Fuel cells are one of the most promising energy conversion devices for providing sustainable and clean energy [1, 2]. The oxygen reduction reaction (ORR) is the cathode reaction in fuel cells and is usually catalyzed by precious metals, especially by state-of-the-art platinum (Pt)-based catalysts, because of their excellent catalytic performance [3]. However, the Pt materials are expensive and suffer from limited natural reserves, sluggish kinetics, poor durability, the crossover effect, and low tolerance to methanol, which pose a key obstacle to its development and commercialization [4-7]. Therefore, the development of low-cost, high-performance non-Pt catalysts to replace Pt-based catalysts is considered as one of the best ways to achieve large-scale commercialization of fuel cells.

To endow non-Pt catalysts with superior ORR performance, various carbon-based materials are a great choice. Carbon-based materials have some unmatched versatility for electrocatalysis including good electrical conductivity, chemical inertness, versatile morphology, and large surface area. Various carbon-based materials have been extensively applied for the ORR such as graphene [8], carbon nanotubes (CNTs) [9], and carbon microspheres [10]. However, pure carbon materials have few active sites and display poor ORR activity. Numerous studies have demonstrated that the doping of heteroatoms (B [11, 12], N [13, 14], P [15, 16], S [17, 18]) within the skeleton of carbon materials can change their electron donor properties and form active sites to boost the ORR [19, 20]. Specifically, N-doped carbon materials with unique structures, favorable flexibility, and excellent mechanical and electrochemical properties have been widely used as effective electrocatalysts for the ORR [21]. In N-modified carbon materials, the nitrogen atoms have a higher electronegativity than that of the carbon atoms. The electron density of the carbon atoms adjacent to the nitrogen atom is lowered, which favors the dissociative chemisorption of oxygen [22]. Furthermore, the earth-abundant transition metals (Fe [23], Co [24], Ni [25], Cu [26]) have also been introduced to enhance ORR activity, which could immobilize the dissociative oxygen species and promote electrical conductivity [23, 27]. In particular, transition metals (such as Co) in N-doped carbon can effectively improve the ORR properties by constructing Co–N_x active sites [28]. Cao et al. [29] synthesized Co and N co-doped bamboo-like CNTs (NBCNT), which showed an enhanced ORR performance compared with individual Co-or N-doped materials. The synergistic effect of the co-doped Co and N promoted the ORR process.

Recently, carbon-based materials derived from carbon-enriched biomass (seaweed [30], fermented rice [31], typha orientalis [32]) have also been widely exploited for the ORR. Biomass is sustainable, renewable, ecofriendly, cheap, and ubiquitous in nature. All these merits imply that biomass is a potential candidate for the ORR for large-scale applications. Paper is one of the most common biomass-derived materials used in daily human life, but its recycling rate is still very low. Wei et al. [33] have prepared Fe-based N-doped carbon materials as ORR catalysts using tissue and filter paper. Carbon materials prepared from waste paper have a large specific surface area. The large specific surface area is favorable for the transmission of ORR-related species (O₂, H⁺, OH^-, and H₂O) and accelerates the reaction rate. Furthermore, it is an environmentally friendly process, which would not only reduce the cost but would also be conducive to large-scale preparation.

Herein, we designed a readily scalable approach for the fabrication of Co and N co-doped porous carbon (Co/N/CNT@PC-800) electrocatalysts, which were prepared through a simple annealing of a mixture of waste paper, dicyandiamide, and cobalt(Ⅱ) acetylacetonate at 800 ℃. Because of the catalytic effect of cobalt, the Co/N/CNT@PC-800 catalyst with a porous structure had many nanotubes on its surface. The numerous micropores and mesopores are beneficial for exposure of the active sites and the transportation of ORR-relevant species. The Co and N co-doped carbon materials can generate more active sites to enhance the ORR activity. For these reasons, the low-cost catalyst exhibited good ORR performance with a quasi-four-electron reaction pathway, good stability, and methanol tolerance in an alkaline medium. This work will open up a new way to utilize biomass waste to obtain value-added materials through an economic and environmentally friendly method.

