Grubb ^[1] first put forward the concept of electrocatalysis in 1963, which he defined as the enhancement of the rate of an electrochemical reaction by a substance. Electrocatalysis has experienced great progress and gradually developed into an interdisciplinary science and comprehensive technology that involves electrochemistry, materials science, surface science, catalytic science and many other branches of science. Electrocatalysis plays an indispensable role in industrial and agricultural production, national defense, economic construction, energy development and environmental protection ^{[2, 3, 4, 5, 6, 7, 8, 9, 10]}. At present, electrocatalysis, as an essential and effective means for an electrochemical and catalytic united-technology, is widespread, and has been applied to the controllable synthesis and functional assembly of nanomaterials ^{[11, 12, 13, 14]}, micro-nano fabrications and electronic devices ^{[15, 16, 17, 18, 19, 20]}, metal processing, forming, plating and finishing ^{[21, 22, 23, 24, 25]}, organic chemical reactions ^{[26, 27, 28]}, nitrogen cycle and chlor-alkali industry ^{[29, 30, 31]}, electrochemical luminescence and photosensitive measurement ^{[32, 33, 34]}, electrochemical sensors and biosensors ^{[35, 36, 37, 38]}, electrochemical engineering for cell biology and medical therapy ^{[39, 40, 41, 42, 43]}, electrochemical process of detection and analysis ^{[44, 45, 46, 47]}, and organic waste degradation and wastewater treatment ^{[48, 49, 50, 51]}. In particular, to alleviate and address the increasingly severe energy and environmental crisis, efforts have been intensively focused on electrocatalysis for energy development, conversion and storage, such as chemical batteries or metal-air batteries ^{[52, 53, 54, 55]}, super-capacitors ^{[56, 57, 58, 59]}, solar water splitting cells ^{[60, 61, 62, 63]}, and low-temperature fuel cells ^{[64, 65, 66, 67, 68, 69]}.

The great advances and broad applications of electrocatalysis are greatly facilitated by the design and controlled synthesis of metal nanocatalysts ^{[52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69]}. The metal nanocatalysts (especially transition metal and noble metal nanocatalysts) have been widely used as electrocatalysts, mainly because the transition metals and noble metals may provide d-orbital lone-pair electrons as nucleophilic reagents or provide empty d-orbitals as electrophilic reagents in chemical reactions. The metals may coordinate with alien species to form intermediates to reduce the reaction activation energy, eventually promoting the catalytic reaction. Meanwhile, they may also provide adjustable d-band locations and electronic structures by tuning their composition and structural features, which alters the adsorption and stripping abilities of reactants/products on their surface/interface and further improves the activity, selectivity and stability of nanocatalysts ^{[70, 71, 72, 73, 74, 75, 76, 77, 78]}. Thus far, various unique transition/noble metal nanocrystals and their hybrid nanostructures have been synthesized through a variety of gas-phase, solid-phase and solution-phase synthetic methods as well as loading and assembly techniques. In particular, the solution-phase synthesis provides a versatile platform for controlled synthesis by manipulating the nucleation and growth process that involves atom generation, nucleation, seed formation, seeding growth and/or epitaxial growth. The experimental parameters that can be tuned in a reaction system include precursors, reducing agents, capping agents, etching agents, solvents, temperature and many others. Furthermore, combined with post-synthesis techniques, such as etching, loading and assembly, the synthesis can yield metal nanocrystals and their hybrid nanostructures with well-defined sizes, shapes and structures ^{[72, 73, 74, 75, 76, 77, 78]}.

Noble metal catalysts (mainly Pt, Pd, Ru and Rh) with high activity, selectivity and stability are prevalent in catalysis and still cannot be completely replaced by other materials ^{[79, 80, 81, 82]}. The major bottleneck limiting the catalytic applications of noble metals is their high cost and unsustainability owing to their relatively low abundance in nature. For this reason, an urgent challenge facing the human race is how to design, synthesize and assemble high-performance and low-cost metal nanocatalysts. In the past decade, great efforts have been made towards this goal, leading to numerous gratifying and encouraging achievements ^{[72, 73, 74, 75, 76, 77, 78, 79, 80, 81, 82]}. The catalytic properties of metal nanocatalysts are closely related to their shapes, structures, sizes and compositions that are essentially correlated with atomic arrangements, molecular adsorption and activation, electronic structures, charge transport, and other surface and interface features. Inspired by previous researches, we can simply define electrocatalysis as the heterogeneous catalytic reaction that occurs on the surface and interface of electrode materials and electrolytes ^{[2, 3, 4, 5, 6, 7, 8, 9, 10, 83]}. Thus, surface and interface engineering is vital to the design, synthesis, construction and optimization of metal nanostructures and their hybrid nanostructures.

Fuel cells (FCs) are a class of electrochemical power-generation devices which can directly and continuously convert the chemical energy of fuels and oxidants into electrical energy. Compared with other traditional power-generation devices, fuel cells have the merits of low noise, fast response, good modularization, high operation quality and high energy conversion efficiency and can adapt well to different load power requirements, waste-heat utilization and catalyst loading/manufacturing solutions ^{[84, 85, 86, 87, 88]}. Fuel cells can be classified as molten carbonate fuel cells (MCFCs), solid oxide fuel cells (SOFCs), alkaline fuel cells (AFCs), phosphoric acid fuel cells (PAFCs), and proton exchange membrane fuel cells (or polymer electrolyte membrane fuel cells, PEMFCs) according to the varied electrolytes used in the cells. Despite the difference in electrolytes, reactions at the anode and the cathode in all FCs are separated, in which ions move through the electrolyte while electrons transport along an external circuit. Among the five types of FCs, the PEMFC is the most promising for efficient energy development given its high power density, small stack size, low cost, low operating temperature, fast response, high energy conversion efficiency and clean environment. It can additionally reduce or eliminate the need for thermal insulation, hydrogen storage and transportation, and safety protection ^{[84, 85, 86, 87, 88]}. As such, the electrocatalytic reactions related to the PEMFCs, such as the formic acid oxidation reaction (FAO), methanol oxidation reaction (MOR), oxygen reduction reaction (ORR), oxygen evolution reaction (OER) and hydrogen evolution reaction (HER), have naturally become the focus of research in recent years. In particular, surface and interface engineering for cost-effective anode or cathode electrocatalysts is the challenge for the emerging and burgeoning investigations on electrocatalysis for fuel cells ^{[89, 90, 91, 92]}.

