mg−1cat at −0.21  V versus RHE. Because of ZrO2, (O, S, Se, and Te).[107] DFT calculations revealed two significant
Ru@ZrO2/NC could suppress HER and give a higher Faradaic effects induced by the heteroatom-doping: charge accumulation
efficiency (21%) than Ru/NC (14%). To confirm the reaction and spin polarization. While the former was found to promote
sites for NRR, DFT calculations on various adsorption sites N2 chemical adsorption on carbon sites, the latter decreased
were conducted. The results suggested that oxygen vacancies the energy barrier of the first protonation step of N2 leading to
in ZrO2 surfaces could anchor the Ru single atoms that serve *N2H (the rate-determining step) in NRR. The authors further
as the main active sites for the NRR process. predicted that among these heteroatom-doped carbon mate-
In another work, Han et  al. synthesized single Mo atoms rials, Se- and Te-doped ones could be more promising NRR cat-
on N-doped carbon with 3D hierarchically porous structure alysts due to their enriched charge and large spin moments at
(SA-Mo/NPC) by decomposing hydroxylamine hydrochlo- their carbon sites. They also prepared these heteroatom-doped
ride precursor containing Mo, and the single Mo atoms real- porous carbon materials by pyrolysis of MOF (ZIF-8) and sub-
ized the formation a high density of atomic active centers sequent dopant vapor diffusion. As expected, the Se- and Te-
for NRR (Figure  16f).[104b] The fact that only MoC or MoN doped porous carbon materials showed much higher intrinsic
bonds observed by XAS measurements confirmed the pres- catalytic activity for NRR than their O- and S-doped counter-
ence of atomically dispersed Mo sites on a carbon support parts. These findings might also be applicable to the efficient
material. Notably also, theoretical studies had previously pre- screening of porous carbon electrocatalysts for NRR.
dicted that N-bonded single Mo atoms could have high catalytic
performance for NRR, but the result was not experimentally
proven till this work. In the work, a high NH3 production rate 7. Porous OER Electrocatalysts
(34 mg h−1 mg−1cat) and a large Faradaic efficiency (14.6%) was
achieved by the dispersed nonprecious metallic active sites (Mo) 7.1. Reaction Mechanism at OER Active Sites
on a porous support material. When compared with the experi-
mental and theoretical investigation carried out for single- The OER is an important anodic half-reaction in many electro-
atom active sites for electrocatalytic HER, ORR, and CO2RR, chemical processes including water splitting and CO2 reduction.
those for NRR are still in their early stages and require more The reaction is sluggish as it involves four-electron/four-proton
emphasis in the near future. coupled processes and oxygen–oxygen bond formation. Thus, it
requires efficient catalysts that can improve the kinetics of the
processes involved in it. In alkaline media, the most common and
6.2.3. Creating Nonmetallic Active Sites in Porous stable OER catalysts are transition metal (hydro)oxides. Although
Carbon Electrocatalysts a growing number of transition metals and their sulfides, phos-
phides, borides, and nitrides are reported to efficiently catalyze
The important findings that metal-free heteroatom-doped OER, they are irreversibly oxidized and form amorphous oxide
carbon materials catalyze ORR and CO2RR have encouraged layers under high working potentials (Figure  17a).[108] Recently,
researchers to investigate the ability of these materials to elec- some nonprecious metal (Fe, Co, Ni)-based catalysts including
trocatalyze NRR, especially over the past two years. In the pio- (hydro)oxides and nonoxide-derived materials, which exhibit
neering work of Liu et al., N-doped porous carbon was found to even better catalytic activities for OER than the benchmark elec-
be effective electrocatalyst for NRR, producing NH3 at a higher trocatalyst IrO2 in alkaline solution, have been developed.[109]
rate than those of metal-based electrocatalysts.[105] The pyridinic However, compared to the Ni-based electrodes, which are widely
and pyrrolic N groups were theoretically and experimentally used in commercial alkaline water electrolysis technologies,
found to be key active sites for the material’s catalytic activity the performances of these OER catalysts are still unsatisfactory.
here. A great advantage of many metal-free NRR electrocata- Of note, their corrosion resistance and long-term stability over
lysts is the weak adsorption affinity of their nonmetallic sites to several thousands of hours, especially at large current densities
H*, making HER unfavorable. Considering Lewis acidity of H+ (e.g., 200–300  mA cm−2 in commercial alkaline electrolyer), are
ion (or a proton), Liu et al. introduced F atoms into 3D porous limited. Moreover, many of the promising nonprecious OER
carbon framework to generate Lewis acidic sites in the mate- catalysts that work well in alkaline media are unstable in acidic
rial, via electron modulation, that can repel a proton and limit media. Note that proton exchange membrane (PEM) electro-
its binding on the material.[106] NH3 temperature-programmed lyzers that operate under acidic conditions are increasingly
desorption (NH3-TPD) proved the presence of a much higher receiving attention due to their higher energy efficiency than
density of Lewis acidic sites in the F-doped carbon than in the alkaline electrolyzers. At present, only iridium-based oxides (e.g.,
pristine carbon. DFT calculations revealed that F-containing IrO2) have been found to show reasonable activity and stability in
graphene model possessed a higher energy barrier for H2 evo- acidic media. In view of the scarcity and high cost of iridium, the
lution as well as a lower energy barrier for N2 adsorption and catalytic properties of some iridium-containing complex oxides
dissociation than the F-free counterpart. Benefiting from these (e.g., pyrochlores, perovskites) as alternatives to IrO2 for OER
well-tuned surface active sites as well as it abundant 3D porous have recently been investigated.[110] These materials can be inter-
structure, the F-doped carbon showed much higher Faraday effi- esting alternatives due to their intrinsic advantages including
ciency (54.8%) than pristine carbon framework for NH3 product. lower iridium content compared with pure IrO2. These iridium-
To understand the origin of catalytic activity in carbon-based based compounds almost always tend to undergo surface recon-
electrocatalysts for NRR, Yang et al. systematically investigated a struction during OER and form amorphous surface IrOx layers,
class of carbon materials doped with various nonmetallic atoms which ultimately serve as the real catalytic phase. Regardless of

Adv. Mater. 2020, 32, 2002435 2002435  (21 of 32) © 2020 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.advmat.de

Figure 17.  a) Surface reconstruction of nonoxide catalysts under OER condition in alkaline solutions. b) Mechanism of OER at catalytic active sites
in acidic (green route) and alkaline (blue route) conditions. c) Free-energy diagram on ideal OER catalyst at U = 0 and U = 1.23 V. d) The relationship
between adsorption energies (ΔE) of HOO* versus those of HO* for various metal oxides. e) The volcano plot for OER between negative values of
theoretical overpotential and ΔGO* − ΔGHO* for several metal oxides. d,e) Reproduced with permission.[112a] Copyright 2011, Wiley-VCH. f) The relation
between the catalytic activity for OER and the metal eg-orbital occupancy of various perovskite oxides. f) Reproduced with permission.[113] Copyright
2011, American Association for the Advancement of Science.

the environments in which OER electrocatalysts work, careful The apparent reaction rate of OER at a catalytic site is deter-
OER mechanistic studies on the surfaces of the catalysts are mined by the slowest step involved in the whole process, which
needed in order to establish design principles of OER catalysts is generally known as the RDS. The possible RDS can be evalu-
with higher activity and stability. ated according to the Tafel slope obtained from the polarization
Figure  17b shows the well-accepted reaction mechanism curve. A general trend is that the last part of the reaction causes
of OER, which includes the formation and transformation of a larger transfer coefficient, further leading to a smaller Tafel
OH, O, and OOH intermediates at a catalytic active site (M). slope and faster kinetics.
The reactions under acidic and alkaline conditions are slightly Accurate computation of OER catalysis on a metal oxide sur-
different from each other due to the different reactive species face is challenging due to the complexity of the the reconstruction
(H2O in acidic media and OH− in alkaline media) they involve. that the metal oxide surface undergoes during the reaction. Based
The OER mechanisms under acidic and alkaline conditions are on a simple thermochemical framework, theoretical method
described with the following equations, respectively: has been developed to reveal the origin of OER overpotential
by evaluating reaction free energies of the individual reaction
a) Under acidic conditions
steps.[111] The specific step with the largest free energy is known
M + H2 O → M − OH + H+ + e − (9) as the potential-determining step (PDS), and the corresponding
theoretical overpotential can be obtained by using the equation
M − OH → M − O + H+ + e − (10)
G OER
η OER = − 1.23 V (17)
M − O + H2 O → M − OOH + H+ + e − (11) e

where ηOER is the theoretical overpotential, GOER is the Gibbs


M − OOH → M + O2 + H+ + e − (12) free adsorption energy, and e is the charge of an electron.
It is worth noting that RDS and PDS are not always the
b) Under alkaline conditions
same. This is because RDS depends on the activation energy
M + OH− → M − OH + e − (13) (i.e., energy of a transition state) of elementary reaction (known
as kinetic bottleneck) whereas PDS depends on the thermo-
M − OH + OH− → M − O + H2 O + e − (14) dynamic reaction energy (known as thermodynamic bot-
tleneck).[111] Compared to RDS, PDS provides a more direct
approach to help catalyst design. Meanwhile, their differences
M − O + OH− → M − OOH + e − (15) can be reflected by the differences in the values of practical
overpotential and theoretical overpotential of the reaction over
M − OOH + OH− → M + O2 + H2 O + e − (16) the catalyst.

Adv. Mater. 2020, 32, 2002435 2002435  (22 of 32) © 2020 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.advmat.de

