Pt-Based Intermetallic Compound Catalysts for the Oxygen Reduction Reaction: Structural Control at the Atomic Scale to Achieve a Win–Win Situation Between Catalytic Activity and Stability

Abstract

The development of ordered Pt-based intermetallic compounds is an effective way to optimize the electronic characteristics of Pt and its disordered alloys, inhibit the loss of transition metal elements, and prepare fuel cell catalysts with high activity and long-term durability for the oxygen reduction reaction (ORR). This paper reviews the structure–activity characteristics, research advances, problems, and improvements in Pt-based intermetallic compound fuel cell catalysts for the ORR. First, the structural characteristics and performance advantages of Pt-based intermetallic compounds are analyzed and explained. Second, starting with 3d transition metals such as Fe, Co, and Ni, whose research achievements are common, the preparation process and properties of Pt-based intermetallic compound catalysts for the ORR are introduced in detail according to element types. Third, in view of preparation problems, improvements in the preparation processes of Pt-based intermetallic compounds are also summarized in regard to four aspects: coating to control the crystal size, doping to promote ordering transformation, constructing a “Pt skin” to improve performance, and anchoring and confinement to enhance the interaction between the crystal and support. Finally, by analyzing the research status of Pt-based intermetallic compound catalysts for the ORR, prospective research directions are suggested.

Graphical abstract

1 Introduction

As a device that can directly convert the chemical energy of fuel oxidation into electrical energy [1], fuel cells exhibit great potential in the field of transportation energy. Compared with a traditional internal combustion engine, as a vehicle power source, fuel cells have the advantages of high efficiency, environmental friendliness, a long driving range, high reliability, no vibration and noise, strong overload capacity, and power changing efficiency, which makes them a research hotspot in the energy field [2, 3]. However, for a long time, the actual application of fuel cells has been directly restricted by their high cost and low performance along with imperfect hydrogen storage technology and inadequate refueling infrastructure [4, 5].

Due to their poor kinetics, the electrode reactions of fuel cells do not occur spontaneously under typical operating conditions; therefore, it is necessary to add catalysts to reduce the activation energy and promote spontaneous electrode reactions. Adsorption is necessary to achieve catalytic reaction. Metal nanocrystals have strong adsorption capacity because of their large specific surface area and high atomic unsaturation; thus, they are considered the best choice for catalysts. Through the explanation of energy band theory and valence bond theory [6,7,8,9,10], it is generally believed that the catalytic capacity of a metal is closely related to its adsorption ability and chemical bond characteristics [11, 12]. Adsorption that is too weak will result in an insufficient number of activated molecules, while adsorption that is too strong will result in intermediates covering the crystal surface and preventing the catalytic reaction [13]. Pt is the commonest active component employed in fuel cell catalysts owing to its moderate adsorption to fuel and oxygen molecules. However, Pt still has many shortcomings. First, Pt catalysts account for a large proportion of the cost of fuel cells, and the rarity and required amount of Pt make its cost stubbornly high. Second, Pt shows good catalytic capacity for the hydrogen oxidation reaction (HOR) in anodes, but the overpotential of the oxygen reduction reaction (ORR) is still large in acidic cathode environments, reaching 300 mV under open-circuit conditions [14]; this high overpotential is one of the important reasons restricting the power generation performance of fuel cells. Third, Pt undergoes dissolution, agglomeration, sintering, and poisoning under the operating conditions [15, 16], which will significantly reduce catalyst performance. In summary, the high cost and insufficient activity and durability of Pt have seriously hindered the commercialization of fuel cells [16]; therefore, a higher requirement for catalytic capacity is proposed as a result of the slow ORR kinetics on Pt cathodes [17].

To solve the above problems of Pt catalysts, two approaches, namely alloying and nanoscale structure design, are adopted. First, many types of nonplatinum [18,19,20] and low platinum [21, 22] catalysts have been developed to reduce the amount of Pt that is used. Among them, the alloying of Pt with transition metals, especially 3d transition metals such as Fe, Co, Ni, Cr, Mn, and Ti, has become an effective means to reduce the cost and improve catalytic activity. In terms of the atomic structure, the Pt–Pt distance on the surface of pure Pt is not ideal for O₂ adsorption, and this adsorption conforms to the Griffiths model. An oxygen molecule is placed transversely on a Pt atom and interacts with its hollow dz² orbital. The interaction between the d electrons of Pt and π electrons of O₂ weakens the O–O bonds, which is conducive to the four-electron reaction process to a certain extent. Different from the single-site adsorption mechanism of pure Pt, according to Vegard’s law, transition metals with smaller atomic radii can shorten the Pt–Pt distance to reach or approach the ideal double-site bridge adsorption distance of the ORR. Therefore, the oxygen molecule can be placed flat on two Pt atoms to simultaneously activate the two oxygen atoms, which is more conducive to the four-electron reaction of O₂ [23]. In terms of the electronic structure, electrons tend to transfer to Pt in alloys formed by Pt and 3d transition metals, which will lower the position of the d-band center of Pt and weaken the adsorption of oxygen-containing intermediates, thus improving the ORR catalytic activity [7, 8, 11, 24].

In addition, because catalytic reactions take place on the crystal surface and near the surface, the crystal morphology and facets significantly affect the catalytic activity and selectivity [25,26,27,28]. Therefore, another feasible way to solve the cost and performance problems of Pt catalysts is to design the structure of the active components of an alloy at the nanoscale. A variety of Pt alloy shape-selective catalysts, such as core–shell [29,30,31,32], nanowire [33,34,35], nanotube [36], nanoframe [37], and polyhedral [38,39,40,41,42] materials, have been developed. For example, the mass and specific activities of core–shell Au/PtCu nanoparticles prepared by Sun et al. [32] reached 1 700 mA mg_Pt⁻¹ and 2.75 mA cm_Pt⁻² at 0.9 V, which are five and six times those of commercial Pt/C, respectively. Sun et al. [35] also obtained a PtFe nanowire catalyst with mass and specific activities of 844 mA mg_Pt⁻¹ and 1.53 mA cm_Pt⁻², which are 5.5 and 4.7 times higher than those of commercial Pt/C, respectively. Owing to the larger active area provided by a nanosegregated open Pt skin structure, the mass and specific activities of the Pt₃Ni nanoframe catalyst developed by Stamenkovic et al. [37] increased by 36 and 22 times compared with those of commercial Pt/C, respectively. The PtNi octahedral catalyst prepared by Huang et al. [40] exhibited 19.3 and 17.6 times the ORR mass activity and specific activity of commercial Pt/C, respectively; this result was due to the more active sites being provided by the highly active PtNi(111) facets. The activities of traditional Pt alloys are usually no more than twice that of commercial Pt/C. Due to special crystal morphology and highly active facets, Pt-based shape-selective catalysts not only show much better performance than commercial Pt/C but also have higher catalytic activity than traditional Pt alloy catalysts.

Although the abovementioned Pt alloy catalysts can indeed reduce cost and improve catalytic activity, the alloying of Pt with Fe, Co, Ni, and other 3d transition metals leads to an excessive compression of the d-band; thus, their ORR activities are still lower than the theoretical optimal value [43]. Moreover, the transition metal atoms in these alloys have a stronger affinity than Pt for oxygen and will gradually migrate to the crystal surface and become ionic states that dissolve in acidic media, resulting in significant changes in the phase composition and facet types near the surface; thus, the performance of these catalysts declines rapidly [44]. Therefore, an important problem in the actual application of fuel cells is to improve the durability of Pt alloy catalysts on the premise of ensuring enhanced catalytic activity. In recent years, some achievements have been made in the field of Pt-based intermetallic compound catalysts. Different from a traditional disordered alloy solid solution, these ordered intermetallic compounds have definite atomic positions and stoichiometry, in which the components form a better combination; thus, these ordered intermetallic compounds can not only further enhance catalytic activity but also effectively inhibit the loss of transition metals. This means that controlling the crystal structure has progressed from the nanoscale to the atomic scale, and structural ordering has become an important method to prepare Pt-based catalysts with high activity and long-term durability.

Herein, the structural characteristics and performance advantages of intermetallic compounds are primarily presented. Then, recent research advances of Pt-based intermetallic compound catalysts for the ORR are reviewed, and the preparation problems and subsequent solutions are summarized. It is believed that this review will help future research in this field.

2 Characteristics and Advantages of Pt-Based Intermetallic Compounds

Intermetallic compounds are metal alloys that have highly long-range ordered superlattice structures at the atomic scale; these structures are formed through a combination of metal bonds and ionic/covalent bonds. Compared with disordered alloys, they usually show higher ORR activity and stability. In this section, the characteristics of the crystal structure and surface composition of Pt-based intermetallic compounds are summarized, and then, the enhancement mechanisms of their ORR activity and stability are explained.

