Catalyst characterization. The Ni-Co-P heterostructure nanoarrays with CoP nanoparticles uniformly distributed on the surface of NiCoP nanowires were in situ synthesized on nickel foam (NF) by ion exchange method as shown in Fig. S1, in which the NF is corroded and oxidized to Ni (II) during the hydrothermal synthesis to form Ni-Co precursor nanoarrays on NF (Ni-Co-Pre/NF), and then the Ni-Co-P/NF was synthesized after phosphating by using NaH2PO228,29.
The Ni-Co-Pre/NF shows a nanowire array morphology of 900 nm in an average length on NF (Fig. 1a, Fig. S2). With the increase of Co source addition amount from 3 to 5 mmol, the nanowire arrays of Ni-Co-Pre/NF gradually accumulate into a sea urchin shape (Fig. S3a and S3b), which adhere on the NF substrate loosely and will easily fall off in subsequent treatment. 3 mmol cobalt source addition amount is optimal, at which the precursor loading on NF achieves maximum according to the XRD patterns and the digital photographs (Fig. S3c). After phosphating, sample Ni-Co-P/NF at the optimal Co source addition of 3 mmol retains the dense and uniform 3D nanowire morphology of precursor (Fig. 1b, Fig. S4), while the surface becomes rougher because of a large number of nanoparticles formed on the surface of nanowires (inset in Fig. 1b). Moreover, the nanoarrays of Ni-Co-P/NF are hydrophilic featuring a much lower water contact angle (36.14°) than that of Pt/C (139.98°) (Fig. S5), which is favorable for the release of absorbed bubbles on discontinuous solid-liquid-gas triple phase contact points30.
In the TEM and high-resolution TEM (HRTEM) images (Fig. 1c and 1d), two different morphologies in Ni-Co-P (the sample separated from NF) were observed: randomly oriented particles from nanowire under ultrasound (area 1) (Fig. 1d) with the selected area electron diffraction (SAED) pattern being assignable to CoP with (011) (211) (013) planes (Fig. 1d illustration), and nanowire main phases (area 2) (Fig. 1e). The energy dispersion spectrum (EDS) (Fig. S6) indicates the component ratio of area 1 is Co:P = 1:1 and that of area 2 is Ni:Co:P = 1:1:1, further confirming the existence of CoP in area 1 and nickel cobalt phosphide (NiCoP) in area 2. While the edge area of the nanowire (Fig. 1e) is Co-rich and Ni-deficient besides P signal according to the element linear scanning profile in Fig. S7, evidencing the existence of CoP nanoparticles outside of the NiCoP nanowires. In addition, the oxygen signal can also be detected, indicating the partial oxidation of the outer layer into an amorphous phosphorus oxide layer31. Two sets of interplanar spacing of 0.202 nm and 0.285 nm can be detected in HRTEM (Fig. 1f), which are in correspondence to the (201) plane of NiCoP and (011) plane of CoP, respectively, indicating the successful synthesis of Ni-Co-P heterostructure nanoarrays with CoP nanoparticles uniformly distributed outside of the NiCoP nanowires in Ni-Co-P/NF (Fig. S8). The element mappings in Fig. 1g demonstrate the homogeneous dispersion of Co, Ni, P on nanowires, while Co is slightly wider than that of Ni, further indicating successful loading of CoP nanoparticles on the NiCoP nanowires surface.
From the XRD patterns in Fig. 2a, the strongest diffraction peaks of Ni-Co-P/NF centered at 40.99°, 44.90° (covered by nickel peaks), and 47.58° can be indexed to the (111), (201), (210) planes of standard hexagonal NiCoP (JCPDS No. 71-2336), and those at 31.59°, 36.32°, 48.12° assigned to the (011), (111), (211) planes of CoP (JCPDS No. 29–0497), further verifying the formation of Ni-Co-P heterostructure nanoarrays on NF with CoP nanoparticles distributed outside of the NiCoP nanowires32.
