Gram-scale production of an Fe single atom catalyst and mass transfer enhancement in PEMFCs

Abstract

The huge advantage of the lower fabrication cost for non-Pt catalysts is attracting increasing attention for fuel cell development. As a potential candidate, Fe single atom (SA) catalysts exhibit remarkable catalytic activity for the oxygen reduction reaction. However, due to the relatively low intrinsic activity, high active site density and an optimized mass transfer path are particularly required for Fe-SA proton exchange membrane fuel cells (PEMFCs). Herein, a Fe-SA catalyst with abundant heteroatoms and a large specific surface area is synthesized based on a lab-made ZIF-derived carbon support via a simple adsorption-annealing method. Benefitting from the advanced carbon support, plenty of Fe atoms can be adsorbed and anchored on the surface of the carbon particles. After careful modulation of the annealing temperature, highly dispersed Fe single atom active sites can be obtained, leading to good catalytic activity (the half-wave potential is more than 0.827 V versus RHE). Furthermore, coordinated with the structure optimization of the gas diffusion layer, the maximum power density can be improved to 803 mW cm⁻², indicating the application potential of this catalyst in PEMFCs. This work not only obtains an advanced Fe-SA ORR catalyst but also provides a demonstration for the research and development of non-Pt fuel cell catalysts.

---

Introduction

Proton exchange membrane fuel cells (PEMFCs) have emerged as a promising clean energy technology for various applications, particularly in the transportation sector, owing to their high energy efficiency, low operating temperature, and zero emissions. Currently, scarce hydrogen infrastructure and the high cost of fuel cell catalysts are the major challenges for the commercial application of PEMFCs.¹ Recently, enormous efforts have been made to reduce PEMFC fabrication costs by using low-Pt² and non-Pt³ catalysts during membrane electrode assembly (MEA) production. Compared with nanoparticle-based oxygen reduction reaction (ORR) catalysts, non-noble metal single atom catalysts (SAC), such as Fe–N–C,^4,5 Co–N–C,^6,7 Mn–N–C,^8,9 Cu–N–C¹⁰ and Zn–N–C,^11,12 show impressive catalytic activity for the ORR, providing candidates for low-cost PEMFCs. Among all the SACs, the Fe–N–C configuration is the most promising, due to the outstanding ability of Fe atoms in the Fe–N–C matrix for adsorbing O₂ and weakening O–O bonds.^13,14 To further improve the catalytic activity, researchers have focused their efforts in two directions, improving the intrinsic activity and increasing the SA density. On the one hand, the coordination environment seriously affects the ORR activity of Fe-SACs. Based on theoretical and experimental analysis, researchers have a better understanding of the reaction mechanism and electronic configurations on Fe-based active sites.^15–17 Compared with FeN₄, FeN₂ on the edge of graphene may show higher ORR activity due to the lower interaction between Fe sites and oxygen-containing intermediates.¹⁸ However, the ORR turnover frequency¹⁹ for Fe-SA is still orders of magnitude lower than that for Pt, which always requires adequate Fe-SA to obtain satisfactory performance. Therefore, optimized carbon supports with abundant heteroatoms (such as N, S, O) and larger specific surface areas are important for high-performance Fe-SA catalysts.²⁰ Recently, zeolitic imidazolate framework (ZIF)-derived non-Pt catalysts have attracted increasing attention because of their high heteroatom content and porous structure. Benefitting from the porous and stable structure of the precursor, ZIF-derived carbon powder with large specific surface area (more than 1000 m² g⁻¹) can be obtained by facile pyrolysis.²¹ To further increase the surface area of the catalyst powder, NaCl was used as the salt encapsulation agent during ZIF-8 pyrolysis.²² Compared with the traditional carbonization process, the Brunauer–Emmett–Teller (BET) specific surface area of the NaCl-treated material increased by 46% to 1240.8 m² g⁻¹, with a larger micropore volume. Although the micropores provide a considerable number of graphene edges for Fe–N–C active site loading, the mass transfer efficiency should also be carefully considered due to the large mass transfer resistance of O₂ molecules and protons in the micropores.²³ Comprehensive analysis indicates that micropores in catalyst particles can increase the physical surface area while severely restricting the electrochemical wettability and accessibility. The existence of mesopores can significantly alleviate the above restrictions.²⁴ Furthermore, a size-sensitive molecular probe method was employed to accurately analyse the influence of pore size. More than 70% of the activity for the ORR was attributed to micropores with diameters from 0.8 to 2.0 nm during a PEMFC test.²⁵ To improve the mass transfer efficiency and increase the utilization of active sites, partial etching of carbon support precursor is an imaginative method.^26,27 As a general strategy, a carboxylate-assisted method has been developed to introduce the in situ formed carbon nanotube and enlarge the proportion of mesopores, leading to high accessibility for Fe-SA in the catalyst micropores.²⁸ During MEA fabrication and fuel cell testing, the complex properties of the catalyst layer (CL), such as ionomer distribution, porosity and interfacial engineering also exert a significant influence on the overall performance. Recent studies have revealed that the gradient ionomer distribution in CL can improve PEMFC performance. For instance, a lower ionomer gradient at the gas transport inlet of the cathode catalyst layer can achieve a more uniform oxygen distribution at high current density.²⁹ On the other hand, a composite catalyst layer consisting of a small-pore Pt-rich layer and a large-pore Pt-poor layer can enhance the transportation of the reactants as well as improve the Pt utilization,³⁰ which further emphasizes the importance of catalyst layer design in MEA.