2.1 Structural Characteristics of Pt-Based Intermetallic Compounds

2.1.1 Crystal Structure

The differences between Pt-based intermetallic compounds and disordered alloys are first reflected in the crystal structures [45]. Taking the most common alloys formed by Pt and 3d transition metal M as an example, as shown in Fig. 1, the disordered Pt_xM_1−x (x = 0 − 1) alloy has a face-centered cubic (fcc) structure (the A1 phase, space group Fm–3m), in which Pt and M atoms can randomly occupy eight vertices and six face centers on the cube. According to their atomic ratios, Pt-based intermetallic compounds are generally divided into three types: Pt₃M, PtM, and PtM₃. The Pt₃M and PtM₃ intermetallic compounds still have fcc structures (the L1₂ phase, space group Pm–3m). However, unlike the disordered A1 phase, the Pt atoms occupy six face centers and the M atoms occupy eight vertices in Pt₃M intermetallic compounds, while PtM₃ is the exact opposite. The PtM intermetallic compounds have a face-centered tetragonal (fct) structure (the L1₀ phase, space group P4/mmm), in which the Pt and M atoms are alternately arranged along the c-axis layer by layer. The crystal structure of a ternary PtMN intermetallic compound is similar to that of a binary compound, and the third component, namely the N atoms, randomly occupies the original positions of the M or Pt atoms in the PtM lattice in accordance with the N content. However, order and disorder are only relative concepts. In real intermetallic compounds, the existence of punctual defects and extended defects causes them to deviate from a fully ordered structure. Punctual defects mainly include a vacancy that forms when a position is not occupied by an atom and an antisite that forms when an original position is occupied by another kind of atom [46]. The extended defect is mainly the antiphase domain boundary [47]; that is, in the long-range structure, quantitative translation takes place along the c-axis starting from the cross section of a certain unit cell. The original Pt layer becomes the M layer and vice versa; thus, the phase distribution on both sides of the cross section is the opposite. The ordering degree of the intermetallic compound can be measured by the ordering parameter S² calculated from the peak intensity ratio of (110)/(111) that is obtained by X-ray diffraction (XRD) [48].

Fig. 1

The difference of crystal structures between the Pt-based intermetallic compounds and the disordered alloys

2.1.2 Surface Composition

The surface composition of Pt-based intermetallic compounds is also different from that of disordered alloys [45]. To reduce the Gibbs energy, the different atoms of the alloy spontaneously redistribute between the surface and bulk due to thermodynamics, resulting in the content of a certain atom on the surface being significantly higher; in other words, composition segregation occurs [49]. This phenomenon is most obvious in Pt₃M. In a study, the energy required for moving Pt atoms from the bulk to the surface, namely the surface segregation energy, is expressed by the difference in free energy between the surface and the bulk. The results of this study demonstrate that the surface segregation energy of Pt is negative in most Pt₃M alloys, indicating that Pt segregates onto the surface [50]. Regarding the disordered alloy, composition segregation tends to form a “Pt skeleton” structure consisting of several Pt layers with a lower Pt coordination number on the crystal surface [24, 51]. However, the long-range ordered structure of intermetallic compounds hinders Pt segregation to a certain extent [52], which makes its surface more inclined to form a “Pt sandwich segregation” structure with a single layer of “Pt skin” and an M-rich subsurface [53, 54].

2.2 Performance Advantages of Pt-Based Intermetallic Compounds

2.2.1 ORR Activity

The ORR on a Pt surface is a complex process involving multiple adsorption and desorption reactions [55, 56]. Due to the strong adsorption of Pt, the protonation of the OH intermediate to generate H₂O becomes the rate determining step of the ORR [57]. On one hand, to improve the ORR rate, it is necessary to weaken the adsorption of OH intermediates on the Pt surface so that desorption can occur smoothly. On the other hand, the adsorption strength of OH on the catalyst surface should be suitable to form Pt–OH bonds. The adsorption strength can be measured by the chemisorption energy of intermediates, which is related to the position of the d-band center [24]. Some studies show that within limits, the higher the position of the d-band center, the stronger the chemisorption energy of the intermediate [8, 58]. The changes in the d-band in an alloy are mainly ascribed to the synergy of the geometric and electronic effects. According to Vegard’s law, the lattice parameters of Pt decrease when alloyed with an M atom that has a smaller atomic radius [23], and the compressive strain results in a decrease in the d-band center and a broadening of the d-band width. When the electronegativity of the M atom is lower, the electrons transfer to Pt, which makes the d-orbit of Pt close to its filled state with low energy; thus, its d-band center also decreases. This synergistic effect helps to weaken the chemisorption of intermediates and enhance the ORR activity of a Pt alloy. Unfortunately, the d-band center of Pt cannot be moved to an appropriate position by alloying with common transition metals. For instance, disordered alloys formed with Fe, Co, Ni, and Pt usually produce excessive compression, resulting in a lower d-band center than the theoretical optimal position. Therefore, their ORR activities still need to be further enhanced.

The ORR activity of Pt-based intermetallic compounds is usually superior to that of disordered alloys due to the suppression of excessive compression by its ordered distribution. Regarding L1₀ intermetallic compounds, the layers of Pt and M atoms in the fct structure are alternately arranged along the c-axis and parallel to the a-axis and b-axis. This structure enhances the compression strain along the c-axis but weakens those along the a-axis and b-axis; when taking everything into account, this structure is conducive to weakening the excessive compression [59]. Thus, the ORR activity can be more effectively improved. This explanation is also applicable to the “Pt sandwich segregation” structure on the surface of intermetallic compounds. Taking the L1₀-PtFe intermetallic compound as an example, the Pt–O binding energy on its surface calculated by density functional theory (DFT) is 0.23 eV, while that of the disordered alloy is 0.26 eV. The Pt–O binding energy of the former is closer to the theoretical optimal value of 0.2 eV, demonstrating that its d-band center is slightly increased and the adsorption of intermediates on Pt is relatively enhanced [59]. Regarding L1₂ intermetallic compounds, unlike the random occupation of atoms in the disordered alloys to confine the active sites in some regions, the uniform distribution of active sites in L1₂ intermetallic compounds can weaken the interference of ion repulsion during adsorption by adjacent atoms; thus, this structure is more conducive to improving the utilization of active components. Furthermore, the position of its d-band center is also optimized. Taking L1₂-Pt₃Co as an example [60], the Pt–O binding energies of ordered and disordered Pt₃Co are 0.259 and 0.278 eV according to DFT calculations, respectively, indicating that L1₂-Pt₃Co can effectively inhibit the excessive compression of the d-band that is caused by alloying. In addition, the abilities of intermetallic compounds with different structures to enhance the ORR activity also vary. By comparing the specific activities of ordered and disordered PtM and Pt₃M, Antolini [45] found that the activity improvement ratio of Pt₃M was 1.7 after ordering, while that of PtM reached 2.7, and the activity ratio of ordered PtM to ordered Pt₃M was 1.3. These values demonstrated that L1₀-PtM had better activity than L1₂-Pt₃M, which was attributed to its more suitable Pt–Pt distance.

2.2.2 Electrochemical Stability

The main factors affecting the electrochemical stability of alloys are the loss of M and the dissolution of Pt. The former changes the alloy composition, and the latter decreases the active surface area, which leads to a gradual decrease in ORR activity in the electrochemical environment. Because the formation energy of Pt-based intermetallic compounds is usually 30% more negative than that of disordered alloys [61], they are more thermodynamically stable and less prone to rearrangement and phase separation. Therefore, they show a higher stability under the operating conditions of fuel cells.

The enhanced electrochemical stability of Pt-based intermetallic compounds is first attributed to the ordered arrangement and stable bonding. Because the Pt–M bond energy is higher than that of M–M, the coordination between M and Pt is relatively stabler [62]. In the alternately layered arrangement of L1₀ Pt-based intermetallic compounds, there are Pt layers above and below the M layer; thus, the M atoms can coordinate more with Pt and obtain a higher stability compared with the disordered alloy. For example, the formation energy of vacancies generated by Fe dissolution in L1₀-PtFe is obtained by using first-principles calculations [63]. The average vacancy formation energy of the first and second layers is approximately 0.87 eV, which exceeds that of the disordered alloy and reveals that Fe atoms are more difficult to release in L1₀ intermetallic compounds. Regarding the L1₂ Pt-based intermetallic compounds, the coordination number of the M atom with Pt is also higher than that of the disordered alloy; consequently, the release of M can be effectively inhibited. In addition, the “Pt skin” formed on the crystal surface of the intermetallic compound can also partly prevent the loss of M [64, 65]. For this reason, L1₂-Pt₃M intermetallic compounds with higher Pt contents and more easily formed “Pt skin” structures generally show better electrochemical stability than L1₀-PtM [66, 67].

The size effect also affects the stability of Pt-based intermetallic compounds. Because of a lower coordination number and a stronger oxygen affinity, Pt atoms at the edges and corners are easier to dissolve than those on the surface. The number of atoms at the edges and corners is large for small crystals; therefore, more Pt atoms dissolve. Additionally, they tend to reduce the surface free energy due to their large specific surface area [68], and hence, the dissolved Pt will redeposit on the surface of the other Pt to form a larger crystal [69, 70]. Regarding the special preparation process, the intermetallic compounds generally show larger crystal sizes than the corresponding disordered alloys, which helps to inhibit the dissolution and redeposition of Pt, thus improving the electrochemical stability of crystals to a certain extent [45].

3 Research Advances of Pt-Based Intermetallic Compound Catalysts for the ORR

As mentioned above, the ORR properties of Pt alloys are related to the atomic radius and the electronegativity of M atoms. To enhance the activity, transition metals with smaller atomic radii and lower electronegativities than Pt are usually employed. In addition, a large electronegativity difference between the bonding atoms can reduce the compatibility of different components, which is more conducive to the formation of an ordered structure. In view of this, Pt-based intermetallic compound catalysts for the ORR that are formed with 3d transition metals, especially Fe, Co, Ni, and Cu, are favored and have been thoroughly researched. Moreover, the application of intermetallic compounds formed by Pt and other metals has also been investigated as a cathode in proton exchange membrane fuel cells (PEMFCs). In this section, the recent research advances of Pt-based intermetallic compound catalysts for the ORR will be introduced in terms of element types, and their preparation conditions and ORR properties are summarized in Table 1.

