Nanostructured intermetallics: from rational synthesis to energy electrocatalysis

Hume-rothery rules

Even the simplest bimetallic systems can be complex due to their different states of the intermediate phase, which include immiscible mixtures and miscible systems. Miscible systems can be further classified into random alloys, in which the atomic positions of the metals are disordered, and intermetallic alloys, in which the atomic positions are ordered. In the 1920s and 1930s, British metallurgist William Hume-Rothery and his colleagues proposed the Hume-Rothery Rules, which is an empirical guideline for determining the intermediate phase state of binary alloy systems^[32,33]. This rule is based on factors such as atomic radius, crystal structure, electronegativity, and electronic structure, and can be a useful tool for designing binary alloys.

According to the Hume-Rothery Rules, when the atomic radii of two elements differ by more than 15%, they are difficult to miscible. For example, atomic radii of Co and Ag are 135 pm and 160 pm, respectively. Regardless of the ratio or temperature of the two metals, it is not possible to obtain a miscible system. Instead, an immiscible solid mixture will form [Figure 2A]. In contrast, Ni (135 pm) has a similar atomic radius to Co, and Ni-Co can form a miscible disordered alloy [Figure 2B]. In addition to atomic radius, the electronegativity difference (Δχ) between the two metals also plays an important role in the formation of intermetallic compounds. A large electronegativity difference strengthens the chemical bond between the two metals, making it more thermodynamically inclined to generate an ordered structure. For example, Ni and Co have similar electronegativities (χ_Ni = 1.92 and χ_Co = 1.88), but Pd has a higher electronegativity of 2.2. Pd and Co can form both random solid solutions and ordered intermetallic compounds, such as Pd₃Co and PdCo, under specific conditions [Figure 2C].

Figure 2. (A-C) Phase diagrams of three binary alloy systems. (A) Co-Ag immiscible system; (B) Co-Ni disordered alloy system; (C) Co-Pd ordered intermetallic compound system; (D) Schematic of the influence of different enthalpy and entropy on Gibbs free energy; (E) Phase diagram and critical transition temperature of typical ordered intermetallic compounds; (F) Comparison of free energies between disordered alloys and ordered intermetallic compounds under A_xB_y stoichiometry in different temperatures.

When Cu family elements form alloys with polyvalent metal elements (such as Zn, Ga, and Ge), the atomic radius and electronegativity difference become secondary factors, and the electron concentration effects become dominant^[34]. The specific phases can be predicted by the average number of itinerant electrons per atom (e/a), as shown in a corresponding between phase and e/a relationship in Figure 3. In other words, the maximum solubility of the solute element will decrease as the e/a increases to ensure that the e/a of alloy is at a specific value. For example, when Zn, Ga and Ge form α-phase alloys with Cu, respectively, the maximum solubility is 38%, 20% and 12%. The e/a of Zn_0.38Cu_0.62 is easily calculated to be 1.38, and the e/a of the remaining two systems is also about 1.4. Different from the first two empirical rules, the Hume-Rothery electron concentration effect rule has been further explained with the development of condensed matter physics. In 1936, Mott and Jones successfully explained this rule by using the nearly free-electron model in the first Brillouin region^[35]. According to this model, the critical e/a [ (e/a)_c ] of the system can be given by the spheroidal electron cloud tangent to the first Brillouin region [Figure 3]. For example, the (e/a)_c values of α-, β- and γ-phase are 1.362, 1.480 and 1.538, which are basically consistent with the experimental results. The electron concentration effect is still widely studied today, for example, by using the total number of electrons as a new descriptor, and by trying to extend the rule to other transition metals.

Figure 3. Schematic of the relationship between electron concentration (e/a) and crystal phase structure, and the First Brillouin Zone and critical electron concentration (e/a)_c values of typical crystal phase structures (α-, β- and γ-phase).

The Hume-Rothery rules are effective for determining the crystal structures of numerous intermetallic compounds. However, these rules have limitations and are not applicable to all binary systems. For instance, the 18-n rule can be employed to ascertain the bond formation of intermetallic compounds composed of transition metals and p-block elements^[36,37]. Meanwhile, the Zintl phase, which is formed by alkali/alkaline earth metals and p-block elements, does not adhere to the Hume-Rothery rules. A typical example is NaTl, which exhibits properties of brittleness, diamagnetism, and low conductivity despite being comprised of two metals. These nonmetallic properties arise from the significant electronegativity difference between sodium and thallium. After thallium accepts an electron from sodium, it still fails to achieve an octet (only four electrons), resulting in the formation of a regular tetrahedral network similar to carbon atoms in diamonds. This is so-called the Zintl-Klemm counting scheme^[38-40], in which the anion of the Zintl phase forms polyanionic substructures resembling elements with the same number of outermost electrons, such as white phosphorus (tetrahedral P₄) and sulfur (circular S₈). However, these compounds are infrequently utilized as electrocatalysts due to their poor electrical conductivity, and thus will not be elaborated upon in this review.

Thermodynamic analysis and phase diagram

The Hume-Rothery rules is an empirical guideline used to predict the conditions under which metallic alloys will exhibit miscibility. The Gibbs free energy of mixing is a key determinant in judging whether the system can spontaneously become miscible. We also take the binary system as a simple example, and the Gibbs free energy change in the mixing process (ΔG_bulk) of the bulk system can be written as follows.

where ΔH is the enthalpy change of mixing, T is the absolute temperature, and ΔS is the entropy change of mixing. It is well established that the entropy of the system must increase during the mixing process, meaning that ΔS is always positive. The sign of ΔH depends on whether the formation of new bonds in the system is exothermic or endothermic. If the ΔH is negative, then the ΔG will be less than zero at any temperature leading to a miscible system, such as the Co-Ni system. However, when ΔH is positive, the effect of temperature should be considered [Figure 2D]. The influence of TΔS on the Gibbs free energy of the system is relatively small at lower temperatures, which is because the enthalpy change is generally 2-3 orders of magnitude larger than the entropy change. For example, in the Co-Ag immiscible system, the ΔH is considerably larger than zero, so the two metals are still not miscible at 2000 K. This formula can also be used to guide the transition process between disordered alloys and ordered intermetallic compounds. When ΔG = 0, the system is in a two-phase mixed equilibrium state, and the temperature is called critical phase-transition temperature (T^c)^[7]. In general, the enthalpy change of the disorder-to-order transition is negative, as intermetallic compounds typically have higher bond energies, making the ordered system more thermodynamically stable. The entropy change of this process is also negative, as the atoms in the system become more ordered. Thus, low temperatures favor the formation of ordered metal compounds, while high temperatures favor the formation of disordered alloys [Figure 2E and F].

