Iron Carbide Nanoparticles Embedded in Edge-Rich, N and F Codoped Graphene/Carbon Nanotubes Hybrid for Oxygen Electrocatalysis

Abstract

Rational design of cost-effective and efficient bifunctional oxygen electrocatalysts for sluggish oxygen reduction reaction (ORR) and oxygen evolution reaction (OER) is urgently desired for rechargeable metal–air batteries and regenerative fuel cells. Here, the Fe₃C nanoparticles encapsulated in N and F codoped and simultaneously etched graphene/CNTs architecture catalyst (Fe₃C@N-F-GCNTs) was synthesized by a simple yet cost-effective strategy. The as-prepared Fe₃C@N-F-GCNTs exhibited excellent ORR and OER performances, with the ORR half-wave potential positive than that of Pt/C by 14 mV, and the OER overpotential lowered to 432 mV at the current density of 10 mA·cm⁻². In addition, the ΔE value (oxygen electrode activity parameter) increased to 0.827 V, which is comparable to the performance of the best nonprecious metal catalysts reported to date. When it was applied in a Zn–air battery as a cathode, it achieved a peak power density of 130 mW·cm⁻², exhibiting the potential for large-scale applications.

1. Introduction

Environmental friendly and sustainable energy conversion and storage systems, such as rechargeable metal–air batteries and fuel cells, are urgently required to effectively mitigate the current energy and environmental crisis [1,2]. However, their development and commercialization are hindered by the sluggish kinetics of oxygen reduction reaction (ORR) and oxygen evolution reaction (OER) [3]. To date, Pt-based materials are the most active catalysts for the ORR, while their OER activities are far from satisfactory. Ir- and Ru-based materials are widely used to accelerate the OER, whereas such catalysts exhibit insufficient ORR performance. Moreover, the pretty price, terrestrial scarcity, and low electrochemical durability of precious metals seriously limit their widespread applications [4]. Therefore, searching for an efficient way to fabricate low-cost, highly efficient, bifunctional oxygen electrocatalysts based on earth-abundant elements still remains a crucial challenge for researchers.

Recently, heteroatom-doped carbon-based materials have emerged as promising ORR/OER bifunctional electrocatalysts because of the reasonable balance among catalytic activity, stability, and cost [5,6,7,8,9]. Since the heteroatom (e.g., N, S, B, P, and F) and carbon atom have different electronegativity and atomic radius, the replacement of carbon atoms in graphitic carbon skeleton by heteroatoms can induce inhomogeneous electron distribution and efficiently modulate electron properties and introduce more active sites, thus boosting the bifunctional electrocatalytic oxygen activities [10,11]. Among various heteroatoms, nitrogen has attracted more attention due to its similar atomic radius and five valence electrons for bonding with carbon atoms. Therefore, N atoms can be efficiently doped into a carbon matrix to enhance the oxygen electrode electrocatalytic activity. Moreover, compared with single N heteroatom doping, multi-heteroatoms codoped into a carbon skeleton can further improve the ORR/OER catalytic kinetic rates due to the synergistic effects arising from the codoped heteroatoms [12,13]. Since fluorine atom has the highest electronegativity, doping fluorine atoms can cause great surface polarization of the carbon matrix [8]. Therefore, because of the synergistic effect between N and F atoms leading to the maximum charge delocalization of C atoms, N and F codoped carbon materials have enhanced electrocatalytic activity [9,14,15,16,17]. In addition, 3D transition metal can promote the graphitization of carbon materials in the carbonization process, thus improving the electrical conductivity and strengthening the antioxidant capacity of the carbon-based catalysts [18]. At the same time, 3D transition metals are often the active components for oxygen electrocatalysis [19,20]. Moreover, 3D transition metals can catalyze carbon precursors to form carbon nanotubes or graphene [21,22]. The graphene/CNTs hybrid is considered as one of the most advanced carbon substrates for ORR/OER due to its corrosion resistance; high electrical conductivity; and special structure, which is conducive to the transmission of oxygen and removal of water [23,24]. In addition to heteroatom doping modification, defect engineering (such as etching or edging) of nanocarbon electrocatalysts could also effectively enhance their electrocatalytic activity in both experimental and theoretical studies [25,26]. However, etching of carbon materials is often prepared by a post-treatment method (such as laser etching or chemical activation) [27,28], and the preparation method is complicated.

