Cobalt Nanoparticles Encapsulated in Nitrogen-Doped Carbon Nanotube as Bifunctional-Catalyst for Rechargeable Zn-Air Batteries

Developing highly efficient and low-cost non-precious-metal electrocatalysts toward oxygen reduction reaction (ORR) and oxygen evolution reaction (OER) is crucial for improving the performance and economic outlook of alkaline-based rechargeable Zinc-air battery technologies. In this work, a nitrogen-doped carbon nanotube encapsulated with metallic cobalt nanoparticles (Co@NC) was synthesized through a facile method and subsequent pyrolysis treatment. The field emission scanning electron microscope, high resolution transmission electron microscopy, Raman spectra investigations demonstrate that the presence of Co induces the formation of carbon nanotube during the pyrolysis process and increase degree of graphitization of carbon nanotubes. The reversible oxygen electrode activity can be evaluated by the difference of OER and ORR metrica ΔE. The Co@NC exhibit a ΔE value of 0.91 V, less than those of Pt/C (1.1 V) and nitrogen-doped carbon (1.17 V). The primary Zn-air battery based on Co@NC display an open-circuit potential 1.4 V, a high power density of 137 mW cm-2, and superior high energy density of (708.3 mAh·kgZn-1 and 868.9 Wh·kgZn-1 at 10 mA·cm-2), superior to that of the commercialized Pt/C.


INTRODUCTION
In response to the growing global energy demand and environmental crisis, extensive research has focused on energy conversion and storage devices. It is well-known that oxygen reduction reaction (ORR) and oxygen evolution reaction (OER) play a vital role in various energy storage systems, for instance, metal-air batteries (Park et al., 2016;Feng et al., 2018;Guo et al., 2018a;Wen et al., 2018) and fuel cells (He et al., 2013;Karim et al., 2017).
For this reason, excellent electrochemical performance Zn-air batteries need a bifunctional catalyst, which reveal excellent catalytic performance and stability, for ORR and OER. Traditionally, Platinum is deemed the best catalyst for ORR, while iridium is reputed the greatest catalyst for OER. However, some shortcomings such as exorbitant price, scarcity and poor stability have prohibited their extensive application (Deng et al., 2013;Dong et al., 2016;Tian et al., 2016;Qiao et al., 2017;Si et al., 2017).
To address above issue, various non-precious metal electrocatalyst, including transition metal oxides, transition metal nitrides, various carbon, and metal-free heteroatoms doped carbon were explored (Deng et al., 2015;Tao et al., 2015;Zhou et al., 2015;Cao et al., 2016;Qu et al., 2016Qu et al., , 2017Yang Z. K. et al., 2016;Mo et al., 2017;Guo et al., 2018b;Liu et al., 2018). Among the various electrocatalysts being developed, CNT can not only improve electron transport, but also maintain and improve the stability of the catalyst under harsh conditions, which due to the metal cobalt encased in carbon nanotubes (Fang et al., 2016a;Guo S. et al., 2017;Sheng et al., 2017). Liu et al. adopted a hybrid strategy to synthesize carbon nanotubes and graphene as the carrier of metallic cobalt, and the catalyst also maintain high catalytic activity and stability under severe acidic conditions. Zheng et al. found that cobalt nanoparticles were wrapped in carbon nanotubes, which could prevent Co nanoparticles from being etched and oxidized in the post-process after pyrolysis, which have a pivotal part in enhancing the electrocatalyst stability to ORR and OER. Deng et al put forward a new idea to synthesize carbon nanotube encapsulated iron as ORR catalyst. In their opinion, the encapsulated metallic Fe can reduce the local work function of the surface carbon, which improve the ORR performance Zhe et al., 2017;Zheng et al., 2017;Liu et al., 2018).
Recently, nitrogen-doped carbon nanotube has the potential to replace the metal-free-based electrocatalysts, which is mainly due to the nitrogen induced charge delocalization (Hao et al., 2017). Deng et al. obtained a suite of NCNTs by controlling the pyrolysis temperature of precursor polyaniline (PANI), which achieved the doping of nitrogen atoms into the carbon skeleton at the atomic level. By researching the unique electron configurations of carbon atoms in combination with different nitrogen functional groups, they found that nitrogen doping can provide atomic-level catalytic activity centers for ORR . Wang et al. developed an economical approach to obtain metallic Co encapsulated in N,S co-doped carbon nanotubes. the porous channels of the carbon skeleton provide more space for the dispersion of cobalt nanoparticles, and the introduction of nitrogen and sulfur into carbon nanotubes can provide more catalytic active sites   Jie et al., 2017;Li et al., 2017;Song et al., 2017;Dilpazir et al., 2018). However, in these experiments, the introduction of a template or the use of ammonia gas increased the cost and risk of the experiment.
Herein, cobalt acetate tetrahydrate (Co(Ac) 2 ·4H 2 O), dicyanamide (DCDA) and Poly(ethyleneglycol)-blockpoly(propyleneglycol)-block-poly(ethyleneglycol) (P123) were elaborately selected to fabricate cobalt nanoparticles@Ndoped carbon nanotubes composites (Co@NC). During the pyrolysis process, dicyanamide not merely do duty for a nitrogen source and carbon source and not only uses its reductivity to form the cobalt nanoparitcles. P123, serve as the main carbon source for the formation carbon nanotube structure under cobalt species inducing effect. In terms of its excellent overall catalytic performance for ORR and OER, the prepared Co@NC electrocatalyst is served as a cathode in rechargeable Zn-air battery and reveals a high specific power density, peak power density and stability, superior to that of the commercialized Pt/C.