In practical systems, it is critical to take into account the effect of surface energy, which is not present in ideal bulk systems. The smaller the particle size, the larger the influence of surface energy on the overall energy. As a result, the equation for Gibbs free energy change in the mixing process needs to be modified to consider surface energy^[41,42]:

where A is the surface area of the particle and Δγ is the change in surface energy. When bulk system and nanoparticle system reach critical phase-transition temperature, ΔG_bulk and ΔG_nano are zero. Combined with equations 1 and 2, the T^c of the two systems can be written as equation 3^[28].

where ΔH_v is volume-specific bulk enthalpy. In the disorder-to-order transition, it is clear that ΔH_v is negative, the same sign as ΔH. And since the average bond energy is higher on ordered surfaces, Δγ is also negative. According to equation 3, as the particle size of nanoparticles decreases, the specific surface area (A/V) increases. The nanoparticle system should have a lower critical phase-transition temperature than the bulk system^[43,44]. A typical example of this conclusion is the discovery by Alloyeau et al. that CoPt nanoparticles have a T^c of 175-325 ℃ lower than bulk materials^[45]. Moreover, the exposed surface of the nanoparticles also influences the energy of the system, as the surface energy of different crystal faces can vary. In conclusion, the Gibbs free energy and the critical phase transition temperature of the system will be impacted by the composition, particle size, shape, and other factors.

Although thermodynamics at a given temperature can provide an accurate indication of whether a system is miscible at a given temperature, it is still not intuitive. The phase diagram can visualize the state of the reaction system and provide a preliminary reference for the stoichiometric ratio and temperature required for the synthesis of metal compounds. The Gibbs phase rule is the basis for phase diagrams, a formula introduced by Josiah Willard Gibbs in the 1870s^[46]. This rule holds that the state of a system is entirely described by its thermodynamic parameters (e.g., temperature and pressure) at equilibrium.

where C is the composition fraction of the system, P is the number of phases, and F is the number of degrees of freedom, that is, the number of independent variations of the system. For a binary system with C = 2, the number of degrees of freedom is given by F = 4 - P. If only one phase is present in the system (P = 1), then F is three, meaning that the system’s pressure, temperature, and composition can vary independently.

H. W. Bakhuis Roozeboom applied this rule and produced the first iron-carbon phase diagram in 1900^[47]. The phase diagram can intuitively show the equilibrium state of the system at any temperature. Additionally, the phase diagram can be used to predict the behavior of the high-temperature system during cooling. Moreover, intermetallic compounds with fixed composition are represented as narrow regions or vertical lines in the phase diagram. This can make it challenging to experimentally plot phase diagrams. By combining theoretical simulation and thermodynamic analysis, phase diagrams can be accurately mapped and the prediction of new intermetallic compounds becomes possible. The thermodynamic parameter, pressure, which is more difficult to control in experiments, can also be easily obtained from computational simulations. For example, Clarke et al. utilized DFT calculations to create the temperature-pressure phase diagram of the Cu-Bi binary system and successfully synthesized superconducting Cu₁₁Bi₇^[48]. Theoretical prediction combined with high-throughput screening and machine learning methods has played an important role in predicting and characterizing the properties of intermetallic compounds^[49,50].

Kinetics analysis

Thermodynamics and phase diagram can provide crucial guidance for the synthesis of metallic compounds, but a range of kinetic factors need to be considered in the actual reaction process. While thermodynamic conditions may be favorable during synthesis, the presence of kinetic energy barriers can result in the formation of metastable structures. In fact, the successful preparation of ordered intermetallic compounds is dependent on balancing the thermodynamic driving force and the kinetic energy barrier, and the system must cross the activation energy barrier to reach the most stable thermodynamic state^[51] [Figure 4A].

Figure 4. (A) Schematic of the free energy of the disordered structure converted to the ordered structure; (B) Schematic of nucleation rate, diffusion rate and total reaction rate as a function of temperature.

The Johnson‐Mehl‐Avrami‐Kolmogorov (JMAK) theory describes in detail the dynamics during the formation of ordered phases^[52-54]. Here we will examine the two fundamental processes in the transformation from disorder to order, the formation of new phases (nucleation) and the rearrangement of local atoms (diffusion). According to the Arrhenius equation, the reaction rates of the two processes are as follows.

where A₁ and A₂ are the respective preexponential factors, k_B is the Boltzmann constant, ΔG is the Gibbs free energy change in phase transition, E_a is the activation energy of atomic diffusion, and T is the absolute temperature. For the reaction to occur, ΔG must be negative, while the remaining parameters (A₁, A₂, k_B, E_a, and T) are all positive. Therefore, k₁ and k₂ have diametrically opposite tendencies with respect to temperature. The reaction rate of k₁ decreases while that of k₂ increases with increasing temperature. The overall transition rate depends on the product of k₁ and k₂, so there exists an optimal transition temperature. In other words, the curve of the total rate as a function of temperature [Figure 4B] shows a volcanic pattern^[55,56].

Particle size also influences the dynamics of the "disorder-to-order" process. Smaller particles will reduce the thermodynamic driving force and thus cause a decrease in k₁, as described in Equation 2. However, diffusion is easier in small particles. Therefore, there may also be an optimal range of particle sizes that is most favorable for the formation of ordered intermetallic compounds. In addition, the hole concentration and the interatomic bond strength also influence the reaction rate. Generally speaking, increasing the concentration of vacancies effectively facilitates the phase transition process because the activation energy of vacancy diffusion is lower^[57]. In summary, several factors including temperature, composition, particle size and shape, and hole concentration must be considered to balance the thermodynamic driving force and kinetic energy barrier during the synthesis of ordered intermetallic compounds^[58-63].

Thermal annealing

Thermal annealing, which heats random alloys at high temperatures for a sufficiently long time, is the simplest and most straightforward way to generate intermetallic compounds [Figure 5A]. According to thermodynamic and kinetic analyses, random alloy structures are more easily generated than intermetallic compounds when the reaction temperature is maintained below the disordered-to-ordered transition temperature. At elevated temperatures, thermal vibration of metal atoms is promoted, facilitating the diffusion of the metal and making it easier for random alloys to cross the barrier to form ordered compounds^[7,28]. Using this method, a variety of bimetallic and trimetallic compounds, as well as intermetallic semiconductors and superconducting nanocrystals, have been synthesized. However, the thermal annealing method also has its disadvantages: higher annealing temperature will promote Ostwald ripening and aggregation, which results in the increase of nanocrystal size and uneven size distribution^[70]. At the same time, the catalytic performance of intermetallic compound nanocrystals is reduced. Therefore, it is necessary to control the annealing temperature and annealing time, and to select appropriate support and coating layers to prevent aggregation of nanocrystals at high temperatures.

