An ultra-dispersive, nonprecious metal MOF–FeZn catalyst with good oxygen reduction activity and favorable stability in acid

As one of the most efficient energy conversion devices, proton exchange membrane fuel cell shows great potentials in the application of hydrogen vehicles, stationary power plants and renewable energy utility [1]. Being a vital bottleneck in achieving the large-scale application of fuel cells, cathodic ORR catalysts with higher activity and durability have been extensively explored and investigated to facilitate the sluggish ORR process [2]. Normally, precious platinum and its alloys are popularly used to accelerate the ORR catalysis. However, the high price, scarity of resources and limited stability of Pt-based catalysts have held back the commercialization [3]. Great attention has been paid to investigate and develop the highly active nonprecious metal or metal-free ORR catalysts [4]. Modification of carbon materials varies from doping the heteroatoms of N, P, S, B, F, etc., incorporating with carbon and creating topological defects on them, to developing the atomically metal induced defective carbons with metal-macrocycle complexes, which are all effective methods to enhance the catalytic activity of carbon catalysts [5‐9].

Modified carbon materials, such as carbon nanotubes, graphene and mesoporous carbons with unique physical and chemical properties after doping with heteroatoms, have drawn intensive interest for their excellent performances in ORR catalysis [10‐13]. Among the generally doped foreign atoms, nitrogen could not only alter the electronic property and crystal structure of catalysts, but also prominently transform the chemical stability, surface polarity and electron-donor properties of the carbon materials, which effectively promoted the ORR catalysis [14]. Nitrogen type in the catalyst modification is one of the most fatal factors for ORR catalysis. Usually, several typical nitrogen structures are believed to be catalytic active for ORR catalysis, namely pyridinic N, pyrrolic N, quaternary N and pyridine-N-oxide [15]. Electronegativity is another key factor to affect the catalytic property of carbon by influencing the surface charge density, which benefits the ORR process through facilitating the formation of oxygen adsorption sites on the modified carbon atoms [1, 4].

Fabrication of effective nonprecious ORR catalysts with earth-abundant metals have been paid much more attention [16, 17]. A lot of pioneering work have been done by Dai and co-workers [18] on ORR catalysis with the modified CNTs. They proved that the generation of positive charges on the surrounding modified C atoms with incorporated high electronegative N atoms was catalytic active for ORR [19‐21]. Sun and Lai et al. [22‐24] demonstrated that both the pyridinic N and graphitic N were beneficial for ORR catalysis. Quaternary nitrogen (graphitic N), located at the edge of graphene planes, is more active for ORR, because the enriched carbon defects on edges facilitate the incorporation of nitrogen and activation of carbon [1, 25]. N-modified graphene can exhibit improved ORR activity, demonstrating the crucial role of nitrogen in creating defective active sites by interrupting the original state of graphene [26, 27]. Lai et al.[23] proved that it was the graphitic N in modified graphene that showed the higher limiting current density in catalysis, and pyridinic N was more beneficial to elevating the ORR onset potentials. Niu et al. [28, 29] demonstrated that mesopores at around 2–50 nm could remarkably contribute to enhancing the mass transportation of relevant electrochemical species, and prominently facilitating the ORR catalysis. The high specific surface area, abundant mesoporous architecture and well-modified nitrogen structure can render the mesoporous carbon materials to the excellent ORR catalysts. Atomic-dispersive metal species are also capable of modulating the electronic structure and configuration of carbon catalysts, producing defective carbon active sites with improved catalytic activity. Most of the metal–nitrogen-based catalysts show excellent ORR activity and durability in alkaline electrolytes. However, only a few works reported the comparable activities in acidic electrolytes [30‐32], because transitional metals in the nonprecious catalysts are liable to be dissolved in acidic electrolytes and aggregate under fuel cell operating conditions, suffering severe damage and serious losses during the potential cycling processes. Therefore, developing the stable and catalytic active nonprecious ORR catalyst remains to be a meaningful and challenge work.