Additionally, higher porosity in the gas diffusion layers (GDLs) reduces mass transport resistance, as well as balancing water retention and oxygen accessibility.³¹ Simultaneously, a thinner GDL is a way of reducing the mass transfer distance. However, under high relative humidity (RH), liquid water drainage is hindered due to the gradient of the liquid water content under the channel and the rib.³² Through a high-temperature sintering process, the hydrophobicity and electrical conductivity of Toray GDL can be finely regulated, leading to optimized PEMFC performance under a wide RH range (0–80%).³³ Beyond that, researchers have also carried out graded design in the microporous layer (MPL), including graded pore size, porosity and hydrophobicity, which also made large improvements to the PEMFC performance.³⁴ Despite the significant progress made with Fe-SA catalysts, several challenges remain to be addressed for their practical implementation in PEMFCs. (1) Due to the harsh acidic environment, the long-term stability of Fe-SA catalysts should be further improved. (2) The relatively lower activity of Fe-SA catalysts necessitates higher catalyst loadings, which also leads to nonnegligible mass transport limitations. (3) As a highly integrated system, the performance of a fuel cell is influenced by the catalyst, gas diffusion layer and operation conditions, requiring systematic research based on newly designed catalysts.

Herein, an Fe-based SAC with a large specific surface area and high heteroatom content was carefully designed for the ORR in PEMFCs. For practical purposes, the catalyst production was extended to 5.13 g for a single batch. Both physicochemical analysis and electrochemical testing indicate the good consistency and high catalytic activity of the Fe-SA catalyst with promising potential for commercial production. Based on the specific characteristics of this catalyst, the structure of the MEA was optimized to accelerate the mass transfer process and improve the utilization of Fe-SA. This work attempts to explore a transformation path for ORR catalyst research and development from the lab to real application.

Results and discussion

As the support for Fe-SA catalysts, metal-nitrogen-doped carbon (MNC) was synthesized based on a ZIF-8 precursor via a carbonization process with KCl protection. As shown in Fig. S1,† the mean particle size of MNC is 233 nm. Fe-based catalysts with different morphology can be easily obtained through a simple adsorption and annealing process at different temperatures. After low-temperature annealing (800 °C), the polyhedron structure of the ZIF-derived carbon particles remained in Fe/MNC-800, as shown in Fig. 1a. Fe and N elements are uniformly distributed over the catalyst particles. Simultaneously, a lot of Zn atoms can also be detected in Fe/MNC-800 (Fig. 1b), due to the relatively lower temperature than the boiling point of Zn metal (907 °C). With the increase of the annealing temperature to 900 °C, the edges of carbon particles become rougher (Fig. 1c), which is similar to our previous research.³⁵ Fe and N elements are still highly dispersed, as shown in Fig. 1d. However, the percentage of Zn element dramatically decreases from 3.7% for Fe/MNC-800 to almost 0 for Fe/MNC-900 (Fig. S2†). As the temperature increases further, some metal particles wrapped by a carbon matrix can be observed in Fig. 1e. Element mapping of Fe/MNC-1000 indicates that the bright spots in the high-angle annular dark field (HAADF) image obtained via scanning transmission electron microscope (STEM) are Fe-based metal particles (Fig. 1f). As a key parameter, the annealing temperature can change the property of a catalyst from two aspects: (1) changing the composition of the catalyst due to the evaporation of low-boiling-point metal atoms and N-containing species; (2) changing the morphology of carbon particles due to the formation of Fe nanoparticles and catalytic rearrangement of carbon atoms at high temperature.