In this review, some recent advances in the surface and interface engineering of noble metal-based nanostructures as well as their applications in electrocatalysis will be briefly introduced. The key factors to the design, synthesis and assembly of metal nanostructures and their hybrid nanostructures will also be simply outlined and discussed. We will use some of our representative cases in electrocatalysis as specific examples, with an emphasis on surface and interface engineering, to shed light on how to rationally design and fabricate these electrocatalytic nanostructures, and thus to achieve optimization of nanocatalysts in terms of both cost and performance.

To rationally implement surface and interface engineering in the electrocatalyst design, an understanding of the effects of surface and interface on reactions is required. In this section, we highlight a few of the key factors that need to be taken into account when designing metal nanostructures and their hybrid nanostructures.

Heterogeneous catalytic reactions, including electrocatalysis, occur on the surface of catalysts, and as a result, a larger effective surface area should be advantageous for catalytic reactions. In principle, four different approaches can be employed to increase the effective surface area of a catalyst. The first method is to reduce the sizes of the catalyst particles, simply because a decrease in size can dramatically boost the surface-to-volume ratio. As a matter of fact, this is one of the main reasons for the widely used size-control strategy in heterogeneous catalysis. As the size of nanoparticles is reduced to 2 nm, their surface-to-volume atomic ratio can be increased to 80% ^{[93, 94, 95, 96]}. The second typical approach is to fabricate catalysts into thin-layer two-dimensional (2D) nanostructures, such as ultrathin 2D nanoplates or even atomically thin 2D nanosheets, which can significantly boost the surface-to-volume ratio of catalysts. Xie’s group carried out systematic researches on the design, synthesis, assembly and application of 2D ultrathin nanostructures, including metal oxides and metal chalcogenides, which were proven to be useful for improving the specific surface area and active reaction sites of catalysts as well as their flexibility and dispersion ^{[97, 98, 99]}. Despite the fast development of 2D compound materials, the major challenge is to synthesize and stabilize 2D metallic nanostructures, as metal atoms have a strong preference to form close-packed crystal structures in three dimensions. Most recently, a number of research groups have made pioneering contributions to the controllable synthesis of Pd, Rh, Ru, Au, Ag, Pd@Ag and Pt-Cu nanosheets or nanoplates with catalytic and optical applications, which has increased the possibility of fabricating 2D metal nanostructures ^{[100, 101, 102, 103, 104, 105, 106, 107, 108]}. The third way is to increase the roughness of the catalyst nanostructures, and to form irregular surfaces with convex, concave, plicated or cracking rough nanostructures. Such unique structures with morphologies analogous to flowers, dendrites, multi-pods, tapers, stars, starfishes, cactuses and islands can naturally promote the effective surface area for catalytic performance enhancement ^{[109, 110, 111, 112, 113, 114]}. The fourth common means is to manufacture hollow, porous or mesoporous nanostructures such as nanotubes, hollow nanospheres, nanoboxes, nanocages, nanoframeworks and metal-organic frameworks (MOFs). These structures with porosities can not only substantially increase the specific surface areas and number of active reaction sites, but also effectively accelerate the exchange and transportation of species on the catalysts ^{[115, 116, 117, 118, 119, 120, 121, 122]}. In addition to borrowing porous structures from templates, such as MOFs, 3D nanoporous structures of noble metals and their alloys can be directly prepared towards electrocatalytic applications through different strategies including in-situ synthesis, electrodeposition, de-alloying, annealing, templating, galvanic replacement, and modification ^{[123, 124, 125, 126, 127, 128, 129, 130, 131, 132, 133, 134, 135]}. For instance, Ding et al. ^{[128, 129, 130, 131, 132, 133, 134, 135]} have developed 3D nanoporous metal structures with a focus on gold and it alloys.

It is widely recognized that the catalytic properties of nanostructures depend on the type of surface facets ^{[136, 137, 138, 139, 140, 141, 142, 143]}. According to Miller indices {hkl}, for a face-centered cubic (fcc) crystal structure, surface facets can be classified as low-index facets including the three basic {111}, {100} and {110} facets, and high-index facets {hkl} (h ≥ k ≥ l > 0) with at least one of the indices greater than unity (including {hk0} (h > k > 0), {hhl} (h > l > 0), {hkk} (h > k > 0), and {hkl} (h > k > l > 0)) (Figs. 1(a), (b)). Various surface facets have different atomic arrangements, and the surface energy (γ) follows the order of γ {111} < γ {100} < γ {110} < γ {hkl} at high-index facets. As a result, the research community faces a challenge-opportunity dilemma. On one hand, nanocrystals tend to form regular shapes (namely, tetrahedral or octahedral enclosed by {111} facets, or Wulff polyhedra composed of multiple {111} and {100} facets) rather than nanocrystals with high-index facets to minimize surface energy. Thus, the formation of high-index facets on nanocrystal surfaces represents a formidable challenge. On the other hand, nanocatalysts with high-index planes usually exhibit enhanced activities, owing to their higher surface energy and lower coordinated atoms, relative to those close-packed low-index facets such as (111), (100) and (110) planes. Given this dilemma, the research community strives to tailor the surface structures of nanocrystals with low-index facets by controlling the nucleation and growth process through synthesis (Figs. 1(c)-(e)), and at the same time, develop nanostructures with high-index surfaces for more valuable and widespread applications. For example, CO adsorbed on Pt (111) terraces has a higher electrocatalytic reactivity compared to that adsorbed on either (110) or (100) steps ^[144]; it has been reported in 2007 that Pt low-index facets in the electrocatalytic formic acid oxidation reaction follow the activity order of Pt (100) < Pt (111) < Pt (110), and the activities of high-index facets, such as Pt (710), Pt (310), Pt (210) and Pt (520) with a large density of atomic steps and dangling bonds, are significantly higher than those of low-index facets ^[145].