Herein, this article illustrates a preparation of an ultra-dispersive Fe- and Zn-doped nonprecious MOF catalyst with a combined hydrothermal crystallization and pyrolysis processes. With the integral uniform catalytic active structures developed both on the surface and in the bulk, this catalyst displays good ORR catalytic activity and prominent stability in HClO₄.

Integral uniform MOF–FeZn catalyst was typically synthesized through the hydrothermal and pyrolysis processes. The added organic materials went through the hydrogen-bond interaction and dehydration condensation reactions with particular functional groups, forming tight conjunction organic skeletons by stacking and covalent bonds [22, 33]. After chelating and complexation with metal ions, integral uniform precursors were obtained. Going through the hydrothermal crystallization, these organic skeletons were well developed and sufficiently interacted. Finally, the collected powder underwent carbonization to form the active components, developing porous structures at the same time.

Morphology characteristics of MOF–FeZn were identified by TEM and HRTEM (Fig. 1). In the low magnification images, we can see that this catalyst is a straticulate folded graphene structure, which is similar to the rime ice flowers (Fig. 1a and b). As shown in the high magnification images, well-distributed metal granules dispersively inlayed in the carbon matrixes (Fig. 1c–f). In Fig. 1d and e, there are many fish scaly structures formed with intensively scattered or tactic distributed metal carbide nanoparticles for the confined carbon lattices. Because of the well-proportioned organic skeletons as well as the good combination and interaction of metal ions, this catalyst develops into an integral uniform structure, which is not only catalytic active on the surface, but also highly active in the matrix. This is very important for MOF–FeZn to provide persistent activity and longer durability in electrochemical catalysis. Moreover, the well contact of nanoparticles and graphitic layers are also meaningful to improving the activity and stability of catalyst. However, there are few well-constructed crystal texture metallic carbides formed in the matrix as shown in (Fig. 1g and h). The clear crystal lattice fringes demonstrate the good crystallization of carbon and metal carbides with the interfringe distances of 0.2075 nm and 0.348 nm in and out the particles, which are correspondingly belonging to the lattice spacing of (021) plane of for Fe₅C₂ and (002) plane for graphitic carbon, respectively. The lattices of ~ 0.21 nm are consistent with the (211) plane of Fe₃C, further demonstrating the well-developed metallic carbide crystal textures. The enlarged carbon interlayer distances (> 0.34 nm) are owing to the heteroadulteration of Fe, Zn and N substances in carbon. These encapsulated nanoparticles are believed to activate the graphitic layers, making positive contributions to catalysis.

Elemental mapping and electron diffraction properties of the prepared MOF–FeZn catalyst were characterized with the HAADF-STEM (Fig. 2). Comprehensively, this catalyst is an integral uniform Fe-, Zn-, N-modified catalyst with numerous dispersive tiny nanoparticles inlayed among graphitic carbon lattices. Its element distribution, revealed by the above EDS mappings, demonstrates the well distribution and good combination of C, O, N, Fe and Zn (Fig. 2c–f, h–i). The relevant dispersive SAED diffraction pattern in Fig. 2g further indicates the existence of well-crystalline metallic compounds.

