100 mg of CNTs without having any pretreatment was dispersed in the solution mixtures of deionized water (40 mL) and ethanol (30 mL). Dopamine (200 mg) in Tris-HCl (30 mL, 50 mM, pH≈8.5) solution was then added. The reaction was allowed to proceed under magnetic stirring for 24 h at room temperature. The obtained CNTs@polydopamine (CNTs@PDA) nanocomposite was collected by centrifugation and washed with deionized water and then dried at 60℃ for 12 h. Then, the CNTs@PDA was stirred with FeCl₃ in deionized water to coordinate Fe (Ⅲ) ions with -OH in PDA structure. After pyrolysis of CNTs@PDA-Fe under N₂ atmosphere for 2 h at 800 ℃, the CNTs@Fe/N/C catalysts were obtained by HCl washing. The resulting materials were designated as "CNTs@XFe/N/C", where X represents the weight ratio of FeCl₃ to CNTs@PDA (X =0, 20%, 40% and 60%).

Transmission electron microscopy (TEM) was carried out on a Zeiss LIBRA 200 FETEM instrument operating at 200 kV. Raman spectrum was recorded by a LabRamHR evolution Raman spectrometer equipped with an Nb-Yag laser excitation source operated at 532 nm. X-ray Photoelectron Spectroscopy (XPS) was conducted on a PE PHI-5400 spectrometer equipped with a monochromatic Al X-ray source (Al KR, 1.486 6 keV).

All electrochemical experiments were performed in a standard three-electrode cell at room temperature. The cell consisted of a glassy carbon working electrode (GC electrode, d=5 mm), an Ag/AgCl reference electrode, and a carbon rod counter electrode. All potentials in this study are given relative to a reversible hydrogen electrode (RHE). The working electrodes were prepared by applying catalyst ink onto glassy carbon (GC) disk electrodes. The loading of the as-prepared catalyst is 0.5 mg·cm^-2.

All of the electrodes were pretreated by cycling between 0 and 1.2 V (vs. RHE) at a sweep rate of 50 mV·s^-1 for 50 cycles in order to remove any surface contamination prior to ORR activity testing.

The activities of catalysts were performed by recording linear sweep voltammetry (LSV) curves in the oxygen-saturated 0.1 M KOH solution. The LSV curves for ORR were recorded at potential scan rate of 10 mV·s^-1. The rotation speed was controlled at 1 600 rpm.

Rotating disk electrode (RDE) measurements were conducted at rotation speeds from 100 to 2 500 rpm using a VersaSTAT3 (V₃). RDE measurements were conducted in O₂-saturated 0.1 M KOH solution at 10 mV·s^-1 scan rate. The electron transfer number (n) was analysed on the basis of Koutecky-Levich equations shown in equations:

in which J is the measured current density, J_K and J_D are the kinetic- and diffusion-limiting current densities, $\omega $ is the angular velocity of the disk ($\omega = 2\pi V$, N is the linear rotation speed), n is the overall number of electrons transferred in ORR, F is the Faraday constant (F = 96 485 C·mol^-1), C₀ is the bulk concentration of O₂, ʋ is the kinematic viscosity of the electrolyte, and k is the electron transfer rate constant. According to Equations (1) and (2), the number of electrons transferred (n) and J_K can be obtained from the slope and intercept of the Koutecky-Levich plots, respectively, by using the values of C₀ = 1.2×10^-3 mol·L^-1, D₀ = 1.9×10^-5 cm s^-1 and $\upsilon $ = 0.1 m²·s^-1 in 0.1 M KOH.

The CV accelerating stress tests (AST) were performed at potentials between 0 and 1.2 V vs. RHE at a scan rate of 50 mV·s^-1 in nitrogen-purged 0.1 M KOH at room temperature. RDE cycling stability tests of the catalysts were performed in N₂-saturated 0.1 M KOH in the potential range of 0 to 1.2 V vs. RHE.

