Production of hydrogen from water electrolysis has stimulated the search of sustainable electrocatalysts as possible alternatives. Recently, cobalt phosphide (CoP) and molybdenum phosphide (MoP) received great attention owing to their superior catalytic activity and stability towards the hydrogen evolution reaction (HER) which rivals platinum catalysts. In this study, we synthesize and study a series of catalysts based on hybrids of CoP and MoP with different Co/Mo ratio. The HER activity shows a volcano shape and reaches a maximum for Co/Mo = 1. Tafel analysis indicates a change in the dominating step of Volmer–Hyrovský mechanism. Interestingly, X-ray diffraction patterns confirmed a major ternary interstitial hexagonal CoMoP2 crystal phase is formed which enhances the electrochemical activity.
Generation of hydrogen fuel from water as alternative to fossil oils without releasing carbonaceous gases, such as carbon monoxide and carbon dioxide , has been considered as a promising green technology. Electrolytic hydrogen evolution reaction (HER) has received great attention, because it can be activated from renewable sources of energy like wind and solar [2,3]. However, water electrolysers are hampered by high costs and limited abundance of electrode materials owing to the use of noble metals like platinum. This makes water electrolysers unfavourable to compete with natural gas reforming. It is of high priority to research earth-abundant catalysts as possible alternatives to noble metals [4–6]. Unfortunately, transition metal-based electrocatalysts have suffered from high overpotentials and corrosion problems in acidic media .
Transition metal-based pnictides [8,9] (nitrogen-group elements) are potential electrocatalysts for the HER, as they possess excellent corrosion resistance in the HER condition and good electrical conductivity as an electrode. Very recently, metal phosphides have received great attention owing to their superior catalytic activity towards the HER which rivals platinum catalysts [10–12]. It has previously been reported by Schaak and co-workers [13–15] that orthorhombic CoP and Co2P nanocrystallines produced cathodic current density of 20 mA cm−2 at overpotentials ranging from 85 to 117 mV and stable over 24 h of operation. Bulk hexagonal molybdenum phosphide (MoP) has shown stable HER activity in acidic solution with overpotentials of 140–246 mV for driving a current density of 10 mA cm−2 [16,17]. Even lower overpotentials were reported on amorphous or nanosized MoP HER when compared with bulk MoP [18,19]. It is established that doping Co to MoP increases the intrinsic activity of MoP (so-called Co-promoted MoP); however, the positioning of Co and the exact structural polytype are still not resolved . A decent comprehension of the correspondence between the HER activity and the crystal structure of these materials is not yet thoroughly developed. For bimetal alloys, material design methodologies for optimizing the electronic structure and electrochemical activity are well investigated [21–25]; however, only a few studies were focused on the modification of structure of mixed metal pnictides to accelerate the electrolytic reaction. Cao et al.  reported a layered Co0.6Mo1.4N2 allows the Co to tune the electronic states of molybdenum at the catalyst surface without disrupting the catalytic activity. Staszak-Jirkovský et al.  demonstrated Con+ cations in a CoMoSx chalcogel structure helped accelerate the rate-determining Volmer step. Ternary Chevrel-phase NiMo3S4 was reported by Jiang et al.  that the interconnected [Mo6S8]2 cluster units allow faster charge transfer.
Herein, we show that this research improves upon previous Co-promoted MoP catalysts towards the HER by optimizing the CoP and MoP compositions. We have used a solid-state synthesis route to prepare hybrids of CoP and MoP. Structural studies indicate that at the ratio of Co/Mo = 1, a new crystalline hexagonal CoMoP2 phase is formed in addition to CoP and MoP phases. The hybrid with Co/Mo = 1 exhibits good HER performance and shows promise as an efficient cost-effective cathode material for water splitting.
