Trace Metals Dramatically Boost Oxygen Electrocatalysis of N-doped Coal-derived Carbon for Zinc-air Battery

The commercialization of metal-air batteries requires efficient, low-cost, and stable bifunctional electrocatalysts for reversible electrocatalysis of the oxygen reduction reaction (ORR) and the oxygen evolution reaction (OER). The modification of natural coal by heteroatoms such as N and S, or metal oxide species, has been demonstrated to form very promising electrocatalysts for the ORR and OER. However, it remains elusive and underexplored on how the impurity elements in coal may impact the electrocatalytic properties of coal-derived catalysts. Herein, we explore the influence of the presence of various trace metals that are notable impurities in coal, including Al, Si, Ca, K, Fe, Mg, Co, Mn, Ni, and Cu, on the electrochemical performance of the prepared catalysts. The constructed Zn-air batteries are further shown to be able to power green LED lights for more than 80 h. The charge-discharge polarization curves exhibited excellent and durable rechargeability over 500 ( ca . 84 h) continuous cycles. The promotional effect of the trace elements is believed to accrue from a combination of electronic structure modification of the active sites, enhancement of the active site density, and formation of a conductive 3-dimensional hierarchical network of carbon nanotubes.

Dear Editor, Enclosed please find our manuscript entitled "Trace metals dramatically boost oxygen electrocatalysis of N-doped coal-derived carbon for zinc-air battery" by Huimin Liu, Xinning Huang, Zhenjie Lu, Tao Wang, Yaming Zhu, Junxia Cheng, Yue Wang, Dongling Wu,*, Zhenyu Sun,* and Xingxing Chen, which we would like to submit for publication in Nanoscale.
Metal-air batteries possess very high theoretical energy densities but their rechargeability is still a challenge due to lack of efficient bifunctional catalysts for reversible oxygen electrodes. It is of paramount importance that bifunctional ORR/OER catalysts are not only efficient but also that they can be produced cheaply from abundant resources in the interest of sustainability, and potential for large-scale use. Coal is an attractive material with low cost and earth abundance that can be directly converted to useful electrocatalysts without the need for costly pre-refining treatment. However, the role of different trace metal elements in pristine coal on the performance of the catalysts still remains unclear. Herein, we make a systematic study on how different trace metal elements may boost oxygen electrocatalysis of coal-derived ORR/OER catalysts in alkaline solution. More importantly, we construct a rechargeable Zn-air battery, which shows excellent stability and can power a green LED bulb for more than 80 h, clearly underlining the robustness of the developed catalysts.
We declare that this manuscript has not been published and it is not under consideration for publication elsewhere. We are confident that this study will be highly interesting for scientists in the field of oxygen electrocatalysis. We are looking forward to hopefully positive referee comments.

