Ⅰ. INTRODUCTION

Energy conversion and storage based on electrochemical processes are considered as a sustainable approach to store renewable energy in the form of clean chemical fuels [1]. Oxygen evolution reaction (OER) is one of the key reaction for water splitting and metal-air batteries [2]. OER process is relatively slow and complicated in mechanism, which makes it the rate-determining step in many processes [3, 4]. Currently, the most widely used electrocatalysts for OER are based on noble metal oxides, such as RuO$_2$ and IrO$_2$ [5], but their high cost and scarcity have limited their large-scale applications. Recently, first row transition metals-based electrocatalysts for OER have attracted much interest for their earth-abundance and impressive electrochemical properties, including oxides, hydroxides, sulfides and phosphides of Ni [6-8], Fe [9, 10], Mn [11], Co [12-16, 17]. On the other hand, Cu is one of the most earth-abundant metal elements and has a wide range of accessible oxidation range (Cu$^0$, Cu$^\textrm{I}$, Cu$^{\textrm{II}}$, Cu$^{\textrm{III}}$, Cu$^{\textrm{IV}}$). Therefore, Cu based materials possess a great potential to catalyze various chemical reactions via single and multi-electron transfer pathway, including OER [18-25]. Several CuO-based OER electrocatalysts have been synthesized by thermal oxidation reactions at high temperatures [24]. Formation of a compact film of CuO on Cu surface prevents anodic corrosion under the OER conditions [22]. However, due to the relatively low catalytic activity of CuO for OER [22], the performance of these CuO-based catalysts for OER is typically much lower than that of noble-metal based electrocatalysts (such as IrO$_2$) [24].

One effective way to improve the performance of electrocatalysts is to assemble the active materials into defined micro- and nano- structure on the electrode, such as hollow structure [26] and vertical array structure [27]. These nanostructure can be synthesized by hydrothermal [28], sol-gel [29], and chemical bath deposition methods [30, 31]. But there is often adhesion issue between the formed nanomaterials and the conductive substrate, which limits the loading amount of active materials and the stability of electrode during long time electrolysis [32]. Electrodeposition of active catalysts on electrodes appears as a cheap, safe, and convenient approach to prepare non-noble-metal based electrocatalysts with defined micro- and nano- structures [33, 34]. The effective adhesion and electrical contact between the active materials and the conductive substrate are ensured. The shape, size, and thickness of the active catalyst layer can be controlled with the adjustment of solution concentration and electrochemical deposition parameters. Herein we report an electrochemical approach to synthesize CuO nanocrystal arrays as an efficient and robust electrocatalyst for OER in an alkaline electrolyte, whose performance exceeds that of IrO$_2$. These CuO nanocrystal arrays are deposited on a Cu foam electrode through a simple, safe, and low-cost two-step procedure: Cu$_2$Se nanoparticles are deposited on a Cu foam electrode and then converted into CuO nanocrystal array by electrochemical oxidation. The formation mechanism of the nanocrystal array structure could be similar to that of nanoporous anodic aluminum oxide [35], where the localized dissolution of Cu$_2$Se and the growth of CuO nanocrystal are balanced to form the NCA structure. With the CuO NCA structure, three highly desired properties for high-performance electrocatalyst have been simultaneously achieved: (ⅰ) high mass-loading of active catalysts on Cu foam electrodes without using any binders, (ⅱ) efficient mass transfer and charge transfer through the electrolyte, CuO NCA and Cu foam as conductive substrate, and (ⅲ) efficient oxygen bubble releasing from the hydrophilic and nanostructured CuO NCA surface. For OER, CuO NCA electrode requires overpotentials of 150 mV and 400 mV to attain current densities of 10 mA/cm$^2$ and 100 mA/cm$^2$, respectively, superior to the performance of IrO$_2$. The performance of CuO NCA at a high current density around 270 mA/cm$^2$ is quite stable during a 10-h OER test. These data demonstrate that Cu NCA is an efficient and robust electrocatalyst for OER in alkaline electrolyte, superior to previously reported Cu-based electrocatalysts (see Table S1 in supplementary materials for comparison).