In this study, we ingeniously choose polyvinylidene fluoride (PVDF), melamine, and iron acetate as precursors to in situ synthesis of Fe₃C encapsulated in edge-rich, N and F codoped graphene/carbon nanotubes hybrid electrocatalysts by an activation-free method. It is worth noting that the PVDF not only plays a role as a carbon and fluorine resource, but also could provide edge-sites and microporous structures by an activation-free method, as reported in the literature [29]. By taking advantage of the hierarchical porosity, enhanced electric conductivity, and the high density and multiple types of active sites, the resulting Fe₃C@N-F-GCNTs hybrid exhibits outstanding ORR and OER performance.

2. Results and Discussion

TEM and high-resolution TEM (HRTEM) images of Fe₃C@N-F-GCNTs, Fe-F-C, and N-F-C are shown in Figure 1 and Figure S1. Fe-F-C (Figure S1a,b) and N-F-C (Figure S1c,d) both exhibit graphenelike morphologies, while Fe₃C@N-F-GCNTs (Figure 1a,b) show an open 3D architecture coexisting of graphene and bamboolike carbon nanotubes. A large number of literatures have proved that the coexistence of melamine and iron salts is conducive to form a bamboolike, Fe₃C-encapsulated, carbon nanotubes structure [30,31,32]. The 3D architecture coexisting of graphene and bamboolike carbon nanotubes effectively inhibits the stacking of individual graphene sheets or carbon nanotubes, benefiting to expose more active sites, promote rapid electron transfer, and supply fast transport channels for O₂ diffusion and electrolyte infiltration, thereby improving the electrochemical performance. It can be seen that some nanosized holes on the surface of the graphene/CNTs substrate (partially highlighted with red open circles in Figure 1b,c) are generated in Fe₃C@N-F-GCNTs, resulting from the release of HF from the carbon chain in PVDF at high temperature under inert atmosphere or some metal particles on the surface getting leached out after subsequent acid treatment process [33]. It could also be confirmed by the TEM images of Fe-F-C and N-F-C (Figure S1), where the nanosized holes could also be seen on the graphene sheet’s surface. Both edge effects and topological defects have been proved to be favorable for both ORR and OER [26,34]. HRTEM images (Figure 1d) show that Fe₃C nanoparticles are encapsulated in a 2.5 nm thick graphite carbon layer with a lattice spacing of 0.34 nm, corresponding to the (002) planes of graphitic carbon. The structure of the graphite carbon layer encapsulating the Fe₃C nanoparticles not only inhibits the dissolution and agglomeration of the Fe₃C nanoparticles under the harsh alkaline/acidic solution, but also adjusts the electron density of the carbon surface to promote the surface catalytic reaction [10]. We further investigated the elemental distribution in Fe₃C@N-F-GCNTs using high-angle, annular, dark-field scanning TEM (HAADF-STEM). As shown in Figure 1e, all elements are evenly distributed throughout the sample except for the nanoparticle region. Two types of Fe species exist in the carbon matrix: one is atomically dispersed and the intensities of Fe and N match up to a certain degree, indicating that it may exist in the form of Fe-N_x, which is generally considered to be the active site of ORR; the other in the nanoparticle region has a few crystalline phases associated with Fe₃C, which has been proved by XRD.

