Introduction

Extensive use and depletion of fossil fuels resulting in serious pollution. Therefore, green and renewable fuel resources are required for continuing sustainable economic development1,2,3. Electrocatalysis acts as a vital role in the conversion of clean energy to achieve a sustainable approach to various commercial processes, including HER4,5. However, electrochemical water splitting is hindered by the large kinetic barrier and slow kinetics6,7,8,9. Pt-based electrocatalysts are recognized as highly efficient electrocatalysts due to good electrical conductivity10, fast kinetics11,12, and the preference to overcome the large kinetic energy barrier involved in the above-mentioned process13. Unfortunately, high price and not desirable stability hinder the extended Pt-based catalysts’ application14. Thus, it is very urgent to develop cost-effective Pt-free electrocatalysts with comparable activity and better stability.

Researchers recently have designed a wide range of low-cost catalysts, including transition-metal chalcogenides (TMDCs)15,16, metal nitrides17,18, metal carbides19,20, and metal phosphides21,22. Among these candidates, MoS2, a typical layered 2D TMDCs formed by Van der Waals interaction and stacking of S–Mo–S layers, attracts extensive interests with its adjustable bandgap, unique band structure, high energy-conversion efficiency, and earth abundance23,24,25. However, the electrocatalytic activity of MoS2 is closely associated with its surface electric structure26,27,28,29,30,31,32,33,34,35,36, many researchers have focused on adjusting the electronic structure of the MoS2 surface to promote electrocatalytic activity, such as surface engineering26, doping27, single-atom anchoring28, phase structure29,30,31,32,33, interface active site34,35, and defect36. Interestingly, two main phases of MoS2 were widely justified: 2H and 1T phases29. 2H phase has the most thermodynamical stability among the molybdenum sulfide family, whose HER activities are restrained by the amount and active site types as well as conductivity. Unlike the 2H phase, 1T-phase one demonstrates higher catalytic activity since it has numerous active sites on the edges and a fast transfer rate. However, it is remaining a giant challenge of directly synthesizing the high percentage 1T-phase molybdenum sulfide due to the thermodynamic instability of 1Tphase-MoS230. To solve this problem, a feasible strategy is to efficiently realize the 2H → 1T-phase transformation to improve HER capability. Wang et al. found that a partial 2H → 1T-MoS2 phase transition by facile one-pot annealing of a large amount of 2Hphase-MoS2 under phosphorus vapor is able to enhance HER catalytic activities31. A synergistic strategy of doping nitrogen and intercalating PO43− is reported, which can convert 2H- to 1T-phase with a conversion rate of up to 41%, and has excellent HER performance32. However, the electronic transport capacity and phase stability of the phase boundary of a single component (pure 1T-phase) are generally poor. In order to overcome the puzzles, the HER activity of the pure phase can be improved by constructing a heterogeneous boundary. Therefore, it is expected to further enhance the HER performance and its stability of traditional single 1T-phase or 2H-phase interface by constructing a composite heterojunction between 1T-phase and the other phases33.

Interface modification could be an effective approach to construct a composite heterojunction34,35. Ni-based materials (such as Ni2P, NiS2, Ni2S3, etc) with high activity and conductivity have been considered as highly efficient electrocatalysis materials for HER21,22,37, as another heterogeneous interface, which is also very important to control the electronic structure of the MoS2 interface. Kim et al. reported that Ni2P nanoparticles were used to activate the MoS2 base surface, which exhibits Pt-like HER performance in 0.5 M HCl solution37. Because the electronic structure of Ni2P is a \(P\bar{6}2m\) space group, which could facilitate recombination at the atomic scale. Moreover, Ni has a unique α and β orbital integral asymmetric d orbital, which makes it easy for the lone pair of electrons to recombine with the d orbital of the exposed Mo atom on MoS2 to generate new interface electrons, thereby improving HER performance. Lin et al. reported that a defect-rich heterogeneous interfacial catalyst (MoS2/NiS2) could provide abundant active sites to promote electron transfer, thereby further rapidly promoting electrocatalytic hydrogen evolution38. More importantly, the introduction of NiS2 hybridization on the surface of MoS2 generates a new form of interface electrons, and Niδ+ is reduced to low-valence Ni to improve the binding energy with hydrogen elements, thereby weakening the Mo–H strength. To sum up, although the heterojunction-phase catalyst synthesized by the above-mentioned approach further improves the HER activity and good stability, the understanding and regulation of the surface electronic structure on the MoS2 interface are still huge challenges, and thus it is very necessary to develop an efficient synthesis approach to obtain stable multi-heterogeneous interface catalyst.

Here, we address these challenges by tuning its electronic structure through phase modulation synergistic with interfacial chemistry and defects of phosphorus or sulfur implantation, and we then successfully design and prepare a series of heterojunction-phase-interface electrocatalysts (denoted 1T0.81-MoS2@Ni2P and 1T0.72-MoS2@NiS2) with an outstanding HER activity and are stable in dual-pH surroundings. The strategies to control the electronic characteristics of the MoS2 surface include surface phase modulation, surface defects, and the construction of hetero-structure (Fig. 1a). Furthermore, we control the hydrogen and hydroxyl adsorption energy through the synergistic effect of heterojunction-phase-interface catalysts (Fig. 1b, c–f) because the energy of the hydroxyl species is very important for the hydrolysis accelerator. Starting from hydrothermally synthesized MoS2 nanosheets, we develop a simple surface electronic structure modulation strategy of constructing multi-heterogeneous-phase-interface 1T0.81-MoS2@Ni2P and 1T0.72-MoS2@NiS2 electrocatalysts (Fig. 1a) by citric acid-induced hydrothermal synthesis, electrodeposition and then phosphorus (or sulfur) vapor thermal treatment approach for the first time. Our approach can not only realize the construction of abundant catalytic reactive sites but also improve the conversion rate of 2H to 1T (81%), and it is also convenient to introduce Ni2P or NiS2 heterogeneous interfaces. As to the surface electronic structure of catalysts, high-resolution transmission electron microscopy (HRTEM) images show that such phase-structures, heterojunction-phase-interface edges, and defects are derived by the featured electronic states and Ni atomic coordination. Additionally, X-ray photoelectron spectra (XPS) showed that citric acid induces hydrothermal synthesis of stable 1T0.41-MoS2 (41% of 1T-phase), and the 1T0.81-MoS2 or 1T0.72-MoS2 (81% or 72% of 1T-phase) conversion rate is further improved after phosphorus or sulfur vapor thermal treatment. As-synthesized 1T0.81-MoS2@Ni2P (or 1T0.72-MoS2@NiS2) multi-heterogeneous catalyst exhibits the remarkable HER catalytic activity, achieving the low overpotentials of 38.9 (or 186) and 98.5 mV (or 128 mV) for HER at a current density of 10 mA/cm2. They also have Tafel slopes of 41 (or 79) and 42 (or 68) mV/dec in 0.5 M H2SO4 or 1.0 M KOH media, and good stability during testing for 16 h in both media, respectively. The 1T0.81-MoS2@Ni2P (or 1T0.72-MoS2@NiS2) catalysts exhibited superior activities with Tafel slope values and the overpotentials lower than the values reported for Mo-base HER catalysts in both alkaline and acidic media30,31,33,37,38,39,40. Moreover, as-synthesized 1T0.72-MoS2@NiS2 (or 1T0.81-MoS2@Ni2P) catalyst also exhibits excellent OER and overall-water splitting catalytic activity. DFT calculation results display that the introduction of 1T-phase MoS2 and Ni-based materials can regulate MoS2 electronic structure effectively for making the bandgap narrower, and decreasing H* and water adsorption energy. In situ electrochemical-Raman spectra results indicate that the OH– ions are driven to be adsorbed on Mo, Ni atoms in the alkaline medium, and then there forms the OOH* intermediates. There is a strong interaction between Ni and Mo on the surface of the catalyst, thereby increasing the local electronic state of Mo atoms, reducing the hydrogen-adsorption energy for protons on Mo atoms, and thus improving its intrinsic catalytic. Moreover, X-ray absorption spectroscopy results imply that reduced Ni supply empty d-orbitals to facilitate H atom capture, and decrease Mo–H strength of 1T0.81-MoS2@Ni2P (or 1T0.72-MoS2@NiS2) catalyst. This work provides useful insights for exploring the enhancement mechanisms of HER with an optimized surface electronic structure on the MoS2 interface, which provides an effective insight of constructing invaluable metal electrocatalysts for HER and other fields.

