With rapidly increasing global population and the developing risk of climate change, the demand for clean and sustainable energy has increased significantly over the past decades. Developing fossil-free pathways for energy consumption and environmentally friendly power sources and chemicals for the industry is critical for reducing global warming [1]. Among renewable clean energy technologies, proton exchange membrane fuel cells (PEMFCs), which directly convert chemical energy from fuels into electricity, have attracted considerable attention. They show many advantages, including high efficiency, high energy density, low operating temperature, and quick start-up. Consequently, they show promise for applications such as powering portable electronic devices, distributed stationary grids, and transportation vehicles [2]. Over the past decades, PEMFC technologies have achieved significant progress and are expected to achieve widespread commercial use, specifically in automobiles and buses. Fuel cell vehicles (FCVs) have been considered as one of the final solutions to the problems associated with the automotive business due to their advantages over secondary batteries. In 2014, the first mass-produced FCV was released, the Toyota Mirai, which was powered by H₂ and had a cruise mileage reaching 650 km [3].

Even so, PEMFCs are still in the early stages of commercialization, and they currently face many issues such as high cost and insufficient stability. As is well known, the sluggish oxygen reduction reaction (ORR) kinetics in PEMFC cathodes requires high Pt loading, as a catalyst, to promote the reaction rate. Due to the scarcity and high price of Pt, the predominant hindrance to the commercialization of PEMFC is their high cost. According to the US Department of Energy (DOE) hydrogen and fuel cells program reports, the current state-of-the-art Pt-based catalyst is projected to be the largest single component of the PEMFC stack cost. According to the 2025 target for automotive application, the PEMFC cost should be dramatically reduced to 40 ＄/kW by 2025, and 30 ＄/kW beyond 2030 [4]. The other obstacle to PEMFC applications is the instability, which can be attributed to several complicated reasons, including catalyst deactivation and membrane/systemic degradation. Catalyst deactivation has been reported as the main cause of the dramatic performance decay in PEMFCs. According to the US DOE, the PEMFC lifetime requirements for automobiles are 8000 h for cars, which means driving 150000 miles with less than 10% performance loss. Developing a high-performance, cost-effective, and highly stable catalyst for the ORR is the key for successful fuel cell commercialization.

After almost half a century of developing ORR catalysts, two main strategies have been determined to reduce the cost of cathode catalysts in PEMFCs. One is developing PtM alloy catalysts (M is other metals, such as Fe, Co, Ni, and Cu), which contain only tiny amounts of precious metal (low-Pt). The other is to explore catalysts that do not contain Pt, aiming at completely replacing Pt with Pt group metal-free (PGM-free) earth-abundant elements. Although significant progress has been achieved in the field of PGM-free catalysts involving their rational design and controllable synthesis, their activities and particularly their stabilities are still not acceptable for the PEMFC industry [5]. For low-Pt catalysts, various PtM alloys or intermetallics with novel structures and morphologies have been investigated that exhibit enhanced catalytic activity and stability toward ORR and could possibly be candidates to replace single metal Pt/C catalysts in the PEMFC industry.

New insights from advanced characterization techniques and the theoretical modeling of the ORR have led to new proposals on the mechanism and nature of catalysts with low-Pt. It is possible to improve the catalytic activity of ORR electrocatalysts by manipulating the atomic structure, tuning the surface electronic structure, and controlling the preparation processes based on the in-depth understanding of the ORR mechanism, and the relationship between the performance and properties of catalysts. Precise design and fabrication at the molecular or atomic level for low-Pt catalysts have proven to be the most crucial principle in the development of catalysts. Many excellent reviews on the development of these catalysts have recently been published; however, most of them focus on detailed preparation methods rather than the performance and structure [6-11]. To this end, we provide a comprehensive review on the nanostructure, morphologies, and performance of the latest highly active low-Pt catalysts, which show promise for replacing commercial Pt/C catalysts. We highlight the PtM alloys, particularly intermetallic, with novel nanostructures, and the effect of supports on their ORR performance to illustrate potential developments in the future. Furthermore, the challenges facing low-Pt catalysts are presented for the further development of ORR catalysts.

The ORR at the cathode has been known as the limiting electrochemical reaction in PEMFCs due to its sluggish reaction dynamics; therefore, it is the deciding factor for cell performance. A direct 4e^- pathway for the ORR with the formation of H₂O in acidic medium was preferred with an ideal electrode potential of 1.23 V, although along this pathway, several different intermediates could be formed. The kinetics of the ORR are still not fully understood due to its high level of complexity. Early research on pure Pt catalysts carried out by Nørskov et al. [12, 13] indicated that the adsorption of intermediate species such as O or OH might be the rate-determining step of the ORR. They calculated the adsorption energy of all intermediates during the ORR on the precious surface by density functional theory (DFT). It was found that Pt and Pd are the best metals for ORR, although there is room for activity improvement. Metals with a somewhat lower oxygen binding energy than that of Pt should have a higher rate of ORR, which binds oxygen more weakly than Pt{111} does by about 0.2 eV [12].

In the 1980s, it was discovered that Pt nanocrystals alloyed with transition metals exhibited superior ORR performance compared to that of pure Pt. Pt-based alloys have attracted significant attention not only due to the reduced usage of Pt but also their unique natures and structures. Thenceforth, intensive efforts have been devoted to the fundamental study of various Pt-based alloyed catalysts with different transition metals, including their components, compositions, morphologies, and structures. The corresponding speculation of the mechanisms and theoretical calculations have also been developed over the years. It is believed that incorporating transition metals causes a series of alternation in lattice structure due to the mismatch of atomic radii, which leads to the lattice strain in the PtM nanocrystals. Strain engineering is becoming increasingly significant for optimizing the electrocatalytic activity [14, 15]. Based on fundamental experiments and DFT modulations, lattice strain plays two roles in improving the catalytic performance. First, the surface strain causes the modification of binding strength between the catalyst surface and the reaction intermediates, which should be suitable not only for the adsorption of reactants but also for the removal of reaction intermediates. This binding strength was confirmed to govern the ORR reaction rate [16]. Second, the lattice strain directly influenced the electronic structures of PtM nanoparticles, thereby reducing the amount of unoccupied projected electronic states, which could be interpreted as the position of the electronic d-band center. In general, tensile strain upshifts the d-band center, while compressive strain downshifts the d-band center, according to the d-band model [17]. The modification in electronic structure actually contributes to the enhancement of catalytic activity [13]. Wadayama and co-workers [18] studied the surface strains and direct electronic effects induced by alloying elements. They evaluated the ORR activity enhancement factors for Pt(111)-shell layers on Pt₂₅Ni₇₅(111) single-crystal surface and found that 2-ML-thick Pt shell catalysts exhibited a remarkably enhanced ORR activity, 25 times higher than that of pristine Pt(111).

