Philosophical Transactions of the Royal Society A: Mathematical, Physical and Engineering Sciences
You have accessResearch articles

Synthesis of unused-wood-derived C-Fe-N catalysts for oxygen reduction reaction by heteroatom doping during hydrothermal carbonization and subsequent carbonization in nitrogen atmosphere

Yasuto Goto

Yasuto Goto

Research Center of Supercritical Fluid Technology, Department of Chemical Engineering, Graduate School of Engineering, Tohoku University, 6-6-11 Aoba, Aramaki, Aoba-ku, Sendai 980-8579, Japan

Google Scholar

Find this author on PubMed

,
Yuta Nakayasu

Yuta Nakayasu

Research Center of Supercritical Fluid Technology, Department of Chemical Engineering, Graduate School of Engineering, Tohoku University, 6-6-11 Aoba, Aramaki, Aoba-ku, Sendai 980-8579, Japan

Frontier Research Institute for Interdisciplinary Sciences (FRIS), Tohoku University, 6-3 Aoba, Aramaki, Aoba-ku, Sendai 980-8578, Japan

[email protected]

Google Scholar

Find this author on PubMed

,
Hiroya Abe

Hiroya Abe

Frontier Research Institute for Interdisciplinary Sciences (FRIS), Tohoku University, 6-3 Aoba, Aramaki, Aoba-ku, Sendai 980-8578, Japan

[email protected]

Google Scholar

Find this author on PubMed

,
Yuto Katsuyama

Yuto Katsuyama

Research Center of Supercritical Fluid Technology, Department of Chemical Engineering, Graduate School of Engineering, Tohoku University, 6-6-11 Aoba, Aramaki, Aoba-ku, Sendai 980-8579, Japan

Google Scholar

Find this author on PubMed

,
Takashi Itoh

Takashi Itoh

Frontier Research Institute for Interdisciplinary Sciences (FRIS), Tohoku University, 6-3 Aoba, Aramaki, Aoba-ku, Sendai 980-8578, Japan

Google Scholar

Find this author on PubMed

and
Masaru Watanabe

Masaru Watanabe

Research Center of Supercritical Fluid Technology, Department of Chemical Engineering, Graduate School of Engineering, Tohoku University, 6-6-11 Aoba, Aramaki, Aoba-ku, Sendai 980-8579, Japan

Environment Conservation Center, Department of Chemical Engineering, Graduate School of Engineering, Tohoku University, 6-6-11 Aoba, Aramaki, Aoba-ku, Sendai 980-8579, Japan

[email protected]

Google Scholar

Find this author on PubMed

Published:https://doi.org/10.1098/rsta.2020.0348

    Abstract

    There is an urgent need to develop renewable sources of energy and use existing resources in an efficient manner. In this study, in order to improve the utilization of unused biomass and develop green processes and sustainable technologies for energy production and storage, unused Douglas fir sawdust (SD) was transformed into catalysts for the oxygen reduction reaction. Fe and N were doped into SD during hydrothermal carbonization, and the N- and Fe-doped wood-derived carbon (Fe/N/SD) was carbonized in a nitrogen atmosphere. After the catalyst had been calcined at 800°C, its showed the highest current density (−5.86 mAcm−2 at 0.5 V versus reversible hydrogen electrode or RHE) and Eonset value (0.913 V versus RHE). Furthermore, its current density was higher than that of Pt/C (20 wt% Pt) (−5.66 mA cm−2 @0.5 V versus RHE). Finally, after 50 000 s, the current density of sample Fe/N/SD (2 : 10 : 10) remained at 79.3% of the initial value. Thus, the synthesized catalysts, which can be produced readily at a low cost, are suitable for use in various types of energy generation and storage devices, such as fuel cells and air batteries.

    This article is part of the theme issue ‘Bio-derived and bioinspired sustainable advanced materials for emerging technologies (part 2)’.

    1. Introduction

    For the realization of sustainable development goals, it is important that regional ecosystems develop and use universal sustainable technologies. It is considered that the widespread use of renewable energy is essential for this purpose. Thus, there is an urgent need to develop renewable energy production and storage devices. At present, however, energy storage devices generally use precious/minor metals, such as platinum, lithium and cobalt, which are scarce, or highly toxic metals, such as selenium and lead. Moreover, the mining of these metals can lead to environmental destruction and pose a hazard to human health [1]. In addition, precious metals such as Pt are used as efficient electrode catalysts for the oxygen reduction reaction (ORR) at the cathode of fuel cells and air batteries. However, owing to price and resource constraints, it is necessary to develop low-cost alternative materials that exhibit high activity and stability. In recent years, to overcome these problems, significant research efforts have been devoted to produce the electrodes for rechargeable batteries [2] and fuel cells [3] from renewable organic substances obtained from biomass resources.