Table 1 The as-reported Pt-based intermetallic compound ORR catalysts

3.1 PtFe-Based Intermetallic Compounds

The diffusion coefficient of Fe in an Au substrate is greater than those of Co and Ni [71]; therefore, it can be considered that Fe more easily diffuses in Pt, which has a similar structure to Au. Therefore, compared with that of Co and Ni, the ordering transformation of a PtFe alloy occurs relatively easily, promoting its further study.

Sun et al. [72] prepared a disordered PtFe alloy by reducing Pt(acac)₂ and Fe(CO)₅ with the oleylamine/oleic acid method, and then, an L1₀-PtFe intermetallic compound was obtained through heat treatment at 500 °C. This L1₀-PtFe catalyst exhibited high ORR activity and low Fe loss. After optimizing the heat treatment temperature, they found that the L1₀-PtFe prepared at 650 °C showed no obvious sintering, and its specific activity at 0.531 V vs. Ag/AgCl reached 2.1 mA cm_Pt⁻², which is more than two times that of a disordered PtFe alloy [59]. Xia et al. [73] synthesized two kinds of disordered PtFe alloys with different components using H₂PtCl₆ and FeSO₄ as precursors, and then, L1₀-PtFe and L1₂-Pt₃Fe intermetallic compounds were prepared by annealing. They investigated the effect of annealing temperature on the ordering transformation. As presented in Fig. 2a, b, the PtFe ordering transformation process began at 400–450 °C, and intermetallic compound nanoparticles with sizes of less than 5 nm and good dispersion could be formed at 500–600 °C. Compared with commercial Pt/C, the two intermetallic compound catalysts exhibited excellent ORR activity, which is attributed to their geometric structure and electronic properties. After an accelerated durability test (ADT), the specific activity attenuation of L1₂-Pt₃Fe was lower, while the pure Pt phase was observed in L1₀-PtFe, indicating that the former could more effectively prohibit the release of Fe.

Fig. 2

XRD patterns of a L1₂-Pt₃Fe and b L1₀-PtFe intermetallic compounds. Reproduced with permission from Ref. [73]. Copyright 2012, Royal Society of Chemistry

The ORR activity and stability of the PtFe alloy can be further optimized by adding an appropriate third element to the binary system. As described in Sect. 2.2.1, although L1₀-PtFe is beneficial for relieving excessive compression strain, there is still a gap from the best position of the d-band center. Theoretically, it is advantageous to partially offset the decrease in the d-band center and enhance the ORR activity by adding an appropriate amount of an element with a larger atomic radius or stronger electronegativity than Fe [8, 24, 43, 59]. Fe is easily dissolved in an electrochemical environment due to its relatively low electronegativity. Therefore, the addition of an appropriate element can prevent the continuous release of internal Fe atoms and obviously improve the stability of PtFe intermetallic compounds by partly replacing the Fe with a more electronegative element. The above strategies have been successfully applied to prepare various ternary PtFe-based intermetallic compound ORR catalysts. By virtue of a slightly larger atomic radius and stronger electronegativity, Cu has advantages in regulating the properties of PtFe intermetallic compounds. Sun et al. [59] obtained an L1₀-PtFeCu intermetallic compound by doping Cu and annealing a PtFe alloy. As shown in Fig. 3a, b, its specific activity was further improved compared with that of L1₀-PtFe. The DFT calculation results confirmed that the Pt–O binding energy on the surface of L1₀-PtFeCu was 0.22 eV, which is closer to the optimum value of 0.2 eV than that of L1₀-PtFe. Therefore, it was proven that Cu doping further weakened the excessive compression of the d-band of Pt. Yamaguchi et al. [74] prepared an L1₀-Pt₂Fe₁Cu₁/C intermetallic compound catalyst by H₂ reduction at 800 °C for 4 h. Its ORR mass activity and specific activity were 2.5 and 4 times higher than those of commercial Pt/C (TKK), respectively. After a 10 000-cycle durability test at 60 °C, due to the lower losses of Fe and Cu in L1₀-PtFeCu, the retention rates of its mass activity and electrochemical active surface area (ECSA) were over 70%. In contrast, these values were only approximately 40% for L1₀-PtFe because of the destruction of the ordered structure caused by Fe loss, as shown in Fig. 3c, d. The power generation performance and durability of this compound as a PEMFC cathode were also subsequently evaluated [75]. The membrane electrode assembly (MEA) fabricated using the L1₀-PtFeCu catalyst not only exhibited a much higher exchange current density than the commercial control but also had lower performance degradation after a 200-h durability test. Therefore, it is suggested that the addition of Cu can significantly enhance the ORR performance of PtFe intermetallic compound catalysts, especially their durability. Mo, Cr, Au, and other elements have also been employed to optimize the properties of PtFe intermetallic compounds owing to their larger atomic radii and better stabilities in acid solution. Wang et al. [76] obtained a Mo-PtFe intermetallic compound containing 2.5 at.% (at.% means atomic percentage) Mo by annealing at 600 °C for 1 h with a ternary disordered alloy as the precursor. Its ORR activity was significantly better than that of the PtFe intermetallic compound prepared with the same process, and its mass activity was still 94.07% of the initial value after 5 000 cycles. The enhanced ORR performance of the Mo-PtFe intermetallic compound was attributed to the addition of an appropriate amount of Mo to inhibit excessive crystal growth and optimize the oxygen binding energy. Zhu et al. [77] synthesized Au-PtFe and Au-PtFe₃ intermetallic compounds with 0.2 wt.% (wt.% means the weight percentage) Au by microwave-assisted polyol coreduction and a subsequent annealing treatment. The ORR activities of the two Au-doped samples were significantly higher than those of the binary controls, which benefitted from Au regulating the surface electronic structure. In addition, Au-PtFe/C and Cr-PtFe/C intermetallic compounds were also prepared by surface doping the binary systems [78]. Their ORR mass activities were improved in comparison to the undoped system, and the activity of Cr-PtFe/C was the highest, while the electrochemical stability of Au-PtFe/C was better than that of Cr-PtFe/C.

Fig. 3

a High-angle annular dark field–scanning transmission electron microscopy (HAADF–STEM) image and b linear sweep voltammetry (LSV) curve of L1₀-PtFeCu. Reproduced with permission from Ref. [59]. Copyright 2014, American Chemical Society. The attenuations of c ECSA and d mass activity of L1₀-PtFeCu in durability cycles. Reproduced with permission from Ref. [74]. Copyright 2015, American Chemical Society

Although the ordered structure of a PtFe intermetallic compound can inhibit Fe release to a certain extent, Fe loss cannot be completely avoided, even after doping to further improve the stability of the compound. In particular, the Fe release can be accelerated during startup/shutdown or under high-potential conditions in actual fuel cells [75]. The dissolved Fe²⁺ can react with H₂O₂, an ORR byproduct, to form hydroxyl radicals with strong oxidizability. These radicals oxidize and decompose the macromolecular organic compounds in the electrolyte membrane and catalyst layers, which eventually leads to fuel cell failure. Therefore, the application of PtFe-based intermetallic compound catalysts for the ORR is greatly limited.

3.2 PtCo-Based Intermetallic Compounds

Co has a smaller atomic radius and higher electronegativity than Fe. Thus, the decrease in the Pt d-band center of a PtCo alloy is slightly less than that of a PtFe alloy, resulting in its ORR activity being higher. When the heat treatment temperature is above the order–disorder phase transition temperature, a disordered structure will still be formed during the cooling process. According to the phase diagram, the order–disorder phase transition temperatures of PtCo and Pt₃Co are approximately 850 and 750 °C, respectively [79], which are above the normally used heat treatment temperatures; thus, PtCo intermetallic compounds with high stability can be obtained under the usual preparation conditions. As mentioned above, excellent catalytic performance and relatively low preparation difficulty make PtCo intermetallic compounds promising ORR catalysts in this field.

Abruña et al. [80] prepared Pt₃Co intermetallic compounds with 2–3 “Pt skins”, as shown in Fig. 4a, b, by annealing the disordered alloy at 700 °C for 2 h in a H₂ atmosphere. The electrochemical test results are shown in Fig. 4c, d. Compared with the disordered alloy and the commercial Pt/C catalyst, the mass activity and specific activity of the Pt₃Co intermetallic compound increased by more than two and three times, respectively, and its activity loss was also lower after 5 000 cycles. Its high activity and high stability were attributed to the Pt-rich shell and stable Pt₃Co intermetallic core. On this basis, to reduce the thickness of the cathode catalyst layer and the mass transfer resistance, a 40 wt.% Pt₃Co intermetallic compound catalyst was also synthesized along with an intermetallic compound catalyst containing 20 wt.% Pt as a seed by seed-mediated growth and a subsequent annealing treatment [81]. Despite the 40 wt.% Pt₃Co intermetallic compound catalyst having a larger crystal size, its mass activity and ECSA were similar to those of the 20 wt.% Pt seed; however, its specific activity increased by two times. After a 4 000-cycle stability test, a slight attenuation was only observed in its half-wave potential and ECSA. The ordering transformations of PtCo alloys with different Co contents were realized by Murray et al. [82]. Their test results demonstrated that the ordering degree and ORR performance were strongly dependent on the annealing temperature and Co content. The PtCo₃/C intermetallic compound annealed at 650 °C for 4 h showed the highest ordering degree, mass activity, and specific activity, while the stability of PtCo/C obtained under the same conditions was the best, showing only an 8 at.% Co loss after testing.

