Optimization Strategies of Preparation of Biomass-Derived Carbon Electrocatalyst for Boosting Oxygen Reduction Reaction: A Minireview

Oxygen reduction reaction (ORR) has attracted considerable attention for clean energy conversion technologies to reduce traditional fossil fuel consumption and greenhouse gas emissions. Although platinum (Pt) metal is currently used as an electrocatalyst to accelerate sluggish ORR kinetics, the scarce resource and high cost still restrict its further scale-up applications. In this regard, biomass-derived carbon electrocatalysts have been widely adopted for ORR electrocatalysis in recent years owing to their tunable physical/chemical properties and cost-effective precursors. In this minireview, recent advances of the optimization strategies in biomass-derived carbon electrocatalysts towards ORR have been summarized, mainly focusing on the optimization of pore structure and active site. Besides, some current challenges and future perspectives of biomass-derived carbon as high-performance electrocatalysts for ORR have been also discussed in detail. Hopefully, this minireview will afford a guideline for better design of biomass-derived carbon electrocatalysts for ORR-related applications.


Introduction
The electrocatalytic activity of oxygen reduction reaction (ORR) significantly determines the performance of current energy conversion and storage devices, including various fuel cells and metal-air batteries [1][2][3]. Due to the sluggish kinetics of the ORR process, electrocatalysts have usually been required to accelerate the reaction rates and decrease the overpotentials [4]. Currently, platinum (Pt)-group metal (PGM) based materials have been broadly utilized and regarded as the most effective electrocatalysts for ORR catalysis [5,6]. However, the disadvantages of scarce resource, high-cost and poor durability greatly limit their further scale-up applications [7]. In this regard, a great number of efforts have concentrated on the development of cost-effective and high-performance candidates to replace the state-of-the-art PGM-based electrocatalysts [8][9][10].

Performance Evaluation of Biomass-Derived Carbon Electrocatalysts for ORR
To reliably evaluate the ORR performance of biomass-derived carbon electrocatalysts [37], constructing a general and standard procedure is highly necessary. So far, the widely adopted method to assess the intrinsic activity of catalysts is based on the rotating disk electrode (RDE) and rotating ring disk electrode (RRDE) measurement [48]. Such method is capable of avoiding the mass transfer concerns and thus affording a stable ORR kinetic current density (jk), because the working electrode rotates at a high speed during the evaluation process ( Figure 2a) [44]. Typically, ORR measurement is carried out in a three-electrode system, in which the electrocatalyst is the working electrode, an Ag/AgCl or Hg/HgO electrode is the reference electrode, and a Pt wire is the counter electrode [49]. The electrocatalysts are usually dispersed in solvent with Nafion as binders to form a uniform slurry. Then the slurry is drop-casted on glassy carbon (GC)-based RDE or RRDE [50]. Of note, the quality of drop-casting film and loading amount of electrocatalyst directly affects the ORR kinetics. Therefore, uniform catalyst films and appropriate catalyst mass loadings on the electrode are essential to obtain an accurate ORR performance, including onset potential (Eonset), half-wave potential (E1/2), limiting current density (jL) and Tafel slope parameters [51]. A positive shift of Eonset or E1/2, large jL, and small Tafel slope indicate high ORR activity and fast ORR kinetics, respectively [52].
Generally, two ORR pathways have been proposed: one is the direct four-electron (4e − ) reduction pathway, the other is the indirect two-electron (2e − ) peroxide pathway (Figure 2b) [44]. O2 can be directly reduced to H2O through the 4e − pathway with a rate constant k1; alternatively, it can also be reduced to adsorbed hydrogen peroxide (H2O2,ad) through the 2e − pathway with a rate constant k2 (HO2 − in alkaline medium and H2O2 in acidic medium). The amount of H2O2 was calculated according to Equation (1). Subsequently, H2O2,ad could either be electrochemically reduced to H2O with a rate constant k3, or could be chemically decomposed to O2,ad with a rate constant k4 on surface, and desorbed into electrolyte with a rate constant k5. The suitable electrocatalyst drives the ORR process towards direct 4e − reduction pathway, which requires high