    Phthalocyanine iron-based electrodes show catalytic activity greater than that of Pt electrodes [4,5]. In addition, Fe- and N-doped carbon materials are attracting attention because of their high activity and durability and low cost and thus have significant potential for use as ORR catalysts instead of Pt [68]. In general, industrially produced organic substances and carbon materials derived from petroleum pitch are used for synthesizing these electrode materials. On the other hand, woody biomass resources are potentially inexpensive carbon raw materials owing to their abundance and sustainability. According to previous reports [911], when the wood is impregnated with Fe ions and carbonized through a two-step process (heating at 500°C for 2 h in an N2 atmosphere followed by heating at 850–900°C for 2 h in an N2 atmosphere), graphitic carbon is formed. It is known that the Fe surface functions as a catalyst and forms a graphitic shell structure with a 002 surface orientation. This suggests that electrically conductive carbon materials can be produced directly from woody biomass. In fact, this type of graphitic carbon has been used as an anode material for Li- and Na-ion batteries, and its storage capacity is similar to that of artificial graphite [12]. In order to synthesize ORR catalysts from biomass with high efficiency, it is necessary to ensure that the dopants Fe and N are present in a high concentration and are well dispersed.

    The hydrothermal carbonization (HTC) technique is the best method for synthesizing these materials. HTC is used for carbonizing organic substances and biomass at low temperatures (less than 300°C). In addition, it is an exothermic reaction and can be self-sustaining in terms of energy, meaning that the carbonization process only requires a small amount of energy input. In recent years, carbon materials fabricated by the HTC method have been employed as electrode materials in energy devices [13]. Moreover, the hydrothermal conditions involved make it easy to dope the biomass with heteroatoms [14]. However, in general, industrially produced organic substances are used as precursors, and there are few studies where woody biomass has been employed as the precursor instead.

    In Japan, 70% of the land area is forest, of which 18% is Japanese cedar [15]. Currently, the number of Japanese cedar trees is increasing as these trees were planted in large numbers after World War II. In addition, the utilization of these trees has decreased sharply because of the reduction in the price of imported materials owing to the oil revolution and a decrease in the number of buildings being built using wood. Because of the increase in the number of Japanese cedar trees, problems such as pollinosis and the inability to grow healthy wood owing to overcrowding have arisen. Thus, it has become essential to increase the utilization rate of these trees. Japanese cedar is mainly used in traditional Japanese architecture. However, the sawdust (SD) produced during its use is usually discarded. Hence, the development of high-value-added utilization methods for SD would be desirable for forestry development as well as from an environmental perspective. In other words, if the unused woody biomass were to be used to produce materials for energy devices, it would contribute to both the utilization of unused resources and the spread of sustainable technologies. While we plan to conduct research using Japanese cedar SD as a raw material in the future, in this study, we used SD from our collaborators as a raw material. Our group has previously used regional-wood-derived carbon (sometimes called biochar) to produce materials for energy storage devices such as organic batteries [16] and Na-ion batteries [17]. In addition, Abe et al. [18] developed ORR catalysts derived from organic substances. A previous report stated that ORR electrodes with good characteristics can be produced by immersing woody biomass in an aqueous solution of phosphoric acid and iron chloride, drying it and then carbonizing it in an NH3 atmosphere at 1000°C [19,20]. However, keeping the factors of safety and a low environmental load in mind, a process that does not involve toxic gases such as NH3 or a high-temperature environment would be desirable.

    In this study, to synthesize sustainable ORR electrodes, N and Fe were doped into SD during HTC, and the N-and Fe-doped carbon material was subsequently carbonized in an N2 atmosphere to produce catalysts for the ORR. This was done with the aim of increasing the utilization of unused SD as well as developing green processes and sustainable technologies.

    2. Methods

    (a) Materials

    SD of Douglas fir (Pseudotsuga menziesii) (particle size ≤ 200 µm) was provided by Futamura Chemical Co., Ltd. and used as a local woody biomass in this study. Potassium hydroxide (KOH) solution, hydrochloric acid, iron (III) nitrate enneahydrate (Fe(NO3)3 · 9H2O), which is an Fe source, and melamine, which is an N source, were purchased from FUJIFILM Wako Chemicals. Pt/C (20 wt% Pt) and an aqueous Nafion solution (5 wt%) were purchased from Sigma Aldrich. All the reagents were used as received.