Figure 2 shows the XRD patterns of Fe₃C@N-F-GCNTs, Fe-F-C, and N-F-C. All the XRD patterns show two broad diffraction peaks centered at 26.4° and 44.4°, corresponding to the (002) and (101) reflections of the graphitic peak, respectively (PDF#41-1487). It is noteworthy that the graphite peak of Fe₃C@N-F-GCNTs is much sharper and stronger than those of Fe-F-C and N-F-C, implying the higher graphitization of Fe₃C@N-F-GCNTs. The other diffraction peaks of Fe₃C@N-F-GCNTs were characteristic of Fe₃C (PDF#35-0772).

The N₂ adsorption–desorption isotherms (Figure 3a) were recorded to investigate the pore structure and surface area of Fe₃C@N-F-GCNTs. A typical type IV characteristic for the N₂ adsorption–desorption isotherms with hysteresis loops is observed, indicating the existence of plentiful mesopores. The BET surface area of Fe₃C@NF-GCNTs is 421 m²·g⁻¹. The corresponding pore size distribution shows the coexistence of micropores, mesopores, and macropores (Figure 3b) and the pore volume is as high as 0.4 cm³·g^–1. This result indicates that the hierarchically porous structure has been successfully fabricated, which facilitates mass transport and active site exposure. As a contrast, the N₂ adsorption–desorption isotherm curves and the corresponding pore size distribution curves of N-F-C and Fe-F-C were also recorded. As shown in Figure S2a, the N₂ adsorption–desorption isotherms of the N-F-C have type I characteristics, with a steep initial region at P/P₀ = 0–0.03, indicating the catalyst’s microporous structure. The corresponding pore size distribution curve shows that the pore diameter is of ca. 1.1 nm (Figure S2b), which confirms that microporous carbon can be prepared by the pyrolysis of PVDF through an activation-free method, as reported in the literature. The N₂ adsorption–desorption isotherm curve of Fe-F-C also shows a typical type IV characteristic, with a hierarchical tetramodal micro–meso–macro pore size distribution (Figure S2c,d). The specific surface area and pore volume of N-F-C and Fe-F-C are found to be 976.0 m²·g⁻¹ and 0.1 cm³·g⁻¹, and 951.2 m²·g⁻¹ and 0.4 cm³·g⁻¹, respectively.

XPS analysis (Figure 4a) shows that Fe₃C@N-F-GCNTs is mainly composed of C (91.87 at.%), O (3.2 at.%), N (3.79 at.%), F (0.72 at. %), and Fe (0.42 at.%), confirming that N, O, F, and Fe were successfully incorporated into the carbon matrix (Table S1). The high-resolution N 1 s spectrum can be fitted to three peaks corresponding to oxidized-N (404.3 eV), graphite N (401.3 eV), pyrrole N (400.1 eV), and pyridine N/Fe-Nx (398.6 eV) (Figure 4b). Among them, graphite N (40.85%) and pyridine N (25.61%) are more catalytically active for ORR than pyrrole N (12.60%) and oxidized-N (20.94%) (Table S2). Figure 4c shows the high-resolution F 1 s spectrum; a main peak at 686.4 eV is attributable to the half-ion C-F bond. As described in the previous literature, charge redistribution induced by electronegativity can promote electrocatalytic reactions. Thus, a significant difference in electronegativity between the C and F atoms can promote electron transfer in the ORR and OER. Figure 4d shows the XPS Fe 2p high-resolution spectrum of Fe₃C@N-F-GCNTs; no obvious Fe peaks are observed for the hybrid, suggesting that the formed carbon shell coatings are hindering core Fe₃C signals [35], in agreement with TEM results. The XPS analyses of Fe-F-C and N-F-C are also presented in Figures S3 and S4.