Ⅱ. EXPERIMENTS A. Reagents and materials

IrO$_2$ was purchased from Johnson-Matthey Inc. All the other chemicals were purchased from Sinopharm Chemical Reagent Co. Ltd. All used chemicals were of analytical grade. The deionized (DI) water was used to prepare solutions.

B. Synthesis of CuO nanocatalyst array (CuO-NCA)

Cu foam was repeatedly washed with acetone, distilled water, and 0.5 mmol/L H$_2$SO$_4$ under ultrasonic radiation to remove the oil layer and oxide on its surface. A solution of 50 mmol/L CuSO$_4$ and 50 mmol/L SeO$_2$ was used for the electrodeposition of Cu$_2$Se, and the area of active electrode is kept at 0.16 cm$^2$. Cu$_2$Se nanoparticles with different particle sizes were deposited on the copper foam at different voltages (-300, -450, and -600 mV vs. SCE (SCE saturated calomel electrode)). After the preparation of the Cu$_2$Se precursor, the Cu$_2$Se-coated Cu foam electrode was electrochemically oxidized in 1 mol/L KOH under 700 mV vs. SCE for 1 h. During this process, Cu$_2$Se nanoparticles were oxidized and converted to CuO nanocrystals. The CuO nanocrystal-coated electrode was washed in DI water and then used in the electrochemical test and characterization without any further processing (FIG. 1).

C. Preparation of IrO$_\textbf{2}$ electrode

The IrO$_2$ electrode was prepared by dispersing 10 mg of IrO$_2$ in 2 mL ethanol through sonication for 10 min. A homogeneous catalyst ink was formed. Then, 64 $\mu $L IrO$_2$ ink was loaded on copper foam by drop-casting (Cu foam surface area: 0.16 cm$^2$). Consequently, the mass loading of IrO$_2$ on Cu foam was 2 mg/cm$^2$.

D. Material Characterization

X-ray powder diffraction (XRD) was carried out on a Rigaku D X-ray diffractometer with Cu K$\alpha$ radiation ($\lambda$=1.54178 Å) to confirm the crystalline structure and chemical composition of the materials. The X-ray photoelectron spectra (XPS) were recorded on a Thermos ESCALAB 250 using Al K$\alpha$ ($h\nu$=1486.6 eV) radiation exciting source to collect the detailed information of the surface of the electrode. Field emission scanning electron microscope (FE-SEM, JEOL JSM-6700F) was used to observe the detailed morphology and structure.

E. Electrochemical test

Electrochemical tests were performed with a three-electrode system using a CHI660 electrochemical workstation. A stainless steel sheet, SCE (saturated calomel electrode), and CuO-NCA were used as the counter electrode, reference electrode, and working electrode respectively. The 5 mV/s scan rate was used for Tafel plot and polarization curves. The electrochemical impedance spectroscopy (EIS) was performed with a 100 kHz to 0.05 Hz frequency range and 5 mV sinusoidal voltage. The ohmic drop during electrolysis process was calculated based on the contact resistance $R_\Omega$ that was obtained through EIS data. The $R_\Omega$ values for CuO NCA and IrO$_2$ were found very similar around 0.6 $\Omega$$\cdot$cm$^2$. The $iR$ corrected potential vs. RHE was calculated via the ($E_{\textrm{RHE}}$=$E_{\textrm{SCE}}$+0.234 V+0.0591 pH-$iR_\Omega$) equation. The overpotential for water oxidation was obtained through $\eta_{\textrm{OER}}$=$E_{\textrm{RHE}}$-1.23 V.

F. Calculation of electrochemical double-layer capacitance (${C}_{\textrm{dl}}$)

The electrochemical active surface area of the catalysts was estimated based on the double-layer capacitance ($C_{\textrm{dl}}$) of the catalysts. Cyclic voltammetry (CV) in a potential window where no Faradaic processes took place was used to test the catalysts under different scan rates. The relation between the scan rate ($V$), double layer capacitance ($C_{\textrm{dl}}$) and charging currents $j_\textrm{a}$-$j_\textrm{c}$ is given in equation: $j_\textrm{a}$-$j_\textrm{c}$=2$V$$\cdot$$C_{\textrm{dl}}$. The $C_{\textrm{dl}}$ can be estimated as the slope of a straight line plot of charging currents $j_\textrm{a}$-$j_\textrm{c}$ vs. scan rate. The test was performed in 1 mol/L KOH as electrolyte.