Based on the above four electron/four proton-coupled mech- the more unstable its surface structure is during OER (e.g., it
anism of OER, one can conclude that i) the ideal OER catalyst undergoes cation dissolution). The Markovic group studied a
should be able to facilitate these four elementary steps with series of monometallic oxides with different crystal structures
the same reaction free energies, just above the equilibrium and degree of crystallinity for OER under acidic condition,
potential, and ii) the value of free energy should be 1.23 eV at and found that their activity was related to the oxophilicity of
zero potential (Figure  17c). However, this ideal model cannot the metals and their surface defects.[115] In all cases the most
be achieved under the current theoretical framework because active oxides were found to be the least stable. Grimaud et  al.
an inevitable scaling relation exists among these three oxygen- reported similar results for Pr0.5Ba0.5CoO3−δ under alkaline
containing intermediates. As reported by two different groups, solution, which exhibited increasingly higher catalytic activity
there is a constant of ≈3.2 eV between the adsorption energies as its covalency was increased compared with LaCoO3.[116]
of M–OOH and M–OH, stemming from the similar bonding However, increasing the covalency of Ba0.5Sr0.5Co0.8Fe0.2O3−δ
type of O species in these intermediates (Figure  17d).[112] This not only resulted in structural amorphization but also ren-
demonstrates that even the most optimized material can give dered the material unstable during catalysis. Taken together,
an overpotential of at least 0.3–0.4  V due to this universal these studies demonstrated that the surface structure should
scaling relation. be carefully addressed to balance the catalytic activity and sta-
This scaling relation also provides an opportunity to cor- bility of OER catalysts. To identify the origin of this relation-
relate the proposed catalytic performance with more simpli- ship, several important research works have been carried out.
fied descriptors such as ΔG(O*) −  ΔG(OH*). As shown in The results revealed that a potential-induced valence-transition
Figure 17e, a recognizable volcano shape can be seen from this were involved on the surface atoms during catalytic reaction,
plot, with highly active catalysts (e.g., RuO2, IrO2) sitting near and the as-formed higher-valence species were the real active
the top of the volcano. This volcano plot also indicates that the sites, albeit their unstable features.[117] As the different surfaces
performance of catalysts follows the Sabatier principle: while on catalysts have different abilities of valence-transition, they
the catalysts on the left side of the volcano plot bind oxygen too govern the trends in catalytic activity and stability. Therefore, the
strongly and the reaction over them is limited by the formation most important issue in OER electrocatalysis entails breaking
of M–OOH species, those on the right side of the volcano plot this recessive “scaling relation” between activity and stability.
bind oxygen too weakly and the oxidation of M–OH species This may require appropriate structural optimization to facili-
in them is the PDS. The adsorption energies of such oxygen-­ tate the formation of higher-valence species while keeping them
containing intermediates are highly correlated with the elec- with stable structures and catalytically active phases.
tronic structure of the catalyst (such as the O p-band center and
occupation of eg orbital).[67] Figure  17f presents that a volcano-
like plot for catalytic activity for OER versus eg-occupancy for 7.2. Engineering Active Sites in Porous OER Electrocatalysts
many perovskite oxides.[113] As this correlation is based on the
hybridization between the π* orbital of adsorbed oxygen and 7.2.1. Porous NiFe-Based Electrocatalysts for Alkaline OER
the empty eg orbital of the catalyst, the level of eg occupancy
can reflect the binding strength of oxygen intermediates to the NiFe-based bimetallic oxides/hydroxides are widely regarded as
catalyst surface. Although these descriptors can guide us to find among the most active OER electrocatalysts in alkaline electro-
good catalysts for OER on the volcano plot, they cannot break lytes. Reports on the synergistic effects exhibited by Ni and Fe
the limitation imposed by the scaling relation. More delicate bimetallic species to catalyze OER can be traced back to 1987.[118]
strategies to stabilize one intermediate with respect to another, Corrigan observed that the overpotential of OER over a nickel
by tuning the electronic structure or/and the geometric struc- oxide thin film was dramatically lowered as iron impurities
ture of catalytic surface, should be developed.[69] Mechanisms made it into the electrode from the electrolytes or via copre-
that lead to the formation of direct oxygen-oxygen coupling cipitation processes. In fact, as little as 0.01% Fe impurity in
on the surfaces of catalysts eliminate the initial hurdles of the the electrode was found to bring about a substantial enhance-
reaction. To make a significant breakthrough, advancements in ment in the material’s catalytic activity toward OER. Thus, to
our current knowledge on the active sites of OER by unraveling uncover the active phases/species responsible for such catalytic
the relationships between the structure of a precatalyst and the enhancement, several groups have focused their attentions on
reconstructed oxide surface, and by identifying key structural NiFe catalyst systems.
motifs on the resulted surface, should be made. In 2012, Boettcher and co-workers synthesized composi-
Besides the scaling relation among different intermedi- tion-controllable NiFeOx thin film and observed that the film
ates, the relationship between experimental activity and sta- underwent an irreversible phase transformation during OER
bility for OER catalysts is another crucial issue that is hard to ­electrocatalysis.[119] Based on this observation, the authors argued
­overcome.[114] This issue arises from the relatively harsh oxi- that the as-formed Ni hydroxide/oxyhydroxide phase was rather
dative condition (especially in acidic media) that the catalysts the actual active catalyst on the electrode. This was confirmed
face during OER. (Note that the catalysts do not face the same by the Bell group.[120] The authors provided a clearer structure
issue in the case of the other four reactions mentioned earlier, of the actual catalytic sites. They found that γ-NiOOH would
i.e., HER, ORR CO2RR, and NRR, because these four entail form over a Ni-containing electrode at potentials approaching
reduction conditions.) There is a conundrum about the rela- the ones resulting in oxygen evolution (Figure  18a). They also
tionship between the activity and the stability of OER catalysts: indicated that γ-NiOOH could be transformed further into
generally, the more catalytic active the material is for OER, β-NiOOH during additional aging process. In their study, Ni

Adv. Mater. 2020, 32, 2002435 2002435  (23 of 32) © 2020 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.advmat.de

Figure 18.  a) TOF of Ni sites of Ni–Fe catalysts during OER and in situ Raman spectra of Ni–Fe catalysts. a) Reproduced with permission.[120] Copyright
2013, American Chemical Society. b) X-ray adsorption spectra at Fe K-edges and Ni K-edges of Ni-Fe catalysts under OER catalysis. b) Reproduced with
permission.[122] Copyright 2016, American Chemical Society. c) Cyclic voltammetry curves of Fe-free and Fe-doped Ni(OH)2/NiOOH catalysts during
OER. c) Reproduced with permission.[123] Copyright 2014, American Chemical Society. d) Computed overpotentials for OER at Fe–Fe bridge sites in
Ni-doped γ-FeOOH. d) Reproduced with permission.[126] Copyright 2015, American Chemical Society.

species were suggested as active sites in NiFe hydr(oxy)oxides, study, the same authors claimed that the Fe species intercalated
because the authors observed that aged Ni films could achieve a at the edges or defects of Ni oxyhydroxide would contribute the
similar catalytic activity as NiFe films. This observation was also most to the apparent catalytic activity of these materials.[125]
in accordance with the one reported by the Markovic group, in Additionally, the Bell group re-examined NiFe oxyhydroxides
which they revealed that 3d M (M = Ni, Co, Fe, Mn) hydr(oxy) using operando XAS, and their results demonstrated an
oxides had catalytic activity in the order of Ni > Co > Fe > increased valence state in both Ni and Fe during OER.[126] Fur-
Mn.[121] In a separate work, Görlin et  al. used operando differ- ther DFT calculations of Fe–Ni oxyhydroxides also pointed to Fe
ential electrochemical mass spectrometry (DEMS) and quasi-in sites (η = 0.43 V) as having a higher intrinsic activity for OER
situ X-ray adsorption spectroscopy to explore the dynamics of than Ni sites (η = 0.59 V) (Figure 18d). Furthermore, Chen et al.
OER over NiFe electrocatalysts.[122] Their results indicated the detected the presence of Fe4+ species on Ni–Fe oxide catalysts
generation of Ni4+ species during the reaction, while Fe species under catalytic condition with operando Mössbauer spectros-
consistently remaining in the +3 state (Figure 18b). In the report copy,[127] while Goldsmith et al. proposed that a NiOOH lattice
by Nocera and co-workers, Ni species were also reported as the enabled a facile oxidation of Fe3+ to Fe4+ based on theoretical
active sites for OER.[117b] The authors argued that the incorpora- and spectroelectrochemical studies.[128]
tion of Fe3+ ions promoted the formation of Ni4+ species in the The debate as to what exactly is/are the active sites in NiFe-
catalysts during OER, explaining further why FeNi oxide films based OER electrocatalysts is still ongoing at the moment.
show good catalytic activity for OER. Another important issue about NiFe-based electrocatalysts is
While some experimental observations were provided to the stability of their “elusive” active sites. This stability issue
support Ni species being the active sites, others continued to has also implications on the prospects of the potential prac-
point rather to Fe species being the active sites. For example, tical application of highly active NiFe-based electrocatalysts. A
the Boettcher group found that the increase in catalytic activity study led by the Liu group found that the layered structures
observed in aged Ni films was originated from the Fe impuri- in NiFe-based electrocatalysts could limit the diffusion of
ties in KOH electrolyte (Figure  18c).[123] They showed this by proton acceptors (e.g., OH−) into the small interlayer spaces
removing the trace Fe in the electrolyte through absorption in the materials during OER.[129] Furthermore, they indicated
using a Ni(OH)2 precipitate and by obtaining pure NiOOH. that the local acidic environment would cause cation dissolu-
The pure NiOOH exhibited only a poor catalytic activity toward tion, making the electrocatalyst unstable. This work provided
OER. Its catalytic activity became even lower upon prolonged an important explanation as to why some previously reported,
anodic polarization. Under rigorous Fe-free electrolytes, the highly active NiFe-based OER electrocatalysts usually had very
researchers found that the catalytic activities of the different limited catalytic lifetimes (<100 h). This work, together with
species for OER were in the order of Ni(Fe)OxHy  > Co(Fe) some other relevant studies,[130] also suggested that simulta-
OxHy  > FeOxHy  > CoOxHy  > NiOxHy  > MnOxHy.[124] These neous improvement of the activity as well as the stability of
results highlighted the important role that Fe would play in Ni– NiFe-based OER electrocatalysts is necessity for such catalysts
Fe catalysts during OER. At the same time, it raised another to find practical use.
reasonable assumption that Fe species might be the ones A representative work recently carried out by Liu et al. demon­
serving as active sites even in NiFe hydr(oxy)oxides. In another strated a possible solution to the activity–stability issue in

Adv. Mater. 2020, 32, 2002435 2002435  (24 of 32) © 2020 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.advmat.de

Figure 19.  a) SEM and b) TEM images of the NiFe LDH material. c,d) Chronopotentiometric curves in 1 and 10 m KOH at a current density of 1000 mA
cm−2. a–d) Reproduced with permission.[131] Copyright 2018, Nature Research. e) Schematic illustration of FeS as a precatalyst for producing catalytically
active species (named Fe–O2cat) for OER. f) Catalytic properties for OER of Fe–O2cat. e,f) Reproduced with permission.[132] Copyright 2018, Elsevier B.V.