Besides the effect of M incorporation, the morphology and structure including both macro- and micro-structures of PtM alloys also affect the catalytic properties because specific structural features such as high-index facets, core-shell structures, nanoframes/nanocages, and ordered intermetallics will ultimately modify the overlapping interaction of electrons between the Pt and transition metals. Therefore, we emphasize on the recently developed Pt-based catalysts with excellent activities and stabilities that show promise for future application in the fuel cell industry.

PtM alloys with various types and amount of transition metals such as Co, Ni, Fe, Cu, Pd, Au, Ag, Mo, Mn, and Al have been synthesized and reported. It is generally believed that the activity enhancement in Pt alloy catalysts is mainly attributed to the modification of the geometric and electronic structures. Consequently, the different kinds of transition metals will cause different effects depending on their intrinsic nature. Stamenkovic and his colleagues [19] reported a volcano-type linear relationship between the catalytic activity of Pt₃M (M = Ni, Co, Fe, Ti, and V) and the experimentally determined surface electronic structure, the d-band center, and found that Pt₃Co exhibited the best specific activity. The optimized catalytic activity was governed by the balance between the adsorption energy of reactive intermediates and surface coverage of spectator (blocking) species. This electrocatalytic trend explained the active pattern of Pt₃M and provided a fundamental basis for the catalytic activity enhancement. Therefore, it is possible to further enhance the activity of PtM alloy catalysts by tuning the electronic properties of nanoparticles based on these trends with engineered nanoscale surfaces [19]. In addition, many other kinds of transition metal elements, including early transition metals and rare earth metals, have been studied in-depth [20, 21]. These catalysts also exhibited enhanced activities and durabilities toward ORR due to the ligand effect and compressive strain. Among all base metals, Fe, Co, and Ni were found to be the most effective alloying elements not only due to their low cost but also to their capability to improve the catalytic activity.

Another factor that affects the alloys catalytic activity is the ratio between Pt and M. Engineering the composition ratio of PtM alloys will change the surface electronic structure and may further improve the catalytic activity. At first, it was reported that PtM alloys with a 3:1 atomic ratio exhibited the highest activity [22, 23], and it was popularly used in the calculation models for theoretical speculation [13]. A new method using an electrochemical activation process was introduced to dealloy the initial PtNi alloy, thereby forming a Pt-rich surface. Dealloying dramatically decreased the content of Ni in the PtNi octahedrons, although it is difficult to establish the exact atomic ratio that would promote the best catalytic activity [24, 25]. Wadayama et al. [26] explored the alloy-composition-dependent ORR activity and the stability of the Pt/Pt_xNi_100-x(111) model system. They found that the ratio of Pt and Ni had considerable influence on the catalytic activity. The initial activity of the as-prepared two-monolayered Pt-covered Pt_xNi_100-x(111) substrates increased with increasing Ni composition in the Pt_xNi_100-x(111) substrate. For 4ML-Pt/Pt_xNi_100-x(111), activity enhancements were insensitive to alloy composition; however, thicker Pt shell layers did stabilize the catalyst against Ni leaching [26].

To further reduce the usage of Pt and improve both the catalytic activity and stability, PtM alloy catalysts with novel nanotextures have been designed and explored. The polyhedron-engineered nanoparticles, core-shell structures, and nanoframe/nanocage morphologies of Pt-based alloys have been developed and are considered promising catalysts for PEMFCs due to their unique morphologies, improved catalytic performance, and electrochemical stability.

It has been proposed that the ORR has a structure-dependent nature and, for single-crystal Pt, the surface facets follow an order of {100} < {111} < {110} for the ORR activity [27], whereas for the polyhedron-designed PtM alloy catalysts, it was commonly believed that the ORR activity on different facets increased in the order {100} < {110} < {111} [11, 27, 28]. Consequently, it was important to control the polyhedron PtM nanoparticles facets when searching for the optimum catalytic performance. As an example, PtNi octahedral nanocrystals demonstrated higher activity than that of PtNi cuboctahedral nanocrystals due to the higher percentage of {111} facets in the former. The octahedral PtNi nanocrystals exhibited a mass activity of 3.3 A/mg_Pt and specific activity of 7.3 mA/cm²_Pt, with well-preserved 8×{111} facets [29]. Interestingly, although the icosahedron is surrounded by 20×{111}, the DFT calculations and molecular dynamic simulations determined a better, strain-induced electronic effect from octahedrons than from icosahedrons, which resulted in a higher catalytic activity for octahedral nanocrystals [30]. Besides Ni, other metals such as Pd [30, 31], Co [32, 33], and Cu [34] were also introduced to study the facet-dependent ORR activity of PtM alloy structures with {100} and/or {111} facets. It was found that among all the polyhedral catalysts, the PtNi octahedral nanocrystals exhibited the most potential as a highly active PEMFC electrocatalysts. Thus, Pt-Ni nanocrystals with abundant {111} facets have become a popular area of research. A third metal doped into PtM alloys might further enhance the activity due to possible synergistic effects combined with facet-control. Huang et al. [35] prepared surface-doped Pt₃Ni octahedral particles supported on carbon with different transition metals. As shown in Fig. 1(a), the Mo-doped Pt₃Ni/C exhibited a specific activity of 10.3 mA/cm² and a mass activity of 6.98 A/mg_Pt at 0.9 V, which had an 81- and 73-fold enhancement compared to those of commercial Pt/C. This catalyst also exhibited excellent stability during the ADT test. After 8000 potential cycles between 0.6 and 1.1 V, the Mo-Pt₃Ni/C exhibited only 3 mV downshift for E_1/2, and the activity was still as high as 9.7 mA/cm² and 6.6 A/mg_Pt for specific and mass activities, respectively. DFT calculations indicate that Mo doping helps to modify the local oxygen binding energies at some specific sites. Consequently, some sites might become highly active for catalysis and enhance both the performance and the stability of the Pt₃Ni catalysts (Fig. 1(b)) [35].