Ⅲ. RESULTS AND DISCUSSION A. Chemical composition

The commercially purchased Cu foam was chosen as the substrate due to its low cost, excellent conductivity and porous structure. The 3D porous structure of Cu foam can provide a large loading of CuO nanocrystals and enough interspaces for the release of generated gas bubbles. The synthesis method of CuO nanocrystals has been illustrated in FIG. 1. Cu$_2$Se was deposited on Cu foam through a cathodic electro-deposition in the aqueous solution of CuSO$_4$ and SeO$_2$. The identity of Cu$_2$Se was confirmed by power X-ray diffraction (XRD) pattern (FIG. 2(a)), which indicates a tetragonal phase Cu$_2$Se (JCPDS No.46-1129). The X-ray photoelectron spectroscopy (XPS) spectrum (FIG. S1 in supplementary materials) shows the characteristic peaks of Cu and Se. The two peaks at 932.3 eV and 952.2 eV with no shoulder peaks in Cu 2p spectrum (FIG. 2(c)) correspond to Cu(I) species. The two broad peaks at 59.4 eV and 54.3 eV in the Se 3d spectrum (FIG. S1 in supplementary materials) correspond to Se 3d$_{5/2}$ and Se 3d$_{3/2}$, which are the characteristic peaks of Se$^{2-}$.

The Cu$_2$Se-coated Cu foam electrode was electrochemically oxidized (0.7 V vs. SCE) in 1 mol/L KOH electrolyte to yield a product, which was identified by X-ray diffraction (XRD) (FIG. 2(b)) to be CuO (JCPDS No.45-0937). The XRD patterns before and after the electro-oxidation clearly show that the dominant component of the deposited precursor Cu$_2$Se has been converted to CuO, as shown in FIG. 2 (a, b). The characteristic satellite peaks in Cu 2p XPS spectrum also indicates that Cu atoms have been oxidized to Cu$^{\textrm{II}}$ (FIG. 2(c)). According to the XPS results, only tracing amount of Se was found in the form of selenide oxide on the surface of CuO nanocrystal after the electro-oxidation (FIG. S1 in supplementary materials), which also indicates the conversion from Cu$_2$Se to CuO. The XPS spectrum of O 1s also provides information about valance state of Cu, the peak at 529.4 eV is possibly from O element in CuO, and the peak at 532.1 eV indicates the chemisorbed water (FIG. S2 in supplementary materials) [36].

B. Control of the morphology of the CuO NCA

By adjusting the voltage for electrodeposition, the size of Cu$_2$Se nanoparticles deposited on the Cu foam can be easily controlled [37]. After a systematic exploration, we have found that the Cu$_2$Se nanoparticles deposited at the voltage of -450 mV vs. SCE are a layer of pyramid-like nanoparticles that are uniformly and densely packed on the copper foam substrate (FIG. 3(a)). The in situ electro-oxidation of Cu$_2$Se nanoparticles results in the ordered flower-like CuO NCA with an average crystal diameter of 200 nm (FIG. 3(b)). The Cu$_2$Se nanoparticle deposited at a higher voltage (e.g. -300 mV vs. SCE) had a smaller particle size than that obtained at the voltage of -450 mV vs. SCE (FIG. 3(c)). This Cu$_2$Se nanoparticle with smaller size could also be oxidized to form CuO nanocrystals with smaller size (10-30 nm in crystal diameter), which showed a loosely packed porous structure (FIG. 3(d)). On the other hand, the Cu$_2$Se nanoparticle deposited at a lower voltage (e.g. -600 mV vs. SCE) had a much bigger particle size than that obtained at the voltage of -450 mV vs. SCE, where some big size polyhedral Cu$_2$Se particles can be observed (FIG. 3(e)). The obtained CuO nanocrystals from this bigger Cu$_2$Se nanoparticles are also bigger in size ($>$400 nm in diameter, FIG. 3(f)). This porous structure of CuO NCA could leave a large number of active sites exposed to the electrolyte solution and allow fast diffusion of solvents and ions through the micropores. The crystal size and morphology of three different CuO NCA materials prepared from different Cu$_2$Se precursors indicate the activity and stability of these CuO NCA electrodes for OER, as discussed below.