Figure 5. Schematic of the main synthesis methods for intermetallic compounds, including (A) thermal annealing; (B) metallothermic reduction; (C) wet chemistry method; and (D) seed-mediated growth.

Direct annealing is currently the most widely used method to generate ordered intermetallic nanocrystals. In this process, random alloys obtained by the liquid phase method were used as primary intermediates, and then they were converted into atomically ordered intermetallic nanocrystals by heat treatment in an atmosphere or vacuum. Due to the relatively high kinetic barriers involved in atomic ordering, annealing at high temperatures (usually > 500 °C) and/or long times is required. For example, Abe et al. converted atomically disordered Pt₃Ti nanoparticles into larger atomically ordered Pt₃Ti nanoparticles by annealing above 500 °C^[71]. The authors annealed Pt₃Ti in vacuum at various temperatures and analyzed the PXRD profiles of the products, and they found that when the annealing temperature exceeds 550 °C, the PXRD of the annealed product shows the superlattice peak of Pt₃Ti in the ordered Cu₃Au-type structure. When the annealing temperature was further increased to 600 °C, the intensity of the above lattice peaks was improved. The results indicate that the annealing temperature is an important parameter in adjusting the order degree of the prepared nanocrystals. However, higher annealing temperatures inevitably lead to a sharp increase in particle size. Atomically disordered Pt₃Ti nanoparticles with a particle size of 3 ± 0.4 nm were annealed in vacuum at 600 °C and ordered Pt₃Ti nanoparticles with a particle size of 37 ± 23 nm were obtained. Chi et al. synthesized a series of Pt₃Co nanocrystals with different atomic arrangements during in situ annealing and observed five distinct stages of element rearrangement on the surface of nanocrystals by scanning transmission electron microscopy^[72]. They found that when the sample temperature was raised to 600 °C and held for 20 min, the ordered Pt₃Co phase began to nucleate on {001}. By prolonging the annealing time at the same temperature, it is advantageous to form fully ordered phases with similar morphology. Therefore, careful selection of annealing temperature, annealing time, and annealing atmosphere plays a crucial role in promoting the degree of order, size, and shape of the final product.

Support-assisted annealing. Unlike direct annealing, the support-assisted annealing method deposits nanocrystals on stable supports with a larger surface area. The catalyst support can prevent unwanted aggregation of the active catalyst during thermal annealing^[73]. In addition, electron rearrangement occurs at the interface between nanocrystals and support, which can increase the electron density of nanocrystals and improve the performance of catalysts. Carbon is one of the most common carriers widely used in fuel cells and batteries^[74-76]. Yang et al. showed that Pt loading on sulfur-doped carbon support produced strong platinum-sulfur bonds, allowing small platinum alloy nanoparticles (< 5 nm in diameter) to be stabilized at temperatures up to 1000 °C^[77]. They use sulfur-doped carbon as a support for synthesized intermetallic libraries of small nanoparticles consisting of 46 combinations of platinum with 16 other metal elements. Other carbon supports, such as porous carbon, porous oxide/carbon composite and graphene, are also commonly used to synthesize various intermetallic nanocrystals. Ji et al. modified the surface of ordered mesoporous carbon (OMC) with sulfur and used the surface-modified OMC-S as catalyst support to prepare OMC-PtBi nanocrystals embedded in the porous structure of OMC^[78]. Because the OMC has a connected pore structure, a high pore density and a specific surface area, and the sulfur has a strong affinity for metals, the metal precursor in solution can be adsorbed and combined into the pores of OMC, and the particle growth can be limited in the reduction process, to obtain ultra-fine dispersed metal nanocrystals. The resulting nanocrystals have a size of about 2-3 nm and a uniform size distribution. In addition, they also synthesized other intermetallic PtM (M = Pb, Pd, Ru) catalysts using this method.

Alternatively, intermetallic can be prepared at the interface of metal and support by strong metal-support interactions^[79]. Generally, noble metal nanocrystals are carried by oxide support and converted into intermetallic compounds by reduction at high temperatures. For example, Pd NPs supported on ZnO, Ga₂O₃, or In₂O₃ were converted into intermetallic ZnPd, Ga₂Pd₅, and InPd. Bernal et al. reduced Pt/CeO₂ in H₂ atmosphere at 200-950 ℃ and observed the structural transformation using HRTEM^[80]. They found that Pt nanocrystals loaded on CeO₂ were converted to the intermetallic compound CePt₅ after annealing at high temperatures. The method of synthesizing intermetallic compounds by the strong interfacial interaction between metal and oxide carrier surface usually requires a long annealing time at high temperatures to transform into an ordered structure, so it is still challenging to apply this method to the synthesis of intermetallic compounds nanocrystals.

Coating-assisted annealing. By coating the nanocrystals with polymers, oxide protective layers (SiO₂, MgO) and KCl substrates, the diffusion between nanocrystals can be eliminated, thus effectively alleviating agglomeration and sintering during thermal annealing and precisely controlling the size of nanocrystals^[81]. Fujimori et al. coated the surface of disordered face-centered cubic (fcc)-FePt with SiO₂ and subsequently thermally annealed to obtain ordered L1₀-FePt nanocrystals^[82]. SiO₂ can limit the diffusion of Fe and Pt atoms during annealing, and the resulting material has an average particle size of 6.4 nm. Kim et al. used a similar approach coating the disordered fcc-FePt with a MgO protective layer (FePt/MgO), and heat-treated it at 800 °C to obtain an intermetallic FePt with an ordered face-centered tetragonal (fct) structure^[83]. Song et al. developed a small molecule-assisted impregnation approach to realize the general synthesis of a family of Pt-based intermetallics^[84]. They utilized molecular additives containing functional groups as ligands to coordinate Pt (IV) in impregnation, and then thermally converted to intermetallic nanoparticles coated on a hybrid atom-doped graphene layer. The method realized the general synthesis of 18 binary intermetallics, including Pt₃M (M = Ti, V, Cr, Mn, Fe, Co, Al, Ga, In, Ge, Sn), PtM (M = Mn, Fe, Co, Ni, Cu, Zn) and PtCu₃. Other inorganic salts such as NaCl and KCl were also able to stabilize nanoparticles during high-temperature annealing. However, these salts do not form a tight shell around each nanoparticle and are less efficient in stabilizing nanoparticles to prevent aggregation/sintering^[85,86].