Fig. 1 Morphology of Fe/MNC catalysts. The TEM images and corresponding STEM-HAADF images and heteroatom distribution mapping of (a and b) Fe/MNC-800, (c and d) Fe/MNC-900 and (e and f) Fe/MNC-1000.

To quantitatively analyse the porous structure evolution of different catalysts, N₂ adsorption and desorption isotherms were recorded and simulated by nonlocal density functional theory (NLDFT) method. The isotherms of the different catalysts are similar, as exhibited in Fig. 2a. From Fe/MNC-800 to Fe/MNC-900, the BET specific surface area (Fig. 2b) increases from 990.4 m² g⁻¹ to 1450.7 m² g⁻¹, which can be attributed to the evaporation of residual Zn metal in the MNC support. As the annealing temperature further increases to 1000 °C, the BET specific surface area conversely decreases to 1016.9 m² g⁻¹. Meanwhile, the micropore area and micropore volume show a similar tendency to the BET specific surface area. As shown in Fig. 2c, the annealing parameters directly affect the micropore distribution of Fe-based catalysts. In Fe/MNC-800, the micropores were occupied by residual Zn, leading to a small micropore volume and low specific surface area. With the increase of the annealing temperature, the amount of micropores obviously increases. However, excessive annealing temperature also leads to the destruction of the micropores in Fe/MNC-1000, because of the damage to the ZIF-derived microstructure, which is accompanied by agglomeration of Fe atoms, as shown in Fig. 2d. No obvious diffraction peaks can be observed in the catalysts that were annealed below 900 °C, indicating the atomically dispersed Fe in these catalysts. For Fe/MNC-950, a narrow peak arises at 44.6°, attributed to the (110) plane of Fe³⁶ (PDF #06-0696). With the further increase of the pyrolysis temperature to 1000 °C, more diffraction peaks are observed at 35.6°, 57.1° and 62.9°, which are respectively attributed to the (311), (511) and (440) planes of Fe₂O₃ crystal³⁷ (JCPDS No. 39-1346). X-ray diffraction (XRD) analysis further confirms the severe loss of Fe-SA active sites due to the agglomeration at high annealing temperature, indicating that the optimized annealing temperature is 900 °C.

Fig. 2 Microstructure analysis of Fe/MNC catalysts. (a) Nitrogen adsorption and desorption isotherms and the corresponding (b) surface area and pore volume parameters and (c) pore size distribution curves. (d) XRD patterns of different Fe/MNC catalysts.

The variation of the content and chemical state of different elements were investigated using X-ray photoelectron spectroscopy (XPS). In Fig. 3a, the Zn signal can be easily distinguished at around 1000 eV in the XPS survey curves, implying the existence of residual Zn in the low-temperature annealed catalysts. The signal completely disappeared in Fe/MNC-900, which was consistent with the STEM-EDS analysis. As shown in Fig. 3b, with the increase of the annealing temperature, the trend for C and N elements is completely the opposite. In addition, the content of N element decreases at an accelerated rate over 900 °C. As the primary active component, the percentage of Fe does not exhibit any significant change (around 1 at%) below 950 °C. For Fe/MNC-1000, the percentage of Fe drops sharply to 0.58 at% due to the formation of Fe-based nanoparticles wrapped by carbon shells, which can be observed in the high-resolution TEM image in Fig. S3.† To gain a deep understanding of the chemical state of the elements, the N 1s XPS curves were deconvoluted into five peaks (Fig. 3c) at 398.4, 399.1, 400.2, 401.1 and 402.6 eV, attributed to pyridinic N (N-1), metal-N (N-2), pyrrolic N (N-3), graphitic N (N-4) and oxidized N (N-5), respectively.²⁷ In primitive MNC, the N content is more than 13 at%, which mainly exists as N-1 and N-3. After annealing with the ferruginous precursor, the percentage of N-2 increased to 21.75 at% for Fe/MNC-800. Meanwhile, N-3 rapidly decreases to 14.06 at%. With increasing annealing temperature, the N-2 content decreases, due to the breaking of the Fe–N bond and agglomeration of metal atoms. For Fe/MNC-1000, only 4.1 at% of N was retained in the final catalyst and most N species are N-4 and N-5. The high-resolution XPS curves of Fe 2p can be deconvoluted into three peaks at 709.8, 711.7 and 716.7 eV, attributed to Fe²⁺, Fe³⁺ and satellite signals. As shown in Fig. 3d, compared with MNC, the Fe signals were significantly enhanced after active site loading. With the increase of the temperature, the ratio of Fe³⁺/Fe²⁺ increased from 1.81 for Fe/MNC-800 to 3.27 for Fe/MNC-900, indicating the optimized Fe-based active sites.^38,39 When the annealing temperature is further increased to 1000 °C, this ratio dramatically decreases to 1.02, due to metal agglomeration, which further verified the TEM and XRD results.