Fig. 4

a HAADF–STEM image, b XRD pattern, c LSV curve and d its change after ADT of L1₂-Pt₃Co. Reproduced with permission from Ref. [80]. Copyright 2015, Springer Nature

Ternary PtCo intermetallic compounds have also been investigated. The geometric structure and electronic characteristics of Pt can be further optimized by doping a third element into the system, such as Cu and Au. Zhu et al. [83] prepared PtCoCu/C intermetallic compounds with different Cu contents by two-step annealing at 200 and 700 °C in an Ar/H₂ atmosphere. The mass and specific activities of PtCo_0.75Cu_0.25/C were significantly higher than those of the PtCo/C and Pt/C (JM) catalysts and reached 1.31 A mg_Pt⁻¹ and 0.59 A cm_Pt⁻², respectively; this result was attributed to the changes in the Pt–Pt distance and surface electronic characteristics that resulted from the ordered phase and alloy composition. Notably, Cu doping also optimized the surface electronic structure, and thus, the stability of the catalyst was further improved. Chen et al. [84] obtained L1₀-PtCoAu by annealing an Au-doped PtCo disordered alloy. Its mass activity and specific activity increased to three and two times those of the Pt/C catalyst, and the performance was almost unchanged after a 10 000-cycle ADT. An L1₀-PtCoW catalyst with 18 times the mass activity and 19.5 times the specific activity of commercial Pt/C was successfully developed by Li et al. [85], and its mass activity remained 86.4% of the initial value after 10 000 cycles. As presented in Fig. 5, DFT calculations were employed to investigate the above Au- and W-doped L1₀-PtCo. The results demonstrate that Au and W atoms play a role in stabilizing the ordered structure and optimizing the Pt lattice; thus, the electrocatalytic activity and durability can be improved.

Fig. 5

The results of DFT simulations and calculations of L1₀-PtCoAu. Reproduced with permission from Ref. [84]. Copyright 2018, American Chemical Society

3.3 PtNi-Based Intermetallic Compounds

Compared with Co, Ni has the advantages of a higher reserve, wider distribution and lower cost. In addition, the activity of the PtNi alloy is also not the lowest when compared to those of PtCo and PtFe according to the “volcano” plots of ORR activity [43]; in particular, the Pt₃Ni(111) facet has been proven to be closest to the peak of the “volcano” plots. Therefore, PtNi has great potential as an intermetallic compound in theory. However, compared with those of Fe and Co, the lower diffusion coefficient of Ni in Pt [71] makes the ordering transformation of a PtNi alloy more difficult, resulting in higher requirements for the preparation conditions. Theoretically, a higher treatment temperature should be adopted to promote atomic diffusion in PtNi alloys. Unfortunately, the order–disorder phase transition temperature of PtNi is only approximately 650 °C, and those of PtNi₃ and Pt₃Ni are lower [86]. Thus, it is almost impossible to form stable PtNi intermetallic compounds above this temperature, while also making it impossible to promote the ordering transformation by increasing the temperature. The above two contradictory reasons raise the difficulty of preparing PtNi intermetallic compounds and lead to relatively rare research results when compared with PtFe and PtCo.

Wang et al. [87] prepared a Pt₃Ni₂/C intermetallic compound catalyst by H₂ solid-phase reduction and a subsequent annealing at 600 °C. Its specific activity is up to four and three times those of Pt/C and the disordered control, respectively, and the retention of its activity is also better after a 10 000-cycle ADT. Yang et al. [88] developed a carbon black-supported L1₀-PtNi intermetallic compound catalyst by annealing disordered PtNi/C. Compared with the disordered Pt/C and commercial Pt/C, the specific activity at 0.85 V vs. reversible hydrogen electrode (RHE) of L1₀-PtNi/C increased by three and six times, respectively, and its activity loss was the lowest after 5 000 cycles. They also obtained slightly active and stable L1₀-PtNi catalysts using graphene and multiwalled carbon nanotubes as supports by annealing aerogels of the supported PtNi alloys at 450 °C [89]. To overcome the ordering difficulty of PtNi alloys, taking into consideration their low phase transition temperature, PtNi alloys usually need to be treated for a long time at temperatures that are not too high, which is unfavorable to optimizing the crystal size and activity. For example, the annealing time used in the above work of Yang et al. [88] was as long as 16 h; during this time, the particle size of L1₀-PtNi increased by three times and the ECSA decreased by 60% due to particle sintering. Therefore, the specific activity improved, but its mass activity showed little improvement compared with the disordered control.

3.4 PtMN (M, N = Fe, Co, Ni) Intermetallic Compounds

As the most common 3d transition metal elements that can form Pt-based intermetallic compounds, Fe, Co and Ni can not only be added alone but also be used together to form ternary or quaternary samples. As a result of the synergistic effect among Fe, Co, and Ni, these ternary and quaternary intermetallic compounds often have different ordering transformation processes, crystal structures, and catalytic performance than the corresponding binary compounds.

Gan et al. [90] systematically investigated the ordering transformation of PtFeCo, PtNiCo, and PtFeNi alloys with the same Pt content during high-temperature annealing, and their crystal structure changes are shown in Fig. 6. Thanks to the fast diffusion of Fe and the effective sintering resistance of Co, it was found that the PtFeCo intermetallic compound obtained by annealing at 600 °C exhibited the highest ordering degree and the smallest particle size; consequently, it had the best ORR activity and stability. Its mass activity at 0.9 V vs. RHE was four times that of a pure Pt catalyst, and its performance decreased by only 16% after 10 000 cycles. Yamaguchi et al. [67] prepared an L1₀-Pt₂FeCo catalyst by a simple solid-phase impregnation method. This catalyst exhibited 2.5 times the ORR activity of the disordered control. After a 5 000-cycle ADT, the retention rates of its mass activity and ECSA were more than 80%. The scanning transmission electron microscopy–energy-dispersive X-ray spectroscopy (STEM–EDS) results demonstrated that L1₀-Pt₂FeCo remained a chemically ordered structure after the durability test, and the Fe and Co contents were similar to those before the test; this result revealed that the L1₀-Pt₂FeCo intermetallic compound effectively hindered the release of Fe and Co. The ORR properties of the L1₀-Pt₂FeCo and L1₂-Pt₆FeCo intermetallic compounds were further compared under actual PEMFC operating conditions [91], and the activity and durability of the former were superior to those of the latter. Zhu et al. [92] synthesized a well-dispersed L1₂-Pt₂Fe₃Ni₃ intermetallic compound catalyst with a particle size of 5 nm and a high transition metal content by the microwave-assisted polyol method and a subsequent annealing treatment at 675 °C. Resulting from the strong diffusion of Fe to partly offset the ordering difficulty of Ni, the ordering transformation of this PtM₃ ternary alloy that contains Ni occurred at a low annealing temperature. Its mass activity reached up to 3.5 times that of commercial Pt/C, and its stability was higher. Beyond PtMN (M, N = Fe, Co, Ni) ternary intermetallics, Tang et al. [93] were the first to develop a series of L1₂-PtFeCoNi quaternary intermetallic compounds, as illustrated in Fig. 7a. The same atomic ratio of Pt, Fe, Co, and Ni was used while varying the annealing temperature; additionally, a combination of spraying dehydration and annealing processes was performed. As presented in Fig. 7b, c, their ORR mass activities at 0.9 V vs. RHE were up to 6.6 times that of commercial Pt/C and showed minimum mass and specific activity losses of only 17% and 1.5% after 10 000 cycles, respectively. The enhanced stability was attributed to the synergistic effect of the particle size and ordered structure.

Fig. 6

XRD patterns of a PtNiCo, b PtCoFe, and c PtFeNi ternary alloys after annealing. Reproduced with permission from Ref. [90]. Copyright 2015, MDPI

Fig. 7

a Atomic-scale high resolution transmission electron microscopy (HRTEM) image, b LSV curves, and c the comparison of LSV curves before and after ADT of L1₂-Pt(FeCoNi)₃. Reproduced with permission from Ref. [93]. Copyright 2019, American Chemical Society

3.5 Pt-Based Intermetallic Compounds Composed of Other Elements

In addition to Fe, Co, and Ni, intermetallic compound catalysts for the ORR formed by Pt and other 3d transition metals, such as Cu, Ti, Cr, Mn, and Zn, have also been investigated. Gaberšček et al. [94, 95] prepared PtCu₃ intermetallic compounds by an improved sol–gel method that had five steps: xerogel preparation, pyrolysis, platination, partial oxidation, and annealing. Its structure and LSV curve are shown in Fig. 8. Due to the ordered structure, “Pt skin”, and enhanced binding between the alloy and support, its specific activity and mass activity increased by 3.5 and 2.5 times compared with the targets of the U.S. Department of Energy; furthermore, the ECSA showed almost no change after 7 000 cycles. In the works of McGinn et al. [96] and Lee et al. [97], Pt₃Ti intermetallic compounds were obtained by modifying a Pt/C catalyst with titanium acetylacetonate and TiO₂, respectively. The activities and stabilities of the two Pt₃Ti samples increased in comparison with those of commercial Pt/C. The effect of the H₂ content in a heat treatment atmosphere on the structure and properties of the Pt₃Ti intermetallic compound was also investigated. It was proven that the sample had the best ordered structure and ORR performance was achieved with 10% H₂ in the heat treatment atmosphere. Yang et al. [65] annealed a disordered alloy in the range of 700 to 800 °C in a H₂ atmosphere to form a Pt₃Cr intermetallic compound. Because the particle size increased from 3.1 to 7.2 nm after annealing, its mass activity was slightly lower than that of the disordered particles. However, the mass activity of the disordered control decreased by approximately 50% after a 5 000-cycle ADT, while this attenuation in the Pt₃Cr intermetallic compound could be ignored. The transformation of the Pt₃Mn alloy from the A1 phase to the L1₂ phase was realized by annealing at 600 °C for 30 min by Murray et al. [98]. The ORR mass activity at 0.8 V vs. RHE of the L1₂-Pt₃Mn catalyst was above three times that of Pt/C (E-TEK). Li et al. [99] prepared an L1₀-PtZn catalyst by annealing at 600 °C for 2 h in a H₂/Ar atmosphere with Pt nanoparticles embedded in ZnO as the precursor. Its mass activity at 0.9 V vs. RHE reached 0.52 A mg_Pt⁻¹, and the loss of mass activity was only 16.6% after 30 000 cycles. According to the DFT calculation results, the enhanced ORR activity of L1₀-PtZn benefitted from the biaxial strain, which could regulate the Pt–Pt distance and optimize the Pt–O binding on the surface; moreover, the excellent stability was ascribed to the increase in the vacancy formation energy of Zn atoms in the ordered structure.