The survey scan of XPS spectrum of the Ni-Co-P/NF indicates the existence of Co, Ni, P, and O element (Fig. S9). Co 2p XPS spectra (Fig. 2b) of Ni-Co-P/NF exhibits two major groups with binding energies of 778.7 and 793.5 eV, respectively, representing Co-P, which are close to those of metallic Co (778.2 and 793.3 eV), indicating the presence of partially charged Co species (Coδ+, δ is close to 0). While the relatively weak bands at 781.8 and 797.6 eV belong to the Co-POx resulting from the Co oxidation state, and the 785.6 and 802.5 eV are the satellite peak of Co. Similarly, there are three groups of peaks appear in the Ni 2p XPS spectra (Fig. 2c), and the binding energies at 853.6 and 870.8 eV represent Ni-P (Niδ+), and those at 56.8 and 874.7 eV belong to Ni-POx, and the 861.3 and 879.6 eV are assigned to the satellite peak of Ni33–36. Compared to CoP, the Co 2p3/2 and 2p1/2 of Ni-Co-P/NF slightly shift toward the lower binding energies (Fig. 2b, 2c, Fig. S9), while the Ni 2p3/2 and 2p1/2 of Ni-Co-P/NF have a higher binding energy shift compared NiP/NF, indicating an increased ionicity of M − P bond in bimetallic phosphides and a promoted electron migration from metal to phosphide37, which makes the electronic structure of the Ni-Co-P more conducive to the HER and HzOR. The deconvoluted P XPS peaks (Fig. 2d) at 128.3 and 130.2 eV belong to P 2p1/2 and P 2p3/2 of MP, which is lower than that of elemental P (130.0 eV), suggesting that the P is partially negatively charged (Pδ−), and the peak at 133.5 eV could be ascribed to a metallic oxidation state, which is associated with MPOx due to exposed to the air38. In addition, the binding energies of P in MP and MPOx in Ni-Co-P/NF also shift to the lower position compared with those of NiP/NF and CoP, indicating the electron-rich character of P. Thus, the P can trap positively charged protons during electrocatalysis, which is responsible for the HER activity39.
X-ray absorption fine structure (XAFS) spectrum was obtained to investigate the local electronic structure and atomic arrangement of Co K-edge and Ni K-edge. In the normalized X-ray absorption near edge structure (XANES) (Fig. 2e), the absorption edge energy of Co K-edge in Ni-Co-P/NF is in between those of CoO and Co foil (Fig. 2g-left), and similar to that of CoP, indicating that the average value state of Co element is in between 0 and + 2, consistent with the XPS results35. Similarly, the absorption edge energy of Ni K-edge of Ni-Co-P/NF gives the average value states of Ni between 0 and + 2 (Fig. 2f), slightly higher than that of NiP according to Fig. 2g-right, suggesting the changed binding energy between Ni and P due to the electronic interaction between Co and Ni. The Fourier-transformed extended X-ray absorption fine structure (EXAFS) spectrum of Ni-Co-P/NF shows the similar radial distribution function to those of CoP and NiP as shown in Fig. 2h and Fig. 2i, respectively. Relative to the reference CoP, a similar peak at the 1.66 Å in Ni-Co-P/NF is assigned to Co − P bond because of the existence of CoP species in Ni-Co-P/NF (Fig. 2h). Notably, the Ni − P peak in Ni-Co-P/NF positively shifts to 1.75 compared to 1.69 Å in NiP reference (Fig. 2i), due to the Co introduction. All above results demonstrate the interaction between Ni and Co atoms in NiCoP lattice and the coexistence of CoP and NiCoP components.