Figure 5a shows the LSV curves of Fe₃C@N-F-GCNTs, Fe-F-C, N-F-C, and 20 wt.% Pt/C in an O₂-gas-saturated 0.1 M KOH solution. The onset potential and half-wave potential of the catalysts for the ORR increase in the following order: Fe-F-C < N-F-C < Fe₃C@N-F-GCNTs. Among them, Fe₃C@N-F-GCNTs showed the best ORR catalytic performance, and its onset potential and half-wave potential were 0.972 V (vs. RHE) and 0.846 V (vs. RHE), respectively, which were 10 and 14 mV higher than those of commercial Pt/C, indicating better intrinsic catalytic activity than commercial Pt/C. Further, the diffusion limiting current density of Fe₃C@N-F-GCNTs at 0.6 V is 15% higher than that of 20 wt% Pt/C, indicating that its hierarchical framework containing graphene and carbon nanotubes is beneficial to the diffusion of reactants and products. The Tafel slope is an important parameter for assessing ORR activity. Figure 5b shows the Tafel slopes of Fe₃C@N-F-GCNTs, Fe-F-C, N-F-C, and 20 wt.% Pt/C, which are 54, 139, 86, and 68 mV·dec⁻¹, respectively. The lower Tafel slope of Fe₃C@N-F-GCNTs indicates their high intrinsic catalytic activity.

To further understand the ORR process of Fe₃C@N-F-GCNTs, Fe-F-C, and N-F-C, we also performed RRDE measurements. As can be seen in Figure 5c,d, Fe₃C@N-F-GCNTs have the highest electron transfer number (~3.92) and the lowest peroxide yield (4–5%) (in the potential range of 0.3–0.6 V) when testing the ring current and the disk current; N-F-C has an electron transfer number of about 3.58 and a peroxide yield of 17–23%; Fe-F-C achieved the lowest electron transfer number (~3.4) and the highest peroxide yield (26–32% in the same voltage window).

Apart from high ORR activity, the excellent durability also plays a significant role for catalyst in practical applications. The durability of Fe₃C@N-F-GCNTs was assessed by current–time (i–t) chronoamperometry measurement in an O₂-saturated 0.10 M KOH aqueous solution with a rotating speed of 900 rpm at a constant cathodic voltage of 0.6 V. As a contrast, we also performed the same measurement for 20 wt.% Pt/C. As can be seen in Figure 6a, after continuous operation for 30,000 s, Fe₃C@N-F-GCNTs retain a relatively stable current density that only dropped by 16%; however, the Pt/C catalyst exhibits a faster and larger decay of ORR current density, illustrating the superior ORR durability of Fe₃C@N-F-GCNTs compared with commercial Pt/C.

We also characterized the methanol positioning effect of the Fe₃C@N-F-GCNTs. As shown in Figure 6b, after introducing methanol at around 200 s, the Pt/C catalyst exhibits an obvious drop in current density, while Fe₃C@N-F-GCNTs achieve an excellent current retention ratio of 95%. This result suggests that the Fe₃C@N-F-GCNTs cathodic ORR catalyst can be used in methanol-based fuel cells as well.

To investigate the electrocatalytic OER activity of Fe₃C@N-F-GCNTs, the OER LSV test was performed at 1600 rpm in 0.1 M KOH electrolyte solution saturated with O₂. Similar tests were performed on Fe-F-C, N-F-C, and 20 wt.% IrO₂/C (Figure 7a) for comparison. Compared with the standard potential to split water (1.23 V), Fe₃C@N-F-GCNTs can afford 10 mA·cm⁻² at a low overpotential (η) of 432 mV, which is close to the value for IrO₂/C (370 mV), while Fe-F-C and N-F-C reached 10 mA·cm⁻² at higher overpotentials (η) of 656 and 621 mV, respectively.

To study the OER kinetics of the catalysts, we plotted their Tafel curves. As shown in Figure 7b, the Tafel slopes of Fe₃C@N-F-GCNTs, Fe-F-C, and N-F-C are ~140, ~501, and ~224 mV·dec⁻¹, respectively. Fe₃C@N-F-GCNTs showed the smallest Tafel slope, indicating that Fe₃C@N-F-GCNTs have excellent intrinsic OER catalytic activity.

For Fe₃C@N-F-GCNTs, an accelerated stability test was performed at room temperature to study the durability of the OER. As can be seen in Figure S5, after 200 repeated cycles, the η of Fe₃C@N-F-GCNTs increased by only 10 mV.