Performance Evaluation of Biomass-Derived Carbon Electrocatalysts for ORR
To reliably evaluate the ORR performance of biomass-derived carbon electrocatalysts [37], constructing a general and standard procedure is highly necessary. So far, the widely adopted method to assess the intrinsic activity of catalysts is based on the rotating disk electrode (RDE) and rotating ring disk electrode (RRDE) measurement [48]. Such method is capable of avoiding the mass transfer concerns and thus affording a stable ORR kinetic current density (j k ), because the working electrode rotates at a high speed during the evaluation process ( Figure 2a) [44]. Typically, ORR measurement is carried out in a three-electrode system, in which the electrocatalyst is the working electrode, an Ag/AgCl or Hg/HgO electrode is the reference electrode, and a Pt wire is the counter electrode [49]. The electrocatalysts are usually dispersed in solvent with Nafion as binders to form a uniform slurry. Then the slurry is drop-casted on glassy carbon (GC)-based RDE or RRDE [50]. Of note, the quality of drop-casting film and loading amount of electrocatalyst directly affects the ORR kinetics. Therefore, uniform catalyst films and appropriate catalyst mass loadings on the electrode are essential to obtain an accurate ORR performance, including onset potential (E onset ), half-wave potential (E 1/2 ), limiting current density (j L ) and Tafel slope parameters [51]. A positive shift of E onset or E 1/2 , large j L , and small Tafel slope indicate high ORR activity and fast ORR kinetics, respectively [52].
Generally, two ORR pathways have been proposed: one is the direct four-electron (4e − ) reduction pathway, the other is the indirect two-electron (2e − ) peroxide pathway (Figure 2b) [53]. The selectivity of the electrocatalyst is usually described with the number (n) of transferred electrons, which can be calculated by K-L equation and RRDE measurement [54]. The number (n) was calculated according to Equation (2).
where I D is the disk current, I R is the ring current, and N is the collection efficiency (≈0.24-0.5).
current efficiency and fewer peroxide species [53]. The selectivity of the electrocatalyst is usually described with the number (n) of transferred electrons, which can be calculated by K-L equation and RRDE measurement [54]. The number (n) was calculated according to Equation (2).
where ID is the disk current, IR is the ring current, and N is the collection efficiency (≈0.24-0.5).

Figure 2. (a)
The rotating ring disk electrode (RRDE) configuration: glassy carbon (GC)-disk is used as a substrate to load electrocatalysts for ORR, and Pt-ring is used to detect the produced peroxide species from electrocatalyst layer. (b) General ORR mechanism via direct four-electron (4e − ) reduction and indirect two-electron (2e − ) peroxide pathways. Reproduced with permission [44]. Copyright 2020, Wiley-VCH.

Optimization Strategies of Pore Structure
Since ORR only occurs at the solid-gas-liquid (electrocatalyst-O2-electrolyte) triple-phase boundary regions, interconnected hierarchical pores in biomass-derived carbon electrocatalysts are beneficial to the penetration of electrolyte and mass transfer, resulting in excellent ORR activity [55,56]. Appropriate porosity and microarchitecture have the ability to increase the amounts of triple-phase boundary region, thus maximizing the utilization of active sites. As a convenient method to increase the exposed active surface, activation has been adopted for carbon electrocatalysts to adjust the meso-/micropore proportion [57]. According to the mechanism, activation processes can be classified to physical activation, chemical activation, template activation and self-activation [58]. In this section, we have summarized such commonly used optimization strategies of pore structure, which is important for ORR performance.

Physical Activation
Physical activation strategy mainly involves N2, oxidizing atmosphere and H2O steam activation, which contributes to the formation of a porous structure owing to the removal of carbon atoms at high temperatures [38,59]. The development of porosity in the physical activation process includes the following three mechanisms: (i) the generation of new pores at the inaccessible early stage; (ii) the formation of new pores with the removal of ordered carbons; and iii) the increase of pore size [45,60]  The rotating ring disk electrode (RRDE) configuration: glassy carbon (GC)-disk is used as a substrate to load electrocatalysts for ORR, and Pt-ring is used to detect the produced peroxide species from electrocatalyst layer. (b) General ORR mechanism via direct four-electron (4e − ) reduction and indirect two-electron (2e − ) peroxide pathways. Reproduced with permission [44]. Copyright 2020, Wiley-VCH.