Electrocatalytic HER performance. The HER electrocatalytic activity of the as-synthesized catalysts were investigated in N2-saturated 1.0 M KOH electrolyte, which was calibrated by reversible hydrogen electrode (RHE) as shown in Fig. S10. From the linear sweep voltammetry (LSV) curves in Fig. 3a, sample Ni-Co-P/NF requires a much lower overpotential (37 mV) to reach 10 mA cm− 2 compared to references NiP/NF (124 mV), Ni-Co-Pre/NF (175 mV), and CoP (173 mV). Though the overpotential of Ni-Co-P/NF below 150 mA cm− 2 is slightly higher than that of the Pt-C/NF (η10 = 28 mV), but become substantially lower than that of Pt-C/NF at current density above 150 mA cm− 2. Importantly, the Ni-Co-P/NF catalyst needs only 280 mV to reach 1000 mA cm− 2, which is better than most reported catalysts (Table S1). A Tafel slope of 33.3 mV dec− 1 was recorded on Ni-Co-P/NF (Fig. 3b), much smaller than those of NiP/NF (84.0 mV dec− 1), Ni-Co-Pre/NF (129.3 mV dec− 1), and CoP (72.6 mV dec− 1), though slightly larger than 23.6 mV dec− 1 of Pt-C/NF in the potential range40, suggesting the Tafel mechanism dominated the reaction on Ni-Co-P/NF.
The electrochemical resistance spectra (EIS) of the catalysts are shown in Fig. 3c, and the resistance of Ni-Co-P/NF (13.09 Ω) is much lower than those of CoP (57.3Ω) and NiP/NF (45.9Ω) though slightly higher than that of the Pt-C/NF (10.73 Ω) according to the fitting data in Table S2. Since the electrochemical double-layer capacitance (Cdl) of the catalyst is proportional to its electrochemical active surface area (ECSA) directly related to catalytic activity, Ni-Co-P/NF shows the Cdl value of 20.6 mF cm− 2 (Fig. 3d and Fig. S11), much higher than Ni-Co-Pre/NF (10.6 mF cm− 2), NF (2.2 mF cm− 2), and Pt-C/NF (2.2 mF cm− 2), indicating the highest ECSA of Ni-Co-P/NF heterostructure nanoarrays than others.
In addition, the Ni-Co-P/NF shows excellent HER catalytic stability, which can be operated stably for 90 h at -50 mV with a limited current attenuation by 6.7% at the end of test (Fig. 3e). The accelerated durability tests (ADTs) in Fig. S12 show no significant change in LSV curves after 10000 cycles, further confirming the excellent stability of Ni-Co-P/NF toward HER. Furthermore, the influence of Co addition on the HER performance of Ni-Co-P/NF composite was also tested (Fig. S13), and the Co-3 with Co(NO3)2·6H2O addition of 3 mmol during the synthesis (as named Ni-Co-P/NF above) is optimal for HER catalytic activity. Moreover, the volume of H2 production was collected by gas chromatography (GC) and compared with the theoretical gas volume obtained from the number of transferred electrons, which gives a HER Faradaic efficiency (FE) of 97.2% on Ni-Co-P/NF (Fig. S14).
Electrocatalytic HzOR performance. The HzOR electrocatalytic activity of the as-synthesized catalysts was tested in an aqueous electrolyte containing 1.0 M KOH and 0.1 M N2H4 using a typical three-electrode system. Figure 4a exhibits the LSV curves with 90% iR compensation at the scan rate of 50 mV s− 1. Compared to NiP/NF (20 mV), CoP (-3 mV) and Ni-Co-Pre/NF (260 mV), sample Ni-Co-P/NF requires a much lower potential (-61 mV) to reach a current density of 10 mA cm− 2, which is superior to the Pt-C/NF (34 mV). Exceptionally, the ultralow working potentials of -24, 78 and 176 mV are required for Ni-Co-P/NF heterostructure nanoarrays to reach current densities of 100, 500 and 1000 mA cm− 2, respectively. From Fig. 4b, Ni-Co-P/NF shows a small Tafel slope of 22.9 mV dec− 1, largely lower than those of Pt-C/NF (33.2 mV dec− 1), NiP/NF (49.5 mV dec− 1) and CoP (92.5 mV dec− 1), indicating its fast catalytic kinetics for HzOR. The excellent HzOR catalytic activity of Ni-Co-P/NF is connected with its much smaller electrochemical resistance. In the simulated equivalent circuit diagram in the inset of Fig. 4c, Rs, Rf and Rct respectively represent the solution resistance, high-frequency semicircle resistance and the charge-transfer resistance, and the QCPE1 and QCPE2 were introduced to simulate the double-layered capacitor at catalyst interface and the electrode/electrolyte interface, respectively41. According to the result of EIS and model fitting (Table S3), Ni-Co-P/NF shows the lowest Rct (2.614 Ω) and Rf (0.990 Ω) values owing to its three dimensional array structure having greatly increased contact area at gas-solid-liquid three-phase, in which the N2H4 molecules adsorbed on the catalyst surface and electrolyte can be oxidized immediately and continuously, meanwhile the electrons generated can be efficiently transferred through conductive NF.
