Saccharina japonica is cleaned with water, and then baked to dryness. After being dried and powdered into a fine powder (5 g), 10 g ZnCl₂ and 500 mL water are combined, and the above mixture are constantly agitated for 48 h. Above mixture is baked in an oven (80°C) until a fully dry colloid form. This colloid undergoes pyrolysis into a quartz at 850°C for 2 h. Above carbon material is combined with NaBH₄ at a 1:1 mass ratio. The heated material is washed repeatedly with 2 M HCl solution and water until it reaches a neutral pH. The final product is labeled S-850-1. The sample, prepared using the same method as described above without the addition of ZnCl₂, is designated as SN-850-1. The pyrolysis temperature of the samples is increased from 850°C to 1,000°C, with holding times of 2 and 3 h, respectively, denoted as S-1000-2 and S-1000-3. S-1000-2 is mixed with NaBH₄ in a 1:1 ratio, with the pyrolysis temperature set at 850°C and held for 2 h, resulting in the sample being named S-850-B.

The population analysis, electron density difference, and adsorption free energy of O*, OH*, and OOH* (descriptors) are calculated using VASP software. The exchange-correlation functional is set to GGA-PBE. The energy cutoff for the plane-wave basis set is 500 eV, and the projector augmented wave method is used for pseudopotentials. A solvation model is applied using implicit solvation, and the dielectric constant for water (H₂O) is set to 78.5. The slab model is based on a graphene structure with a (4 × 4 × 3) supercell, with the bottom two layers of the slab fixed during relaxation. A vacuum layer of 20.0 Å is included, and the cleave plane is oriented along the (002) direction.

D/MAX-Ultima + diffractometer with Co. Kα radiation is used for XRD. Using a WBL-810 apparatus, N₂ adsorption-desorption tests are carried out at 77 K. The microstructural morphology of the produced samples is examined with a Supra-55 sapphire fitted FE-SEM.

The electrochemical studies are done using a Corrtest CS310 M electrochemical station. Following a combination of 10 mg sample, 5 mL isopropanol, and 20 µL 5% Nafion, the mixture underwent a 20 min ultrasonography treatment. Experiments using LSV (linear sweep voltammetry) and CV (cyclic voltammetry) are conducted at scan speeds of 10 and 20 mV s⁻¹, respectively. At rotating speeds from 400 to 2,500 rpm, LSV curves are recorded in alkaline electrolyte.

To get the transferred electron number (N), the Koutechy-Levich (K-L) curve’s slope is calculated using Equations 1, 2. 1 J = 1 J L + 1 J K = 1 B ω 1 2 + 1 J K (1) B = 0.2 n F C 0 D 2 3 ν − 1 6 (2)

The current density measured during the electrochemical experiments is denoted as J, while J _k represents the kinetic current density and J _L corresponds to the limited current density. n is the representation of the transmitted electron number. B represents the Levich slope, and the electrode rotating rate is indicated by ω. The Faraday constant, denoted as F. The diffusion coefficient of O₂ is represented as D, and the bulk concentration of O₂ is denoted as C. v is the kinetic viscosity representation.

This study uses DFT calculations to confirm that the ORR activity of the N, B dual-doping carbon model is higher than that of the N or B mono-doping carbon models. Saccharina japonica and NaBH₄ are employed as sources of N and B, respectively. High-temperature pyrolysis and washing with 2 M HCl solution are then applied to produce carbon materials doped with N (4.54 at%) and B (1.05 at%) as shown in Scheme 1.

Reports commonly identify graphitic and pyridinic nitrogen as the most effective active sites for ORR (Guo et al., 2020; Liang and Yuan, 2024). Figures 1, 2 depict reaction energy barriers and the charge distribution for models of graphene, pyridinic-N, graphitic-N, and B-doping carbon. The O*, OH*, and OOH* species exhibit favorable adsorption and desorption energies, as shown by their adsorption at the reaction sites marked by red circles in these models. The graphene model demonstrates a homogeneous charge distribution, as seen in Figures 1A, 2A, leading to a substantial energy barrier for the reaction (1.21 eV in step 2) and limited adsorption capacity. As shown in Figure 1B, graphitic-N (N₁₃) is bonded to three carbon atoms (C₁₂, C₁₄, C₂₃), forming three C-N bonds. The electronegativity of N atom (3.04) is higher than that of C atom (2.55), causing the N atom to attract the valence electrons of the C atoms. As a result, N atom accumulates electrons, gaining a charge of −0.64 e, while the three surrounding C atoms (C₁₂, C₁₄, C₂₃) lose valence electrons, each acquiring a charge of +0.21 e. The reaction energy barrier of the graphitic-N doped carbon model is reduced to 0.70 eV in step 2 (Figure 2B), attributed to the uneven distribution of the electron cloud.