Physical features and properties of MOF–FeZn were sufficiently characterized in Fig. 3. The XRD pattern matched well with the iron carbide and nitride structures (Fig. 3a). Specifically, the broad diffraction peak at around 26.9° is associated with the graphite carbon, and the peaks at around 27.61°, 30.97°, 41.31°, 51.03°, 53.26°, 74.44° and 83.78° are ascribed to Fe₃C (JCPDS no. 03-065-0393, 89-2876). The peaks at 52.40° and 59.50° are belonging to Fe₅C₂ (JCPDS no. 01-089-8968). The 52.40° and 77.43° peaks are assigned to the metallic Fe (JCPDS no. 01-087-0721). The formation of Fe₄N substance can be proved by the 48.16°and 89.27° peaks (JCPDS no. 01-086-0231, 01-077-2006). As we can see, most of the doped metal ions in MOF–FeZn are developed into iron carbides and nitrides, further conforming the well combination of Fe–C, N (Fig. 2). The iron carbides or iron nitrides display relative weak diffraction peaks in XRD pattern, further confirming the tiny crystal or even amorphous structure of the doping substances, and some of which are not detectable through XRD. Raman spectrum was recorded to characterize and testify the graphitization property of MOF–FeZn (Fig. 3b). Accordingly, this catalyst has obviously high graphitic property with a typical I_D/I_G ratio of 0.945, indicating the ordered graphitic carbon structure of MOF–FeZn due to the sufficient pyrolysis under 800 °C in N₂. The intensive D band at around 1345.2 cm⁻¹ implies the rich disordered defects in catalyst. Moreover, the noteworthy broad band 2D is from 2328 cm⁻¹ to 3275 cm⁻¹, demonstrating that the straticulate folded graphene is in multilayer structure. Numerous graphitic fringes and defects in catalyst will activate the ORR catalytic activity of MOF–FeZn [34‐38]. As shown in Fig. 3c and d, the N₂ adsorption/desorption isotherm of MOF–FeZn belongs to the prominent type IV isotherm with remarkable hysteresis loops, which indicate the mesoporous structure of catalyst. Porosity properties were assessed and analyzed by BET method. The catalyst surface area is up to 529.92 m² g⁻¹. Based on the BJH analysis, the total pore volume is 0.729 cm³ g⁻¹, and the mesopore volume can be as many as 0.662 cm³ g⁻¹. Some of the mesopores are intensively located at around 4 nm as shown in the insert graph in Fig. 3d. This typical pore size is significant for electrocatalysis [37‐39]. Some pores are widely distributed from 10 to 50 nm, which are beneficial for the rapid transportation of reactant species. In brief, this catalyst exhibits high surface area, desired mesoporous structures and large pore volume, which are all favorable for active site exposure and reactants transport [38, 40].

The exact mass contents of Fe and Zn in MOF–FeZn catalyst were analyzed to be 33.92 wt.% and 1.05 wt.%, respectively, using the ICP-OES technique. This content is very different from the initial addition of Fe and Zn (n_Fe:n_Zn = 1:4.84, mol/mol). Because ZnCl₂ is easily evaporated under 800 °C during pyrolysis, demonstrating the existence of Zn in precursor is Zn²⁺ instead of the metal (boiling point of metallic Zn is 907 °C). ZnCl₂ in precursor, acting as the pore-creating agent, is mainly sacrificed in the carbonization. Moreover, specific surface element contents were characterized by XPS, which were conducted to elucidate the composition and bonding configuration of different elements on the surface of MOF–FeZn (Fig. 4). Generally, this catalyst is composed of C (86.81 at.%), O (7.57 at.%), N (4.67 at.%), Fe (0.67 at.%) and Zn (0.27 at.%) on the surface, demonstrating the successful doping of heteroatoms. Compared to the ICP analysis, we can see most of the doping metals are encapsulated deep in the carbon matrix with abundant active sites developed both in the matrix and out on the surface of MOF–FeZn. For C 1 s spectra, the intensive sp2 C peak, centered at 284.6 eV (Fig. 4a), implies the high graphitization of MOF–FeZn coincident to the Raman analysis, enhancing the catalytic activity by increasing the electrical conductivity [41]. According to the O1s spectra (Fig. 4b), oxygen content in this catalyst is relatively high, because of the formation of oxygen-rich functional groups in organics. Certain amount of C–O and C=O groups enhance the transport of electrolytes, benefiting the interaction between catalysts and electrolytes. The deconvoluted N 1 s spectrum is composed of five peaks, assigning to pyridinic N, Fe–N, pyrrolic N, graphitic N and oxidized N, respectively (Fig. 4c), most of which are recognized as active components except for the small amount oxidized N. Particularly, the deconvolution peaks clearly indicate the high ratio of Pyridinic N and pyrrolic N with relatively rich Fe–N and graphitic N in this catalyst, implying the rich active sites formed. The deconvolution peak of Fe2p reveals two pairs of doublets of the Fe2p_3/2 and Fe2p_1/2 peaks, demonstrating the combination state of Fe and further confirming the present of iron carbide and nitride species in catalyst at around 711.1 eV and 724.6 eV as well as at 713.7 eV and 728.8 eV with the satellite peak at 719.1 eV (Fig. 4d). The content of Zn is very low and mostly in the ion state.