Cobalt(Ⅱ) acetylacetonate (99%), dicyandiamide, and methanol were obtained from Sinopharm Chemical Reagent Co., Ltd. (Shanghai, China). Nafion (10% wt%) was obtained from Sigma-Aldrich (Missouri, USA). Commercial 20 wt% Pt/C was obtained from Alfa Aesar (China) Chemicals Co., Ltd. (Shanghai, China). The waste paper obtained from our laboratory. (Zhenjiang, China). All these chemicals were used as delivered without further treatment.

The shredded paper was washed with an appropriate amount of distilled water to remove the impurities. Then, the paper pulp was filtered by suction and vacuum dried at 60 ℃ before further use.

The synthesis of Co/N/CNT@PC-800 is shown in Scheme 1. In a typical synthesis, 2.0 g dicyandiamide and 0.5 g of cobalt(Ⅱ) acetylacetonate were dissolved in 50 mL methanol and defined as solution A. Then, 1.0 g pretreated paper was distributed in 50 mL of distilled water to obtain mixture B. Solution A was added into mixture B and then stirred for 12 h. The intermediate product was dried in an oven. Afterwards, the obtained dark-green material was ground and placed into a ceramic boat. The material was heated to 500 ℃ for 2 h with a ramp rate of 5 ℃ min^-1 under an argon atmosphere in a tubular furnace. Then, the temperature was raised to 800 ℃ and maintained for 2 h. Finally, the carbon composite was immersed in 6 mol L^-1 HCl for 12 h to remove any unstable and inactive species.

C-800 was fabricated by using the same procedure as that for Co/N/CNT@PC-800 without adding cobalt(Ⅱ) acetylacetonate and dicyandiamide.

N/C-800 was fabricated by using the same procedure as that for Co/N/CNT@PC-800 without adding cobalt(Ⅱ) acetylacetonate.

Co/PC-800 was fabricated by using the same procedure for Co/N/CNT@PC-800 without adding dicyandiamide.

The morphology and microstructure of the materials were verified by scanning electron microscopy (SEM; JEOLJSM-7001F) and transmission electron microscopy (TEM; JEOL6JEM-2010). The elemental composition of the electrocatalysts was determined by element mapping (JSM-6010PLUS/LA). The crystalline structures of the products were recorded with a Shimadzu XRD-6000 X-ray diffractometer in the 2θ range of 10°–80° with Cu K_α radiation at room temperature (λ = 0.15418 nm). The surface states of the samples were analyzed by X-ray photoelectron spectroscopy (XPS) with a monochromatic Mg K_α source operated at 20 kV. The specific surface area and pore size distribution of the products were measured by adsorption–desorption isotherms of N₂ on a TriStar Ⅱ 3020 surface area and porosity analyzer (Micromeritics Instrument Corporation, USA). The Raman spectra were obtained with a micro Raman spectrometer (Renishaw Invia) in a backscattering geometry with a 532 nm laser as the excitation source.

The ORR experiments were performed in a three-electrode cell connected to an electrochemical analyzer (CHI 760E) at room temperature. A rotating disk electrode (RDE, 3 mm diameter), Pt wire, and Ag/AgCl electrodes were used as working electrode, counter electrode, and reference electrode, respectively. 4 mg of prepared catalyst was distributed in 1 mL mixed solvent (2-propanol/water = 1:3 (v/v)) with 15 μL of 10% Nafion. 20 μL of the catalyst ink was loaded on the surface of the RDE and dried at room temperature. Cyclic voltammetry (CV) measurements were recorded at 50 mV s^-1 from 0.2 to -0.8 V vs. Ag/AgCl. Linear sweep voltammetry (LSV) measurements were performed at a rotating speed from 400 to 2500 r min^-1 at 5 mV s^-1 in 0.1 mol L^-1 KOH solution. O₂ or N₂ were introduced into the electrolyte 30 min before the test.

Koutecky-Levich (K-L) plots were analyzed at different rotation speeds of the ORR polarization curves. The electron transfer number (n) was calculated by using the K-L equation [34]:

(1)

(2)

(3)

in which J is the measured current density, J_k and J_L are the kinetic-and diffusion-limiting current densities, respectively. B is the Levich slope, ω is the angular velocity, F is the Faraday constant (F = 96485 C mol^-1), C_O2 is the concentration of dissolved O₂(C_O2 = 1.2 × 10^-6 mol cm^-3), D_O2 is the diffusion coefficient of O₂ in 0.1 mol L^-1 KOH (D_O2 = 1.9 × 10^-5 cm² s^-1), ν is the kinetic viscosity of the electrolyte (ν = 0.01 cm² s^-1), and k is the electron transfer rate constant.