Fig. 1: Motivation and design of electrocatalyst.
figure 1

a Tuning strategy of the electronic structure of the MoS2 surface. b Design ideas of hydrogen evolution catalyst. cf the energetics of hydroxyl species on 2Hphase-MoS2 (002), 1Tphase-MoS2 (002), NiS2 (210), and Ni2P (111) HER electrocatalyst surfaces. g Schematics of the 1T0.72-MoS2@NiS2 and 1T0.81-MoS2@Ni2P synthesis steps.

Results

Preparation and characterizations of multi-heterojunction interface electrocatalysts

The formation process of multi-heterojunction interface electrocatalysts is schematically illustrated in Fig. 1g. 1T0.81-MoS2@Ni2P and 1T0.72-MoS2@NiS2 catalysts were synthesized by a three-step procedure. First, 1T0.41-MoS2 nanospheres were obtained on carbon cloth (CC) by acid-induced hydrothermal approach at 200 °C for 12 h (see details in “Methods” section). The as-obtained 1T0.41-MoS2 shows a large number of microspheres (Supplementary Fig. 1b–d) with a narrow diameter distribution of 2.0–4.0 µm distributed uniformly on the surface of CC substrate. Flower-shaped MoS2 microspheres consist of many aligned 1T0.41-MoS2 nanosheets, on which the Ni(OH)2 nanoparticles were then electrodeposited (see details in “Methods” section). 1T0.41-MoS2@Ni(OH)2 material inherited its morphology from spherical MoS2. When being electrodeposited for 100 s, a small amount of Ni(OH)2 nanoparticles can be anchored on the surface of MoS2 nanospheres (Supplementary Fig. 2). As the electrodeposition time increases to 300 s, a large number of Ni(OH)2 nanoparticles can be observed to adhere to the MoS2 surface (Supplementary Fig. 3). Subsequently, as-prepared 1T0.41-MoS2@Ni(OH)2 was loaded into a quartz tube mixed with red phosphorus or sulfur powder and sealed by oxyacetylene flame. Finally, these were heated to 600 °C for the reaction with red phosphorus or sulfur to synthesize 1T0.81-MoS2@Ni2P and 1T0.72-MoS2@NiS2 catalysts, respectively (Supplementary Figs. 4 and 5). As to 1T0.81-MoS2@Ni2P catalyst, the MoS2 microspheres are very rough, on which there distribute many random Ni2P nanoparticles (Supplementary Fig. 5). It is because that the 1T/2H-mixed phase and heterojunction-interface structure reduces the adhesion of the gas-solid interface and facilitates releasing hydrogen from the catalyst surface, which is essential for enhancing HER34.

Next, the phase composition and crystal structure of 1T0.81-MoS2@Ni2P and 1T0.72-MoS2@NiS2 were obtained by X-ray diffraction (XRD) and Raman spectroscopy. There are some obvious characteristic diffraction peaks of 14.3°, 33.4°, and 59.2° (Supplementary Fig. 6b, c), which can be ascribed to 2Hphase-MoS2 (JCPDS#75-1539). However, the XRD peak of 1T0.81-MoS2@Ni2P and 1T0.72-MoS2@NiS2 located at 2θ ≈ 28.8° can be indexed as the (004) peak of 1Tphase-MoS2, which indicates that 1T- and 2H-mixed phases were successfully hydrothermally synthesized41. The other characteristic peaks (2θ ≈ 31.3°, 35.2°, 38.8°, 44.9°, and 53.3°) demonstrate that the 1T0.72-MoS2@NiS2 is a hybrid of NiS2 (JCPDS#11-0099), which verifies the presence of NiS2 nanoparticles. Similarly, as to 1T0.81-MoS2@Ni2P catalyst, its XRD results also showed the presence of Ni2P nanoparticles (JCPDS#21-0590) on the 1T0.41-MoS2 surface. Raman spectroscopy showed E2g1 and A1g vibrational bands at 376.2 and 402.9 cm−1 peaks typical for 2Hphase-MoS242. J1, J2, and J3 vibrations at 147.3, 235.4 and 335.2 cm−1 are characteristic for 1Tphase-MoS243 (Supplementary Fig. 6b). These results prove that the 1T-phase of MoS2 is formed by the hydrothermal reaction induced by organic acids (e.g., citric acid)41. 1T0.72-MoS2@NiS2 or 1T0.81-MoS2@Ni2P demonstrated three characteristic peaks of 1Tphase-MoS2 and the two characteristic peaks (E2g1 and A1g) of 2Hphase-MoS2. Additionally, they showed a vibrational peak (437.3 cm−1) of Ni–S38 or three vibrational peaks (216.2, 249.7, and 269.5 cm−1) of Ni–P37. More importantly, the E2g1 and A1g vibrations of 1T0.72-MoS2@NiS2 at 382.2 and 408.1 cm−1 were red-shifted by 6.0 and 5.2 cm−1, respectively (Supplementary Fig. 6d). This could be attributed to the exploits the S layer of MoS2 as an external S source to grow NiS2 in situ. Therefore, it changes the original vibration mode of the Mo–S bonds, and the out-of-plane vibration mode has a more significant change44,45,46. Similarly, the E2g1 and A1g peaks for the 1T0.81-MoS2@Ni2P catalyst slightly red-shifted by 7.3 and 3.0 cm−1, respectively, because of interfacial stress between Ni2P and MoS2, indicating that the formation of MoS2@Ni2P heterojunction leads to the Raman shift of MoS244,45,46. These results confirm that rich multi-heterojunction interface edges active sites catalysts were successfully synthesized.