Generally, electrocatalytic reactions take place on the surface of nanoparticles because only the surface atoms are accessible to the electrolytes and participants of the reaction, leading to the low utilization of valuable Pt. To further reduce the noble metal usage as well as improve the catalytic activity and stability of PtM alloy catalysts, active shell structures encapsulating various core substrates have received tremendous attention. The composition of the core and the morphology of the shell can be carefully designed to obtain desirable interactions between core and shell, leading to enhanced ORR activity [36]. It was constructed as a thin Pt-based shell coating and a less expensive core consisting of transition metal nanoparticles. Recent theoretical studies predicted that many bimetallic Pt-based core-shell structures would exhibit relatively high ORR activity and stability [37]. Many experimental studies also demonstrated their potential application as ORR catalysts in PEMFCs [38-40]. The octahedral Pd@Pt_1.8Ni core-shell nanocrystals with ultrathin PtNi alloy shell structure exhibited a mass activity of 0.79 A/mg_Pt, which was 4.9 times higher than that of Pt/C (0.16 A/mg_Pt) at 0.9 V. The enhancement of the catalytic performance could be attributed to the alloying effect of Pt with Ni, as well as the interfacial interactions between the thin shell and the Pd core, both of which might modify the electronic and geometric structures of the shell atoms [41]. Expectedly, the character of the shell, including the component and thickness of the shell, directly influences the surface properties of nanoparticles. By controlling the amount and relative concentrations of the metal precursors, Xu et al. [42] manipulated the shell composition and thickness of Au@Ni_mPt₂ nanoparticles, as well as the core-shell interaction and surface electronic structures as illustrated in Fig. 1(c). After thermal pretreatment in flowing 2% CO/Ar, the obtained Au@Ni₂Pt₂/C catalysts exhibited a specific activity of 0.8 mA/cm²_Pt and mass activity of 0.56 A/mg_Pt at 0.9 V (Fig. 1(d)), about 3-fold higher than those of commercial Pt/C catalysts. This catalyst also exhibited much higher stability during the potential cycling tests; the specific and mass activities remained 0.72 mA/cm²_Pt and 0.42 A/mg_Pt, respectively [42].

Nanoframe and nanocage structures, which are typically open three-dimensional (3D) nanostructures that consist of cellular architecture with nano- and micro-lattice framework, are becoming an important alternative catalyst strategy and a dominant research focus for advanced fuel cell technologies. Consequently, both the exterior and interior surfaces are accessible to the reactant during the electrocatalytic reaction, potentially improving the electrocatalytic performance. Moreover, the usage of Pt can be reduced to the minimum in the Pt-based nanoframe structure due to the increased Pt utilization. It has been reported that Pt atoms are prone to segregate on the outer surface of the edges in the nanoframe structure, accompanied by the incorporation of transition metals [43]. By controlling the surface component and configuration for these nanoframe/nanocages at the atomic level, the catalytic properties can be tuned precisely to achieve enhanced activity and stability.

Chemical etching or corrosion of erosion was manipulated to selectively remove definite facets and undesirable elements by controlling the etching strength and reaction environment. Several previous studies reported Pt-based nanocage catalysts prepared by selectively etching the Pd core of Pd-Pt core-shell structure with the formation of ultrathin Pt-enriched nanocages [44-46]. Mustain et al. [47] recently synthesized a type of Pt-Ni nanocage electrocatalyst, which was transformed from a Pt-Ni alloy by applying a complicated two-phase corrosion process. They tested their catalytic activity and stability in both a three-electrode half-cell and an MEA single cell. Although the Pt-Ni nanocage catalysts exhibited much higher activity and stability toward ORR than those of Pt/C in the half-cell test, the cell performance was still short of the DOE 2020 target [47]. Stamenkovic and his cooperators [48] synthesized Pt₃Ni nanoframes using PtNi₃ solid polyhedral as a precursor, which has a hollow 3D structure and a {111}-like Pt skin (Fig. 1(e)). This open architecture resulted in a 22-fold improvement in mass activity (5.7 A/mg_Pt) over that of Pt/C (Fig. 1(f)), and exhibited remarkable durability with almost negligible activity loss after 10000 potential cycles [48]. The enhanced durability was ascribed to the electronic structure modification of the Pt-skin surface resulting from a relatively low coverage of oxygenated intermediates, which diminished the probability of Pt dissolution.

Materials with one-dimensional (1D) anisotropic structures such as nanowires [49], nanotubes [50], and nanorods [51, 52] have attracted considerable attention in recent years owing to their high flexibility, high surface area, and high conductivity. In general, these 1D nanostructures are obtained by the orientation growth of crystals with a rough exterior surface covered in high-index facets. There exist many edges, corners, and stepped atoms on these rough surfaces, which are highly reactive, and therefore, beneficial to the catalytic reaction [53-55]. A class of 1D high-index faceted PtNi alloy nanowires (NWs) with an uneven ultrathin diameter, and the rough surface was prepared by a wet chemical method. Many small facets with high-indexes such as {211} and {311} can be clearly observed in the high-resolution transmission electron microscopy (HRTEM) and scanning transmission electron microscopy (STEM) images, shown in Figs. 2(a)-2(c). The PtNi NWs supported on the Vulcan XC72R (PtNi/C) catalysts exhibited a specific activity of 9.2 mA/cm² at 0.9 V and a mass activity of 4.15 A/mg_Pt, which were 51- and 34-fold greater than those of the Pt/C catalyst, respectively. After 10000 potential cycles, the specific and mass activities were still as high as 8.6 mA/cm² and 3.96 A/mg_Pt, more than a 90% retention rate, indicating excellent durability [53]. DFT calculations indicate that the oxygen adsorption energy (E_O) at atomic steps on high-index facets were closer to the optimal E_O than the pure Pt catalysts were (Fig. 2d). In another study regarding PtNi NWs, a Pt/NiO core/shell nanowire alloy was synthesized and loaded onto carbon black, which was then annealed in an Ar/H₂ atmosphere. After electrochemical dealloying to gradually remove Ni atoms from the alloy, the Pt atoms on the surface rearranged to form an ultrafine jagged rhombic configuration. These jagged PtNi NWs delivered specific and mass activities of 11.5 mA/cm² and 13.6 A/mg_Pt at 0.9 V, respectively. It was found that the surface Pt atoms on these jagged features were mostly undercoordinated, which was a crucial factor for enhancing ORR activity. Additionally, the mechanical strain of the surface atoms decreased the binding energy of adsorbents and made the surface more active, further contributing to the activity enhancement [55].