Defect-assisted annealing. The complete transition from disorder to order can also be achieved by introducing defects to lower the atomic diffusion barrier and promote mutual diffusion between metal atoms^[87]. At present, defects are mainly introduced into the crystal interior by the generation of oxygen vacancies or by the introduction of a third element^[88]. Sun et al. prepared disordered fcc-FePt-Fe₃O₄ nanocrystals by controlling the decomposition of Fe(CO)₅ and reduction of Pt(acac)₂, and obtained fully ordered fct-FePt nanocrystals by subsequent thermal annealing^[89]. In the synthesis process, fcc-FePt NPs were first synthesized and then oxidized in air to obtain dumbbell fcc-FePt-Fe₃O₄ NPs, which were surrounded by a polycrystalline Fe₃O₄ shell. Then defects are generated when Fe₃O₄ is reduced to Fe, and the dumbbell structure ensures that Fe/Pt is easy to diffuse to form a fully ordered fct structure^[90]. Wang et al. used a bifunctional core/shell Pt/NiCoO_x as a precursor system to prepare sub-6 nm monodisperse ordered L1₀-PtNi_0.8Co_0.2 nanocrystals. NiCoO_x can provide abundant oxygen vacancies to promote the diffusion of Pt/Ni/Co atoms and act as a protective shell to inhibit the sintering of nanocrystalline during thermal annealing^[91].

Metallothermic reduction

Thermal reduction is a method of producing intermetallic compounds by reducing a metal source at high temperatures with a reducing agent [Figure 5B]. Usually, metal oxides, halides, or sulfides are used as metal sources, reducing gases (H₂, B₂H₆), and active metals (K, Mg, Al, Sn) as reducing agents. Due to the wide variety of reducing agents available in this method, many typical reactions have been developed, including solid-state metathesis reactions and metal thermal reduction reactions, to achieve the simple, rapid or universal synthesis of intermetallic compounds. Recently, our group has prepared a series of intermetallic borides and silicides by thermal reduction^[92,93]. Our group employed the boron thermal reaction between boron powder and metal oxide under KCl-NaCl molten salt conditions to prepare phase-pure borides. Twelve single metals MB₂ (M= Ti, V, Cr, Mn, Zr, Nb, Mo, Hf, Ta, W, Re and Ru) were prepared by this method at relatively low temperatures (900-1000 ℃)^[94]. The twelve metal diborides adopt four types of crystal structures. Their conspicuous structural difference lies in the different two-dimensional covalent boron sheet subunits. Later, we developed a solid-state metathesis reaction using magnesium diboride (MgB₂) and anhydrous metal chloride as feedstocks to produce intermetallic borides^[95]. Due to the formation of a thermodynamically stable byproduct (MgCl₂), only a lower heating temperature is required. In addition, additional reducing agents can reduce residual boron impurities or other unwanted boride phases in the target product^[23]. Our group reported the synthesis of mono-borides with Mg as an additional reducing agent to reduce MCl_x under relatively mild conditions^[96].

Wet chemistry method

Compared with the thermal annealing method, the wet chemical method is easier to control the morphology and size of metal compounds [Figure 5C]. In wet chemistry, intermetallic compounds are synthesized by direct reaction with the addition of precursors, reducing agents, capping agents, and surfactants in solution. Because high-temperature annealing is not required, particle agglomeration during annealing is avoided. However, due to the limitation of the boiling point of the solvent, the reaction can only be carried out at a low temperature, so it is necessary to adopt a suitable strategy to reduce the barrier of the transition from disordered alloys to ordered intermetallic compounds. The following two strategies can be adopted for the preparation of intermetallics^[97].

(i) Adding halide ions (Cl⁻, Br⁻, I⁻) or strong reducing agents (ascorbic acid (AA), tetrabutylammonium bromide (TBAB), NaBH₄) to the reaction solution to slow down the crystal growth rate. Rong et al. synthesized surface defect-rich Pt₃Sn using Pt(acac)₂ and SnCl₂·2H₂O as metal precursors, polyvinylpyrrolidone (PVP) as surfactant, and N, N-dimethylformamide (DMF) as solvent and reducing agent^[98]. During synthesis, Cl⁻ released from SnCl₂ slows crystal growth and cooperates with O₂ to etch the nanocrystals. According to the growth-etching synergetic growth mechanism reported by Tilley, the corrosion process proceeds synchronously with the growth process in the presence of higher concentrations of corrosive agents. At the beginning of the reaction, the etching process is slow because the concentration of Cl⁻ in the solution is low. However, with Sn atoms reduced, more Cl⁻ is released into the solution, which leads to a faster etching rate and more defect formation. The newly generated Sn atoms can diffuse from the surface into the Pt-rich lattice to form PtSn solid solution. Liao et al. synthesized novel PtBi nanoplatelets (NPLs) and PdBi nanowires (NWs) with controlled shape, size, and composition in the presence of oleylamine (OAm) and NH₄Br^[99]. Using a strong reducing agent, the simultaneous reduction of two metal precursors is also a method of synthesizing intermetallic compounds. Maksimuk et al. used TBAB as a reducing agent to simultaneously reduce Pt(acac)₂ and Pb(acac)₂ to synthesize ordered metal-interstitial PtPb nanorods^[100]. Cable et al. reported the synthesis of a series of tin intergeneric compounds using NaBH₄ as a strong reducing agent^[101].