2. Material and methods
2.1. Synthesis of hybrids of CoP and MoP
Hybrids of cobalt phosphide and molybdenum phosphide were prepared by a two-step thermal treatment of mixtures of cobalt nitrate (Co(NO3)2.6H2O, Acros), ammonium molybdate ((NH4)6Mo7O24.4H2O, Aldrich) and ammonium dihydrogen phosphate (NH4H2PO4, Aldrich). A typical procedure is shown as follows. For preparing Co0.5Mo0.5P catalyst, 10 µmol ammonium molybdate, 70 µmol cobalt nitrate and 140 µmol ammonium dihydrogen phosphate were mixed in 50 ml water and were ultrasonicated until all salts were dissolved. The ratio of cobalt nitrate to ammonium molybdate was varied to create molar ratios [Co]/[Mo] of 0.1/0.9, 0.3/0/7, 0.5/0.5, 0.7/0.3 and 0.9/0.1. After mixing, the transparent solution was dried at 80°C in an oven. The solid mixture was then annealed in a quartz tube furnace with a 100 sccm Ar flow from ambient to 350°C at a rate of 10°C min−1 and then held at 300°C for 1 h. Then the gas flow was switched to H2/Ar mix flow (H2 50 sccm; Ar 50 sccm). The temperature was increased to 800°C at a ramping rate of 15°C min−1 and held at 800°C for 2 h. Bulk CoP and MoP were prepared by the same procedure using corresponding precursors with [Co or Mo]/[P] = 1. To study the effect of annealing temperature, the precursors of Co0.5Mo0.5P were treated at 650°C and 1000°C under H2/Ar mix flow for 2 h after the annealing at 350°C under Ar.
2.2. Structural characterization
The micromorphology of hybrid catalysts were observed on a field emission scanning electron microscope (JOEL JSM−6700F) and on a JEOL JEM-2100 transmission electron microscope. The crystalline compositions of the CoφMoτP hybrids were verified by X-ray diffraction (XRD) patterns with a Bruker D2 PHASER diffractometer. X-ray photoelectron spectra (XPS) were collected on a VG Scientific ESCALAB 250 and the binding energy was calibrated using the C 1s peak at 284.6 eV.
2.3. Electrochemical measurements
Electrochemical measurements were performed in a three-electrode electrochemical cell using a Zahner ZENNIUM E electrochemical workstation. The electrodes for the electrochemical measurements were fabricated as follows. Catalyst ink was prepared by mixing catalyst powder with milli-Q water solution (1 ml for 10 mg of electrocatalyst). In total, 5% Nafion dispersion (DuPont) was added (50 mg solid Nafion for 100 mg of catalyst) to the catalyst slurry. Catalyst coating on glassy carbon electrode with 0.196 cm2 active area was fabricated by drop-casting the catalyst ink on with a micropipette. The catalyst loading was 0.4 mg catalyst cm−2. The electrode was then dried under vacuum at room temperature. The electrolytes for electrochemical measurements were prepared with perchloric acid (Aldrich) and Milli-Q water (Millipore). Pt foil (purity 99.999%) was purchased from Aldrich. All the electrochemical measurements were performed in 0.1 M HClO4 (aq) electrolyte, which was deaerated with hydrogen gas before use. A platinum wire, and an Ag/AgCl reference electrode (3 M KCl) were used as the counter electrode and reference electrode, respectively. All potentials, E, are quoted with respect to reversible hydrogen electrode (RHE). The electrocatalysis was studied using linear sweeping voltage in the range of +0.2 V to −0.5 V (versus RHE). Electrochemical impedance spectroscopy (EIS) was performed in potentiostatic mode at an applied overpotential of 100 mV with frequency from 10 mHz to 0.1 MHz and amplitude of 5 mV. The accelerated deterioration test of electrocatalysts was examined by applying cyclic voltammetry in the potential range from +0.1 to −0.5 V versus RHE for 1000 cycles.
3. Results and discussion
The goal of this investigation is combining Mo with Co in order to tune the properties of their phosphides. Early attempts found enhancement of the HER activity on MoP by the doping of small amounts of Co onto its hexagonal structure . The synthesis was reported to produce hexagonal MoP, while the state of cobalt was not known. In this study, a set of catalysts was prepared with different cobalt-to-all-metal atomic ratios, φ (φ = [Co]/[Co + Mo], τ = [Mo]/[Co + Mo]). CoφMoτP (e.g. Co0.1Mo0.9P) is defined as the sample code of the obtained catalysts; it does not represent any crystalline phase. The CoφMoτP catalysts investigated were Co0.1Mo0.9P, Co0.3Mo0.7P, Co0.5Mo0.5P, Co0.7Mo0.3P and Co0.9Mo0.1P as shown in scheme 1.