Fig. 8

a STEM–HAADF image and b LSV curve of L1₂-PtCu₃. Reproduced with permission from Ref. [94]. Copyright 2014, Royal Society of Chemistry

Although 4d transition metals and Pt can theoretically form intermetallic compounds, it is difficult to experimentally obtain their ordered structure under normal conditions because of their ordering temperatures are too low (usually below room temperature). Therefore, the as-reported alloys formed by Pt and 4d transition metals are almost disordered [100]. Herein, they will not be introduced in detail.

Intermetallic compounds composed of Pt and 4f transition elements, namely lanthanides, have also been systematically investigated. Escudero-Escribano et al. [101] analyzed the ORR properties of Pt₅M (M = lanthanide elements, La, Ce, Sm, Gd, Tb, Dy, and Tm) intermetallic compounds through experiments and theoretical calculations. Their specific activities at 0.9 V were three–six times higher than that of a Pt catalyst. Different from the gradually enhanced lanthanide contraction with increasing atomic numbers, their ORR activities in descending order were Pt₅Tb, Pt₅Gd, Pt₅Sm, Pt₅Dy, Pt₅Tm, Pt₅Ce, and Pt₅La. The relationship between their activity and atomic number resulted in a “volcano” plot where the apex was between Pt₅Tb and Pt₅Gd. The DFT calculations showed that this phenomenon could be explained by using the OH binding energy ΔE_OH. On the left of the “volcano” plot, the interaction between the Pt₅M samples and OH was too weak, while the binding on the right was too strong; therefore, their ORR activities showed the above trend.

Because post-transition metals and Pt can naturally form intermetallic compounds and will not form disordered alloys, intermetallic compounds composed of Pt and post-transition metals such as Ga, In, Pb, and Bi have begun to attract more attention. Gedanken et al. [102] synthesized a graphene-supported PtGa intermetallic compound by dissolving molten Ga into H₂PtCl₆ aqueous solution and then performing ultrasonic treatment for 6 min. The catalyst was a mixture of Pt₃Ga and Pt₂Ga. Its mass activity and specific activity at 0.85 V vs. RHE increased to 1.8 and 2.3 times those of commercial Pt/C, and the half-wave potential negatively shifted by 22 mV after 5 000 cycles; this result was less than the attenuation of 36 mV for Pt/C under the same conditions. As illustrated in Fig. 9, InCl₃ and H₂PtCl₆ were used as precursors and annealed at 700 °C for 2 h in a H₂/Ar atmosphere to obtain a Pt₃In intermetallic compound by Gu et al. [103]. This compound exhibited a 4.1-fold increase in the mass activity and a 2.7-fold increase in the specific activity when compared with commercial Pt/C at 0.9 V vs. RHE. After a 20 000-cycle ADT, its activity decay and structure change could be ignored. The oxygen binding energy on the Pt₃In(111) facet calculated by DFT was 0.33 eV lower than that of pure Pt, so it had better ORR activity. Matsumoto et al. [104, 105] doped Pb into Pt nanoparticles by reducing Pb(CH₃COO)₂ with ethylene glycol to prepare a PtPb intermetallic compound catalyst. The half-wave potential of the PtPb intermetallic compound increased by nearly 50 mV compared with that of the Pt/C catalyst. Guo et al. [106] prepared a PtBi intermetallic compound nanoplate catalyst with 5.2 times the specific activity of the Pt/C catalyst by an oleylamine reduction method. Although the ECSA was smaller, its mass activity still reached more than two times that of the Pt/C catalyst. After 5 000 cycles, its ECSA was 90% of the initial value, and only a slight decrease in the specific activity was observed after a small amount of Bi dissolution.

Fig. 9

a, c, e STEM images, and b, d, f corresponding fast Fourier transform (FFT) of individual Pt₃In intermetallic compound particle along different zone axes. g DFT calculation of oxygen binding energy on Pt₃In (111) facet. h ORR activity and i the comparison of LSV curves before and after ADT of Pt₃In intermetallic compound. Reproduced with permission from Ref. [103]. Copyright 2019, John Wiley & Sons, Inc

In the abovementioned Pt-based intermetallic compound catalysts for the ORR, to explain the mechanism of activity enhancement, some researchers have calculated and analyzed the structure–activity relationship of intermetallic compounds with specific composition on the basis of the crystal structure model built with DFT. In addition, DFT has also been applied to predict the ORR activities of Pt-based intermetallic compound catalysts with unknown composition. Rankin et al. [107] established a new fitting model based on DFT to study the oxygen adsorption capacity of Pt₃(MN)₁ intermetallic compounds, in which the distance between metals and peaks in ORR “volcano” plots was normalized as a function of the valence state and atomic weight. It was predicted that Pt₃(MN)₁ intermetallic compounds might have excellent ORR activity when the MN elements were RhPd, RuPd, RhIr, RhOs, RuOs, CuOs, and so on. This work provides inspiration and reference for selecting appropriate elements and expanding the types of Pt-based intermetallic compounds.

4 Problems and Improvements in the Preparation of Pt-Based Intermetallic Compounds

Although many kinds of Pt-based intermetallic compound catalysts with enhanced ORR activity and durability have been developed, there are still some problems during their preparation process. For instance, they easily sinter at high temperatures, have difficulty undergoing ordering transformation, require further improvements in performance, and demonstrate weak interactions between the crystal and support; these problems are unfavorable to improving performance and enriching the types of Pt-based intermetallic compounds. Therefore, they have been studied. Herein, the above problems and subsequent research advances for overcoming them will be introduced in this section, and the related results are summarized in Table 2.

Table 2 Problems and improvement in preparation of Pt-based intermetallic compound catalysts

4.1 Size Growth and Sintering—Crystal Size Control

Because the growth kinetics of metal crystals are much faster than the ordering kinetics, the nanocrystals synthesized at room temperature are nearly disordered. To change from an initial disordered state to an ordered state, a very high activation energy barrier in regard to interdiffusion must be overcome. At present, the main preparation methods of Pt-based intermetallic compounds include liquid-phase reduction, solid-state annealing, and vapor deposition. An ordered structure can be obtained at a lower temperature in the liquid-phase reduction method, but it is usually necessary to select specific precursors and solvents; thus, the product shows a lower ordering degree due to the reaction temperature being lower than the ordering transformation temperature [108, 109]. In the vapor deposition method, a second metal component M must be easily evaporated when heated, and the deposition amount of M atoms is greatly affected by temperature and time. Therefore, it is difficult to control M deposition to meet the requirements of intermetallic compounds for a specific stoichiometry [110]. Compared with the above two methods, the solid-state method can realize a specific stoichiometry by adjusting the elemental ratio of the disordered alloy before annealing; moreover, its high temperature is more conducive to the ordering transformation. Therefore, solid-state annealing is the most widely employed method to prepare Pt-based intermetallic compound catalysts. However, a high temperature or long treatment time is usually necessary to obtain a high ordering degree during annealing, which will lead to particle growth and sintering that decreases the specific surface area and results in a poor dispersion of the catalyst; the above behavior affects the ORR performance of the catalyst. Based on the as-reported work, the physical barrier effect of coatings including metal oxides, carbon shells, and inorganic salts can effectively control the particle size and avoid sintering during annealing.