Electronic structure characterizations of 1T0.72-MoS2@NiS2 and 1T0.81-MoS2@Ni2P catalysts

To further identify the surface electronic structure of multi-heterogeneous interface catalysts, we applied the high-resolution transmission electron microscopy (HRTEM) to assess the morphology and crystal structures of 1T0.81-MoS2@Ni2P and 1T0.72-MoS2@NiS2 catalysts. Supplementary Fig. 7a, b shows the typical low-magnification image of the 1T0.72-MoS2@NiS2 on the Cu grid, which confirms the flower-like nanosphere morphologies of 1T0.72-MoS2@NiS2. TEM and corresponding elemental distribution map obtained for the 1T0.72-MoS2@NiS2 sample demonstrated uniformly distributed Mo, Ni, and S (Supplementary Fig. 7c-c4). As revealed by the HRTEM image (Fig. 2a–c and Supplementary Fig. 7e, f), NiS2 nanoparticles are decorated on MoS2 nanosheets edge (Supplementary Fig. 10a, b). The HRTEM image of 1T0.72-MoS2@NiS2clearly shows the crystal lattice of 0.25 nm, referring to the NiS2 (210). Interestingly, Fig. 2a shows the HRTEM image of 1T0.72-MoS2@NiS2 flower-like nanosheets, which there demonstrate the lattice fringes perpendicularly to the electron beam direction circled by blood color, justifying the S defect (Fig. 2c). The trigonal lattice in the yellow circle implies the presence of 1T-phase MoS2, while the hexagonal lattice in the blue circle suggests the presence of 2H phase MoS2. The above-described results further confirm the successful preparation of the 1T0.72-MoS2@NiS2 multi-heterojunction interface catalyst. The anion is changed to be P to produce 1T0.81-MoS2@Ni2P multi-heterojunction interface catalyst by phosphorus vapor thermal treatment. Supplementary Fig. 8a, b displays the morphologies of 1T0.81-MoS2@Ni2P catalyst, overlapping nanosheets with many embedded particles can be clearly identified. There is an obvious alternation of 1T and 2H phases, and a large number of defects or disorder (Fig. 2f and Supplementary Fig. 9). As shown in Supplementary Fig. 8c, there are the distributions of Mo, Ni, S, and P over the whole 1T0.81-MoS2@Ni2P, verifying that Ni2P nanoparticles are encapsulated by MoS2 edges (Supplementary Fig. 10c, d). The interplanar spacings of 0.62 and 0.22 nm are assigned to (002) and (111) interplanar distances of MoS2 and Ni2P, respectively (Fig. 2d, e). Similarly, Fig. 2e, f displays two amplified HRTEM images truncated from Fig. 2d, in which Fig. 2f demonstrates some hexagonal and trigonal lattice areas of semiconductor 2Hphase- and metallic 1Tphase-MoS2, respectively. The HRTEM results further confirm the successful preparation of the 1T0.81-MoS2@Ni2P multi-heterojunction interface catalyst.

Fig. 2: Electronic structure characterizations of 1T0.72-MoS2@NiS2 and 1T0.81-MoS2@Ni2P catalysts.
figure 2

ac HRTEM image of 1T0.72-MoS2@NiS2. a MoS2 lattice, scale bars are 5 nm. b NiS2 lattice and heterojunction interface, scale bars are 1 nm. c shows 2H and 1T lattices, scale bars are 2 nm. df Typical HRTEM image of 1T0.81-MoS2@Ni2P. d shows MoS2 lattice and heterojunction interface, scale bars are 5 nm. e shows Ni2P lattice fringes, scale bars are 1 nm. f shows 2H and 1T lattices, scale bars are 2 nm. g HR Mo 3d core-level XPS spectra of 1T0.72-MoS2@NiS2, 1T0.81-MoS2@Ni2P, 1T0.41-MoS2, and 2Hphase-MoS2. h S 2p core-level XPS spectra of 1T0.72-MoS2@NiS2, 1T0.81-MoS2@Ni2P, 1T0.41-MoS2 and 2Hphase-MoS2, respectively. i Ni 2p XPS spectrum for 1T0.72-MoS2@NiS2, 1T0.81-MoS2@Ni2P, pure Ni2P, and Ni2S.

Next, we performed XPS measurement to assess the elemental valence states of all the as-synthesized samples (Fig. 2g–i and Supplementary Fig. 13a). Full XPS spectrum for 1T0.72-MoS2@NiS2 (Supplementary Fig. 13a) showed that atomic ratios of Mo, S, and Ni were equal to 13.96%, 36.96%, and 4.39%, respectively, and close to that measured by HRTEM elemental mapping (~14.30%, 35.87%, and 4.76%). Mo 3d spectra obtained for the 1T0.41-MoS2 sample shows Mo4+ 3d3/2 and Mo4+ 3d5/2 peaks at 232.68 and 229.43 eV (Fig. 2g), respectively, confirming the existence of Mo4+ for the 1T0.41-MoS2. As to the 1T0.72-MoS2@NiS2, or 1T0.81-MoS2@Ni2P heterostructures catalyst, the high-solution Mo 3d XPS spectrum shows that both Mo4+ 3d3/2 and Mo4+ 3d5/2 peaks for mixed-phase MoS2 has a shift of 0.23 eV and 0.15 eV to lower binding energy compared with 1T0.41-MoS2 (Supplementary Fig. 11a), which is attributed to the existence of 1Tphase-MoS247. In addition, two peaks of 163.41 and 162.22 eV are observed in the 1T0.41-MoS2, corresponding to S2− 2p1/2 and S2− 2p3/2, respectively (Fig. 2h). However, the binding energies of S2− 2p1/2 and S2− 2p3/2 in 1T0.72-MoS2@NiS2, or 1T0.81-MoS2@Ni2P heterostructures catalyst shift to 163.28 and 162.10 eV, respectively (Supplementary Fig. 11b). This negative-shift (0.13 eV) suggests little electron transfer between NiS2 (or Ni2P) and MoS2, also suggesting the reconfiguration of the electronic structure during the transferring of electron from Mo4+ to the surrounding Ni sites48. Interestingly, the 1T-phase contents in the 1T0.81-MoS2@Ni2P and 1T0.72-MoS2@NiS2 samples (81% and 72%, respectively) were higher than the 41% value observed for the 1T0.41-MoS2. Thus, phosphorus or sulfur implantation further facilitates the phase transformation of 1Tphase-MoS224,32. The reason may be that phosphorus can be simultaneously inserted into S–Mo–S atomic planes, inducing the glide of S atomic planes, affording in-plane heterostructures between 1T and 2H MoS2 domains (Supplementary Fig. 14), which is consistent with previous reports31,32. As shown in Fig. 2i, the Ni 2p spectrum of 1T0.81-MoS2@Ni2P shows two spin–orbit doublets at 856.6 and 874.9 eV corresponding to Ni2+ 2p3/2 and Ni2+ 2p1/2 oxidation states in Ni2P, respectively, and two satellite peaks (identified as “Satellite.”)49. Notably, compared with the binding energies of Ni 2p3/2 (857.4 eV) and Ni 2p1/2 (875.4 eV) of pure Ni2P, the two binding energies of Ni 2p3/2 and Ni 2p1/2 have a significant negative-shift of approximately 0.8 and 0.6 eV in 1T0.81-MoS2@Ni2P (Fig. 2i), respectively. This result implies the transfer of electrons from Mo4+ to Ni2+ sites in the 1T0.81-MoS2@Ni2P sample, resulting in a low-valence state and electron-rich structure of Ni2+ sites50. For pure NiS2 sample, the peaks of Ni 2p3/2 and Ni 2p1/2 located at 854.7 and 872.2 eV, and the corresponding satellites appear at 858.7 and 878.6 eV, respectively. However, the binding energies of Ni 2p3/2, Ni 2p1/2 and satellite in 1T0.72-MoS2@NiS2 sample (Fig. 2i) are positive-shifted to 856.6 (by 1.9 eV), 875.3 (by 3.1 eV), 861.8 (by 3.1 eV) and 862.2 eV (by 1.9 eV), respectively. The positive-shift of the Ni 2p binding energies peaks manifest a higher valence state, which are ascribed to Ni bonded to S and O atoms, such as sulfides or surface oxides/hydroxides51. For the 1T0.81-MoS2@Ni2P, the P 2p spectrum shows two peaks at 130.4 and 129.5 eV corresponding to P 2p1/2 and P 2p3/2, respectively, suggesting the existence of Ni2P. In addition, it also can be observed another peak at 134.7 eV of oxidized phosphate (P–O) species (Supplementary Fig. 13b), which is due to the partial oxidation of Ni2P in air. Notably, the binding energies of Ni 2p3/2 (852.7 eV) and P 2p3/2 (129.5 eV) are both shifted, indicating that charge transfer occurs from Ni to P, which can greatly promote the catalytic activity of 1T0.81-MoS2@Ni2P.

