A Low‐Temperature Synthetic Route Toward a High‐Entropy 2D Hexernary Transition Metal Dichalcogenide for Hydrogen Evolution Electrocatalysis

Abstract High‐entropy (HE) metal chalcogenides are a class of materials that have great potential in applications such as thermoelectrics and electrocatalysis. Layered 2D transition‐metal dichalcogenides (TMDCs) are a sub‐class of high entropy metal chalcogenides that have received little attention to date as their preparation currently involves complicated, energy‐intensive, or hazardous synthetic steps. To address this, a low‐temperature (500 °C) and rapid (1 h) single source precursor approach is successfully adopted to synthesize the hexernary high‐entropy metal disulfide (MoWReMnCr)S2. (MoWReMnCr)S2 powders are characterized by powder X‐ray diffraction (pXRD) and Raman spectroscopy, which confirmed that the material is comprised predominantly of a hexagonal phase. The surface oxidation states and elemental compositions are studied by X‐ray photoelectron spectroscopy (XPS) whilst the bulk morphology and elemental stoichiometry with spatial distribution is determined by scanning electron microscopy (SEM) with elemental mapping information acquired from energy‐dispersive X‐ray (EDX) spectroscopy. The bulk, layered material is subsequently exfoliated to ultra‐thin, several‐layer 2D nanosheets by liquid‐phase exfoliation (LPE). The resulting few‐layer HE (MoWReMnCr)S2 nanosheets are found to contain a homogeneous elemental distribution of metals at the nanoscale by high angle annular dark field‐scanning transmission electron microscopy (HAADF‐STEM) with EDX mapping. Finally, (MoWReMnCr)S2 is demonstrated as a hydrogen evolution electrocatalyst and compared to 2H‐MoS2 synthesized using the molecular precursor approach. (MoWReMnCr)S2 with 20% w/w of high‐conductivity carbon black displays a low overpotential of 229 mV in 0.5 M H2SO4 to reach a current density of 10 mA cm−2, which is much lower than the overpotential of 362 mV for MoS2. From density functional theory calculations, it is hypothesised that the enhanced catalytic activity is due to activation of the basal plane upon incorporation of other elements into the 2H‐MoS2 structure, in particular, the first row TMs Cr and Mn.

: Electrochemical double-layer capacitance of the MoS 2 @20% CB and HEDS@20% CB electrodes Table S6: Comparison oft he HER performance of previously-reported MoS 2 -based materials and the HES@20% CB material in an acidic medium Table S7: Summary of the surface-slab models examined in the modelling study

A note on oxidation of the Mn precursor
The Mn precursor was initially synthesised using 2 equivalents of diethyldithiocarbamate ligands to balance the charge of the starting Mn salt. The product was analysed with elemental analysis and found to fit with the anticipated Mn(DTC) 2 product. However, this species rapidly oxidises to Mn(DTC) 3 . 1-3 Indeed, many synthesis methods have been reported which report the use of Mn(DTC) 2 , but we would like to point out that these have been wrongly reported and are likely to be Mn(DTC) 3 in fact. 3 In our study, elemental analysis was found to fit Mn(DTC) 2 , which did not align with literature, in particular the studies by Eagle et al and Hendrickson et al. 1,2 This prompted further investigation into the exact complex which we made. The magnetic susceptibility (μ eff ) was measured. The 'Mn(DTC) 2 ' was found to have a μ eff of 5.34 BM. This value is far from the expected value for tetragonal Mn(DTC) 2 of 5.92 BM or oxtahedral Mn(DTC) 3 4.9 BM for the expected high spin complexes, respectively, indicating that this is not the species present.
Crystals grown and solved were found to be a new species with an [O 2 ] 2bridging dimer ( Figure S2), this appears to inadvertently be the originally synthesised (and pure) 'Mn(DTC) 2 ' S7 as the elemental analysis results are found to be an excellent fit: Found: C: 34.3%, H: 5.6%, N: 7.4%. Expected for Mn 2 O 2 (DTC) 4 0.75(CH 3 ) 2 CO: C: 34.3%, H: 5.7%, N: 7.2%. The proposed synthetic mechanism for formation of this species is shown below ( Figure S1).