We investigated the HER activities of the CoφMoτP catalysts, MoP (the same process for making CoφMoτP) and CoP in 0.1 M HClO4 solution using a typical three-electrode electrochemical cell. The HER activity is compared using the overpotential (η) at 10 mA cm−2 of cathodic current density (η10) and the current density at η = 200 mV. For assessing the electrochemical analysis, in figure 1a, a Pt foil exhibits HER activity with η10 of 50 mV which is comparable to other studies. The polarization curve recorded with CoφMoτP showed good activity for the HER. It is observed clearly in the electrochemical cell that the hydrogen bubbled more vigorously with increasing overpotentials. The η10 of the MoP and the CoP catalysts were read at 250 mV and 283 mV, respectively, which are similar to the overpotentials reported previously . For CoφMoτP catalysts, the polarization curves for the HER in figure 1a changed with the composition. In figure 1b, the overpotential, η10, is plotted versus the cobalt content, φ. The overpotential data of bulk MoS2 are collected from the literature . It is observed that both bulk CoP and MoP show better HER activity than bulk MoS2. MoS2 is known as an edge-active material owing to its unique two-dimensional structure [31–34], while CoP and MoP are active in most facets. In figure 1b, it is shown that on adding Co to the MoP crystal (φ = 0 − 0.5), the η10 decreases (better activity) from 250 to 165 mV, and then it forms a volcano shape with maxima at φ = 0.5 as φ increases to 1.0. This result indicates that when mixing CoP and MoP, the HER activity is enhanced. The Co0.5Mo0.5P catalyst showed an η10 at 165 mV which rivals other bulk catalysts reported. Semimetallic MoP2 nanoparticles showed an η10 of 143 mV . On CoS2 thin film, an η10 of 192 mV was observed by Faber et al. . Very recently, Jiang et al.  demonstrated that Chevrel-phase bimetal sulfide NiMo3S4 possesses good activity with an η10 of 257 mV. Thus, the Co0.5Mo0.5P catalyst presented here is one of the top non-Pt HER catalysts in the bulk form. In figure 1c, the current density at η = 200 mV of the CoφMoτP catalysts showed a volcano shape as a function of the Co content. The Co0.5Mo0.5P catalyst showed an activity of 34.3 mA cm−2 which is 5.9 and 16.3 times the activity of MoP and CoP catalysts, respectively.
To understand how the composition affects the HER activity, XRD patterns were collected from these catalysts. The powder XRD studies in figure 2 evidence the presence of cobalt phosphide and molybdenum phosphide, and their relative peak intensities differ with the various ratios of Co/Mo. For φ from 0 to 0.3, the patterns suggest that the major crystal phase is WC-type MoP (ICSD No: 186 874), with a = 3.25 Å, b = 3.25 Å, c = 3.24 Å, and space group symmetry P6̅m2 (#187). For φ = 0.1, new diffraction peaks appeared at 2θ = 31.3, 39.7, 42.0, and 45.2° (marked as asterisk) with a very low intensity. The intensity of these peaks increases as φ increases and reaches maximum at φ = 0.5. These peaks were confirmed as hexagonal CoMoP2 structure (hereinafter h-CoMoP2 is used to represent this crystal phase in order to distinguish the difference from the hybrid composition Co0.5Mo0.5P) with a = 3.33 Å, c = 11.22 Å, and a space group P63/mmc (ICSD No: 624219, see the electronic supplementary material, figure S1). In addition to MoP and h-CoMoP2 phases at φ = 0.5, minor signals related to MnP-type orthorhombic CoP (ICSD No. 43249) also exist. For spectra with φ = 0.5–1.0, the intensity of CoP peaks increase, but h-CoMoP2 peaks decrease and MoP peaks disappear completely. The XRD pattern of the sample with φ = 0.9 is not shown here because it showed a similar pattern to CoP. The formation of ternary h-CoMoP2 phase was previously explained by the composition of trigonal-prismatic MoP6 and octahedral CoP6 prism to form intermetallic linear –Mo–Co–Mo– chains along the c-axis by Guérin & Sergent . Interestingly, the h-CoMoP2 crystal possesses a similar trigonal-prismatic MoP6 prism to MoP crystal. X-ray photoelectron