Figure 4d gives the comparison between HzOR by Ni-Co-P/NF (in 1.0 M KOH and 0.1 M N2H4) and OER by RuO2 catalyst (in 1.0 M KOH). An ultralow potential of 90 mV is required for Ni-Co-P/NF to reach an elevated current density of 200 mA cm− 2, which is 1.54 V lower than that of RuO2 catalyst (1.63 V) at the same current density, suggesting the significantly reduced energy consumption for hydrogen production by replacing OER with HzOR. To further investigate the HzOR kinetics for Ni-Co-P/NF, varied scanning rates were adopted. It can be seen from Fig. 4e that the current changes in LSV curves are negligible at the varied scanning rates from 10 to 50 mV s− 1, suggesting the efficient charge and mass transfer at the electrode-electrolyte-gas three-phase interface on Ni-Co-P/NF, and resultantly, the fast HzOR kinetics. The Ni-Co-P/NF with Co source addition of 3 mmol shows the best HzOR performance (Fig. S15), coincident with the best HER activity. Moreover, the LSV curve of Ni-Co-P/NF after 10000 cycles shows no significant shifts compared to the initial cycle according to the ADTs (Fig. 4f), indicating its excellent catalytic stability. From the i-t test of Ni-Co-P/NF toward HzOR (Fig. 4g), a large current density above 100 mAcm− 2 can be retained, which is 91% of the initial value at the end of 100 h. Furthermore, the phase structure, composition and morphology of the Ni-Co-P/NF catalyst after stability test (name as Ni-Co-P/NF-used) remain almost unchanged after the test according to the XRD patterns (Fig. S16), Raman spectra (Fig. S17)42 and TEM images (Fig. S18), further confirming the excellent HzOR catalytic stability of Ni-Co-P/NF heterostructure nanoarrays. The HzOR activity parameters of Ni-Co-P/NF, such as the potential to reach 100 mAcm− 2 and Tafel slope in this study, are even better than those of the most reported non-noble metal-based catalysts (Fig. 4h, Table S4).
HzOR mechanistic insight. Based on above results, the HzOR catalytic mechanism of the obtained catalyst was investigated. The CV measurements on Ni-Co-P/NF electrode were firstly carried out in the voltage range of -0.2 V to 0.4 V in 1 M KOH and − 0.2 V to 0.8 V in 1.0 M KOH + 0.1 M N2H4 to determine the active sites of bimetallic phosphides and the potential changes on HzOR (Fig. 5a). A pair of redox peaks, A1 and B1 on the dotted blue CV curve, corresponds to the conversion between MPOx (the valence state of M is in between + 2 and + 3) and MP (M: δ+, δ is close to 0) in the 1 M KOH, indicating the surface oxidation/reduction of the metal phosphide in the anodic oxidation range as shown in Eq. (1).