To date, considerable progress has been achieved for PtM alloys catalysts for the ORR, particularly in regard to the catalytic activity improvement. The activity enhancement originates from the electronic and geometric modification of the Pt lattice due to M incorporation, including the Pt 5d vacancy, Pt-Pt interatomic distance, and the number of Pt nearest neighbors. As a peculiar member of PtM alloys, structurally ordered PtM intermetallic is becoming the most possible ORR electrocatalysts due to its specific properties.

In a typical PtM alloy, Pt and M atoms are randomly distributed in crystallographic sites, still showing the face-centered cubic (fcc) structure of Pt [56]. This random site occupancy of alloys limited the availability of catalyst active sites to certain regions, consequently restricting the improvement of activity. Moreover, the disordered solid solution alloy structure cannot stabilize M atoms under high potential and oxidative acidic circumstance in PEMFC due to the lack of bonding between Pt and M atoms. Consequently, M etches away easily, causing performance degradation of catalysts and fuel cells. Finding a way to simultaneously improve the ORR activity and stability in PtM system is critical.

In 2004, Abruña et. al. [57] reported the electrocatalytic activity of ordered intermetallic phases for fuel cell applications including both the oxidation of fuels (formic acid, methanol, ethanol, and ethylene glycol) and the ORR. They prepared a significant number of ordered intermetallic phases (PtBi, PtPb, PtIn, PtMn, PtSn, and PtSb) that exhibited electrocatalytic activities superior to those of pure polycrystalline Pt and their corresponding PtM alloys [57]. In an ordered intermetallic crystal lattice, the Pt and M constituent atoms occupy the specific sites, with the formation of defined stoichiometry and an ordered atomic structure [58-60]. Consequently, the intermetallic phase shows the properties of definite stoichiometry, long-range ordered crystal structure, and predictable control of the electronic structure, which are not afforded by the PtM alloys [57, 61]. Therefore, these PtM intermetallic compounds exhibited dramatically higher electrocatalytic activities and stabilities than those of the disordered alloys toward the ORR. Many researchers have verified that this performance enhancement can be attributed to the definite composition and the synergistic structural, geometric, and electronic effects [62]. To date, several PtM intermetallic phase systems have been studied for bi-metallic [63-69] and tri-metallic systems [70, 71].

Presently, the most common method for synthesizing these intermetallic compounds involves co-reduction or impregnation reduction, followed by high-temperature annealing under inert or reductive gas. The PtM alloys nanoparticles were obtained first, and annealing at a high temperature (> 500 ℃) is an indispensable process for the phase transformation from disordered to ordered intermetallic structure. However, the disadvantageous sintering of nanoparticles during heat treatment, with the increase in particle size, usually caused limited enhancement of activity in early research on intermetallics [72-75]. Therefore, it is important to control the particle size of PtM intermetallic nanoparticles during phase transformation for the synthesis of PtM intermetallic catalysts. Recently, an electrodeposition method was explored to synthesize intermetallic Pd₃₁Bi₁₂ catalysts with atomic scale ordering at room temperature, demonstrating a new method of preparing ordered intermetallics with extraordinary catalytic activities and providing a new direction in catalyst discovery and synthesis [76].

Over the last decade, several strategies have been employed successfully to reduce the intermetallic particle size and narrow the size distribution during heat treatment. These strategies include oxide (MgO, Fe₃O₄) and carbon layer coatings on the surface of PtM particles, and protective agent (KCl) assistance during the annealing process. Sun and co-workers reported that the fcc-PtFe nanoparticles transformed into face-centered tetragonal-PtFe (fct-PtFe) with a protective MgO coating at the high temperature of 750 ℃. Compared with the case with the particles before annealing, the fct-PtFe nanoparticles exhibited no morphology change without aggregation, maintaining an average particle size of ~8 nm. This MgO coating was removed by a dilute acid wash, affording "clean" fct-PtFe nanoparticles for catalytic studies [77, 78]. Unfortunately, the transformation from disordered to ordered structure was incomplete because the protection limited atom mobility within the MgO shell. A modified strategy was proposed to increase the degree of ordering of the FePt particles. In this method, dumbbell-shaped PtFe-Fe₃O₄ nanoparticles were synthesized and then coated by a MgO shell. The fcc-FePt-Fe₃O₄/MgO nanoparticles were annealed at 700 ℃ under mixed gas of Ar and H₂. With the protection of MgO, the Fe₃O₄ was reduced to Fe, which diffused and facilitated the formation of the fully ordered fct-structure, with an average size of 8.8 ± 0.5 nm without aggregation, inside the MgO enclosure (Fig. 3(a)). When suspended in 0.5 mol/L nitric acid for 1 h, the fully ordered fct-FePt nanoparticles exhibited a negligible Fe/Pt composition change. Expectedly, the high degree of fct-ordering structure tended to stabilize Fe in the FePt against Fe leaching in the acidic solution. This fully ordered catalyst exhibited a half-wave potential (E_1/2) of 0.958 V vs. RHE, much higher than that of the partially ordered one (0.927 V). The specific and mass activities of the fully ordered fct-PtFe catalysts reached 3.16 mA/cm² and 0.69 A/mg_Pt at 0.9 V, respectively. When tested in identical conditions, the partially ordered fct-PtFe showed a slight performance loss after 10000 cycles, while the fully ordered-one was the most durable, showing no obvious performance loss, no particle morphology changes, and very small leaching of Fe after 20000 cycles [79].

