Owing to its high energy density and clean-burning characteristics, hydrogen is an attractive energy carrier for replacing traditional fossil fuels. Electrochemical reactions are potentially effective methods to meet environmental requirements and production demands. Hydrogen production by electrochemical water splitting is therefore considered an ideal choice for meeting energy demands [1]. Electrocatalysts based on noble metals such as Pt and Ru are highly active in water splitting. However, the scarcity and high cost of noble metals present significant barriers to their large-scale practical usage. The overall water-splitting process consists of two parts: the oxygen evolution reaction (OER) for water oxidation and the hydrogen evolution reaction (HER) for proton reduction. The OER is the rate-determining step in water splitting; it involves the transfer of four electrons and is accompanied by the breakage of an O−H bond and the formation of an O-O bond [2, 3]. Generally, HER catalysts operate under acidic conditions [4-8], whereas the OER is generally conducted in alkaline media [9-13]. Therefore, it is difficult to assemble cells with both high HER and OER activities in the water splitting process, due to the different conditions involved in the two reactions.
The development of non-noble metal HER catalysts with high activity for water splitting in alkaline media is therefore an urgent and challenging task. Significant progress has been made in recent years; molecular catalysts [14-17] and various materials (e.g., cobalt phosphide [18-21], nickel phosphide [5, 22, 23], iron phosphide [24], cobalt selenide [25], Pd catalyst [26-29], and CoMoS4 [30]) have been identified as electrocatalysts with excellent water splitting performances in alkaline media. These achievements prompted us to develop better catalysts. Metal-organic framework (MOF)-derived composite catalysts show excellent catalytic properties and display great potential for applications in various fields. MOFs are receiving increasing attention and offer several advantages in catalytic reactions [31-37]. Nitrogen-doped porous carbon (NC) derived from MOFs has a large surface area and hierarchical pore structure, both of which are beneficial for catalytic reactions. Additionally, MOF-derived composites can stabilize nanoparticles for catalytic applications in ambient atmosphere and promote electron transfer during electrocatalytic reactions. Therefore, MOF-derived composites provide superior performances in catalytic reactions [38]. Recently, earth-abundant transition metal-based phosphides (TMPs), e.g., NixPy [39, 40], Cu3P [41], CoxPy [18, 42-44], FexPy [4, 19, 24, 45], and MoPx [46] have demonstrated great potential as alternatives to state-of-the-art noble metal-based catalysts for electronic water splitting. MOF-derived NC/metal phosphide composites could therefore serve as excellent HER electrocatalysts with high activities and stabilities. However, unsatisfactory performances were achieved in previous studies [38, 47-49], and the ineffective electrode assembly methods may be responsible for the low activities observed.
Generally, MOF-derived composites are assembled on conductive substrates by using auxiliary materials such as Nafion, poly(methyl methacrylate), or other polymers [31, 50-54]. These binding agents are acidic or weakly conductive, which affects the activities of the electrocatalysts by reducing their conductivity and blocking the active sites. Controlling the uniform growth of MOFs on the substrates is also challenging because of the rapid production and precipitation of MOF crystals, which may also affect the catalytic activity. Therefore, the synthesis of MOF-derived composites by in situ assembly on a substrate is a challenging task, and a suitable method for assembling MOF-derived composite electrodes is required. Electrophoretic deposition (EPD) is an effective method for assembling electrodes [55, 56], but, to our knowledge, it has never been used for assembling MOF-derived electrodes. As previously reported, metal foam, carbon cloth, or carbon paper (CP) electrodes provide very large supporting surface areas and high non-Faradaic background currents [17, 57-59]. However, it is difficult to assemble carbon materials on metal foam to produce highly stable electrodes; hence, carbon paper was selected as an alternative carbon-based supporting material.
In this study, an MOF-derived NC/Co/CoP/CP material was assembled as shown in Scheme 1. ZIF-67 was synthesized in advance and assembled on CP by the EPD method. The MOF-derived NC/Co/CoP electrocatalyst was formed via carbonization and partial phosphating of Co2+-ZIF-67/CP. The resulting NC/Co/CoP/CP electrode required overpotentials of 208 and 350 mV to achieve a current density of 10 mA/cm2 for the HER and OER, respectively, and is thus clearly superior to traditional electrodes. A water-splitting current density of 10 mA/cm2 was achieved at a low potential of 1.72 V with a catalyst loading of only 0.17 mg/cm2. Although the overpotential of NC/Co/CoP/CP is not comparable to that of the noble-metal catalyst Pt/C, this represents a substantial improvement with respect to the performance of previously reported similar catalysts [47, 49].