NiFe-based electrocatalysts.[131] The authors prepared a porous generated on nonoxide surfaces are in metastable and amor-
thin film of NiFe layered double hydroxides (LDH), which was phous state, which are hard to access directly. iii) The under-
assembled using ultrathin nanosheets consisting of ultrasmall, lying nonoxide materials can serve as conductive media
crystalline nanodomains (Figure  19a,b). The combination their while the less conductive oxides/hydroxides serving as the
porous structure and an abundant amount grain boundaries catalytically active phases.
between their nanodomains endowed the material with highly Some nonmetallic elements (e.g., P and B), and their asso-
exposed catalytic active sites and sufficient mass transfer capa- ciated anions, in the corresponding nonoxide-derived electro-
bility for electrolytes. As a result, this porous electrocatalyst exhib- catalysts are found to participate in the construction of cata-
ited extraordinary catalytic stability for over 6000 h (>8 months) lytically active phases. For example, the Lou group reported a
at a current density of 1000 mA cm−2 (Figure 19c,d). This was the porous iron cobalt (oxy)phosphide catalysts (Fe–Co–P), which
first time that a NiFe-based catalytic material presented both high show excellent OER performance with a small overpotential of
efficiency and stability during OER for several months. 269 mV at 10 mA cm−2 in alkaline solution.[134] During the OER,
the Fe–Co–P was found to undergo structural evolution, where
some of P atoms were substituted by O atoms producing Fe–
7.2.2. Porous Nonoxide-Derived Electrocatalysts for Alkaline OER Co–P–O species. First-principle calculations attributed the high
catalytic activity to electron transfer between Fe and Co atoms
In addition to metallic oxides, transition metal-based (Fe, Co, via the O/P bridges. In another important study, the Zou group
Ni) chalcogenides, nitrides, phosphides, and borides have also showed that metal borides could catalyze OER and that surface
been extensively explored for electrocatalysis of OER. How- oxidized layers containing stable boron species would form on
ever, they are more unstable under OER, and easily transform the material during the reaction.[109a] The authors further iden-
to their corresponding oxides/hydroxides, making the latter tified that metaborate-containing oxyhydroxide films on boride
the real catalytically active species. This transformation can surfaces were the catalytic active phases. The electron redistri-
entail either their surfaces or the entire bulk structures. For bution induced by metaborate species on the material led to
example, Zou et al. identified that FeS nanosheets were totally optimized surface adsorption of oxygen intermediates and high
converted into porous amorphous FeOx film when being used catalytic activity for the reaction. These studies demonstrate
as catalysts in OER (Figure  19e,f).[132] Another report showed that the roles that anions associated with nonmetallic elements
that NiSe nanorods underwent a deselenization process during in the derived catalytic phase are a potential reason for reducing
OER, and then surface oxidation forming NiSe/NiOx core/shell the OER reaction barrier of nonoxide electrocatalysts.
structure.[133] Nonetheless, these nonoxide-derived electrocata-
lysts continue to attract a great interest, mainly because they
often present superior apparent OER activities than the corre- 7.2.3. Porous Iridium-Based Electrocatalysts for Acidic OER
sponding oxides/hydroxides that are directly prepared. Their
improved activity may be due to several scenarios.[108a] i) The Although various types of nonprecious electrocatalysts,
surface reconstruction during OER induces surface roughening including the above-mentioned NiFe-based materials, present
and high surface area. ii) The highly active oxides/hydroxides effective and stable OER catalytic activity in alkaline media,

Adv. Mater. 2020, 32, 2002435 2002435  (25 of 32) © 2020 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.advmat.de

they are unstable in acidic media. At present, only iridium- area, many of these materials exhibited higher catalytic activity
based oxides, typically IrO2, have been found to show reason- for OER than IrO2 nanocatalyst. Based on the studies conducted
ably good activity as well as reasonably good stability in acidic on such materials so far, two acceptable conclusions have been
solution. This is also why they are currently employed in many made about their surface active phases: i) some less stable con-
commercial PEM electrolyzers. However, more efficient elec- stituents (e.g., Ni and Co) in the Ir-based alloys leach into the
trocatalysts with enhanced Ir mass activity are required for the electrolyte during electrocatalysis, producing Ir-rich surfaces,
next-generation PEM water electrolysis technologies. and ii) Ir-rich surfaces should be present in the oxidized form
To this end, Ortel et  al. synthesized mesoporous IrO2 rather than in metallic state during OER. The discussion and
via soft templating method using an amphiphilic triblock theoretical calculations presented for multimetallic alloy surface
poly(ethylene) oxide (PEO)-poly(butadiene) (PB) copolymer, without considering surface reconstruction cannot account for
PEO–PB–PEO, for efficient electrocatalysis of OER.[135] The the activity improvement exhibited by these Ir-based nanoalloy
material had crystalline walls comprising 4 nm crystals, ordered catalysts. Recently, using IrNi alloy-derived core–shell IrNi@IrOx
mesopores with 16  nm in diameter, and a surface area of nanocatalyst as a model system, Nong et  al. investigated the
≈100 m2 g−1. Thanks to its mesoporous structure, the material local coordination environment and electronic states of Ir atoms
exhibited a 200% higher catalytic activity for OER compared under OER conditions with X-ray absorption spectro­scopy, dif-
with nonporous IrO2 nanocrystals, demonstrating the structural ferential atomic pair correlation analysis, and resonant high-
advantages of having a porous structure in such catalysts. This energy X-ray diffraction.[138] The authors found that the leaching
was further corroborated by the study conducted by Oakton et al. of Ni atoms during OER produced lattice vacancies and short-
for a porous IrO2–TiO2 composite material, which possessed a ened IrO bonds (Figure 20a,b), thus leading to a considerable
high surface area of 245 m2 g−1 and ultrasmall IrO2 nanoclus- amount of d-band holes in the IrOx shells. Based on this, the
ters (1–2 nm) dispersed on the surfaces of TiO2 nanoparticles.[136] authors considered that the oxygen ligands, which have strong
Although the composite material contained only 40 mol% of Ir, electrophilicity, were the catalytic active sites. These sites were
it exhibited a better activity and stability during OER in acidic expected to be involved in the nucleophilic acid–base-type
media than the benchmark commercial OER electrocatalyst IrO2. oxygen evolution reaction with a reduced kinetic barrier. This
Their study also revealed that the iridium hydroxo surface spe- work provided some important insights into the origins of the
cies transformed into the oxo-terminated surface groups at the catalytic activity of Ir-based nanoalloy OER catalysts. However,
OER regime, and the latter species were considered to be respon- there are still some unresolved questions about such OER cata-
sible for the material’s good catalytic activity for the reaction. lysts: e.g., Are the oxygen active sites the only ones operating
Recently, several groups have made attempts to synthesize during OER? Can there be surface Ir atoms coserving as active
porous Ir-based nanoalloys for OER electrocatalysis in acidic sites? Can two different active sites contribute to the overall
media. The materials included IrCo, IrNi, IrCu, IrRu, and IrOs catalytic activity? How the presence of such active sites balance
alloys.[137] Because of their porous structure and high surface the activity and stability of the catalysts?

Figure 20.  a) Schematic of catalytic active sites in IrNiOx core–shell nanoparticles. Oxygen atoms, iridium atoms, and nickel atoms are represented
in red, gray, and green, respectively. b) IrO bond distances in IrOx and IrNiOx nanoparticles at different applied potentials. a,b) Reproduced with
permission.[138] Copyright 2018, Nature Research. c) Schematic illustration of the synthesis of porous hollow RuIrOx nanonetcages. d) OER polarization
curves for RuIrOx, IrO2, and RuO2. e) The free-energy profiles of OER on RuO2 (110) surface, as well as Ir and Ru active sites on RuIrOx (110) surfaces
at U = 1.23 eV. c–e) Reproduced with permission.[139] Copyright 2019, Nature Research.

Adv. Mater. 2020, 32, 2002435 2002435  (26 of 32) © 2020 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.advmat.de

7.2.4. Porous Ruthenium-Based Electrocatalysts for Acidic OER electrocatalysts, providing atomic-level insights into the design
of next-generation electrocatalysts. We have also highlighted the
Comapred with Ir-based oxides, Ru-based oxides generally notable recent advancements made in active site engineering
have higher activity, but lower stability due to their tendency to in porous electrocatalysts to improve or maximize their perfor-
over-oxidize into water-soluble RuO4 species during electroca- mances in various electrochemical reactions including HER,
talysis of OER. To improve their stability as well as their activity, ORR, CO2RR, NRR, and OER. Some of the issues and chal-
making them in the form of bimetallic Ru–Ir oxide has been lenges related to porous electrocatalysts have been discussed
reported to be a good strategy. For example, the Li group pre- and also summarized and emphasized below.
pared porous hollow RuIrOx nanonetcages (with Ru and Ir in
a molar ratio of 2:1) by using a dispersing–etching–holling syn- i) Innovation of synthetic methods. Innovative and “green”
thetic strategy (Figure 20c).[139] The resulting material, RuIrOx, synthetic approaches enabling the development of structurally
exhibited not only remarkably improved stability relative to precise porous electrocatalysts should be promoted for
Ir-free RuOx catalyst, but also significantly higher activity, in several reasons. 1) Efficient electrocatalytic reaction takes place
OER (η10 = 233 mV) than commercial IrO2 (η10 = 433 mV) and at some specific sites with appropriate adsorption properties
RuO2 (η10  = 392  mV) in Figure  20d. The authors’ DFT results (i.e., active sites), such as the edges in MoS2 during HER.
in Figure 20e revealed that Ir and Ru sites in RuIrOx had syn- So synthetic pathways that create and control porous struc-
ergistic effect enabling the decrease in the energy barrier of the tures, and expose more catalytically active sites, are of practi-
potential limiting step for OER (O* → HOO*). Furthermore, cal significance. The catalytic behaviors of porous electro-
their porous 3D nanonetcage structure increased the exposure catalysts are also strongly dependent on the structures, sizes,
and utilization of Ru and Ir sites in the material. and morphology of the pores. Some well-designed porous
To reduce the amount of noble metal in efficient OER cat- structures, including hierarchical porous structure, ordered
alysts, the introduction of nonprecious metals into RuO2 has mesoporous structure, and ultrathin porous nanosheets,
been widely considered. Several groups recently prepared offer advantages of increased accessibility of active sites;
porous bimetallic oxides such as Mn–RuO2, Cu–RuO2, and they should thus receive more attention. 2) There are usual-
Cr0.6Ru0.4O2, by annealing Ru-exchanged MOF precursors.[140] ly many types of active sites in a given porous electrocatalyst
These transition metal-doped-RuO2, which have lower amounts (e.g., heteroatoms, edge sites, and intrinsic defects in a porous
of Ru, exhibited higher performances for OER in acidic condi- carbon-based electrocatalyst). The presence of diverse possi-
tion as compared to RuO2. In addition, more complex multi- ble catalytic sites makes it difficult to identify the key micro­
metal oxides (e.g., pyrochlore A2Ru2O7−δ, perovskite ARuO3) are structural fragments responsible for or contributing to the
also emerging as potential electrocatalysts for OER.[141] Their catalytic properties of the electrocatalyst. To address this prob-
structural diversity and compositional richness offers additional lem, high-precision synthetic methods that can produce high
advantage to improve their catalytic activity. For example, Yang quality electrocatalysts with homogeneous active sites that can
and co-workers reported a porous Y2[Ru1.6Y0.4]O7−δ pyrochlore easily be detected and studied should be developed. 3) While
oxide via sol–gel method using citric acid as chelating agent.[11a] some metal-based catalysts such as intermetallic borides and
The authors found that partial substitution of Ru4+ cations by iridium-based perovskites possess outstanding intrinsic activi-
the larger Y3+ cations at B-sites led to lattice oxygen vacancy, ties for HER and OER,[142] creating porosity in these materials is
optimizing their energy band structure and electrocatalytic inherently difficult. This is because conventional syntheses for
performances toward OER. A more flexible way to tailor the these materials often involve high temperatures and/or pres-
oxygen vacancies and improve their performance is by substi- sures, which tend to produce dense rather than porous struc-
tuting the A-site in porous Y1.8M0.2Ru2O7−δ catalysts with transi- tures. Thus, new approaches and milder synthetic conditions
tion metals (where M = Fe, Co, Ni, Cu, Y), as demonstrated by that can give these materials with porous structures need to be
Kuznetsov et  al.[141a] Despite these great advances, current Ru- explored. 4) Environmentally friendly, cost-effective, scale-up
based OER catalysts still hardly meet the critical requirements synthetic methods that help the transition of good catalytically
of high activity, low cost and, especially stability to be applicable active materials proven in academic laboratory settings into
in large-scale commercial PEM electrolyzers. Moreover, key practical electrochemical devices also need to be developed.
structural subunits and reaction mechanism for most studied ii) Better understanding of reaction mechanisms. Although re-
Ru-catalyzed OER process are not well understood, hampering cent advances in synthetic strategies have allowed a variety of
the rational modulation of Ru-based active sites. electrocatalysts with good catalytic properties to be obtained,
our understanding of how the catalysts work during the cata-
lytic processes is still limited in many cases. For example,
8. Summary and Outlook some N-doped carbon materials with subtle difference in mi-
crostructures have been reported to produce vastly different
Electrocatalysts are key to efficiently interconvert between products, product distributions, and Faradaic efficiency dur-
electrical and chemical energies and to realize renewable ing electrocatalytic CO2RR.[94,95] DFT calculations are widely
energy-based water, carbon, and nitrogen cycles. Porous elec- used to understand the nature of active sites and to investi-
trocatalysts are a family of important materials that can simul- gate catalytic reaction pathways, but in situ characterizations
taneously possess multiple features enabling high-performance to directly observe the active sites and reaction intermediates
electrocatalysis. In this review, we have summarized the fun- under actual catalytic conditions for many catalysts are still
damental aspects of active sites in different types of porous quite difficult. It is especially complicated for some catalysts

