Proton exchange membrane fuel cells (PEMFCs) are widely recognized as promising renewable energy sources that can play a role in overcoming the current global energy crisis and address environmental concerns. In practice, oxygen reduction reaction (ORR), which occur at the PEMFC cathode are often limited by high activation barriers, and hence require the presence of an electrocatalyst, which severely hinders PEMFC commercialization [1, 2]. There has been great progress in the development of non-precious metal ORR catalysts [3-5], and carbon-supported Pt catalysts, which hitherto remain the-state-of-the-art ORR catalysts. However, their use is strictly limited by the prohibitive cost and that further improvements to activity/stability are required [6-10]. Nørskov et al. [11] reported that during ORR, the Pt-oxygenated species interaction was very strong, and thus, hindered reactant product desorption. In principle, the binding strength, which is directly related to the d-band center position of Pt catalysts, can be tuned by introducing a second metal, in which the bimetallic alloy results in changes to the Pt electronic state, and concomitantly its interaction. Based on this paradigm, a wealth of interdependent research has focused on developing MPt (M = Fe, Co, Cu, Ni, etc.) catalysts [12-15]. However, these MPt nanocrystals (NCs) are often in a solid-solution structure because of the existing diffusion barrier. In PEMFCs, which operate in acid environments, the M is subject to etching, leaving a low-coordinated Pt surface and structural integrity loss of the NCs.

There are cases where MPt alloys can convert into intermetallic structures according to thermodynamic principles. Compared with disordered alloys, the atoms in intermetallic NCs (iNCs) arrange alternately and interact strongly because of the 3d–5d orbital interactions between M and Pt, thereby preventing transition metal etching. Furthermore, iNCs provide predictable control over structural, geometric, and electronic effects, which allows rationale MPt catalyst design with enhanced performance. In theory, the formation of the intermetallic structure is rather complicated in having multiple influencing variables such as composition, electronegativity, and electron density. The Gibbs free energy of intermetallic (G_{intermetallic}) and disordered (G_disorder) structures can be employed to evaluate the thermodynamic stability of a binary system and mapped onto a phase diagram. If G_{intermetallic} of an MPt alloy is smaller than G_disorder at the working temperature, such an MPt system could be converted into an intermetallic phase through external energy inputs (e.g., temperature and pressure) [16].

MPt iNCs were first explored for ORR in fuel cells in 1994 [17]. Typically, MPt iNCs can be prepared via thermal annealing of disordered NCs above the phase transformation temperature, which can be found from their phase diagrams. However, as a result of the difficulties in synthesizing monodisperse and uniform iNCs without particle agglomeration during high-temperature annealing, only a very limited number of MPt iNCs have been successfully prepared [7, 18, 19]. In this mini-review, we highlight the representative examples of MPt iNCs, i.e., FePt, CoPt, and PbPt, and correlate composition and morphology to catalytic performance.

According to the theory put forward by Nørskov, alloying Pt with a late transition metal can lower the d-band center, and thus decrease the bonding energy resulting in improved activity. The theory is also an effective approach to reduce Pt usage. FePt iNCs have been widely investigated as magnetic materials for data storage and high performance magnets [20, 21]. With exact compositional control (Fe/Pt atomic ratio should be ~1/1 according to the phase diagram), the face-centered cubic (fcc) or L1₀-FePt can be converted to a chemically-ordered intermetallic structure, denoted as face-centered tetragonal (fct) FePt. Recent studies have proven that these ordered fct-FePt NCs can be used as active and stable catalysts towards ORR [22]. Sun et al. [22] prepared ~8 nm disordered fcc-FePt polyhedral NCs by the controlled decomposition of iron pentacarbonyl (Fe(CO)₅) and reduction of platinum acetylacetonate (Pt(acac)₂) (Fig. 1(a)). After supporting on carbon, the nanoparticles (NPs) were annealed at 650 ℃ for 1 h, and the disordered fcc structure was converted to an ordered fct structure, in which Fe and Pt stacked alternately along the c-direction (Fig. 1(b)). After electrochemical dealloying, a FePt/Pt core/shell structure having a ~0.6 nm Pt shell was confirmed by scanning transmission electron microscopy-electron energy loss spectroscopy (STEM-EELS) line scans (Fig. 1(c)). ORR polarization curves revealed that the intermetallic NCs exhibited greater activity than the corresponding disordered analogues (Fig. 1(d)). Density functional theory (DFT) calculations indicated that the Pt–O binding energies (ΔE_O) of fct-FePt is 0.23 eV, smaller than that of fcc-FePt (0.26 eV) and closer to the optimal ΔE_O (0.2 eV). The change in ΔE_O may originate from the compressed surface strain. When fcc-FePt transforms to fct-FePt in the core, the surface strain of Pt was smaller, and thus reduced ΔE_O and promoted ORR activity.