Rotating ring-disk electrode (RRDE) measurements: In the RRDE system, the rotating speed of the working electrode was fixed at 1600 r min^-1 with a scan rate of 5 mV s^-1. The potential of the ring electrode was set to 0.5 V vs. Ag/AgCl. The electron transfer number and hydrogen peroxide yields were calculated by the following equations [35]:

(4)

(5)

in which I_D and I_R are the disk and ring currents and N is the collection efficiency, which was reported to be 42.4% by the manufacturer.

The SEM and TEM images were used to elucidate the morphology and microstructure of the prepared samples. As shown in Fig. S1(a) and (b), C-800 and N/C-800 exhibited a similar microstructure with a wrinkled surface. This indicated that the introduction of dicyandiamide did not change the original microstructure of the waste paper. After the addition of Co, many pore structures appeared on the surface of the Co/PC-800 catalyst (Fig. S1(c) and (d)). Therefore, the doped Co played a critical role in the formation of the porous structures. As shown in Fig. 1(a) and (b), the surface of the Co/N/CNT@PC-800 catalyst had a porous structure and many carbon nanotubes with diameters of approximately 20 nm. The synergetic coupling of doped Co and N was essential for the formation of carbon nanotubes [29]. As shown in Fig. 1(c) and (d), Co/N/CNT@PC-800 had an obvious hollow tube structure and some residual Co nanoparticles were wrapped in the carbon layer and CNTs. Moreover, the element mapping ascertained that both Co and N were uniformly distributed in the Co/N/CNT@PC-800 catalyst (Fig. S2).

The crystal structure of the C-800, N/C-800, Co/PC-800 and Co/N/CNT@PC-800 catalysts was further characterized by powder X-ray diffraction (XRD) patterns (Fig. 2(a)). The well-defined diffraction peaks of the XRD patterns of the Co/N/CNT@PC-800 and Co/PC-800 catalysts at approximately 2θ = 44.2°, 51.5°, 75.9° were assigned to the (111), (200), (200) plane of Co [28]. All the carbon materials showed a sharp peak corresponding to the (002) plane of the graphite structure. In stark contrast, the C-800 catalyst displayed a broad peak at approximately 23.7°, which signified a low graphitization degree in this sample [36]. Raman spectra were obtained to further analyze the graphitization degree and defects of these catalysts (Fig. 2(b)). The peaks at approximately 1350 cm^-1 (D-band), 1580 cm^-1 (G-band), and 2680 cm^-1 (2D-band) in the Raman spectra indicated a typical characteristic signature of crystalline graphite [37, 38]. The G-band indicated the presence of sp²-hybridized carbon atoms. The presence of defects in the doped carbon materials was indicated by the D-band [39]. The relative intensities of the G-band (I_G) and D-band (I_D) were used to study the characteristics of the carbon materials. A lower I_D/I_G ratio indicates a higher degree of graphitization of carbon materials [40]. As displayed in Fig. 2(b), the I_D/I_G values of the C-800, N/C-800, Co/PC-800, and Co/N/CNT@PC-800 catalysts were 0.98, 1.02, 1.10, and 1.04, respectively. The increase in defects and disorder for N/C-800, Co/PC-800, and Co/N/CNT@PC-800 after Co or N doping was clearly observed. The I_D/I_G ratio of the Co/N/CNT@PC-800 catalyst was lower than that of Co/PC-800. This may be owing to the formation of a large number of carbon nanotubes, which further improved the graphitization degree.