Fig. 3 Element content and chemical states analysis of MNC and Fe/MNC catalysts. (a) XPS survey curves and (b) corresponding C, N, O and Fe percentages of Fe/MNC catalysts. The high-resolution XPS curves and corresponding deconvoluted curves of (c) N 1s and (d) Fe 2p.

The XPS depth profiles of different catalysts were investigated by sputtering method. As shown in Fig. 4a, a Zn signal can be detected both on the surface and in the deep layers of the Fe/MNC-800 catalyst. With the increase of the annealing temperature, this signal disappears completely, indicating the complete removal of Zn species, which further validates the XPS conclusion above. Based on the XPS depth profiles (Fig. 4b), the distribution of different elements in the catalyst particles can be summarized into the following three characteristics. (1) The vast majority of oxygen elements are distributed on the surface layer, which may be due to the long-term exposure to air. (2) From the surface layer to the deepest layer, the content of N decreases gradually, and lower temperature annealing results in a higher proportion of N. (3) Fe is evenly distributed at different depths. However, harsh annealing conditions will result in a rapid decrease in the detectable proportion.

Fig. 4 Structure analysis of active components in different catalysts. (a) The XPS survey curves and (b) element content curves of Fe-based catalysts with different etching levels (EL). (c) The normalised Fe K-edge X-ray absorption spectra of Fe reference materials and Fe/MNC-900. (d) The corresponding Fe K-edge extended X-ray absorption fine structure (EXAFS) shown in R-space. (e) Wavelet transform (WT) of k²-weighted EXAFS signal of reference materials and sample using Morlet wavelet with κ = 10, σ = 1.

To analyse the coordination environment of Fe sites accurately, X-ray absorption spectroscopy (XAS) tests, including X-ray absorption near-edge structure (XANES) and extended X-ray absorption fine structure (EXAFS), were conducted with Fe/MNC-900 and reference samples. Compared with Fe phthalocyanine (FePc), the Fe absorption edge of Fe/MNC-900 shows an obviously higher first pre-edge peak (Fig. 4c), indicating that the Fe-based active sites have a higher oxidation state and lower symmetry owing to the strong electronic interaction between the Fe species and the defect-rich carbon support.⁴⁰ The corresponding Fourier transform (FT) curves are shown in Fig. 4d. Only a prominent peak can be observed at 1.46 Å for the Fe/MNC-900 catalyst, indicating the existence of an Fe–N bond and the absence of an Fe–Fe bond (2.18 Å). Furthermore, the wavelet transform (WT, Fig. 4e) of Fe/MNC-900 exhibits only one intense peak at around 4 Å⁻¹, which is attributed to an Fe–N bond. The EXAFS fitting results are shown in Fig. S4.† The key parameters are summarized in Table S3.† The coordination number of Fe–N is 4.2 ± 0.7, suggesting a deduced structure of Fe–N₄. Based on the TEM, XRD and XAS analysis, Fe/MNC-900 shows many attractive features for ORR application, including a large specific surface area and high N and Fe-SA contents.