    (b) Preparation of C-N-Fe catalysts

    To synthesize the C-N-Fe catalysts, we used a two-step carbonization process, as shown in figure 1. To begin with, 0–20 g of SD, 0–15 g of melamine, and 100 ml of an aqueous Fe(NO3)3 solution (0, 43.3, 86.6 or 173.2 g l−1) were placed in a 500-ml autoclave (TPR-1, Taiatsu Techne Co. LTD., Tokyo, Japan), and Fe ions were doped into SD overnight at room temperature. To subject the precursors to a hydrothermal treatment, the autoclave was maintained at 250°C for 4 h. After the hydrothermal treatment, the resulting solid was filtered, washed several times with ultrapure water and dried overnight in an oven at 105°C. Next, the solid was carbonized at 700–950°C for 1 h in a tube furnace in an N2 atmosphere. After being cooled to room temperature, the solid was washed overnight with 2 M HCl at 105°C. After filtration, the solid was washed with ultrapure water and EtOH and dried at 105°C overnight. The input ratios of the source materials and the annealing conditions used are given in the electronic supplementary material, tables S1 and S2. For comparison, a carbon material without N and Fe was also synthesized using only 20 g of SD and 100 ml of ultrapure water. All the other treatments were the same as those for the synthesis of the C-N-Fe catalysts.

    Figure 1.

    Figure 1. Two-step carbonization process used in this study: (a) simultaneous hydrothermal carbonization and N and Fe doping and (b) subsequent high-temperature carbonization in N2 atmosphere. (Online version in colour.)

    (c) Material characterization

    The crystalline structures of the synthesized samples were analysed using θ-2θ X-ray diffraction (XRD; Ultima, Rigaku) analysis, which was performed with Cu-Kα radiation, and Raman spectroscopy, which was performed using a laser with a wavelength of 532 nm. X-ray photoelectron spectroscopy (XPS; PHI 5000 VersaProbe II) was carried out to elucidate the ionic valences of N1s and Fe2p on the catalyst surfaces. The specific surface areas of the catalysts were measured using an N2 adsorption/desorption analyser (Bellsorp Mini II). The specific surface area was calculated from the adsorption points for relative pressures of 0.1–0.3 using the Brunner–Emmett–Teller (BET) theory. The compositions of the synthesized samples were analysed through CHNS analysis (Elementar Vario EL Cube). The morphologies of the catalysts were observed by transmission electron microscopy (TEM; Hitachi HT-7500). The compositions of the samples were also analysed by transmission electron microscopy coupled with energy-dispersive X-ray spectroscopy (TEM-EDS; JEOL JEM-2100F). Evaluation of iron configuration of the samples was carried out using X-ray absorption near edge structure (XANES) analysis at the SAGA Light Source.

    (d) Oxygen reduction reaction activity test

    First, 6 ml of a Nafion® perfluorinated resin solution (5 wt% solution in a mixture of lower aliphatic alcohols and water; Sigma Aldrich), 84 ml of ultrapure water and 334 ml of isopropyl alcohol were mixed by sonication. Then, 0.82 mg of the synthesized carbon catalysts and Pt/C (20 wt% Pt) being tested was added to 424 µl of the mixed solvent, and the slurry was dispersed by sonication. Then, 20 µl of the resulting catalyst ink was applied on a glassy carbon electrode and dried at room temperature and then the areal mass loading of all samples was 0.31 mg cm−2. A potentiostat (2325, BAS Co., Ltd, Japan) was used for the ORR activity tests. A 0.1 M KOH aqueous solution was used as the electrolyte, and O2 bubbling was performed for 30 min. Linear sweep voltammetry (LSV) was performed for the potential range of 0.2 V to −0.8 V at 1600 rpm using rotating ring-disc electrodes. The catalyst ink, Pt/C (20 wt% Pt) and Ag/AgCl electrodes (saturated aqueous KCl solution) were used as the working, counter and reference electrodes, respectively. The potential of the ring disc was maintained at 0.5 V, and the unreduced oxygen was detected. All the potentials were converted into the corresponding reversible hydrogen electrode (RHE) potentials using the following expression: ERHE=EAg/AgCl+0.0591×pH+EAg/AgCl0. We used a pH of 13 and an EAg/AgCl0 value of 0.199 V. Hence, the potential shift between the Ag/AgCl electrode and the RHE was 0.967 V.