XPS measurements were performed to further elaborate the doped element and bonding states. The survey XPS spectrum of the Co/N/CNT@PC-800 catalyst demonstrated the presence of C (89.75 at%), N (4.43 at%), O (5.6 at%), Co (0.22 at%) (Fig. 3(a) and (b)). The high-resolution N 1s spectrum showed four peaks at 398.45 ± 0.3, 399.53 ± 0.3, 400.73 ± 0.3, and 402.50 ± 0.3 eV, which were pyridinic-N (42.15%), pyrrolic-N (12.79%), graphitic-N (32.70%), and oxidized-N (12.36%), respectively (Fig. 3(c), 3(d) and S3). This suggested that pyridinic-N and graphitic-N were the dominant species, which has been demonstrated to boost O₂ reduction [41]. As shown in Fig. 3(e), there were four types of bonding states in the C 1s spectrum: C =C, C=N & C–O, C=O & C–N, and O=C–O, which indicated the existence of N and O heteroatoms in the hybrids. The high-resolution spectrum of Co 2p_3/2 and Co 2p_1/2 can be fitted with three components in accordance with Co ⁽⁰⁾, Co–N_x and the shake-up peaks (Fig. 3(f)). The peaks at 780.8 and 796.7 eV were assigned to the zerovalent state cobalt, which was consistent with the results of the XRD. The peak at 779.5 eV was the Co–N_x peak, which demonstrated the existence of a coordination bond between Co and N atoms [28].

The specific surface area and pore size of the hybrids were determined by the nitrogen adsorption isotherms. The surface area of the Co/N/CNT@PC-800, Co/PC-800, N/C-800, and C-800 catalysts was 421.146, 304.801, 358.220 and 476.675 m² g^-1 respectively (Fig. 4 and Table S1). The type IV isotherm with a distinct hysteresis loop in the medium-and high-pressure regions (P/P₀= 0.4–1.0) implied the presence of multiple micropores and mesopores in the structure of the prepared catalysts. The pore size distributions diagrams (inset) indicated that the size of the micro/mesopores was between 1 and 10 nm. The number of mesopores increased after doping of Co. The results also confirmed the appearance of the porous structure in the SEM patterns after the doping of Co, which indicated that the Co was beneficial for the formation of a porous structure. It has been reported that a hierarchical porosity is favorable for mass transfer of ORR-related species and the formation of abundant active sites for ORR [42].

The ORR activities of the Co/N/CNT@PC-800, Co/PC-800, N/C-800 and C-800 catalysts were investigated by rotating disk electrode measurements. Fig. S4(a) and (b) showed the cyclic voltammogram tests of the Co/N/CNT@PC-800 and Pt/C catalysts in a O₂-saturated and N₂-saturated solution. No cathodic peaks could be observed in the N₂-saturated solution for the two samples. In contrast, when the electrolyte solution was saturated with O₂, cathodic peaks appeared for those catalysts in the CV potential range, demonstrating the ORR activity of the catalysts, which was attributed to the reduction of O₂. As presented in Fig. 5(a), Co/N/CNT@PC-800 showed a better oxygen reduction performance with a more positive onset potential, half-wave potential (E_1/2), and higher limited current density than the other three prepared materials (Table S2). Compared with the C-800 curve, the N/C-800 showed a more positive initial potential and the Co/PC-800 displayed a more positive initial potential and higher limited current density. This indicated that the N-or Co-doped carbon materials can boost the ORR properties, but the activity promotion was limited. The Co/N/CNT@PC-800 possessed optimized ORR activity than that of Co or N single-doped materials, which was owing to the synergetic interaction between both N and Co co-doped in the carbon materials. The Co/N/CNT@PC-800 catalyst showed a superior ORR performance than most catalysts derived from biomass under similar conditions (Table S2). The (K-L) plots of the Co/N/CNT@PC-800 and Pt/C catalysts were obtained from the LSV curves at different rotation speeds; both showed a good linearity (Fig. S5(c) and (d)). The electron transfer number of the Co/N/CNT@PC-800 catalyst was calculated to be approximately 3.4–3.5 at potentials ranging from -0.4 to -0.8 V (Fig. 5(b)). The result suggested that the Co/N/CNT@PC-800 exhibited a quasi-four-electron pathway in the ORR process. All the results indicated that both the Co-and the N-functionalized carbon species were essential for the active site in M–N_x/C type catalysts [43].