Adv. Mater. 2020, 32, 2002435 2002435  (27 of 32) © 2020 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.advmat.de

that undergo dynamic surface reconstruction during elec- non-noble transition metal-based HER catalysts including
trocatalysis. For example, iridium-based oxides can undergo sulfides and phosphides have been found to possess high
chemical dissolution and self-reconstruction forming amor- apparent activity (current per geometric surface area of the
phous layers of iridium oxides during OER in acidic media. electrode), which are comparable to those of Pt-based cata-
There is still disagreement as to what exactly is/are the OER lysts, their intrinsic activities for HER are still 1–2 order
active sites on iridium oxides. While some researchers have lower than that of Pt-based catalysts under both acidic and
claimed that the OER takes place at Ir sites with the Ir un- alkaline conditions.[145] Exploring earth-abundant HER elec-
dergoing changes in oxidation state,[143] others have proposed trocatalysts with Pt-like intrinsic activity should definitely be
that active oxygen atoms on the surfaces of the catalyst are an area of research in the near future. Oxygen electrocata-
where the reaction occurs.[144] These different results high- lytic reactions (OER/ORR) are currently the bottleneck reac-
light the need for more research works to unravel the origin tions limiting the realization of the water cycle, through wa-
of catalytic activity and reaction mechanism on many electro- ter electrolysis and H2-powered fuel cells. Many nonprecious
catalysts, including some of the most widely used materials catalysts have been demonstrated to display better catalytic
in electrocatalysis, such as iridium oxides. activity than precious metal-based catalysts (Ir/Ru-based
iii) Standardized electrochemical tests. Establishing standardized catalysts for OER and Pt for ORR). However, even the most
test protocols is crucial to reasonably compare the performances reactive OER/ORR catalysts require a minimum theoretical
of various electrocatalysts obtained from different sources overpotential of, at least, 0.3–0.4 V due to the inherent scal-
and with different synthetic strategies with one another. This ing relation of oxygen-containing intermediates. Designing
helps screening catalysts and determining the most optimal OER/ORR catalysts that can break the adsorption-energy
ones for different reactions as well as improving others. For scaling relations is an important, future research direction
HER and OER catalysts, their electrochemical activities are to achieve breakthrough performances in oxygen electro-
often evaluated by the overpotential required to yield a current catalysis. Additionally, novel strategies that can break the
density of 10 mA cm−2 (i.e., current per projected area of the inverse activity–stability relation of OER electrocatalysts are
electrode). However, the apparent catalytic activities of many urgently needed. Finally, NRR and CO2RR catalysts are still
electrocatalysts are affected by many factors, including their far from being satisfactory due to their high overpotential
morphology, porosity, size, and loading amount. To evaluate and poor selectivity toward the formation of ammonia and
the intrinsic catalytic activity of a catalyst, the current density C2+ products.
it produces should be normalized to the actual surface area
of the catalyst. Specifically, in the case of thin film-based elec-
trodes with flat surfaces, the actual scan areas determined by Acknowledgements
atomic force microscopy (AFM) or the electrochemically active
surface areas can be used for these purposes. In the case of X.Z. thanks for the financial supports from the National Natural
catalysts in the form of nanoparticles, the electrochemically Science Foundation of China (NSFC): Grant Nos. 21922507 and
21771079; the Fok Ying Tung Education Foundation: Grant No.161011.
active surface areas or the BET surface areas determined by H.C. acknowledges the financial supports from the NSFC (Grant No.
N2 adsorption and other porosimetric techniques are more 21901083), the Postdoctoral Innovative Talent Support Program (Grant
suitable. In addition, the mass activity (current per unit mass) No. BX20180120), and China Postdoctoral Science Foundation (Grant
of a catalyst is an important metric to evaluate catalysts, espe- No. 2018M641771). The authors also thank the NSFC (Grant No.
cially those containing noble metals. The TOF of the catalytic 21621001) and the 111 Project (No. B17020) for financial support.
reaction is highly useful to quantify the intrinsic activity of an
active site; however, the number of active sites in a catalyst is
sometimes hard to determine. Using these and any possible Conflict of Interest
standardized tests, benchmarking existing electrocatalysts in
The authors declare no conflict of interest.
terms of activity and stability is also highly recommended. For
ORR and CO2RR catalysts, good standardized tests are hard
to come by due to their multiple possible reaction pathways,
intermediates, and products. More efforts should be made to Keywords
find good standardized testing protocols in order to evaluate active sites, electrocatalysis, electronic structures, energy conversion,
the performances of many catalysts for these two reactions. porous materials
Finally, for a promising electrocatalyst discovered in the labora-
tory, further performance evaluation in a real electrochemical Received: April 9, 2020
device, such as an actual fuel cell and electrolyzer, is necessary. Revised: May 22, 2020
This is because the working environment of catalysts in the Published online: July 14, 2020
model reaction system applied by researchers in an academic
laboratory is not necessarily the same as one employed in a
real electrochemical device.
[1] B. Obama, Science 2017, 355, 126.
iv) Electrochemical performance requirements. The electro- [2] X. Zou, Y. Zhang, Chem. Soc. Rev. 2015, 44, 5148.
catalytic performances of many known earth-abundant cat- [3] a) P.  De Luna, C.  Hahn, D.  Higgins, S. A.  Jaffer, T. F.  Jaramillo,
alysts are still below the requirements needed to meet the E. H.  Sargent, Science 2019, 364, eaav3506; b) X.  Cui, C.  Tang,
practical commercial applications. For HER, although some Q. Zhang, Adv. Energy Mater. 2018, 8, 1800369.

Adv. Mater. 2020, 32, 2002435 2002435  (28 of 32) © 2020 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.advmat.de