The formation of surface facets is mainly determined by the surface energies, and thus can be altered by introducing foreign species into the synthetic systems, as the specific adsorption ability of species to the metal surface modifies the surface energies. It should be pointed out that the surface structure of a nanocrystal has a strong correlation with its particle size (namely, the length of edge) (Fig. 1(f)). Use of surface capping agents can tune both the facets and sizes of nanostructures in the synthesis. In the synthesis, one can control the growth rates of various facets by selecting appropriate surface capping reagents, regulating the “two rates” (i.e., nucleation and growth rates) and “two mechanics” (i.e., thermodynamic control or kinetic control), as well as inducing anisotropic growth or epitaxial growth to obtain various polyhedrons as well as 1D nanostructures (including nanowires and nanorods) and 2D nanostructures (including nanoplates, nanosheets and nanobelts) deviated from the Wulff polyhedra. Implementing oxidation etching or epitaxial growth in the synthesis, one can also achieve the preparation of novel nanostructures with high-index facets.

Fig. 1 . (a) Triangular diagram showing fcc metal polyhedrons bounded by different crystallographic facets. (b) Atomic models of two low-index {100} and {110} planes as well as five representative high-index {210}, {520}, {310}, {720} and {410} planes viewed from the <100> directions. The h/k value of the {hk0} planes and the atomic length of the (100) terraces increase. (a, b) Reproduced with permission from Scheme 1 and Figure S8 (Supporting Information) of Ref. [137]. (c) Schematic illustration for the reaction pathways that lead to fcc metal nanostructures with different shapes. Reproduced with permission from Ref. [138]. (d) Atomic model of a three-shell cuboctahedron, showing the different types of surface atoms and their coordination numbers. Reproduced with permission from Ref. [139]. (e) Atomic models of cubic and octahedral nanocrystals. (f) Proportions of different types of surface atoms (vertex, edge, and face) as a function of edge length for Au cuboctahedrons. (e, f) Reproduced with permission from Ref. [140].

The concept of “active sites”, which was first introduced to heterogeneous catalysis by Taylor in 1925, has played an important role in explaining and understanding the nature of heterogeneous catalytic reactions ^[146]. The active sites, rather than all sites on the surface of catalysts, are responsible for catalytic reactions ^{[97, 147, 148, 149, 150]}. In general, the active sites of a catalyst are regarded as low-coordinated surface atomic sites, including the low-coordinated atoms at terraces, steps, kinks, edges, vertexes, corners, defects, vacancies, pits, and holes that often favor the catalytic reactions. The number of active sites heavily depends on the features of surface and interface, and has a key effect on the interactions of molecules and ions with nanocatalysts. First, active sites are strongly correlated with the surface roughness of a catalyst. A metal surface generally consists of two types of surface atoms: (1) well-embedded, predominant and stable equilibrated metal surface (EMS) atoms, and (2) protruding, low-population and high-energy metastable metal surface (MMS) atoms. The MMS atoms are prone to undergo reactions with the involved species at low potentials formulating the basis for pre-monolayer oxidation/reduction, which thus often function as surface active sites and play a vital role in surface/interfacial electrocatalysis ^{[151, 152, 153]}. Second, active sites are intuitively associated with the surface facets of a nanocrystal (Figs. 1(d)-(f)) ^{[139, 140]}. The surface of a metal nanocrystal is enclosed by different types of atoms (including face, edge and vertex atoms) whose coordination numbers (CNs) follow the order of face > edge > vertex with the reverse order in terms of catalytic activity. For an fcc metal nanocrystal, the faces of a perfect cube and octahedron are exclusively covered by {100} and {111} facets with square and hexagonal atomic arrangements, respectively. For different well-defined shapes, the variations in surface facets would alter the proportions of various surface atoms with different coordinated states, which also, in turn, affect the catalytic properties. Third, active sites are related to surface defects or vacancies. In principle, breaking chemical bonds to endow atoms with low-coordinated states is beneficial to creating active sites. For instance, defect or vacancy engineering on the basal planes of catalysts can increase the exposure of active edge sites by forming cracks on the surfaces, and as a result, may dramatically enhance electrocatalytic performance ^{[154, 155, 156, 157]}. Last, the number of active sites naturally depends on surface areas and particle sizes based on the discussion above (Figs. 1(d)-(f)).

Obviously, interfaces will be formed where different objects or phases contact each other. Thus, heterogeneous catalytic reactions can essentially be regarded as various interfacial effects. In terms of electrocatalysis, three interfaces need to be presumed and understood: the interface between electrolyte and catalyst, the internal interface between components in a catalyst with hybrid structures, and the interface between catalyst and electrode. In this review, we will limit our discussion to the internal interface in the catalyst, because this interface is a versatile center for engineering catalysts towards heterogeneous electrocatalytic reactions. As illustrated in Fig. 2(a), an interface is naturally formed between two components A and B around which a dynamic triple-phase structure involves the electron path, proton path and solution phase. Engineering the interface can potentially alter interfacial compositions and geometry structures, electronic structures and electron transport, proton transfer and species exchange, crystal structures and dangling bonds, Fermi levels and band alignments, atomic arrangements and coordination numbers, and selective adsorption and desorption, which all may have tunable and synergistic effects on the heterogeneous electrocatalysis occurring at the active sites of catalysts ^{[158, 159, 160, 161, 162, 163, 164, 165, 166]}. For instance, the dual-function active sites in Pt-Pd alloy nanopolyhedrons can enhance the electrocatalytic performance in the methanol oxidation reaction ^[164]; Pt₃Y nanostructures can modify the electronic structure, d-band depth and surface adsorption ability of Pt, leading to improved performance in the ORR ^[165]; Pt-Co alloy nanowires, depending on their compositions and structures, can indeed serve as promising electrocatalysts with enhanced catalytic activity and improved durability for practical electrocatalytic applications ^[166].