4.1.1 Metal Oxide Coating

In the work of Sun et al. [111, 112], PtFe/Fe₃O₄ nanoparticles were first coated by using Mg(acac)₂ as the precursor, and then, MgO-coated PtFe/Fe₃O₄ was annealed at 650–700 °C for 6 h in a H₂/Ar atmosphere. The L1₀-PtFe intermetallic compound was obtained after the MgO coating was removed by acid etching for 20 min in 0.1 M (1 M = 1 mol L⁻¹) HClO₄. Compared with the nanoparticles before heat treatment, the MgO-coated sample exhibited no discernible change in crystal size after annealing and thus exhibited excellent ORR activity and stability. Its half-wave potential reached 0.945 V vs. RHE, which was a shift in the positive direction by 82 mV relative to that of commercial Pt/C; moreover, there was no obvious activity loss after a 5 000-cycle ADT. Using the preparation procedure in Fig. 10a, Wei et al. [113] prepared the L1₀-PtFe intermetallic compound in Fig. 10b, c by annealing the SiO₂-coated disordered alloy and washing with 0.2 M HF. The L1₀-PtFe catalyst showed uniform dispersion and a narrow size distribution. These results demonstrated that the SiO₂ coating could effectively prevent the sintering of alloys at a high temperature. Therefore, L1₀-PtFe exhibited 6.73 times the mass activity and 6.55 times the specific activity of commercial Pt/C, as well as better stability. Joo et al. [60] successfully prepared a Pt₃Co intermetallic compound nanowire catalyst with SiO₂ as both a hard template to control the crystal morphology and a coating to prevent particle agglomeration. Due to the enhanced ligand effect and agglomeration-resistant structure, its ORR activity and durability further improved after the sample was washed with 10 wt.% HF solution to etch the template. Lee et al. [114] introduced aluminosilicate (xAl₂O₃·ySiO₂) to control the particle size of Pt₃Co intermetallic compounds and employed HF etching to remove aluminosilicate or silica. Owing to a strong interaction formed by the charge transfer from Pt to aluminosilicate, the sintering of Pt₃Co nanoparticles was effectively inhibited, and the average crystal size was controlled below 5 nm. Therefore, its mass activity and specific activity reached 3.25 and 2.59 times those of the Pt/C catalyst, respectively. Moreover, due to the additional MgO and SiO₂, the metal oxide formed by the second component could also directly control the crystal size; therefore, the acid etching step could generally be omitted. For example, in the work of Lee et al. [97], TiO₂ played a dual role as a Ti source and coating, and the average particle size of the as-prepared Pt₃Ti intermetallic compound remained 4.2 nm after heat treatment. Li et al. [115] prepared a series of PtNiCo intermetallic compound catalysts by annealing the Pt@NiCoO_x core–shell structure precursors with different Ni and Co contents at 600 °C. The NiCoO_x shell effectively prevented sintering and agglomeration and ensured a small crystal size and good dispersion. Among them, L1₀-PtNi_0.8Co_0.2 exhibited the best performance. Its mass and specific activities were 23 and 19 times higher than those of commercial Pt/C, and the performance curves showed almost no change after a 10 000-cycle ADT.

Fig. 10

a Schematic illustration of a SiO₂ coating preparation process, b HRTEM and c HAADF–STEM images of L1₀-PtFe catalyst. Reproduced with permission from Ref. [113]. Copyright 2019, Royal Society of Chemistry

Although the metal oxide coating method can inhibit crystal sintering, high-temperature resistant metal oxide coatings on the surface increase the difficulty of atomic diffusion and reduce the ordering degree. In particular, as mentioned in the above examples, the additional metal oxide usually needs to be removed by acid leaching, which easily causes pollution and changes in the surface composition of the catalyst.

4.1.2 Carbon Shell Coating

The oxidative self-polymerization of dopamine can occur on the surface of nearly any solid to form a polydopamine (PDA) coating under specific conditions; because of this behavior, it has been widely used for material surface functionalization in various fields [116]. Therefore, it is not surprising that PDA coatings can be employed to control the sizes of intermetallic compounds. In the work of Wang et al. [117], PDA-coated disordered alloys were directly annealed at 900 °C in Ar for 30 min. A Pt₃Co/C intermetallic compound with a size of 4 nm was then obtained after further heat treatment, at 350 °C in air for 30 min, to remove the carbon shell. Compared with the control prepared without the coating, the coated compound showed a similar stability, a higher ECSA and a sevenfold higher ORR activity. As presented in Fig. 11, the size control of a PtFe intermetallic compound was also realized with a PDA coating by Hyeon et al. [63]. The thickness of the carbon shell could be controlled below 1 nm by precisely adjusting the amount of PDA, and thus, a small crystal size and high activity could be achieved without removing the carbon shell. The mass and specific activities of the L1₀-PtFe catalyst prepared by annealing at 700 °C for 2 h were up to 11.4 and 10.5 times those of commercial Pt/C. In addition to PDA, other organic compounds have also been studied as carbon sources for coatings. For instance, Wang et al. [118] prepared a fully ordered PtFe intermetallic compound catalyst embedded in carbon nanocages with C₂H₂ as the carbon source. Because the metal crystals were blocked by the carbon nanocages, the crystal size of L1₀-PtFe was only 3.6 nm after heat treatment at 900 °C. Its ORR activity was 8–10 times higher than that of the Pt/C catalyst, and its structure and activity were only slightly changed after the ADT. Subsequently, an Au-PtFe intermetallic compound with good size and dispersion was also obtained using the same process [119]. With hexadecanediol as the organic carbon source of the coating, an L1₂-Pt₃Fe intermetallic compound with a crystal size of 6.77 nm was obtained by annealing at 700 °C by Lee et al. [120]. The size of the uncoated sample increased to 10.74 nm for comparison, indicating that the carbon coating effectively prevented agglomeration and ensured enhanced ORR activity and durability.

Fig. 11

a Schematic illustration of a PDA coating preparation process, b, c transmission electron microscopy (TEM), d high resolution powder diffraction (HRPD), e HAADF–STEM, f model, and g FFT images of L1₀-PtFe catalyst. Reproduced with permission from Ref. [63]. Copyright 2015, American Chemical Society

To ensure that the carbon shell does not hinder mass transfer but prevents particle sintering, it is necessary to accurately control the amount of the coating material to obtain a carbon shell with moderate thickness; this control remains difficult to achieve. Another method is heating in an air atmosphere after ordering to remove the carbon shell, as described in the work of Wang et al. [117]; however, this method affects the structure of the catalyst support.

4.1.3 Inorganic Salt Coating

In the original inorganic salt coating method, an inorganic salt with a high melting point is directly mixed with the as-prepared disordered alloy nanoparticles. For example, as early as 2008, ball-milled NaCl was dispersed into oleylamine to encapsulate unsupported disordered PtFe alloy nanoparticles by Liu et al. [121], and then an L1₀-PtFe intermetallic compound was prepared by annealing at 700 °C in a H₂/Ar atmosphere for 4 h. When the mass ratio of the PtFe alloy to NaCl was 1:400, the crystal size only slightly increased after annealing. To better disperse the disordered alloy crystals into the inorganic salt matrix, a special reaction system has been developed to achieve an in situ coating for alloy nanoparticles. DiSalvo et al. [122] developed an in situ KCl coating method to control the crystal size of Pt₃Fe intermetallic compounds, as shown in the schematic diagram in Fig. 12a. The reaction system consisted of PtCl₄ and FeCl₃ as the precursors, potassium triethylborohydride (KEt₃BH) as the reducing agent, and tetrahydrofuran (THF) as the solvent. LiCl was also added to achieve a high proportion of Cl⁻, and the equivalent amounts of K⁺ and Cl⁻ were realized by adjusting the amount of KEt₃BH to ensure the formation of KCl. The disordered Pt₃Fe nanocrystals were directly embedded in the KCl matrix owing to the insolubility of KCl in THF. The KCl, with its high melting point, acted as a barrier in the subsequent annealing process. The effect of the Pt₃Fe/KCl molar ratio on the crystal size is compared in Fig. 12b, c. When the molar ratio of Pt₃Fe to KCl reached 1:80, the agglomeration and sintering of alloy crystals could be effectively prevented. Therefore, the L1₂-Pt₃Fe catalyst exhibited an average size of 4 nm and excellent ORR performance. They subsequently adopted this process to prepare Pt₂FeCo, Pt₂FeNi, Pt₂CoNi, and Pt₃Cr intermetallic compound catalysts by adjusting the annealing conditions and replacing FeCl₃ with other metal chlorides, such as NiCl₂, CoCl₂, and CrCl₃. These catalysts also showed good dispersion and small sizes when coated with an appropriate amount of KCl [123, 124]. Goodenough et al. [125] prepared PtFe₃ intermetallic compounds in a similar reaction system. After annealing, a crystal size of nearly 5 nm remained almost unchanged. Moreover, its mass activity at 0.9 V vs. RHE increased to five times that of commercial Pt/C, while its activity loss was only 9.7% after a 5 000-cycle ADT.

Fig. 12

a Schematic illustration of a KCl coating preparation process, TEM images of L1₂-Pt₃Fe catalysts prepared with Pt₃Fe/ KCl mass ratios of b 1:15 and c 1:160. Reproduced with permission from Ref. [122]. Copyright 2012, American Chemical Society

Unlike the former two coating methods, it is not necessary to remove the coatings by acid leaching or heat treatment in this method; only an appropriate solvent is needed to remove the inorganic salt. However, the employed organic solvents, such as furans, ketones, and alkanes, are toxic and can cause pollution. In addition, the alloy crystals and inorganic salts must form at the same time. Thus, a specific reaction system must be employed, which limits the types of available precursors, reducing agents and solvents.

4.2 Harsh Ordering Transformation Conditions—Promoting the Ordering Process

There are evident differences in the ordering transformation difficulties of PtM alloys in light of the different kinds and contents of transition metal M. For example, Gan et al. [126] investigated the ordering transformation conditions of disordered PtFe₃, PtCo₃, and PtNi₃ alloys. After annealing at 600 °C, ordered L1₂ phases were formed in the PtCo₃ and PtFe₃ samples, but obvious sintering was observed in L1₂-PtFe₃, while the PtNi₃ sample retained a disordered structure under the same treatment conditions. Strasser et al. [127] studied the structural changes of disordered PtNi and PtNi₃ alloys at different annealing temperatures. No ordered phase was found in PtNi₃ at all temperatures, indicating that the ordering difficulty increased with the Ni content. Elements with high ordering difficulty required harsh conditions, such as a high annealing temperature or a long annealing time, to overcome the high diffusion activation energy barrier. However, these conditions usually promoted the sintering of nanoparticles, resulting in an unsatisfactory ORR activity. It also restricted the variety of Pt-based intermetallic compounds because many metals could not form ordered structures with Pt in the normal preparation process.