[4] a) M. K. Debe, Nature 2012, 486, 43; b) L. Feng, J. Chang, K. Jiang, [22] B. Y.  Xia, H. B.  Wu, X.  Wang, X. W.  Lou, Angew. Chem., Int. Ed.
H. Xue, C. Liu, W.-B. Cai, W. Xing, J. Zhang, Nano Energy 2016, 30, 2013, 52, 12337.
355. [23] a) Q. Wang, Y. Lei, D. Wang, Y. J. E. Li, Energy Environ. Sci. 2019, 12,
[5] Z. W.  Seh, J.  Kibsgaard, C. F.  Dickens, I.  Chorkendorff, 1730; b) C. Xie, D. Yan, W. Chen, Y. Zou, R. Chen, S. Zang, Y. Wang,
J. K. Nørskov, T. F. Jaramillo, Science 2017, 355, eaad4998. X.  Yao, S.  Wang, Mater. Today 2019, 31, 47; c) Y.  Wang, P.  Han,
[6] a) Y.  Zheng, A.  Vasileff, X.  Zhou, Y.  Jiao, M.  Jaroniec, S.-Z.  Qiao, X.  Lv, L.  Zhang, G.  Zheng, Joule 2018, 2, 2551; d) D.  Yan, Y.  Li,
J. Am. Chem. Soc. 2019, 141, 7646; b) B. H. R. Suryanto, H.-L. Du, J. Huo, R. Chen, L. Dai, S. Wang, Adv. Mater. 2017, 29, 1606459.
D.  Wang, J.  Chen, A. N.  Simonov, D. R.  MacFarlane, Nat. Catal. [24] J.  Duan, S.  Chen, C. A.  Ortíz-Ledón, M.  Jaroniec, S. Z.  Qiao,
2019, 2, 290. Angew. Chem., Int. Ed. 2020, 59, 8181.
[7] a) J. Guo, J. Huo, Y. Liu, W. Wu, Y. Wang, M. Wu, H. Liu, G. Wang, [25] Y. He, P. Tang, Z. Hu, Q. He, C. Zhu, L. Wang, Q. Zeng, P. Golani,
Small Methods 2019, 3, 1900159; b) W.  Li, D.  Wang, Y.  Zhang, G. Gao, W. Fu, Nat. Commun. 2020, 11, 57.
L.  Tao, T.  Wang, Y.  Zou, Y.  Wang, R.  Chen, S.  Wang, Adv. Mater. [26] X. Liu, L. Dai, Nat. Rev. Mater. 2016, 1, 16064.
2020, 32, 1907879; c) S.  Ding, M. J.  Hülsey, J.  Pérez-Ramírez, [27] D.  Guo, R.  Shibuya, C.  Akiba, S.  Saji, T.  Kondo, J.  Nakamura,
N. Yan, Joule 2019, 3, 2897. Science 2016, 351, 361.
[8] C. Tang, H.-F. Wang, Q. Zhang, Acc. Chem. Res. 2018, 51, 881. [28] J.  Ortiz-Medina, Z.  Wang, R.  Cruz-Silva, A.  Morelos-Gomez,
[9] a) R.-Q.  Yao, Y.-T.  Zhou, H.  Shi, Q.-H.  Zhang, L.  Gu, Z.  Wen, F.  Wang, X.  Yao, M.  Terrones, M.  Endo, Adv. Mater. 2019, 31,
X.-Y.  Lang, Q.  Jiang, ACS Energy Lett. 2019, 4, 1379; b) Q.  Wu, 1805717.
M.  Luo, J.  Han, W.  Peng, Y.  Zhao, D.  Chen, M.  Peng, J.  Liu, [29] C.  Tang, H. F.  Wang, X.  Chen, B. Q.  Li, T. Z.  Hou, B.  Zhang,
F. M. F. de Groot, Y. Tan, ACS Energy Lett. 2020, 5, 192; c) X. Wang, Q. Zhang, M. M. Titirici, F. Wei, Adv. Mater. 2016, 28, 6845.
Y. Jia, X. Mao, D. Liu, W. He, J. Li, J. Liu, X. Yan, J. Chen, L. Song, [30] C. Tang, Q. Zhang, Adv. Mater. 2017, 29, 1604103.
A. Du, X. Yao, Adv. Mater. 2020, 32, 2000966; d) X. Zhang, Y. Zhao, [31] a) Y.  Jia, L.  Zhang, L.  Zhuang, H.  Liu, X.  Yan, X.  Wang, J.  Liu,
Y. Zhao, R. Shi, G. I. N. Waterhouse, T. Zhang, Adv. Energy Mater. J.  Wang, Y.  Zheng, Z.  Xiao, E.  Taran, J.  Chen, D.  Yang, Z.  Zhu,
2019, 9, 1900881. S. Wang, L. Dai, X. Yao, Nat. Catal. 2019, 2, 688; b) Y. Jia, L. Zhang,
[10] a) B.  Jiang, Y.  Guo, J.  Kim, A. E.  Whitten, K.  Wood, K.  Kani, A.  Du, G.  Gao, J.  Chen, X.  Yan, C. L.  Brown, X.  Yao, Adv. Mater.
A. E. Rowan, J. Henzie, Y. Yamauchi, J. Am. Chem. Soc. 2018, 140, 2016, 28, 9532.
12434; b) Y. Pi, J. Guo, Q. Shao, X. Huang, Chem. Mater. 2018, 30, [32] a) Y.  Lei, Y.  Wang, Y.  Liu, C.  Song, Q.  Li, D.  Wang, Y.  Li,
8571. Angew. Chem., Int. Ed. https://doi.org/10.1002/anie.201914647;
[11] a) J.  Kim, P.-C.  Shih, Y.  Qin, Z.  Al-Bardan, C.-J.  Sun, H.  Yang, b) Q.  Wang, X.  Huang, Z. L.  Zhao, M.  Wang, B.  Xiang, J.  Li,
Angew. Chem., Int. Ed. 2018, 57, 13877; b) W. Wang, L. Kuai, W. Cao, Z.  Feng, H.  Xu, M.  Gu, J. Am. Chem. Soc. 2020, 142, 7425;
M. Huttula, S. Ollikkala, T. Ahopelto, A.-P. Honkanen, S. Huotari, c) Q. He, D. Tian, H. Jiang, D. Cao, S. Wei, D. Liu, P. Song, Y. Lin,
M. Yu, B. Geng, Angew. Chem., Int. Ed. 2017, 56, 14977. L. Song, Adv. Mater. 2020, 32, 1906972.
[12] a) F.  Zhang, S.  Xi, G.  Lin, X.  Hu, X. W.  Lou, K.  Xie, Adv. Mater. [33] a) X.  Zhang, S.  Zhang, Y.  Yang, L.  Wang, Z.  Mu, H.  Zhu, X.  Zhu,
2019, 31, 1806552; b) B. Ni, T. He, J.-o. Wang, S. Zhang, C. Ouyang, H. Xing, H. Xia, B. Huang, J. Li, S. Guo, E. Wang, Adv. Mater. 2020,
Y. Long, J. Zhuang, X. Wang, Chem. Sci. 2018, 9, 2762. 32, 1906905; b) C.  Tang, B.  Wang, H.-F.  Wang, Q.  Zhang, Adv.
[13] T.  Sun, S.  Zhao, W.  Chen, D.  Zhai, J.  Dong, Y.  Wang, S.  Zhang, Mater. 2017, 29, 1703185.
A.  Han, L.  Gu, R.  Yu, X.  Wen, H.  Ren, L.  Xu, C.  Chen, Q.  Peng, [34] Y. Chen, S. Ji, C. Chen, Q. Peng, D. Wang, Y. Li, Joule 2018, 2, 1242.
D. Wang, Y. Li, Proc. Natl. Acad. Sci. USA 2018, 115, 12692. [35] H.  Chen, X.  Ai, W.  Liu, Z.  Xie, W.  Feng, W.  Chen, X.  Zou, Angew.
[14] a) T.  Zhao, A.  Elzatahry, X.  Li, D.  Zhao, Nat. Rev. Mater. 2019, Chem., Int. Ed. 2019, 58, 11409.
4, 775; b) K. A.  Cychosz, R.  Guillet-Nicolas, J.  García-Martínez, [36] Y. Li, Y. Sun, Y. Qin, W. Zhang, L. Wang, M. Luo, H. Yang, S. Guo,
M.  Thommes, Chem. Soc. Rev. 2017, 46, 389; c) X.-Y.  Yang, Adv. Energy Mater. 2020, 10, 1903120.
L.-H. Chen, Y. Li, J. C. Rooke, C. Sanchez, B.-L. Su, Chem. Soc. Rev. [37] J. Mao, C.-T. He, J. Pei, W. Chen, D. He, Y. He, Z. Zhuang, C. Chen,
2017, 46, 481; d) V. Malgras, H. Ataee-Esfahani, H. Wang, B. Jiang, Q. Peng, D. Wang, Y. Li, Nat. Commun. 2018, 9, 4958.
C.  Li, K. C. W.  Wu, J. H.  Kim, Y.  Yamauchi, Adv. Mater. 2016, [38] a) T. F. Jaramillo, K. P. Jørgensen, J. Bonde, J. H. Nielsen, S. Horch,
28, 993. I. Chorkendorff, Science 2007, 317, 100; b) Y. Chen, G. Yu, W. Chen,
[15] J. K.  Nørskov, T.  Bligaard, B.  Hvolbæk, F.  Abild-Pedersen, Y.  Liu, G.-D.  Li, P.  Zhu, Q.  Tao, Q.  Li, J.  Liu, X.  Shen, H.  Li,
I. Chorkendorff, C. H. Christensen, Chem. Soc. Rev. 2008, 37, 2163. X. Huang, D. Wang, T. Asefa, X. Zou, J. Am. Chem. Soc. 2017, 139,
[16] a) K. Zhu, X. Zhu, W. Yang, Angew. Chem., Int. Ed. 2019, 58, 1252; 12370.
b) X.  Li, X.  Yang, J.  Zhang, Y.  Huang, B.  Liu, ACS Catal. 2019, 9, [39] B.  Hinnemann, P. G.  Moses, J.  Bonde, K. P.  Jørgensen,
2521; c) J. H. K. Pfisterer, Y. Liang, O. Schneider, A. S. Bandarenka, J. H.  Nielsen, S.  Horch, I.  Chorkendorff, J. K.  Nørskov, J. Am.
Nature 2017, 549, 74. Chem. Soc. 2005, 127, 5308.
[17] J.  Greeley, I. E. L.  Stephens, A. S.  Bondarenko, T. P.  Johansson, [40] J.  Kibsgaard, Z.  Chen, B. N.  Reinecke, T. F.  Jaramillo, Nat. Mater.
H. A.  Hansen, T. F.  Jaramillo, J.  Rossmeisl, I.  Chorkendorff, 2012, 11, 963.
J. K. Nørskov, Nat. Chem. 2009, 1, 552. [41] J.  Deng, H.  Li, S.  Wang, D.  Ding, M.  Chen, C.  Liu, Z.  Tian,
[18] D. Xu, X. Liu, H. Lv, Y. Liu, S. Zhao, M. Han, J. Bao, J. He, B. Liu, K. S.  Novoselov, C.  Ma, D. Deng, X. Bao, Nat. Commun. 2017, 8,
Chem. Sci. 2018, 9, 4451. 14430.
[19] B. Garlyyev, J. Fichtner, O. Piqué, O. Schneider, A. S. Bandarenka, [42] Q. Xu, Y. Liu, H. Jiang, Y. Hu, H. Liu, C. Li, Adv. Energy Mater. 2019,
F. Calle-Vallejo, Chem. Sci. 2019, 10, 8060. 9, 1802553.
[20] S.  Liu, H.  Tao, L.  Zeng, Q.  Liu, Z.  Xu, Q.  Liu, J.-L.  Luo, J. Am. [43] a) H.  Zhou, F.  Yu, Y.  Huang, J.  Sun, Z.  Zhu, R. J.  Nielsen, R.  He,
Chem. Soc. 2017, 139, 2160. J.  Bao, W. A.  Goddard Iii, S.  Chen, Z.  Ren, Nat. Commun. 2016,
[21] a) W. Zhu, L. Zhang, P. Yang, C. Hu, Z. Luo, X. Chang, Z. J. Zhao, 7, 12765; b) Y. Liu, J. Wu, K. P. Hackenberg, J. Zhang, Y. M. Wang,
J.  Gong, Angew. Chem., Int. Ed. 2018, 57, 11544; b) D.  Gao, Y. Yang, K. Keyshar, J. Gu, T. Ogitsu, R. Vajtai, J. Lou, P. M. Ajayan,
H.  Zhou, J.  Wang, S.  Miao, F.  Yang, G.  Wang, J.  Wang, X.  Bao, J. Brandon C.  Wood, B. I.  Yakobson, Nat. Energy 2017, 2, 17127;
Am. Chem. Soc. 2015, 137, 4288; c) S.  Back, M. S.  Yeom, Y.  Jung, c) H.  Li, Y.  Tan, P.  Liu, C.  Guo, M.  Luo, J.  Han, T.  Lin, F.  Huang,
ACS Catal. 2015, 5, 5089; d) R.  Reske, H.  Mistry, F.  Behafarid, M.  Chen, Adv. Mater. 2016, 28, 8945; d) J.  Sun, F.  Tian, F.  Yu,
B. Roldan Cuenya, P. Strasser, J. Am. Chem. Soc. 2014, 136, 6978. Z. Yang, B. Yu, S. Chen, Z. Ren, H. Zhou, ACS Catal. 2020, 10, 1511.

Adv. Mater. 2020, 32, 2002435 2002435  (29 of 32) © 2020 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.advmat.de