Fig. 2 . (a) Illustration for an ideal simple triple-phase boundary structure where electron path, proton path and solution phases are presented together with active sites. (b) Schematic representation of various hybrid configurations: core-shell, subcluster, multishell and embedded structures in the cross-section view. Reproduced with permission from Ref. [167]. (c) Illustrative schemes of some surfactant assemblies and lyotropic liquid crystalline (LLC) phases that can be designed on electrode surfaces. For the top LLC cartoons, blue, white and yellow colors represent water domain, hydrophilic head groups and hydrophobic alkyl chains, respectively. Reproduced with permission from Ref. [181]. (d) Illustration for two types of realistic metal-graphene contact structures. Reproduced with permission from Ref. [191].

Adding a new metal or other element into the electrocatalytically active metal (e.g., Pt and Pd) represents a powerful method to tune the electrocatalytic performance in various reactions. Thus far, the obtained bimetallic or multimetallic nanostructures, including alloys, can be mainly classified as a core-shell structure, subcluster structure, multishell structure and embedded structure according to their atomic locations (Fig. 2(b)) ^[167]. The interface engineering is usually accomplished through structure engineering and/or composition engineering on the components, constructing novel nanostructures for specific catalytic reactions.

In addition to the multimetallic structures, hybrid structures can also be formed between metals and other materials such as graphene and semiconductors. The metal-based hybrid structures (also called metal matrix nanocomposites, MMNCs) represent the most popular development direction in materials science, since the concept was put forward in 1984 by Roy et al. ^[168]. The hybrid structures can also be referred to assembly nanostructure and integrated nanostructure in terms of their preparation process and resulting configuration, respectively. From the viewpoint of synthesis, they can be made by the following approaches.

In-situ growth and coating: In-situ synthesis is a straightforward approach to prepare various metal-based hybrid structures, in which synthetic conditions, such as precursors, should be adjusted according to the specific situation ^{[169, 170, 171]}. In specific cases, coating can be achieved through electrochemical deposition, epitaxial growth and surface modification in aqueous-phase synthesis, or other synthetic methods such as gas evaporation, sputtering, sol-gel synthesis and microemulsion synthesis. Coating has played an effective role in processing metal surfaces and assembling functional composites. Note that the coating approach has often been employed at the early stage of building metal-organic frameworks (MOFs) on electrodes ^{[172, 173, 174, 175]}.

Etching: The function of etching is to intentionally remove and subtract materials from existing structures. Etching has been widely employed to prepare unique well-defined or ordered metal-based hybrid nanostructures including low- or high-index, porous, mesoporous, hollow or framework structures ^{[176, 177, 178, 179, 180]}. Among the obtained structures, the hybrid structures, with porosity and hollow nature, are particularly active to heterogeneous electrocatalysis given their improved solubility and dispersion, larger surface areas as well as increased number of surface active sites and dangling bonds. By building various spatial arrangements of surfactant phases on the electrode surface (Fig. 2(c)), proton transfer and species exchange can be facilitated ^{[181, 182, 183, 184]}.

Loading: Loading metal nanostructures on support materials has played an important role in the fabrication of metal-based hybrid structures in recent years. 2D materials, such as graphene (including N, P and O doping), g-C₃N₄, hexagonal-BN and transition metal chalcogenide nanosheets, which possess unique periodic plane structures and excellent physical properties (e.g., large specific surface area, high tensile strength and elastic modulus, and/or high thermal conductivity and electronic mobility), can act as the support materials for manufacturing such unique hybrid structures. These hybrid nanostructures have been used in electrocatalytic reactions, such as FAO, ORR, OER and HER, and show high activity and high stability in contrast to bare metal nanocrystals. Graphene is a typical representative as a support to construct metal-graphene hybrid nanostructures ^{[185, 186, 187, 188, 189, 190]}. Despite the great progress, there are still many issues that need to be addressed such as geometry optimization for realistic metal-graphene contact. In 2014, Wallace et al. proposed two contact models, side-contact model and end-contact model, for metal-graphene interfaces (Fig. 2(d)). They showed that Pd- and Ti-graphene hybrid structures adopted the end-contact model owing to the formation of Pd- and Ti-carbides associated with C-C bond breaking in graphene ^[191]. This work provides important references and insights for assembling the metal-graphene hybrid structures on electrodes.

In this section, we will overview our recent progress on surface engineering for electrocatalyst design in terms of surface facets. Thus, the discussions of materials design, synthesis and electrocatalytic performance are sorted into low-index facets and high-index facets.

The low-index facets include {111}, {100} and {110} facets with their surface energies in the order of γ {111} < γ {100} < γ {110}. Owing to their different atomic arrangements on a surface, various low-index facets exhibit differentiated activities in electrocatalysis. Certainly, the activities of different surface facets relative to each other are highly dependent on the types of reactions as well as the reaction media (e.g., ORR in H₂SO₄ versus HClO₄). Based on the established synthetic methodology for fcc metal nanocrystals ^{[72, 78, 138]}, we have been able to successfully engineer the low-index facets exposed on the surface of well-defined metal nanocrystals. We have performed studies on the surface engineering of low-index facets from three angles.

The first method involves the acquisition of metal (e.g., Pd) nanocrystals bound by different low-index facets with comparable sizes, which allows us to investigate their facet-dependent effects free from size-effect interference. For instance, we have obtained two types of well-defined Pd nanocrystals (namely, Pd nanocubes enclosed by {100} facets and Pd octahedrons covered by {111} facets) by tuning the reaction kinetics. The sizes of the two Pd nanostructures are precisely controlled by employing oxidative etching or adjusting the reaction time. The study clearly demonstrated that the Pd {100} facet is significantly more active than Pd {111} in electrocatalytic formic acid oxidation ^{[192, 193]}. It is believed that this catalytic performance difference can be ascribed to Pd {100} having a higher ability to activate oxygen species. Similarly, when Pd nanocrystals are implemented in glucose oxidation, the Pd nanocubes with {100} facets exhibit dramatically higher activity than the {111}-octahedrons at similar sizes ^[194].