Linear sweep voltammetry (LSV) was used to investigate the ORR catalytic activity of the Fe-SA catalysts. As shown in Fig. 5a, Fe/MNC-900 exhibits the highest half-wave potential (E_1/2), reaching 0.825 V versus a reversible hydrogen electrode (RHE). As shown in Fig. S5,† both excessively high and excessively low annealing temperatures will lead to obvious attenuation of the ORR activity. Combining the results of physicochemical characterization and electrochemical testing, annealing at 900 °C for 1 h is considered optimal for the coordination structure and dispersibility of the Fe-SA active site. To meet the requirements of PEMFC application, single-batch production of the ZIF precursor and catalyst was increased to >20 g (Fig. 5c) and >5 g (Fig. 5d), respectively. The scanning electron microscopy image of the gram-level produced ZIF (Fig. 5e) indicates the relatively wide range of ZIF particle size. However, the mean particle size is still 279 nm (Fig. S6†), which is almost the same as that of the mg-level ZIF sample. After pyrolysis and active site loading, ∼25% of the mass of the ZIF precursor is retained, 5.13 g of Fe/MNC-900 catalyst can be obtained in a single batch (denoted as g-Fe/MNC-900, Fig. 5f). During the electrochemical activity test, the E_1/2 of g-Fe/MNC-900 reaches 0.827 V versus RHE, which can be further increased to 0.838 V versus RHE by increasing the catalyst loading to 0.89 mg cm⁻². Even though the catalyst loading on RDE is reduced to 0.38 mg cm⁻², the E_1/2 is maintained at 0.812 V versus a RHE (Fig. 5b), indicating the outstanding repeatability of this Fe-SA catalyst for scale production. An accelerated degradation test (ADT) was conducted by cyclic voltammetry (CV) method from 0.3 V to 1.1 V versus a RHE. As shown in Fig. S7,† after 5000 cycles of CV scanning, the E_1/2 of Fe/MNC-900 only decreased from 0.833 V to 0.821 V versus a RHE, indicating the outstanding electrochemical stability.

Fig. 5 Electrochemical performance analysis and gram-level production of Fe/MNC-900 catalyst. (a) LSV curves of Fe/MNC catalysts prepared with different annealing temperatures. (b) LSV curves of Fe/MNC-900 catalyst obtained by gram-level production. The photos of (c) the ZIF precursor and (d) the Fe/MNC-900 catalyst synthesized in a single batch. The corresponding SEM images of (e) the ZIF precursor and (f) the Fe/MNC-900 catalyst.

Due to the relatively lower intrinsic activity of the ORR, a non-Pt-derived catalyst layer is always thicker than a Pt-based catalyst layer, leading to considerable mass transfer problems. To enhance the PEMFC performance, the GDL and cathode catalyst loading were carefully adjusted in this work. As shown in Table 1, four GDLs with different thickness, porosity, microporous layer (MPL) and hydrophobicity were employed as the GDL in the cathode side for PEMFC testing (Fig. S8†). Firstly, two commercial GDLs (AvCarb P40T and Toray 060) without MPL were used during MEA fabrication. As shown in Fig. 6a and a', the arrangement of carbon fibers in AvCarb P40T is relatively loose (calculated porosity is 85.07%), with obvious hydrophilicity. During the contact angle test, the water droplet can be absorbed rapidly. In comparison, Toray 060 is relatively hydrophobic with a larger contact angle of 110.6°, as shown in Fig. 6b and b'. During the PEMFC test (Fig. 6e), MEA-2 (based on Toray 060) shows a much higher maximum power density (MPD) compared with that of MEA-1 (based on AvCarb P40T). Clearly, hydrophilicity impedes the efficient mass transfer of water in the GDL and catalyst layer, leading to poor PEMFC performance. The introduction of MPL significantly reduces the percentage of large pores in the GDL (10 to 100 μm). Meanwhile, some smaller pores with diameter of 100 nm can be observed in Fig. 6f and S9,† which may be derived from the carbon particle packing in the MPL. The lab-made TF140 (thickness: 140 μm, Fig. 6c and c') and commercial SGL 22BB (thickness: 215 μm, Fig. 6d and d') with super-hydrophobicity were also employed as GDLs for PEMFC testing, denoted as MEA-3 and MEA-4, respectively. Unexpectedly, MEA-3 shows the lowest MPD (284 mW cm⁻²) of all the MEAs. Conversely, the MPD of MEA-4 can reach 803 mW cm⁻², which is 2.8 times that of MEA-3. As the contact angles and porosity of these GDLs are similar, the primary factor may come from the GDL thickness and MPL structure. To obtain in-depth insights, electrochemical impedance spectroscopy (EIS) analysis of different MEAs was carried out at 0.1 A cm⁻² and 0.5 A cm⁻². Compared with MEA-2 and MEA-4, the resistance of MEA-1 and MEA-3 in the high-frequency region is relatively larger at 0.1 A cm⁻² (Fig. 6g), indicating the low ionic conductivity of the hydropenic PEM. When the current density increases to 0.5 A cm⁻² (Fig. 6h), the high-frequency resistance of MEA-1 decreases dramatically, due to the hydrophilic GDL, which also presents a serious mass transfer problem, leading to unsatisfactory PEMFC performance. Conversely, the PEM in MEA-3 is still hydropenic because the ultra-thin and super-hydrophobic GDL cannot retain the necessary water. The hydrophobicity of GDL also has a significant impact on the stability of the PEMFC (Fig. 7a). At 500 mA cm⁻², MEA-1 shows smaller voltage attenuation (∼34%) after a 15 h galvanostatic test compared with that of MEA-4 (∼62%). The improvement of the stability can be attributed to the optimization of the water management in both the gas diffusion layer and the catalyst layer of the cathode side. Therefore, the appropriate thickness of the GDL and superhydrophobicity are the key parameters to enhance the performance of Fe-SA-based PEMFCs.

