A universal pH range and a highly efficient Mo2C-based electrocatalyst for the hydrogen evolution reaction

We designed a series of transition metal-modified molybdenum carbides in nitrogen doped carbon matrix for investigating and boosting the performance of Mo2C-based electrocatalysts in all-pH media by increasing the Mo2+ content in catalysts.


Introduction
Owing to its high energy density and ability to be generated from renewable sources of energy, hydrogen (H 2 ) is a promising alternative to fossil fuels to meet future global energy demands. [1][2][3] Electrochemical water splitting is considered to be one of the most effective means of 'green' hydrogen generation. [4][5][6][7] The production of hydrogen from water is a two-electron process, 2H + + 2e À ¼ H 2 , the electro-kinetics of which is highly pH dependent. Industrial production of hydrogen by water electrolysis can take place in acid or alkaline media, each having its relative merits and challenges, with the effect on the kinetics of the complementary oxygen evolution reaction (OER) playing an important role. 8,9 Ruthenium (Ru)-based materials for the OER and HER are very promising because of their prominent catalytic activity and universal pH application. Various modication strategies on Ru-based materials have been summarized and the general description about water splitting has been presented to understand the reaction mechanism, and key stability issues for Ru-based materials have been further investigated. 10,11 Additionally, it is desirable to achieve effective HER performance in neutral electrolytes for sea-water and microbial electrolysis. [12][13][14] Therefore, to meet practical requirements, an ideal HER electrocatalyst should operate effectively over the entire pH range. Pt-based materials are the benchmark for the HER in acid media owing to their low overpotential and suitable hydrogen adsorption Gibbs free energy; however, their cost and scarcity are a barrier to industrial use. 15,16 Therefore, extensive effort has been devoted to the development of hydrogen evolution catalysts based on noble-metal-free and earth-abundant resources. 17 Mo-based catalysts are regarded as favorable candidates for the HER due to the modication of the d-band structure induced by the formation of a metal-carbon bond. 18,19 In the past few decades, various Mo-based materials, such as molybdenum sulphides, 20-23 oxides, 24 nitrides, [25][26][27] phosphides, [28][29][30] and carbides, [31][32][33] have been emerging as prospective electrocatalysts for hydrogen production. 2,34,35 Representatively, molybdenum carbide (Mo 2 C) has received wide attention and been intensively investigated as an HER catalyst on account of its abundance, low toxicity, similar Fermi level and d-electronic structure to Pt, and high resistance in various electrolytes. [36][37][38] Particularly, b-phase molybdenum carbide (b-Mo 2 C) has demonstrated the highest activity among different crystalline phases of molybdenum carbides. 36 Previous research has shown that Mo 2 C has a relatively undesirable hydrogen binding energy (DG H * ) due to Mo atoms with a high density of an empty dorbital; thus, it is easy to form strong Mo-H bonds on a Mo 2 C surface which is unfavorable for the desorption of active hydrogen, making the Heyrovsky/Tafel step partially restricted. 19,28,39,40 Many modication methods to optimize the activity of molybdenum carbide have been summarized; specic nanostructures and carbon-based hybrids are introduced to increase the active sites and promote mass transfer, and heteroatom doping is encouraged to improve the chemical congurations of active sites towards intrinsically boosted HER kinetics. 41,42 Elemental doping has been established to be a reliable method to regulate the electronic structure of materials. 43 Recently, the modication of Mo 2 C-based materials, especially via introducing different elements into the Mo 2 C lattice, has been widely explored. 44,45 This method strongly affected their HER properties by regulating the electron density of such materials. Ma et al. reported porous molybdenum carbide (b-Mo 2 C) spheres with novel surface engineered structures (coexistence of Co doping and Mo vacancies) to enhance the HER performance over the b-Mo 2 C-based catalyst surface and tested their hydrogen evolution activity in acidic media. 46 Wan et al. reported a wide range of magnetic iron-doped molybdenum carbide (Mo 2Àx Fe x C) nanomaterials synthesised by the decomposition of a novel amine-metal oxide composite. They evaluated their hydrogen evolution activity in an acid environment, which required a $300 mV overpotential to achieve a current density of 10 mA cm À2 . 47 Li et al. reported a twodimensional coupled hybrid of molybdenum carbide and reduced graphene oxide with a ternary polyoxometalatepolypyrrole/reduced graphene oxide nanocomposite as a precursor and tested its hydrogen evolution activity in acidic media. 48 Xiong et al. reported Ni-doped Mo 2 C nanowires directly grown on Ni foam via a hydrothermal reaction combined with a carbonization process and tested their hydrogen evolution activity in alkaline media. 49 Furthermore, Mo 2+ sites have been proven to be important for promoting catalytic hydrogen evolution, 50,51 but their abundance among the Mo 2 C structures is difficult to control. Even though great effort has been made to the improvement of Mo 2 C-based catalyst performance, tuning of properties for effective operation in neutral electrolytes, and further expanding the range of applications, is seldom achieved. Thus, a strategy to optimise the structure and electronic properties of Mo 2 C-based materials is urgently required to implement these materials as an efficient noble metal-free 'pH-universal' HER catalyst.