However, one drawback in this case, is that the Pt/Fe atom mobility is strictly confined to the polyhedral precursor structure, resulting in partially-ordered NCs. Recently, we circumvented this problem by using dumbbell-like fcc-FePt-Fe₃O₄ as a precursor to achieve full crystal ordering in monodispersed 8.8 nm fct-FePt NPs (Fig. 2(a), (b)) [7]. In this approach, the Fe₃O₄ comprising the dumbbell-like FePt-Fe₃O₄ (Fig. 2(b)) structure, can be reduced to Fe during gas annealing, which creates defects upon O removal (from Fe₃O₄) and facilitates Fe and Pt reorganization into the fully-ordered fct-FePt (Fig. 2(a)). Magnetic measurements show the fully-ordered fct-FePt NC hysteresis loop to have a single-phase loop with the coercivity reaching 33 kOe, indicating the presence of a perfect L1₀-ordered structure. Compared with the partially-ordered FePt, disordered FePt, and commercial Pt/C, the fully-ordered FePt NPs exhibit the highest E_1/2 (Fig. 2(c)) and excellent stability in 0.1 mol L^-1 HClO₄ (Fig. 2(d)). The stable structure of the fct-FePt NPs is confirmed by high angle annular dark field-STEM (HAADF-STEM) imaging after 20, 000 potential cycles under acidic conditions. The Fe/Pt ordering is clearly indicated by the high (Pt) and low (Fe) Z contrast (Fig. 2(e)), providing evidence that the fct structure is well maintained after potential cycling. Importantly, the composition exhibited negligible change after potential cycling (Fe/Pt: 44/56), indicating the stability of Fe in fct iNCs, even when subjected to acid.

Besides the degree of crystal ordering, another challenge that should be addressed is the MPt iNC size, which is larger than 8 nm in the majority of the reported studies, and results in lower mass activity and increases cost. Relative utilization of Pt and surface area availability of NC catalysts can be significantly enhanced by reducing the NC size. Wang et al. [23] have successfully synthesized 3–4 nm FePt NCs by the galvanic replacement between Fe and Pt⁴⁺ embedded in a carbon matrix (Fig. 3(a)). After annealing, an fct-FePt iNC structure was obtained with well-defined ordering, as clearly indicated by the high (Pt) and low (Fe) Z contrast in the STEM image (Fig. 3(b)). The protection afforded by the amorphous carbon matrix prevented severe aggregation of the FePt NPs during annealing at 900 ℃ (Fig. 3(c)). Compared with the aforementioned FePt iNCs, particle size was significantly smaller (3.6 vs. 8.8 nm), and the measured mass activity was enhanced by a factor of 7.4 relative to the commercial Pt/C catalyst (Fig. 3(d)). A recent study has reported a new approach to synthesize iNC/carbon core/shell catalysts [13]. In this work, FePt iNCs were coated with N-doped carbon shells derived from polydopamine and the obtained catalysts exhibited enhanced activity and good stability. However, in this case, there is difficulty to identify the real active sites and how FePt influences activity.

