Development of Highly Active Bifunctional Electrocatalyst Using Co3O4 on Carbon Nanotubes for Oxygen Reduction and Oxygen Evolution

Replacement of precious platinum catalyst with efficient and cheap bifunctional alternatives would be significantly beneficial for electrocatalytic oxygen reduction reaction (ORR) and oxygen evolution reaction (OER) and the application of these catalysts in fuel cells is highly crucial. Despite numerous studies on electrocatalysts, the development of bifunctional electrocatalysts with comparatively better activity and low cost remains a big challenge. In this paper, we report a nanomaterial consisting of nanocactus-shaped Co3O4 grown on carbon nanotubes (Co3O4/CNTs) and employed as a bifunctional electrocatalyst for the simultaneous catalysis on ORR, and OER. The Co3O4/CNTs exhibit superior catalytic activity toward ORR and OER with the smallest potential difference (0.72 V) between the \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${{\boldsymbol{E}}}_{{{\boldsymbol{j}}}_{{\bf{10}}}}$$\end{document}Ej10 (1.55 V) for OER and E1/2 (0.83 V) for ORR. Thus, Co3O4/CNTs are promising high-performance and cost-effective bifunctional catalysts for ORR and OER because of their overall superior catalytic activity and stability compared with 20 wt% Pt/C and RuO2, respectively. The superior catalytic activity arises from the unique nanocactus-like structure of Co3O4 and the synergetic effects of Co3O4 and CNTs.


Experimental Section
The Co 3 O 4 /CNTs were synthesized by a simple chemical method. In a typical preparation, CoCl 2 .6H 2 O (62 mg) was mixed with 30 mL of water. Then, 30 mg of multiwalled CNTs, which were treated with acid for 30 min 45 , were mixed with 30 mL of water. The solution was added into the CoCl 2 .6H 2 O solution and vigorously stirred for 1 h. The pH was controlled by slowly adding a solution of 0.21 mol L −1 NaOH and 0.066 mol L −1 Na 2 CO 3 under vigorous stirring until pH 10 was reached at room temperature (RT). The prepared suspension was kept at 60 °C for next 24 h under gentle stirring. Finally, the Co 3 O 4 /CNTs were filtered and washed with water for and dried at 60 °C. The bare CNTs were prepared using the same protocol but without addition of CoCl 2 .6H 2 O solution. Also, the Co 3 O 4 /CNTs in various pH values (i.e. pH 7,12,14) were prepared by controlled addition of aforementioned alkaline solution. The electrochemical and instrumental characterizations were described in the supporting information. Figure 1 shows that the Co 3 O 4 precursor and clean CNTs were mixed with a simple cooperative assembly of prepared alkaline solution in water at RT. The solution was then kept at 60 °C under gentle stirring for next 24 h to form Co 3 O 4 nanocactus onto CNTs. The growth of Co 3 O 4 nanocactus probably goes through a modified mechanism as below 46  However, the as prepared Co 3 O 4 /CNTs were found with a highly crystalline form. The growth of Co 3 O 4 nanocactus grown on CNTs ( Figure S1) was confirmed by transmission electron microscopy (TEM) analysis in Fig. 2. TEM revealed that the numerous nano-sized Co 3 O 4 cactus were grown onto CNTs (Fig. 2a) and the average size of a single unit of nanocactus was ~25 nm in length with ~5 nm thick sidewall (Fig. 2b). High resolution TEM (HRTEM) showed the crystalline spinel structure of the Co 3 O 4 nanocactus (Fig. 2c) and the lattice spacing of 2.4 Å and 2.8 Å can be assigned to the (311) and (220) planes of typical Co 3 O 4 47 . The bulk elemental component of Co 3 O 4 /CNTs was investigated by energy dispersive X-ray spectroscopy (EDX) in Fig. 2d. The C peak at 0.2 keV was accompanied by an O peak in the EDX spectra. Three Co peaks at ~0.77, 6.9 and 7.63 keV corresponding to CoL α1 , CoL β1 and CoL γ1 , respectively, were also obtained in the EDX spectra. The as-prepared Co 3 O 4 /CNTs consisted of 6.93 wt% Co, 81.93 wt% C, and 11.14 wt% O. Also, Fig. 2 shows bright-fiend TEM image (e) and C (f), Co (g) elemental mapping of Co 3 O 4 /CNTs sample which confirming once again the presence of C and Co elements.