Optimization Strategies of Pore Structure
Since ORR only occurs at the solid-gas-liquid (electrocatalyst-O 2 -electrolyte) triple-phase boundary regions, interconnected hierarchical pores in biomass-derived carbon electrocatalysts are beneficial to the penetration of electrolyte and mass transfer, resulting in excellent ORR activity [55,56]. Appropriate porosity and microarchitecture have the ability to increase the amounts of triple-phase boundary region, thus maximizing the utilization of active sites. As a convenient method to increase the exposed active surface, activation has been adopted for carbon electrocatalysts to adjust the meso-/micropore proportion [57]. According to the mechanism, activation processes can be classified to physical activation, chemical activation, template activation and self-activation [58]. In this section, we have summarized such commonly used optimization strategies of pore structure, which is important for ORR performance.

Physical Activation
Physical activation strategy mainly involves N 2 , oxidizing atmosphere and H 2 O steam activation, which contributes to the formation of a porous structure owing to the removal of carbon atoms at high temperatures [38,59]. The development of porosity in the physical activation process includes the following three mechanisms: (i) the generation of new pores at the inaccessible early stage; (ii) the formation of new pores with the removal of ordered carbons; and iii) the increase of pore size [45,60] graphene are significantly enhanced due to the beneficial role of the simultaneously doped N and S, and the effect of 3D porous morphology with large surface area (319.93 m 2 g −1 ) to expose more active sites after activation. When used as a high-performance electrocatalyst for ORR, the E 1/2 and j L for NSG is even more positive than commercial Pt/C, demonstrating a remarkable electrocatalytic activity of NSG.
In an activated biomass-derived carbon, pore structure is dependent on several parameters including carbonization temperatures and times, activating agents, post-treatments, etc. [62]. H 2 O steam and CO 2 are two commonly used activation agents, which can react with carbon to produce H 2 and CO gases, respectively [63,64]. To achieve a high specific surface area of biomass-derived carbon electrocatalyst, the reaction between activation agents and carbon is required to occur inside of carbon atoms [65]. Therefore, choosing appropriate synthesis conditions is necessary for the removal of carbon and development of porosity. Although physical activation is an effective strategy to prepare biomass-derived carbon with high yield and bulk density, it is still unable to regulate the electronic structure and surface chemistry properties of carbon, which hinders further application to some extent [66].

Chemical Activation
Compared with the physical activation method, chemical activation is a one-step strategy that involves the impregnation of biomass precursors with activation agents prior to pyrolysis at the desired temperature and inert atmosphere [67]. In this specific method, biomass precursors are usually treated with dehydrating chemicals such as ZnCl 2 , ZnO, H 3 PO 4 , KOH and alkaline carbonates, then are treated by carbonization and a further washing step to remove the residual activation agents and by-products [45]. Although such a strategy has been adopted for several years, the activation mechanisms are still elusive [58]. Taking KOH activation as an example, three possible mechanisms have been proposed: (i) the abundant micro/mesopores are generated via redox reactions between K-containing species (KOH, K 2 CO 3 , K 2 O) and the carbon matrix at high temperature [68]; (ii) the water vapor at high temperatures can increase porosity due to its gasification [69]; and (iii) the metallic K would intercalate into the carbon skeleton to expand the interplanar spacing, thus resulting in a highly porous structure [70].
Except for KOH activation, ZnCl 2 activation is another typical strategy used for chemical activation [71]. Owing to the Lewis acid feature, the ZnCl 2 agent has the capability to facilitate the dehydration and aromatic condensation reaction of biomass precursors at low temperatures and high temperatures, respectively [72]. Recently, Gao and coworkers developed a general ZnCl 2 activation strategy to successfully prepare a N,O codoped carbon electrocatalyst with hierarchical porosity [73]. The prepared N 0.54 -Z 3 /M 1 -900 sample exhibits a high ORR activity in terms of E onset of 0.94 V vs. reversible hydrogen electrode (RHE) and E 1/2 of 0.824 V. In chronoamperometric responses test, 97.4% of the original current density is retained for the N 0.54 -Z 3 /M 1 -900 electrode after 24 h, whereas the commercial Pt/C catalyst electrode shows much higher current loss of 15% after only 8 h. These results suggest that the electrocatalytic activity strongly depends on the hierarchically pore architectures and nitrogen active sites in the carbon matrix. As shown in Figure 3b, such a strategy can be easily applied to synthesize other biomass-derived carbon materials owing to a versatile dehydration-aromatization-condensation carbonization mechanism. Intriguingly, chemical activation is capable of significantly increasing the specific surface area and pore volume of biomass-derived carbon due to the high degree of carbon etching.

