1. Introduction

The oxygen reduction reaction plays a vital role in energy conversion and storage applications such as fuel cells and metal– air batteries. The electrocatalysts for ORR are crucial in determining the energy conversion efficiency because the ORR is a kinetically sluggish reaction [1, 2]. Up to now, platinum (Pt) based catalysts are regarded as the best catalysts [3, 4]. However, poor durability and high cost of the available Pt-based catalysts restricts the commercialization and application of fuel cell power systems [5–7]. Therefore, seeking for suitable alternatives with low cost, high efficiency and durability for ORR has become a major research focus in the past decades.

Highly efficient nonprecious metal and metal-free catalysts have attracted numerous researchers' attention due to their superior activity, comparable stability and enhanced durability against CO or methanol [8–11]. Many research efforts have been devoted to develop porous carbon based materials as catalyst supports [12–16]. High surface area, porous architecture and high conductivity facilitate gas diffusion, electron transfer [17–20]. Nitrogen-doped carbon materials represent one of the most promising electrocatalysts for ORR due to their unique electronic and structure properties [21–25]. The strong electron-withdraw property of N atom could alter the charge density and spin density of the adjacent C atoms, which facilitate the adsorption of O₂ and weaken the O-O bond [26]. Apparently, four-electron pathway is preferential to proceed based on the above mentioned conditions. However, the ORR performance of Ndoped carbons is still inferior to the commercial Pt-based catalysts. An agreement has been reached that Fe is important for the ORR activity of the N-doped carbons [27–31]. However, the real active sites for N-dope carbon with Fe catalysts are still under discussion [32].

Herein, we prepare a red-blood-cell (RBC) like Fe, N-C@carbon sphere ORR catalyst (detailed synthetic scheme are shown in Scheme 1). Nitrogen-containing active sites could be incorporated into the adjacent carbon matrix with pyrolysis treatment. Polystyrene (PS) are chemically modified with sulfonate group and used as the template. The polyaniline (PANI)@PS composite is pyrolyzed to prepare the catalyst Fe, N-C@carbon. The high surface area and core–shell interconnection could provide rich active sites and good electronic transfer pathway. This catalyst has the similar morphology with RBC to adsorb and diffuse O₂. The catalyst exhibits superior catalytic activity for ORR with the onset potential (E_onset) of 0.87 V and the half-wave potential (E_1/2) of 0.78 V in 0.1 mol/L KOH solution, which is close to those of commercial Pt/C. It is found that the carbon sphere support of the catalyst ruptured at high pyrolysis temperature. The ruptured morphology dramatically increases the active sites and the catalytic performance of the Fe, N-C@Carbon catalyst improves significantly. This work provides the solid evidence that this novel morphology and the introduction of Fe boost the catalytic activity of Fe, N-C@carbon catalyst.

2. Results and discussion

As shown in Fig. 1, the morphology of the resulting RBC-like PANI@PS composites were characterized by SEM and TEM. A thin hydrophilic interior layer and a permeable transverse channel, comprised of poly(methyl methacrylate)-poly(methacrylic acid) (PMMA-PMA), were embedded in the PS shell of the hollow template spheres with an average diameter of 400 nm (Fig. 1a and b) [38]. As shown in Fig. 1c–f, nanosized uniform trichoid PANI was formed on the exterior surface of the PS template after polymerization. Repeated centrifugation and wash with deionized water before polymerization ensured that PANI pillars mostly grew on the PS latex. The resulting RBC-like PANI@PS composites were obtained with the PANI pillars growing radially [34, 38]. Some hollow spheres were broken to confirm that the spheres were hollow (Fig. 1g–i). We could easily see that the ruptured morphology dramatically increased the active sites and the catalytic performance of the Fe, N-C@carbon-900 catalyst improved significantly.

The emeraldine form of polyaniline was confirmed by the FT-IR spectra shown in Fig. 2a. The two moderate peaks at 1570 cm^-1 and 1490 cm^-1 corresponded to the C=C stretching deformation of quinoid and benzenoid rings, respectively. The absorption bands at 1297 cm^-1 and 1132 cm^-1 were assigned to the stretching of secondary aromatic amine C-N and in-plane bending of the aromatic C-H. The typical out-of-plane deformation of C-H of 1, 4-disubstituted benzene ring were observed at 797 cm^-1 and 505 cm^-1, respectively. These characteristic peaks indicated that PANI possessed the conductive emeraldine structure.