The crystal structure and chemical composition were analyzed by X-ray diffraction (XRD) and X-ray photoelectron spectroscopy (XPS). As shown in Fig. 1(A), the XRD pattern of CP was consistent with that of standard graphite (JCPDS No. 41-1487). The strong diffraction peak at 26.4° can be indexed to the (002) plane of graphite. The characteristic peaks of Co/CoP and Co were barely discernible from the baseline, due to their low content in the sample and the strong background signals from CP. The catalyst loading was 0.17 mg/cm2 for Co/CoP, as determined by inductively coupled plasma-atomic emission spectrometry (ICP-AES). The diffraction peaks between 10° and 30° in the XRD pattern of Co2+-ZIF-67/CP largely corresponded to those of ZIF-67 [50], which also confirmed the successful assembly of ZIF-67 on CP. The XRD patterns of the NC/Co/CoP and NC/Co particles, also shown in Fig. 1(A), suggest the coexistence of metallic Co and CoP phases in NC/Co/CoP/CP. The strong diffraction peaks at 44.3° and 51.6° can be indexed to the (111) and (200) planes of Co (JCPDS No. 15-0806), respectively, while the peaks at 31.6°, 36.3°, and 46.2° can be assigned to the (011), (111), and (112) planes of CoP (JCPDS No. 65-2593), respectively. After phosphating, the characteristic diffractions of the CoP phase appeared in the XRD patterns of the NC/Co/CoP samples. The XPS results further confirm the formation of the Co/CoP composite after the low-temperature phosphating of the carbon-coated metallic Co. The results indicate the presence of Co, P, N, and C elements, as shown in Fig. 1(D). The Co peaks of NC/Co/CP at 778.5 and 793.2 eV were attributed to the Co 2p state, consistent with the reported values for Co0 [60]. For the NC/Co/CoP/CP electrode, the Co peaks at 777.6 and 792.9 eV were attributed to Co0; a slight negative shift in binding energy, observed after the formation of the CoP catalyst, can be ascribed to the formation of Co/CoP [47]. The peaks at 782.3 and 798.4 eV were attributed to the Co 2p state in highly-charged CoP. The presence of the peaks at 129.1 and 133.2 eV is consistent with the phosphating of NC/Co/CP to form NC/Co/CoP/CP. The single peak at 133.9 eV indicates the presence of some partially oxidized phosphate species.
Further characterizations by transmission electron microscopy (TEM) and high-resolution TEM (HR-TEM) are shown in Fig. 2 and Figs. S1-S6. The SEM images of Co2+-ZIF-67/CP in Fig. 2(A) and Fig. S1 reveal that regular nano-polyhedra of ZIF-67 were assembled on CP in a uniform and compact morphology. The SEM images of NC/Co/CoP/CP in Fig. 2(B) and Fig. S5 show that the hierarchically-structured carbon-coated Co/CoP composite was embedded on the surface of CP. The HRTEM images of a NC/Co/CoP particle show well-resolved lattice fringes with interplanar distances of 0.24 and 0.28 nm, which were indexed to the (111) and (200) planes of Co, respectively, while a further interplanar distance of 0.17 nm was indexed to the (011) plane of CoP. Furthermore, the elemental mapping images shown in Fig. 2(D) and Fig. S6 reveal phosphorus enrichment in the catalyst, as well as a homogeneous distribution of Co and N.
HER performance tests were carried out in a three-electrode system with NC/Co/CoP/CP as cathode, carbon rod as the counter electrode, and Ag/AgCl as the reference electrode. The alkaline electrolyte was a 1 mol/L KOH solution, which was purged with Ar before electrolysis to avoid interferences from N2 and O2. Figure 3(A) shows polarization curves obtained from linear sweep voltammetry (LSV) measurements, where the geometric current density (j) is plotted against the potential (vs. reversible hydrogen electrode, RHE). The CP sample required an overpotential of 470 mV to drive a current density of 10 mA/cm2, higher than the value of 320 mV obtained for NC/Co/CP. In contrast, NC/Co/CoP/CP showed the highest HER activity, requiring a lower overpotential of 208 mV. Figure 3(B) shows plots of the current density as a function of the overpotential (i.e., Tafel plots). A Tafel slope of 239 mV/dec was observed for the CP sample, while that of NC/Co/CoP/CP was 126 mV/dec. This value was much lower than the slope obtained for NC/Co/CP (181 mV/dec), but higher than that observed for Pt/C (81 mV/dec). The higher current density and lower Tafel slope of NC/Co/CoP/CP suggest a faster HER kinetics. Figure S8 shows the cyclic voltammetry (CV) curves of NC/Co/CP and NC/Co/CoP/CP electrodes, recorded at different scan rates in the non-Faradaic region (0.09-0.22 V) and in KOH electrolyte. The corresponding capacitive currents at 0.165 V as a function of the scan rates, displayed in Figure 3(C), give a double-layer capacitance (Cdl) of 15.22 mF/cm2, greater than 12.33 mF/cm2. In structural terms, the metal and phosphorus atoms in TMPs can act as hydride-acceptor and proton-acceptor sites, respectively, resulting in high electrocatalytic activity for the HER [61].