Electrocatalytic HER performances in alkaline and acidic media

1T0.72-MoS2@NiS2 and 1T0.81-MoS2@Ni2P electrodes exhibited attractive multi-heterogeneous interface edges, plentiful active sites, and abundant mass transfer and gas release channels and are expected to be used as very effective and stable catalysts for H2 production. First, we analyzed HER activities (in 1.0 M KOH) of the electrodes containing these electrodes. The 1T0.72-MoS2@NiS2 and 1T0.81-MoS2@Ni2P electrodes exhibit small overpotentials of 95 and 170 mV at 10 mA/cm2, respectively (see linear sweep voltammetry (LSV) results in Fig. 3a), which are better than the commercial Pt/C electrode (127 mV). To in-depth understand the HER kinetic mechanism, we calculated the Tafel slopes of these electrodes using the Tafel equation52 and obtained the smallest slopes equal to 68 and 79 mV/dec for the 1T0.72-MoS2@NiS2 and 1T0.81-MoS2@Ni2P electrodes, respectively (Supplementary Fig. 15a). These values are even closer to the Tafel slope of the Pt/C electrode (56 mV/dec). Thus, the 1T0.72-MoS2@NiS2 and 1T0.81-MoS2@Ni2P electrodes as active electrocatalysts exhibit the fastest HER processes and better reactivity, which is attributed to the multi-heterogeneous interface effect, a large number of defects, and a higher proportion of 1Tphase-MoS2. Next, we evaluated the long-term cycling stability of the as-prepared electrodes using the chronopotentiometry technique at 10 and 30 mA/cm2, respectively. The 1T0.72-MoS2@NiS2 and 1T0.81-MoS2@Ni2P electrodes were very robust and exhibited negligible damping after 16 h measurement (Supplementary Fig. 15b), and the LSV curves measured before and after the long-term tests are almost the same (Supplementary Fig. 15c), demonstrating excellent long-term stability. Supplementary Fig. 15d lists the overpotential values for the 20.0 wt % Pt/C, 1T0.72-MoS2@NiS2, and 1T0.81-MoS2@Ni2P electrodes in 1.0 M KOH at various current densities. 1T0.72-MoS2@NiS2 electrodes exhibited lower overpotential. Generally, low overpotential and Tafel slope values demonstrated the superior HER catalytic activities, which was the case for our 1T0.72-MoS2@NiS2 and 1T0.81-MoS2@Ni2P electrodes. Moreover, 1T0.72-MoS2@NiS2 electrode has such excellent HER activity comparable to those of as-reported Mo-based materials (Fig. 3b) and composites and various representative catalysts30,31,33,37,38,39,40 (Supplementary Table 3). Thus, 1T0.72-MoS2@NiS2 electrode is a catalyst with the best HER activity in alkaline solutions.

Fig. 3: HER was performed in alkaline and acidic electrolytes.
figure 3

a LSV curves in 1 M KOH. b η10 and Tafel slopes for various Mo-based HER electrocatalysts in 1.0 M KOH. c LSV curves in 0.5 M H2SO4. d η10 and Tafel slopes for various Mo-based HER catalysts in 0.5 M H2SO4. (All LSV curves were corrected without iR-compensation).

To obtain the electrochemically active area (ECSA) of the 1T0.72-MoS2@NiS2 and 1T0.81-MoS2@Ni2P electrodes, the double-layer capacitance (Cdl) was calculated because the two values are proportional to each other. Therefore, we tested their cyclic voltammetry (CV) by continuously increasing scanning speed (Supplementary Fig. 16a-c) in order to obtain the CV curve of the electrode materials in the non-Faraday region (−0.2 to 0.4 V). Then, as shown in Supplementary Fig. 16d, the Cdl was calculated from the plot slope (slope = 2Cdl) between current-density difference (∆j) (0.15 V vs. RHE) and scan rate. The 1T0.72-MoS2@NiS2 electrodes possessed the highest Cdl value (Cdl = 359.7 mF/cm2), suggesting a multi-heterogeneous interface could be effectively enhanced conductivity and exposed more active sites of as-prepared electrodes. We recorded the electrochemical impedance spectra (EIS). The corresponding Nyquist (Supplementary Fig. 17) of the 1T0.72-MoS2@NiS2 electrode showed the lowest value for the charge transfer resistance (Rct). Thus, it possessed very favorable charge transfer kinetics. To further reveal the intrinsic catalytic activity of each active sites, the turnover frequency (TOF) is also calculated53. Based on the above-mentioned analysis, CV approach is regarded as the promising way to determine reasonable results (Supplementary Fig. 18). The TOF value of 1T0.81-MoS2@Ni2P (3.56 S−1) and 1T0.72-MoS2@NiS2 (2.26 S−1) heterojunction catalyst at the overpotential of 200 mV is 18.7 and 11.9 times higher than of 2Hphase-MoS2 catalyst (0.19 S−1) for HER, respectively (Supplementary Table 1). Typically, the amount of hydrogen evolution was measured of 1T0.72-MoS2@NiS2 catalyst in 1.0 M KOH solution (Supplementary Fig. 19), presenting HER Faraday efficiency of 97.6 ± 0.6%, owing to the synergistic effect of the phase, defect and interface engineering of electrocatalyst.