The Raman spectra of carbon sphere and the catalyst Fe, NC@carbon-900 displayed two typical peaks about 1320 cm^-1 and 1592 cm^-1, corresponding to the D bands and G bands, respectively (Fig. 2b) [39, 40]. The D bands were assigned to the defective nature of carbon. The G bands denoted to graphitic carbon which were attributed to the coupling effect between the in-plane-bond stretching of C=C and the hexagonal sheets of graphene [41–43]. The intensity ratio of D band and G band (I_D/I_G) elucidated information about the degree of the defects and disordered structures. The decreased I_D/I_G ratio from 1.315 to 1.086 for the carbon sphere and the catalyst Fe, N-C@carbon-900 gave the evidence that higher degree of defects existed in the carbonized PS template and the introduction of Fe and N contributed to produce the graphitic structure. As for the catalyst Fe, N-C@carbon-900, relatively stronger peak intensity of the D band than the G band also provided the information that a certain amount of edges/ defects existed in the Fe, N-C@carbon. It might also play a crucial role to enhance the ORR catalytic activity [44, 45]. Furthermore, we carried out XRD analysis (Fig. 2c) to understand structural transformation involved during the pyrolysis process. The NC@carbon-900 showed a characteristic peak at about 27.3°, which corresponded to the stacking peak of π-conjugated layers and (002) phase of graphitic materials [46]. Graphite structure show good electronic conductivity which is essential for the electrocatalysts. When introducing FeCl₃, peaks of FeCl₃·nH₂O, iron hydroxide, ferric oxide, Fe–C and ferriccarbide are clearly found according to the XRD results of Fe, N-C@carbon-900. As the annealing temperature increased, Fe was gradually generated. The higher catalytic activity of the Fe, N-C@carbon heat-treated at 900 ℃ indicated that the forms of Fe in the catalyst at this temperature enhanced the number of active sites of the catalysts. It is speculated that, during the heat treated process, iron might react with the nitrogen contained carbon derived from PANI in situ and catalyzed to form more active sites with the existence of Fe and N at the same time under the proper annealing temperature.

Furthermore, TGA analysis has been performed with Fe, NC@carbon-900 precursor under N₂ atmosphere with a scan rate of 10 ℃/min in the range of 25–900 ℃. The possible mechanism involves the following reactions [47]:

(1)

(2)

(3)

(4)

As shown in Fig. 2d, weight loss below 158.6 ℃ is due to the loss of absorbed water molecules. At this stage, FeCl₃·nH₂O is hydrolyzed into amorphous Fe(OH)₃ (as shown in the Eq. (1)). The weight loss is rather obvious at between 158.6 ℃ and 453.9 ℃, which is assigned to the decomposition of low molecular weight polymers or oligomers of PS and PANI, along with the loss of H₂O released from the decomposition of Fe(OH)₃ (Eq. (2)). The weight loss starts at 453.9 ℃ and ends at around 563.4 ℃, due to the decomposition of FeO(OH) (Eq. (3)). During the range of 158.6– 563.4 ℃, the total weight loss of Fe, N-C@carbon is 61.6%. The weight loss between 563.4 ℃ and 755.6 ℃ indicated that Fe₂O₃ reacted with carbon (Eq. (4)). At higher temperature from 755.6 ℃, Fe₂O₃ gradually transformed into Fe.

Cyclic voltammetries (CVs) (Fig. 3a) of the prepared electrocatalysts were carried out in 0.1 mol/L saturated KOH. The catalyst Fe, N-C@Carbon-900 showed a remarkable reduction peak at 0.820 V and the highest peak current -303.92 μA, which were almost as better as those of commercial 20 wt.% Pt/C (0.780 V, -316.63 μA, respectively). In contrast, carbon-900 exhibits ORR peak potential at 0.750 V and a small peak current -141.93 μA, which meant poor ORR catalytic activity. The catalyst N-C@carbon-900 showed more positive peak potential (0.770 V). Fe, NC@carbon-700 (0.758 V) and Fe, N-C@carbon-800 (0.763 V) showed more positive peak potential than the carbon sphere and less positive peak potential than the Fe, N-C@carbon-900. The results indicated that both nitrogen-doped carbon and Fe enhanced ORR catalytic activity and the proper pyrolysis temperature of the catalysis was 900 ℃.