(ii) By seed-mediated synthesis, metal nanocrystals with well-defined size, shape and composition are first synthesized, and then the second metal is diffused into the synthesized seeds by diffusion growth^[102] [Figure 5D]. This synthesis strategy was pioneered by Cable and Schaak, who first reported the generation of intermetallic M-Zn nanocrystals (M = Au, Cu, and Pd) by seed-mediated diffusion growth using zero-valent Zn precursors in a hot organic amine solvent. Samanta et al. synthesized ultra-small monodisperse intermetallic NiZn nanoparticles by converting Ni nanoparticles into an ordered alloy using diethylzinc [Zn(C₂H₅)₂] as precursor^[103]. During the transformation, the particles retained their morphology, and the estimated particle size of the NiZn NPs was ~4.5 nm. The seed-mediated growth approach enabled the preparation of a variety of noble (gold, silver, platinum, and palladium) and non-noble (tin, plumbum, and copper) intermetallic compounds, and can realize the control of morphology. An excellent example was demonstrated in the solution-phase synthesis of core/shell PbPt/Pt with PbPt core being in hexagonal structure^[104]. The core/shell NPs were made by using Pt(acac)₂ and Pb(acac)₂ as the metal precursors, oleylamine (OAm)/octadecene (ODE) mixture as solvents and surfactants, and AA as the reducing agent. Under the reaction condition, Pb was first reduced and Pt was deposited on the Pb seeds, followed by controlled Pt diffusion into the Pb seeds. With the Pb/Pt ratio controlled at about 1:1, hexagonal PbPt was obtained. Meanwhile, the authors found that using AA as a reducing agent was the key to forming a well-organized Pt atomic layer, because AA can be used as a weak acid to remove Pb during synthesis, allowing Pt atoms to diffuse and rearrange at higher temperatures.

Using Au nanocrystals as precursors, Chen et al. prepared monodisperse CuAu and Cu₃Au nanocrystals by seed diffusion method^[105]. They prepared Cu₃Au and CuAu using 6.3 nm and 8.5-9.5 nm Au seeds at different temperatures, respectively. According to powder X-ray diffraction (XRD) patterns of the product, Cu₃Au intermetallic compounds of cubic order were formed at 300 °C when the Cu/Au atomic ratio was 3:1. The intermetallic compound CuAu was obtained at 280 °C when an equal molar amount of raw material was used. In addition, Clarysse et al. report the development of a colloidal synthesis method based on amalgamation of monometallic nanocrystal seeds with low melting point metals, and have used this method to obtain uniform intermetallic nanocrystals of Au-Ga, Ag-Ga, Cu-Ga, Ni-Ga, Pd-Ga, Pd-In and Pd-Zn compounds^[106]. They showed that besides Ga, other low melting point metals (In and Zn) could be used to synthesize intermetallic compounds.

Although many methods have been used to synthesize intermetallics, they have sufficiently different advantages and disadvantages. For example, metallothermic reduction can synthesize highly-crystalline intermetallics that are difficult to obtain by wet chemical methods, such as silicates and borides, but the size of the as-synthesized products is still relatively large, dozens or hundreds of nanometers^[93]. Wet chemical method can accurately control the morphology and size of intermetallics, but may involve some expensive metallic reagents [e.g., Pt(acac)₂, Pd(acac)₂] and auxiliary solvents (e.g., oleylamine, benzyl alcohol), resulting in high synthesis cost and difficulty in large-scale synthesis^[97]. A significant trend for large-scale production of intermetallic catalysts is the development of green and low-cost synthesis of small-size intermetallic nanoparticles. In addition, the current synthesis of intermetallics mainly relies on trial‐and‐error approaches, which are inefficient and tedious. More research is needed on general methods and design principles for the synthesis of intermetallic phases.

Intermetallic electrocatalysts for HER

Water electrolysis is one of the promising technologies that can produce clean fuel hydrogen on a large scale. As a key half-reaction in water electrolysis, the HER at cathode usually requires an efficient electrocatalyst to reduce the reaction barrier and energy consumption. Noble metal Pt is readily recognized as the most active HER catalyst, but its high cost and scarcity hinder the wide commercialization of water electrolysis technology^[110]. Alloying Pt with non-noble metals is an effective way to reduce Pt content and improve the intrinsic activity of electrocatalysts. In recent years, researchers have developed a series of Pt-based intermetallics, such as Pt₃Ti^[111], PtFe^[112], PtNi^[113], Pt₃Co^[114], and so on, whose ligand effects and stress effects between Pt and metal elements can regulate the surface adsorption properties, thus achieving the purpose of enhancing HER activity. For example, Zhang et al. reported an IrMo intermetallics supported on carbon nanotubes by a hydrogel-freeze-drying method, whose specific activity for HER was 9.5 folds that of Pt/C^[115]. The stably-adsorbed OH spectator on the Mo site optimized the surface chemical environment and electronic structure of the catalyst, resulting in superior hydrogen adsorption energy and accelerating the water dissociation kinetics.

Developing non-Pt HER catalysts with Pt-like activity has been an important research direction in recent years. Especially, some noble metal-based intermetallics, such as Ru (RuB₂, RuB, RuSi, RuGe, etc.),^[93,116,117] Pd (Pd₂B, PdSi, etc.)^[118,119], Rh (RhB, RhSi, etc.)^[120] and Ir (IrMo, IrSi, etc.)^[121-123] have attracted wide attention. Our group studied a series of light element (e.g., Si, B)-containing intermetallics (silicide and boride), and found that the hybridization between sp orbital of light elements and d orbital of transition metals dominates the electronic structure characteristics [Figure 6A]^[96,118,124]. This sp-d orbital hybridization can reduce the hydrogen adsorption energy of metal surface and regulate the catalytic activity for HER [Figure 6B]. Based on these understandings, several non-Pt HER electrocatalysts with Pt-like activity, such as RuB and RuSi, have been confirmed by experiments and theoretical calculations [Figure 6B]^[96,118,124]. Inspired by these works, the discovery of new synthesis methods and novel electrocatalysts based on borides and silicide intermetallics has become a hot topic in recent years^{[120,121,125,126]}.

Figure 6. (A) Calculated pDOS of Pd and Pd-B alloy models; (B) A volcano plot showing the ΔG_H* values of Pd-B alloy models containing different subsurface B concentrations as a function of their surface d-band centers. The inset shows the structural model of Pd-B systems with different subsurface B concentrations; (C) Schematic illustration of ordered interstitial B atom in Pd₂B electrocatalysts and the optimal near-surface electronic structure for HER^[118]; (D) Schematic illustration of the synthesis of ordered mesoporous PtZnM trimetals^[128]; (E) Scanning electron microscopy (SEM) image of ordered mesoporous PtZnCo trimetals.