Results and Discussions
The TEM results are consistent with the X-ray diffraction (XRD) data. XRD was performed to investigate the phase structure of Co 3 O 4 /CNTs. In Fig. 3a, several peaks of the pristine Co 3 O 4 were consistent with the standard Co 3 O 4 (ICDD: 98-008-8940, red line). Except for the broad peak (002) at ~25°, which may be ascribed to disordered stacked graphitic structure of CNTs, the major diffraction peaks of Co 3 O 4 /CNTs were in good agreement with those of Co 3 O 4 48-50 . The type IV N 2 adsorption/desorption isotherm curve with a distinct hysteresis loop in the relative pressure range of 0.45-0.99 confirmed the presence of mesopores in Co 3 O 4 /CNTs and bare CNTs samples (Fig. 3b). The Brunaue-Emmett-Teller specific surface area (SSA) for Co 3 O 4 /CNTs was measured to be 373 m 2 g −1 , which was approximately 3-magnitudes higher than the corresponding typical values for Co 3 O 4 -decorated carbon nanomaterials (i.e., Co 3 O 4 /N-rGO, 103.9 m 2 g −1 ; Co@Co 3 O 4 /NC-1, 111 m 2 g −1 ; Co 3 O 4 / CNW-A, 166 m 2 g −1 ) 28,30,32 . On the contrary, the SSA for bare CNTs was 133.2 m 2 g −1 . Barrett-Joyner-Halenda pore size distribution curves confirm the presence of the main mesopores with various sizes between 3 nm and 25 nm (average pore diameter, 6.9 nm) and a pore volume of 1.32 cm 3 g −1 . The average pore diameter and pore volume were much higher than those of bare CNTs (Fig. 3b inset). Therefore, a large SSA, high pore volume, and wide pore size distribution are the clear indication of facile electrocatalysis on Co 3 O 4 /CNTs sample.
X-ray photoelectron spectroscopy (XPS) was performed to elucidate the chemical changes and confirmed the cobalt state during Co 3 O 4 growth on CNTs. The peaks obtained in the XPS spectra at 284.2, 531.0 and 780.7 eV (Fig. 3c) could be ascribed to C1s, O1s and Co2p, respectively, due to the existence of carbon, oxygen and cobalt  52 . Moreover, a tiny shakeup peak was obtained at 292.2 eV for π-π*, signifying higher degree of graphitization 53,54 . The tiny peak at low binding energy of 284.1 eV could probably be attributed to the C-Co bond in Co 3 O 4 /CNTs 55 . The overall elemental composition of Co 3 O 4 / CNTs is listed in Table S1. Furthermore, XPS confirmed the oxidized state of the Co-species with the detection of binding energies of 781.7 eV and 797.6 eV which were attributed to Co2p 3/2 and Co2p 1/2 peaks, respectively ( Fig. 3f) 56,57 . However, the Co 0 , Co 3+ and Co 2+ species were detected at 781.4, 781.6, and 783.8 eV in Co2p 3/2 with corresponding satellite peak (786.7 eV) due to the presence of Co 3 O 4 in the Co 3 O 4 /CNTs sample. At Co2p 1/2 , the Co 3+ and Co 2+ species also appeared at 797.45 eV and 799.6 eV with its shakeup satellite at 803.3 eV. Moreover, the numerical analysis of XPS data was also recorded, and Co was detected as 6.94 wt% with a good ratio of Co 3+ / Co 2+ (1.1) at pH 10 which was the lowest value among all pHs ( Figure S2 and Table S2).  