Template Activation
Template activation strategies, including hard template and soft template, have been widely employed to modulate the pore structures of biomass-derived carbon electrocatalysts for ORR [58]. The morphologies of hard templates greatly determine the obtained architectures of the final electrocatalysts, such as zero-dimensional (0D) ordered mesoporous carbon sphere, two-dimensional (2D) carbon nanosheet, and 3D hierarchical carbon monolith, etc. [76][77][78]. Briefly, introducing hard templates in biomass precursors can generate a large number of mesopores in the resulting carbon skeleton. N-doped porous carbon (NPC) derived from alfalfa, for example, has been synthesized through a facile copyrolysis method using a CaCO3 hard template (Figure 3c). In their preparation, CO2 gases can be in situ generated from a CaCO3 template at high temperature, thus forming abundant pore structures. Unfortunately, the use of hard templates is time-consuming and expensive, which is the largest challenge for large-scale application. As comparison, soft templating strategies including block copolymers, surfactants and micelles have also been employed to

Template Activation
Template activation strategies, including hard template and soft template, have been widely employed to modulate the pore structures of biomass-derived carbon electrocatalysts for ORR [58]. The morphologies of hard templates greatly determine the obtained architectures of the final electrocatalysts, such as zero-dimensional (0D) ordered mesoporous carbon sphere, two-dimensional (2D) carbon nanosheet, and 3D hierarchical carbon monolith, etc. [76][77][78]. Briefly, introducing hard templates in biomass precursors can generate a large number of mesopores in the resulting carbon skeleton. N-doped porous carbon (NPC) derived from alfalfa, for example, has been synthesized through a facile copyrolysis method using a CaCO 3 hard template (Figure 3c). In their preparation, CO 2 gases can be in situ generated from a CaCO 3 template at high temperature, thus forming abundant pore structures. Unfortunately, the use of hard templates is time-consuming and expensive, which is the largest challenge for large-scale application. As comparison, soft templating strategies including block copolymers, surfactants and micelles have also been employed to manufacture biomass-derived carbon electrocatalysts [79]. The soft-template activation avoids the disadvantage of removing template step, thus simplifying the production procedure of porous carbon.

Self-Activation
Unlike other activation methods mentioned above, the self-activation method has the unique advantage of no additional activation reagents, which can further decrease the cost and simplify the process [80]. The self-activation method usually involves two possible mechanisms. One is a physical self-activation process, which utilizes the emitted gases from biomass precursors at high temperatures for the activation of carbon [81]. The other is a chemical self-activation process, which employs some inorganic compounds that already exist in biomass precursors to activate carbon in situ [82]. Kong and coworkers have prepared a N-doped meso-cellular graphene foam (SF-NMGF) by employing sludge flocs as carbon and nitrogen sources, where numerous microorganisms in biomass precursors can serve as templates to produce pore structure (Figure 3d) [75]. As shown in Figure 3e,f, numerous mesocellular pores are identified in graphene nanosheets with interleaved structure, which is beneficial to mass transfer in ORR. The resultant SF-NMGF had a large Brunauer-Emmett-Teller (BET) surface area of 370 m 2 g −1 , a pore volume of 0.69 cm 3 g −1 and pore diameter of 15 nm. It demonstrates an excellent catalytic activity with an E onset only negatively shifted by 50 mV compared to commercial Pt/C, a j L of 5 mA cm −2 , a four-electron transferred electrons process, superior operational stability and methanol-tolerance in comparison with Pt/C in alkaline electrolyte. The excellent catalytic performance could be a result of (1) the high percentage of graphitic-N in the SF-NMGF, and (2) the foam-like mesopores with a large surface area and the graphitic framework with uniform N-doping, which both offer a high probability of exposure of the active sites. In this particular strategy, in situ emitted gases from biomass during the carbonization process can effectively etch carbon frameworks. Therefore, compared with other activation methods, the self-activation approach is more fitting to large-scale production and industrial requirements.