The morphology of the samples was first investigated by transmission electron microscopy (TEM). The typical TEM image in Fig. 2a evidences that the product has cable-like one-dimensional (1D) structures. Under a higher magnification (Fig. 2b), the core region of the 1D structure can be identified to be CNTs, and the surface was uniformly covered by a layer of NC with a thickness of~6 nm. Fig. 2c shows that the morphology of the product does not change much after Fe introduction, except that the thickness of the PDA layer is slightly decreased to~4 nm. This could be ascribed to the decomposition and gasification of PDA with the aid of FeCl₃ at 800 ℃. Moreover, high-resolution TEM image in the insert of Fig. 2d clearly reveals the lattice fringes in the core region have an interplanar distance of ca. 0.34 nm, corresponding to the (002) plane of graphitic carbon. On the other hand, the outer sheath is more disordered, suggesting the amorphous nature of the carbon coating layer^[14]. These CNTs could serve as a three- dimensional conductive network in catalysts which facilitates the electron migration.

Figure 2.  TEM images

The defected structures of CNTs@Fe/N/C were evaluated by Raman spectra. As shown in Fig. 3a, there are two peaks at around 1 340 cm^-1 and 1 590 cm^-1 for all as-prepared samples, assigned to D band and G band of graphite carbon, respectively. The D band suggests the existence of substantial defects or disordered sites in the composite, whereas the G band refers to the graphitic structure. The calculated I_D/I_G ratio for CNTs@Fe/N/C catalyst is about 1.09~1.15, higher than that for CNTs@N/C (1.07) samples, implying that the number of defects in CNTs@Fe/N/C increased after the Fe introduction during the carbonization process.

Figure 3.  Physical characterization

The porous nature of these catalysts was further characterized by N₂ adsorption-desorption experiments. The adsorption isotherm obtained for CNTs@Fe/N/C catalyst exhibits a behavior of type-IV with H4 hysteresis loop (Fig. 3b) with all the important values summarized in Table 1. The Brunauer-Emmett-Teller (BET) specific surface area of pristine CNTs was measured to be only 84.3 m²·g^-1. When the pyrolysis process was carried without Fe, CNTs@N/C shows a very low S_BET of only 137.9 m²·g^-1, suggesting a poorly developed porosity. With the increasing Fe content, the S_BET dramatically increases from 213.6 m²·g^-1 for CNTs@20%Fe/N/C to 301.4 m²·g^-1 for CNTs@40%Fe/N/C, and it drops to 256.5 m²·g^-1 for CNTs@60%Fe/N/C (Table 1). The pore size distribution curves of CNTs@Fe/N/C (Fig. 3c, 3d) exhibit sharp peaks centered at about 2.0 and 15 nm, confirming their mesoporous structures. Remarkably, the high BET surface area and the enriched porous structure are favorable for mass transfer of reactants and products, and able to offer more exposed active sites. All these profitable features could contribute to an enhanced ORR activity.

The chemical states of the elements of these samples were determined by X-ray Photoelectron Spectroscopy (XPS). The relative contents of N element and different N species of CNTs@N/C and CNTs@Fe/N/C catalysts are summarized in Table 1. It is apparent that the N content of CNTs@N/C without Fe remains at a high level around 2.25 at%, however, when the Fe content is increased from 20% to 60%, the N content exhibits a gradually decreasing trend. It can be observed that the increase of Fe content will increase the BET specific surface area of the materials and lead to the decrease of N doping content, which is consistent with our previously reported result^[15]. The N 1s XPS spectrum (Fig. 5c) can be deconvoluted into three main peaks centered at 398.6, 400.5, and 401.3 eV, which can be assigned to pyridinic-N, pyrrolic-N and graphitic-N, respectively. These N species play a very important role as an active center for ORR^[16-18].