Introduction
A looming energy crisis and environmental pollution are major problems facing humanity today, consequently, the development and use of non-polluting and sustainable renewable energies has received extensive attention. Achieving efficient, economical, and environmentally friendly energy conversion and storage systems is inevitably urgent for both current and future development [1][2][3][4] . The oxygen reduction reaction (ORR) and the oxygen evolution reaction (OER) are a pair of reversible electrochemical reactions involved in electrochemical energy conversion devices such as fuel cells and metal air batteries, among others, that are key to future green energy systems. However, slow kinetics and frustratingly poor reversibility are key hindrances that limit the energy efficiency of metal-air batteries and regenerative fuel cell systems [5,6] .
The best catalysts for ORR invariably consist of Pt, however, the OER activity of Pt is unsatisfactorily poor [7,8] , meanwhile, oxides of Ru and Ir are benchmark OER catalysts but exhibit dismal ORR activity [8,9] . A commonly adopted strategy for designing effective bifunctional ORR/OER catalysts involves the integration of individually excellent ORR and OER catalysts into a composite material embodying both functions. To this end, alloys of Pt, Ir, and Ru have been widely evaluated as bifunctional oxygen electrodes and they represent the archetypical benchmark, especially when used under acidic conditions. However, wide-scale technological application of such materials is evidently prohibitive due to their high cost and scarcity [10][11][12][13] . As such, the pursuit of non-Page 4 of 48 Nanoscale precious and effective bifunctional ORR/OER electrocatalysts is a research undertaking of topical importance.
Some of the most promising potential replacements that have been developed in recent years include heteroatom (e.g. N, S, P, B and Si) and single metallic atom (e.g. Fe, Mn and Co)-doped carbon materials [14][15][16][17][18][19] , among others. Among the carbon-based materials, coal has the advantages of low cost and abundant availability. Nagai et al. reported the use of coal as a raw material to prepare electrocatalysts for the ORR in acidic media [20][21][22][23] . We have also reported the feasibility to convert coal into promising ORR, OER and bifunctional ORR/OER electrocatalysts in alkaline media [11,[24][25][26] . It was previously demonstrated that coal can be directly modified into ORR, OER or bifunctional ORR/OER active catalysts without prior purification. However, natural coal has a very complex composition, consisting of C, H, O, N and S together with other inorganic elements (e.g. Si, Fe, Mn, Al and Mg) [20,24] . The electrocatalytic properties of carbon can be tremendously altered by the mere presence of traces of particular elements, either individually or through their interaction with other elements present in the coal. It would thus be prudent to understand how the various impurities in coal impact its electrocatalytic properties.
Here, we study for the first time the influence of traces of Fe, Al, Ca, K, Co, Si, Mg, Cu, Ni, and Mn, the notable impurity elements in coal, on its electrocatalytic ORR and OER performance. To achieve this, natural coal was first purified through treatment with hydrochloric acid and hydrofluoric acid to remove  [27] . But the growth of CNTs from purified coal samples impregnated with Al, and K after treatment in ammonia at 1050 °C was surprising as these elements are not known to catalyze the growth of CNTs.
Although it is possible that the purified coal could still have contained residual Page 6 of 48 Nanoscale transition metallic species that facilitated growth of the CNTs in the case of the Al, Ca, and K containing samples, it is evident to infer that growth of the CNTs must have contributed to enhancement of the ORR and OER activity. In addition, much as Al, Ca and K may not directly function as catalysts to grow the CNTs, they are well-known reduction agents [28][29][30] , and could possibly facilitate reduction of the real trace catalyst prior to growth of CNTs. Secondly, in the case of growth of the CNTs, Al, Ca and K, may contribute to enhancement of the ORR and OER activity of the catalyst through electronic modification of the active sites, either when encapsulated inside the CNTs, or in tandem through direct interaction with the active sites. Thirdly, transition metals and Ca have been reported to favor the reactivity of coal pyrolysis and the decomposition of unstable volatiles [11,26,31] . The existence of those trace metals would lead to surface defects such as multiple atom vacancies arising from structural reconstruction of the coal-based pyrolyzed carbon by the removal of heteroatoms (e.g. S and N). Meanwhile, a more porous structure with higher surface area is expected to be formed. Fourth, the melting points of the used