Another method to suppress the growth and agglomeration of nanoparticles is to use protective agents during the annealing process. A surfactant-free KCl-assisted method (Fig. 3(b)) for the synthesis of intermetallic nanoparticles with controlled size and structure has been developed [80, 81]. KCl, which was formed in-situ as the by-product of the reduction reaction, worked as a physical barrier, mitigating agglomeration/sintering during the high-temperature treatment. The annealing temperature should be lower than the melting point of KCl to avoid agglomeration of nanoparticles due to the melting of the KCl matrix [82]. Using this method, an ordered Pt₃Cr structure was obtained, which exhibited a relatively narrow particle size distribution with an average size of 5.2 ± 0.5 nm. Consequently, heat treatment within a KCl matrix might be an effective way to prevent aggregation and control the size distribution of nanoparticles for the phase transformation for Pt-based intermetallic catalysts.

Carbon shell coatings are an effective way to suppress the coalescence of nanoparticles by blocking the migration of nanoparticles. The as-prepared disordered fcc-PtFe nanoparticles supported on carbon black were coated using polydopamine, which then underwent thermal annealing at 700 ℃ for 2 h as shown in Fig. 3(c). Subsequently, a very thin N-doped carbon shell was formed in situ after heat treatment, derived from the carbonization of polydopamine, which effectively prevented PtFe nanoparticles from coalescing with a particle size of 6.5 nm. This carbon layer coated ordered fct-PtFe/C catalyst showed higher performance than both the disordered one and Pt/C did, with a specific activity and mass activity of 2.3 mA/cm² and 1.6 A/mg_Pt, respectively. Moreover, it was found that N-doped carbon shells also acted as a highly effective protective coating, protecting nanoparticles from sintering/aggregation during the stability test. The ordered fct-PtFe/C catalysts demonstrated almost negligible activity loss after ADT in both Ar and O₂ conditions after 10000 potential cycles (Fig. 3(d)). The membrane electrode assembly (MEA) was performed under continuous operation for 100 h and exhibited high stability, with a loss in maximum power density of only 3.4%, which is better than Pt/C, which lost 27%, as shown in Figs. 3(e) and 3(f) [83].

Besides the strategies that reduce the particle size of PtM intermetallics during the annealing process, morphology control during synthesis is also important to further improve the performance since the electrocatalytic reaction is sensitive to structure. PtM intermetallic catalysts with specific morphologies and structures have achieved high activities and stabilities, such as core-shell, nanowires, nanoplates, and mesostructured skeletons.

Similar to PtM alloy catalysts, the core-shell structures were introduced to maximize the utilization efficiency of Pt, which consisted of several layers of Pt atoms sitting on the surface of a metal intermetallic core. The main methods for synthesizing the core-shell structure involve either a displacement or dealloying reaction, followed by heat treatment. Underpotential deposition (UPD) of Cu, followed by galvanic displacement leads to the formation of a Pt monolayer on certain substrates through a relatively complex electrochemical process [84, 85]. Recently, a Pt skin on AuCu intermetallic nanoparticles was obtained by a simple Pt-Cu displacement reaction. It was found that three layers of surface Cu were displaced during the formation of the Pt skin, which had an average thickness of 1.5 monolayers. Density functional theory (DFT) calculations confirmed that when three layers of Cu atoms were displaced, the oxygen adsorption energy of the Pt segregated surface was closest to the optimal value of Pt(111) (–3.81 eV). The change in the electronic structure in Pt due to the charge reception from the AuCu substrate, which would lower the d-band reactivity and the surface oxygen affinity, was confirmed by the large negative shift of the binding energy in X-ray photoelectron spectroscopy (XPS) [86].

For the dealloying method, a PtM composition was often synthesized and heated at high temperature, after which a chemical/electrochemical leaching process was conducted in an acidic solution to dissolve the M on the surface. Mukerjee et al. [87] reported a dealloyed Pt₁Co₁ nanoparticle catalyst that possessed a unique ordered PtCo core/ultrathin nanoporous Pt shell structure. MEA tests (0.2 mg_Pt/cm²) in H₂/O₂ revealed an initial mass activity of 0.49 A/mg_Pt at 0.9 V, which declined to 0.32 A/mg_Pt after 30000 voltage cycles (0.6–1.0 V vs. RHE). Modeling calculations based on XANES revealed that the cooperative compressive-strain and ligand effects induced by the Co introduction might contribute to the enhanced activity and stability [87]. Sun et al. [88] reported a core/shell L1₀-PtFe/Pt structure with a 5 Å thick Pt shell obtained by acid etching and post thermal annealing (Fig. 4(a)). The L1₀-PtFe/Pt showed an initial E_1/2 of 0.945 V and excellent stability at 60 ℃ without obvious ORR performance loss after 10000 cycles (4 mV negative shift). The mass activity dropped only from 0.70 to 0.68 A/mg_Pt at 0.9 V, and TEM images showed nearly unchanged morphologies after ADT testing, even at 60 ℃. The MEA and ADT tests were conducted following the DOE protocol. H₂/air polarization curves showed little degradation after 30000 cycles in MEA, and the mass activity of the L1₀-FePt/Pt was stable, with no drop during the ADT (Fig. 4(b)). However, its initial mass activity in fuel cell was still below the DOE 2020 target (0.44 A/mg_Pt) [88]. Subsequently, a typical hard-magnet L1₀-CoPt/Pt core/shell nanoparticle with 2–3 atomic layers of Pt (Fig. 4(c)) achieved a mass activity of 0.56 A/mg_Pt initially and 0.45 A/mg_Pt after 30000 voltage cycles in the MEA test at 80 ℃ (Fig. 4(d)), which represents good activity and decent stability in real cell operation conditions. It was believed that the L1₀ ordering in the core structure and the thin Pt shell structure facilitated the ORR performance in terms of both activity and durability. This novel structure not only stabilized Co from leaching in acidic media but also optimized the oxygenated species-binding energies on the compressed Pt surface [89]. Atomic rearrangement during heat treatment helps to form the core-shell structure through a simple method involving impregnation, reduction, and annealing processes. In this method, dealloying is excluded and the core-shell structure was formed during heat treatment [90-92]. High temperature induced the inter-diffusion of M atoms and the selective segregation of Pt atoms at the surface of the nanoparticles, with the formation of nanoparticles with several atomic-layers of Pt shell covering the ordered PtM intermetallic core. Different from the core-shell PtM alloys, the ordered intermetallic structure in the core enhances the effects of lattice contraction and electronic state modification. Moreover, the Pt atomic layers on the intermetallic surface prohibited the M atoms in the core from dissolving into the acidic environment.