Furthermore, to clearly understand the intrinsic electrocatalytic activity, the ORR activities of the catalysts were characterized by linear sweep voltammograms (LSVs). Specifically, a series of RDE studies of carbon sphere-900, N-C@carbon-900, Fe, N-C@carbon-900 and Pt/C were collected on RDE at 1600 rpm (Fig. 3b). The onset potential (E_onset) of the LSVs curves was used to evaluate the activity of catalysts. And the E_onset is defined as the potential is 5% of the ORR current measured at 0.33 V, at which the diffusionlimiting current for all the catalysts was obtained. The E_onset of Fe, N-C@Carbon-900 is 0.87 V, slightly more negative than Pt/C (0.96 V). The catalyst Fe, N-C@carbon-900 showed superior activity in comparison to carbon sphere-900 and N-C@carbon-900, with the onset potential of 0.82 V and 0.85 V, respectively. The half-wave potential (E_1/2) of the Fe, N-C@carbon-900 (0.78 V) was very close to the E_1/2 of Pt/C (0.82 V). The comparative E_1/2 of carbon sphere-900 and N-C@carbon-900 were 0.69 V and 0.75 V. The current density at 0.4 V of the Fe, N-C@carbon-900 (5.20 mA/cm²) was close to Pt/C (5.85 mA/cm²) and much higher than carbon sphere-900 (3.43 mA/cm²) and N-C@carbon-900 (4.34 mA/cm²). Moreover, the unique and the wide current plateau from 0.2 V to 0.6 V of the Fe, NC@carbon-900 indicated a diffusion-controlled process for the four-electron-dominated ORR pathway [48]. The remarkable catalytic performance of the Fe, N-C@carbon-900 was due to its large surface areas, unique porous structure and accessible catalytic active sites. It could be concluded that both Fe and N improved the catalytic activities.

To further understand ORR activity of Fe, N-C@carbon-900, LSVs studies at different rotation rates were performed (Fig. 3c). The catalytic current density increased with the increase of rotation rate. The electron transfer numbers (n) and kinetic limiting current density are also calculated based on Koutecky–Levich (K–L) equation. All K–L plots displayed good linearity from 0.2 V to 0.6 V. The electron transfer number for Fe, N-C@carbon-900 was 3.94 on average and suggested good selectivity for the fourelectron dominated pathway for ORR, close to that of the commercial Pt/C with an average n value of 4.00 (Fig. 3d). The large n value and high limiting current density (J_K) with a wide current plateau from 0.2 to 0.6 V confirmed the superior and stable electrochemical performance for Fe, N-C@carbon-900.

Furthermore, RRDE measurements were employed to assess the electrocatalytic activity of the catalysts (shown in Fig. 4a and b). If oxygen is reduced through a two-electron pathway, hydrogen peroxide (H₂O₂) would generate as the intermediate. The Fe, NC@carbon-900 exhibited an ORR process through a four-electron transfer pathway with the H₂O₂ yield of 9.3% and the average n value of 3.83 under the potential range from 0.55 to 0.90 V. By comparison, Pt/C produced 3.2% H₂O₂ with the average n of 3.94 under the same condition.