The second study is to maintain the same low-index facets but tune the particle sizes of metal nanocrystals, providing a platform for investigating size effects. Two approaches can make it possible to tune the sizes of metal nanocrystals (e.g., Pd nanocubes, octahedrons and cuboctahedrons): controlling the number of seeds formed in the nucleation stage by employing oxidative etching ^[195], or manipulating the kinetics of atomic addition by adjusting the feeding rates of precursors with a syringe pump ^[196]. The particle sizes can also be controlled by adjusting the reaction time or performing a seeding process; however, the applications of these two synthetic methods are limited by shape inconsistency and low product yield, respectively. Enabled by this size control, we have been able to observe that the activities of Pd nanocubes in electrocatalytic formic acid oxidation increase with the shrinkage of particle sizes mainly as a result of the boosted surface-to-volume ratios ^[193]. This size control has provided an excellent platform for the investigation of size-dependent catalytic activities in various reaction systems such as the catalytic efficiency improvement in the Heck coupling reaction by reducing the sizes of Pd octahedrons ^[196].

The third investigation involves evaluating the effect of electrical contact between low-index facets and electrodes on electrocatalytic performance. To perform this study, we have successfully achieved the gradual evolution from Pt multipods to Pt multicubes by increasing the coverage of {100} facets on the surface, all of which show significantly higher ORR activities than Pt nanocubes despite their {100} coverage (Figs. 3(a)-(e)) ^[197]. This shape evolution takes place through a simple Ni²⁺-underpotential deposition and galvanic replacement process. The resulting Pt multicube is analogous to an assembly of multiple cubes enclosed by {100} facets but contains a tiny junction area with {311} facets between the adjacent cubic components. In sharp contrast, the Pt multipods are enriched with {311} high-index facets. Such {311}-to-{100} facet evolution is mainly caused by the fast growth on (111) steps in the presence of Ni (Figs. 3(f)-(h)). Despite their high {100} coverage, the Pt multicubes exhibit excellent ORR performance. The half-wave potential (E_1/2) value of Pt multicubes in ORR has been determined to be as high as 0.955 V, which is comparable to 0.957 V of Pt multipods and is 56 mV higher than that of Pt nanocubes (Fig. 3(i)). The mass and specific activities of Pt multicubes reach 0.562 A/mg_Pt and 0.729 mA/cm², which are about 5.1 and 6.7 times higher than those of Pt nanocubes, respectively (Fig. 3(j)). The blank CVs verify that the Pt multicubes have distinct (100) features similar to those for the {100}-enclosed Pt nanocubes (Fig. 3(k)). Pt electrocatalysts should have half-wave potentials (E_1/2) in the order of low-index facets (100) << (111) < (110) << high-index facets (311) for ORR in the HClO₄ system ^{[198, 199, 200]}; however, the Pt multicubes achieve high ORR activity but are enclosed with high {100} surface coverage. This exceptionally high ORR performance is enabled by the low contact resistance of a large flat {100} surface, which has been supported by electrochemical impedance spectroscopy (EIS) measurements (Fig. 3(l)).

Fig. 3 . (a)-(d) TEM and HRTEM images showing the morphological evolution from Pt multipods to multicubes in the presence of Ni2+ cations for 0, 4, 12 and 20 h. (e) TEM and HRTEM images of Pt nanocubes used as a reference sample for the electrochemical measurements. (f) Top and side views of Pt (311) surface consisting of Pt (100) (in red box) and Pt (111) (in green box) surface. (g) Top and side views of Pt (111) and Pt (100) slabs. The atoms in brown boxes are the atoms in the outermost Pt layer, which will be substituted by Ni in the calculations. (h) Dependence of surface formation energy on Ni coverage for Pt (111) and Pt (100). (i) ORR polarization curves and (j) mass activity and specific activity at 0.9 V versus RHE in O2-saturated 0.1 mol/L HClO4 solution at a scan rate of 20 mV/s and a rotation speed of 1600 revolutions per minute (rpm). (k) Blank CV curves recorded for different evolution samples at a scan rate of 20 mV/s and (l) EIS Nyquist plots of the electrodes made from different evolution samples at a bias of −0.25 V in Ar-saturated 0.1 mol/L HClO4 solution. Reproduced with permission from Ref. [197].

Since the 1970s, fundamental investigations on the single-crystal surfaces of bulk Pt have revealed that high-index planes exhibit much higher catalytic activity than low-index planes, such as {111}, {100} and even {110}, which usually act as distinct catalytic active sites for breaking chemical bonds ^{[199, 200, 201, 202, 203, 204]}. After the pioneering work of synthesizing cubic and tetrahedral Pt nanoparticles by El-Sayed and co-workers ^[205], tremendous efforts have been devoted to prepare metal nanocrystals with different shapes by changing metal precursors, reducing reagents, stabilizing reagents, and solvents ^{[72, 137, 141, 145, 206, 207, 208, 209, 210, 211, 212, 213, 214]}. In recent years, intensive studies on electrocatalysis have been performed using metal nanocrystals with well-defined morphologies and high-index facets, which exhibit significantly enhanced catalytic performance relative to their counterparts with low-index facets ^{[137, 209, 210, 211, 212, 213, 214]}. Thus far, the metal nanocrystals bounded by high-index facets can be obtained through various wet-chemistry routes, such as oxidation etching, epitaxial growth, galvanic replacement, templating, de-alloying, electrochemistry, amine or ion-assisting, and manipulating reaction kinetics or thermodynamics. These nanocrystals are classified into two types in terms of the elemental compositions: monometallic (e.g., Pt, Pd and Au) nanocrystals and multimetallic (including bimetallic) nanostructures ^{[137, 209, 210, 211, 212, 213, 214]}. For instance, unique Pt nanocrystals with high-index facets, such as tetrahexahedrons, concave polyhedrons and concave nanocubes, all show improved electrocatalytic activity and stability for formic acid oxidation, ethanol oxidation and ORR, respectively ^{[145, 215, 216]}. Fabricating Pt-based bimetallic or multimetallic nanostructures, such as PtPd, PtAu, PtFe, PtNi, PtCo, PtAg, PtCu, PtPdCu and PtPdAu, enclosed by high-index facets in the forms of alloys or core-shell structures is another key approach to boost the PEMFC electrocatalytic performance ^{[80, 120, 165, 217, 218, 219, 220, 221, 222, 223, 224, 225, 226]}.