The introduction of phosphorus can provide a higher density of exposed active sites for the HER. Therefore, the larger active surface area and higher density of exposed active sites could explain the higher catalytic activity of NC/Co/CoP/CP. We then compared the HER activities of binder-free NC/Co/CoP/CP and NC/Co/CoP/CP-Nafion, obtained from LSV curves recorded with the same catalyst loading at a scan rate of 5 mV/s. As shown in Fig. 3(E), the binder-free NC/Co/CoP/CP required an overpotential of 208 mV to achieve a current density of 10 mA/cm2, which was lower than the value of NC/Co/CoP/CP-Nafion (275 mV) assembled with same catalyst loading. Therefore, the facile synthesis method adopted in this work exhibited significant advantages in the assembly of MOF-derived electrodes. Electrochemical impedance spectroscopy (EIS) measurements were then used to investigate the HER kinetics of the NC/Co/CoP/CP and NC/Co/CP electrodes. The corresponding Nyquist plots, displayed in Fig. 3(F), were fitted with the equivalent circuit shown in the inset of Fig. 3(G). In particular, the plots were fitted by a model consisting of two parallel constant phase elements: the one in the high-frequency region is related to the surface porosity (R1), while that at low frequencies corresponds to the charge-transfer resistance at the interface between electrolyte and catalyst (Rct) [62, 63]; the latter is generally used to represent the electrocatalytic activity. As the overpotential increased from 100 to 300 mV, Rct decreased from 146.98 to 3.42 Ω, with nearly the same Rs (approximately 1.21 Ω) and R1 (approximately 4.25 Ω) values. At an overpotential of 200 mV, the Rct value of NC/Co/CoP/CP (29.44 Ω) was much lower than that of the NC/Co/CP electrode (74.27 Ω).
The formation of the Co/CoP catalyst reduced the charge-transfer resistance of the cobalt/carbon composites, which resulted in a faster HER kinetics. So, the higher HER activity of the NC/Co/CoP/CP electrode was attributed to its combination of large active surface area, high density of exposed active sites, and high electrical conductivity, which enhanced the catalytic activity at each active site. As the long-term stability is another important property of HER catalysts, we investigated the durability of the NC/Co/CoP/CP electrode. In particular, stability tests were carried out at a fixed current density of 10 mA/cm2 for 20 h, as shown in Fig. 3(H). The overpotential remained at approximately 215 mV, with only a very slight increase over the duration of the experiment. The LSV curves for the NC/Co/CoP/CP electrode showed a negligible decrease in current density after 1000 CV cycles (Fig. 3(I)).
The OER is the rate-determining step in water splitting, involving the transfer of four electrons. The OER performance was examined in the same three-electrode system used for the HER activity tests, but with NC/Co/CoP/CP as the anode and a carbon rod as the cathode. As shown in Fig. 4(A), the NC/Co/CoP/CP sample exhibited an improved OER performance, with an overpotential of 350 mV to deliver a current density of 10 mA/cm2, much smaller than those of NC/Co/CP (434 mV) and CP (562 mV). As shown in Fig. 4(B), the Tafel slopes of NC/Co/CoP/CP, NC/Co/CP, and CP were 94, 133, and 210 mV/dec, respectively, indicating a favorable OER kinetics.
The OER kinetics was further analyzed via the Nyquist plots. At an overpotential of 400 mV, the Rct of NC/Co/CoP/CP was much lower than that of NC/Co/CP. The long-term stability was also investigated: the LSV results for the NC/Co/CoP/CP electrode (Fig. 4(D)) show a negligible decrease in current density after 1000 CV cycles. The Faradaic efficiency for oxygen evolution during the OER was then evaluated. As shown in Fig. S9, after 10 h of testing, the amount of generated O2 measured by gas chromatography (GC) was 0.905 mmol, together with a passed charge of 360 C (equivalent to 0.9325 mmol of O2 and a 100% Faradaic efficiency). In addition, no characteristic peaks for CO2 caused by the oxidation of carbon materials were detected from GC. The Faradaic efficiency for oxygen evolution was calculated to be higher than 97%; the excess O2 may dissolve in the electrolyte, indicating the occurrence of few side reactions during the electrolysis period. During the stability tests, carried out at a fixed current density of 10 mA/cm2 for 20 h (inset of Fig. 4(D)), the potential remained at approximately 1.58 V. A two-electrode device for water splitting in 1.0 mol/L KOH solution was then fabricated. Compared to NC/Co/CP and CP, NC/Co/CoP/CP exhibited the best performance, with potentials of 1.72, 1.80, and 1.87 V at the current densities of 10, 20, and 30 mA/cm2, respectively.
A binder-free MOF-derived NC/Co/CoP/CP electrode was assembled by EPD and post-processing reactions. The composite electrode showed high catalytic activity and long-term stability, superior to those of traditional electrodes. The electrode required overpotentials of 208 mV for the HER and 350 mV for the OER to achieve a current density of 10 mA/cm2, delivering a water-splitting current density of 10 mA/cm2 at a low potential of 1.72 V. In summary, we reported a novel and practical method for assembling MOF-derived carbon-coated metal/metal phosphide electrodes for electronic water splitting with high stability. This facile synthesis method also provides a promising avenue for assembling metal-doped and multi-metal phase MOF-derived composite electrodes involving foreign metal ions for energy storage and conversion devices, such as supercapacitors, proton-exchange membrane fuel cells, and Li-ion batteries.