The heat-treated temperature played a crucial role on the electrocatalytic activity of the catalysts [49, 50]. The precursors of the catalysts were treated at 700 ℃, 800 ℃ and 900 ℃ (denoted as Fe, N-C@carbon-700, Fe, N-C@carbon-800, Fe, N-C@carbon-900, respectively) to confirm the proper pyrolysis temperature for the Fe, N-C@carbon catalysts. The LSVs studies at different rotation rates of Fe, N-C@carbon-700 and Fe, N-C@carbon-800 were performed. Because too high pyrolysis temperature was not beneficial to incorporate N with C, but helped to increase the electrical conductivity for the carbon matrix. The ORR activities for the catalyst Fe, N-C@carbon-X (X = 700, 800, 900) increased with increasing the treatment temperature. The ORR E_onset and E_1/2 in the RDE polarization curves at 1600 rpm (Fig.S1 in Supporting information) were 0.67 and 0.50 V for Fe, N-C@carbon-700 and 0.72 and 0.56 V for Fe, N-C@carbon-800, which were both inferior to that of Fe, N-C@Carbon-900, which agreed well with the CV results. Thus we chose 900 ℃ as the optimal pyrolysis temperature for the catalysts.

Next, we also carried out the comparison of the catalyst NC@carbon-900 of which the precursors contained N but Fe and the catalyst carbon-900 without Fe or N at different rotation rates (Fig. S2 in Supporting information). Apparently, the catalyst carbon-900 proceeds mainly by a two-electron ORR mechanism. After introducing N, the catalyst N-C@carbon-900 were shown through a four-electron process with more positive potential and higher limiting current density. The higher activity of the N-C@carbon-900 than carbon-900 were attributed to that electron-rich N atom induced charged sites of the adjacent carbon favorable adsorption of O₂ [51–53]. In association with the ORR performance of Fe, NC@carbon-900, we could reach a conclusion that Fe and N exhibited synergistically promotion in the ORR activity forthe catalyst Fe, N-C@carbon-900.

3. Conclusion

In summary, we have developed a high-performance ORR catalyst with a novel RBC-like structure through pyrolysis treatment of the PANI@PS composites. The unique flat sphere shells afforded abundant catalytic sites of the catalyst Fe, NC@carbon-900 and facilitated the mass and electron transfer process during the oxygen reduction process. The Fe, N-C@carbon-900 exhibited the ORR E_onset at 0.87 V, E_1/2 at 0.78 V and high diffusion-limiting current density (5.20 mA/cm²), which were approaching that of Pt/C in alkaline medium. Furthermore, this work indicated that both N and Fe accounted for the high activity of the catalyst Fe, N-C@carbon-900 toward the oxygen reduction process. The precursor PANI was heat treated at 900 ℃ to ensure the nitrogen sites distributed uniformly on the carbon matrix. It could come to the conclusion that Fe and N exhibited synergistically promotion in the ORR activity for the catalyst Fe, N-C@carbon-900. We also provided a rational design of electrocatalysts with high ORR activity to further clarify the essential ORR sites of heteroatom doped carbon materials for fuel cells and metal–air battery applications.

4. Experimental 4.1. Materials and chemicals

Aniline, hydrochloric acid and ferric chloride were purchased and used as received. PS (polystyrene) hollow spheres were purchased from Rhom & Haas Company. Nafion 117 solution (5 wt. %) was purchased from Alfa Aesar (USA). Pt/C catalyst (20 wt.% Pt on Vulcan XC-72R) was obtained from ETEK Inc (USA). All the aqueous solutions were prepared by the ultra-pure water ( > 18 MΩ cm) from a Millipore system.

4.2. Materials characterization

The microstructures of PANI@PS composites and the nitrogen doped RBC-like carbon catalyst were represented by field-emission scanning electron microscopy (SEM; HITACHI, SU-8010) and transmission electron microscopy (TEM; JEOL, JEM-2100). The crystalline phases were further confirmed by X-ray diffraction (XRD) pattern (XRD; SHIMADZU, Lab X XRD-6000). Thermogravimetric analysis (TGA) was conducted on a PerkinElmer TGA7 thermogravimetric analyzer with a heating rate of 10 ℃/ min under nitrogen flowing from 30 ℃ to 900 ℃. Raman scattering spectra were conducted on a HORIBA JOBIN YVON Instrument HR 800 with a laser wavelength of 514 nm. The FT-IR of PANI layer on PS surface was performed by Fourier transform infrared spectroscopy (FT-IR, BRUKER, TENSOR 27).