Table 1 The parameters of different cathode GDLs in PEMFCs

	Anode	Cathode catalyst loading (mg cm^−2)	Cathode GDL	CGDL thickness (μm)	MPL	Calculated porosity	Contact angle (°)
MEA-1	Johnson Matthey 40 wt% Pt/C, 0.1 mgPt cm^−2, GDL: SGL 22BB	1.2	AvCarb P40T	200	No	85.07%	—
MEA-2		1.2	Toray 060	190	No	75.51%	110.6
MEA-3		1.2	TF140	140	Yes	74.98%	152.9
MEA-4		1.2	SGL 22BB	215	Yes	75.34%	155.8
MEA-5		0.8	SGL 22BB	215	Yes	75.34%	155.8
MEA-6		1.6	SGL 22BB	215	Yes	75.34%	155.8

---

Fig. 6 Gas diffusion layer and performance analysis of PEMFC. SEM images of (a) AvCarb P40T, (b) Toray 060, (c) lab-made TF140 and (d) SGL 22B. (a'–d') Corresponding contact angle analysis of different GDLs. (e) The polarization curves of MEAs with different GDLs. (f) Cumulative intrusion curves of different GDLs. EIS curves of different PEMFCs at a constant current density of (g) 0.1 A cm⁻² and (h) 0.5 A cm⁻².

Fig. 7 Galvanostatic curve, polarization curve and EIS of g-Fe/MNC-900-based PEMFC. (a) The galvanostatic curves of MEA-1 and MEA-2 under H₂–O₂ supplied without backpressure. The polarization curves of g-Fe/MNC-900-based PEMFC with different catalyst loading under (b) H₂–O₂ and (c) H₂–air supplied with backpressure of 100 kPa. The EIS curves of Fe-based PEMFCs with different catalyst loadings under H₂–O₂ supply without backpressure at (d) 0.1 A cm⁻² and (e) 0.5 A cm⁻².

The relationship between the cathode catalyst loading of g-Fe/MNC-900 and the PEMFC performance was also investigated by polarization curve and EIS. As shown in Fig. 7b, the MPD reached 803 mW cm⁻² when the cathode catalyst loading was set at 1.2 mg cm⁻², indicating outstanding activity compared with previous research (Table S4†). Under H₂ and air supply, the MPD also can reach 410 mW cm⁻² (Fig. 7c), exhibiting the application potential in low-cost PEMFC fabrication. In comparison, lower catalyst loading (0.8 mg cm⁻²) leads to insufficient active site density, resulting in larger charge transfer resistance in the cathode (Fig. 7d). Higher catalyst loading (1.6 mg cm⁻²) leads to a thicker catalyst layer, causing larger ohmic resistance and obvious mass transfer resistance, as shown in Fig. 7e. Based on the above analysis, the remarkable fuel cell performance of the g-Fe/MNC-900 catalyst can be attributed to the following three characteristics: (1) Large specific surface area (1450.7 m² g⁻¹) providing adequate sites for Fe-SA loading, leading to high intrinsic activity for ORR. (2) Optimized GDL with high hydrophobicity and appropriate thickness enhances the mass transfer process and increases the utilization of active sites. (3) The lower catalyst loading with a thin catalyst layer reduces the ohmic and mass transfer resistance and improves the performance of the fuel cell.

Conclusions

Optimizing the structure of Fe-SA-based ORR catalysts is essential for high PEMFC performance. However, the MEA design is also important to ensure the accessibility of Fe-SA active sites in the catalyst layer. Based on a ZIF-derived carbon support, Fe-based ORR catalysts with high specific surface area and plenty of single atom active sites were designed and synthesized. After careful modulation of the annealing temperature, the half-wave potential of Fe/MNC-900 increased to >0.82 V versus a RHE, with good stability and high activity retained during gram-level production. To accelerate the transformation from catalyst design to fuel cell application, the g-Fe/MNC-900 catalyst was evaluated in a PEMFC and adapted with different GDLs and catalyst loading. Under H₂–O₂ supply, the maximum power density can reach 803 mW cm⁻² with only 1.2 mg cm⁻² of catalyst in the cathode, demonstrating the attractive potential of Fe/MNC-900 for practical application in PEMFCs.