In Figure 1C, the pyridinic-N atom forms C-N bonds with two carbon atoms (C₁₂ and C₁₄). Due to the presence of a lone pair of electrons, pyridinic-N exhibits a weaker electron-attracting ability compared to graphitic-N. Consequently, C₁₂ and C₁₄ carry a charge of +0.17 e, while C₂₃ exhibits a charge of +0.06 e. Due to the uneven electron cloud distribution, the reaction energy barrier of pyridinic-N doped carbon is reduced to 0.80 eV (step 1) in Figure 2C.

Electronegativity of B atom (2.04) is lower than that of C atom, causing its electrons to be attracted by the surrounding C atoms (C₁₃, C₂₂, and C₂₄). B atom carries a charge of +0.184 e, while the neighboring C atoms (C₁₃, C₂₂, and C₂₄) carry a charge of −0.059 e in Figure 1D. The reaction energy barrier of B-doped carbon is further reduced to 0.60 eV (step 2) in Figure 2D. In Figure 1E, the dual doped graphitic-N and B atoms further disrupts the electron cloud distribution. The electron density of graphitic-N atom increases to −0.65 e, while the electron density of B atom decreases to +0.15 e. The surrounding carbon atoms, C₁₂ and C₁₄, carry a charge of +0.20 e, C₂₃ carries +0.18 e, while C₂₅ and C₃₄ carry −0.04 e. These changes in electron distribution are attributed to the coupling effect between the graphitic-N and B atoms. In Figure 2E, the reaction energy barrier of graphitic-N and B doped carbon is further reduced to 0.53 eV in step 2. Similarly, pyridinic-N and B dual doping also further disrupts the electron distribution in the carbon structure. The electron density of pyridinic-N atom increases to −0.64 e, while the electron density of B atom decreases to +0.15 e. The surrounding carbon atoms, C₁₂ and C₁₄, carry a charge of +0.17 e, C₂₃ carries +0.03 e, while C₂₅ and C₃₄ carry −0.04 e. In Figure 2F, the reaction energy barrier of graphitic-N and B doped carbon is further reduced to 0.51 eV in step 2. According to the DFT calculation results, the graphitic-N, pyridinic-N and B dual-doped carbon structure further disrupts the electron cloud distribution, adjusts the adsorption energies of intermediate species O*, OH*, and OOH*, thereby lowering the energy barrier for the ORR, which is beneficial for enhancing ORR catalytic performance.

DFT findings suggest that N, B dual-doping carbon may exhibit exceptional ORR performance. N, B dual-doping carbon ORR catalysts are synthesized from saccharina japonica.

SEM pictures provide visual information on the shape and structure of the as-synthesized samples. In Figure 3A, S-850-1 exhibits a porous surface morphology, with interconnected carbon particles forming the surface porosity. The porous morphology enhances the exposure of active sites and improves O₂ molecule transport. As shown in Figure 3B, SN-850-1 demonstrates a large, solid bulk structure at the micrometer scale, with a non-porous surface. SEM analysis indicates that the porous structure of the sample is attributed to the influence of ZnCl₂ activator. N₂ adsorption-desorption study assesses the surface area and distribution of pore sizes. In Figure 3C, mesopores are evident, with both S-850 and S-850-1 exhibiting type-IV isotherms and a distinct hysteresis loop in the medium and high-pressure regions (Wang et al., 2011). However, SN-850-1 exhibits a type II isotherm, indicating a non-porous structure (Li et al., 2019). After doping with B, S-850-1 shows a modest increase in specific surface area from 1,108 to 1,173 m² g⁻¹, which is likely attributable to the B atomic incorporation. SN-850-1 exhibits a comparatively low specific surface area of 89 m² g⁻¹. Non-local density functional theory, the distribution of pore size shown in Figure 3D is predominantly in the 15–50 nm range, indicating the presence of mesopores. SN-850-1 is characterized by a non-porous structure. Based on the above physical characteristics, ZnCl₂ activator is identified as a key factor in the development of the porous structure of biomass-derived carbon-based ORR catalyst, potentially impacting ORR performance.

XPS measurements can be used to analyze the types and quantities of doping N and B atoms. Figure 4A reveals the presence of doped N atoms through peaks at approximately 187 eV (B), 284 eV (C), 532 eV (O), and 400 eV (N) for the as-synthesized samples (Lee et al., 2016). S-850 contains only doped C, O, and N atoms, whereas S-850-1 exhibits significant B doping, confirming the incorporation of B atoms into the N-doped carbon matrix. S-850-1 shows a higher O content than N and B. The pyrolysis temperature of the samples is increased to 1,000°C and maintained for 2 h to ensure the complete volatilization of doped N atom. S-1000-2 is mixed with NaBH₄ in a 1:1 ratio, with the pyrolysis temperature set at 850°C and held for 2 h, resulting in the sample being named S-850-B.