spectroscopy (XPS) was used to probe the surface electronic properties in the Co0.5Mo0.5P hybrid. The Co 2p spectrum (see the electronic supplementary material, figure S2a) showed two dominating peaks appeared at 777.0 and 791.8 eV corresponding to the Co 2p3/2 and Co 2p1/2 signals, respectively. The minor broad peak at 780 eV indicates the presence of surface oxidized Co species resulting from the contact with air. The Mo 3d spectrum, as shown in electronic supplementary material, figure S2b, showed a strong 3d5/2 peak at 227.8 eV and 3d3/2 peak at 231.1 eV. These peaks are assigned to zero valence state metallic Mo which indicates no or very little oxidation occurs on the Mo site. In the P 2p XPS spectrum shown in the electronic supplementary material, figure S2c, the strong peak at 129.4 eV is assigned to negative charged (metal-Pδ−) phosphide. The minor peak at 133.4 eV is referred to the P–O species. The crystal field model describes that the strong P ligands split the Co 3d states in the octahedral CoP6 prism into high-spin t2g5eg2 ground state. Kibsgaard et al.  have shown that smaller differential hydrogen adsorption-free energies ΔGH of the Co-bridge phosphorus site on CoP surface at a low coverage than that of the P site on MoP surface at a high coverage makes the CoP6 prism a high turnover-frequency towards the HER.
The reaction between the cobalt and/or molybdenum precursors with ammonium hydrogen phosphate under hydrogen environment at 800°C generated uniform phosphide microparticles. The SEM image in figure 3 shows that the CoP and the Co0.5Mo0.5P samples are composed of primary particles, but the MoP contains mostly secondary aggregated particles. These results indicate that the ratio of Co and Mo precursors affects the basic morphology of the catalysts. This can be ascribed to the complicated decomposition process regarding the reduction of MoOx in the high-temperature reduction reaction. The Co0.5Mo0.5P hybrid was characterized by transmission electron microscopy (TEM). The TEM image (see the electronic supplementary material, figure S3) showed interconnected structure composed of large particles (0.3–0.6 µm) which is in good agreement with the SEM result. The energy dispersive X-ray (EDX) spectra (as shown in electronic supplementary material, figure S4) collected on the TEM showed signals from Co, Mo and P elements. The Cu signals resulted from the Cu grid. The atomic percentages of Co, Mo and P elements obtained are 26.9, 24.2 and 48.9%, respectively, which answers to the composition of Co0.5Mo0.5P well.
The effect of annealing temperature on the crystal structure and the HER activity of the Co0.5Mo0.5P catalyst are studied as shown in figure 4. The Co0.5Mo0.5P catalyst annealed at 650oC (Co0.5Mo0.5P-650, pink line in figure 4a) presents diffraction signals composed of h-CoMoP2 (black asterisk) and CoP (red hollow square) crystals. After annealed at 1000°C (Co0.5Mo0.5P-1000, black line in figure 4a), a new orthorhombic CoMoP with a Pnma space group (ICSD No. 2421) is formed (marked as green hollow triangle). The result indicates that increasing annealing temperature to 1000°C induced a phase transformation from h-CoMoP2 to CoMoP, and the excess phosphorus was evaporated. The polarization curves for the HER of Co0.5Mo0.5P-650, Co0.5Mo0.5P-800 and Co0.5Mo0.5P-1000 are compared in figure 4b. The HER activities of these catalysts are in series of Co0.5Mo0.5P-1000 < Co0.5Mo0.5P-650 < Co0.5Mo0.5P-800. The Co0.5Mo0.5P-1000 catalyst showed a poor HER activity with η10 of 338 mV, which indicates that CoMoP structure is less active than the P-rich h-CoMoP2. This can be explained by the higher phosphorus content in the h-CoMoP2 (28.6%) than that in CoMoP (16.7%). Callejas et al.  reported that CoP showed significantly lower overpotential than Co2P to produce the same current density. Xiao et al.  compared the HER activity between bulk MoP and Mo3P and concluded that MoP is more active than Mo3P. Thus, in this study, the h-CoMoP2 with higher P content showed better HER activity.