Herein, we develop a facile annealing strategy to synthesise a series of transition metal (TM)-doped molybdenum carbide modied within porous nitrogen-doped carbon frameworks (TM-Mo 2 C@NCF). Among them, iron-doped molybdenum carbide (Fe-Mo 2 C) showing the most abundant Mo 2+ active species exhibits the highest catalytic HER activity and long-term stability for the HER under all media, thus making it easy to couple with any suitable OER catalysts in different electrolytes. This is ascribed to the merits of high intrinsic activity from the combination of Fe 2+ and Mo 2+ in Mo 2 C, which exposes more Mo-Mo bridge active sites, as well as the stabilising effect of the nitrogen-doped carbon frameworks. Density functional theory (DFT) calculations are also applied to elucidate the underlying mechanism behind the distinct performance by comparing the electronic structure, hydrogen chemisorption process and Gibbs free energy of pure Mo 2 C and Fe-Mo 2 C@NCF.

Synthesis of TM-Mo 2 C@NCF
The preparation of the various TM-Mo 2 C@NCF materials all followed a similar procedure. The amount of C 2 H 2 O 4 $2H 2 O and melamine remained unchanged with a molar ratio of 2 : 1. Typically, Fe-Mo 2 C@NCF was synthesised as follows. Firstly, 2.48 g (2 mmol) of (NH 4 ) 6 Mo 7 O 24 $4H 2 O and 2.52 g (0.02 mol) of C 2 H 2 O 4 $2H 2 O were dissolved in 40 mL of deionized water. 1.26 g (0.01 mol) of melamine was added to the above clear solution, and then the mixture was kept at 70 C for 6 h under magnetic stirring. Then, 5 mL of Fe 2 (SO 4 ) 3 solution (0.07 mmol of Fe 3+ ) was added into the above mixed solution, stirring at 70 C for 12 h progressively. Aer ltration, the precipitate was washed with deionized water and ethanol, then dried under vacuum at 60 C. During this process, a composite containing Fe salts, oxalate and molybdate was formed, which was used as the precursor. Subsequently, the precursor was calcined at 800 C for 3 h with a heating rate of 5 C min À1 under a continuous N 2 ow. To investigate the effects of different doped metals (Mn/Fe/Co/Ni) on the hydrogen evolution performance, Mn/Co/Ni-doped materials were prepared according to Table S1 in the ESI. † The obtained products were denoted as sample xTM-Mo 2 C@NCF, where x is the molar ratio of TM : Mo, respectively.