The lattice parameters and interatomic interaction of Pt alloys can be adjusted by doping the appropriate element, which can reduce the diffusion activation energy and realize ordering transformation under mild conditions. Therefore, doping under mild conditions is very important to ensure the high ordering degree and high activity of Pt-based intermetallic compounds. Although there are few reports in the field of ORR catalysts, this doping process has been successfully applied to other catalysts, which can provide references for promoting the ordering of Pt-based intermetallic compound catalysts for the ORR. For example, Sun et al. [128] prepared disordered Au-doped PtFe nanocrystals by the coreduction of Pt(acac)₂ and HAuCl₄ and the pyrolysis of Fe(CO)₅; then an L1₀-PtFeAu intermetallic compound catalyst for the formic acid oxidation reaction (FAOR) was obtained by annealing at 600 °C for 2 h. Au segregation contributed to the formation of vacancies in the PtFeAu lattice structure and promoted the rearrangement of Pt and Fe. This work demonstrated that the effect of Au on the ordering transformation was composition dependent. The most obvious promotion of the ordering process occurred with 20 at.% Au; however, too much Au was not conducive to the ordering transformation. Wang et al. [129] realized the ordering of a PtFeCu alloy at 500 °C, while a PtFe sample treated under the same conditions remained disordered, indicating that Cu doping promoted the transformation of the PtFe alloy from a disordered structure to an ordered intermetallic compound at a lower annealing temperature. Their work showed that the PtFe_0.7Cu_0.3 intermetallic compound catalyst had the best activity and durability for the methanol oxidation reaction (MOR).

4.3 Properties Need to be Improved—Building a “Pt Skin”

As mentioned in Sect. 2.1, during high-temperature treatment when preparing intermetallic compounds, Pt atoms preferentially occupy surface positions to spontaneously form a “Pt skin” structure due to the strong surface segregation of Pt [80, 130]. On the one hand, the Pt atom layer on the surface can prevent the release of internal transition metals, which helps to enhance the stability of the catalyst. On the other hand, the lattice mismatch between the “Pt skin” and internal PtM intermetallic compound can cause lattice compression and a change in the electronic state of Pt atoms on the surface. Compared with the disordered structure, the “Pt skin” structure can more effectively optimize the Pt d-band [131], so the ORR activity can be further enhanced. However, there is usually a high atomic defect density in the spontaneously formed “Pt skin”, and thus, the M atoms in the subsurface are not effectively stabilized with a “Pt skin” that is too thin; in contrast, too many Pt layers forms a pure Pt surface, weakening or even causing the geometric and electronic effects of transition metals to disappear [132]. Therefore, it is necessary to take extra steps in the preparation process when building a “Pt or Pt alloy skin” with a suitable thickness on the surface of an intermetallic compound core; with proper preparation, the ORR activity and stability of Pt-based intermetallic compounds can be further improved. Currently, surface engineering methods, such as dealloying and galvanic replacement, are mainly employed.

4.3.1 Dealloying

Dealloying refers to a technique to selectively remove one or more components of an alloy by a chemical or electrochemical corrosion process. It has been widely used in the preparation and surface modification of metal nanomaterials.

In the chemical dealloying method, a reagent that can dissolve target elements is usually employed to etch the excess transition metals on a crystal surface [51, 133]. Acid leaching is the commonest chemical dealloying method because most transition metal elements are soluble in strong acids. However, the use of only acid leaching forms a “Pt skeleton” structure; the lower coordination number and higher oxidation state of Pt on the surface affects the stability and activity of the catalyst. It has been demonstrated that an additional annealing step after acid leaching is more advantageous for constructing a “Pt skin” structure on the surface of Pt-based intermetallic compounds. For instance, in the work of Popov et al. [134], thanks to secondary annealing in a reducing atmosphere, a “Pt skin” was formed by the surface relaxation and reconstruction of atoms on the topmost surface; moreover, the oxidized Pt on the surface was reduced. Therefore, the stability of the L1₀-PtCo/Pt intermetallic compound catalyst was significantly improved in comparison with the sample using only acid leaching, and its ECSA loss was only 9% after a 30 000-cycle ADT. Hu et al. [135] also prepared an L1₀-PtCo/Pt intermetallic compound with an average particle size of approximately 4.8 nm and a “Pt skin” of three atomic layers by acid leaching and a subsequent heat treatment. Compared with the disordered PtCo and commercial Pt/C catalysts, L1₀-PtCo/Pt showed higher ORR activity and durability. It was thought that its high activity was related to lattice contraction and the enhanced d-orbital hybridization of the “Pt skin”, and its high durability was mainly attributed to the ordered intermetallic core. As shown in Fig. 13a ,a prepared L1₀-PtFe intermetallic compound was stirred in HClO₄ at 60 °C for 12 h by Sun et al. [112], and then, L1₀-PtFe/Pt with a “Pt skin” of two atomic layers was obtained by annealing at 400 °C for 2 h in a H₂/Ar atmosphere, as shown in Fig. 13b–e. Its mass activity reached 0.7 A mg_Pt⁻¹. The half-wave potential showed no shift after a 5 000-cycle ADT, while that of L1₀-PtFe without dealloying treatment shifted negatively by 36 mV under the same conditions. Using a similar method, they also obtained an L1₀-PtCo/Pt catalyst with excellent ORR performance and a “Pt skin” of two atomic layers [132]. In addition to acid leaching, an alkali solution can also be employed to dealloy an intermetallic compound formed with Pt and acid–base amphoteric post-transition metals. For example, Jiang et al. [136] realized the chemical dealloying of Pt₁₂Al₈₈ in a N₂-saturated NaOH aqueous solution at room temperature to obtain an L1₂-Pt₃Al/Pt intermetallic compound with a Pt monolayer skin. As a result of the strong interaction, ligand effect and compressive strain between internal Al atoms and surface Pt atoms, L1₂-Pt₃Al/Pt showed excellent durability and a 6.3-fold increase in the specific activity at 0.9 V vs. RHE compared with a Pt/C catalyst.

Fig. 13

a Schematic illustration of building “Pt skin” by chemical dealloying method, b TEM, c electron energy loss spectroscopy (EELS) mapping, and d, e HAADF–STEM images of an L1₀-FePt/Pt, where Fe is colored green and Pt is colored red. Reproduced with permission from Ref. [112]. Copyright 2018, American Chemical Society

The electrochemical dealloying method, in which the sample is swept in a certain potential range in the electrolyte to remove transition metal components, is also employed to construct the “Pt skin” structure of Pt-based intermetallic compounds. Both potential cycling and potentiostatic methods were conducted to regulate the surface structure of PtCu₃ intermetallic compounds with a low Pt content by Abruña et al. [137, 138]. Regarding the potentiostatic method, it was found that the ORR activity of the catalyst was improved when the potential reached 1.0 V; however, the crystal showed a “spongy” structure, and the ordered structure disappeared. Regarding the potential cycling method, an ordered PtCu₃ core–shell structure with a thin “Pt skin” could be formed by reasonably adjusting the electrochemical cycling parameters. The elemental distributions of the catalyst before and after the potential cycles are shown in Fig. 14a, b. A PtCu₃/Pt intermetallic compound catalyst with a 0.6–1.0-nm-thick “Pt skin” was obtained by sweeping between 0.05 V and 1.0 V at a scan rate of 50 mV s⁻¹ for 5 000 cycles in N₂-saturated HClO₄. As shown in Fig. 14c, d, its mass and specific activities were 2.5 and 6 times higher than those of commercial Pt/C. The ORR activity of PtCu₃/Pt could be further enhanced by increasing to 100 000 cycles, and the above two parameters increased to five and seven times those of commercial Pt/C, respectively. Li et al. [139] performed a potential cycling dealloying treatment for L1₂-PtCu₃ by sweeping in the range from 0.06 to 1.30 V for 20 cycles, and a PtCu₃/Pt intermetallic compound catalyst with a “Pt skin” of three atomic layers was formed. Due to the lattice mismatch between the core and shell, its mass activity was 6.9 times higher than the control without dealloying, and its mass and specific activities were 5.2 and 8.2 times higher than those of commercial Pt/C, respectively. In addition, the low formation energy of the ordered core and the enhanced oxidation resistance of the “Pt skin” provided excellent durability. After 5 000 cycles, the attenuation rates of its mass activity and specific activity were only 12.5% and 9.2%, respectively.