The dependence of the gradient of logarithm of current density versus potential (so-called Tafel plot) corresponds to the reaction mechanism of the HER. The Tafel curves recorded on the CoφMoτP catalysts (annealed at 800°C) as shown in figure 5a proceed in accordance with the classical two-electron-reaction model. The cathodic hydrogen evolution in acidic conditions involves either Volmer–Tafel (proton discharge followed by the recombination of two bound hydrogen) or Volmer–Heyrovský mechanism (proton discharge followed by electrochemical hydrogen desorption). Tafel slopes were obtained by linear fitting of the low current density region (−0.1 to +0.5 log[mA cm−2]). Figure 5b shows the dependence of the obtained Tafel slopes on the cobalt content, φ. For Pt foil, the Tafel slope of 30.1 mV dec−1 suggests a Volmer–Tafel mechanism that the recombination of two Hads is rate-determining. The data in figure 5b showed decreasing Tafel slopes from 84.0 to 60.5 mV dec−1 for samples from φ = 0 to 0.5. These results suggest that at a small φ (φ = 0, high Tafel slope) hydrogen evolution occurs via a Volmer–Heyrovský mechanism in which slow adsorption of proton dominates the kinetics. For φ = 0.5, the small Tafel slope of 60.5 mV dec−1 indicates that the electrochemical desorption of hydrogen influences the reaction kinetics through the Volmer–Heyrovský mechanism much more than the case of MoP catalyst. Comparing the electrochemical properties of Co0.5Mo0.5P catalyst to bulk CoP and MoP catalysts, the small overpotential and the low Tafel slope of Co0.5Mo0.5P demonstrate that the presence of h-CoMoP2 crystal phase in the catalyst favours proton adsorption kinetics. To further ensure that h-CoMoP2 crystal phase enhances HER activity, we collected polarization curves from three 50/50 mixtures of CoP/MoP powder. The obtained polarization curves of the physical mixtures (see the electronic supplementary material, figure S5) showed HER activity ranging between pure CoP and MoP catalysts. This result indicates clearly that the physical mixture of CoP and MoP powder does not enhance the HER activity, while the h-CoMoP2 crystal phase does.
We carried out EIS analyses on these catalysts to study the charge transfer in the catalytic turnover. The measurements were conducted at overvoltage of 100 mV. The Nyquist plots of the Co0.5Mo0.5P, CoP and MoP samples as shown in figure 6 revealed classical two time-constants characteristics that correspond to a combination of kinetic and diffusion response of the HER on rough electrode surfaces. This two time-constants equivalent circuit is included in the electronic supplementary material, figure S6. The series resistance, Rs, is a sum of all ohmic resistances that the electron passed through from wires to the catalyst surface. The double-layer capacitance, Cd1, in parallel with the charge transfer resistance, Rct, contributes to the charge-transfer process; the capacitance of the catalyst coating, Cd2, in parallel with the resistance of ion conducting paths that develops in the porous surface, Rp. The curves in figure 6 were fitted with the above two time-constant equivalent circuit. The fitting agrees well with the experimental results (see the electronic supplementary material, figure S7). These electrodes showed small series resistances between 2.5 and 4 Ω which indicates good adhesion of the powder on glassy carbon electrode. In figure 6, the pronounced semicircle at low frequencies (high Z’) returns estimates of the charge-transfer resistance, Rct. The Rct of the Co0.5Mo0.5P catalyst (92.0 Ω) is found much lower than CoP (154.1 Ω) and MoP catalysts (107.5 Ω). As reported recently, the Rct of bulk MoS2 was found to be 150 Ω at η = 150 mV . MoS2/RGO has an Rct of 250 Ω at η = 120 mV ; Mo2C nanowires showed a low Rct of 90 Ω at η = 150 mV . Thus, the present Co0.5Mo0.5P catalyst is regarded as a catalyst with efficient charge transfer property. Efficient charge transfer kinetics reflects to the accelerating of the proton discharge step in the Volmer–Heyrovský mechanism and thus lowers its Tafel slope as aforementioned. The enhancement in Rct can be ascribed to the modification of d-band structure owing to the formation of Co–P–Mo linkage.