Next, we also studied the HER performance of all the as-prepared electrodes in 0.5 M H2SO4 (Fig. 3c). The HER catalytic performance of the 1T0.81-MoS2@Ni2P and 1T0.72-MoS2@NiS2 electrodes was significantly improved their HER activities according to the LSV data: their overpotential values at 10 mA/cm2 were as low as 38.5 and 152 mV, respectively, which is lower than the values for the electrodes containing 1T0.41-MoS2@Ni(OH)2 (236 mV), 1T0.41-MoS2 (389 mV), 1Tphase-MoS2 (392 mV), and 2Hphase-MoS2 (354 mV). The Tafel slopes for the 1T0.81-MoS2@Ni2P and 1T0.72-MoS2@NiS2 electrodes were 41 and 42 mV/dec (Supplementary Fig. 20a). These values were lower than the values obtained for 1T0.41-MoS2 (169 mV/dec), 1Tphase-MoS2 (163 mV/dec), and 2Hphase-MoS2 (189 mV/dec) electrodes and were better than the electrode based on 20 wt% Pt/C (86 mV/dec). It is probably because, in the acidic environment, the H2 desorption is the limiting step because H+ are abundant. The 1T0.81-MoS2@Ni2P electrode had a weaker adsorption capacity toward Hads so it exhibits a better catalytic effect than 2Hphase-MoS254. Meanwhile, compared to the other electrodes, 1T0.81-MoS2@Ni2P also has a higher ECSA because it has a larger Cdl (Cdl = 106.15 mF/cm2, Supplementary Fig. 21) and, as a result, more catalytical sites, which significantly contributed to the overall activity. Furthermore, 1T0.81-MoS2@Ni2P also possesses a much smaller Rct, in contrast to other electrodes at 300 mV overpotential vs. RHE (Supplementary Fig. 22), revealing satisfied electron transport and good catalytic kinetics, which leads to high activity and low Tafel slope. Supplementary Fig. 20b shows that at 10 and 45 mA/cm2, 1T0.81-MoS2@Ni2P and 1T0.72-MoS2@NiS2 electrodes were very durable and possesses negligible damping after 16 h measurement, which displays excellent long-term stability. In addition, even after 16 h of a chronoamperometric stability test of the electrodes, the current density remains above 95% (Supplementary Fig. 20c), and there is only a slight deviation for the LSV recorded after the stability test, indicating that as-prepared electrodes have very good stability in an acidic environment. As to 20.0 wt% Pt/C, 1T0.72-MoS2@NiS2, and 1T0.81-MoS2@Ni2P electrodes in 0.5 M H2SO4, Supplementary Fig. 20d shows overpotentials vs. various current densities. 1T0.81-MoS2@Ni2P exhibits lower overpotential. We also compared the overpotentials (at 10 mA/cm2 in acidic medium) and Tafel slopes with previously excellent Mo-based electrocatalysts8,32,37,55,56,57 (Fig. 3d and Supplementary Table 4). Catalytic HER performance of 1 T0.81-MoS2@Ni2P is also superior. Afterward, the amount of hydrogen evolution of 1T0.81-MoS2@Ni2P catalyst was given in Supplementary Fig. 23, demonstrating a promising Faraday efficiency of 98.7 ± 0.5% towards real water splitting into hydrogen. Based on the above-mentioned results, 1T0.81-MoS2@Ni2P multi-heterogeneous interface catalyst shows the remarkable intrinsic HER activities in acidic medium mainly attributed to multi-heterointerface interface edges active sites. In addition, as-synthesized 1T0.72-MoS2@NiS2 (or 1T0.81-MoS2@Ni2P) catalyst also exhibits excellent OER and overall-water splitting catalytic activity (Please see Supplementary Information for details, Supplementary Figs. 2527).

Theoretical calculation and mechanisms analysis of the surface electronic structure and HER activation energy for the as-prepared electrocatalysts

To explain the distinguished synergistic effect of 1T0.72-MoS2@NiS2 (or 1T0.81-MoS2@Ni2P) multi-heterogeneous interface catalysts, Density functional theory (DFT) calculations were also performed. Model building and computational parameters can be seen in the “Methods” section. Firstly, the interfacial electron interaction was investigated. The charge difference images (Fig. 4a, b and Supplementary Fig. 37) reveal the charge transfer from 1T0.41-MoS2 to the Ni2S or/and Ni2P interface, and the introduction of 1T-phase is more conducive to charge transfer from MoS2 to NiS2 or Ni2P interface, which significantly increases the interface electron concentration and thus improves its activity. To better understand the surface electronic structure reconfiguration of MoS2 through a coordinated phase transition and interface regulation in theory, the band structure and density of states (DOS) of bare NiS2, Ni2P, 2Hphase-MoS2, 1Tphase-MoS2, 2Hphase-MoS2@NiS2, 2Hphase-MoS2@Ni2P, 1Tphase-MoS2@NiS2, and 1Tphase-MoS2@Ni2P (Fig. 4c–e and Supplementary Figs. 3840) obtained using the hybrid DFT-HSE06 exchange–correlation functional, which is presented in the Supplementary Information. The calculation results show that the bare NiS2 exhibits typical semiconductor characteristics (Fig. 4c), with a narrow bandgap equal to 0.68 eV (Supplementary Figs. 38 and 39a). The band structure of 1Tphase-MoS2 (Fig. 4d) and 1Tphase-MoS2@NiS2 (Fig. 4e) exhibited a certain zero bandgap, indicating a complete transition from the semiconductor phase (0.91 eV) to the metallic phase (0 eV) with improved conductivities27. Notably, the intensity of PDOS of 1Tphase-MoS2@NiS2 was higher than that of 1Tphase-MoS2 and NiS2 at the Fermi level (Supplementary Figs. 38 and 39). Thus, the electron mobility of the 1Tphase-MoS2@NiS2 catalysts was more favorable for the efficient charge transfer, which agrees consistent with the EIS test results58. Moreover, the PDOS results imply that the NiS2 interface hybrid generates some new interface electronic states in 1Tphase-MoS2 (Supplementary Fig. 39c), which was very likely because of the hybridization of the d-orbital of Mo and an empty d-orbital of Ni. Thus, higher HER activity of 1Tphase-MoS2@NiS2 in comparison to 1Tphase-MoS2 agrees with the Fermi level DOS (Fig. 4d, e). Thus, the actual electrochemical performance would show even faster conductivity and charge transfer kinetics.

Fig. 4: Theoretical calculation and mechanisms analysis of the surface structure and HER activation energy of the as-prepared electrocatalysts.
figure 4

the deformation of the electronic density of a 2Hphase-MoS2@NiS2 and b 1Tphase-MoS2@NiS2 interface, in which yellow/green isosurfaces correspond to positive/negative spin densities (0.00295308 e/Å3). Band structure and density of states (DOS) for c NiS2, d 1Tphase-MoS2 and e 1Tphase-MoS2@NiS2. f Free-energy diagrams for HER on the 2Hphase-MoS2, 1Tphase-MoS2, pure Ni2P, pure NiS2, 1T/2Hmix-MoS2, 2Hphase-MoS2@Ni2P, 2Hphase-MoS2@NiS2, 1Tphase-MoS2@NiS2 and 1Tphase-MoS2@Ni2P interface edge. g Schematics showing water activation, *H intermediate formation and hydrogen generation on multi-heterojunction electrocatalysts.