The tolerance to methanol and stability are significant criteria to evaluate the performance of an ORR catalyst. As shown in Fig. 6(a) and (b), after the addition of methanol (1 mol L^-1), the CV curve of the Pt/C catalyst clearly changed and displayed a typical inverse methanol oxidation peak. However, there was only a slight effect on the CV curve of the Co/N/CNT@PC-800 catalyst. The results showed that the Co/N/CNT@PC-800 catalyst had a tolerance to methanol. Finally, the durability tests of the Co/N/CNT@PC-800 and Pt/C catalysts were recorded by performing 3000 cycles of cyclic voltammogram measurements in O₂-saturated 0.1 mol L^-1 KOH with a scan rate of 50 mV s^-1. The Co/N/CNT@PC-800 catalyst could maintain high activity with a limited current density loss of 0.27 mA cm^-2 and there was no significant change in the initial potential (Fig. 6(c)). Nevertheless, the limited current density of the Pt/C catalyst showed a decrease of approximately 0.80 mA cm^-2 at -0.8 V, and the initial potential became more negative (Fig. 6(d)). In addition, the E_1/2 of the Co/N/CNT@PC-800 catalyst showed a negative shift of 15 mV, which was much smaller than that of commercial Pt/C (ΔE_1/2 = 37 mV). As shown in the inset of Fig. 6(c), a well-defined reduction peak in the CV curves of Co/N/CNT@PC-800 was still evident after 3000 cycles. However, the peak potential of Pt/C shifted negatively from -0.138 to -0.168 V, almost 30 mV (inset of Fig. 6(d)). These results indicated that the Co/N/CNT@PC-800 demonstrated a more remarkable methanol crossover effect and long-term stability for the ORR.

To further study the ORR performance of Co/N/CNT@PC-800 in an alkaline environment, rotating ring-disk electrode (RRDE) measurements were performed to investigate the electron transfer number and H₂O₂ yield. Fig. 7 recorded the ring (Fig. 7(a)) and disk (Fig. 7(b)) currents of the Co/N/CNT@PC-800 and Pt/C catalysts. The limiting current of Co/N/CNT@PC-800 was -553 μA at -0.8 V, which was slightly different from that of Pt/C (-625 μA). The electron transfer number and the H₂O₂ yield was directly calculated from the disk currents and the ring currents by Eqs. (4) and (5), respectively. As shown in Fig. 7, the electron transfer number of the Co/N/CNT@PC-800 and Pt/C catalysts was approximately 3.5–3.7 and 3.8–3.9 (Fig. 7(c)), respectively. The H₂O₂ yields of the Co/N/CNT@PC-800 and Pt/C catalysts were 26.6% and 8.9%, respectively, at -0.8 V in 0.1 mol L^-1 KOH solution (Fig. 7(d)). All the electrochemical experiments suggested that the oxygens were directly reduced to hydroxyls through a quasi-four-electron reaction in the ORR process and the Co/N/CNT@PC-800 is a good choice as a cathode catalyst.

Based on above results, the Co/N/CNT@PC-800 catalyst showed an optimized ORR performance, which illustrated the mechanism controlling the ORR active sites in the electrocatalyst. First, doped N could improve the electron donor-acceptor properties, enhance the adsorption of dissociative oxygen, and weaken the O–O bonding [44-46]. Second, doped metal atoms can form active sites and intimate bonds (M–N_x) more efficiently and boost the ORR activity of the material [23]. Third, the Co nanoparticles also activated the protective carbon layers, which allowed the carbon layer of the outer surface to participate in the ORR process [28]. Finally, the Co nanoparticles catalyzed the dicyandiamide into CNTs on the surface of the porous carbon, affording more exposed active sites and facilitating the transportation of ORR-relevant species.

We successfully employed an uncomplicated process that effectively combines the advantages of porous carbon and biomass to obtain catalysts for the ORR. Cheap waste paper was used as carbon source, which was doped with Co and N to prepare a Co/N/CNT@PC-800 electrocatalyst. The Co and N co-doped carbon material greatly improved the ORR performance. The Co catalyzed the in situ formation of carbon nanotubes, which promoted the exposure of active sites. The porous carbon substrate was favorable for mass transfer. The Co/N/CNT@PC-800 catalyst exhibited excellent ORR activity, a positive half-wave potential, and a quasi-four-electron reaction pathway. The tolerance of methanol and stability of Co/N/CNT@PC-800 was also better than that of commercial Pt/C. This work will provide a new idea for the use of biomass for the synthesis of low-cost electrocatalysts with effective performance for metal-air batteries and fuel cells.