Optimization Strategies of Active Site
It has been widely accepted that the ORR electrocatalytic activities of carbon catalysts are greatly dependent on their spin densities and atomic charge [25]. Up to now, some carbon materials with graphitic structure and abundant edge-plane sites have been confirmed to be active in ORR electrocatalysis [83][84][85]. However, the majority of pristine carbon electrocatalysts possess inferior performance in aqueous electrolytes due to their unsatisfactory active sites [86]. Typically, active sites in carbon-based materials (especially carbon nanomaterials) include intrinsic carbon defects (existed in the conjugated network without any dopants, such as edges, vacancies, holes or topological defects), extrinsic defects (mainly heteroatoms or single metal atoms doping) and combination sites between each other [87][88][89][90]. As we known, the intrinsic defect sites in architectures greatly affect physicochemical properties of carbon nanomaterials, and the introduced intrinsic carbon defect sites in carbon frameworks can directly serve as the potential active sites [90]. Since "defect sites" are a big topic, with regard to this minireview, the optimization strategy of active sites discussed in this paper only involves the aspect of heteroatoms doping. Recently, introducing heteroatoms to the molecular backbone of biomass-derived carbon materials has been reported as an effective optimization strategy for the enhancement of ORR activity [91,92]. Such heteroatoms are capable of breaking the electroneutrality of pristine carbon, creating additional charge sites for O 2 adsorption, and efficiently utilizing the π electrons of carbon skeleton. In this section, we have summarized some commonly used optimization strategies of active site, which is another important factor for the enhancement of ORR performance.

Nitrogen Doping
The atomic size of nitrogen (N) and carbon is similar, but the electronic configuration of them is much different. Consequently, N doping in pristine carbon molecules has the capability to alter the electronic structure and minimize the lattice mismatch simultaneously [93]. Normally, the N dopant in carbon nanostructure can be divided into four parts: pyridinic N, pyrrolic N, graphitic N and quaternary N [94]. Interestingly, some biomass precursors containing abundant N element can be directly used to fabricate N-doped carbon electrocatalysts [95][96][97]. For example, Wang and coworkers have developed a cost-effective SiO 2 template-assisted approach for the fabrication of shrimp-shell derived N-doped carbon electrocatalysts (Figure 4a) [98]. The prepared 3D N-doped porous carbon obtained at 800 (NPC-800) with a surface area of 360.2 m 2 g −1 exhibits excellent catalytic activity in alkaline medium with an E onset of −0.06 V vs. Ag/AgCl (Pt/C −0.03 V), an E 1/2 of −0.21 V (Pt/C −0.17 V) and a j L of 5.3 mA cm −2 (Pt/C 5.5 mA cm −2 ) at −0.4 V vs. Ag/AgCl. It also displays superior durability and high methanol tolerance in alkaline medium than the commercial Pt/C catalyst. It has been found that pyridinic N and quaternary N moieties with electron-withdrawing ability function as the active centers for ORR electrocatalysis. However, the activity origin of catalytic mechanism of N-doped carbon electrocatalysts for ORR is still unclear due to the difficulty of precisely synthesizing the form of N dopant in biomass-derived carbon materials.

Phosphorus Doping
Phosphorus (P) exhibits a larger atomic size and lower electronegativity (2.19) than pristine carbon (2.55), which gives rise to a stable combination between P and C atoms [99]. Therefore, P element has been broadly studied as a promising dopant to substitute sp 2 carbon atom and modulate electronic configuration. Yang and coworkers have developed P-doped carbon hollow spheres (P-CHS) through a hydrothermal method, in which biomass glucose, tetraphenylphosphonium bromide and anionic surfactant sodium dodecyl sulfate are selected as carbon source, P source and soft template, respectively (Figure 4b) [100]. As shown in Figure 4c-e, C, O and P elements are uniformly distributed in the skeleton of P-CHS, indicating P atoms are successfully doped into the carbon nanostructure. The results confirmed the charged sites P + and/or asymmetric spin density in carbon atoms induced by doping were the catalytic sites for the ORR [101,102]. The study reveals that P content is crucial for the improvement of the ORR activity of carbon owing to the large distortion and partial destruction of the sp 2 -carbon network [103] caused by the high content P doping since P has a much larger covalent radius (107 ± 3 pm) than C (73 ± 1 pm). Even so, the of P-CHS catalyst exhibits an inferior ORR activity, whose E onset and E 1/2 are about 78 mV and 52 mV lower than that of Pt/C. Despite a great number of efforts on P-doped biomass-derived carbon have been developed as efficient ORR catalysts, their performances are still underwhelming compared to N-doped carbon materials, probably because P sites limit the desorption of ORR intermediates and thus decrease the reaction kinetics [104].