4a) and signify the electrocatalytic ORR performance on all electrodes. The superior electrocatalytic ORR was observed on Co 3 O 4 /CNTs in terms of the improved onset potential (E onset ) of 0.93 V and a half-wave potential (E 1/2 ) of 0.83 V than the CNTs (E onset , 0.83 V and E 1/2 , 0.76 V ( Figure S3). These values were also superior to those of commercially available Pt/C (E onset of 0.91 V and E 1/2 of 0.83 V) and several other reported Co 3 O 4 -based catalysts for ORR 26,58 . Moreover, the current density (j, normalized by electrode area, 0.196 cm 2 ) at the Co 3 O 4 /CNTs electrode was higher than that of the CNTs modified electrode and closer to that of Pt/C. Thus, the Co 3 O 4 /CNTs showed better electrocatalytic activity for ORR in terms of E onset , E 1/2 , and j. This result highlights the importance of the incorporation of nanocactus-shaped Co 3 O 4 with CNTs that have mesoporous structure and higher SSA. The ORR dynamics at the Co 3 O 4 /CNTs electrode were then investigated by RDE, and the results are shown in Fig. 4b. Figure 4b displays a series of RDE curves for ORR using the Co 3 O 4 /CNTs catalyst at various rotation speeds in same electrolyte at 5 mV s −1 scan rate. The obtained data were analyzed using Koutecky-Levich (K−L) equation as follows 58-60 :

Electrochemical ORR on Co
where j, j k , and j L are the measured, kinetic, and diffusion limiting current densities (mA cm −2 ), respectively; n is the electron transfer number per O 2 , and A is the surface area of the working electrode, Moreover, F and T are Faraday constant (96485.3 C mol −1 ) and temperature, respectively; D O 2 and C O 2 are the oxygen diffusion coefficient (1.9 × 10 −5 cm 2 s −1 ) and the bulk concentration (1.2 mM L −1 ), respectively, in 0.1 M KOH 60 ; v is the kinetic viscosity of the electrolyte (1 × 10 −2 cm 2 s −1 ); ω is the angular velocity of electrode (2π*rpm), and k is the electron-transfer rate constant. Based on the K-L equation, a plot of j k −1 vs. ω −1/2 was yield a straight line and the slopes of those plots reflect the B factor in equation (7).
However, Fig. 4b inset shows the K-L plots for Co 3 O 4 /CNTs electrode and the slopes of all K-L plots remain approximately constant over the studied potential range. This result indicates the number of electrons transferred in the ORR remained constant. Based on equations (6) and (7), the average n value in ORR was estimated to be 4, suggesting a four-electron (4e − ) pathway for electrocatalytic ORR [61][62][63] . The ORR dynamics on the CNTs electrode were also investigated by RDE and the average n value in ORR was estimated to be 3.6 from corresponding K-L plots ( Figure S4). The j k obtained from the intercept of the K-L plots for the Co 3 O 4 /CNTs (16.5 mA cm −2 at 0.8 V) was 3.1-magnitudes larger than that of bare CNTs (5.2 mA cm −2 ) catalyst and similar to that of Pt/C (17.3 mA cm −2 ). The ORR activities on as-synthesized Co 3 O 4 /CNTs at various pH values were also investigated ( Figure S5a). Although the Co 3 O 4 /CNTs @ pH 12 has the highest Co 3 O 4 (Table S2), the relatively low Co 3+ /Co 2+ might lead to a high charge-transfer ( Figure S5b). Hence, a relatively better electrocatalytic activity was observed at Co 3 O 4 /CNTs @ pH 10 in based on the higher j k among all pH-dependent Co 3 O 4 /CNTs (Figure S5a inset).
The rotating ring-disk electrode (RRDE) measurement was performed to further evaluate the ORR pathway on Co 3 O 4 /CNTs, bare CNTs, and Pt/C electrodes. The Co 3 O 4 /CNTs electrode exhibited high disk current density (j d ) for ORR and much lower ring current density (j r ) than CNTs. The j r profiles accompanied with further reduction of peroxide species synthesized during ORR process are shown in the upper curves. Both j d and j r from Pt/C are very similar to that of Co 3 O 4 /CNTs.