    The number of electrons (n) involved in the ORR was calculated using the following equation:

    n=4×JDJD+JR/N, 2.1
    where JD and JR are the current densities of the disc and ring electrodes, respectively, and N′ is the capture efficiency (0.42). The durabilities of the catalysts were evaluated from their i–t chronoamperometric responses, which were measured using the Pt/C (20 wt% Pt) catalyst in O2-saturated 0.1 M KOH at +0.6 V (versus RHE) for 50 000 s at a rotational speed of 1600 rpm.

    3. Results and discussion

    (a) Effects of N and Fe doping on oxygen reduction reaction activity

    First, we investigated the effects of the N and Fe contents on the ORR activity. As mentioned above, the LSV curves of the synthesized catalysts were measured in O2-saturated 0.1 M KOH. The ratio of N and Fe as dopants was changed, and HTC was performed for semicarbonization at 250°C for 4 h, after which high-temperature carbonization was carried out at 850°C for 1 h. The labels ‘Fe’, ‘N’ and ‘SD’ in the figures indicate the weight of iron alone contained in the aqueous Fe(NO3)3 solution, the weight of melamine and the weight of SD, respectively.

    Figure 2a,b shows the results of the ORR activity tests for the catalysts with different N contents. The maximum current density and Eonset values of all the heterodoped catalysts were higher than those of the nondoped carbon sample. The current density was the highest (−4.65 mA cm−2 @0.5 V versus RHE) when a greater amount of N was doped into the carbon (Fe/N/SD ratio of 1 : 10 : 10). The onset potential was also high in this case (0.894 V versus RHE). On the one hand, an Fe/N/SD ratio of 1 : 15 : 5 resulted in a high current density (−4.21 mA cm−2 @0.5 V versus RHE) as well as the highest onset potential (0.904 V versus RHE). From the XRD patterns, which are shown in figure 2c, it can be seen that crystals of the (002) graphite phase grew when the amount of melamine added was small, such as when the Fe/N/SD ratio was 1 : 0 : 20 or 1 : 5 : 15. In general, when graphite crystals are formed, the ORR activity tends to decrease. On the other hand, when the amount of melamine added was large, as in the case of Fe/N/SD ratios of 1 : 10 : 10 and 1 : 15 : 5, the crystal growth of the (002) graphite phase was suppressed. As a result, the ORR activity was relatively high. In addition, the catalysts contained a large amount of Fe3C, which enhanced the ORR activity [2123]. The Raman spectra of the various catalysts, which are shown in the electronic supplementary material, figure S1, were similar. Furthermore, figure 2d shows the BET surface areas of the catalysts, as calculated from the N2 adsorption measurements. The catalyst with the Fe/N/SD ratio of 1 : 10 : 10 had a higher specific surface area than that of the catalyst with the Fe/N/SD ratio of 1 : 15 : 5. Generally, as the specific surface area increases, the number of active sites also increases and so does the ORR activity [24,25].

    Figure 2.

    Figure 2. Results of ORR activity tests: (a) effect of the amount of melamine added on LSV measurement results. (b) E1/2 is potential corresponding to half of the maximum current density while current density at 0.5 V was determined from analysis of data in (a). (c) XRD patterns of catalysts with different melamine (N) contents. (d) BET surface areas of catalysts with different melamine (N) contents. (Online version in colour.)

    Figure 3a shows the fitting results for the deconvoluted N1s XPS spectra of the four catalysts with different N contents; the result for the catalyst formed using SD only, Fe/N/SD ratio of 0:0:20, is not shown. The N1s peaks could be deconvoluted into five peaks, which were located at approximately 398.6, 400.5, 401.3, 402.0 and 404.5 eV. These corresponded to pyridinic N, pyrrolic N, graphitic N, oxidized pyridinic N and N-O, respectively [26]. For the sample with an Fe/N/SD ratio of 0 : 0 : 20, an N component was not detected, in contrast with the case for the other three catalysts. This indicates that N was doped into the latter samples during the HTC process owing to the addition of melamine. Figure 3b shows the elemental compositions of the surfaces of the catalysts. It can be seen that the N content of the catalysts increased with the increase in the amount of N dopant added. In addition, as the amount of Fe dopant added was increased, the Fe content of the catalysts also increased. This suggests that Fe was also incorporated into the catalysts by the HTC process. Figure 3c shows the contributions of the different N components to the N1s spectrum. It was confirmed that the sample with the Fe/N/SD ratio of 1 : 10 : 10 contained more pyridinic N than did the sample with the Fe/N/SD ratio of 1 : 15 : 5. The presence of pyridinic N enhances the ORR activity [27].