Figure 4.  High-resolution N 1s spectra

Figure 5.  ORR catalytic activities

The ORR catalytic activities of the as-prepared catalysts were first investigated using cyclic voltammetry (CV) in 0.1 M KOH electrolyte. As shown in Fig. 5a, the CV curve of CNTs@40%Fe/N/C obtained in O₂-saturated KOH electrolyte displayed a well-defined cathodic peak at approximately 0.81 V, which is absent in the CV obtained in the N₂-saturated electrolyte, indicating its excellent electrocatalytic activity for ORR. To investigate its comprehensive ORR catalytic activity, linear sweep voltammetry (LSV) curves of CNTs@Fe/N/C and commercial Pt/C catalyst were obtained using a rotating disk electrode (RDE). The ORR activity of CNTs@Fe/N/C increases quickly with the increasing Fe content and reaches the maximum for CNTs@40%Fe/N/C with regard to both diffusion limiting current density at 0.4 V (5.12 mA cm^-2) and half-wave potential (0.881 V), which is comparable to that of commercial Pt/C. Further increasing the Fe content to 60% leads to a slight decrease in the onset/half-wave potential of the CNTs@60%Fe/N/C catalyst, which can be ascribed to the decreased specific surface area (see Table 1). The superior ORR activity of CNTs@40%Fe/N/C can thus be attributed to the optimal exposure of the ORR active sites as a result of the porous micro/mesoporous structure.

To gain further insight into the electron transfer process in ORR catalysis of the CNTs@40%Fe/N/C catalyst, a serial of RDE measurements on CNTs@ 40%Fe/N/C in 0.1 M KOH at different rotating speeds was carried out (Fig. 5c). Fig. 6d presents the Koutecky-Levich (K-L) plots at 0.40 V, 0.50 V and 0.60 V calculated from the LSV curves at the various rotating speeds (100~2 500 rpm). The K-L lines for 0.40 V, 0.50 V and 0.60 V are almost superimposed, indicating that ORR occurs on these potentials in a similar way. The average number of transferred electrons during the ORR was 3.50, indicating the near four electron selectivity of the catalyst in alkaline media.

The stabilities and possible crossover effects of catalytic materials are also very important for practical applications. The LSV curves of the CNTs@40%Fe/ N/C was evaluated at 10 mV·s^-1 in an O₂-saturated 0.1 M KOH solution as well as in an O₂-saturated 0.1 M KOH solution with the addition of 0.5 M CH₃OH. As shown in Fig. 5e, no noticeable changes were observed in the ORR polarization curves on CNTs@40%Fe/N/C catalyst after adding 0.5 M CH₃OH. This implies that this catalyst has a high selectivity for ORR with a remarkable ability to avoid the crossover effect and holds high promise for use in direct methanol fuel cells. Meanwhile, the stability of the CNTs@40%Fe/N/C catalyst was determined before and after proper potential sweeps between 0 and 1.2 V with a scan rate of 50 mV·s^-1 in N₂-saturated 0.1 M KOH. The ORR half-wave potential still remained the same as the original value after 5 000 cycles and only the diffusion limiting current decreases ca. 0.26 mA·cm^-2, evidently highlighting the outstanding electrocatalytic stability of the sample for ORR in an alkaline electrolyte.

In summary, we reported a simple method to synthesize core-shell nanowire CNTs@Fe/N/C as an efficient electrocatalyst for ORR. The CNTs@Fe/N/C catalysts are composed of CNT cores and porous Fe/N-doped carbon shells derived by pyrolysis of PDA at 800 ℃ with the aid of FeCl₃. The CNT core plays a role of a three-dimensional conductive network for fast electron transfer. At the same time, porous Fe/N-doped carbon layer offers numerous catalytic active sites and fast mass transfer for ORR. Benefited from the well-defined structure and rich actives sites, the CNTs@40%Fe/N/C catalyst presented an ORR activity comparable to the commercial Pt/C in alkaline media and also showed superior durability and resistance to the methanol crossover. This work may shed light on the future development of practical non-noble metal- based catalysts for fuel cells.