Similar to FePt, Co is also a transition metal that can improve activity, or as a result of highly optimized tuning of the d-band center, can improve ΔE_O over that of FePt [11]. According to the phase diagram, CoPt can also be converted into iNCs when subjected to appropriate treatment, such as annealing at 700 ℃. The influence of how the CoPt system orders with respect to ORR catalysis has been investigated at the very early stages [17]. However, results show that the ordered CoPt NPs suffer from severe activity loss, compared with only 1% activity loss in the corresponding disordered alloys. Recently, the ordering transformation of Pt₃Co iNCs, which were synthesized using an impregnation-reduction method, demonstrated new results by annealing H₂PtCl₆∙H₂O, CoCl₂∙6H₂O and carbon at 700 ℃ for 2 h [18]. Owing to the unique super periods that are not present in the disordered alloy phase, the presence of the L1₂-ordered intermetallic structure in Pt₃Co can be directly identified from atomic-resolution ADF-STEM imaging (Fig. 4(a)). Unlike the aforementioned L1₀ structure, the projected L1₂ unit cell is composed of a periodic square array of pure Co columns surrounded by Pt atoms along the [001] axis. Among Pt/C, Pt₃Co/C-400 and Pt₃Co/C-700, the latter showed a marked positive shift in ORR E_1/2 of ~70 mV relative to Pt/C (Fig. 4(b)). Furthermore, Pt₃Co/C-700 displayed higher mass activity (Fig. 4(c)) and enhanced stability after potential cycling compared with the corresponding disordered analogue (Fig. 4(d)). However, it should be noted that it is difficult to evaluate the degree of crystal ordering in this work.

Another strategy to enhance ORR electrocatalysis is to develop one-dimensional (1D) nanowires (NWs). NWs provide a higher surface area that interacts with the carbon support compared with NCs. This advantage not only enhances electron transfer between oxygen and the catalyst surface, but also facilitates bonding between the NWs and the carbon support, thus achieving high structural stability and electro-conductivity. However, high-temperature annealing often destroys the NW structure, yielding spherical nanoparticles. Huang et al. [24] circumvented the issue of NW structural integrity loss by employing a seed-mediated method. They reported the synthesis of crenel-like Pt₃Co hierarchical NWs with an L1₂ intermetallic structure, high-index facets, and a Pt-rich surface in oleylamine (OAm) at 160 ℃ for 8 h (Fig. 5(a)). In this reaction, glucose served as a reducing agent to first reduce Pt and subsequently Co, with cetyltrimethylammonium chloride (CTAC) acting as the structure-directing agent. The presence of the two agents resulted in the formation of 2 nm Pt NWs. Thereafter, the newly-reduced Co species diffused into the Pt NWs. After aging for 8 h, intermetallic structures were formed and the ordering of Pt and Co can be identified by HAADF-STEM imaging (Fig. 5(b)). With respect to ORR testing, Pt₃Co exhibited a specific activity of 7.12 mA cm^-2 and a mass activity of 3.71 A mg_Pt^-1 at 0.9 V (Fig. 5(c)). The catalyst exhibits good ORR stability after long-term cycling without sacrificing ORR activity (Fig. 5(d)). This excellent stability is thought to be attributed to the 1D morphology and the intermetallic structure.

Controlling surface strain is an alternative strategy to tune the Pt d-band structure and thus optimize the catalysis. As a result of the lattice mismatch between the surface Pt shell and the core, the surface Pt is usually under a strained state. According to theoretical calculations, it is widely accepted that subjecting the Pt surface to an appropriate compressed strain can improve catalytic activity [25]. Because a Pb atom is larger than a Pt atom, tensile strain would result in a PbPt system and is usually disadvantageous for the catalyst.