[44] Q.  Lu, G. S.  Hutchings, W.  Yu, Y.  Zhou, R. V.  Forest, R.  Tao, Z. Chen, T. F. Jaramillo, J. Am. Chem. Soc. 2012, 134, 7758; c) W. Liu,
J.  Rosen, B. T.  Yonemoto, Z.  Cao, H.  Zheng, J. Q.  Xiao, F.  Jiao, P.  Rodriguez, L.  Borchardt, A.  Foelske, J.  Yuan, A.-K.  Herrmann,
J. G. Chen, Nat. Commun. 2015, 6, 6567. D. Geiger, Z. Zheng, S. Kaskel, N. Gaponik, R. Kötz, T. J. Schmidt,
[45] Z. Fang, L. Peng, Y. Qian, X. Zhang, Y. Xie, J. J. Cha, G. Yu, J. Am. A. Eychmüller, Angew. Chem., Int. Ed. 2013, 52, 9849.
Chem. Soc. 2018, 140, 5241. [71] X.  Tian, X.  Zhao, Y.-Q.  Su, L.  Wang, H.  Wang, D.  Dang, B.  Chi,
[46] J. Jia, T. Xiong, L. Zhao, F. Wang, H. Liu, R. Hu, J. Zhou, W. Zhou, H. Liu, E. J. M. Hensen, X. W. Lou, B. Y. Xia, Science 2019, 366, 850.
S. Chen, ACS Nano 2017, 11, 12509. [72] a) R.  Jiang, L.  Li, T.  Sheng, G.  Hu, Y.  Chen, L.  Wang, J. Am.
[47] Y.-R.  Zheng, P.  Wu, M.-R.  Gao, X.-L.  Zhang, F.-Y.  Gao, H.-X.  Ju, Chem. Soc. 2018, 140, 11594; b) M.  Qiao, Y.  Wang, Q.  Wang,
R.  Wu, Q.  Gao, R.  You, W.-X.  Huang, S.-J.  Liu, S.-W.  Hu, J.  Zhu, G.  Hu, X.  Mamat, S.  Zhang, S.  Wang, Angew. Chem., Int. Ed.
Z. Li, S.-H. Yu, Nat. Commun. 2018, 9, 2533. 2020, 59, 2688; c) X.  Wan, X.  Liu, Y.  Li, R.  Yu, L.  Zheng, W.  Yan,
[48] Z.  Luo, R.  Miao, T. D.  Huan, I. M.  Mosa, A. S.  Poyraz, H.  Wang, M.  Xu, J.  Shui, Nat. Catal. 2019, 2, 259; d) S. H.  Lee,
W.  Zhong, J. E.  Cloud, D. A.  Kriz, S.  Thanneeru, J.  He, Y.  Zhang, J.  Kim, D. Y.  Chung, J. M.  Yoo, H. S.  Lee, M. J.  Kim, B. S.  Mun,
R. Ramprasad, S. L. Suib, Adv. Energy Mater. 2016, 6, 1600528. S. G. Kwon, Y.-E. Sung, T. Hyeon, J. Am. Chem. Soc. 2019, 141, 2035;
[49] R. Wu, J. Zhang, Y. Shi, D. Liu, B. Zhang, J. Am. Chem. Soc. 2015, e) H. T. Chung, D. A. Cullen, D. Higgins, B. T. Sneed, E. F. Holby,
137, 6983. K. L.  More, P.  Zelenay, Science 2017, 357, 479; f) Y.  Chen, S.  Ji,
[50] T.  Zheng, C.  Shang, Z.  He, X.  Wang, C.  Cao, H.  Li, R.  Si, B.  Pan, Y.  Wang, J.  Dong, W.  Chen, Z.  Li, R.  Shen, L.  Zheng, Z.  Zhuang,
S. Zhou, J. Zeng, Angew. Chem., Int. Ed. 2019, 58, 14764. D. Wang, Y. Li, Angew. Chem., Int. Ed. 2017, 56, 6937.
[51] Y. Hou, M. Qiu, T. Zhang, X. Zhuang, C.-S. Kim, C. Yuan, X. Feng, [73] a) K.  Yuan, D.  Lützenkirchen-Hecht, L.  Li, L.  Shuai, Y.  Li, R.  Cao,
Adv. Mater. 2017, 29, 1701589. M.  Qiu, X.  Zhuang, M. K. H.  Leung, Y.  Chen, U.  Scherf, J. Am.
[52] Y.  Yin, J.  Han, Y.  Zhang, X.  Zhang, P.  Xu, Q.  Yuan, L.  Samad, Chem. Soc. 2020, 142, 2404; b) Y.-C.  Wang, Y.-J.  Lai, L.  Song,
X.  Wang, Y.  Wang, Z.  Zhang, P.  Zhang, X.  Cao, B.  Song, S.  Jin, Z.-Y.  Zhou, J.-G.  Liu, Q.  Wang, X.-D.  Yang, C.  Chen, W.  Shi,
J. Am. Chem. Soc. 2016, 138, 7965. Y.-P.  Zheng, M.  Rauf, S.-G.  Sun, Angew. Chem., Int. Ed. 2015, 54,
[53] B. Liu, Y. Wang, H.-Q. Peng, R. Yang, Z. Jiang, X. Zhou, C.-S. Lee, 9907; c) K.  Yuan, S.  Sfaelou, M.  Qiu, D.  Lützenkirchen-Hecht,
H. Zhao, W. Zhang, Adv. Mater. 2018, 30, 1803144. X. Zhuang, Y. Chen, C. Yuan, X. Feng, U. Scherf, ACS Energy Lett.
[54] H.  Zhang, P.  An, W.  Zhou, B. Y.  Guan, P.  Zhang, J.  Dong, 2018, 3, 252.
X. W. Lou, Sci. Adv. 2018, 4, eaao6657. [74] C. Hu, L. Dai, Adv. Mater. 2019, 31, 1804672.
[55] H.  Wei, K.  Huang, D.  Wang, R.  Zhang, B.  Ge, J.  Ma, B.  Wen, [75] H. Jiang, J. Gu, X. Zheng, M. Liu, X. Qiu, L. Wang, W. Li, Z. Chen,
S. Zhang, Q. Li, M. Lei, C. Zhang, J. Irawan, L.-M. Liu, H. Wu, Nat. X. Ji, J. Li, Energy Environ. Sci. 2019, 12, 322.
Commun. 2017, 8, 1490. [76] R. Silva, D. Voiry, M. Chhowalla, T. Asefa, J. Am. Chem. Soc. 2013,
[56] H. J. Qiu, Y. Ito, W. Cong, Y. Tan, P. Liu, A. Hirata, T. Fujita, Z. Tang, 135, 7823.
M. Chen, Angew. Chem., Int. Ed. 2015, 54, 14031. [77] Y. Meng, D. Voiry, A. Goswami, X. Zou, X. Huang, M. Chhowalla,
[57] K.  Jiang, B.  Liu, M.  Luo, S.  Ning, M.  Peng, Y.  Zhao, Y.-R.  Lu, Z. Liu, T. Asefa, J. Am. Chem. Soc. 2014, 136, 13554.
T.-S. Chan, F. M. F. de Groot, Y. Tan, Nat. Commun. 2019, 10, 1743. [78] D.  Li, Y.  Jia, G.  Chang, J.  Chen, H.  Liu, J.  Wang, Y.  Hu, Y.  Xia,
[58] a) X. Tian, X. F. Lu, B. Y. Xia, X. W. Lou, Joule 2020, 4, 45; b) M. Xiao, D. Yang, X. Yao, Chem 2018, 4, 2345.
J. Zhu, L. Feng, C. Liu, W. Xing, Adv. Mater. 2015, 27, 2521. [79] X. Yan, Y. Jia, X. Yao, Chem. Soc. Rev. 2018, 47, 7628.
[59] H.  Peng, F.  Liu, X.  Liu, S.  Liao, C.  You, X.  Tian, H.  Nan, F.  Luo, [80] Y. Jiang, L. Yang, T. Sun, J. Zhao, Z. Lyu, O. Zhuo, X. Wang, Q. Wu,
H. Song, Z. Fu, P. Huang, ACS Catal. 2014, 4, 3797. J. Ma, Z. Hu, ACS Catal. 2015, 5, 6707.
[60] K. Gong, F. Du, Z. Xia, M. Durstock, L. Dai, Science 2009, 323, 760. [81] S. K.  Singh, K.  Takeyasu, J.  Nakamura, Adv. Mater. 2019, 31,
[61] L.  Zhang, C.-Y.  Lin, D.  Zhang, L.  Gong, Y.  Zhu, Z.  Zhao, Q.  Xu, 1804297.
H. Li, Z. Xia, Adv. Mater. 2019, 31, 1805252. [82] A. Bagger, W. Ju, A. S. Varela, P. Strasser, J. Rossmeisl, ChemPhy-
[62] A.  Kulkarni, S.  Siahrostami, A.  Patel, J. K.  Nørskov, Chem. Rev. sChem 2017, 18, 3266.
2018, 118, 2302. [83] S.  Nitopi, E.  Bertheussen, S. B.  Scott, X.  Liu, A. K.  Engstfeld,
[63] a) Y.  Jiao, Y.  Zheng, M.  Jaroniec, S. Z.  Qiao, J. Am. Chem. Soc. S.  Horch, B.  Seger, I. E. L.  Stephens, K.  Chan, C.  Hahn,
2014, 136, 4394; b) H. J. Yan, B. Xu, S. Q. Shi, C. Y. Ouyang, J. Appl. J. K. Nørskov, T. F. Jaramillo, I. Chorkendorff, Chem. Rev. 2019, 119,
Phys. 2012, 112, 104316. 7610.
[64] J. K. Nørskov, J. Rossmeisl, A. Logadottir, L. Lindqvist, J. R. Kitchin, [84] X.  Liu, J.  Xiao, H.  Peng, X.  Hong, K.  Chan, J. K.  Nørskov, Nat.
T. Bligaard, H. Jónsson, J. Phys. Chem. B 2004, 108, 17886. Commun. 2017, 8, 15438.
[65] A. Vojvodic, J. K. Nørskov, Natl. Sci. Rev. 2015, 2, 140. [85] H. Liu, Y. Zhu, J. Ma, Z. Zhang, W. Hu, Adv. Funct. Mater. 2020, 30,
[66] a) J. H.  Zagal, M. T. M.  Koper, Angew. Chem., Int. Ed. 2016, 55, 1910534.
14510; b) M. Li, L. Zhang, Q. Xu, J. Niu, Z. Xia, J. Catal. 2014, 314, [86] a) B. A.  Rosen, A.  Salehi-Khojin, M. R.  Thorson, W.  Zhu,
66; c) G.-L. Chai, K. Qiu, M. Qiao, M.-M. Titirici, C. Shang, Z. Guo, D. T.  Whipple, P. J. A.  Kenis, R. I.  Masel, Science 2011, 334, 643;
Energy Environ. Sci. 2017, 10, 1186; d) S. Liu, Z. Li, C. Wang, W. Tao, b) A.  Salehi-Khojin, H.-R. M.  Jhong, B. A.  Rosen, W.  Zhu, S.  Ma,
M.  Huang, M.  Zuo, Y.  Yang, K.  Yang, L.  Zhang, S.  Chen, P.  Xu, P. J. A. Kenis, R. I. Masel, J. Phys. Chem. C 2013, 117, 1627.
Q. Chen, Nat. Commun. 2020, 11, 938. [87] Q. Lu, J. Rosen, Y. Zhou, G. S. Hutchings, Y. C. Kimmel, J. G. Chen,
[67] J.  Liu, H.  Liu, H.  Chen, X.  Du, B.  Zhang, Z.  Hong, S.  Sun, F. Jiao, Nat. Commun. 2014, 5, 3242.
W. Wang, Adv. Sci. 2020, 7, 1901614. [88] a) B.  Kumar, V.  Atla, J. P.  Brian, S.  Kumari, T. Q.  Nguyen,
[68] a) Y.  Feng, Q.  Shao, Y.  Ji, X.  Cui, Y.  Li, X.  Zhu, X.  Huang, Sci. M.  Sunkara, J. M.  Spurgeon, Angew. Chem., Int. Ed. 2017, 56,
Adv. 2018, 4, eaap8817; b) M.  Escudero-Escribano, P.  Malacrida, 3645; b) G.  Hyun, J. T.  Song, C.  Ahn, Y.  Ham, D.  Cho, J.  Oh,
M. H.  Hansen, U. G.  Vej-Hansen, A.  Velázquez-Palenzuela, S.  Jeon, Proc. Natl. Acad. Sci. USA 2020, 117, 5680; c) P.-P.  Yang,
V.  Tripkovic, J.  Schiøtz, J.  Rossmeisl, I. E. L.  Stephens, X.-L.  Zhang, F.-Y.  Gao, Y.-R.  Zheng, Z.-Z.  Niu, X.  Yu, R.  Liu,
I. Chorkendorff, Science 2016, 352, 73. Z.-Z. Wu, S. Qin, L.-P. Chi, Y. Duan, T. Ma, X.-S. Zheng, J.-F. Zhu,
[69] Z.-F. Huang, J. Song, S. Dou, X. Li, J. Wang, X. Wang, Matter 2019, H.-J. Wang, M.-R. Gao, S.-H. Yu, J. Am. Chem. Soc. 2020, 142, 6400.
1, 1494. [89] a) Q.  Gong, P.  Ding, M.  Xu, X.  Zhu, M.  Wang, J.  Deng, Q.  Ma,
[70] a) H. Cheng, S. Liu, Z. Hao, J. Wang, B. Liu, G. Liu, X. Wu, W. Chu, N. Han, Y. Zhu, J. Lu, Z. Feng, Y. Li, W. Zhou, Y. Li, Nat. Commun.
C. Wu, Y. Xie, Chem. Sci. 2019, 10, 5589; b) J. Kibsgaard, Y. Gorlin, 2019, 10, 2807; b) N.  Han, P.  Ding, L.  He, Y.  Li, Y.  Li, Adv. Energy