Synthesis of Co@NC Electrocatalysts
Co(Ac) 2 ·4H 2 O and P123 were dissolved with H 2 O and HCl in the 200 mL beaker. After stirring at 50 • C for 12 h, the DCDA was added. The final molar ratio of Co(Ac) 2 ·4H 2 O: P123: DCDA: HCl: H 2 O in the mixture was 0.003: 0.001: 1: 0.8: 13. Then the solution was dried at 50 • C to obtain precursor. The bulk precursor was calcined in a tube furnace under the protection of a N 2 atmosphere. Firstly, it was arised to 400 • C at a heating rate of 5 • C/min and maintained 1 h, then the temperature further arised to 900 • C at the same heating rate and maintained 1 h. In the preparation of the precursor, we performed an experiment without adding a cobalt source under the same synthesis conditions, and the obtained sample was named NC.

Electrochemical Measurements
Electrochemical measurements were performed on a Metrohm Autolab (Swiss Wantong Co. Ltd, Switzerland), using a typical three-electrode device. The rotation ring-disk electrode (RRDE, ϕ = 5 mm, Swiss Wantong Co., Ltd.) was used as working electrode, Ag/AgCl electrode was used as referode electrode and platinum foil was used as counter electrode. The fabrication of the working electrode was carried out as following: catalyst (5 mg) was mixed with H 2 O (0.8 mL), CH 3 CH 2 OH (0.2 mL) and Nafion solution (50 µL) in an ultrasonic bath to get evenly dispersed ink.
Then, 10 µL of the ink was dripped onto the glass carbon surface and natural drying at room temperature. The mass loading of catalyst on the surface of glass carbon electrode was 0.24 mg·cm −2 . All potentials in this paper are based on reversible hydrogen electrode potentials (RHE): .
Cyclic voltammetry (CV) were measured in N 2 -saturated or O 2 -saturated 0.1 M KOH solution with a sweep rate of 10 mV·s −1 . The RDE and RRDE tests were carried out in O 2saturated 0.1 M KOH solution at 1600 rpm with a scan rate of 10 mV·s −1 . The electrochemical impedance spectroscopy (EIS) of catalysts were measured in O 2 -saturated KOH with 5 mV amplitude from 100 kHz to 0.01 Hz. NC and Pt/C catalyst were also measured using the same methods for comparison. All of the above electrochemical tests were performed at room temperature.