However, a recent report by Huang et al. [19] presented unconventional results by applying biaxial strain to intermetallic B8₁ PbPt nanoplates, in which 7.5% tensile strain along the [001] direction and compressed strain along the [110] direction were induced. In the synthesis, ascorbic acid (AA) was used as a weak reducing agent, which was key for NC growth into core/shell nanoplates (Fig. 6(a)). Unlike the aforementioned iNCs, in which high-temperature annealing was required, the transformation temperature is relatively low (160 ℃). The ordered core/shell structures are clearly identified by HAADF-STEM imaging (Fig. 6(b)). The degree of strain was controlled in a different direction as a result of the ordering of the structure, and therefore, systematic illustration was feasible. DFT calculations suggest that an appropriate tensile strain along the [001] direction can weaken the bonding energy between Pt and O at the (110) facets, and thus increase catalytic activity. As a result, PbPt nanoplates/C exhibited the highest catalytic activity among Pt/C, PbPt NPs/C and PbPt nanoplates/C, with a specific activity and mass activity of 7.8 mA cm^-2 and 4.3 A mg_Pt^-1 at 0.9 V, respectively (Fig. 6(c), (d)). After 50, 000 sweeping cycles, there was only a 7.7% loss of mass activity for the PbPt nanoplates (Fig. 6(e), (f)).

In a later report, Huang et al. [26] reported morphological changes from plate to PbPt octahedra simply by reducing the presence of AA (Fig. 7(a)). Because the PtNi(111) facet is suggested to be highly active for ORR [12, 27-30], a PbPt/NiPt core/shell structure with a thin NiPt shell was obtained (Fig. 7(b)). Combining the collective benefits of intermetallic structures, the NiPt shell, with induced tensile strain along [001] direction, the catalysis was greatly improved. Further increasing Ni doping resulted in further improvements to the activity and the PbPt_1.12Ni_0.14 octahedra/C showed the highest activity with a specific activity and mass activity of 5.16 mA cm^-2 and 1.92 A mg_Pt^-1 at 0.9 V, respectively (Fig. 7(c)). Both PbPt and PtPb_1.12Ni_0.14 octahedra/C were relatively stable. However, PtPb_1.12Ni_0.14 octahedra/C displayed enhanced stability compared with PbPt, with less activity loss after potential cycling (Fig. 7(d), (e)).

Intermetallic NCs are receiving ever-increasing attention in electrocatalysis. The iNCs adopt well-defined structure and composition that allow for NC catalyst rationale design. Compared with randomly alloyed NCs, MPt iNCs, e.g., FePt, CoPt, PbPt, exhibit superior ORR catalytic activity because of their optimal ΔE_O at the Pt surface. Because ORR performance of intermetallic MPt catalysts is dependent on numerous factors, including the composition, size and shape of the iNCs, it is still too early to draw the conclusion of which transition metal would best enhance activity and stability after alloying with Pt. Furthermore, the strong interaction between Pt and M in the intermetallic structure helps to stabilize M, and hinders etching when subjected to acids, thus improving long-term stability. However, intermetallic catalyst development is still at an early stage and there remains several challengers to be addressed. Generally, high temperatures are required to overcome the diffusion barrier to produce iNCs. However, such treatment can also destroy the NC morphology, and therefore, controlling the NC shape is difficult, with the majority of scientific studies reporting nanosphere iNC systems. Heading forward, there is a great opportunity to develop new methodologies, through thermodynamic and kinetic control, to better design shape control during the synthesis of intermetallic crystals. Conversely, reducing particle size is a promising approach to improve mass activity, in addition to reducing Pt usage. MPt iNCs typically possess particle sizes > 6 nm, as a result of the challenges to prevent particle agglomeration in sub-5 nm NCs and the ease of phase transformations. Additionally, the reported majority of iNCs remain partially-ordered, which seriously impacts their inherent performance, and therefore, advanced synthetic strategies toward fully-ordered iNCs of various size, composition, and shape are highly desirable. It is of great fundamental and practical significance to continuously investigate this new class of electrocatalysts to improve on the delicate structural features, and concomitantly, the catalytic activity.