4.3. Electrochemical measurements

Electrochemical measurements were performed on a Pine biopotentiostat (Pine Instrument Co., USA) using a conventional three-electrode system with the platinum wire as the counter electrode, Ag/AgCl as the reference electrode and glassy carbon electrode (D = 5 mm) modified with a certain load of 4 mg/mL catalyst suspension as the working electrode. The ORR activity and four-electron selectivity of catalysts were evaluated using a rotating disk electrode (RDE) and a rotating ring disk electrode (RRDE), respectively.

A mixture of 4 mg catalysts in 1 mL of water/ethanol (4/1, v/v) and 10 mL 5% Nafion was sonicated for 30 min to form a homogeneous suspension. 20 mL of the suspensionwas cast on glassy carbon electrode surface and then dried at room temperature. As comparison, a commercial Pt/C catalyst suspension was prepared following the same procedure. The rotation rate of RDE was set from 400 rpm to 2025 rpm. The electron transfer number (n) was calculated according to the Koutechy–Levich equation:

Where J stands for the measured current density, J_L the limiting diffusion current density, J_K the kinetic current density, ω the RDE rotation rate, n the electron transfer number, F the Faraday constant (96485C/mol), D_O2 the diffusion coefficient of O₂ in 0.1 mol/L KOH (1.73 ×10^-5 cm²/S), ν the kinematic viscosity of 0.1 mol/L KOH (0.01 cm²/S) and C_O2 the bulk concentration of the dissolved O₂ (1.2 ×10^-6mol/cm).

The RRDE (D = 5.61 mm) measurements was conducted at a rotating rate of 1600 rpm. The ring potential was 0.5 V (vs. Ag/ AgCl). The electron transfer number (n) and percentage content (X) of the intermediate production HO₂^- were calculated according to the following equation:

Where I_d is the disk current, I_r is the ring current and N is the current collection efficiency (37%) of the Pt ring.

All measurements were carried out in 0.1 mol/L KOH aqueous solution under Oxygen (or nitrogen) atmosphere. O₂ (or N₂) was bubbled into the electrolyte solution for at least 30 min to ensure the saturation of oxygen (or nitrogen). For CV measurements, at least 10 cycles of CV, ranging from -0.8 V to 0.2 V (vs. Ag/AgCl), were performed before the data collected. The Ag/AgCl electrode was calibrated against the reversible hydrogen electrode (RHE) according to the previously reported procedure [33]. E_RHE = E_{Ag/ AgCl} + 1.0225 V in 0.1 mol/L KOH.

4.4. Synthesis of the RBC-like PANI@PS composites

The PANI@PS was synthesized through ferric chloride induced oxidative polymerization of aniline [34, 35]. For a typical procedure, 0.04 mol of aniline was added into HCl aqueous solution (30 mL, 2 mol/L) under stirring. Afterwards 30 mg crosslinked sulfonated PS hollow spheres [36, 37] were added into this solution. The mixture was kept for 24 h at room temperature to ensure aniline diffusion into the PS hollow spheres via the channels within the shell. The aniline hydrochloride coated sulfonated PS were centrifuged and washed by deionized water for several times. In situ oxidative polymerization was triggered when the treated monomers were added into 100 mL FeCl₃ aqueous solution (0.1 mol/L) and stirred for 24 h at room temperature. The emeraldine PANI@PS composites were obtained after centrifugation with deionized water and dried in air.

4.5. Preparation of the nitrogen-doped carbon catalyst

The obtained composites were centrifuged and washed with FeCl₃ solution and dried in air. The FeCl₃ adsorbed RBC-like PANI@PS composites were subjected to heat treatment at 700 ℃ (Fe, N-C/carbon-700), 800 ℃ (Fe, N-C/carbon-800) and 900 ℃ (Fe, N-C/carbon-900) for 2 h under nitrogen atmosphere, respectively. For comparison, the RBC-like PANI@PS composites without FeCl₃ solution treatment and the crosslinked sulfonated PS template without PANI or Fe³⁺ were pyrolyzed at the same condition as the Fe, N-C/carbon-900 and were denoted as N-C/carbon-900 and carbon sphere-900, respectively.

Acknowledgment

This work was partially supported by National Natural Science Foundation of China (Nos. 51273158, 21303131).

Appendix A. Supplementary data

Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.cclet.2016.12.006.