Data availability

The data supporting this article have been included as part of the ESI.†

Author contributions

W. Z. conceived and designed the experiments. W. Z. and B. Z. synthesized the MNC carbon support and Fe/MNC catalysts. S. Y. carried out the electrochemical tests of different catalysts. A. D. and Y. L. assisted with the SEM and XRD tests. X. L. assisted with the data analysis. W. Z. Y. S. and Z. L. wrote and revised the manuscript with the contributions from all authors.

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

This work was financially supported by CATARC Automotive Test Center (Tianjin) Co., Ltd Youth Fund (TJKY2425005) and CATARC Automotive Test Center (Tianjin) Co., Ltd Cultivation Fund (TJKY2425009).

Notes and references

1. Y. Wang, Y. Pang, H. Xu, A. Martinez and K. S. Chen, Energy Environ. Sci., 2022, 15, 2288–2328.
2. D. Xue, Y. Yuan, Y. Yu, S. Xu, Y. Wei, J. Zhang, H. Guo, M. Shao and J. N. Zhang, Nat. Commun., 2024, 15, 5990.
3. Q. Chen, Z. Zhang, R. Zhang, M. Hu, L. Shi and Z. Yao, Processes, 2023, 11, 361.
4. Y. Zhao, Z. Shen, J. Huo, X. Cao, P. Ou, J. Qu, X. Nie, J. Zhang, M. Wu, G. Wang and H. Liu, Angew. Chem., Int. Ed., 2023, 62, e202308349.
5. D. Xue, P. Yuan, S. Jiang, Y. Wei, Y. Zhou, C.-L. Dong, W. Yan, S. Mu and J.-N. Zhang, Nano Energy, 2023, 105, 108020.
6. D. Li, Q. Zhang, Z. Shen, K. Siddharth, L. Chen, M. Shao and Z. Shi, Nano Energy, 2022, 91, 106644.
7. W. Qi, W. Huang, J. Niu, B. Zhang, Z. Zhang and W. Li, Appl. Surf. Sci., 2021, 540, 148325.
8. L. Ran, Y. Xu, X. Zhu, S. Chen and X. Qiu, ACS Nano, 2024, 18, 750–760.
9. Y. Li, Z. Li, K. Shi, L. Luo, H. Jiang, Y. He, Y. Zhao, J. He, L. Lin, Z. Sun and G. Sun, Small, 2024, 20, 2309727.
10. H. Zhang, Y. Li, L. Zeng and Y. Pan, Small, 2024, 20, e2307384.
11. P. Song, M. Luo, X. Liu, W. Xing, W. Xu, Z. Jiang and L. Gu, Adv. Funct. Mater., 2017, 27, 1700802.
12. T. Zhang, F. Wang, C. Yang, X. Han, C. Liang, Z. Zhang, Y. Li, A. Han, J. Liu and B. Liu, Chem Catal., 2022, 2, 836–852.
13. Z. Miao, X. Wang, M.-C. Tsai, Q. Jin, J. Liang, F. Ma, T. Wang, S. Zheng, B.-J. Hwang, Y. Huang, S. Guo and Q. Li, Adv. Energy Mater., 2018, 8, 1801226.
14. Z. Wen, S. Ci, F. Zhang, X. Feng, S. Cui, S. Mao, S. Luo, Z. He and J. Chen, Adv. Mater., 2012, 24, 1399–1404.
15. Z.-y. Mei, S. Cai, G. Zhao, Q. Jing, X. Sheng, J. Jiang and H. Guo, Energy Storage Mater., 2022, 50, 12–20.
16. P. Nematollahi, B. Barbiellini, A. Bansil, D. Lamoen, J. Qingying, S. Mukerjee and E. C. Neyts, ACS Catal., 2022, 12, 7541–7549.
17. Y. Zhu, J. Li, Y. Chen, J. Zou, Q. Cheng, C. Chen, W. Hu, L. Zou, Z. Zou, B. Yang and H. Yang, ACS Catal., 2021, 11, 13020–13027.