Electrocatalytic HER
The electrochemical measurements are carried out at room temperature using a CHI 660E potentiostat. Ag/AgCl (3 M KCl) and a graphite rod are used as a reference electrode (RE) and a counter electrode (CE) respectively. According to the formula E (RHE) ¼ E (Ag/AgCl) + 0.0591 Â pH + 0.2 V, all potentials can be converted to the reversible hydrogen electrode (RHE). The catalyst ink can be prepared by mixing a solution containing 190 mL of distilled water, 50 mL of ethanol, 5 wt% Naon solution (10 mL) and the catalyst (2.0 mg) aer an hour of sonication in an ice water bath. Thereaer, the surface of carbon paper (coated area: 0.5 Â 1 cm 2 ) is loaded with 100 mL of the prepared catalyst ink, wherein 1.6 mg cm À2 is calculated to be the loading amount. To compare the properties of the as-prepared catalysts, linear sweep voltammetry (LSV) is conducted at a scan rate of 5 mV s À1 . A commercial 20 wt% Pt/C catalyst is used as the reference material for comparison of performance. The polarization curves were plotted as overpotential (h) versus log current (log J) to get Tafel plots for evaluating the HER kinetics. The Tafel slope (b) as an intrinsic property of the catalyst was obtained by tting the linear portion of Tafel plots according to the Tafel equation (h ¼ b log(J) + a) that can describe the ratedetermining step of the HER process. Regarding the evaluation of the electrochemically active surface areas (ECSAs) of the catalysts, CV tests were performed by measuring the electrical double-layer capacitances (C dl ) with different scan rates (10,20,30,40,50,60,70,80, 90 and 100 mV s À1 ). The chronoamperometry measurements of 0.5% Fe-Mo 2 C@NCF are conducted at an overpotential of 105 mV and negligible degradation can be observed aer a 12 h process. In addition, aer continuous CV scans at a scan rate of 100 mV s À1 , the overpotential differences of the polarization curves before and aer the 1000 CV cycles between À0.1 and 0.1 V (vs. Ag/AgCl) were compared. Note that all the as-measured current densities were corrected to reduce the effect of ohmic resistance.

Characterization
XRD patterns were recorded using a Bruker D8 Advance X-ray diffractometer with a Cu Ka source. The 2q angular region of the XRD pattern between 20 and 80 was studied at a scan rate of 0.1 s À1 with each step of 0.02 . The microstructures and compositions of samples were analyzed using scanning electron microscopy (SEM, Zeiss/Auriga Focused ion beam SEM), transmission electron microscopy and high-resolution transmission electron microscopy (TEM and HRTEM, FEI TalosF200S). X-ray photoelectron spectroscopy (XPS) measurements were performed on an ESCALAB 250 instrument (Thermo Electron) with Al Ka radiation. Raman spectra were collected using a LabRAM HR Evolution Laser Raman Spectrometer with a 532 nm laser diode as the excitation source. Brunauer-Emmett-Teller (BET) surface area measurements were obtained using a Micromeritics ASAP2460 with nitrogen adsorption at 77k. The amount of doped metal (Mn, Fe, Co, and Ni) in the samples was determined by ICPAES (ICPE-9820, SHIMADZU).

Results and discussion
Fe-Mo 2 C@NCF nanostructures were synthesised by the solidstate thermolysis of a melamine-related composite of oxalate, Mo sources and Fe precursors. As for a typical synthesis, 0.5% Fe-Mo 2 C@NCF was produced from annealing the precursor containing Fe salts, oxalate and molybdate. The crystalline phases of xFe-Mo 2 C@NCF characterised via XRD are shown in Fig. 1a 55 No other diffraction peaks of impurities were detected and the as-prepared catalysts all display a high crystallinity. However, the patterns of b-Mo 2 C in the 0.5% Fe-Mo 2 -C@NCF catalyst shows a slight shi to a higher diffraction region, in comparison with pure Mo 2 C (Fig. S2, ESI †). This shi of the diffraction peaks is attributed to the shrinkage of the Mo 2 C unit cell because the radius of iron is smaller than that of molybdenum. The absolute values of lattice parameters and unit cell volumes for undoped and Fe modied samples are listed in Table S2. † Both lattice parameters (a/b and c) and unit cell volume follow similar trends, decreasing with the increase of Fe doping amount from 0 to 0.31%, and the replacement of some Mo atoms with smaller Fe in the lattice contributes to the shrinking of the Mo 2 C unit cell. The Mn/Fe/Co/Ni doping amount of all the as-prepared catalysts is determined by inductively coupled plasma atomic emission spectroscopy (ICPAES) ( Table S3, ESI †). Moreover, the BET surface area of 0.5% Fe-Mo 2 C@NCF (29.5 m 2 g À1 ) is higher than that of undoped Mo 2 C (23.9 m 2 g À1 ), which could increase the active sites.