Through an alternative method, the core-shell structures with a specific morphology of PtPb/Pt hexagonal nanoplates were synthesized by liquid reduction. The monodispersed edge length of nanoplates was determined to be ~16 nm and the thickness to be 4.5 ± 0.6 nm. HRTEM and STEM-electron energy-loss spectroscopy (EELS) techniques revealed a Pt edge layer wrapped around a PtPb core with a thickness of about 0.8 to 1.2 nm (four to six atomic layers). The atomic resolution High-angle annular dark-field (HAADF)-STEM images displayed the Pt and PtPb phases from different stacking sequences, indicating an obvious intermetallic structure of PtPb in the core. Consequently, two types of interfacial planes formed in the PtPb nanoplates with both compressive and tensile strain in different crystal orientation, which are {010}PtPb//{110}Pt between the PtPb and the edge-Pt layer, and {001}PtPb//{110}Pt between PtPb and the top (bottom)-Pt layer. The unique Pt {110} surface would be beneficial for ORR activity enhancement because the Pt{110} facet was intrinsically more active than the Pt{111} facet in perchloric acid. The DFT calculations (Fig. 4e) showed that tensile strain on the Pt(110) facet could also increase the ORR activity and that the low-coordinated surface atoms could be activated by large tensile strains. Consequently, the specific activity of PtPb nanoplates/C reached 7.8 mA/cm² and achieved a mass activity of 4.3 A/mg_Pt at 0.9 V, much higher than those of PtPb nanoparticles/C and Pt/C. After 50000 potential cycles, there was almost no negative shift for E_1/2 and only a 7.7% loss of mass activity [93]. Another specific octahedral core-shell structure with intermetallic PtPb as the core and atomic layers of PtNi as the shell was achieved by introducing additional Ni precursor into the synthesis of PtPb octahedral without heat treatment. The line scans and the XPS spectra showed that the PtNi shell was located outside the octahedron and was distributed homogeneously on the surface. It was believed that the integration of strong tensile strain of the PtPb to the Pt(110) facet, the ordered intermetallic phase core, and the active PtNi shell were highly beneficial for the enhancement of ORR performance [94].

For PtM intermetallic catalysts, various special structures with unique morphologies have been studied, many of which have been confirmed to be beneficial for the enhancement of catalytic activity. The 3D skeleton framework and rough surfaces increase the accessibility of the active sites to the reactants, O₂ and H⁺, for the ORR, which improves the utilization factor of Pt in PtM intermetallics [68, 69, 95]. Unique mesostructured Pt-Al skeletons, consisting of several atomic layers of Pt encompassing a PtAl intermetallic compound, were prepared by dealloying the alloy of Pt and Al. Due to the dissolution of Al, a bimodal mesoporous architecture with quasi-periodic Pt₃Al/Pt ligaments and mesoporous channels was obtained as shown in Fig. 5(a). The HAADF-STEM images illustrate the L1₂ ordered structure of intermetallic Pt₃Al with a periodic square of pure Al columns (low intensity) surrounded by Pt columns (high intensity) (Figs. 5(b) and 5(c)), similar to the Pt₃Co structure. This mesoporous Pt₃Al/Pt/C catalyst exhibited a remarkable positive shift of ≈52 mV in the E_1/2 relative to that of Pt/C, with a specific activity of 1.23 mA/cm²_Pt, which was about 6.3 times higher than that of Pt/C. It showed only a 6-mV negative shift in E_1/2 after 10000 cycles and 3.7% Al dissolution during stability testing. The enhancement of catalytic activity and stability were attributed to not only the bimodal mesoporous architecture but also the strong Pt-Al covalent bonds at the Pt_xAl/Pt-skin interface [96].

As discussed in PtM alloys, the 1D nanowire (NWs) structure sheds light on the development of advanced catalysts due to its particular structural properties, inherent anisotropic morphology, high flexibility, and high surface area compared to the 0D structure. Compared to that of the alloyed nanowires, the intermetallic NWs with Pt atomic-layer skins exhibited enhanced stability. A class of hierarchical PtCo intermetallic NWs enclosed with high-density and high-index faceted Pt skins were synthesized through a robust wet-chemical approach. The length of the NWs was in the range of several micrometers and the average aspect ratio was ~50, exhibiting highly uneven and crenel-like hierarchical nanostructures. Both X-ray diffraction (XRD) and HAAD-STEM measurements confirmed their ordered intermetallic structure (Figs. 5(d) and 5(f)). It was found that the formation of such hierarchical Pt₃Co NWs relied on the initial formation of Pt NWs, followed by the reduction of Co species onto the preformed NWs, followed by inter-diffusion to form Pt₃Co NWs. The resulting Pt₃Co NWs/C catalysts exhibited a specific activity of 7.12 mA/cm² and a mass activity of 3.71 A/mg_Pt at 0.9 V vs. RHE. This hierarchical Pt₃Co NWs/C catalyst also demonstrated excellent electrochemical stability. After 20000 potential cycles, the mass activity was still as high as 3.41 A/mg_Pt, maintaining 91.9% of the initial value (Figs. 5(g) and 5(h)). DFT studies indicate that the high ORR activities on the PtCo NWs were mainly attributed to the hollow sites on the [110] and [310] high-index facet of the NWs [66].

To compare the present performance of PtM catalysts in the fuel cells, the MEA results for both PtM alloys and intermetallic are included in Table 1. Although much effort has been devoted to the research on ORR catalysts, few studies have reported the catalytic performance of PtM in a real PEMFC. Moreover, the activity gap between RED and MEA results is still large, which is attributed to the complex differences between the operating conditions of these two systems. Presently, the DOE 2020 target for Pt-based catalysts is a Pt mass activity @0.9 V of 0.44 A/mg_Pt and < 40% loss in initial mass activity after 30 K potential cycles with total Pt loading of 0.125 mg/cm² [97]. Fortunately, several promising catalysts, including PtNi alloys and PtCo intermetallics, show potential for application in PEMFCs, which would meet the requirement of these challenging targets [89, 98, 99].