As shown above, this as-synthesized MOF–FeZn catalyst is a well-established nonprecious metal organic framework catalyst, on which the effective adulteration of N, proper combined Fe–N, and the highly dispersive embedded metal carbides and nitrides, as well as the integral uniform distribution of active sites, display synergetic effects for catalysis. Meanwhile, the relatively high specific surface area, typical mesoporous features, unique pore sizes and large pore volumes are all beneficial for the improvement of catalyst activity, due to their great contribution to the development of active sites and fast transport of reactants.

Compared with the commercial Pt/C, this as-prepared integral uniform MOF–FeZn catalyst has noteworthy ORR activity and favorable stability in acidic electrolyte (Fig. 5), which are evaluated by the rotating disk electrode in 0.1 M HClO₄. Linear sweep voltammetry curves (LSV; Fig. 5a) of MOF–FeZn clearly demonstrate the good ORR activity in acidic electrolyte with the onset potential (E_onset) of 0.93 V @ 0.1 mA cm⁻², the half-wave potential (E_1/2) of 0.768 V@ 2.757 mA cm⁻² and the limited diffusion current density (j_d) of 5.5 mA cm⁻², only slightly inferior to the commercial Pt/C (E_onset = 0.959 V; E_1/2 = 0.833 V; j_d = 5.5 mA cm⁻²). The high ORR catalytic performance of MOF–FeZn can be ascribed to the good electrical conductivity, high specific surface area and well-developed mesoporous architecture with the successful modified N and Fe substances. In Fig. 5b, cyclic voltammetry (CV) curves reveal the high specific surface area of MOF–FeZn with a large double electric layer capacitance, indicating the porous structure and good electrical conductivity of this catalyst. The long-term stability of MOF–FeZn and Pt/C was compared, following the accelerated durability test (ADT) by cycling the potentials up to 10,000 times between 0.6 and 1.0 V in N₂-saturated HClO₄ under 100 mV s⁻¹. CV and LSV tests were performed before and after the ADT protocols, respectively (Fig. 5a and b). After cycling in the wide potential range, MOF–FeZn still remains good electrochemical features as shown in Fig. 5b with no special peaks appeared in the CV curves. The slightly enlarged double electronic layer of CV curve indicates the enhanced hydrophilia of MOF–FeZn catalyst during the ADT process, which is obviously different form the remarkable reduction of electrochemical active specific surface area of commercial Pt/C. The changed hydrophilia properties and adsorbed functional groups can change the adsorption and desorption of reactants, influencing the catalytic process. Foremostly, this nonprecious MOF–FeZn catalyst well sustains the harsh electrochemical potential cycling process in acid with the half-wave potential only changed 21 mV, superior to the reduction of 149 mV for Pt/C. The limited diffusion current of MOF–FeZn is also relatively stable than Pt/C (Fig. 5a). The remarkably stable ORR activity after 10,000 potential cycles in a wide oxidation potential range can be ascribed to the ultra-dispersive integrative structure and high graphization of MOF–FeZn, preventing the active sites from serious destroyed. The less prominent ORR activity attenuation on MOF–FeZn after the potential cycles is because this catalyst displays an integral uniform structure with abundant active sites not only on the surface but also in the bulk matrix of the catalyst, so the CV curve before and after ADT is similar and the half-wave potentials are close. However, the Pt/C underwent significant decrease in the ADT process with more obviously negative shifted half-wave potentials and significantly reduced limited diffusion current density. In brief, the results prove that this integral uniform MOF–FeZn catalyst has greatly improved ORR activity and favorable durability in acid with its homogeneously dispersive metal carbides and nitrides protected by high graphitic carbon layers. To further recognize the kinetic property of electrocatalysts, transferred electrons of MOF–FeZn were calculated based on the LSV curves under different rotation speeds in O₂-saturated HClO₄. The rotation speed of RDE was adjusted from 400 to 2000 rpm with continuous flowing of O₂ bubbles during the test. Koutecky–Levich (K–L) equation is applied to analyze the kinetic activity and transferred electrons of MOF–FeZn (Fig. 5c). The number of transferred electrons was calculated based on:

where ω is the rotating speed in rpm. j is the measured current density; j_K and j_L are the kinetic and limited diffusion current densities, respectively; n is the number of transferred electrons; F is the Faraday constant (96 485 C mol⁻¹); C₀ is the bulk concentration of O₂ (1.2 × 10^–6 mol cm⁻³); D₀ is the diffusion coefficient of O₂ (1.93 × 10^–5 cm² s⁻¹); and υ is the kinematic viscosity of electrolyte (0.01 cm² s⁻¹). The corresponding K–L plots show the relatively good linearity (Fig. 5c). B derived from the slope of K–L plot of MOF–FeZn is 0.125. Accordingly, the number of transferred electrons (n) is calculated to be 3.82 in 0.1 M HClO₄ on MOF–FeZn, which is close to the ideal four-electron pathway of Pt/C for oxygen reduction in acid, revealing relative fast ORR kinetic process on MOF–FeZn. Furthermore, Tafel analysis was also performed on MOF–FeZn in 0.1 M HClO₄ (Fig. 5d). There is an obvious change at around 0.81 V, above which the Tafel slop is 40.58 mV dec⁻¹, and below which it transforms to 104.6 mV dec⁻¹, revealing the relative fast kinetic process and significant change of kinetic activity in this region.

According to the above analysis, MOF–FeZn, an integral uniform nonprecious catalyst, exhibits good ORR catalytic activity and favorable stability in HClO₄, which may be ascribed to the integral uniform structure of MOF–FeZn with sufficient active sites not only on the surface but also deep in the bulk parts, so the catalyst can still display excellent ORR activity after the potential cycling processes.

Comparing to the reported state-of-the-art nonprecious catalysts (Table 1), the preparation processes of MOF–FeZn are relative simple and easy to mass production. The raw materials for MOF–FeZn synthesis are relatively low toxicity and easily access, and the solvent used is just water. The catalysis of oxygen reduction on MOF–FeZn follows nearly four-electron process, and the half-wave potential is slightly higher than the reported first-class nonprecious-metal catalysts. The stability of MOF–FeZn is among the best platinum-free catalysts in acid electrolyte. In a word, with the mild preparation method, MOF–FeZn catalyst develops into an integral uniform structure with highly dispersive active sites not only on the surface but also in the bulk, displaying good ORR catalytic activity and excellent stability in acid.

In summary, this as-synthesized integral uniform MOF–FeZn catalyst is a promising ORR candidate catalyst in acid. The high surface area and well-distributed nanoparticles inlayed in the N-modified carbon matrix provide sufficient active sites not only on the surface but also deep in the bulk. The well-exposed high contents of pyridinic N, Fe–N, pyrrolic N, graphitic N active sites and porous skeleton structures facilitate the mass-transfer and kinetic property in the ORR catalysis. Due to the synergetic contribution of the catalyst composition and structure, MOF–FeZn follows a nearly four-electron path in ORR catalysis with high limited diffusion current and half-wave potentials in 0.1 M HClO₄. This integral uniform catalyst also displays favorable durability in acid for its homogeneously dispersive metal carbides and nitrides protected by high graphitic carbon layers both on the surface but also in the matrix.