    Figure 3.

    Figure 3. Results of XPS analysis: (a) fitting results of deconvoluted N1s XPS spectra of four catalysts with different N contents (results for SD only are not shown). N1s spectra were deconvoluted into five peaks, located at approximately 398.6, 400.5, 401.3, 402.0 and 404.5 eV, which correspond to pyridinic N, pyrrolic N, graphitic N, oxidized pyridinic N, and N-O, respectively. (b) Elemental compositions of catalyst surfaces. (c) Contributions of various N components to N1s spectrum. (Online version in colour.)

    Thus, based on the above-stated results, it can be concluded that the activity of the sample with an Fe/N/SD ratio of 1 : 10 : 10 was higher than that of the sample with an Fe/N/SD ratio of 1 : 15 : 5, even though the former had lower Fe and N contents. This is because it had a higher specific surface area and greater Fe3C and pyridinic N contents.

    Figure 4a,b shows the results of the ORR activity tests for the catalysts with different amounts of the Fe dopant. The ORR activity of the catalysts increased with the increase in the Fe content, with the sample with the Fe/N/SD ratio of 2 : 10 : 10 exhibiting the highest current density (−5.23 mA cm−2 @0.5 V versus RHE) and Eonset value (0.900 V versus RHE). However, a further increase in the amount of Fe doped (i.e. an Fe/N/SD ratio of 4 : 10 : 10) caused a significant decrease in the ORR activity. Figure 4c shows the XRD patterns of the various catalysts. Focusing on the peak of the (002) phase of graphite, it can be seen that the peak in the case of the Fe/N/SD (4 : 10 : 10) sample was much sharper than those of the other samples, whereas that of the Fe/N/SD (2 : 10 : 10) sample was broader than that of the Fe/N/SD (1 : 10 : 10) sample. Furthermore, as can be seen from the Raman spectra, which are given in figure 4d, the intensity of the D band, which originates from defects in the graphene structure, was very high in the cases of samples Fe/N/SD (1 : 10 : 10) and Fe/N/SD (2 : 10 : 10). The two-dimensional band is related to second-order processes and can be attributed to two-phonon lattice vibrations. However, unlike the D band, it is not related to proximity to a defect. In general, the existence of a two-dimensional band suggests the presence of multilayer graphene. The intensity of the two-dimensional band in the case of the Fe/N/SD (4 : 10 : 10) sample was extremely high while the band was the broadest in the case of the Fe/N/SD (2 : 10 : 10) sample. Thus, sample Fe/N/SD (2 : 10 : 10) contained the smallest amount of graphite. In fact, from the BET surface areas shown in the electronic supplementary material, figure S2, it was confirmed that sample Fe/N/SD (2 : 10 : 10) had a specific surface area twice as large as that of sample Fe/N/SD (1 : 10 : 10). This probably resulted in poor growth of the graphite crystals, which, in turn, increased the ORR activity. Thus, based on the above-discussed results, it can be concluded that the growth of the graphite crystals suppressed the ORR activity.

    Figure 4.

    Figure 4. (a) Effects of amount of Fe(NO3)3 added on LSV measurement results. (b) E1/2 is potential corresponding to half of the maximum current density for each sample while current density at 0.5 V was determined by analysis of results of ORR activity tests. (c) XRD patterns of catalysts formed using different amounts of Fe(NO3)3. (d) Raman spectra of catalysts formed using different amounts of Fe(NO3)3. (Online version in colour.)