Synthesis of CrL 3 (5)
Tris(diethyldithiocarbamato) chromium (Figure S1(e)) was synthesised following a previously-reported literature procedure. [4] Chromium trichloride hexahydrate (2.0 g, 9 mmol) was dissolved in deionized water (250 mL) to form a green solution. The monosodium salt of diethyl dithiocarbamate (12.0 g, 54 mmol) was added in the solution to generate a blue precipitate, which was collected by vacuum filtration to obtain crude CrL 3 . The crude product was purified using column chromatography on silica, eluting with dichloromethane or chloroform, and collecting the rapidly-eluted blue band. A vacuum evaporator was used to isolate the ultramarine solid from the solvent before drying in vacuum oven overnight. Anal.  Ref.

X-Ray Photoelectron Spectroscopy:
The Mo 3d spectra (Figure S8(a) & S9(a)) show three chemical species, which can be attributed to Mo 4+ from MoS 2 and Mo 6+ from MoO 3 . [11] The third chemical species is attributed to another sulfide environment. It is noteworthy that the peaks at 232.8 eV (MoO 3 ) and 229.6 eV (MoS 2 ) in MoS 2 shifted to 232.2 eV and 228.6 eV, respectively in the high-entropy disulfide (HEDS). This shift indicates that the chemical environment of Mo is altered by the introduction of the other metals. [12] The fitted S 2p spectra (Figure S8 attributed to the Mn 2p 1/2 and 2p 3/2 species, respectively. [15] Similarly, Cr 2p 1/2 and Cr2p 3/2 species were observed in the Cr 2p spectra ( Figure S9(d)) with peaks at 584.2 and 575.3 eV, respectively. [16] The valence states of Mn and Cr could not be accurately quantified due to the complex chemical surroundings. [17] There are two characteristic doublet peaks in the W 4f spectra ( Figure S9(e)). The peak at 32.2 eV can be attributed to W 4+ from WS 2 and 35.2 eV to the W 6+ of WO 3 . The core-level peaks in the Re 4f spectrum (Figure S9(f)) are located at 41.4 eV for ReS 2 , [14] and 48.8 eV for Re 2 O 7 . A further sulfide environment was found for Re at 42.3 eV. The presence of oxide species for most of the metals is not uncommon for metal-sulphide materials, [18] but to establish whether these species arise from surface oxidation, or whether the oxides are present as bulk materials, additional characterization techniques were used. S15 Figure S8. Adventitious carbon and oxygen contaminants are usually observed on samples that have been exposed to air, however the O 1s peak could also result from a minor amount of surface oxidation in air of the as-prepared samples. The C 1s peak at 284.8 eV was used to calibrate the binding energy scale. [2,18] S17