Compared with the disordered alloys, the intermetallic structures would not only enhance the catalytic activity, but also possess greater structural stability during HER. A remarkable example is Pd-B alloys reported by our group [Figure 6C]^[118]. Controlled preparation of intermetallic Pd₂B and disordered Pd-B alloys was achieved by temperature control in a metallothermic reduction reaction. Despite a similar stoichiometric ratio, the intermetallic Pd₂B exhibits significantly higher catalytic activity for HER than the disordered Pd-B alloy, because the interstitial boron ordering of the former yields an optimal near-surface electronic structure. Moreover, the catalytic and structural stability of intermetallic Pd₂B is significantly enhanced relative to the disordered Pd-B alloy. (i) The leaching amount of B from Pd₂B (1.4%) is less than that of disordered Pd-B alloy (13.3%) during HER. (ii) Pd₂B possesses excellent catalytic stability for more than 7 days, while the disordered Pd-B alloy suffers severe deactivation.

In order to further reduce the amounts of noble metals and optimize catalytic performance in noble metal-based intermetallics, several important strategies have been developed. One strategy is to load noble metal intermetallics onto a suitable support, for example, IrGa supported on nitrogen-doped reduced graphene oxide,^[127] Pt₃Ti nanoparticles supported on Ti₃C₂T_x^[111]. In these supported catalysts, in addition to reducing the amount of noble metals, the support effects can improve HER catalytic performance through metal-support electronic interaction. Another effective strategy to promote the catalytic performance of intermetallics is accurate nanostructure design. For example, Liu et al. prepared a series of ordered mesoporous PtZnM trimetals (M= Sc, V, Cr, Mn, Fe, Co, Ni, Cu or Ga) through a concurrent template route [Figure 6D and E]^[128]. Among them, the PtZnCo trimetals display a higher mass activity (0.36 A mg_pt^-1 at -0.05 V) for HER than that of commercial Pt/C (0.22 A mg_pt^-1) and strong catalytic stability for 50000 cycles, attributing to the large specific surface area and abundant low coordination active sites.

The ultimate pursuit of HER electrocatalysts is to develop highly active, noble metal-free catalysts. Intermetallic electrocatalysts based on Mo, W, Fe, Co, Ni and other non-noble metals have been widely studied, such as Gd₂Co₁₇^[129], Co₇Mo₆^[130], NiZn^[131], and Co₃W^[132]. Although an enormous number of non-noble metal intermetallic electrocatalysts have been reported, it should be pointed out that their intrinsic activities for HER are usually 1-2 orders of magnitude lower than those of Pt-based catalysts^[133]. Their high catalytic activities are often achieved at high loadings or/and large surface areas of catalysts to compensate for relatively worse intrinsic activity. Moreover, these non-noble metal electrocatalysts face a larger challenge in long-term catalytic stability relative to Pt/C catalysts.

Intermetallic electrocatalysts for ORR

ORR is an important half-reaction in fuel cell technology. The reaction largely determines the overall efficiency of fuel cells, because of multiple proton-electron transfer and sluggish reaction kinetics^[134]. Similar to HER electrocatalysis, noble metal Pt is traditionally the most active electrocatalyst for ORR^[43]. Alloying is an effective strategy to reduce the amount of Pt and increase the catalytic activity of ORR^[135]. The strain and ligand effects in Pt alloys can influence the electronic structure and surface adsorption behavior, leading to greater catalytic activity for ORR. However, Pt alloys often suffer from poor stability, which is assigned to corrosion and leaching of transition metals during the catalytic process. The dissolved metal cations hinder proton conduction and accelerate the decomposition of the membrane and catalytic surface. Unlike the disordered Pt-M alloys, Pt-based intermetallics exhibit enhanced chemical stability in operating conditions due to ordered atomic arrangement and strong bonding interactions. The Pt-based intermetallics are expected to be new-generation ORR catalysts to realize low cost, greater catalytic activity and long-term durability simultaneously.

The exploration of Pt-based intermetallics as ORR electrocatalysts has progressively covered Pt-transition metal alloys (e.g., Pt₃Ni, Pt₃Co, Pt₃Mn, Pt₃Cu),^[136-138] Pt-rare earth alloys (e.g., Pt₃Y, Pt₅La)^[139-141], Pt-alkaline earth alloys (e.g., Pt₅Ca and Pt₅Sr)^[142] and so on. A typical class of Pt-based intermetallic catalysts for ORR is binary Pt₃M (M = 3d-transition metals). In 2007, Stamenkovic et al. demonstrated a volcano-type relationship between the surface electronic structure and ORR activity for Pt-transition metal alloys Pt₃M (M=Ni, Co, Fe, Ti, V)^[143]. The Pt₃Ni and Pt₃Fe are found to be at the very top of the volcano plot, exhibiting better ORR activities compared with pure Pt. In 2014, Yang et al. synthesized hollow Pt₃Ni nanoframes with open structures and high surface areas, which displayed 36 times higher mass activity than Pt/C catalysts^[144]. Subsequently, some doping and heterostructure strategies are developed to realize the component and structure optimization of Pt₃Ni catalysts for ORR, such as Mo-doped Pt₃Ni octahedron, Ni₃C@Pt₃Ni core-shell nanoparticles^[145].

Constructing of ultrathin Pt shell on the surface of Pt-based intermetallic nanoalloys is an important strategy to improve the ORR activity. The formation of core-shell structure (such as PtPb/Pt, PtCo@Pt, etc.) can regulate the d-band center of the surface Pt atoms, thus changing the binding strength between the electrocatalyst surface and oxygen intermediates^{[138,146,147]}. Recently, the Yang group proposed a Co₃O₄-assisted structural evolution strategy [Figure 7A] to controllably synthesize sub-6 nm core-shell Pt₁Co₁ intermetallic compound(Pt₁Co₁@Pt)^[148]. Experiments and theoretical calculations showed that the ordered arrangement of Pt-Co atoms reduced the d-band center of surface Pt relative to the pure Pt and enhanced the oxidation resistance of Pt/Co sites, thus improving the ORR activity and durability. The proton exchange membrane fuel cell (PEMFC) using Pt₁Co₁@Pt as cathode catalyst layers exhibited a record power density (2.30/1.23 W cm^-2 under H₂-O₂/air conditions) and extraordinary stability, making it the highest performance fuel cell of the time.

Figure 7. (A) Schematic illustration of the synthesis of Pt₁Co₁@Pt^[148]; (B) Schematic illustration of the general synthesis of 46 Pt-based intermetallic nanoparticles on porous sulfur-doped carbon supports^[77]; (C) The transmission electron microscopy (TEM) image of intermetallic PtCo nanoparticles; (D) Schematic illustration of the template-directed synthesis of porous intermetallic Pd-M nanosheets^[153]; (E) The TEM image of intermetallic Pd₃Pb nanosheets.