Pyrolysis at 1,000°C, with hold times of 2 and 3 h (noted as S-1000-2 and S-1000-3), is conducted to fully volatilize N and B, aiming to investigate the effect of doped O atom on ORR performance. In the S-1000-2 and S-1000-3 samples, N and B are indeed absent, with only O remaining, which is attributed to the volatilization of N and B at high temperatures. The O contents of S-1000-2 and S-1000-3 are 5.34 at% and 3.21 at%, respectively. Subsequent electrochemical performance tests will be conducted to analyze the effect of doped O atom on ORR performance. N (1.45 at%) and B (0.23 at%) contents in SN-850-1 are lower than those in S-850-1 (N: 4.54 at% and B: 1.05 at%), indicating that the porous structure facilitates the exposure of more heteroatoms. Figure 4B shows doping N and B atomic content of the as-synthesized samples. The high-resolution N_1s spectra of S-850-1 reveal peaks corresponding to graphitic-N (401 eV), pyridinic-N (398 eV), pyrrolic-N (399 eV), and oxidized-N (402 eV) (Zhang et al., 2023b; Huang et al., 2019). The doped B atoms are represented by peaks at 187 eV (B₄C) and 189 eV (BC₃) in Figures 4C, D. According to the XPS results, saccharina japonica may serve as a precursor for the synthesis of N mono-doping carbon materials. Additionally, combining saccharina japonica with NaBH₄ enables the effective production of N and B dual-doping carbon, which may show promising activity for ORR.

XRD patterns of S-850-1 and S-850 exhibit two distinct peaks at 2θ = 30.7° and 49.4°, which match to the (002) and (100) crystallographic planes of a standard hexagonal carbon structure (JCPDS card no. 41–1,487) in Figure 4E (Yu et al., 2014). The diffraction peaks of the (002) plane in S-850 and S-850-1 show a slight negative shift relative to the standard peak, indicating that the diffraction is performed by ions (N, and B) of different radius or the defects formed due to incorporation of N or B. Both S-850 and S-850-1 display D and G bands in Raman spectra (Figure 4F). The extent of disorder in the carbon material is reflected in the I_D/I_G ratio of these bands (Kim et al., 2011). For S-850-1, the I_D/I_G ratio is 1.23, showing a slight increase compared to the N mono-doped S-850, which has an I_D/I_G ratio of 1.22. This increases likely results from the enhanced defect density due to B doping.

This work assessed the ORR performance of the as-synthesized samples and 20% Pt/C using CV and LSV tests in 0.1 M KOH electrolyte, with recorded potentials plotted against the RHE scale. Figure 5A shows clear reduction peaks for the above samples, demonstrating their ORR activity in an alkaline electrolyte. Compared to S-850 (0.881 V) and S-850-B (0.805 V), S-850-1 exhibits a higher peak potential (0.895 V), closely matching that of 20% Pt/C (0.899 V). The results show that the N, B dual-doped carbon catalyst has significantly higher ORR performance than the N or B mono doped carbon catalyst, even approaching that of 20% Pt/C. The peak potentials of the N and B mono doped carbon catalysts (S-850 and S-850-B) are higher than those of the N, B dual-doped SN-850-1 (0.795 V) with low surface area and non-porous structure, indicating a direct relationship between surface area, porosity, and ORR performance.

Among the carbon catalysts with different doped O contents, S-1000-2 (O: 5.34 at%) and S-1000-3 (O: 3.21 at%) without N and B doping exhibit similar peak potentials of 0.682 V, which are lower than those of the N- and B-doped carbon catalysts, indicating that O doping has a minimal effect on ORR performance. Furthermore, the peak potential of the most effective S-850-1 exceeds that of previously reported N and B dual-doped carbon catalysts (BN-GNRs: 0.870 V (Tu et al., 2022), B-NHMC: 0.881 V (Al-Enizi, 2022), BNC-600: 0.759 V (Zhou et al., 2021)), which can be attributed to its lower N and B content, as well as its smaller surface area and pore structure. The LSV curves, shown in Figure 5B, demonstrate that the performance of S-850-1 is like that of 20% Pt/C. Specifically, S-850-1 shows improvements of 23 mV and 0.36 mA cm⁻² in half-wave potential and limited current density (0.861 V and −5.60 mA cm⁻²), respectively, compared to S-850 (0.838 V and −5.24 mA cm⁻²). The half-wave potential and limited current density of S-850-1 are nearly identical to those of 20% Pt/C. While S-850-1 starts with an initial potential of 0.962 V, it shifts negatively by 83 mV compared to the 1.050 V of 20% Pt/C. The trends observed in the onset potential, half-wave potential, and limited diffusion current density of the other samples are consistent with the above CV results. The detailed electrochemical data are listed in Table 1.