Density Functional Theory Calculations
To investigate the impact of alloying on the catalytic activity of the HE MoS 2 nanoparticles, we performed density-functional theory (DFT) calculations on a series of slab models of the basal plane (001) surface. The 2H structure of bulk MoS 2 was taken from the Materials Project database and fully optimised. The 2H structure comprises MoS 2 layers stacked along the axis and separated by a van der Waal's gap (Figure S17(a)). This bulk structure was used to create a slab model consisting of 3 × 3 unit cells along the and directions and five layers stacked along the direction with a 15 Å gap between periodic images ( Figure S17 (Table S7). Across the five sets of models we considered a total of 130 configurations from which 19 were selected (Figures S19(a)-S37(a)).
For each selected configuration we calculated the electronic density of states (DoS), including the projections onto the different atomic species, in order to assess the impact of the substitutions on the electronic structure (Figures S38-S56). To determine whether the substitutions affect the relative positions of the valence-and conduction-band edges, we also We calculated the H binding energies for our pristine MoS 2 surface and the 19 substituted surfaces (Figure S18(a), Figures S19(a)-S37(a)). For each model, a H atom was placed at an initial distance of 2 Å above each of the symmetry-unique surface sites and the resulting models optimised. In total, we considered 266 initial configurations across pristine MoS 2 and the 19 substituted models. The ads were then calculated from the energies of the optimised slab models with and without adsorbed H and the energy of a reference gas-phase H 2 molecule. This procedure gives us a range of H binding energies, which are displayed in Figure 5 in the main text and summarised in Table S7. The optimised models of the MoS 2 slab with H atoms adsorbed at the two unique surface sites are shown in Figures S17(b) and S18(c), and the H adsorption configurations with the lowest energies for each of the substituted surfaces are shown in Figures S19(b)-S37(b). We note that the positions of the H atoms were not constrained during the optimisations, so it was possible for H atoms placed initially above one surface site to move to a different (e.g. neighbouring) site during the structural relaxation.
All modelling was performed using pseudopotential plane-wave density-functional theory (DFT), as implemented in the Vienna Ab initio Simulation Package (VASP) code. [39] Electron exchange and correlation were modelled using the PBE generalised-gradient approximation (GGA) exchange-correlation functional [40] with the DFT+ correction applied S28 to the transition-metal d orbitals according to the method of Dudarev et al. (i.e. PBE+ ). [41] The values were taken from the Materials Project (MP) [42] and are calibrated against redox energies according to the method in Ref. [43] : Cr -3.7 eV, Mn -3.9 eV, Mo -4.38 eV, and W -6.2 eV. The MP does not list a value for Re. Extensive testing on the singly-substituted MoS 2 + Re slab model found that DFT+ calculations with a correction of = 1-8 eV applied to the Re d orbitals were numerically unstable, and we were therefore forced to disable the correction for this element. respectively, after geometry optimisation, so these were set as the initial moments on these atoms in all other calculations. For the H-adsorption calculations we set an initial moment of = 1 BM on the H atom. For simplicity, we set an initial ferromagnetic ordering in calculations on models containing multiple atoms with non-zero initial moments. We note that we did not constrain the magnetic moments, and in some models the initial moments relaxed to different values during the geometry optimisations.
Electronic-structure calculations were performed on the bare slab models using the r 2 SCAN meta-GGA functional, [44] which we previously found to give better bandgaps than GGA+ at a more manageable computational cost than a hybrid functional such as HSE06. [45] The ion cores were modelled using projector augmented-wave (PAW) pseudopotentials [46] with the following electrons included in the valence shells: H -1s 1 , S -3s 2 3p 4 , Cr -3p 6 4s 1 3d 5 , Mn -3p 6 4s 2 3d 5 , Mo -4p 6 5s 1 4d 5 , W -5p 6 6s 2 5d 4 , Re -5p 6 6s 2 5d 5 . These pseudopotentials correspond to those recommended by the Materials Project. [42] For bulk MoS 2 , the Kohn-Sham wavefunctions were expanded in a plane-wave basis with a kinetic-energy cutoff of 500 eV and the electronic Brillouin zone was integrated using a Γ-centered Monkhorst-Pack -point mesh [47] with 10 × 10 × 2 subdivisions. These parameters were determined based on explicit convergence testing to converge the absolute S29 total energy to < 1 meV atom -1 and the cell pressure to < 1 kbar (0.1 GPa). A full geometry optimisation was performed to tolerances of 10 -8 eV on the electronic energies and 10 -2 eV Å -1 on the ionic forces. The PAW projection was performed in reciprocal space, the precision of the charge-density grids was set automatically to avoid aliasing errors, and non-spherical contributions to the gradient correction were accounted for inside the PAW spheres.
The slab models were generated from bulk MoS 2 using the Transformer code. For these models we used the same cutoff but a slightly reduced -point sampling of 2 × 2 × 1 as opposed to the 4 × 4 × 1 or 3 × 3 × 1 meshes that should be used based on the -point mesh used for bulk MoS 2 and the three-fold expansion in the and directions.
The optimisations were performed at fixed volume with reduced tolerances of 10 -6 eV on the electronic energy and 10 -3 eV on the total energy and using real-space PAW projection. These   Figure  These images were prepared using VESTA. [48]  These images were prepared using VESTA. [48]  The energies are referenced to the average 1s core level of the nine Mo atoms in the central layer of the slabs (c.f. Figs. S10(b)/S11(a)) and the energy zero is set to the F of the pristine