The synthesis of Pt-based intermetallics often requires high-temperature annealing to overcome the kinetic energy barrier of atomic ordering rearrangement in the solid phase. However, high-temperature synthesis inevitably leads to serious sintering of metal particles and a decrease in surface area, which ultimately results in low overall activity. It is highly desirable yet challenging to develop synthetic methods for small-sized Pt-based intermetallic catalysts. The Liang group developed a high-temperature sulfur anchoring synthesis strategy to construct an intermetallic library composed of 46 small-sized Pt-based intermetallic nanoparticles (< 5 nm), including 20 binary and 26 multicomponent intermetallics on sulfur-doped carbon supports [Figure 7B and C]^[77]. The strong interaction between Pt atoms and S atoms on carbon supports inhibits metal sintering during high-temperature annealing. Among them, the PtCo and PtNi catalysts showed mass activities of 1.52 and 1.84 A mg_Pt⁻¹, which was much higher than commercial Pt/C catalyst (0.29 A mg_Pt⁻¹) in H₂-O₂ cell tests at 0.9 volts. Subsequently, the Liang group realized the grams scale production of Pt-intermetallic catalysts on commercial carbon black supports by a small molecule-assisted impregnation method^[84,149]. This work is an important step towards the practical application of nano-sized Pt-intermetallic catalysts.

Pd is considered the most suitable alternative to Pt catalyst for ORR, because Pd has similar physical and electrochemical properties to Pt and is relatively inexpensive. The ORR properties of Pd-based intermetallics have been widely studied, such as PdCu,^[150] Pd₃Pb^[151], and Pd₆Ni^[152]. Huang et al. reported a template-directed rapid synthesis method [Figure 7D and E] for a series of Pd-based ultra-thin porous intermetallic Pd-M nanosheets (M = Pb, Sn and Cd)^[153]. The mass activity and specific activity of Pd₃Pb catalysts are 6.8 times and 9.8 times that of commercial Pt/C in the alkaline medium for ORR, respectively. Recently, Hu et al. reported a novel technique for the synthesis of intermetallic Pd₃Pb nanoparticles by rapid joule heating within 60 s^[154]. The short heating time can prevent particle aggregation, thus realizing the loading of ultra-small Pd₃Pb nanoparticles with a narrow size distribution (6 nm) on carbon fiber supports. The synthesized Pd₃Pb nanoparticles showed greater electrocatalytic activity for ORR than traditionally heated Pd₃Pb.

Intermetallic electrocatalysts for CO₂RR

Electrochemical CO₂RR provides an attractive approach for the reduction of greenhouse gas emissions and the production of high-value-added industrial products^[155-157]. The CO₂RR in H₂O electrolyte can be described as multiple steps of proton-coupled electron transfer and yields a range of different products, such as formates, CO, and hydrocarbons^[158]. Many metal catalysts for CO₂RR have been explored, which can be classified into three categories according to their main products: CO (Ag, Au, Zn, etc.), formates (Bi, Sn, Hg, In, etc.) ,and hydrocarbons (Cu)^[159]. However, most monometallic catalysts are undesirable for CO₂RR in terms of efficiency and selectivity. Intermetallic alloys provide unique electronic structures and well-defined coordination environments for precise tuning of catalytic activity and selectivity of CO₂RR^{[21,28,29,153,160]}. A broad range of nanostructured intermetallic catalysts has been developed for electrochemical CO₂RR^[161].

Compared to monometallic catalysts, intermetallic alloys can improve the catalytic activity. For example, Sn catalysts can only exhibit low CO₂RR selectivity within a narrow potential window for formate generation, due to the large energy barriers and slow kinetics of the oxygen-breaking hydrogenation and desorption steps^[162,163]. Tan et al. reported an ordered nanoporous intermetallic SnTe, which has better electrochemical CO₂ to formate reduction performance compared to monometallic Sn catalysts^[164]. This SnTe maintains more than 90% Faraday efficiency (FE) over a voltage range of -1.0 to -1.3 V (vs RHE). Combining operando Raman spectroscopy analysis with DFT calculation, it is found that the strong orbital interaction between Sn and neighboring Te in the intermetallic SnTe promotes bond breaking between metal and oxygen, which significantly increases the yield of formate. In addition, Jiang et al. synthesized a self-supported nanoporous Cu-Sn hybrid electrode, which is an intermetallic Cu₃Sn in situ formed on Cu skeleton (Cu₃Sn/Cu)^[165]. Due to the enhanced adsorption ability of CO on Cu atoms, the intrinsic activity of Cu₃Sn is 80 times higher than that of monometallic Cu, with a FE of 91.5% and excellent selectivity at an overpotential of 0.59 V.

Besides catalytic activity, the product selectivity of CO₂RR can be changed by the rational construction of intermetallic catalysts. For example, Yang et al. investigated the effect of disorder-to-order transformation for AuCu bimetallic nanoparticles on their CO₂RR selectivity [Figure 8A]^[166]. While the intermetallic AuCu electrocatalysts selectively converted CO₂ to CO with a high Faraday efficiency of 80%, the disordered counterpart was mainly catalytically active for HER [Figure 8B and C]. DFT calculation and X-ray absorption analysis further confirmed that the increased activity and CO selectivity of ordered AuCu nanoparticles are due to the formation of a compressive strain Au shell layer over the AuCu intermetallic core. Another example is intermetallic Pd₃Bi nanocrystals and chemically disordered Pd₃Bi nanoparticles [Figure 8D and E] that were selectively synthesized by Li et al.^[167]. The intermetallic Pd₃Bi exhibited excellent selectivity for formate generation (≈ 100%), while the formate selectivity of disordered Pd₃Bi reached only 60%. DFT calculation indicated that the crystallographic ordering of Pd and Bi atoms in Pd₃Bi can inhibit the *CO poisoning and improve *OCHO adsorption during CO₂RR to formate.

Figure 8. (A) TEM images of AuCu bimetallic nanoparticles with different ordering degrees^[166]; (B) Electrochemical CO₂ reduction activities of AuCu nanoparticles evaluated by Faraday efficiencies of CO and H₂; (C) Calculated limiting potentials for CO₂ reduction and H₂ evolution; (D) Schematic synthetic procedures for intermetallic Pd₃Bi and disordered Pd₃Bi alloys^[167]; (E) Site-dependent projected partial density of states of Pd-4d in intermetallic Pd₃Bi. (F) Crystal structure of CuGa₂^[168]; (G) Comparison of normalized XANES spectra of Ga K-edge of CuGa₂ and Cu₉Ga₄ with the standard sample.