18. J. Hu, W. Liu, C. Xin, J. Guo, X. Cheng, J. Wei, C. Hao, G. Zhang and Y. Shi, J. Mater. Chem. A, 2021, 9, 24803–24829.
19. H. A. Gasteiger, S. S. Kocha, B. Sompalli and F. T. Wagner, Appl. Catal., B, 2005, 56, 9–35.
20. Z. Yang, Y. Chen, S. Zhang and J. Zhang, Adv. Funct. Mater., 2023, 33, 2215185.
21. Y. Deng, Y. Dong, G. Wang, K. Sun, X. Shi, L. Zheng, X. Li and S. Liao, ACS Appl. Mater. Interfaces, 2017, 9, 9699–9709.
22. J. Yan, X. Zheng, C. Wei, Z. Sun, K. Zeng, L. Shen, J. Sun, M. H. Rümmeli and R. Yang, Carbon, 2021, 171, 320–328.
23. C. Gao, S. Mu, R. Yan, F. Chen, T. Ma, S. Cao, S. Li, L. Ma, Y. Wang and C. Cheng, Small, 2022, 18, 2105409.
24. W. Zhu, Y. Pei, Y. Liu, J. Zhang, Y. Qin, Y. Yin and M. D. Guiver, ACS Appl. Mater. Interfaces, 2020, 12, 32842–32850.
25. Y. C. Wang, W. Huang, L. Y. Wan, J. Yang, R. J. Xie, Y. P. Zheng, Y. Z. Tan, Y. S. Wang, K. Zaghib, L. R. Zheng, S. H. Sun, Z. Y. Zhou and S. G. Sun, Sci. Adv., 2022, 8, eadd8873.
26. X. Ao, Y. Ding, G. Nam, L. Soule, P. Jing, B. Zhao, J. Y. Hwang, J.-H. Jang, C. Wang and M. Liu, Small, 2022, 18, 2203326.
27. W. Zhu, H. Liu, Y. Pei, T. Liu, J. Zhang, X. Liu, L. Wang, Y. Feng, Y. Yin and M. D. Guiver, Small, 2023, 19, e2302090.
28. Y. Li, P. Zhang, L. Wan, Y. Zheng, X. Qu, H. Zhang, Y. Wang, K. Zaghib, J. Yuan, S. Sun, Y. Wang, Z. Zhou and S. Sun, Adv. Funct. Mater., 2021, 31, 2009645.
29. R. Ding, Y. Cheng, X. Fan, N. Wang and L. Zhang, Int. J. Hydrogen Energy, 2024, 79, 208–220.
30. L. Ye, Y. Gao, S. Zhu, J. Zheng, P. Li and J. P. Zheng, Int. J. Hydrogen Energy, 2017, 42, 7241–7245.
31. P. Abraham B and K. Murugavel K, Electrochim. Acta, 2021, 389, 138793.
32. H. Liu, Z. Hu, J. Li, L. Xu, Y. Shao and M. Ouyang, Energy, 2023, 282, 128780.
33. X. L. Wang, Z. G. Qu and G. F. Ren, J. Power Sources, 2023, 575, 233077.
34. J. Zhang, B. Wang, J. Jin, S. Yang and G. Li, Renewable Sustainable Energy Rev., 2022, 156, 111998.
35. W. Zhu, Y. Shao, B. Zhou, S. Yin, J. Zhang, Y. Yin, A. Dong, Y. Liu, X. Liu and Z. Li, Small, 2024, e2403098.
36. B. Jiang, N. Jiang, Y. Cui, H. Wang, G. Zhang, J. Li and Y. Zhang, Adv. Sci., 2024, 11, e2402610.
37. W. Yan, Y. Wu, Y. Chen, Q. Liu, K. Wang, N. Cao, F. Dai, X. Li and J. Jiang, J. Energy Chem., 2020, 44, 121–130.
38. J. Wang, C. Hu, L. Wang, Y. Yuan, K. Zhu, Q. Zhang, L. Yang, J. Lu and Z. Bai, Adv. Funct. Mater., 2023, 33, 2304277.
39. E. Murphy, B. Sun, M. Ruscher, Y. Liu, W. Zang, S. Guo, Y. H. Chen, U. Hejral, Y. Huang, A. Ly, I. V. Zenyuk, X. Pan, J. Timoshenko, B. R. Cuenya, E. D. Spoerke and P. Atanassov, Adv. Mater., 2024, 36, e2401133.
40. Y. Wang, K. Li, R. Cheng, Q. Xue, F. Wang, Z. Yang, P. Meng, M. Jiang, J. Zhang and C. Fu, Chem. Eng. J., 2022, 450, 138213.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d4ta08289c

---