Fig. 14

ADF–STEM and EELS mapping images of L1₂-PtCu₃/Pt a before and b after electrochemical dealloying. c Cyclic voltammogram (CV) and d LSV curves of L1₂-PtCu₃/Pt obtained under different potential cycles. Reproduced with permission from Ref. [138]. Copyright 2015, American Chemical Society

4.3.2 Galvanic Replacement

Galvanic replacement is a redox reaction between metal atoms in a sacrificial template and the ions of another metal in solution. Driven by their reduction potential difference, the template metals are oxidized and dissolved, and the ions in solution are reduced and deposited on the template surface [140, 141]. This method has also been applied for modifying the surface of Pt-based intermetallic compounds. The surface composition of crystals can be controlled by regulating the ratio of the two metals. Taking the work of Adzic et al. [142] as an example, a monolayer of Cu atoms was first deposited on the surface of a PtPb intermetallic compound by underpotential deposition, and then, the Cu was completely replaced in 3 min by the Pt in a K₂PtCl₄ solution to prepare a PtPb/Pt intermetallic compound catalyst with a monolayer “Pt skin”. Compared with the control without dealloying, its specific activity and mass activity increased to 3.9- and 2-fold, respectively, and no activity attenuation was observed after a 6 000-cycle ADT. Gaberšček et al. [143] realized galvanic replacement between Cu and Au by adding HAuCl₄ into a suspension containing a PtCu₃/C catalyst; the surface segregation of Au and Pt was promoted by annealing at 300 °C in a H₂/Ar atmosphere for 12 h to form a “PtAu skin”, as shown in Fig. 15. The “PtAu skin” containing only 1 at.% Au provided the as-prepared L1₂-PtCu₃/PtAu catalyst better resistance to Pt and Cu dissolution. In particular, the attenuation of the ECSA was only 14% in the actual fuel cell at 60 °C after 10 000 cycles, which was significantly lower than the 38% of the sample before forming the “PtAu skin”.

Fig. 15

HAADF–STEM image with EDS elemental mapping of L1₂-PtCu₃/PtAu. Reproduced with permission from Ref. [143]. Copyright 2016, American Chemical Society

4.4 Weak Interaction between the Crystal and Support—Anchoring and Confinement Effects

To enhance the dispersion of nanoparticles and the conductivity of catalysts, active component crystals are usually supported on porous substrates with high specific surface areas in the actual application of intermetallic compound catalysts; carbon materials are the most commonly employed support. However, because the interaction of the traditional physical adsorption mode between a crystal and support is weak, the active component crystal easily sheds, migrates, and agglomerates under dynamic operating conditions, resulting in the performance degradation of fuel cells [144]. The anchoring and confinement of intermetallic compounds can enhance the interaction between nanocrystals and supports, which can not only effectively avoid crystal shedding and migration but also change the crystal structure and adsorption capacity to promote interfacial charge transfer. Therefore, it has become a feasible way to improve the activity and stability of Pt-based intermetallic compound catalysts [145].

The anchoring effect refers to the interaction between the active component crystal and support being enhanced by the function of the anchoring species, thereby inhibiting the shedding and migration of catalyst nanoparticles. It is the common means of achieving an anchoring effect to introduce atoms, such as O, N, P, S, B, and their groups that have a strong binding force with Pt, by the surface functionalization of supports [146, 147]. For example, Popov et al. [148] prepared a N-doped graphite carbon matrix embedded with Co particles by the Co-catalyzed pyrolysis of chelates and activated carbon at high temperatures. Subsequently, Pt was deposited and anchored onto the N of the carbon surface and then Co diffused into the Pt particles during heat treatment at 800 °C to form a PtCo intermetallic compound catalyst with three times the mass activity of commercial Pt/C. Benefitting from the ordered structure and strong interaction between the crystal and support, the crystals remained uniformly distributed on the support, and the average particle size was increased by only approximately 1 nm after 30 000 cycles of the MEA durability test. The attenuation rates of its ECSA and mass activity were 36% and 26%, respectively, which are lower than those of commercial Pt/C.

The confinement effect means that the metal crystal is confined in a small area by substances such as ceramics, mesoporous materials or hollow materials [149,150,151] to inhibit the migration of catalyst crystals. In recent years, metal–organic framework (MOF) materials have developed rapidly, and various MOF-derived materials have served as supports for Pt-based intermetallic compound catalysts. These carbon supports can play a dual role in anchoring and confinement because of their heteroatoms and porous structures. Wu et al. [152] synthesized a Co-doped zeolite imidazolate framework (ZIF) by mixing Zn(NO₃)₂, Co(NO₃)₂ and 2-methylimidazole at 60 °C for 1 h, and then, the Co-doped ZIF was heat treated at 900 °C for 1 h to obtain N- and Co-doped carbon supports. A Pt₃Co intermetallic compound catalyst with a half-wave potential of 0.92 V was prepared by annealing these ZIF-derived carbon-supported Pt nanoparticles at 900 °C for 30 min. As a result of the anchoring effect of the doped N and the confinement effect of the porous structure, this catalyst exhibited good stability, and its half-wave potential only decreased by 12 mV after 30 000 cycles. Adopting the preparation procedures in Fig. 16a, Liao et al. [153] impregnated a ZIF-8-derived mesoporous carbon support with Pt and Co precursors. After freeze-drying, as illustrated in Fig. 16b, c, a Pt₃Co intermetallic compound with a crystal size of 3 nm was synthesized in the porous structure of the carbon support by annealing at 750 °C for 2 h. Due to the special structures of MOF-derived materials, as shown in Fig. 16d, e, the mass and specific activities of the L1₂-Pt₃Co catalyst were 5.2 and 4.2 times higher than those of commercial Pt/C, respectively, and its half-wave potential was up to 0.93 V. After 5 000 cycles of the rotating disk electrode (RDE) test and 20 000 cycles of the MEA test, the losses in these activities were only 13.3% and 15.4%, respectively. There are also other reports on a similar method. For instance, Kuang et al. [154] prepared a PtZn intermetallic compound catalyst by the direct pyrolysis of ZIF-8-supported Pt nanoparticles. Peng et al. [155] supported Pt on ZIF-8 partially replaced with Fe and then prepared a PtFe₃ intermetallic compound catalyst by a pyrolysis process. These catalysts also showed good ORR activity and stability owing to the dual effects of confinement and anchoring provided by the supports.

Fig. 16

a Schematic of preparation procedures of ZIF-8-derived supported L1₂-Pt₃Co catalyst. b TEM, c HAADF–STEM, d mass and specific activities of L1₂-Pt₃Co catalyst, and e the comparison of its mass activities before and after ADT. Reproduced with permission from Ref. [153]. Copyright 2020, Royal Society of Chemistry

5 Conclusion and Prospects

Although alloying with 3d transition metals and nanoscale structural design can solve the cost and performance problems of Pt catalysts for fuel cells to a certain extent, the insufficient activity and stability of Pt alloys caused by the excessive compression of the d-band and the release of transition metals still cannot be ignored. Different from disordered Pt alloys, Pt-based intermetallic compounds are long-range ordered alloys with definite atomic positions and specific stoichiometric ratios; additionally, these alloys are a combination of metal bonds and ionic/covalent bonds. Due to the suppression of excessive compression, ordered and uniform composition distribution and enhanced atomic bonding, the ORR activity can be further enhanced and the loss of transition metals can be effectively prevented. Therefore, Pt-based intermetallic compound catalysts for the ORR have received more attention. To date, various Pt-based intermetallic compound catalysts have been developed, and some research advances have been gained in regard to inhibiting high-temperature sintering, reducing ordering difficulty, improving ORR performance, and enhancing crystal–support interactions. These achievements have not only helped to solve the scientific problems and technical bottlenecks of preparing Pt-based intermetallic compound catalysts but also provided theoretical and technological guidance for the development of other catalysts, which is of great significance to the application of high-performance fuel cells.

To promote the actual application of Pt-based intermetallic compound catalysts in the cathodes of fuel cells, future research is expected to focus on the following aspects.

• (1) In view of various disadvantages, such as complex operation, difficult control, and environmental pollution, the as-reported coating processes are unfavorable for large-scale preparation. Therefore, a preparation strategy with wide applicability, simple posttreatment operation and little impact on structure and performance is urgently needed to control the size of intermetallic compound catalysts.

• (2) At present, research advances in promoting the ordering transformation are mainly focused on the initially developed PtFe intermetallic compounds that are easier to prepare. Henceforth, the ordering promotion strategy should be applied to the preparation of other Pt alloys to expand the types of Pt-based intermetallic compounds. In addition, it is necessary to clarify the influence of dopants on the crystal structures and atomic interdiffusions of Pt alloys and determine the mechanism of promoting the ordering transformation to select the appropriate dopants and optimize the ordering conditions.

• (3) Limited by the preparation difficulty and product stability, most of the as-reported Pt-based intermetallic compound catalysts are Pt₃M and PtM types, and few works on PtM₃ intermetallic compounds have been reported. Consequently, to further reduce the Pt content and the cost of intermetallic compound catalysts, it is necessary to develop a new preparation technology that reduces the preparation difficulty and improves the stability of intermetallic compounds with a high transition metal content.

• (4) Although DFT calculations have been employed in a few works to explain the mechanism of the improved ORR activity of Pt-based intermetallic compounds, powerful experimental evidence is still lacking to prove the relationship among the crystal structure, electronic state, and ORR performance improvement, especially the effect of changes in catalyst structure on the durability during actual electrode reactions. Therefore, in-situ characterization and real-time detection technology should be developed to combine theoretical calculations with experimental characterization and deepen the understanding of the mechanism by which Pt-based intermetallic compound catalysts improve ORR performance.

• (5) Most of the as-reported Pt-based intermetallic compound catalysts are investigated at the half-cell level. However, the influencing factors of catalyst performance in actual single cells are much more complex than those in the RDE test. In addition to the intrinsic performance of catalysts, they are also affected by the MEA structure, preparation process, and operating conditions, such as the cell temperature, relative humidity, and back pressure. Hence, to optimize the single-cell preparation process and give full play to the performance advantages of Pt-based intermetallic compound catalysts, the structural characteristics and catalytic process in single cells should also be investigated.