Material Characterization of the Co@NC
The overall synthetic process of Co@NC is shown in Figure 1a.
In the first step, nanorod colloid formed by P123 self-assembly in 2 M HCl solution containing amount of Co 2+ . Secondary, DCDA with rich -NH 2 , -NH and -C≡N functional group was added to obtain the precursor. Finally, the bulk precursor was annealed under N 2 atmosphere to obtain Co@NC. During precursor pyrolysis, carbon nanotube formed in the presence of cobalt, and Co 2+ was reduced to Co nanoparticles encapsulated in carbon nanotube.
Field emission scanning electron microscope (FESEM) and TEM were used to investigate the as-obtained samples microstructure (Figures 1b-d and Figures S2, S3). The low resolution FESEM and TEM image (Figures S2a, S3a) indicate Co nanoparticles was encapsulated in carbon nanotube successfully, and the particle size of Co nanoparticles is within the scope of 20-100 nm. The width and length of carbon nanotube is about 200 nm and 10 µm, respectively. the interplanar spacing was about 0.203 nm (Figure 1d), corresponding to the distance between the (111) planes of cobalt particles. Moreover, cobalt nanoparticles can be observed clearly encapsulated by several graphene layers. The TEM elemental mapping images show an atomically uniform distribution of the C, N and Co elements in prepared Co@NC sample (Figure 1e).
Raman measurements were then used to investigate the carbon defects and the degree of graphitization. In Figure 2B, two distinct peaks at 1,360 and 1590 cm −1 are associated with D and G bands, respectively. The area ratio between D and G bands (S D /S G ) were calculated as 3.047 and 1.489 for NC and Co@NC, respectively. The results indicate that the existence of cobalt could improve the degree of graphitization, which will have a positive effect on its conductivity and catalytic activity . Three characteristic peaks of Co were observed around 475.6, 516, and 676 cm −1 by Raman spectroscopy, which correspond to classical vibration modes E g , F 2g , and A 1g of Co, respectively (Jie et al., 2017;Li et al., 2017).
X-ray photoelectron spectroscopy (XPS) was carried out to analyse the chemical valence state and surface element composition of Co@NC and NC in Figures 2C,D and  Figures S6, S7. The survey spectra revealed that NC only contains three elements of C, N, O, and C, N, O, Co were coexistent in Co@NC ( Figure S6). The specific contents of all elements are summarized in Table S1. The high resolution C1s spectrum were deconvoluted into three peaks ( Figure S7a): C-C (284.6 eV), C-O (285.4 eV), and C=O (286.5 eV) (Dilpazir et al., 2018). The high resolution C1s spectrum ( Figure 2C) exhibits four types of N species at 398.5, 400.1, 401, 403.6 eV, which correspond to pyridinic N, pyrrolic N, graphitic N, and oxidized N (Zhao and Chen, 2015;Fang et al., 2016a;Zhou T. et al., 2017). The content of pyridinic N was also higher than that of the other three nitrogen species in Co@NC, higher content of pyridine N is expected to exhibit superior ORR performance . As shown in Figure 2D, high resolution Co2p spectrum shows two types of Co species at 780.3 and 795.8 eV, corresponding to metallic Co2p3/2 and Co2p1/2 (Zhou L. et al., 2017). This further suggested that Co 2+ was induced to cobalt successfully used DCDA as reducing agent. The presence of Co 2+ and Co 3+ may be due to the oxidation of the surface of the sample exposed to air during the preparation process . According to the XPS results (Zhou L. et al., 2017), the relative content of Co in the Co@NC sample was calculated to be about 0.91 at%.