Boron Doping
Owing to lower electronegativity than P and C elements, boron (B) doping into biomass-derived carbon can also break the electroneutrality of pristine carbon, thus creating a good microenvironment for ORR process [107]. Generally, B atoms doping into carbon structures are in the form of BC2O, BCO2, B4C and BC3 [108]. The B4C and B3C species change the valence band structure and improve the density of states close to the Fermi level of carbon-based materials, which could enhance the electronic conductivity of the carbon material [109]. Furthermore, the B4C and B3C species introduced in carbon lattice increase the number of hole-type charge carriers and the electron density of nearby active carbon sites, which is helpful to improve the conductivity and electrochemical activity of the B-doped carbons [110]. In the past decade, B-doped carbon materials prepared from biomass have been thoroughly investigated as electrocatalysts for ORR, which exhibits superior methanol and CO tolerances than commercial Pt/C [111]. However, the overall performance of B-doped carbon materials is still insufficient in comparison with N-doped carbon. Yang and coworkers have reported a self-assembly approach for the synthesis of B-doped ordered mesoporous carbons (B-OMCs) with 2D hexagonal mesostructured and tunable doping content (Figure 4f) [105]. In their preparation, phenolic resin and H3BO3 are used as carbon and boron sources, respectively. The obtained B-OMCs characterized with an Eonset of −0.14 V vs. Ag/AgCl, E1/2 of -0.12 V (negatively shifted compared to that of the commercial Pt/C), a jL of 5.42 mA cm −2 , 3.71-3.85 of transferred electrons, and Tafel slope of 78.5 mV dec −1 . It has been demonstrated that

Boron Doping
Owing to lower electronegativity than P and C elements, boron (B) doping into biomass-derived carbon can also break the electroneutrality of pristine carbon, thus creating a good microenvironment for ORR process [107]. Generally, B atoms doping into carbon structures are in the form of BC 2 O, BCO 2 , B 4 C and BC 3 [108]. The B 4 C and B 3 C species change the valence band structure and improve the density of states close to the Fermi level of carbon-based materials, which could enhance the electronic conductivity of the carbon material [109]. Furthermore, the B 4 C and B 3 C species introduced in carbon lattice increase the number of hole-type charge carriers and the electron density of nearby active carbon sites, which is helpful to improve the conductivity and electrochemical activity of the B-doped carbons [110]. In the past decade, B-doped carbon materials prepared from biomass have been thoroughly investigated as electrocatalysts for ORR, which exhibits superior methanol and CO tolerances than commercial Pt/C [111]. However, the overall performance of B-doped carbon materials is still insufficient in comparison with N-doped carbon. Yang and coworkers have reported a self-assembly approach for the synthesis of B-doped ordered mesoporous carbons (B-OMCs) with 2D hexagonal mesostructured and tunable doping content (Figure 4f) [105]. In their preparation, phenolic resin and H 3 BO 3 are used as carbon and boron sources, respectively. The obtained B-OMCs characterized with an E onset of −0.14 V vs. Ag/AgCl, E 1/2 of -0.12 V (negatively shifted compared to that of the commercial Pt/C), a j L of 5.42 mA cm −2 , 3.71-3.85 of transferred electrons, and Tafel slope of 78.5 mV dec −1 . It has been demonstrated that B-OMC catalyst exhibits an inferior ORR activity but superior ORR durability, compared with commercial Pt/C electrocatalyst. The enhanced catalytic performance of the B-OMCs could be ascribed to the synergistic effects of the ordered mesostructure, specific surface area (470-626 m 2 g −1 ) and moderate boron doping (1.17 wt %).