For intermetallic catalysts, phase structures have a significant effect on the electrocatalytic performance of CO₂RR. Recently, Peter et al. prepared two intermetallic phases (CuGa₂ and Cu₉Ga₄) formed by copper and gallium through a high-temperature solid-state reaction for electrocatalytic CO₂RR [Figure 8F and G]^[168]. While the CuGa₂ achieved a selective reduction of CO₂ to methanol at a very low potential of -0.3 V with a Faraday efficiency of 77.26%, the Cu₉Ga₄ showed a much lower Faraday efficiency (37.75%) to produce methanol. The higher catalytic performance of CuGa₂ is found to attribute to its unique two-dimensional structure, which preserves surface and subsurface oxide species (Ga₂O₃) in the reducing atmosphere.

Intermetallic electrocatalysts for NRR

Industrial ammonia synthesis relies substantially on the Haber-Bosch reaction, which involves a massive energy input and emits large amounts of greenhouse gases^[169]. The electrochemical NRR under ambient conditions provides an attractive route to realize green ammonia synthesis. Efficient NRR electrocatalysts are highly required to break the inert nonpolar N≡N triple bond of a N₂ molecule and to inhibit severe competition from the HER in H₂O electrolyte^[170-172]. Intermetallic nanocrystals have been widely investigated in the NRR field to exhibit excellent catalytic properties and good stability. The synergistic atomic interaction in intermetallic compounds can modulate the adsorption and desorption of N-containing intermediates while suppressing the competing process of HER, resulting in a lower reaction energy barrier and superior catalytic activity^[51].

Many theoretical computational studies have been reported to accelerate the discovery of intermetallic catalysts with desirable NRR properties^[19,173]. Yuan et al. calculated the N adsorption behavior on the surface of 29 intermetallics with Cu₃Au-type structures^[174]. Among the calculated intermetallics, Pd₃Mo was considered as an attractive catalyst with a low limiting potential of -0.31 V and the ability to suppress the competing HER. Some theoretical studies have suggested that some unique structures of intermetallic compounds can provide ideal single-atom nonmetallic sites for efficient NRR. For example, the Qiao group simulated molybdenum boride models (e.g., Mo₂B, α-MoB and MoB₂) to explore their NRR activities^[175]. Electronic structure studies showed that the occupancy of the hybridized orbital between the p-orbital of the boron site and the antibonding π*-orbital of N₂ determined the binding strength of N₂, so that the filling of the p_z-orbital could be accepted as a descriptor of N₂ activation [Figure 9A and B]. With increased boron content, the filling of p_z-orbital increased in different Mo-B phases. The p_z-orbital of Mo₂B phase containing isolated boron sites is less filled, thus promoting the activation of N₂. Moreover, this conception of isolated boron site is equally applicable in other metal borides in the form of M₂B (M=Ti, Cr, Mn, Fe, Co, Ni, Ta, W). Due to the moderately filled p_z-orbital and the high adsorption strength of the reaction intermediates, Mo₂B, Fe₂B and Co₂B can be potential catalyst candidates for NRR.

Figure 9. (A) Schematic illustrations of hybridization interactions between the boron p-orbital of Mo₂B and the π*-orbital of the N₂ molecule; (B) Correlation between p_z-orbital filling with ICOHP and adsorption energy of N₂^[175]; (C) Schematic illustrations of single-boron sites confined in antiperovskite borides for NRR^[176]; (D) Schematic illustration of the synthesis of nanoporous intermetallic Pd₃Bi^[180]; (E) The SEM image of nanoporous Pd₃Bi; (F) schematic illustration of the NRR process on Pd₃Bi.

Very recently, our group simulated anti-perovskite boride models with metal-coordinating single boron sites (M₃M'B) to reveal the effect of a boron-centered local environment [Figure 9C] on the catalysis of N₂-NH₃^[176]. DFT calculation demonstrates that whereas the adsorption and activation of dinitrogen depend on the presence of isolated boron atoms on the surface, the adsorption conformation of dinitrogen is strongly dependent on the type of M atoms. The M atoms and isolated boron atoms work collectively to stabilize the N_xH_y intermediates for diazo conversion, and M' atoms can modulate the orbital hybridization of the B-M complex, thereby controlling the surface chemisorption and catalytic activity of the *N intermediates. Finally, we predicted that Rh₃GeB, Rh₃SbB and Ni₃LiB are the promising catalysts for ammonia synthesis with higher activity than the Ru (0001) surface.

Apart from theoretical studies, it was confirmed by experimental studies that some intermetallic compounds can work as highly active NRR catalysts^[177-179]. Tan et al. prepared nanoporous intermetallic Pd₃Bi with strong Pd-Bi coupling by transforming chemically etched PdBi₂ [Figure 9D and E]^[180]. The Bi sites in Pd₃Bi can adsorb N₂ molecules and lower the energy barrier of *N₂ [Figure 9F]. Simultaneously, the bicontinuous nanoporous structure of the Pd₃Bi can accelerate the electron transfer in NRR and further improve NRR performance. As a result, the Pd₃Bi can exhibit a high NH₃ yield rate of 59 µg h^-1 mg_cat^-1 and a high Faraday efficiency of 21.5% at -0.2 V. Additionally, Sun et al. reported the preparation of Mo₃Si thin films on graphite paper by the magnetron sputtering technique^[181]. Electrochemical tests demonstrated that Mo₃Si is an efficient NRR catalyst with high structural stability and exceptional selectivity for NH₃ production. DFT calculation indicated that the exposed Mo ions show high chemical activity to facilitate the activation and adsorption of N₂.

Although great achievements have been made, several unresolved questions still should be addressed in the catalytic application of intermetallics. (i) Current electrocatalytic studies of intermetallics mainly focus on HER and ORR. There is still insufficient research on CO₂RR and NRR electrocatalysis of intermetallic catalysts. (ii) Mechanistic investigations are scant and unsystematic for intermetallic catalysts, which generally possess complex surface structures and diverse active site motifs. Combining advanced characterizations with density functional theory (DFT) calculation, an in‐depth understanding of structure‐performance relationships at the atomic and electronic levels is highly required. (iii) Although a variety of intermetallic catalysts have been developed, their high activities are measured by three-electrode system, and have rarely translated into electrochemical devices such as fuel cells and water electrolyzers. Thus, establishing reasonable performance evaluation criteria is essential to bridge the gap between three-electrode and device tests.