Due to the significant role catalysts play in PEMFCs, the design of catalysts, including not only PtM nanoparticles but also the support materials, has an important effect on the catalytic activity and durability. Many non-carbonaceous materials, which show enhanced corrosion resistance, have been proposed to work as catalysts supports such as carbides and oxides. However, several significant drawbacks still exist compared to the case with carbon materials such as lower electronic conductivity and specific surface area. Moreover, due to their higher intrinsic density, it is difficult for them to disperse uniformly in solution during the fabrication of ink and MEA. Therefore, carbon materials are still the optimal supports for electrocatalysts. Currently, the most commonly used supports are carbon black (Vulcan XC-72) and Ketjenblack. Although they showed reasonable performance in many research works, the corrosion of carbon support materials occurs at high potentials in the presence of oxygen, resulting in the detachment of metal particles and degradation of the fuel cell performance. The main reason for carbon support corrosion might be their non-long-range order in the graphitic lattice, which means their degree of graphitization is not well developed. Moreover, the metal particles were often prepared by chemical method and subsequently deposited onto the carbon supports, leading to a weaker interaction between them. This often led to the migration and sintering of metal particles on the carbon supports, decreasing the active surface. Consequently, the catalytic performance degraded during long-term operation. Furthermore, the poor interaction caused the high electron-transfer resistance and an undesired increase in ORR overpotential.

To solve the issues of corrosion in carbon supports, developing highly graphitic carbon materials has been proposed, including graphene, carbon nanofibers, and carbon nanotubes. Graphene has been extensively studied in the past years owing to its well-developed graphitic structures, which feature many benefits for the ORR, such as high surface area, high conductivity, and high stability. Many researchers have demonstrated that graphene can enhance the ORR activity as well as the stability. Sun et al. [104] synthesized 7 nm FePt nanoparticles and assembled them on graphene by a solution-phase self-assembly method. The graphene/FePt catalysts exhibited high catalytic activities and durabilities toward ORR. In-depth studies indicated that the close contact between graphene and FePt nanoparticles facilitated the p-electron polarization of graphene to FePt, making the FePt surface more easily accessible to O₂ for absorption and activation [104]. Recently, Wu and co-workers [105] reported a large type of nitrogen-doped graphene tubes (NGTs) as a support for Pt alloy nanoparticles, as shown in Fig. 6(a). The NGTs were synthesized by simple heat treatment of metal salts (Co, Ni), carbon, and nitrogen precursors. After high-temperature annealing and acid leaching, NGTs with diameters ranging from 150 to 350 nm were obtained. Thereafter, Pt nanoparticles were deposited on them. In this method, a post-treatment was needed to further improve the catalytic activity and stability by strengthening the interaction between Pt nanoparticles and NGT, with the formation of a PtM alloy. Most importantly, the abundant mesopores and macropores of NGT supports facilitate mass transfer for the ORR. The stability test for carbon supports against oxidation was carried out by applying high potentials ranging from 1.0 to 1.5 V at both room temperature and 60 ℃. The results indicate that NGTs were the most stable carbon support under extremely severe conditions in terms of corrosion resistance. The abundant π sites on the graphitized carbon surface interacted strongly with metals as shown by XPS and Raman data. Simultaneously, the introduction of nitrogen into the graphitic carbon matrix proved a useful strategy for stabilizing the Pt nanoparticles. Consequently, it was believed that the enhancement to activity and durability could be attributed to the novel structural and chemical properties of NGTs [105].

Except for the high degree of graphitization of carbon supports, further engineering of the nanostructure of these carbon materials offers the possibility of improving catalytic features. For example, ordered graphitic mesoporous carbon (OMC), mesostructured hollow graphitic spheres, and carbon nanotubes were developed to act as robust carbon supports for the ORR [6, 106]. Common characterization of these supports includes highly porous morphology and high surface area, which facilitate the mass transfer during the reaction process. Most recently, a novel kind of carbon support was introduced, which was derived from metal-organic frameworks (MOFs) [107]. The original porosity of MOFs could be partially retained in the resulting carbon supports after a thermal conversion. Consequently, the carbon derived from MOFs possesses not only high specific surface area but also high porosity, making them suitable as metal catalysts supports [108].

As for the PtM alloy/intermetallic catalysts mentioned above, most of them were synthesized separately and thereafter attached to the carbon materials by physical adsorption. Consequently, there is relatively weak control over the interface between intermetallic nanoparticles and carbon supports. It would be desirable to optimize the synthesis of the ordered structure by enhancing the bonding between the PtM nanoparticles and carbon supports. Recently, we developed a new approach for preparing ordered Pt₃Co intermetallic catalysts on a zeolitic imidazolate framework (a kind of MOF) derived (ZIF-8-derived) nanocarbon. In this method, carbon derived from Co-doped ZIF-8 with atomically dispersed Co sites was used as a support for depositing Pt nanoparticles, after which subsequent annealing treatment facilitated the diffusion of Co atoms into Pt nanocrystals with the formation of highly ordered Pt₃Co intermetallic structures. It was found that the content of Co in the ZIF precursors and annealing temperature were very crucial for the structure transformation because they are determined by the Pt/Co atomic ratio and the diffusion kinetics of Co in carbon supports. The STEM images coupled with electron energy loss spectroscopy (EELS) exhibited different Pt and Co distributions before and after heat treatment and confirmed the formation of ordered Pt₃Co structures (Fig. 6(b)). The optimized catalysts showed an E_1/2 of 0.92 V and only 12 mV negative shift in E_1/2 after 30000 potential cycles (Fig. 6c). MEA tests revealed encouraging fuel cell performance in the kinetic range and yielded a current density of 0.27 A/cm² at 0.8 V. In the future, this method might be extended to studies of materials other than PtM catalysts, potentially for electrocatalysis applications [103].

Most recently, an ultralow-loading Pt-Co catalyst using Co or Co/Zn zeolitic imidazolate frameworks as precursors was reported (Fig. 6d). It was found that the synergistic effects between strained Pt-Co core-shell nanoparticles and a ZIF-derived carbon substrate led to excellent catalytic activity for the ORR. RDE test results showed an E_1/2 as high as 0.96 V and a mass activity of 12.36 A/mg_Pt at 0.9 V. In particular, the MEA test demonstrated a much higher current density than that of the commercial one. The fuel-cell-based Pt mass activity for the optimized catalyst reached 1.77 A/mg_Pt at 0.9 V, and the maximum power density for the fuel cell was about 1.4 W/cm² under H₂-O₂ circumstance. The durability test based on MEA also indicated a promising result, retaining 64% of the initial mass activity, and surpassing the catalysts durability goal of < 40% set by the DOE. Notably, MEA with catalysts using high surface area carbon as supports showed a high current density at a low voltage range, which can be attributed to the improved mass transfer, where more active sites are more accessible to the reactants [99].