Multiple Doping
Compared with single-heteroatom doping, multiple heteroatoms codoping into biomass-derived carbon nanostructure can further enhance the ORR activity owing to the synergistic effect between two or more heteroatom dopants [112]. The multiple heteroatoms impart the formation of multiactive centers in biomass-derived carbon, which has been confirmed as a promising electrocatalyst. For example, Wang and coworkers have reported a N and P co-doped hierarchically porous carbon (N,P-HPC) catalyst, which is synthesized through the pyrolysis of phytic acid, dicyandiamide, and N-doped carbon derived from cattle bones (Figure 4g) [106]. As shown in Figure 4h, the N and P heteroatoms are evenly distributed on the N,P-HPC network, indicating a homogeneous distribution of active sites. After optimization, the obtained N,P-HPC electrocatalyst demonstrated a better ORR performance than the monoelement doped counterparts. Moreover, the introduction of another dopant can also modulate the electronic structure [113]. The resultant N, P-HPC exhibited excellent ORR catalytic activity with an E onset of 0.924 V (comparable to commercial Pt/C), an E 1/2 of 0.853 V (12 mV higher than that of Pt/C), a kinetic current density of 38.2 mA cm −2 at 0.8 V (1.9-times that of Pt/C), better electrochemical stability and methanol tolerance in comparison with Pt/C in alkaline electrolyte. The superior performance of N, P-HPC was attributed to the increased number of active sites, such as pyridinic N and P-C species, the favorable three-dimensional hierarchically porous structure and the large specific surface area (1516 m 2 g −1 ). The N, S and P co-doped porous carbon electrocatalyst, synthesized by pyrolysis of ginkgo leaves, exhibits superior ORR activity to undoped counterpart and two heteroatoms-doped carbon materials [114]. Obviously, the coordination structure and local environment of heteroatom dopants play a key role on intrinsic activities of biomass-derived carbon electrocatalysts.

Conclusion and Perspective
In summary, we have afforded a comprehensive minireview on the recent advance of biomass-derived carbon catalysts for oxygen electroreduction. The ORR activity significantly determines the performance of various energy-conversion devices, such as polymer electrolyte membrane fuel cells, microbial fuel cells and metal-air batteries. Therefore, biomass materials from different sources have been widely considered as scalable and sustainable catalysts [115,116]. That pore structure and active sites affect the physicochemical properties and electrocatalytic performance has been emphasized, and thus giving a guideline for rational design of biomass-derived carbon materials. Compared with the state-of-the-art Pt/C catalyst, biomass-derived carbons exhibit some intriguing advantages of lower production cost, better methanol and CO tolerance, as well as better stability [117,118]. However, several great challenges still remain for designing active electrocatalysts that exhibit comparable activity to Pt/C in acidic environment.
Although biomass-derived carbon electrocatalysts show excellent ORR activity in RDE measurements, it is still difficult to integrate such catalysts into energy conversion devices, especially in H 2 -O 2 fuel cells. Biomass materials are highly abundant and economical, but there has not been much progress yet in large-scale applications. Using biomass-derived carbon materials as substrates to reduce the noble metal loading is a promising strategy, which can simultaneously enhance the electrocatalytic activity and electrochemically active surface area. Here, we have demonstrated that the rational design of pore structure and active site plays a crucial role for synthesizing biomass-derived carbon electrocatalysts with satisfactory physicochemical and electrochemical properties. It is believed that developing effective and green synthetic methods will be a promising strategy to achieve more sustainable platforms.
On the other hand, in order to increase the competitiveness of biomass-derived carbon electrocatalysts for ORR, enhancing the intrinsic activity through optimizing active sites is urgently required. Consequently, understanding the catalytic mechanism of active sites and the relationship between mass transfer and pore structure for ORR process is important [119]. The in-depth investigation on these mechanisms helps to decide what types of biomass precursors should be chosen and what preparation methods should be performed. With the rapid development of materials science and technology, the community should further broaden the practical applications of biomass-derived carbon electrocatalysts in ORR-related devices, such as water splitting, supercapacitors [120,121], lithium-ion batteries [122,123], and CO 2 reduction, etc. In this way, large volumes of biomass and biowaste can be truly turned into a valuable resource and a sustainable society can be really established.