Investigating the development of metal nanocrystals bounded by high-index facets, it is straightforward to conclude that high-index facets possess a large number of terraces, steps, kinks, edges, vertexes and corners, and as a result, they have more low-coordinated active sites on the surface to enable high electrocatalytic activities as compared with low-index facets. For this reason, the preparation of high-index facets represents a highly important solution to improve the electrocatalytic performance of metal nanocrystals.

Towards the formation of high-index facets, we have employed oxidation etching to modulate the branched growth of Pt nanocrystals by simply adjusting the concentration of HCl. This modification of etching strength can simultaneously tailor several important parameters in the synthesis, including the structures of seeds, the growth sites on the seeds and the rates of atomic addition on the seeds. As a result, tunable Pt branched structures, including tripods, tetrapods, hexapods and octopods, can be selectively obtained in a single system with identical chemical environment (Figs. 4(a)-(e)) ^[7]. The tunable structures of Pt branched nanocrystals provide a platform to investigate the effect of branches on their electrocatalytic performance in formic acid oxidation. The blank CVs clearly show that Pt hexapods and octopods have higher active surface areas and roughness than those of Pt/C catalysts (Fig. 4(f)). Surface area-specific activities toward formic acid oxidation are in the order of octopods > hexapods > tetrapods > tripods, and the onset potentials are slightly negatively shifted in the same order (Fig. 4(g)). The superior electrocatalytic performance of Pt octopods to their counterparts is ascribed to their high densities of {311} and {211} stepped surfaces.

Fig. 4 . TEM and STEM images of Pt tripods (a), tetrapods (b), hexapods (c) and octopods (d) prepared in the presence 0, 112.5, 225, 675 mmol/L HCl, respectively. (e) Schematic for the shape dependence of Pt branched nanocrystals on HCl concentrations. (f) Blank CVs of electrodes composed of Pt tripods, tetrapods, hexapods and octopods, in 0.5 mol/L H2SO4 solution at a scan rate of 50 mV/s. (g) Surface area-specific activities towards formic acid oxidation at electrodes composed of Pt tripods, tetrapods, hexapods and octopods, in 0.5 mol/L H2SO4 + 0.1 mol/L HCOOH solution at a scan rate of 50 mV/s. Reproduced with permission from Ref. [7].

Alternatively, we can form high-index facets on nanocrystal surface by inducing uneven anisotropic growth. For instance, we have obtained a class of unique five-fold twinned structures—Pd nanotapers by manipulating the anisotropic growth kinetics. The Pd nanotapers enriched with {311} and {611} high-index facets exhibit high electrochemical activities so that they can react with HAuCl₄ through galvanic replacement to form Au islands on the surface, whereas other low-index Pd nanocrystals cannot ^[227]. We have further systematically investigated their structure-property relationships in electrocatalytic formic acid oxidation. The electrocatalytic performance of Pd nanocrystals is observed to increase in the order of octahedrons < nanowires < nanocubes < nanotapers. The results prove that the electrocatalytic activities of Pd nanocrystals are mainly determined by surface facets in the order of {111} < {100} < high-index facets ^[193].

The electrocatalytic reaction also occurs at active sites on the interface as well as the surface of the nanocatalyst. The interfaces in hybrid structures are the locations relevant to electronic structures, charge transport and species exchange, whose design should be crucial to promoting their electrocatalytic performance. In addition, the interfaces can serve as active sites for electrocatalytic reactions in specific cases (particularly for exposed interfaces). In this section, we will discuss interface engineering through three different combinations and mechanisms.

Multimetallic (including bimetallic) nanostructures can be formed with two different configurations: hybrid structures and alloys which both can interface with different compositions, electronic structures, d-band depths, atomic arrangements and functional sites. Development of multimetallic nanostructures has become a widely used approach to improve the electrocatalytic performance ^{[159, 160, 161, 162, 163, 164, 165, 166, 167]}. To reduce Pt usage, our research group has developed a selective-activation method to synthesize Pd-Pt concave nanocubes (Figs. 5(a)-(d)) ^[228]. In this approach, we firstly selectively activate the corners and edges of Pd nanocubes by HCl etching, and then perform Pt epitaxial growth at the activated sites by controlling Pt diffusion and growth rates ^[229]. As a result, the Pd-Pt concave nanocubes with a very limited amount of Pt (only 3.3 wt%) on Pd nanocrystals can be obtained. The Pd-Pt concave nanocubes display a high specific activity of 4.7 mA cm^-2 for electrocatalytic ORR, which is 2.5 and 1.9 times greater than those of Pd nanocubes and Pt/C catalysts, respectively (Figs. 5(e), (f)). The Pd-Pt interface and {730} high-index facets should be the key to the catalytic performance optimization. This method also offers the possibility of fabricating high-performance and low-cost nanocatalysts through controllable growth of noble metals on the surface of base metal nanocrystals by simply modifying the surface and interface conditions.

Fig. 5 . (a) TEM image of the Pd nanocube seeds used for Pt epitaxial growth. TEM (b), STEM (c) and HRTEM (d) images of the PdPt nanocubes obtained through HCl surface activation and Pt epitaxial growth. (e) ORR polarization curves for the four samples in an O2-saturated 0.1 mol/L HClO4 solution with a sweep rate of 20 mV/s and a rotation rate of 1600 rpm. (f) Kinetic current densities (jk) at 0.9 V versus RHE for the four samples normalized with reference to the geometric area of a GC RDE. Reproduced with permission from Ref. [228].