$$MP+{OH}^{-}\leftrightarrow {MPO}_{x}+{H}_{2}O+{e}^{-}$$
1
When adding 0.005 M N2H4 in electrolyte, a new oxidation peak A2 appears in the CV curves, which is related to the preferential oxidation of N2H4 to the MP oxidation. The area of oxidation peak A2 decreases with the increasing cycling number due to the decreased N2H4 concentration, while that of oxidation peak A1 increases gradually because of the consumption of local N2H4 resulting in the following oxidation of the catalyst itself into MPOx. In particular, the reduction peak (B1) disappears in N2H4-containing electrolyte even under the existence of oxidation peak A1, indicating that no MPOx species is present on the catalyst surface, or can be spontaneously reduced back to MP, if there is any, by N2H4 because of its strong reductivity as shown in Eq. (2). The result suggests that the in situ electrochemically oxidized Ni-Co-P/NF in anodic potential range can be recovered back to active MP species again by N2H4.
$${MPO}_{x}+{N}_{2}{H}_{4}+{H}_{2}O\to MP+{N}_{2}+{OH}^{-}$$
2
However, after the long-time CV measurements of Ni-Co-P/NF under a lowered N2H4 concentration or without N2H4, the HzOR activity of Ni-Co-P/NF will decline rapidly and cannot be recovered even at increased N2H4 concentration in the following (Fig. S19), which is due to the over-oxidation of catalyst (named Ni-Co-P/NF-inactivated) leading to the deactivation of Ni-Co-P/NF in the absence of N2H4. From the XPS spectra of Ni-Co-P/NF-inactivated (Fig. 5b, Fig. S20c, d), it can be found that the ratio of MPOx to MP significantly increased, proving the over-oxidation of Ni-Co-P/NF responsible for the deteriorated HzOR performance, further confirming that the MP is the active sites for HzOR, which can be in situ recovered by N2H4 molecules in the electrolyte in the case of being transiently oxidized of MP.
In order to further exploring these changes, in situ Raman measurements were carried out. In 1 M KOH electrolyte, the catalyst Ni-Co-P/NF shows a significantly broadened band at around 400 cm− 1 corresponding to the vibration of M − P bond43. When adding 0.1 M N2H4 in electrolyte, three strong N2H4 related peaks appear at 677, 1116, 1598 cm− 1 (Fig. 5c), respectively assigned to N − H, N − N stretching modes of adsorbed *NH2NH2, and N − H bending modes of intermediate *NH2 on the Ni-Co-P/NF surface44–46. Fig. 5d shows the real-time Raman spectra at varied applied voltages for monitoring the catalytic process by Ni-Co-P/NF for HzOR in 1 M KOH + 0.1 M N2H4 (-0.05 V to 0.5 V vs. RHE). At the voltage higher than 0.1 V, the peak density of M − P (400 cm− 1) gradually becomes weakened with the increase of voltage, and a weak peak appears near 980 cm− 1 assigned to the stretching bond of P − O in MPOx, indicating that part of MP species is gradually oxidized to MPOx during the HzOR47. From the in situ Raman measurements at varied reaction time intervals at constant 0.2 V (Fig. 5e), when the voltage application was stopped for 10 minutes, the peak intensity of M-P bond (400 cm− 1) increased again, while the peak of P − O bond (980 cm− 1) disappeared, confirming that the slightly oxidized species (MPOx) could be reduced and recovered back to active MP species due to the presence of N2H4, which agrees well with the CV measurement in Fig. 5a.
Furthermore, with the increase of applied voltage from − 0.05 V to 0.5 V (Fig. 5d), the N2H4 adsorption peak becomes gradually decreased, indicating that the surface adsorbed N2H4 is consumed rapidly leading to the fast HzOR kinetics of Ni-Co-P/NF. However, when the applied voltage is greater than 0.2V, the peak of *NH2 becomes stronger because of the accumulation of reaction intermediate *NH2, indicating the existence of another HzOR path at 0.2 V and above. That is to say that the N2H4 molecule may firstly be adsorbed on the bimetallic sites of Ni-Co-P/NF to form *NH2NH2, in which the N − N bond will break to form two *NH2. Then, each *NH2 gradually dehydrogenates to form N2. This HzOR path is similar to the reaction mechanism on the surface of noble metals48,49.