To assess the stability of the Co0.5Mo0.5P electrodes, we performed the accelerated deterioration experiment by sweeping the applied potential from −0.5 to + 0.2 V versus RHE in 0.1 M HClO4 solution. After 1000 cycles on the Co0.5Mo0.5P electrode, only a slight shift (11 mV increase in η10) in the polarization curve is observed as shown in figure 7a, indicating a good stability of the Co0.5Mo0.5P catalyst in the operating conditions. Faradaic efficiency was determined by performing a controlled potential electrolysis experiment using an H-type cell. The amount of hydrogen gas produced was calculated from the volume of gas at a fixed cathodic current density of 10 mA cm−2 in 0.1 M HClO4 solution for 60 min. The experimentally determined hydrogen quantity was compared to the calculated amount based on the charge consumed as shown in figure 7b. The Faradaic efficiency reached 98% after 60 min of operation. The cathodic current density recorded for driving the HER at overpotential of 165 mV as a function of time is plotted in the electronic supplementary material, figure S8. The current density slightly decreased about 5% after 6 h of electrolysis. These results imply that the Co0.5Mo0.5P catalyst is a highly efficient and cathodically stable HER catalyst in the acidic environment.
We have studied the relationship between the composition of CoφMoτP hybrids and their activity in the HER. The HER activity of the CoφMoτP hybrids showed a volcano shape with a maxima at Co content of 0.5. The Co0.5Mo0.5P catalyst possesses a smaller overpotential (165 mV for driving 10 mA cm−2 of current density) when compared with CoP and MoP and is comparable to bulk non-precious catalysts reported. The hexagonal CoMoP2 phase was found to be responsible for the HER activity in the Co0.5Mo0.5P catalyst. The Tafel analysis showed a change in the dominating step in Volmer–Heyrovský mechanism. The low Tafel slope of the Co0.5Mo0.5P catalyst suggests that the electrochemical desorption of adsorbed hydrogen is the rate-determining step in the HER. The Co0.5Mo0.5P catalyst showed low charge transfer resistance, high stability and high Faradaic efficiency; all demonstrate that the Co0.5Mo0.5P catalyst is a potential, high-performance electrocatalyst for water electrolysis in acidic environments.
Electronic supplementary material including models and fitting of the EIS, XRD, XPS, EDX spectra and polarization curves.
W.-F.C. conceived and designed the experiments. S.-L.F. carried out the synthesis and characterization. S.S. contributed to the SEM analysis and the discussion. L.-C.C. and K.-H.C. contributed to the discussion of the results. W.-F.C. and T.-C.C. contributed to the analysis of the data and the writing of the manuscript.
The authors declare that they have no competing interests.
This work was supported by Ministry of Science and Technology (104-2113-M-002-013-MY2). S.-L.F., T.-C.C. and L.-C.C. are grateful for the support by Academia Summit Program from Ministry of Science and Technology (104-2745-M-002-004-ASP; 105-2745-M-002-003-ASP).
This article has been edited by the Royal Society of Chemistry, including the commissioning, peer review process and editorial aspects up to the point of acceptance.
Electronic supplementary material is available online at https://dx.doi.org/10.6084/m9.figshare.c.3691936.
- Received December 8, 2016.
- Accepted January 30, 2017.
- © 2017 The Authors.
Published by the Royal Society under the terms of the Creative Commons Attribution License http://creativecommons.org/licenses/by/4.0/, which permits unrestricted use, provided the original author and source are credited.