To identify the chemical environment of xFe-Mo 2 C@NCF on the surface, XPS analysis was conducted. Fig. S3 † shows the full XPS proles of 0.5% Mn-Mo 2 C@NCF, 0.5% Fe-Mo 2 C@NCF, 0.5% Co-Mo 2 C@NCF and 0.5% Ni-Mo 2 C@NCF. From the XPS survey analysis shown in Fig. S3b, † the surface of 0.5% Fe-Mo 2 C@NCF is composed of Mo, N, C and O. There is no obvious Fe peak because the doping content of Fe is relatively low. The peak tting of Mo 3d veries the existence of Mo 2 C (Mo 2+ at 228.5 and 231.7 eV). Moreover, Mo 4+ at 228.9 and 235.6 eV and Mo 6+ at 232.7 and 236.1 eV originate from surface oxidation (Fig. 1b). The presence of higher oxidation states from XPS is acceptable owing to the unavoidable surface oxidation layers of carbide materials, particularly for the nanosized ones, which are easily oxidized on the surface. 56 Similarly, Mo 3d spectra of 0.5% Mn-Mo 2 C@NCF, 0.5% Co-Mo 2 C@NCF and 0.5% Ni-Mo 2 -C@NCF are displayed in Fig. S4. † As can be seen from Fig. 1c, the N 1s XPS spectrum is deconvoluted into ve peaks at 394.57, 396.21, 398.23, 399.84 and 401.49 eV, which correspond to Mo 3p, Mo-N, pyridinic-N, pyrrolic-N, and graphitic-N, respectively, implying N-doping in the carbon matrix and the joint binding of Mo with the carbon skeleton. 35,57,58 In particular, pyridinic-N exhibits the highest peak from the spectrum and it was veri-ed as advantageous for enhancing the catalytic performance of water splitting. 59 The increase of the Mo 2+ content should be attributed to the addition of transition metals, which is proved to be benecial for the HER. 60 By comparing the peak area of Mo 2+ in materials with different molar ratios of Fe/Mo (Fig. 1d) and different types of TM doping (Fig. S4d, ESI †), it is obvious that Mo 2+ species are the most abundant when the molar ratio of Fe to Mo is 0.5 : 100, according to the XPS analysis, which indicates that doped iron is combined tightly with Mo atoms that leads to an increase of the Mo 2+ content in the molybdenum carbide catalysts. The increased Mo 2+ species further enrich the active sites on the surface of the materials, and also improve the electron density, which has a positive effect on the hydrogen evolution catalytic properties. Fig. S5 and S6 † show scanning electron microscope (SEM) and TEM images of 0.5% Mn-Mo 2 C@NCF, 0.5% Fe-Mo 2 -C@NCF, 0.5% Co-Mo 2 C@NCF and 0.5% Ni-Mo 2 C@NCF, respectively. It can be seen that the morphology of these materials has negligible differences because all materials were synthesised under the same conditions, except using different TM precursors. In particular, 0.5% Fe-Mo 2 C@NCF has a rougher surface, which consists of nanoparticles with several pits and is benecial to exposing more active species. TEM images of both pure b-Mo 2 C (Fig. 2a) and 0.5% Fe-Mo 2 C@NCF (Fig. 2d) show the analogous nanoparticle shape with the particle size of about 15 nm. The d-spacing (2.3Å, Fig. 2b) and SAED (Fig. 2c) pattern are consistent with the (101) plane of b-Mo 2 C. In Fig. 2e, b-Mo 2 C nanoparticles are wrapped by amorphous carbon. The existence of graphitic carbon has also been conrmed by the XRD patterns in Fig. 1a. The presence of a graphitic carbon support is another crucial factor to improve catalytic properties for the entire system. 60-62 Raman spectroscopy was performed to reveal the carbon structure of the catalysts by the ratio of I D /I G . The Raman spectra (Fig. S6, ESI †) exhibit double peaks at around 1347 cm À1 and 1593 cm À1 , where the D-band (1347 cm À1 ) was associated with sp 2 nanocrystalline carbon while the G-band (1593 cm À1 ) corresponded to sp 2 amorphous carbon materials, the large I D /I G implying that more defects formed on the NCF, thus favoring the accessibility of more active sites and enhancing the catalytic properties. EDS mapping via HRTEM in Fig. 2g-i proves that Mo, Fe and N are uniformly distributed and overlapped in the particles.