Electrochemical Performance for ORR and OER
The ORR catalytic activity of Co@NC and NC were measured by cyclic voltammetry (CV) in N 2 -saturated and O 2 -saturated 0.1 M KOH solution with a sweep rate of 10 mV·s −1 , firstly. The diffusion current density of Co@NC, NC and Pt/C ( Figure S8) both increases significantly in the case of the introduction of O 2 , which indicate the intrinsic ORR catalytic activity of the Co@NC and NC.
To gain further insights into the electrocatalytic behavior of Co@NC and NC toward ORR and OER, LSV curves were acquired in a conentional three-electrode device in the O 2 -saturated 0.1 M KOH solution. As observed, Co@NC electrocatalyst has a conspicuous electrocatalytic performance, the onset potential and half-wave potential (E 1/2 ) are 0.89 and 0.81 V, respectively, which both higher than NC and IrO 2 /C (0.84 V, 0.7 V). Compared with the review results in the literature (Table S3), the ORR activity of Co@NC is one of non-precious metal electrocatalysts in alkaline media. The Tafel slope was obtained from ORR polarization curves to evaluate the kinetic features of electrocatalysts, as shown in Figure 3B. The calculated Tafel slope of Co@NC (71 mV/dec) was close to Pt/C (61 mV/dec) and lower than NC (82 mV/dec) and IrO 2 /C (103 mV/dec). The results suggested that the presence of cobalt can improve the degree of graphitization, thus increase the conductivity and decrease the slope of Tafel of Co@NC (Wang et al., 2016).
The catalytic activity of OER is customary assessed by comparing the potential at 10 mA·cm −2 . As shown in Figure 3A, the Co@NC show the OER onset potential (1.47 V) and overpotential (0.5 V), which is comparable to the previously reported catalyst (Table S3). In addition, the Tafel slope was also calculated (Figure 3C), indicating the OER kinetics of the NC, Co@NC and Pt/C. The lower value of Tafel slope means a higher electrocatalysis performance for the OER. The Tafel slope of Co@NC is 110 mV/dec, lower than NC (163 mV/dec) and Pt/C (291 mV/dec), which suggest that the OER kinetics at Co@NC is faster than the other samples . The turnover frequency (TOFs) was calculated to compare intrinsic activity (Supporting Information), the TOFs of the Co@NC and NC are 0.117 s −1 and 0.214 s −1 at an overpotential of 400 mV, respectively, and comparable to the previously reported noble-metal-free electrocatalyst in Table S3. Electrode activity can also be assessed by OER and ORR indicators differences ( E = E j=10 -E 1/2 ), where the E j=10 value is the corresponding OER potential at 10 mA·cm −2 (Meng et al., 2016). Co@NC shows the lowest E value (0.91 V) compared with NC (1.1 V), Pt/C (1.17 V), and IrO 2 /C (0.92 V), confirming the remarkable reversible oxygen electrode characteristics of Co@NC, which directly affects the chargedischarge performance of rechargeable Zn-air batteries (Kashyap and Kurungot, 2018).
To obtain more electrochemical kinetic information, the K-L plots of Co@NC, NC and Pt/C are obtained by K-L equation. Figure 3D shows the K-L plots calculated at 0.5 V at different speeds LSV curves. The K-L plots shows a good linearity between j −1 and ω −1/2 , which indicated that the first-order reaction kinetics is closely related to the concentration of dissolved O 2 . The electron transfer numbers (n) of Co@NC, NC and Pt/C are calculated to be 3.74, 3.67, and 3.91, respectively, suggesting n is close to 4e − pathway for ORR . To further understand the reaction pathways and monitor the formation of peroxide, RRDE tests were carried out accurately calculate the amount of H 2 O 2 generated. The proportion of H 2 O 2 yield ( Figure S10) and the electron transfer numbers ( Figure 4B) of Co@NC, NC and commercial Pt/C were calculated according to disking current density and ring current density ( Figure 4A). In the case of Co@NC electrocatalyst, the proportion of H 2 O 2 yield is lower than 20%, and the n is higher than 3.5, which showed electrocatalytic selectivity was slightly lower than commercial Pt/C, consistenting with K-L plots results. Further suggest that Co@NC ideally performs oxygen reduction via 4e − transfer pathway.
In addition, the stability of Co@NC electrocatalyst is a typical index to evaluate ORR performance, which is of great significance for the safe and stable operation of Zn-air batteries (Park et al., 2016). After accelerating aging process by continuously performing 200 cycles at 10 mV·s −1 , the half-wave potential of Co@NC catalyst was merely 7 mV negative shift and its diffusion current density show no obvious decay (Figure 4C), which indicating significant electrochemical stability of Co@NC electrocatalyst. Comparatively speaking, the half-wave potential of NC catalyst has a negative shift of 10 mV and its diffusion current density decreased significantly.
To better understand the catalytic performance of Co@NC and the role played by N and Co, we further probed the active sites. Transition metals are very sensitive to SCN − ions and can selectively attack the metal centers, impeding their catalytic activity. The ORR activity of Co@NC was re-examined in an electrolyte containing 5 mM KSCN. As shown in Figure 4D, we found that the onset potential, half wave potential and limited diffusion current density of Co@NC decreases in the presence of KSCN, suggesting a strong SCN − ion coordination and blockage of the Co active sites (Amiinu et al., 2018). However, the onset potential, half wave potential and limited diffusion current density of NC was almost unchanged in the presence of SCN − ions. We believe that both Co NPs and N-doped carbon play an important role in enhancing ORR catalytic performance. For the removal of Co NPs, the asobtained material was also immersed in 0.5 M H 2 SO 4 for 24 h. After washing with deionized water and ethanol for three times, the N-doped CNTs were obtained. As shown in supporting information Figure S11, we can know that Co@NC is more electrocatalytically active than N-doped CNTs. This comparison clearly shows that the N-C active site and Co NPs active site existed in Co@NC can significantly improve the catalytic activity (Guo et al., 2018a).
Electrochemical impedance spectroscopy (EIS) showed that the equivalent series resistance of Co@NC was smaller than NC (Figure S12), and there was no high frequency semicircle indicating that charge transfer resistance at the NC/electrolyte and Co@NC/electrolyte interface can be ignored, corresponding to a faster reaction rate. This indicates that the Co@NC has higher conductivity than NC, which corresponds to the Raman results.