DFT calculations were adopted to analyze the HzOR electrocatalytic thermodynamics and kinetics. A microstructure of NiCoP/CoP heterointerface (named as < NiCoP|CoP>) was constructed virtually to simulate the Ni-Co-P/NF catalyst as shown in Fig. S21. The density of states (DOS) of the < NiCoP|CoP> (Fig. S22) shows clearly the metallic nature, and the d-band center is located at -1.549 eV (spin-up) and − 1.384 eV (spin-down), which are close to the values of other heterostructures enabling excellent bifunctional activity50. From the Bader charge analysis (Fig. S23), it can be found that more charges from Co than Ni can transfer to P, specifying the site preference for phosphides adsorption onto the heterostructure. Thus, different adsorption sites for N2H4 were examined including NiCoP-side, CoP-side and < NiCoP|CoP> (Fig. S24). Obviously, the ΔG*N2H4 value of < NiCoP|CoP> (-0.85 eV) indicates the highly favorable *N2H4 adsorption kinetics on the heterostructure for HzOR. Furthermore, the charge density difference analysis shows that the N atom in *N2H4 has a significant charge transfer to the nearby Co and Ni atoms, leading to the bond length increasing and the resultant activation of *N2H4 molecules(Fig. S25)51.
It has been reported that the oxidation of N2H4 on the catalyst surface is usually a four-continuous-step proton coupled electron transfer (PCET) process52,53. Therefore, the free energies of dehydrogenation in each step at the < NiCoP|CoP > were calculated as shown in Fig. 5f. The rate-determining step (RDS) for HzOR is the dehydrogenation from *NH2NH2 to *NHNH2, and the ΔG value of this step on the heterostructure is as low as 0.10 eV, indicating that the NiCoP/CoP heterointerface can offer a fast kinetics for HzOR by effectively modulating the electronic structure. Besides this, according to the in situ Raman spectroscopic result, another theoretically possible reaction path is proposed: *N2H4 →2*NH2→2*NH→2*N→N2. To verify the feasibility of this speculation, the free energies in each step were calculated as shown in Fig. 5g. At U = 0 V, the N − N bond breakage (ΔG = 0.45 eV) forming *NH2NH2 is the RDS, while the N − H bond is much easier to break. However, with the increase of the potential to 0.2 V (U = 0.2 V), the N atoms can be more strongly bound onto the catalyst surface, leading to the much lowered ΔG value down to 0.05 eV, implying the greatly weakened N − N bonding, therefore in this case, HzOR may follow a new reaction path of N-N bond breaking in hydrazine molecules in addition to the traditional four-continuous-step PCET path.
(a) LSV curves of the electrolyzer equipped with different electrode catalysts in 1 M KOH without or with the addition of N2H4 at anode; (b)The proposed competing electroless decomposition path (red, path 1) and electrochemical path for H2 production from hydrazine (green, path 2); (c) Relationship between gas volume and transferred charge number at anode and cathode under different anodic N2H4 concentrations; (d) Stability measurements at 100 mA cm− 2 with the ADTs in the inset; (e) Discharge polarization curve and power density plots; (f) Schematic illustration of a HE unit self-powered by a home-made DHzFC for H2 production; (g) Digital photograph of H2 production system with HE unit self-powered by DHzFC.
HE Unit Powered by DHzFC. Considering the excellent electrocatalytic performance of the Ni-Co-P/NF for both HER and HzOR, a H-type double-chamber electrolyzer (HE unit) was assembled using the Ni-Co-P/NF as both cathodic and anodic catalysts. The HE unit takes 1.0 M KOH + 0.1 M N2H4 and 1 M KOH as the anodic and cathodic electrolytes, respectively, separated by anion exchange membrane (AEM). The separation design can avoid the pollution of N2H4 to the cathode solution to obtain high-purity H2 at cathode. As shown in Fig. 6a, a current density of 200 mA cm− 2 can be obtained at the cell voltage of 0.88 V when using Ni-Co-P/NF as electrodes for HE unit, while it can only reach 50 mA cm− 2 under the same voltage by using Pt-C/NF electrodes (Fig. 6a). In sharp contrast, the traditional OER + HER requires 1.92 V to achieve 200 mA cm− 2, which is 1.04 V higher than that with Ni-Co-P/NF electrodes (0.88 V) under the same conditions, indicating the greatly reduced energy consumption by using HzOR assistant hydrogen production.