Adv. Mater. 2020, 32, 2002435 2002435  (30 of 32) © 2020 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.advmat.de

Mater. 2020, 10, 1902338; c) H.  Yang, N.  Han, J.  Deng, J.  Wu, Chem 2019, 5, 204; b) L.  Han, X.  Liu, J.  Chen, R.  Lin, H.  Liu,
Y. Wang, Y. Hu, P. Ding, Y. Li, Y. Li, J. Lu, Adv. Energy Mater. 2018, F.  Lü, S.  Bak, Z.  Liang, S.  Zhao, E.  Stavitski, J.  Luo, R. R.  Adzic,
8, 1801536; d) N. Han, Y. Wang, H. Yang, J. Deng, J. Wu, Y. Li, Y. Li, H. L. Xin, Angew. Chem., Int. Ed. 2019, 58, 2321; c) Q. Qin, T. Heil,
Nat. Commun. 2018, 9, 1320. M. Antonietti, M. Oschatz, Small Methods 2018, 2, 1800202.
[90] a) Y. Pan, R. Lin, Y. Chen, S. Liu, W. Zhu, X. Cao, W. Chen, K. Wu, [105] Y. Liu, Y. Su, X. Quan, X. Fan, S. Chen, H. Yu, H. Zhao, Y. Zhang,
W.-C. Cheong, Y. Wang, L. Zheng, J. Luo, Y. Lin, Y. Liu, C. Liu, J. Li, J. Zhao, ACS Catal. 2018, 8, 1186.
Q.  Lu, X.  Chen, D.  Wang, Q.  Peng, C.  Chen, Y.  Li, J. Am. Chem. [106] Y. Liu, Q. Li, X. Guo, X. Kong, J. Ke, M. Chi, Q. Li, Z. Geng, J. Zeng,
Soc. 2018, 140, 4218; b) Y. Liu, X. Fan, A. Nayak, Y. Wang, B. Shan, Adv. Mater. 2020, 32, 1907690.
X.  Quan, T. J.  Meyer, Proc. Natl. Acad. Sci. USA 2019, 116, 26353; [107] Y. Yang, L. Zhang, Z. Hu, Y. Zheng, C. Tang, P. Chen, R. Wang, K. Qiu,
c) L.  Zhang, J. M. T. A.  Fischer, Y.  Jia, X.  Yan, W.  Xu, X.  Wang, J. Mao, T. Ling, S.-Z. Qiao, Angew. Chem., Int. Ed. 2020, 59, 4525.
J.  Chen, D.  Yang, H.  Liu, L.  Zhuang, M.  Hankel, D. J.  Searles, [108] a) S.  Jin, ACS Energy Lett. 2017, 2, 1937; b) L.-A.  Stern, L.  Feng,
K.  Huang, S.  Feng, C. L.  Brown, X.  Yao, J. Am. Chem. Soc. 2018, F. Song, X. Hu, Energy Environ. Sci. 2015, 8, 2347.
140, 10757; d) N.  Leonard, W.  Ju, I.  Sinev, J.  Steinberg, F.  Luo, [109] a) F.  Guo, Y.  Wu, H.  Chen, Y.  Liu, L.  Yang, X.  Ai, X.  Zou, Energy
A. S.  Varela, B.  Roldan Cuenya, P.  Strasser, Chem. Sci. 2018, 9, Environ. Sci. 2019, 12, 684; b) A. Grimaud, K. J. May, C. E. Carlton,
5064; e) T.  Asset, S. T.  Garcia, S.  Herrera, N.  Andersen, Y.  Chen, Y.-L.  Lee, M.  Risch, W. T.  Hong, J.  Zhou, Y.  Shao-Horn, Nat.
E. J. Peterson, I. Matanovic, K. Artyushkova, J. Lee, S. D. Minteer, Commun. 2013, 4, 2439; c) C. C. L. McCrory, S. Jung, I. M. Ferrer,
S. Dai, X. Pan, K. Chavan, S. Calabrese Barton, P. Atanassov, ACS S. M. Chatman, J. C. Peters, T. F. Jaramillo, J. Am. Chem. Soc. 2015,
Catal. 2019, 9, 7668. 137, 4347; d) H. Xu, H. Shang, C. Wang, L. Jin, C. Chen, C. Wang,
[91] H.  Yang, Y.  Wu, G.  Li, Q.  Lin, Q.  Hu, Q.  Zhang, J.  Liu, C.  He, Y. Du, Appl. Catal., B 2020, 265, 118605; e) H. Xu, H. Shang, L. Jin,
J. Am. Chem. Soc. 2019, 141, 12717. C. Chen, C. Wang, Y. Du, J. Mater. Chem. A 2019, 7, 26905.
[92] Y.  Wang, Z.  Chen, P.  Han, Y.  Du, Z.  Gu, X.  Xu, G.  Zheng, ACS [110] a) L. C.  Seitz, C. F.  Dickens, K.  Nishio, Y.  Hikita, J.  Montoya,
Catal. 2018, 8, 7113. A.  Doyle, C.  Kirk, A.  Vojvodic, H. Y.  Hwang, J. K.  Norskov,
[93] a) L.  Ye, Y.  Ying, D.  Sun, Z.  Zhang, L.  Fei, Z.  Wen, J.  Qiao, T. F. Jaramillo, Science 2016, 353, 1011; b) Y. Chen, H. Li, J. Wang,
H.  Huang, Angew. Chem., Int. Ed. 2020, 59, 3244; b) B.  Zhang, Y. Du, S. Xi, Y. Sun, M. Sherburne, J. W. Ager, A. C. Fisher, Z. J. Xu,
J. Zhang, F. Zhang, L. Zheng, G. Mo, B. Han, G. Yang, Adv. Funct. Nat. Commun. 2019, 10, 572; c) X.  Liang, L.  Shi, Y.  Liu, H.  Chen,
Mater. 2020, 30, 1906194. R. Si, W. Yan, Q. Zhang, G.-D. Li, L. Yang, X. Zou, Angew. Chem.,
[94] W.  Wang, L.  Shang, G.  Chang, C.  Yan, R.  Shi, Y.  Zhao, Int. Ed. 2019, 58, 7631; d) L. Yang, G. Yu, X. Ai, W. Yan, H. Duan,
G. I. N.  Waterhouse, D.  Yang, T.  Zhang, Adv. Mater. 2019, 31, W.  Chen, X.  Li, T.  Wang, C.  Zhang, X.  Huang, J.-S.  Chen, X.  Zou,
1808276. Nat. Commun. 2018, 9, 5236.
[95] Y. Song, W. Chen, C. Zhao, S. Li, W. Wei, Y. Sun, Angew. Chem., Int. [111] R. L. Doyle, M. E. G. Lyons, in Photoelectrochemical Solar Fuel Pro-
Ed. 2017, 56, 10840. duction: From Basic Principles to Advanced Devices (Eds: S. Giménez,
[96] M. A. Shipman, M. D. Symes, Catal. Today 2017, 286, 57. J. Bisquert), Springer International Publishing, Cham 2016, p. 41.
[97] a) E.  Skúlason, T.  Bligaard, S.  Gudmundsdóttir, F.  Studt, [112] a) I. C. Man, H.-Y. Su, F. Calle-Vallejo, H. A. Hansen, J. I. Martínez,
J.  Rossmeisl, F.  Abild-Pedersen, T.  Vegge, H.  Jónsson, N. G. Inoglu, J. Kitchin, T. F. Jaramillo, J. K. Nørskov, J. Rossmeisl,
J. K.  Nørskov, Phys. Chem. Chem. Phys. 2012, 14, 1235; ChemCatChem 2011, 3, 1159; b) M. T. M.  Koper, J. Electroanal.
b) S. L.  Foster, S. I. P.  Bakovic, R. D.  Duda, S.  Maheshwari, Chem. 2011, 660, 254.
R. D. Milton, S. D. Minteer, M. J. Janik, J. N. Renner, L. F. Greenlee, [113] J. Suntivich, K. J. May, H. A. Gasteiger, J. B. Goodenough, Y. Shao-
Nat. Catal. 2018, 1, 490. Horn, Science 2011, 334, 1383.
[98] J. H.  Montoya, C.  Tsai, A.  Vojvodic, J. K.  Nørskov, ChemSusChem [114] C. Yang, G. Rousse, K. Louise Svane, P. E. Pearce, A. M. Abakumov,
2015, 8, 2180. M.  Deschamps, G.  Cibin, A. V.  Chadwick, D. A.  Dalla Corte,
[99] A. J.  Medford, A.  Vojvodic, J. S.  Hummelshøj, J.  Voss, F.  Abild- H.  Anton Hansen, T.  Vegge, J.-M.  Tarascon, A.  Grimaud, Nat.
Pedersen, F. Studt, T. Bligaard, A. Nilsson, J. K. Nørskov, J. Catal. Commun. 2020, 11, 1378.
2015, 328, 36. [115] V. R. Stamenkovic, D. Strmcnik, P. P. Lopes, N. M. Markovic, Nat.
[100] a) M. A. Shipman, M. D. Symes, Electrochim. Acta 2017, 258, 618; Mater. 2017, 16, 57.
b) B.  Hu, M.  Hu, L.  Seefeldt, T. L.  Liu, ACS Energy Lett. 2019, 4, [116] A.  Grimaud, O.  Diaz-Morales, B.  Han, W. T.  Hong, Y.-L.  Lee,
1053. L. Giordano, K. A. Stoerzinger, M. T. M. Koper, Y. Shao-Horn, Nat.
[101] a) S. Z. Andersen, V. Čolić, S. Yang, J. A. Schwalbe, A. C. Nielander, Chem. 2017, 9, 457.
J. M. McEnaney, K. Enemark-Rasmussen, J. G. Baker, A. R. Singh, [117] a) N.  Danilovic, R.  Subbaraman, K.-C.  Chang, S. H.  Chang,
B. A.  Rohr, M. J.  Statt, S. J.  Blair, S.  Mezzavilla, J.  Kibsgaard, Y. J.  Kang, J.  Snyder, A. P.  Paulikas, D.  Strmcnik, Y.-T.  Kim,
P. C. K.  Vesborg, M.  Cargnello, S. F.  Bent, T. F.  Jaramillo, D.  Myers, V. R.  Stamenkovic, N. M.  Markovic, J. Phys. Chem.
I. E. L. Stephens, J. K. Nørskov, I. Chorkendorff, Nature 2019, 570, Lett. 2014, 5, 2474; b) N. Li, D. K. Bediako, R. G. Hadt, D. Hayes,
504; b) C. Tang, S.-Z. Qiao, Chem. Soc. Rev. 2019, 48, 3166. T. J.  Kempa, F.  von  Cube, D. C.  Bell, L. X.  Chen, D. G.  Nocera,
[102] a) W. Xu, G. Fan, J. Chen, J. Li, L. Zhang, S. Zhu, X. Su, F. Cheng, Proc. Natl. Acad. Sci. USA 2017, 114, 1486.
J.  Chen, Angew. Chem., Int. Ed. 2020, 59, 3511; b) J.  Wang, [118] D. A. Corrigan, J. Electrochem. Soc. 1987, 134, 377.
B.  Huang, Y.  Ji, M.  Sun, T.  Wu, R.  Yin, X.  Zhu, Y.  Li, Q.  Shao, [119] L. Trotochaud, J. K. Ranney, K. N. Williams, S. W. Boettcher, J. Am.
X. Huang, Adv. Mater. 2020, 32, 1907112; c) X. Yang, F. Ling, X. Zi, Chem. Soc. 2012, 134, 17253.
Y.  Wang, H.  Zhang, H.  Zhang, M.  Zhou, Z.  Guo, Y.  Wang, Small [120] M. W. Louie, A. T. Bell, J. Am. Chem. Soc. 2013, 135, 12329.
2020, 16, 2000421. [121] R.  Subbaraman, D.  Tripkovic, K.-C.  Chang, D.  Strmcnik,
[103] Y.  Wang, M.-M.  Shi, D.  Bao, F.-L.  Meng, Q.  Zhang, Y.-T.  Zhou, A. P.  Paulikas, P.  Hirunsit, M.  Chan, J.  Greeley, V.  Stamenkovic,
K.-H.  Liu, Y.  Zhang, J.-Z.  Wang, Z.-W.  Chen, D.-P.  Liu, Z.  Jiang, N. M. Markovic, Nat. Mater. 2012, 11, 550.
M.  Luo, L.  Gu, Q.-H.  Zhang, X.-Z.  Cao, Y.  Yao, M.-H.  Shao, [122] M.  Görlin, P.  Chernev, J.  Ferreira de Araújo, T.  Reier, S.  Dresp,
Y.  Zhang, X.-B.  Zhang, J. G.  Chen, J.-M.  Yan, Q.  Jiang, Angew. B. Paul, R. Krähnert, H. Dau, P. Strasser, J. Am. Chem. Soc. 2016,
Chem., Int. Ed. 2019, 58, 9464. 138, 5603.
[104] a) H.  Tao, C.  Choi, L.-X.  Ding, Z.  Jiang, Z.  Han, M.  Jia, Q.  Fan, [123] L.  Trotochaud, S. L.  Young, J. K.  Ranney, S. W.  Boettcher, J. Am.
Y. Gao, H. Wang, A. W. Robertson, S. Hong, Y. Jung, S. Liu, Z. Sun, Chem. Soc. 2014, 136, 6744.