The bifunctional oxygen electrode electrocatalytic activity of the catalyst can be evaluated by measuring the potential difference (ΔE) between an OER potential at 10 mA·cm⁻² and an ORR potential (−3 mA·cm⁻²). The smaller the ΔE value, the higher the bifunctional electrocatalytic performance. The bifunctional electrocatalytic activity of the above catalyst is shown in Figure 7c. The ΔE value of Fe₃C@N-F-GCNTs is 0.827 V, and their ΔE value is comparable to or even smaller than that of most nonprecious metal-based ORR/OER bifunctional electrocatalysts reported in literature (Table S3) [10,36,37,38,39,40,41], indicating their distinguished bifunctional activity. We also assembled a home-made Zn–air battery using Fe₃C@N-F-GCNTs as the cathode catalyst and Zn plate as the anode to further evaluate the possibility of practical application of Zn–air batteries. As can be seen in Figure 7d, the Zn–air battery with Fe₃C@N-F-GCNTs as cathode catalyst delivers a high power density up to 130 mW·cm⁻², which ranks among the performance of the state-of-the-art carbon-based catalysts reported in the literatures (Table S4) [42,43,44,45,46,47]. Therefore, this catalyst is highly promising for practical applications in a Zn–air battery.

3. Materials and Methods

3.1. Materials

All purchased chemical reagents are of analytically pure grade and can be used directly. Melamine, PVDF, iron acetate, and concentrated sulfuric acid were purchased from Sinopharm Group (Shanghai, China). Nafion solution (5 wt.%) was purchased from Alfa Aesar. Commercial Pt/C (20 wt.%) was purchased from Johnson Matthey. IrO₂/C (20 wt.%) was prepared according to a previously reported method [48].

3.2. Preparation of the Catalysts

A total 2 g of melamine, 1 g of PVDF, and 0.4 g of iron acetate were added into 25 mL of deionized water; then, the suspension was magnetic stirred for 30 min and dried at 80 °C to obtain a catalyst precursor. The precursor was placed in a quartz boat and heat-treated at 900 °C for 2 h under argon atmosphere in a tube furnace. Followed by naturally cooling to room temperature, the sample was leached in a 1.0 M H₂SO₄ aqueous solution at 80 °C for 12 h and washed with deionized water and ethanol three times. Finally, the sample was dried at 60 °C for 12 h and subjected to secondary heat treatment under the same heat treatment conditions as described above; after cooling down to room temperature, the iron carbide nanoparticles embedded in edge-rich, N and F codoped graphene/carbon nanotubes hybrid (Fe₃C@N-F-GCNTs) was obtained. As a comparison, under the same conditions, we prepared Fe-F-C and N-F-C catalysts without melamine or iron acetate, respectively.

3.3. Physical Characterization

X-ray diffraction (XRD) patterns were recorded on a TD-3500 powder diffractometer (Tongda, Dandong, China). Transmission electron microscopy (TEM) images were acquired using a JEM-2100HR transmission electron microscope (JEOL, Tokyo, Japan). X-ray photoelectron spectroscopy (XPS) studies were conducted on an ESCALAB 250 X-ray photoelectron spectrometer (Thermo VG Scientific, Waltham, MA, USA). The specific surface area was analyzed using an adsorption–desorption isotherm curve obtained by a Micromeritics Tristar II 3020 (Micromeritics, Norcross, GA, USA).

3.4. Electrochemical Tests

Electrochemical tests were conducted in a conventional three-electrode electrolytic cell equipped with an electrochemical station (Ivium, Eindhoven, Netherlands). A Pt wire and an Hg/HgO electrode were used as the counter electrode and reference electrode, respectively. A catalyst-loaded glassy carbon electrode (GCE) (diameter, 5 mm) was used as the working electrode. The electrolyte was an aqueous 0.1 M KOH solution. All potentials in the text are quoted with respect to the reversible hydrogen electrode (RHE). The catalyst ink was prepared by dispersing 5 mg of active material in 1 mL Nafion ethanol solution (0.25 wt.%). Then, 20 μL homogeneous ink was dropped onto the GCE surface and dried in air at room temperature. Prior to testing, high-purity N₂ or O₂ gas was purged into the electrolyte solution to saturate it.