The RRDE data were used to further verify the transferred electron number and monitor the corresponding H 2 O 2 formation on aforementioned three electrodes during ORR process from equations (9) and (10) 28,64 in Fig. 4d. The average n value for ORR at the Co 3 O 4 /CNTs electrode (3.96) was consistently higher than that at the CNTs (3.6) over the tested potential range of 0.7-0.2 V (vs. RHE). The corresponding H 2 O 2 yields were 3.5% and 9.6% for Co 3 O 4 /CNTs and CNTs, respectively, over the same potential range. The calculated n value (3.98) and H 2 O 2 (2.9%) yield on Pt/C were slightly higher than the Co 3 O 4 /CNTs. The calculated n values are similar to the result obtained from the K-L plots, signifying that the ORR on Co 3 O 4 /CNTs hybrid was mainly by 4e − involved pathway and the main byproduct was H 2 O.
where N is the collection efficiency of the RRDE (0.37), and i d and i r are the disk and ring electrode currents, respectively. The estimated j k values were plotted against the electrode potential to investigate the Tafel behavior of Co 3 O 4 / CNTs, CNTs and Pt/C (Fig. 4e). The better ORR activity on Co 3 O 4 /CNTs was further confirmed by the lower Tafel slope of 63 mV dec −1 at low overpotential (η) than the CNTs (73 mV dec −1 ) and similar to that of Pt/C (61 mV dec −1 ). Furthermore, the LSV curves before and after the accelerated degradation test (ADT) on Co 3 O 4 /CNTs and Pt/C in Fig. 4f. It was found that the E 1/2 shifted largely at the negative direction (21 mV), and j L lost 17.9% on Pt/C after 3000 consecutive cycles. By contrast, the 3.5-and 9-magnitudes lower E 1/2 shift (6 mV) and j L loss (2%) were observed on the Co 3 O 4 /CNTs under the same conditions. Moreover, Figure S6a shows the current density as the function of time by chronoamperometry technique for Co 3 O 4 /CNTs and Pt/C at an applied potential of 0.8 V. The j was maintained up to 93% after 20 h run in real condition, indicating that the catalytic activity on Co 3 O 4 /CNTs could be sustained for a long time. For Pt/C, the catalytic activity was then maintained up to 76% with the same period of time. In addition, the Co 3 O 4 /CNTs electrode demonstrated good methanol tolerance than the Pt/C ( Figure S6b). The TEM image of used Co 3 O 4 /CNTs displays the decay morphology with the crystalline nature of Co 3 O 4 after 20 h of real-time continuous monitoring. These results indicate that the Co 3 O 4 /CNTs are a competent ORR electrocatalyst because of its better electrocatalytic activity, fuel selectivity, and operational stability than the Pt/C.

Electrochemical OER on Co 3 O 4 /CNTs.
To evaluate the potential use of our hybrid catalyst, we employed Co 3 O 4 /CNTs electrode to evaluate the electrocatalytic OER. The OER catalytic activities of all catalysts were studied by LSV at 5 mV s −1 . The Co 3 O 4 /CNTs were used with same mass loading and afforded higher OER activity than either bare bulk Co 3 O 4 , CNTs or RuO 2 in Fig. 5. Figure 5a shows that the E onset for bulk Co 3 O 4 was 1.55 V and the maximum j was 13.6 mA cm −2 at 1.7 V. The E onset for RuO 2 was 1.36 V with maximum of j = 31 mA cm −2 at the same electrode potential. However, considerable negative shifted in the E onset was observed at Co 3 O 4 /CNTs (1.43 V) with highest j = 70.8 mA cm −2 at 1.7 V and the CNTs showed the lowest performance than all electrodes. However, the Co 3 O 4 /CNTs electrode showed a potential of 1.55 V at the current density of 10 mA cm −2 ( ) E j 10 , which was lower than that of bulk Co 3 O 4 (1.68 V) and RuO 2 (1.61 V). Moreover, as shown in Fig. 5a inset, the η required to drive a j 10 for the Co 3 O 4 /CNTs was 280 mV, which was also significantly lower than that for the bulk Co 3 O 4 and RuO 2 (440 mV and 380 mV, respectively). Thus, the Co 3 O 4 /CNTs exhibited higher OER activity than the bulk Co 3 O 4 , CNTs and RuO 2 electrodes in respect to the η and j. This result indicating that the Co 3 O 4 /CNTs have higher density of active sites than bulk Co 3 O 4 and CNTs, and the porous Co 3 O 4 served as the better active catalytic site for the superior OER even better than benchmark RuO 2 which resulted in the synergic effect of Co 3 O 4 and CNTs, the porous nanocactus-like structure, and better Co 3+ /Co 2+ ratio. These characteristics allowed improved ability for electron transfer. The poor current densities of bare CNTs and Co 3 O 4 were probably due to the degradation nature of carbon 32 and aggregation with less conducive nature, respectively 26 .