Rechargeable Zn-Air Batteries
Superior bifunctional catalysts can accelerate the process of wide application of Zn-air batteries, and thus Co@NC was synthesized to build a typical Zn-air battery. In which, Co@NC electrocatalyst was loaded on conductive carbon paper (0.5 mg·cm −2 ), Zn plate and 6 M KOH containing 0.2 M zinc acetate were used as the cathode, anode and electrolyte, respectively. Figure 5A presents polarization and power density curves of Co@NC, NC, and Pt/C electrocatalyst, Co@NC electrocatalyst exhibited a higher peak power density (137 mW·cm −2 ) and a larger current density (246 mA·cm −2 ) at 0.56 V, both of these are apparently higher than NC and Pt/C. Moreover, the Zn-air battery based on Co@NC electrocatalyst exhibits a large specific capacity of 708.3 mAh·g −1 Zn with a discharge platform of 1.27 V at 10 mA·cm −2 (Figure 5B), which is comparable to the specific capacity of Pt/C (767.1 mAh·g −1 Zn ), but is much larger than the specific capacity of NC (674.9 mAh·g −1 Zn ). By calculation we can also find that the energy density of Co@NC (868.9 mWh·g −1 Zn ) is slightly smaller than Pt/C (892.6 mWh·g −1 Zn ) but higher than NC (829.7 mWh·g −1 Zn ). Two Zn-air batteries were connected in series (Figure 5B), which can provide the required working voltage of the light-emitting diode (LED) and illuminate the rad LED in practical application. The results show that the air cathode of Co@NC can significantly improve the performance of the Zn-air battery than the others, which is attributed to the superior ORR performance of Co@NC catalyst in alkaline medium (Fang et al., 2016b). Furthermore, in the charging process (OER process), Co@NC exhibited higher current density than Pt/C catalyst (Figure 5C), corresponding to the OER test result in Figure 3A.
Durability is an important indicator for future commercial applications of Zn-air batteries. Therefore, galvabostatic charge and discharge battery cycle were test at 10 mA·cm −2 in a 20 h cycle period in Figure 5D. The Zn-air battery with Co@NC electrocatalyst exhibit superior cycling stability. After 6 discharge/charging cycles, the total time exceeds 120 h, and the discharge/charging overpotential slightly changes. Comparatively speaking, the Zn-air battery based on Pt/C requires higher discharge/charge overpotential and <2 discharge/charge cycles (40 h). Besides that, the stability of Zn-air batteries with Co@NC, NC, and Pt/C catalysts were also evaluated at 10 mA·cm −2 with a 10 min cycle period ( Figure 5E and Figure S13). It is worth noting that the charge/discharge overpotential (0.93 V) of Co@NC based Zn-air battery decreased slightly after 1,000 cycles. Which is smaller than the Zn-air battery with Pt/C (1.57 V) and NC(1.1 V). Although the Pt/C catalyst shows higher ORR kinetics, the degradation of the catalyst in charging process leads to a decline in continuous cycle activity. Above all, the as-prepared Co@NC electrocatalyst can be considered as a potential and more environmentally friendly catalyst that can be used as a substitute for noble metal catalysts for rechargeable metal-air batteries.

CONCLUSION
In summary, we developed a facile method for fabrication of nitrogen-doped carbon nanotube encapsulated with cobalt nanoparticles. In this process, DCDA and P123 were rational designed as the precursors and reducing agent to construct carbon nanotube and form cobalt nanoparitcles. The degree of graphitization of nitrogen-doped carbon strengthened in the presence of cobalt nanoparticles. The as obtained Co@NC electrocatalyst showed an excellent ORR and OER bifunctional electrocatalytic performance. In addition, the typical twoelectrode Zn-air battery based on Co@NC electrocatalyst exhibited higher peak power density, large specific capacity and durability. The results of this paper provide a new strategy for the efficient production of nitrogen-doped carbon nanotube encapsulated cobalt. This method can be used in other energy storage devices, for instance, supercapacitors and fuel cells.

AUTHOR CONTRIBUTIONS
LH conceived the study designed the experiments. YL performed the experiments work, analyzed the data, and wrote the manuscript with the help from LH and YZ. HW performed the FESEM testing. PD, CZ, and ML analyzed the TEM data. All authors approved the manuscript for publication.

FUNDING
This work was supported by the National Natural Science Foundation of China (51802134, 51764029, and 51864024).