    Figure 5a shows the fitting results for the deconvoluted N1s XPS spectra of the four catalysts with different amounts of the Fe dopant; the result for the catalyst consisting only of SD is not shown. The N1s peaks could be deconvoluted into five peaks, as shown in figure 3a. The presence of N was confirmed in the case of all the catalysts. From figure 5b, it was confirmed that the N content decreased as the proportion of the Fe dopant was increased. On the other hand, as the amount of the Fe dopant was increased, the proportion of Fe increased in the case of sample Fe/N/SD (2 : 10 : 10) but decreased for sample Fe/N/SD (4 : 10 : 10). Electronic supplementary material, figure S3 shows the pre-HCl treatment XRD patterns of the catalysts after the completion of the high-temperature carbonization process. It can be seen that a peak related to α-Fe was dominant in the pattern of sample Fe/N/SD (4 : 10 : 10) before the HCl treatment. However, the α-Fe content was low while Fe3C was abundant in samples Fe/N/SD (1 : 10 : 10) and Fe/N/SD (2 : 10 : 10). Since Fe3C does not dissolve readily in HCl and α-Fe does, this may be the reason that almost no Fe was detected in sample Fe/N/SD (4 : 10 : 10) after the HCl treatment. Furthermore, the extent of crystal growth of the (002) graphite phase in sample Fe/N/SD (1 : 10 : 10) was greater than that in sample Fe/N/SD (2 : 10 : 10). Therefore, in sample Fe/N/SD (2 : 10 : 10), the amount of Fe3C was higher than the amount of graphite. Hence, the Fe content in sample Fe/N/SD (2 : 10 : 10) was higher than that in sample Fe/N/SD (1 : 10 : 10). Figure 5c shows the contributions of the various N components to the N1s spectrum. The presence of graphitic N suggests that N was incorporated into the graphene structure. On the other hand, the content of the most active N component, namely, pyridinic N, in sample Fe/N/SD (2 : 10 : 10) was less than that in sample Fe/N/SD (1 : 10 : 10). Since the amount of oxidized pyridinic N in sample Fe/N/SD (2 : 10 : 10) was high, we believe that the pyridinic N was oxidized, that is, converted into oxidized pyridinic N.

    Figure 5.

    Figure 5. Results of XPS analysis: (a) fitting results for deconvoluted N1s XPS spectra of four catalysts formed using different amounts of N (results for SD only not shown). N1 s peaks could be deconvoluted into five peaks, located at approximately 398.6, 400.5, 401.3, 402.0 and 404.5 eV, which correspond to pyridinic N, pyrrolic N, graphitic N, oxidized pyridinic N and N-O, respectively. (b) Elemental compositions of catalyst surfaces. (c) Contribution of each N component to N1s spectrum. (Online version in colour.)

    Thus, given the results discussed above, since sample Fe/N/SD (2 : 10 : 10) had a relatively higher Fe3C content and lower graphite content, its ORR activity was higher than that of sample Fe/N/SD (1 : 10 : 10). Therefore, it can be concluded that the doping of Fe and N efficiently improved the ORR activity of the synthesized catalysts.

    Next, we focused on sample Fe/N/SD (2 : 10 : 10) and subjected it to a detailed analysis because this catalyst showed the highest ORR activity.

    (b) Effects of the temperature of second carbonization process on oxygen reduction reaction activity

    We investigated the effect of the carbonization process performed after HTC on the ORR activity. The ORR activities of the catalysts synthesized at different temperatures were measured in an O2-saturated 0.1 M KOH solution. As stated previously, the labels ‘Fe’, ‘N’ and ‘SD’ in the figures indicates the weight of the Fe alone, that of melamine, and that of SD, respectively.

    Figure 6a,b shows the ORR activities of the samples carbonized at different temperatures. The ORR activity increased with the carbonization temperature, and the catalyst carbonized at 800°C showed the highest current density (−5.86 mA cm−2 at 0.5 V versus RHE) and Eonset value (0.913 V versus RHE). In particular, the current density was higher than that of Pt/C (20 wt% Pt) (−5.66 mA cm−2 @0.5 V versus RHE). On the other hand, the Tafel slope plotted from the LSV measurement with 1600 rpm, shown in the electronic supplementary material, figure S4 and the XRD patterns, shown in figure 6c, indicated that there was almost no difference between the samples. Furthermore, figure 6d shows the BET surface areas as calculated from the N2 adsorption measurements. It was confirmed that the specific surface area decreased gradually as the carbonization temperature was increased. This can be attributed to the crystal growth of graphite being greater at higher temperatures. However, no significant difference was observed between the samples whose carbonization temperatures varied by 50°C. Thus, we only focused on the samples carbonized at 700, 800 and 900°C.

    Figure 6.

    Figure 6. Results of ORR activity test. (a) Effects of carbonization temperature on LSV measurement results. (b) E1/2 is potential corresponding to half of the maximum current density and current density at 0.5 V was determined from data shown in (a). (c) XRD patterns of catalysts formed at different carbonization temperatures. (d) BET surface areas of catalysts formed at different carbonization temperatures. (Online version in colour.)