C. OER activity of CuO NCA electrodes

The OER activity of as-prepared CuO NCA electrodes was evaluated in 1.0 mol/L KOH with the Tafel method under a scan rate of 5 mV/s (FIG. 4(a)). As expected, the CuO NCA sample prepared from the Cu$_2$Se precursor deposited at -600 mV vs. SCE showed the lowest OER activity, due to the large size of CuO nanocrystals. The other two CuO NCA samples prepared from the Cu$_2$Se precursor deposited at -300 mV and -450 mV vs. SCE showed similar OER activities. However, the CuO NCA prepared from the Cu$_2$Se precursor deposited at -300 mV vs. SCE showed a significant pseudocapacitance in the range of 1.3-1.6 V on the polarization curve, possibly due to small size of CuO nanocrystals. The contact resistance of this small size CuO NCA (0.82 $\Omega$$\cdot$cm$^2$) was higher than the other two CuO NCA materials ($\sim$0.6 $\Omega$$\cdot$cm$^2$) (FIG. S3 in supplementary materials). Therefore, the optimal CuO NCA material was the one prepared from the Cu$_2$Se precursor deposited at -450 mV vs. SCE. And this optimal CuO NCA was used for the following tests.

A bare Cu foam, IrO$_2$ loaded on Cu foam, and the optimal CuO-NCA were tested under the same condition for comparison (FIG. 4(b)). As expected, Cu foam showed a very low activity for OER in alkaline solution. Although the performance of previous CuO-based OER catalysts is not as good as that of IrO$_2$ [24], CuO NCA showed a much better performance than IrO$_2$. The CuO NCA electrode can reach 100 mA/cm$^2$ with an overpotential of 400 mV, while most of the Cu based OER catalyst can only reach 10 mA/cm$^2$ at the similar overpotentials (Table S1 in supplementary materials). And the CuO NCA can reach 10 mA/cm$^2$ with an overpotential of only 150 mV. On the other hand, the IrO$_2$ on Cu foam required a 570 mV overpotential to achieve a current density of 100 mA/cm$^2$, while CuO-NCA can drive the same current density at an overpotential of 400 mV, which is much lower than that of IrO$_2$ (FIG. 4(b)). The Tafel slope of CuO NCA was 98 mV/dec, which was similar to that of IrO$_2$ (84 mV/dec) (FIG. 4(c)), also indicating the high activity of CuO NCA electrode for OER. Based on the Nyquist plots (FIG. 4 (d, e)), the contact resistance of CuO NCA (0.59 $\Omega$$\cdot$cm$^2$) and IrO$_2$ (0.57 $\Omega$$\cdot$cm$^2$) are similar to each other, which indicates that both catalysts have a good electrical contact with the Cu foam.

For many OER catalysts, the generated O$_2$ bubbles often stick to the electrode surface, hence limit the mass transport and reduce the active surface area [38]. The electrode rotation and modification of the electrode surface can be used to partially solve this problem [39]. The CuO NCA possesses a 3-dimensional flower-like structure with the hydrophilic nature of CuO, which can help the quick release of O$_2$ as tiny bubbles from the electrode surface. Thus, the CuO NCA electrode can drive high current densities at low overpotentials without stirring the solution or rotating the electrode, which makes it practically favorable for water oxidation.