Alloying represents another approach to fabricate multimetallic nanostructures, providing a platform for performance tuning through atomic interfacing. In the synthesis of Cu nanocrystals, Cu nanocubes with small sizes are likely made of Cu₂O rather than Cu ^{[230, 231]}. To improve their resistance to oxidation, we should alloy Cu with more passive metals. For this reason, we have successfully synthesized a series of small-scale and well-defined Cu_xPt_100−x alloy nanocubes with a relatively broad range of compositions and high Cu contents by precisely controlling the stabilizing/capping agents, reducing agents and reaction rates (Figs. 6(a)-(f)) ^[232]. The Cu_xPt_100−x nanocubes, with a consistent size of 10 nm, offer an ideal platform to study the composition-dependent effects on the electrocatalytic CO₂ reduction related to the fuel generation for PEMFCs. The Cu_xPt_100−x nanocubes exhibit improved catalytic activities and gradually positive-shifted onset potentials in the electrocatalysis of the CO₂ reduction reaction with increasing Cu content; however, pure Cu nanocrystals cannot survive from oxidation (Figs. 6(g), (h)). Thus, alloying of Cu nanostructures with Pt can largely improve their chemical stability, and this atomic interfacing may also potentially lower the decomposition activation energy of formate up to 13% ^[233].

Fig. 6 . TEM images of Cu22Pt78 (a), Cu32Pt68 (b), Cu68Pt32 (c) and Cu85Pt15 (d) nanocubes. (e) Schematic illustrating the growth m echanisms for the CuxPt100-x nanocubes prepared in the presence or absence of TDD, taking the comparison of Cu68Pt32 (with TDD) with Cu53Pt47 (without TDD) as an example. (f) XRD patterns of Cu xPt100−x nanocubes with different compositions. (g) Electrocatalytic CO2 reduction performance recorded on CuxPt100−x nanocubes with different compositions benchmarked against pure Pt nanocubes and pure Cu nanocubes with a blank electrode as a reference. (h) Faradaic efficiencies of the different-composition CuxPt100−x nanocubes measured at −1.75 V for 0.5 h, benchmarked against pure Pt nanocubes and pure Cu nanocubes (electrolyte: 0.5 mol/L KHCO3, CO2 purged). Reproduced with permission from Ref. [232].

2D materials may have prominent geometrical, mechanical, thermal and electrical characteristics as well as exhibit excellent solubility, dispersion and adhesion. For this reason, we have used 2D materials, such as graphene, as support materials to fabricate unique metal-based hybrid structures. As such, we have further achieved the optimization of electrocatalytic performance, taking advantage of the merits of components and their synergistic effects.

Based on the synthesis of Pt branched nanostructures ^[7], we have developed a modified protocol for synthesizing Pt highly concave cubic nanocrystals (Pt HCCs) ^[8]. The Pt HCCs with {311}, {100} and {111} facets display excellent activity and durability in electrocatalytic ORR, in sharp contrast to the commercial Pt/C catalysts. In particular, Pt HCCs have achieved a half-wave potential (E_1/2) of 0.960 V, which is comparable to that observed for the most active porous Pt-Ni catalyst reported in the literature ^[234]. To further improve the ORR performance of Pt HCCs, we have assembled the Pt HCCs on reduced graphene oxide (rGO) to form Pt HCCs-rGO hybrid structures (Figs. 7(a), (b)). In the hybrid structures, Pt HCCs enclosed by {311} high-index facets are well supported on the rGO substrate with clear interfaces along the {100} direction (Figs. 7(c)-(g)). The Pt HCCs-rGO hybrid structures exhibit mass activity of −0.847 A/mg_Pt and specific activity of −1.02 mA/cm², which are 5.7-fold and 7.0-fold greater than those of Pt/C, respectively. Impressively, the half-wave potential has been improved to 0.967 V by the integration of Pt HCCs with rGO, which is among the highest ORR E_1/2 values for non-alloy Pt catalysts at present. After durability tests, the Pt HCCs-rGO hybrid structures show electrochemically active area loss (6.8%) in contrast to Pt HCCs (11.2%) and Pt/C (19.5%). The results indicate that Pt HCCs-rGO hybrid structures have very high activity and durability for ORR (Figs. 7(h), (i)). The Pt-rGO interface may have made contributions to enhance electron conductivity, promote catalyst-electrode adhesion as well as suppress the agglomeration and oxidation erosion of catalysts ^{[188, 189, 190]}.

Fig. 7 . (a, b) TEM images of Pt HCCs-rGO hybrid structures at different magnifications. (c)-(g) HRTEM images of a HCC supported on the rGO substrate, focused on each single corner. The insets show the corresponding Fourier-transformed diffraction patterns. (h) ORR polarization curves of Pt/C, Pt HCCs and Pt HCCs-rGO samples before and after 10,000 cycles. (i) Mass activity and specific activity at 0.9 V versus RHE. (h, i) Measurements in an O2-saturated 0.1 mol/L HClO4 solution with a sweep rate of 20 mV/s and a rotation rate of 1600 rpm. Reproduced with permission from Ref. [8].

In parallel, we have also successfully fabricated PtPd nanocages-rGO hybrid structures by a three-step synthetic process, using Pd nanocubes as templates and rGO nanosheets as supports ^[235]. In the synthesis, Pd nanocubes are grown on rGO nanosheets, followed by epitaxial growth of Pt shells and etching removal of Pd cubic cores (Fig. 8). In the hybrid structures, the cubic nanocages that are enclosed by five Pt {100} faces with some pinholes have direct contact with rGO through the other face. The achieved half-wave potential of 0.953 V are shown to be significantly higher than other counterparts in ORR electrocatalysis. The mass activity of 0.534 A/mg_Pt and specific activity of 0.482 mA/cm² are 4.4 times and 3.9 times greater than those of Pt/C, respectively. In comparison with Pt nanocubes and Pd-Pt core-shell nanocubes supported on rGO, we have concluded that both the hollow structure, which increases surface area, and the presence of trace Pd, which prevents Pt dissolution and oxidation erosion, contribute to the ORR performance enhancement, in addition to the roles of the Pt-rGO interface that have been discussed above.

Fig. 8 . (a) Illustration for the three-step synthesis of PtPd nanocages-rGO hybrid structures. TEM and HRTEM (inset) images of Pd nanocubes-rGO (b), Pd-Pt-rGO (c), and PtPd nanocages-rGO (d) hybrid structures. (e) Illustration of PtPd nanocages-rGO hybrid structure for ORR electrocatalysis. (f) ORR polarization curves of as-obtained catalysts. (g) Mass activity and specific activity of Pt-based catalysts at 0.9 V versus RHE. Reproduced with permission from Ref. [235].