Results and discussion
The original brown coal (OC) was first crushed by ball milling followed by drying in an oven at 378 K for 4 h to obtain air-dried coal. After acid pretreatment with  Table 1.
The content of impurities was drastically decreased after the acid purification Page 8 of 48 Nanoscale steps, and even decreased after thermal treatment. Decrease of the H and O contents after thermal treatment was due to the release of unstable volatile components [11] . The N content clearly increased after thermal treatment of the samples under ammonia at 1050 °C, demonstrating successful introduction of nitrogen in the carbon. It seems that the acid pre-purification processes did not facilitate the removal of S, however, a significant decrease of the sulphur content was observed after the thermal treatment step, suggesting the removal of unstable sulphur components. The morphology of the PC, NPCC, and NPCC incorporated with trace elements (Fe, Al, Ca, K, Co) was studied by scanning electron microscopy (SEM) ( Figure   1). Interestingly, although no extra carbon source was introduced during the thermal treatment, bamboo-like CNTs ranging from a few tens to hundreds of nanometers in diameter were clearly observed when the above trace metal elements were introduced into PC before thermal treatment. It is believed that the PC itself was the carbon source, and the introduced trace elements were the possible active domains for catalyzing or promoting the growth of CNTs from volatile organic compounds arising from decomposition of the coal. It is known that transition metals such as Fe, Co, Ni promote CNT growth [32][33][34] . Interestingly, here we found that Al, Ca, and K can also facilitate the growth of CNTs. It is well known that Al, Ca, and K are reduction agents, which may have an effect on metals could be discerned in these two samples, demonstrating that most of the trace elements in the pristine coal were washed away during acid pretreatment. Raman spectroscopy provides insight into the graphitization degree of the prepared carbon materials (Figure 3a). The two strong peaks at 1342 (D band) and 1586 cm -1 (G band) correspond to disordered sp 3 structures and sp 2hybridized graphitic carbon, respectively [35] . Generally, the higher the intensity ratio of the D-to the G-band (ID / IG), the higher the defect degree and content of disordered carbon in the carbon materials [36] . The ID / IG value of PC (0.73) was considerably much smaller than that of the other samples (NPCC: 1.25; Fe-NPCC: 1.00; Al-NPCC: 1.14; Ca-NPCC: 1.10; K-NPCC: 1.10; Co-NPCC: 1.06). This indicates introduction of defects when PC was pyrolyzed in ammonia at 1050 o C. Such observation is in accordance with previous results that incorporation of nitrogen in carbon can occupy its lattice positions, thus leading to increase of disorder in the carbon structure [11,37] . Upon introduction of trace metals prior to the pyrolysis, the intensity ratio of D-to-G-bands slightly decreased. This is due to the decomposition of unstable volatiles generated from the PC carbon matrix catalyzed by the trace metals during the thermal treatment [25,38] . However, no clear peak shifts or line broadening of the G-and D-bands was observed, indicating that the properties of the carbon structure remained apparently unchanged after the doping [39] . The 2D-band (2680 cm -1 ) and (D + G)-band (2920 cm -1 ), which originate from the crystalline graphitic domains and structural disorder or defects, respectively, could be seen for the trace metal-containing NPCC samples. This suggests that stronger structural distortion of the aromatic graphitic carbon domains took place when trace metals were introduced into NPCC [25,26,40,41] .
The XRD patterns of the PC and NPCC-based samples are presented in Figure   3b. Both PC and NPCC samples have broad reflections at ca. 2θ of 18 to 29 o .
In the case of PC, this was because of presence of carbon with a predominantly  [38] . The presence of a pronounced peak at ca. 44 o is indicative of the presence domains of ordered graphene sheets [20,32,42] . Unfortunately, the presence of the trace metals could not be revealed by XRD at such low concentrations.   Table 2 and Figure 4e. The total oxygen content was drastically decreased after high-temperature treatment. This is mainly attributed to its loss as part of the high-temperature decomposition products, and the exchange with nitrogen atoms during N-doping treatment [24] .
Among the three oxygen species, the relative contribution of O 2groups in the NPCC-based samples generally increased after the thermal treatment, indicating the formation of metal oxides. Three nitrogen species, namely, pyridinic-N, pyrrolic-N, and graphitic-N were deconvoluted from the highresolution N 1s spectra, as shown in Figure 4c. As summarized in Table 2 and Figure 4f, the total nitrogen content increased after thermal treatment in NH3.
Among the three species, the content of pyridinic-N significantly increased, and the content of graphitic-N also increased in general after the high-temperature Page 15 of 48 Nanoscale treatment, which has been reported to promote the oxygen reduction [7,8] . The N 1s XPS spectrum of NPCC was deconvoluted into three peaks at 398.5, 400.5, and 401.5 eV, corresponding to pyridinic-N, pyrrolic-N, and graphitic-N, respectively. However, an apparent peak shift in the characteristic binding energy of pyridinic-N was observed in the NPCC samples containing trace metals, which could arise from the formation of metal-N-C moieties. Figure 4d presents high-resolution Fe 2p, Al 2p, Ca 2p, K 2p and Co 2p XPS spectra of NPCC-based samples. The main signal of the Al 2p spectrum in Al-NPCC was at 73.8 eV, which is attributed to the presence of Al2O3 [43] . The Ca 2p spectrum of Ca-NPCC consisted of two bands associated with Ca 2p3/2 (347.4 eV) and Ca 2p1/2 (350.9 eV) [44] . However, no clear characteristic 2p signals of Fe, K, and Co could be observed in the respective Fe-NPCC, K-NPCC, and Co-NPCC samples. Because XPS is a surface-sensitive technique with a maximum detection depth of ca. 10 nm on the sample surfaces [45] , the results imply that the surface concentration of the metals in these catalysts was too low to be detectable by XPS, or the metals were most likely encapsulated inside graphene layers as observed by TEM in our previous studies.  Normalized Intensity / a.u.  [a] The concentration of total element O and N with the sample as the basis.   A Tafel slope between 60 and 120 mV dec −1 implies that the ORR process is governed by both mechanisms [26] . The addition of Mg, Mn, Ni, Cu and Si was found to be unable to improve the ORR performance of NPCC ( Figure S3).