The number of electron transfers (n) per oxygen molecule for ORR of the catalyst is calculated by the Koutecky–Levich (K–L) equation [49]:

$$j^{-1}=j_k^{-1}+{\left(0.62nF{C}_0{D}_0{}^{\raisebox{1ex}{$2$}\!\left/ \!\raisebox{-1ex}{$3$}\right.}{\gamma }^{\raisebox{1ex}{$-1$}\!\left/ \!\raisebox{-1ex}{$6$}\right.}{\omega }^{\raisebox{1ex}{$1$}\!\left/ \!\raisebox{-1ex}{$2$}\right.}\right)}^{-1}$$

(1)

where j is the current density, j_k is the kinetic current density, n is the number of electron transfers during the ORR reaction, F is the Faraday constant (F = 96485 C·mol⁻¹), C₀ is the volume concentration of O₂ (C₀ = 1.2 × 10⁻³ mol·L⁻¹), D₀ is the diffusion coefficient of O₂ (D₀ = 1.9 × 10⁻⁵ cm²·s⁻¹), γ is the dynamic viscosity of the 0.1 M KOH solution (V = 0.01 cm²·s⁻¹), and ω is the rotation speed of the disk.

The Tafel curve is obtained by mass-transfer-corrected linear sweep volt–ampere (LSV) data, where the kinetic current density is obtained by diffusion current correction of the current in the hybrid region, which is calculated as follows:

$$j_{kin}=\frac{j{j}_{diff}}{j_{diff}-j}$$

(2)

In the rotating ring disk (RRDE) test, the ring voltage was set to 1.5 V (vs. RHE), and the transferred electron number (n) and H₂O₂ yield could be calculated by the following formula [50]:

$$\begin{gathered} {\mathrm{H}}_2{\mathrm{O}}_2\%=200I_r{\left(N{I}_d+I_r\right)}^{-1} \\ n=4I_d{\left(I_d+I_r{N}^{-1}\right)}^{-1} \end{gathered}$$

(3)

where I_r is the ring current, I_d is the disk current, N is the capture rate of the Pt ring, and the value is 0.37.

4. Conclusions

In summary, we designed and successfully prepared a heterogeneous nano-oxygen double function of the Fe₃C@C nanoparticles embedded in edge-rich, N and F codoped graphene/carbon nanotubes frame structure catalyst (Fe₃C@N-F-GCNTs) by a simple, scalable method. The specific surface area of the catalyst is as high as 421 m²·g⁻¹ with a hierarchical micro–meso–macroporous structure. In alkaline electrolyte, Fe₃C@N-F-GCNTs exhibit outstanding ORR catalytic activity, and their half-wave voltage and diffusion limiting current density are 14 mV and 15% higher than those of the Pt/C catalyst, respectively, and the catalyst is also highly active for OER. Its overvoltage at 10 mA·cm⁻² is only 432 mV, and its ΔE value is only 0.827 V, which is lower than that of commercial Pt/C and most other nonprecious metal bifunctional oxygen electrocatalysts. Furthermore, the home-made Zn–air battery with Fe₃C@N-F-GCNTs as the cathode catalyst delivers a high power density of up to 130 mW·cm⁻². Detailed studies show that the prominent catalytic performance of the material is due to: (1) a synergistic effect between various active sites (including Fe₃C@C, N and F dopant, Fe-N_x-C); (2) the surface etching on the carbon substrate could introduce more active edge sites; (3) high surface area and fast mass transfer caused by the hierarchical pore structure of the graphene/carbon nanotube framework. This work may provide a simple and scalable approach towards the preparation of 3D-etched graphene–nanotube hybrids for energy conversion and storage applications.