Based on the preceding physical and electrochemical test results, the N, B dual-doped carbon structure demonstrates a notable enhancement in ORR performance, surpassing that of the mono N and B doped carbon structures. Furthermore, the ORR activity of the N, B dual-doped carbon catalyst S-850-1 exceeds that of previously reported N, B dual-doped materials in the literature, suggesting that S-850-1 holds substantial potential for commercialization as an ORR catalyst. The LSV curves (rotation rates: 400–2,500 rpm), shown in Figure 5C, demonstrate that higher rotation rates increase the limiting diffusion current density, reduce diffusion distance, and enhance oxygen transport (Tan et al., 2012). K-L plots within the 0.3–0.7 V potential range reveal strong linearity for S-850-1, suggesting first-order reaction kinetics for oxygen reduction (Xu et al., 2017). Moreover, S-850-1 has an electron transfer number (n) of 3.85, indicating a predominant 4-electron ORR way.

The longevity of catalysts is a key challenge for fuel cell commercialization (Wu et al., 2011). After 5,000 CV cycles, S-850-1 shows a negative shift of 26 mV in half-wave potential (@1,600 rpm) and a decrease of −0.52 mA cm⁻² in limiting diffusion current density from its initial values. In comparison, commercial 20% Pt/C experiences a more significant decline, with shifts of 161 mV and −1.64 mA cm⁻², respectively (Figure 6A). The outstanding stability of S-850-1 may be ascribed to the robust covalent C-N bonds and the lack of complications such as Pt nanoparticle aggregation and dissolution (Zhong et al., 2013). Additionally, under chronoamperometric testing at 0.5 V for 18,000 s, the current density of 20% Pt/C drops to 0.8 mA cm⁻², while S-850-1 maintains a current density of approximately 1.1 mA cm⁻² (Figure 6B). CO gas poses a significant challenge for fuel cells by increasing the risk of ORR catalyst poisoning. Figure 6C illustrates the catalysts’ tolerance to CO. Upon CO introduction at 4,000 s, 20% Pt/C’s current density decreases to about 56%, indicating a higher susceptibility to CO poisoning compared to S-850-1, which retains about 79% of its current density. In direct methanol fuel cells, methanol may migrate from the anode to the cathode, where it can potentially poison the ORR catalyst (Gao et al., 2015c). Therefore, methanol tolerance is crucial for optimal catalyst performance. As shown in Figure 6D, S-850-1 demonstrates high selectivity for oxygen and shows no significant activity loss upon the addition of 10 mL methanol, indicating effective resistance to methanol crossover. In contrast, the ORR performance of 20% Pt/C is adversely affected by methanol, as it is more prone to methanol oxidation reactions (MOR). These results highlight that S-850-1 surpasses 20% Pt/C in terms of stability and resistance to methanol and CO poisoning.

This study constructs a Zn-air battery to further investigate the performance of S-850-1. The Zn-air battery features a 4.5 cm² operating surface, with a 6 M KOH solution circulating at 20 mL per minute as the electrolyte. The oxide layer is removed by sanding the zinc plate (length: about 8.5 cm, width: about 3 cm), as shown in Figure 7A. S-850-1 coated on stainless steel mesh (approximate length: 7.5 cm, approximate width: 3 cm). The Zn-air battery’s open circuit voltage and current density in Figure 7B, c are 1.40 V and 0.72 A, respectively, which illuminates the LED light. The findings suggest that S-850-1 may be used as an ORR catalyst for Zn-air batteries.

Future research directions will focus on exploring novel biomass that is inherently rich in various heteroatoms, enabling the formation of self-doped carbon-based ORR catalysts with high heteroatom content. This approach will avoid the use of external doping chemicals and reduce preparation costs. In the process of scaling up the industrial production of biomass-derived carbon-based ORR catalysts, challenges may arise due to variations in the heteroatom content caused by differences in the origin and seasonal changes of biomass materials across different batches. Additionally, it is crucial to develop scalable and cost-effective processes, such as ball milling, to produce heteroatom-doped carbon-based electrocatalysts. These processes should ensure the catalysts possess optimal structural and surface chemical properties to meet the demands of industrial applications (Sekhon et al., 2022; Koolen et al., 2023).