To reveal further the relationship of HER activity of catalysts with phase structure and heterojunction-interface, we used DFT to calculate the optimized structures and free-energy diagrams for HER on 2Hphase-MoS2, 1Tphase-MoS2, 1T/2Hmix-MoS2, pure Ni2P, pure NiS2, 2Hphase-MoS2@NiS2, 2Hphase-MoS2@Ni2P, 1Tphase-MoS2@NiS2, and 1Tphase-MoS2@Ni2P catalysts with partially multi-heterojunction interface modification. As shown in Fig. 4f and Supplementary Fig. 42, the reaction pathway for alkaline HER is constructed59,60, including prior H2O dissociation to form H* intermediates (Volmer step) and hydrogen generation (Tafel step or Heyrovsky step). However, the energy of the intermediate state H*(ΔGH*) is a critical indicator of the ability of hydrogen evolution (Tafel step or Heyrovsky step)35,59. Figure 4f displays the calculated free-energy diagram on the most stable energy of the 2Hphase-MoS2, 1Tphase-MoS2, Ni2P, NiS2, 2Hphase-MoS2@Ni2P, 2Hphase-MoS2@NiS2, 1Tphase-MoS2@NiS2, and 1Tphase-MoS2@Ni2P catalysts (Supplementary Fig. 41). For 2Hphase-MoS2, the ΔGH* is very positive (1.49 eV), indicating that there is a strong interaction between H* and 2Hphase-MoS2, showing poor HER reaction kinetics. More importantly, MoS2 shows unfavorable catalyst-OHad energetics (ΔGH2O = 0.82 eV), suggesting that the relatively high activated H2O-adsorption energy will hinder the decomposition of H2O into H* intermediates and results in slow HER kinetics. The introduction of the 1T/2Hmix-phase into MoS2 can obviously decrease the value of ΔGH* to 0.97 eV and ΔGH2O to 0.16 eV, implying promoted HER activity compared to 2Hphase-MoS2. Notably, constructing multi-heterointerface interface edges active sites with NiS2 can provide the active sites for –OH adsorption, and the followed ΔGH2O and ΔGH* are decreased to 0.10 and −0.12 eV on the 1Tphase-MoS2@NiS2 interface, indicating the 1T/2Hmix-phase and NiS2 nanoparticles are effective for cleaving HO–H bonds and weaker interaction between H*. Also, the charge transfer from Ni2P to the MoS2 is verified by the DFT calculations, and hence there is a more optimal ΔGH* value of about −0.09 eV. Hence, the NiS2 (or Ni2P) can act as a promoter of H2O dissociation and form hydrogen intermediates which then adsorb on nearby MoS2 catalyst sites. In this way, the multi-heterointerface can also accelerate the subsequent generation of H2. The reaction pathways on the single side (such as Ni2P, NiS2, and MoS2) of the interface have also been shown in Fig. 4f and Supplementary Fig. 42. These both show there is more unfavorable energetics than that of the synergetic pathway on MoS2@NiS2 or MoS2@Ni2P interface. The reason is that H* adsorbed on the surface of 2Hphase-MoS2 bounds to Mo atoms, and strong Mo–H strength and poor conductivity. However, H* can be absorbed not only by the 1Tphase-MoS2@NiS2 surface. Ni atoms possess empty d orbitals capable of binding H atoms, thereby weakening the Mo–H strength. More importantly, the introduction of the 1T-phase not only increases its electrical conductivity but also creates abundant active sites at the multi-heterojunction interface edges, which synergistically promote HER activity (Fig. 4g). Thus, our work demonstrates a novel and efficient design to create multi-heterogeneous interfacial electrocatalysts without noble metal materials and with excellent HER activity.

In situ electrochemical-Raman spectroscopy

To better understand the active sites of the 1T0.81-MoS2@Ni2P and 1T0.72-MoS2@NiS2 electrodes during the HER process, we used in situ Raman spectroscopy to study it (Supplementary Fig. 46). The in situ Raman test system is shown in Supplementary Fig. 46a. The as-prepared sample, Ag/AgCl, and Pt wires were used as working electrode, reference electrode, and counter electrode, respectively. In addition, the electrolyte was 1.0 M KOH. As shown in Supplementary Fig. 46c, d, the Raman spectra of 1T0.72-MoS2@NiS2 electrode collected under potentiostatic conditions at stepped potential values from 0 V to −1.5 V. The results show that three characteristic peaks (147.3, 235.4, and 335.2 cm−1 are attributed to J1, J2, and J3 vibrations) of 1T-MoS2, two characteristic peaks (382 and 407 cm–1 are attributed to the E2g1 and A1g vibrational bands) of 2H-MoS2, and a vibrational peak (437.3 cm−1) of Ni–S. However, when the 1T0.72-MoS2@NiS2 sample was put in the electrolyte solution (1.0 M KOH solution), at different applied voltages from the −0.4 to −1.5 V during electrocatalytic HER, these Raman peaks are significantly enhanced. In addition, many new peaks aroused at 152, 188, 222, 284, 322, 453, 500, 548 cm–1 of MoS2, respectively61. The changes of these Raman peaks indicate that new chemical bonds are formed between our samples and the functional groups of –OH, H+, and H2O molecules in the electrolyte, suggesting that it has a strong absorption capacity of ions and H2O molecules. In addition, it can be observed for two slight new peaks of 429 and 488 cm−1 under the bias potential of −0.4 V (Supplementary Fig. 46d), corresponding to the vNi-OH band of our samples. This result indicates the adsorbed H2O molecules during the cathodic polarization process are decomposed into Hads species and OH ions62. More importantly, the intensity of these characteristic peaks of 1T0.72-MoS2@NiS2 are increased significantly as the potential varies from −0.4 to −1.5 V. It may be due to the OH being driven to adsorb on Mo, Ni, S atoms in the alkaline medium, and then OOH* intermediates are formed63. As to 1T0.81-MoS2@Ni2P sample, we also obtained similar results (Supplementary Fig. 46e, f). We used in situ growth of NiS2 (or Ni2P) nanoparticles on the entire surface of 1T-2H MoS2 microspheres to construct multi-heterojunction interface, which may generate Ni–Mo metal bonds, thereby increasing the number of effective active sites of the catalyst. Due to the introduction of Ni atoms, there is a strong interaction between Ni and Mo atoms on the surface of the catalyst, thereby increasing the local electronic state of Mo atoms, reducing the hydrogen-adsorption energy of the H+ on Mo atoms, and thus improving its intrinsic catalytic activity.