Adv. Mater. 2020, 32, 2002435 2002435  (31 of 32) © 2020 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.advmat.de

[124] M. S.  Burke, S.  Zou, L. J.  Enman, J. E.  Kellon, C. A.  Gabor, S. H.  Chang, Y.  Kang, J.  Snyder, A. P.  Paulikas, D.  Strmcnik,
E. Pledger, S. W. Boettcher, J. Phys. Chem. Lett. 2015, 6, 3737. Y. T.  Kim, D.  Myers, V. R.  Stamenkovic, N. M.  Markovic,
[125] M. B.  Stevens, C. D. M.  Trang, L. J.  Enman, J.  Deng, Angew. Chem., Int. Ed. 2014, 53, 14016; d) Y.-T.  Kim, P. P.  Lopes,
S. W. Boettcher, J. Am. Chem. Soc. 2017, 139, 11361. S.-A. Park, A. Y. Lee, J. Lim, H. Lee, S. Back, Y. Jung, N. Danilovic,
[126] D.  Friebel, M. W.  Louie, M.  Bajdich, K. E.  Sanwald, Y.  Cai, V.  Stamenkovic, J.  Erlebacher, J.  Snyder, N. M.  Markovic, Nat.
A. M. Wise, M.-J. Cheng, D. Sokaras, T.-C. Weng, R. Alonso-Mori, Commun. 2017, 8, 1449.
R. C. Davis, J. R. Bargar, J. K. Nørskov, A. Nilsson, A. T. Bell, J. Am. [138] H. N.  Nong, T.  Reier, H.-S.  Oh, M.  Gliech, P.  Paciok, T. H. T.  Vu,
Chem. Soc. 2015, 137, 1305. D. Teschner, M. Heggen, V. Petkov, R. Schlögl, T. Jones, P. Strasser,
[127] J. Y. C. Chen, L. Dang, H. Liang, W. Bi, J. B. Gerken, S. Jin, E. E. Alp, Nat. Catal. 2018, 1, 841.
S. S. Stahl, J. Am. Chem. Soc. 2015, 137, 15090. [139] Z. Zhuang, Y. Wang, C.-Q. Xu, S. Liu, C. Chen, Q. Peng, Z. Zhuang,
[128] Z. K.  Goldsmith, A. K.  Harshan, J. B.  Gerken, M.  Vörös, G.  Galli, H. Xiao, Y. Pan, S. Lu, R. Yu, W.-C. Cheong, X. Cao, K. Wu, K. Sun,
S. S.  Stahl, S.  Hammes-Schiffer, Proc. Natl. Acad. Sci. USA 2017, Y. Wang, D. Wang, J. Li, Y. Li, Nat. Commun. 2019, 10, 4875.
114, 3050. [140] a) Y.  Lin, Z.  Tian, L.  Zhang, J.  Ma, Z.  Jiang, B. J.  Deibert, R.  Ge,
[129] R. Chen, S.-F. Hung, D. Zhou, J. Gao, C. Yang, H. Tao, H. B. Yang, L.  Chen, Nat. Commun. 2019, 10, 162; b) S.  Chen, H.  Huang,
L.  Zhang, L.  Zhang, Q.  Xiong, H. M.  Chen, B.  Liu, Adv. Mater. P.  Jiang, K.  Yang, J.  Diao, S.  Gong, S.  Liu, M.  Huang, H.  Wang,
2019, 31, 1903909. Q.  Chen, ACS Catal. 2020, 10, 1152; c) J.  Su, R.  Ge, K.  Jiang,
[130] a) D. Y.  Chung, P. P.  Lopes, P.  Farinazzo Bergamo Dias Martins, Y.  Dong, F.  Hao, Z.  Tian, G.  Chen, L.  Chen, Adv. Mater. 2018, 30,
H. He, T. Kawaguchi, P. Zapol, H. You, D. Tripkovic, D. Strmcnik, 1801351.
Y. Zhu, S. Seifert, S. Lee, V. R. Stamenkovic, N. M. Markovic, Nat. [141] a) D. A.  Kuznetsov, M. A.  Naeem, P. V.  Kumar, P. M.  Abdala,
Energy 2020, 5, 222; b) M. Gong, H. Dai, Nano Res. 2015, 8, 23. A.  Fedorov, C. R.  Müller, J. Am. Chem. Soc. 2020, 142, 7883;
[131] Y. Liu, X. Liang, L. Gu, Y. Zhang, G.-D. Li, X. Zou, J.-S. Chen, Nat. b) M.  Retuerto, L.  Pascual, F.  Calle-Vallejo, P.  Ferrer, D.  Gianolio,
Commun. 2018, 9, 2609. A. G.  Pereira, Á.  García, J.  Torrero, M. T.  Fernández-Díaz,
[132] X. Zou, Y. Wu, Y. Liu, D. Liu, W. Li, L. Gu, H. Liu, P. Wang, L. Sun, P.  Bencok, M. A.  Peña, J. L. G.  Fierro, S.  Rojas, Nat. Commun.
Y. Zhang, Chem 2018, 4, 1139. 2019, 10, 2041.
[133] R. Gao, G.-D. Li, J. Hu, Y. Wu, X. Lian, D. Wang, X. Zou, Catal. Sci. [142] a) X.  Ai, X.  Zou, H.  Chen, Y.  Su, X.  Feng, Q.  Li, Y.  Liu, Y.  Zhang,
Technol. 2016, 6, 8268. X.  Zou, Angew. Chem., Int. Ed. 2020, 59, 3961; b) Q.  Li, X.  Zou,
[134] H.  Zhang, W.  Zhou, J.  Dong, X. F.  Lu, X. W.  Lou, Energy Environ. X. Ai, H. Chen, L. Sun, X. Zou, Adv. Energy Mater. 2019, 9, 1803369;
Sci. 2019, 12, 3348. c) O.  Diaz-Morales, S.  Raaijman, R.  Kortlever, P. J.  Kooyman,
[135] E. Ortel, T. Reier, P. Strasser, R. Kraehnert, Chem. Mater. 2011, 23, T. Wezendonk, J. Gascon, W. T. Fu, M. T. M. Koper, Nat. Commun.
3201. 2016, 7, 12363.
[136] E.  Oakton, D.  Lebedev, M.  Povia, D. F.  Abbott, E.  Fabbri, [143] Z. Pavlovic, C. Ranjan, M. van Gastel, R. Schlögl, Chem. Commun.
A.  Fedorov, M.  Nachtegaal, C.  Copéret, T. J.  Schmidt, ACS Catal. 2017, 53, 12414.
2017, 7, 2346. [144] a) A.  Grimaud, A.  Demortière, M.  Saubanère, W.  Dachraoui,
[137] a) J. Feng, F. Lv, W. Zhang, P. Li, K. Wang, C. Yang, B. Wang, Y. Yang, M.  Duchamp, M.-L.  Doublet, J.-M.  Tarascon, Nat. Energy 2017,
J. Zhou, F. Lin, G.-C. Wang, S. Guo, Adv. Mater. 2017, 29, 1703798; 2, 16189; b) V.  Pfeifer, T. E.  Jones, J. J.  Velasco Vélez, R.  Arrigo,
b) Q. Shi, C. Zhu, H. Zhong, D. Su, N. Li, M. H. Engelhard, H. Xia, S.  Piccinin, M.  Hävecker, A.  Knop-Gericke, R.  Schlögl, Chem. Sci.
Q. Zhang, S. Feng, S. P. Beckman, D. Du, Y. Lin, ACS Energy Lett. 2017, 8, 2143.
2018, 3, 2038; c) N.  Danilovic, R.  Subbaraman, K. C.  Chang, [145] J. Kibsgaard, I. Chorkendorff, Nat. Energy 2019, 4, 430.

Adv. Mater. 2020, 32, 2002435 2002435  (32 of 32) © 2020 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

View publication stats

You might also like