We also test the stability of CuO NCA during a long-term OER test at a large current density. An electrolysis voltage of 700 mV vs. SCE was applied on both CuO NCA and IrO$_2$ electrodes in 1 mol/L KOH electrolyte to guarantee that the catalyst can work under a large current density. As shown in FIG. 4(f), the CuO NCA presents a superior performance as compared to IrO$_2$. The CuO NCA can easily reach 280 mA/cm$^2$ at the start. After 1 h, the current density reached 300 mA/cm$^2$ and remained at a high current density above 270 mA/cm$^2$ throughout the 10-h electrolysis test. In contrast, IrO$_2$ can only reach 150 mA/cm$^2$ at the beginning, and the activity shows a fast decline to 40 mA/cm$^2$ in 1 h and remained at a low current density $\sim$25 mA/cm$^2$ throughout the 10-h electrolysis test. The comparison of Cu NCA with IrO$_2$ clearly reveals the superior stability of Cu NCA as electrocatalyst for OER in alkaline electrolyte.

Theoretically, IrO$_2$ possesses better OER activity than CuO NCA, whereas the experimental results present an opposite conclusion. This is might be due to the limited mass loading of IrO$_2$ on the substrate (2 mg/cm$^2$). At a higher mass loading of IrO$_2$, the catalytic activity of IrO$_2$ compromises, because the electronic connection between IrO$_2$ nanoparticles and the Cu foam substrate is not sufficient. In addition, the activity of IrO$_2$ decreases during the long term OER test is due to the nanoparticles aggregate [27]. In contrast, the CuO NCA flower-like structure is firmly connected to Cu foam, so their nanoparticles will not aggregate throughout the long-term electrolysis. The robust connection between the CuO nanocrystals and Cu foam electrode also enables a high mass loading of CuO NCA. The excellent OER activity of CuO NCA is also because of its flower-like morphology, which could provide a larger surface for catalysis. The electrochemical double layer capacitance method ($C_{\textrm{dl}}$) [40] was used to estimate the active surface area of CuO NCA electrode. The high $C_{\textrm{dl}}$ electrochemical active surface of 327 mF/cm$^2$ as compared to Cu foam (5.8 mF/cm$^2$) was credited to flower-like morphology and the high mass loading of CuO NCA on Cu foam (FIG. 5). The hierarchical nanostructure, the high mass loading, and the efficient electrical contact with the Cu foam account for the high OER activity of CuO NCA electrode.

Ⅳ. CONCLUSION

In summary, we present a general strategy of fabricating a CuO-based nanocrystal array as highly active electrocatalysts for OER in alkaline electrolytes. The two-step sacrifice-template method efficiently builds up a 3-D hierarchical CuO NCA with flower-like structure that is firmly connected to the Cu foam as a highly conductive substrate. Owing to the hierarchical nanostructure, the high mass loading and the efficient electrical contact with the Cu foam, the CuO NCA needs a low overpotential of 400 mV to drive a high current density of 100 mA/cm$^2$ and presents a great stability at a heavy-loading long-term OER test, whose performance is much better than that of IrO$_2$, and also superior to other Cu based electrocatalysts for OER (Table S1 in supplementary materials). The high activity of CuO NCA is well retained during a 10-h OER test at a high current density around 270 mA/cm$^2$, which is about 10 times higher than the current density of IrO$_2$ ($\sim$25 mA/cm$^2$) with the same applied overpotential. In general, this facile sacrifice-template strategy may help in the modification and enhancement of many other electroactive materials. Cu based catalyst has always been considered as a good candidate for the electrochemical capacitor and catalyst for CO$_2$ reduction [41]. Therefore, our CuO catalysts with promising activity and stability may be used in many other areas as an efficient electrocatalyst.

Supplementary materials: FIG. S1 and FIG. S2 present the change of the XPS spectrum of the CuO-NCA and Cu$_2$Se precursor. $R_\Omega$ value of the CuO-NCA deposited under different voltage can be found in FIG. S3. The OER performance of CuO-NCA stands out in the Cu-based OER electrocatalysts, which can be found in the Table S1 in the supplementary materials.

Ⅴ. ACKNOWLEDGMENTS

This work was supported by the National Natural Science Foundation of China (No.21474094 and No.21722406). Muhammad Imran Abdullah acknowledges the Chinese Academy of Science (CAS) and TWAS for supporting him for a Ph.D. degree from University of Science and Technology of China in the category of 2016 CAS-TWAS President's Fellowship Awardee (Series No.2016-171).