As for the development of electrocatalytic technology, catalysts that are only suitable for acidic environments limit the range of practical applicability. [63][64][65][66] For example, microbial electrolysis and alkaline electrolysis need catalysts to be efficient in neutral and alkaline media, respectively. 64,65 Therefore, more electrocatalysts with high activity and stability in alkaline or neutral media need to be developed. 67 To get a more comprehensive evaluation of the electrocatalytic HER activity, the performance was measured in alkaline solution (1.0 M KOH, pH ¼ 13.9), 1.0 M phosphate buffer solution (pH ¼ 7) and acid solution (0.5 M H 2 SO 4 , pH ¼ 0.1) with a xed mass loading of 1.6 mg cm À2 on the carbon paper. For a comparative study, Pt/C (20 wt%) was also examined for HER activity.
To obtain further insight into the HER activities of the electrocatalysts, the kinetic parameters evaluating the performance are summarised in Table S2 and Fig. S7. † The electrocatalytic activity of the materials was primarily investigated in alkaline media. In alkaline electrolyte, the 0.5% Fe-Mo 2 C@NCF afforded a relatively low h of 65 mV to reach the current density of 10 mA cm À2 and 172 mV for 100 mA cm À2 (Fig. 3a), which is 16 and 244 mV less than that for undoped Mo 2 C, respectively. For highly active catalysts, the cathode current rises rapidly when a more negative potential is applied. Comparatively, 0.2% Fe-Mo 2 C@NCF, 1% Fe-Mo 2 C@NCF and undoped Mo 2 C show higher overpotentials of 199, 235 and 416 mV at a current density of 100 mA cm À2 , respectively; indicating that 0.5% Fe-Mo 2 C@NCF exhibits the highest activity in alkaline environments. Consistent with the XPS results, the 0.5% Fe doping content results in the most abundant Mo 2+ species on the surface of Mo 2 C, which exposes more active sites, and reaches the peak of the Fe-N-Mo cooperative coordination and shows the best HER activity. 68 For the integral HER process, the reaction step is the combination of the Volmer reaction and the Heyrovsky or Tafel reaction to generate H 2 , and generally referred to as the Volmer-Heyrovsky or Volmer-Tafel process. By plotting overpotential (h) against log current (log J), the Tafel slope was obtained and kinetic parameters for the HER were calculated. The linear part of the Tafel plots (Fig. 3b) reveals Tafel slopes of $80, $76, $141 and $178 mV dec À1 for 0.2% Fe-Mo 2 C@NCF, 0.5% Fe-Mo 2 C@NCF, 1% Fe-Mo 2 C@NCF and undoped Mo 2 C, respectively. This illustrates that hydrogen evolution on 0.5% Fe-Mo 2 C@NCF proceeds via the Volmer-Heyrovsky mechanism and that the electrochemical desorption step is rate-limiting. Compared with other electrocatalysts, except Pt/C, the exchange current density of 0.5% Fe-Mo 2 -C@NCF is the highest ($1.02 mA cm À2 ). It is logical to ascribe the increased activity to the larger electrochemically active surface area (ECSA). One method to estimate the ECSA is to measure the double-layer capacitance (C dl ). 69,70 Based on the cyclic voltammograms (CVs) at different scan rates (Fig. S8, ESI †) in the region of +0.81 and +0.86 V (vs. Ag/AgCl), where the current response is due to the charging of the double-layer capacitance (non-faradaic current region), the capacitances of 0.2% Fe-Mo 2 C@NCF, 0.5% Fe-Mo 2 C@NCF, 1% Fe-Mo 2 C@NCF and undoped Mo 2 C are calculated to be 15.9, 21.4, 13.8 and 8.1 mF cm À2 , respectively. This reveals the highest ECSA, and consequently the greatest number of catalytic active sites, for 0.5% Fe-Mo 2 C@NCF (Fig. 3c). As a potentially effective candidate for the HER, the stability of a catalyst is also a key index. The durability of the 0.5% Fe-Mo 2 C@NCF electrode was tested by conducting continuous cyclic voltammetry (CV) scanning between À0.1 V and 0 V vs. RHE at a scan rate of 10 mV s À1 in 1.0 M KOH. Aer 1000 cycles, the 0.5% Fe-Mo 2 C@NCF still performs efficiently with negligible loss of cathodic current density (Fig. 3d). Moreover, the chronoamperometric test of this electrode by electrolysis at a xed overpotential of 105 mV also suggests its good durability (Fig. 3d).