    Figure 7a shows the fitting results for the deconvoluted N1s XPS spectra of the catalysts carbonized at the three above-mentioned temperatures. The N1s peaks could be deconvoluted into five peaks, as shown in figure 3a. The peak shape did not change significantly between the three samples. Figure 7b shows the elemental compositions of the catalyst surfaces. It was confirmed that, as the carbonization temperature was increased, the C content increased, the N and O contents decreased and the Fe content increased, while the similar trend for the N and Fe content was confirmed from the CHNS analysis results listed in the electronic supplementary material, table S3. This indicates that the number of functional groups derived from the organic substances in SD was reduced as carbonization was performed at higher temperatures. Figure 7c shows the contributions of the different N components to the N1s spectrum. It was found that, as the carbonization temperature was increased, the contents of the active N components, namely, graphitic N and pyridinic N, decreased. On the other hand, that of pyrrolic N increased. In addition, since the content of the inactive oxidized pyridinic N increased, it is speculated that the pyridinic N was oxidized by the high-temperature carbonization process.

    Figure 7.

    Figure 7. Results of XPS analysis: (a) fitting results for deconvoluted N1s XPS spectra of catalysts carbonized at 700, 800 and 900°C. N1s peaks could be deconvoluted into five peaks, located at approximately 398.6, 400.5, 401.3, 402.0 and 404.5 eV, which correspond to pyridinic N, pyrrolic N, graphitic N, oxidized pyridinic N and N-O, respectively. (b) Elemental compositions of catalyst surfaces. (c) Contribution of N components to N1s spectrum. (Online version in colour.)

    The ORR activity of the catalyst carbonized at 800°C was higher than that of the catalyst carbonized at 700°C, although the former showed the highest active pyridinic N content as well as the highest specific surface area. Therefore, it is likely that the increase in the electrical conductivity owing to the formation of graphitic carbon and the increase in the relative amount of Fe3C present also raised the ORR activity. This is supported by the fact that the amount of charge during ORR at 0 rpm in the range between 0.6 and 1.0 V (versus RHE) shown in the electronic supplementary material, table S4 was the largest in the catalyst calcined at 800°C.

    Figure 8a shows TEM images of the carbon material with an Fe/N/SD ratio of 2 : 10 : 10 after HTC and the catalysts formed by its carbonization at 700, 800 and 900°C. The dark areas contain Fe and graphitic structures, such as graphite shell chains, in keeping with a previous report [9]. We further analysed the catalyst carbonized at 800°C using TEM/EDS and XANES analysis; the results are shown in figure 8b–d. We analysed one point in the dark area in the TEM image of the catalyst using EDS. The result of the EDS analysis showed that the analysed area contained Fe, O and N, suggesting that the dark area did not consist only of pure Fe with a zero valence. Electronic supplementary material, figure S5 shows the fitting results for the deconvoluted Fe2p XPS spectra of the catalysts carbonized at 700, 800 and 900°C. Fe with valences ranging from 0 to 3 was present in all the catalysts. Further, the peak of zero-valent iron (ZVI) became sharper as the temperature was increased. In general, ZVI does not show ORR activity. Thus, it is likely that the activity decreased as the carbonization temperature was increased. In addition, the XANES spectrum at the Fe K-edge of the catalyst calcinated at 800°C is shown in figure 8d. The pre-edge and edge features of the catalyst are clearly different from those of Fe-foil, Fe(II) Phthalocyanine, FeO, Fe3O4 and α-Fe2O3, while the shape matches well with that of Fe3C shown in a previous report [28]. This trend is corresponding to the result of XRD.

    Figure 8.

    Figure 8. (a) TEM images of carbon sample after HTC and catalysts formed by its carbonization at 700, 800 and 900°C. (b) TEM/EDS analysis results for catalyst carbonized at 800°C. (c) A result of EDS analysis of area enclosed by the red circle in (b). (d) Results of XANES spectrum for catalyst carbonized at 800°C and other iron-related standard samples. (Online version in colour.)

    Finally, we analysed the Fe/N/SD (2 : 10 : 10) catalyst carbonized at 800°C in detail and also investigated its stability.

    There are two possible reaction routes for oxygen reduction in a basic solution, as described in [29]):

    O2+2H2O4e+4OHE˚=1.23 V vs. RHE 3.1
    and
    O2+H2O+2eHO2+OHE˚=0.76 V vs. RHE. 3.2

    The four- (equation 3.1) and two-electron (equation 3.2) electrochemical reactions listed above are important with respect to energy devices such as fuel cells and metal-air batteries as well as industrial production. Rotating ring-disc electrodes were used to evaluate both reactions. In this study, a disk electrode was used to measure the LSV curves for analysing the ORR kinetics while a ring electrode collected the HO2 produced at the disc electrode. The electron reaction number was calculated from the reduction and oxidation currents at the disc and ring electrodes, respectively (equation 2.1). A comparison of the results for sample Fe/N/SD (2 : 10 : 10) and Pt/C (20 wt% Pt) is shown in figure 9a. The electron number for both catalysts was close to four over a wide range.