X-ray absorption spectroscopy

To investigate electronic states of catalysts, X-ray absorption near-edge structure (XANES) spectra were measured on the fresh catalysts and those after being used in the HER process at three representative potentials (−0.04, −0.1, and −0.2 V), near the onset potential and the overpotential at the current densities of 5 and 10 mA cm–2 (for 1T0.81-MoS2@Ni2P sample), respectively. Figure 5a presents Ni K-edge XANES spectra of 1T0.81-MoS2@Ni2P catalyst recorded at different applied potentials and reference spectra of Ni foil, NiO, and NiO2. From the fresh catalyst to that under the −0.04 V potential condition, the absorption edge is shifted to the lower energy side by ~0.5 eV, along with a broadening of the white-line peak, meaning a decrease of the Ni oxidation state. Moreover, when cathodic potentials of −0.04 and −0.1 V versus RHE were applied, a further shift of the absorption edge towards lower energy by ~0.2 eV occurs in relation to the case under the −0.04 V potential condition, implying a distinct decrease in the Ni valence state in 1T0.81-MoS2@Ni2P during the HER. Notably, all catalyst spectra exhibit white line at 8350.2-8350.9 eV (Supplementary Fig. 47), corresponding to the 1s to 4p electronic transition, indicating Ni–O local coordination similar to NiOOH and Ni oxides64. Using the edge positions of NiO and NiO2 as references, the Ni average valence state is determined as +3.3, +2.2, +1.8, and +2.0 for the fresh and those used at −0.04, −0.1, and −0.2 V, respectively (Supplementary Table 9). Therefore, Ni cations are reduced under working conditions, which is consistent with the Ni 2p XPS results (Supplementary Fig. 28d). In addition, 1T0.72-MoS2@NiS2 (Supplementary Fig. 48) and 1T0.41-MoS2@Ni(OH)2 (Supplementary Fig. 49) show similar behavior as Ni valency is decreased from +3.6 (fresh) to +2.4 (−0.1 V) and from +2.7 (fresh) to 1.8 (−0.2 V), respectively. In addition, oxidation of Mo from +4 to +6 after catalysis is observed as shown in Fig. 5b. The fresh 1T0.81-MoS2@Ni2P exhibit much broader Mo L3 XANES absorption than those of the Mo standards (MoS2, MoO2, and MoO3), and its lower edge position indicates reduced Mo admixture. In most cases, the broadening of XANES relates to a lack of crystallinity. Interestingly, after the HER reaction, the broad peak is shifted to higher energy and split, indicating 4dt2g and 4deg absorption bands of MoO365 located at 2524 and 2526 eV, respectively. These results indicate the valence states of Mo cations are increased from approximately +4 to a higher oxidation state (+6) under working conditions. Moreover, sulfur does not take part in catalysis as shown in Fig. 5c. All S K-edge XANES spectra of catalysts before and after the reaction are similar and agree well with that of MoS2 standard. P K-edge XANES spectra of 1T0.81-MoS2@Ni2P and FePO4 are shown in Fig. 5d. The white line at 2154 eV belongs to PO42− associated with hybridized O 2p- P 3p absorption band66 whereas the original Ni2P species appear only as a minor peak at 2146 eV. The peak position and absorption line shape of Ni2P are close to that of Co2P indicates the valence state P3-67. The ratio of Ni2P to PO42− changes with the cathodic potential, and the highest ratio is 1:3 at −0.2 V. As to the fresh catalyst, the front peak is much weaker, which indicates that although there is a Ni–P bond, its signal is changed by the presence of some other elements. When cathodic potentials of −0.1 and −0.2 V versus RHE were applied, showing the highest intensity of the front peak, which indicates that both P and Ni are involved in the reaction during the HER process.

Fig. 5: XANES spectra measurements of fresh 1T0.81-MoS2@Ni2P catalyst and after being used in the HER process at −0.04, −0.1, and −0.2 V vs RHE, respectively.
figure 5

a Ni K-edge XANES spectra of 1T0.81-MoS2@Ni2P catalyst and standards Ni foil, NiO, and NiO2. Inset, Magnified pre-edge XANES region. b Mo L-edge XANES spectra of 1T0.81-MoS2@Ni2P catalyst and standards MoS2, MoO2, and MoO3. c S K-edge XANES spectra of 1T0.81-MoS2@Ni2P catalyst and standards MoS2, and FeSO4. d P K-edge XANES spectra of 1T0.81-MoS2@Ni2P catalyst and standards CH2-P(OH)3, and FeSO4.

Overall, the XANES spectra studies provide clear evidence that the structures of as-prepared catalysts can drastically change under realistic catalytic conditions. The Ni site at the interface of heterojunction is most susceptible to low-valence induced by chemisorbed OH under electrochemical conditions. While the valence state of the Mo site at the interface increases, suggesting that the charge transfer (electron transfer from the Mo site to the Ni site) on the surface of the heterojunction catalyst is accelerated during the HER reaction. Therefore, reduced nickel possesses empty d-orbitals, which is beneficial to additional H binding ability. Moreover, it can decrease Mo–H bond strength, and so greatly enhance the HER catalytic activity of as-prepared catalysts.

Discussion

In summary, we have constructed multi-heterogeneous-interface catalysts (1T0.81-MoS2@Ni2P and 1T0.72-MoS2@NiS2) by tuning its electronic structure of phase modulation synergistic with interfacial chemistry and defects to phosphorus or sulfur implantation strategies, which is an efficient approach to obtain abundant reactive sites of long-cycling and stable electrocatalysts for HER in alkaline and acidic surroundings. The as-achieved 1T0.81-MoS2@Ni2P and 1T0.72-MoS2@NiS2 electrodes only require small overpotentials of 38.9 (or 186) and 98.5 (or 128) mV to drive HER at 10 mA/cm2 and have low Tafel slopes: 41 (or 79) and 42 (or 68) mV/dec in 0.5 M H2SO4 (or 1.0 M KOH). Accordingly, these results show varieties of multi-heterogeneous interfaces in 1T0.81-MoS2@Ni2P and 1T0.72-MoS2@NiS2 electrodes, which can be considered versatile electroactive sites and facilitate electron transfer because of their unique heterogeneous effects. DFT calculation results also display that the introduction of metallic-phase MoS2 and intrinsic HER-active Ni-based materials can regulate MoS2 electronic structure effectively for making the bandgap narrower. In situ electrochemical-Raman spectroscopy indicates that the OH ions are driven to be adsorbed on Mo, Ni atoms in the alkaline medium, and then there form the OOH* intermediates. There is a strong interaction between Ni and Mo on the surface of the catalyst, thereby increasing the local electronic state of Mo atoms, reducing the hydrogen-adsorption energy for protons on Mo atoms, and thus improving its intrinsic catalytic. Additionally, XANES spectroscopy results imply that reduced Ni supply empty d-orbitals to facilitate H atom capture, and decrease Mo–H strength of 1T0.81-MoS2@Ni2P (or 1T0.72-MoS2@NiS2) catalysts that account for the outstanding HER properties with lower Tafel slopes and overpotentials compared with 2Hphase-MoS2, 1Tphase-MoS2 counterparts and other Mo-based catalysts. Thus, our work provides a new horizon for rationally designing multi-heterogeneous interfaces of non-precious electrocatalysts to realize excellent HER activities.

Methods

Synthesis of 1T0.41-MoS2

MoS2 microspheres were grown on carbon cloth (CC) hydrothermally. First, a CC (2 × 4 cm) was cleaned (for 15 min) using acetone and then sonicated in deionized water and ethanol for 10 min. Then, sodium molybdate (Na2MoO4·2H2O, 411.9 mg) and thiourea (CS(NH2)2, 608.96 mg) were added to deionized water (40 mL) and citric acid (20 mL). The mixture was magnetically stirred to form a cleaning solution, then placed into a 100 mL Teflon-lined autoclave and held in it at 180 °C for 12 h. Finally, the CC substrates with 1T0.41-MoS2 microspheres (denoted through the paper as 1T0.41-MoS2) were rinsed using deionized water and ethanol and vacuum-dried for 6 h at 60 °C. For comparison, deionized water was used as the solvent, and 2Hphase-MoS2 microspheres were synthesized hydrothermally at 220 °C for 24 h from the same precursors.

Synthesis of 1Tphase-MoS2

We used Li-intercalated bulk MoS2 to prepare 1Tphase-MoS268. In an Ar-filled glove box, bulk MoS2 (1.0 g) prepared by stripping were dispersed in 15 mL of 2 M n-BuLi/hexane solution and stirred at ambient conditions for 48 h. The resulting black materials were repeatedly rinsed with anhydrous n-hexane and then centrifuged to eliminate n-butyl lithium excess and other solution impurities. The 1Tphase-MoS2 powder was prepared and was then coated on the CC substrate. In order to promote better contact between 1Tphase-MoS2 and CC substrate, we annealed (500 °C) the CC loaded with 1Tphase-MoS2 sample under the protection of Ar gas.