It is of importance to mention that the 0.5% Fe-Mo 2 C@NCF electrode also performs well in 1.0 M PBS for the HER. It is clear that pure Mo 2 C has poor catalytic activity towards the HER. The polarization curves of 0.5% Fe-Mo 2 C@NCF in 1.0 M PBS are shown in Fig. 4a, which demonstrate an overpotential of 130 mV vs. RHE to reach a current density of 10 mA cm À2 and relatively low overpotential of 275 mV to reach 100 mA cm À2 . In contrast, 0.2% Fe-Mo 2 C@NCF, 1% Fe-Mo 2 C@NCF and undoped Mo 2 C, showed higher overpotentials of 406, 378 and 484 mV, respectively, at a current density of 100 mA cm À2 in buffer solution. Fig. 4b shows the Tafel plots of 0.2% Fe-Mo 2 C@NCF, 0.5% Fe-Mo 2 C@NCF, 1% Fe-Mo 2 C@NCF and undoped Mo 2 C, which are consistent with LSV test results that 0.5% Fe-Mo 2 C@NCF provided the smallest Tafel slope (109 mV dec À1 ) compared to other electrodes. It is noted that the 0.5% Fe-Mo 2 C@NCF electrode maintains long-term durability for the HER in 1.0 M PBS (Fig. 4g); the polarization curves of 0.5% Fe-Mo 2 C@NCF exhibit a subtle decay aer 1000 cycles of continuous cyclic voltammograms, illustrating the stability of the prepared catalysts and the resistance to accelerated degradation. The HER performances for the above catalysts were also tested in 0.5 M H 2 SO 4 . For 0.5% Fe-Mo 2 C@NCF, it requires an overpotential of 129 mV to drive 10 mA cm À2 and 242 mV for 100 mA cm À2 (Fig. 4d), which is much smaller than that of undoped Mo 2 C at 10 mA cm À2 (162 mV). 0.5% Fe-Mo 2 C@NCF shows a lower Tafel slope (110 mV dec À1 ) than pure Mo 2 C (131 mV dec À1 ), as shown in Fig. 4e. Its stability in acid media was also evaluated. As observed, there is negligible loss in current density aer 1000 continuous CV cycles (Fig. 4h), and its activity can be maintained for at least 12 h in both neutral and acidic media (Fig. 4i).
The hydrogen evolution properties of three groups of materials doped with manganese, cobalt, and nickel atoms in different proportions were also tested in acidic, neutral, and alkaline media (Fig. S9-S11 and Tables S6-S8, ESI †). Comparing the performance of the most active materials in the other three groups (0.5% Mn-Mo 2 C@NCF, 0.5% Co-Mo 2 C@NCF, and 0.5%  Ni-Mo 2 C@NCF) with 0.5% Fe-Mo 2 C@NCF, as shown in Fig. 5, it can be seen that the performance of 0.5% Fe-Mo 2 C@NCF in three different pH media is still the best and attributed to the more abundant Mo 2+ active sites. As further compared with the current noble-metal-free catalysts (Tables S9 and S10, ESI †), the 0.5% Fe-Mo 2 C@NCF material shows the best-in-class performance.