    Figure 9.

    Figure 9. (a) Electron reaction number as calculated from reduction and oxidation currents at disc and ring electrodes, respectively. Comparison of results for sample Fe/N/SD (2 : 10 : 10) and Pt/C (20 wt% Pt). (b) Chronoamperograms of sample Fe/N/SD (2 : 10 : 10) and Pt/C (20 wt% Pt) for 50 000 s to evaluate durability. (Online version in colour.)

    The chronoamperograms of sample Fe/N/SD (2 : 10 : 10) and Pt/C (20 wt% Pt) for 50 000 s (over 13 h) are shown in figure 9b. The chronoamperograms were recorded to evaluate the durabilities of the catalysts. After 50 000 s, the current density of sample Fe/N/SD (2 : 10 : 10) remained at 79.3% of the initial value while that of Pt/C (20 wt% Pt) decreased to 80.6%. These results indicate that the Fe- and N-doped catalyst was as stable as Pt/C (20 wt% Pt). In the case of Pt/C (20 wt% Pt), the desorption and aggregation of Pt occur upon the long-term application of a voltage [30]. However, in the case of the synthesized catalysts, N atoms were embedded within the carbon material and were responsible for the observed high stability of the catalysts.

    4. Conclusion

    In this study, to improve the utilization of unused resources as well as to develop green processes and sustainable technologies, unused Douglas fir SD was transformed into catalysts for the ORR. Fe and N were doped into SD during HTC, and the N- and Fe-doped wood-derived carbon (Fe/N/SD) was carbonized in a N2 atmosphere. It was found that crystal growth on the (002) graphite phase suppressed the ORR activity, while the existence of Fe3C enhanced the ORR activity. The catalyst having an Fe/N/SD ratio of 2 : 10 : 10 and carbonized at 800°C showed the highest current density (−5.86 mA cm−2 at 0.5 V versus RHE) and Eonset value (0.913 V versus RHE), with its current density being higher than that of Pt/C (20 wt% Pt) (−5.66 mA cm−2 @0.5 V versus RHE). After 50 000 s, the current density of the sample with the Fe/N/SD ratio of 2 : 10 : 10 remained at 79.3% of the initial value. In the future, to enhance the ORR activity of the thus-fabricated catalysts, we aim to optimize the carbonization process such that the pyridinic N is not oxidized, no elemental Fe remains and the degree of graphitization is low.

    Data accessibility

    Supporting information has been provided in the electronic supplementary material.

    Authors' contributions

    Y.G. performed the experiments, improved the study design and drafted the Experimental section. Y.N. conceived and designed the study, performed the data analysis and drafted the manuscript. H.A. designed the study, trained Y.G. on performing the experiments and analysing the results of the ORR activity tests, and drafted the section discussing the properties of the ORR catalysts. Y.K. drafted the section that discusses the electrochemical and structural properties of the catalysts. T.I. trained Y.G. on the material characterization techniques and drafted the section that discusses the results of the XRD and Raman analyses. M.W. designed and supervised the study, set up the experimental equipment and edited the manuscript. All authors read and approved the manuscript.

    Competing interests

    We declare we have no competing interests.

    Funding

    This study was financially supported by the Program for Creation of Interdisciplinary Research, Frontier Research Institute for Interdisciplinary Sciences, Tohoku University.

    Acknowledgements

    We thank Dr Takamichi Miyazaki of the Technical Division, School of Engineering, Tohoku University, for assistance with the TEM and TEM/EDS analyses, and Dr Ryotaro Kumashiro and Dr Takaaki Tomai of Advanced Institute for Materials Research, Tohoku University, for assistance with the N2 adsorption/desorption measurements.

    Footnotes

    One contribution of 11 to a theme issue ‘Bio-derived and bioinspired sustainable advanced materials for emerging technologies (part 2)’.

    These authors contributed equally to this work.

    Electronic supplementary material is available online at https://doi.org/10.6084/m9.figshare.c.5506754.

    Published by the Royal Society. All rights reserved.

    References