Synthesis of 1T0.41-MoS2@Ni(OH)2

We use a standard three-electrode system to prepare 1T0.41-MoS2@Ni(OH)2. 1T0.41-MoS2 acted as a working electrode, while Pt sheet and Ag/AgCl/3.5 M KCl acted as counter and reference electrodes. Ni(OH)2 was electrodeposited on the 1T0.41-MoS2 using 0.1 M Ni(NO3)2 at 5.0 mA/cm2 cathode current density applied for 300 s. 1T0.41-MoS2@Ni(OH)2 samples were rinsed with deionized water and ethanol several times and vacuum-dried at 60 °C.

Synthesis of 1T0.72-MoS2@NiS2

The 1T0.72-MoS2@NiS2 multi-heterogeneous interfaces were prepared by the solid-vapor reaction method. First, a piece of 1T0.41-MoS2@Ni(OH)2 grew on CC was put into the quartz tube with 32.0 mg S powder and was then sealed. Secondly, the quartz tube was positioned inside a tube furnace and was calcinated at 500 °C for 60 min to obtain a 1T0.72-MoS2@NiS2 electrode.

Synthesis of 1T0.81-MoS2@Ni2P

Similarly, the 1T0.81-MoS2@Ni2P multi-heterogeneous interfaces were also obtained by the solid-vapor reaction method. First, a piece of 1T0.41-MoS2@Ni(OH)2 grew on CC was put into the quartz tube with 31.0 mg red phosphorus and was then sealed. Secondly, the quartz tube was also calcinated at 580 °C for 1.0 h to prepare the 1T0.81-MoS2@Ni2P electrode. Additionally, 20 wt% Pt/C was also coated on CC substrate (2.0 mg/cm2) and was labeled as 20% Pt/C for comparison.

Materials characterization

All as-synthetized electrodes were characterized by XRD (performed by Bruker D8 Advance instrument) and Raman spectroscopy (performed using Horiba LabRAB HR800 instrument). The sample morphologies were studied using SEM performed by Hitachi SU8010 instrument and TEM (performed by FEI Tecnai F30 instrument). XPS spectra were collected by the ESCALAB 250Xi instrument manufactured by ThermoFisher using Al Kα radiation.

Electrochemical measurements

All electrochemical measurements were performed with a CHI 660E Electrochemical Workstation (CHI Instruments, Shanghai Chenhua Instrument Corp., China). The HER performance of different catalysts (1.0 cm2) was characterized using a three-electrode electrochemical cell in N2-saturated 1.0 M KOH and 0.5 M H2SO4 electrolyte, respectively. Before testing the polarization curve, we first perform cyclic voltammetry (CV) for more than 20 cycles to activate the as-prepared catalysts with a scan rate of 50 mV/s. The EIS tests were measured by AC impedance spectroscopy at the frequency ranges 106 to 1.0 Hz at 300 mV. According to the Nernst equation (ERHE = EHg/HgO + 0.059 pH + 0.098), where ERHE was the potential vs. a reversible hydrogen potential, EHg/HgO was the potential vs. Hg/HgO electrode, and pH was the pH value of electrolyte. The electrochemical stability was evaluated by chronoamperometry measurements at a static overpotential, during which the current variation with time was recorded. The ECSA values were measured through CV in the selected non-faradaic range. The current densities have a linear relationship against different scan rates (10–60 mV/s) and the values of the slop were considered as twice of Cdl. The Faraday efficiency of the as-fabricated electrodes was determined by the water drainage method, which can be found in Supplementary Note 2 in Supplementary Information for details. The OER tests were performed in O2-saturated 1.0 M KOH solution, and the others are the same as HER test conditions. The overall-water splitting performance was characterized in 1.0 M KOH using a two-electrode configuration, and the polarization curve was recorded at a scan rate of 5 mV/s. In order to better compare, Pt/C and IrO2 ink were also synthesized by placing 8 mg Pt/C and 8 mg IrO2 powder in the mixture of 700 µL ethanol, 300 µL deionized water, and 50 µL Nafion followed by ultrasonication for 30 min, respectively. Then the as-obtained ink was coated onto the carbon cloth (CC) with the loading mass density of about 3.0 mg/cm2 and was then dried at 60 °C. The long-term stability measurements were carried out using the chronoamperometry measurements. All polarization curves at 5 mV/s were corrected without iR-compensation.

DFT theoretical calculation

Model building

According to the HRTEM micrographs, 1Tphase-MoS2, 2Hphase-MoS2, Ni2P, and NiS2 formed a multiphase heterojunction. 1Tphase-MoS2@Ni2P interface, 1Tphase-MoS2@NiS2 interface, 2Hphase-MoS2@Ni2P interface, 2Hphase-MoS2@NiS2 interface, bulk 1Tphase-MoS2, 2Hphase-MoS2 were also constructed as comparisons. Considering the Van der Waals forces between the two phases, the unrelaxed heterojunction interface distance was set to 3.0 Å. These original structures were obtained from Materials Project Database69.

Computational parameters

DFT calculation was applied to calculate electronic structures of two crystal structures by the partial augmented plane-wave method (PAW) implemented in the VASP70 using VASPKIT code for post-processing. Considering the heterojunction structure, the long-range force correction was considered by using the DFT-D3 correction method of Grimme71. The Perdew–Burke–Ernzerhof (PBE) generalized gradient approximation72 was implemented for exchange-correlation energy calculations using 550 eV kinetic energy cut off for the plane-wave basis. Then structural optimizations using a conjugate gradient (CG) method based on the pre-optimized structure were repeated until the maximum force component on each atom remained below 0.01 eV/Å. Monkhorst-Pack k-point meshes in the first Brillouin zone of the primitive cell were used the VASPKIT code recommended accuracy levels of 0.04 for the optimization calculation and 0.02 for the static calculation, respectively. After fully relaxing the structures, one final (electronic scf) step with the tetrahedron method using Blöchl corrections and denser k-meshes was employed for DOS calculation. In addition to the H adsorbed energy calculations, the frequency calculation of free H and free-energy correction at 298.15 K (including the entropy and zero-point energy contributions) were also calculated. To avoid abnormal entropy contribution, frequencies < 50 cm−1 are set to be 50 cm−1.

XANES spectra measurements

The Mo L3-edge, P, S, and Ni K-edge XANES spectra were measured at the BL8 beamline of Synchrotron Light Research Institute (SLRI), Thailand73. The SLRI storage ring was operated at 1.2 GeV with an electron current of 80–150 mA. The incident X-ray beam was monochromatized with a double-crystal monochromator equipped with InSb (111) and Ge (220) crystals. XANES measurements were carried out in air at Ni K-edge and under He atmosphere at the lower edges on as-prepared catalysts embedded on carbon cloth and those used as the working electrode in electrochemical reaction with 1 M KOH solution. Cathode voltages from −0.04 V to −0.2 V vs. RHE (ERHE = EAg/AgCl + 0.059 pH + 0.197) were applied for 160 s using an electrochemical workstation (Autolab PGSTAT204) before the XANES experiment. All XANES spectra were collected in fluorescence-yield mode using a 13-element Si drift detector. Ni and foils, and elemental S and P were used for photon energy calibration. The edge position was defined as the point corresponding to the maximum value in the derivative curves of the XANES spectra. Data normalization was carried out using the Athena software74.