To gain further insight into the high electrocatalytic activity of the 0.5% Fe-Mo 2 C@NCF electrocatalyst, DFT calculations were performed by constructing correlative theoretical models (Fig. S12-S19 †). The adsorption free energy of H (DG H *) can serve as an accepted measure for the activity of a catalytic material for the HER, for which a catalyst that gives DG H * close to 0 is considered a good candidate for the HER. The calculations of DG H * for 0.5% Fe-Mo 2 C@NCF are described in Fig. 6. For comparison, a similar calculation was performed on pristine Mo 2 C and 1% Fe-Mo 2 C@NCF. The inuence of Fe dopants is investigated by replacing Mo atoms by Fe atoms. Four different doping sites for single Fe dopants are considered (Fig. S13 †) to simulate the 0.5% Fe-Mo 2 C@NCF material. The structure with the lowest formation energy is used for the following HER study ( Fig. S13d and S14 †). Eight different initial H* adsorption sites are considered and ve different adsorption models were obtained aer relaxation, as shown in Fig. S15. † The site with the strongest binding between H* is shown in Fig. S15c and S16. † The limiting potential of the HER on 0.5% Fe-Mo 2 C@NCF is 0.39 V (Fig. 6a), which shows a better HER catalytic activity than pristine Mo 2 C, and the HER catalytic activity decreases with a higher Fe concentration (two Fe dopants, 0.46 V). This is in agreement with the experimental results discussed above. In fact, the Mo-Mo bridge site in Fig. S13d † was found to be the most energetically favorable adsorption site for the H* intermediate in 0.5% Fe-Mo 2 C@NCF (shown in the ESI †). To reveal the mechanism of how Fe dopants enhance the HER activity on Mo 2 C, the electronic properties of the atoms on the adsorption sites have been studied. As shown in Fig. 6b, the d-band center of Mo in pristine Mo 2 C is 0.45 eV closer to the Fermi level than that of Mo 2 C:1Fe. Thus, Mo in Mo 2 C:1Fe will bind with H* weaker than that in pristine Mo 2 C and have a better HER catalytic activity. Besides *H adsorption strength, water dissociation is also an important step for HER activity under alkaline conditions. Our computational results indicate that the water dissociation step is an exothermic reaction for both Mo 2 C and Mo 2 C:1Fe, while they are more energetically favorable on Mo 2 C:1Fe as shown in Fig. S21. † Considering *H adsorption and water dissociation, we can conclude that Fe doping is an effective approach to enhance the HER activity under alkaline conditions. This result reinforces the presence of the synergy between Mo 2 C and Fe dopants, providing outstanding catalytic activity to the Mo-Mo bridge site in Mo 2 C and making 0.5% Fe-Mo 2 C@NCF a highly efficient HER catalyst.

Conclusions
In conclusion, Fe-doped molybdenum carbide on a nitrogendoped carbon substrate (Fe-Mo 2 C@NCF) with a loosely linked and channel-rich structure was prepared and proven to be a highly efficient electrocatalyst for the hydrogen evolution reaction over a wide pH range. Benetting from the exposure of abundant active sites and excellent structural stability, the Fe-Mo 2 C@NCF exhibits enhanced activity for the HER, such as a low h 10 of 65, 130 and 129 mV as well as h 100 of 172, 275 and 242 mV dec À1 in 1.0 M KOH, 1.0 M PBS and 0.5 M H 2 SO 4 , respectively. In addition, the synergistic catalytic effects between the N-doped carbon substrate and Fe-doped Mo 2 C nanoparticles were realised, which can boost the continuous electron transport between metal dopants and transition metal carbide, optimise active species and enhance the corrosion resistance and stability of the catalyst.

Conflicts of interest
There is no conicts of interest.