As is well known, for graphitic carbon material, the high degree of graphitization and high surface area with the porous structure are contradictory, since microporous features contribute most of the specific surface area. Micropores exist in graphitic layers with the formation of defects, which breaks the long-range ordered structure. These defects lead to a large number of edge plane sites, which often suffer from carbon oxidation, leading to carbon corrosion at a high potential in an O₂ environment. Conversely, porous structures (micro-, meso-, and macro-) are beneficial to the formation of the "so-called" three-phase interface, which is the interface where the reactant gas (O₂) meets with protons (H⁺) in the liquid phase and electrons (e^-) in the solid phase. Kongkanand et al. [109] systematically studied the fuel cell performance using accessible mesoporous carbon as catalyst supports. They considered that the ideal carbon supports should possess internal porosity protecting metal nanoparticles from ionomer adsorption as well as allow protons and O₂ access to catalyst nanoparticles. The accessible carbon pores should have a small opening to restrict ionomer penetration but not so small that it restricts O₂ transport. Porous carbon with a preferred pore opening of 4-7 nm can produce catalysts with both excellent ORR activities and transport properties [109]. Consequently, the rational design of carbon support materials with both well-developed graphitic and porous structures might be a possible and effective way to explore novel support materials for PtM catalysts.

Significant progress has been made in the development of low-Pt electrocatalysts with enhanced ORR performance. Through fundamental research, which focused on the composition, morphology, and atomic structure, the related electrocatalytic performance has indeed been considerably improved and the usage of valuable metals has been effectively minimized, particularly for the PtCo intermetallic with highly ordered structures, which exceeded DOE 2020 activity and stability targets in MEAs. Despite these significant achievements, there still exist some serious challenges in the development of catalysts that must be resolved before the commercialization of PEMFCs is feasible.

In-depth understanding of the ORR mechanism for these catalysts is still the most significant challenge. It is important to further explore the ORR mechanism using highly sophisticated modeling on the complex reaction kinetics, which will help to obtain profound insights into the dominating factors involved in activity enhancement. The catalytic properties for ORR are related to the geometric/composition/size and are associated with the electronic structures of the PtM catalysts, which consequently influences the adsorption/desorption nature of the reactant (O₂) and intermediates (O^*, OOH^*, OH^*) on the catalysts surface, which determines the reaction rate. Besides improving the intrinsic catalytic activity by the controlled synthesis of PtM nanoparticles, modifying their surface with carefully selected chemicals might be a promising method, which sheds light on the design of novel structures for the ORR. Conversely, more attention should be paid to the interaction between metal nanoparticles and the carbon supports, as well as the structural characterization of supports including the degree of graphitization and porosity, both of which influence the activity and durability of PtM catalysts. The investigation into the feasibility of utilizing these novel PtM catalysts in practical fuel cell applications to meet the required performance and durability is required.

Another common challenge for the present catalysts is how to bridge the ORR performance gap between RDE and MEA measurements. When used in practical fuel cells, it is often found that these catalysts exhibited less efficient activity than the expected performance according to RDE testing. The reason for this might be lack of versatility since the operating conditions in fuel cells are significantly different from those in RDE tests using aqueous electrolytes. The transport properties for both O₂ and protons are partly restricted in MEAs due to solid electrolyte ionomers and the thicker catalyst layers compared to the case with the electrode in the solution. Furthermore, MEA fabrication technologies such as the ratio of catalyst to ionomer, hot press temperature, and time, as well as bipolar plates and measurement conditions have important effects on the performance of fuel cells. Therefore, efforts should be devoted to optimizing the fabrication of MEAs based on different PtM catalysts according to their unique surfaces and structural characterizations, originating from their specific compositions.

The issue of stability in present PtM catalysts still requires more attention. Although they exhibited excellent durability in laboratory ADTs, the research on their durability in a real fuel cell environment is insufficient and limiting. Furthermore, the degradation mechanism of present Pt-based catalysts in MEA might not be identical to those in solution. Thus far, few studies on the in-situ deterioration of MEAs including catalysts for long-term operation have been reported due to the complicated structure and harsh environment inside operating fuel cells. Except for the degradation of catalysts, the degradation of the ionomer exchange membrane might be another reason for the loss of fuel cell performance. Advanced characterization techniques such as in-situ synchrotron small-angle X-ray scattering, extended X-ray absorption fine structure, and monocrystal surface XRD should be introduced to analyze the problems associated with the structural transformation of nanoalloy catalysts as well as ionomer exchange membranes.

In summary, highly active PtM alloy catalysts are discussed in this review, including disordered and ordered atomic structures. For disordered PtM alloys, the catalysts with specific morphologies such as nanowire and nanoframeworks exhibited particularly high activity and stability in RDE test. However, studies on their practical MEA performance are still needed to judge the potential application in real PEMFC industry. As for ordered intermetallic catalysts, PtCo nanoparticles with core-shell structure exhibited excellent performance not only in RDE but also in MEA test. Some of them exceeded the DOE 2020 activity and stability targets (MA: 0.44 A/mg_Pt, MA loss < 40%, after 30 000 voltage cycles) based on MEA testing. Although considerable progress has been achieved in the catalyst design and preparation, additional work is still required to bridge the gap between the RDE and MEA performance. Most importantly, fundamental theoretical research on ORR pathways and insightful understanding of the active sites are the keys to designing advanced catalysts with high activities and stabilities. Systematic research on the catalyst degradation mechanisms in MEAs during PEMFC operation is in high demand to address stability issues.

X. X. Wang thanks National Natural Science Foundation of China under Grant No. 21805089, Shanghai Natural Science Foundation of China under Contract No. 16ZR1408600, and the Fundamental Research Funds for the Central Universities under Grant No. 222201814024. G. Wu acknowledges the financial support from U.S. Department of Energy, Energy Efficiency and Renewable Energy, Fuel Cell Technologies Office.