Surface charge state is a highly important parameter to catalytic activities, as it can affect molecular adsorption and species-catalyst charge transfer. For instance, the graphene doped with nitrogen has shown enhanced activities in ORR owing to the electronegativity difference between N and C ^{[236, 237]}. Thus far, a set of different approaches have been developed to tune the surface charge state of catalysts including polarization. Originally, the polarization can be induced by the external electric field or ferroelectric substrates ^{[238, 239]}. Most recently, we have demonstrated that interfacial polarization may take place when two materials with different work functions are put together. As long as the travel length is short enough, the charges accumulated by the polarization at the interface can be extended to the surface, modifying the surface charge state of catalysts as a surface polarization. We have implemented this strategy in various reaction systems such as CO oxidation ^[143].

In terms of electrocatalysis, we have demonstrated this concept through the design of Pt-Pd-rGO stack nanostructures for electrocatalytic HER (Fig. 9) ^[9]. The Pt-Pd-rGO nanostructures with a well-defined cubic profile (namely, {100} facets) are fabricated by a two-step synthetic process that involves the in-situ growth of Pd nanocubes and selective coating of Pt cubic shells on the Pd (Figs. 9(a)-(d)). In the hybrid structures, the bottom surface of Pd nanocubes is in contact with the rGO and thus not covered by Pt atoms. The contacting Pd and Pt components have different work functions, which causes a flow of electrons from Pd to Pt to equilibrate the electron Fermi distribution and induces polarization at their interface. Once the distance between the polarized interface and the reactive surface is reduced to the atomic level (1-2 atomic layers), negative charges will accumulate on the surface in favor of HER (Fig. 9(g)). Thus it is important to tune the thickness of Pt shells for achieving effective surface polarization. In the synthesis, we can realize the tunable thickness of Pt shells on Pd nanocubes by simply altering the ratio of added Pd-rGO to Pt precursor. As a result, a series of Pt-Pd-rGO hybrid structures (namely, Pt-Pd-rGO I, II, and III) with increased Pt shell thickness have been obtained, providing an ideal platform for HER investigations. As anticipated, the current densities at the same potential are found to be in the order of Pt-Pd-rGO I > Pt-Pd-rGO II > Pt-Pd-rGO III with a nearly zero overpotential. Meanwhile, both the over-potentials and Tafel slopes are in the reverse order, further verifying that the HER performance can be enhanced with Pt thickness reduction (Figs. 9(e), (f)). This design enables reducing Pt usage while achieving high HER performance.

Fig. 9 . (a) Schematic illustration for the synthesis of Pt-Pd-rGO structures. TEM images of Pt-Pd-rGO I (b), Pt-Pd-rGO II (c) and Pt-Pd-rGO III (d) with tunable Pt thickness. Polarization curves (e) and Tafel plots (f) of Pt-Pd-rGO I, II, and III. (g) Differential charge density by first-principles simulations illustrating the alterations of electron distributions with the Pt thickness. The olive and cyan areas represent the increase and decrease in electron density, respectively. In the simulations, Pd substrates are covered by various layers of Pt atoms to assess the dependence of surface polarization on Pt thickness. Reproduced with permission from Ref. [9].

The electrocatalytic activities of metallic nanocrystals can be facilely tuned by tailoring their surface facets (low-index facets, Fig. 10(a) versus high-index facets, Fig. 10(b)). To further improve their activity and durability, one can form core-shell structures (Fig. 10(c)) or alloys (Fig. 10(d)) by integrating multiple metals together, or/and load the nanocrystals on the supports of other materials (Fig. 10(e)). From the discussions, it is not difficult to recognize that there remain many challenges in this research field. Here, we name some issues that have been brought to our attention from the previous research in the field:

Fig. 10 . (a) Conventional shapes of fcc metals whose surfaces are enclosed by {100} and/or {111} facets. Black and gray colors represent the {100} and {111} facets, respectively. (b) Unconventional shapes of fcc metals whose surfaces are enclosed by high-index facets. The Miller indices {hkl} obey the order h > k > l. (a, b) Reproduced with permission from Ref. [141]. Schematic configurations for the surfaces and interfaces in a bimetallic core-shell structure (c), an alloy (d) and a metal nanocrystal loaded on a nonmetal support (e).

(1) Thus far, the control over surface facets still heavily relies on the use of capping agents. The presence of capping agents on a nanocrystal surface inevitably has various effects on the catalytic activities. In our research, we mainly employ oxygen plasma cleaning and blank electrochemical cycles to remove organic residues. According to the blank CV characterizations, this method should be able to clean the nanocrystal surface; however, it cannot be guaranteed that the nanocrystal surface structure is fully preserved after the cleaning process given the difficulty of performing HRTEM on the samples stripped from electrodes. Thus, it is imperative to develop new methods for facet control. Essentially, oxidative etching should be a relatively clean technique for modifying the nanocrystal surface as it does not introduce organic ligands onto the surface. The remaining problem for this technique is that capping agents are still indispensable for oxidative etching although the same capping agent can be used in an etching system to obtain various nanostructures.

(2) The quality of the interface in hybrid structures should attract sufficient attention from the electrocatalysis research community. This issue has actually been well-recognized in photocatalysis research, as any defects at the interface may become charge recombination centers. In terms of electrocatalysis, charges also need to travel across the interfacial boundary of hybrid structures, in which interfacial defects may scatter or trap electrons. Both syntheses and characterizations would be the focus of research towards the improvement of interface quality.

(3) The sample uniformity and dispersion for hybrid structures are often critical issues to limiting their electrocatalytic performance. In particular, when metal nanocrystals are loaded on 2D nanomaterials, the nanocrystals can hardly be uniformly dispersed on the nanosheets. The aggregation of samples may reduce the number of active sites as well as affect charge transport and species exchange. Improving the integration quality of hybrid structures would be another focus of research, which may arise from surface functionalization and chemical bonding.

We have provided an overview of electrocatalyst design with a focus on surface and interface engineering, using our own researches as examples for demonstration. This review will aid in the engineering of efficient and cost-effective nanocatalysts in electrocatalysis.