When no voltage is applied, the HE unit still exhibits an open circuit voltage (OCV) of -0.1 V and current output (Fig. S26), indicating the spontaneous H2 evolution in the device. As a matter of fact, both HzOR and HER will take place concurrently on the Ni-Co-P/NF catalyst surface in anode chamber (Fig. 6b, path 1), leading to the electroless “spontaneous decomposition” of N2H4 without current (electron) output towards cathode, which is actually not desirable for anode N2H4 oxidation-coupled cathode HER, but widely ignored and indeed inevitable54. In order to further study this phenomenon, the gases produced at the cathode and anode were collected by drainage collection method to analyze the actual utilization rates of N2H4 (Fig. S27). The electroless “spontaneous decomposition” of N2H4 can be inhibited by controlling the N2H4 concentration as shown in Fig. 6c, so as to promote the electrochemical oxidation rate of N2H4 (Fig. 6b path 2). The "spontaneous decomposition" can be largely prevented at the N2H4 concentrations lower than 0.05 M, and in this case the electrochemical utilization rate of N2H4 (in path 2 in Fig. 6b) reaches 93.0%. Comparatively, the “spontaneous decomposition” reaction will take place vigorously at increased N2H4 concentrations, and the utilization rates of N2H4 were calculated to 90.5% and 67.6% at 0.1 M and 0.2 M N2H4, respectively.
The HE unit exhibits excellent stability for H2 evolution (Fig. 6d), and the current density of 100 mA cm− 2 can be driven at the voltage lower than 0.63 V and operated stably for nearly 30 h. Even after ADTs for 10000 cycles, the performance at the current density of 100 mA cm− 2 attenuates by only 11.6% according to the inset in Fig. 6d. The regular voltage fluctuations in the stability curve in Fig. 6d mainly comes from the supplementations of N2H4 and the release of bubbles on the electrode (Fig. S28). Because of the large resistance (17 Ω) of AEM, the performance of Ni-Co-P/NF in HE unit has been operated with iR compensation (Fig. S29), which shows a high current density of 500 mA cm− 2 at 0.498 V, higher than the reported capacities of HzOR-assisted HER in membrane-free or double-chamber electrolyzer (Table S5).
Based on the excellent HzOR performance of Ni-Co-P/NF, a direct hydrazine fuel cell (DHzFC) equipped with Ni-Co-P/NF as the anode immersed in 1.0 M KOH + 0.1 M N2H4 electrolyte and 20 wt.% Pt/C as cathode immersed in 1.0 M KOH electrolyte was constructed. As show in Fig. 6e, the DHzFC exhibits a peak power density of 96 mW cm− 2 at 0.75 V, which can drive the LED by two identical DHzFC in series (Fig. S30). Moreover, the DHzFC can run stably at the current densities of 1, 5, 10 and 15 mA cm− 2 for 4 h (Fig. S31), indicating its excellent operation stability at high output voltages. Furthermore, a self-powered H2 production system has been constructed by integrating a DHzFC and the HE unit using Ni-Co-P/NF as the bifunctional catalysts (Fig. 6f). The self-made DHzFC successfully drives the HE unit for H2 production, resulting in the production of a large number of bubbles (see the Movie in SI), and the H2 evolution current is as high as 26.2 mA (Fig. 6g), corresponding to a hydrogen generation rate up to 19.6 mol h− 1 m− 2, proving the great potential in the utilization of self-powered hydrogen production during the non-carbon energy system.