The Tafel slope of each catalyst beyond the E onset was calculated to understand in detail the OER mechanism and the result are shown Fig. 5b. Tafel plots display a lower Tafel slope of 64 mV dec −1 for Co 3 O 4 /CNTs than those of RuO 2 (71 mV dec −1 ), bulk Co 3 O 4 (73 mV dec −1 ) and CNTs (98 mV dec −1 ), indicating more favorable kinetics toward OER on the Co 3 O 4 /CNTs electrode 32,65 . The Tafel slope of Co 3 O 4 /CNTs was also comparable to other reported Co 3 O 4 -based OER catalysts 26,66 . The OER mechanism can be assumed as follows according to the Tafel slope of Co 3 O 4 /CNTs. The OER on active Co−O-system 65 in Co 3 O 4 /CNTs catalyst was initiated by water adsorption and the formation of adsorbed (ads) reactive intermediate, ⁎ OH abs , by releasing a proton and electron in equation (12). Afterwards, this ⁎ OH abs was converted to another type OH, ⁎ OH abs , in equation (13) (both OH are chemically same but energetically different). In equation (14), a second proton and electron transfer yielded an oxide intermediate, and this step is a rate-determining step (RDS). Recombination of two oxide intermediates completed one reaction turnover in equation (15) abs a bs 2 Furthermore, the electrochemical stability of Co 3 O 4 /CNTs electrode was also compared with RuO 2 under a fixed η and electrolyte conditions. Good stability of Co 3 O 4 /CNTs electrode was confirmed by the similar LSV curves measured at the 1 st and 1000 th potential cycles (Fig. 5c). The used catalyst was characterized by XPS, and the results suggest that amperometric operation did not change significantly in the chemical states except the increase in C-O bond in CNTs and satellite band in cobalt ( Figure S7). These results also suggest that Co 3 O 4 /CNTs have longer stability in electrochemical process than the RuO 2 . The OER processes at Co 3 O 4 /CNTs electrode in 0.1 and 1 M KOH exhibit the same E onset of 1.43 V (Fig. 5c inset). At higher potentials, the more rapid increase in current density was observed for 1 M KOH owing to higher conductivity of the electrolyte. At 270 mV overpotential, j = 22 mA cm −2 was obtained in 1 M KOH solution, while it was 10 mA cm −2 in 0.1 M KOH. The overall oxygen activity of the Co 3 O 4 /CNTs as a bifunctional catalyst could be evaluated (Fig. 5d) by the potential difference (ΔE) between the E j 10 for OER and E 1/2 for ORR 68 . However, the Co 3 O 4 /CNTs catalyst showed the smallest ΔE of 0.72 V and this value was markedly lower than the ΔE obtained using commercial Pt/C (0.85 V) and many other Co-and Co 3 O 4 -based materials i.e., Co-N/G-600, 0.96 V; Co@Co 3 O 4 /NC-1, 0.85 V; Co 3 O 4 /N-Gas, 0.79 V 21,30,67 . This result signifies better reversible oxygen electrode. The detailed comparison with various Co-and Co 3 O 4 -based materials is shown in Table 1. These results clearly indicate that the Co 3 O 4 /CNTs catalyst is a promising low-cost and efficient catalyst for both ORR and OER.

Conclusion
We demonstrated an easy and generic method to synthesize a unique nanocactus-like structure of Co 3 O 4 material embedded onto CNTs for bifunctional electrocatalysis. The newly developed Co 3 O 4 /CNTs were an effective bifunctional ORR and OER electrocatalyst with comparatively better activities and stability than the Pt/C or RuO 2 because of their unique architecture with large surface area, rich active sites, and good electron transfer properties. The excellent catalysis and stability of Co 3 O 4 /CNTs with abundant active sites could be attributed to the strong interaction between the nanocactus-shaped Co 3 O 4 and CNTs. Thus, Co 3 O 4 /CNTs are promising alternatives to noble metal-based catalysts for FCs and water splitting applications because of their low-cost, facile synthesis, and excellent catalysis and stability.