Binding Energy / eV
The long-term stability of ORR electrocatalysts is crucial for their practical application as cathode materials. Therefore, ORR durability tests for the three best catalysts (Ca-NPCC, Fe-NPCC and Al-NPCC) were performed galvanostatically at a constant current density of -1 mA cm -2 . As displayed in Figure 5e, all the three catalysts showed very promising stability with the potential remained almost unchanged after 280 hours. The potential retention was 98%, 98%, and 93% for Ca-NPCC, Fe-NPCC, and Al-NPCC, respectively.
The three samples were further exposed to methanol to evaluate their response to possible fuel crossover effects (Figure 5f). The presence of methanol is notoriously known to rapidly poison the ORR activity of Pt group metals due to strong adsorption of methanol oxidation products, or intermediates such as CO molecules that block ORR active sites. When methanol (3 M) was added to the electrolyte, the ORR currents for Ca-NPCC, Fe-NPCC, and Al-NPCC were unaffected, thus revealing their excellent tolerance to methanol crossover.
Whereas the commercial reference catalyst Pt/C suffered from a sharp decrease in ORR current as a result of the oxidation of methanol on the catalyst  Al-NPCC, defined as the potential corresponding to a current density of 1 mA cm -2 , were 1.40 and 1.34 V, respectively. The potentials at the current density of 10 mA cm -2 , a metric relevant to solar fuel synthesis [8] , were 1.77 and 1.72 V, respectively, for Fe-NPCC and Al-NPCC, better than that for Pt/C (1.90 V) [47] . The Tafel plots for the OER derived from Figure 8a are shown in Figure 6b.
A Tafel slope of ca. 60 mV dec −1 usually suggests that a chemical step after the first electron transfer is rate limiting, while a Tafel slope close to 40 mV dec −1 means that the second electron transfer step of the OER determines the reaction rate [25] . We note that the Fe-NPCC and Al-NPCC had the smallest  Figure 6c and 6d, respectively. Excellent preservation of the OER activity of the catalysts can be clearly observed. In the case of Al-NPCC, a slight but noticeable improvement of the OER performance was observed after the long-term CV experiment, probably due to morphological changes of the catalyst film or change in the surface properties such as gradual surface oxidation, leading to activation of more surface sites for the OER.
The bifunctionality of the catalysts for both ORR and OER were judged by the difference between the potential required for ORR at a current density of -1 mA cm -2 and the OER at a current density of 10 mA cm -2 . Fe-NPCC and Al-NPCC achieved a voltage difference of only 0.780 and 0.743 V, respectively, indicating a remarkable performance for a reversible oxygen electrode. Although Ca-NPCC showed excellent ORR performance, the catalyst presented rather dismal OER activity. The excellent bifunctional ORR/OER activity of the metal-doped NPCC catalysts stimulated us to further examine their practical application as reversible oxygen electrodes for rechargeable Zn-air batteries. We therefore constructed a rechargeable Zn-air battery, which consisted of a Zn-foil as an anode, metal-containing NPCC catalysts loaded on carbon paper as the air electrodes, and 6 M KOH with 0.2 M ZnCl2 as the electrolyte. The addition of ZnCl2 was to ensure occurrence of a reversible Zn/Zn 2+ electrochemical redox reaction at the anode [5,48] . oxygen reaction, thereby ensuring long-term stability [49][50][51][52] . This superior rechargeability of the batteries fabricated in this study unequivocally demonstrates the potential application of coal derived carbon catalysts incorporated with trace metals in building Zn-air batteries.

Conclusion
To summarize, we made a systematic study the role of various trace metals on the performance of N-doped coal-derived carbon (NPCC) materials as bifunctional ORR/OER electrocatalysts for rechargeable metal-air batteries. We found that modification of purified coal-derived carbon doped with nitrogen Koutecky Levich (K-L) analysis using the equation given below [11] : where j is the measured current density at the glassy carbon, j K and j L are the kinetic and diffusion-limiting current densities, respectively, ω is the angular velocity of the electrode, B is the Levich slope, n is the number of electrons transferred in the ORR process, F is the Faraday constant (96485 C mol -1 ), D is the diffusion coefficient of dissolved oxygen in the electrolyte, which for a 0.1 the electrolyte (1.2x10 -6 mol cm -3 ), v is the kinematic viscosity of the electrolyte (0.01 cm 2 s -1 ), and k is the rate constant for the ORR (m s -1 ).
The yield of the intermediate product (% HO 2 -) and the number of transferred electrons (n) during the ORR were calculated from the RRDE data using the following two equations [11] : converted to the reversible hydrogen electrode (RHE) by the equation [11] :

Characterization of samples
Scanning electron microscopy (SEM, HITACHI SU8010) and high-resolution transmission electron microscopy (TEM, HITACHI H-600) were used to characterize the morphology and structure of the as-produced samples. Raman spectra were recorded using a Bruker Vertex 70 system (532 nm laser). X-ray diffraction (XRD) measurements were carried out on a Bruker D8 X-ray diffractometer in the 2θ range from 10 to 80 o , with a Cu K α radiation (λ = 1.5418 Å) source. X-ray photoelectron spectroscopy (XPS) measurements were carried out using an ESCALab 250Xi electron spectrometer equipped with a 300 W Al K α X-ray source. Elemental analysis of C, N, H, S, and O was performed on an elemental analyzer Vario EL.