Enhancement of the Hydrogen Evolution Reaction from Ni-MoS2 Hybrid Nanoclusters

This report focuses on a novel strategy for the preparation of transition metal–MoS2 hybrid nanoclusters based on a one-step, dual-target magnetron sputtering, and gas condensation process demonstrated for Ni-MoS2. Aberration-corrected STEM images coupled with EDX analysis confirms the presence of Ni and MoS2 in the hybrid nanoclusters (average diameter = 5.0 nm, Mo:S ratio = 1:1.8 ± 0.1). The Ni-MoS2 nanoclusters display a 100 mV shift in the hydrogen evolution reaction (HER) onset potential and an almost 3-fold increase in exchange current density compared with the undoped MoS2 nanoclusters, the latter effect in agreement with reported DFT calculations. This activity is only reached after air exposure of the Ni-MoS2 hybrid nanoclusters, suggested by XPS measurements to originate from a Ni dopant atoms oxidation state conversion from metallic to 2+ characteristic of the NiO species active to the HER. Anodic stripping voltammetry (ASV) experiments on the Ni-MoS2 hybrid nanoclusters confirm the presence of Ni-doped edge sites and reveal distinctive electrochemical features associated with both doped Mo-edge and doped S-edge sites which correlate with both their thermodynamic stability and relative abundance.


Further Experimental Details: Cluster source
Briefly, Ni-MoS 2 hybrid nanoclusters are generated within the first section of the source by sputtering with dual, independent magnetrons and gas condensation. After leaving the magnetron sputtering chamber via a small nozzle (5 mm in diameter), clusters with a positive charge are accelerated and steered by the ion optical electrostatic lenses which surround the beam in the second vacuum chamber. In the next step, the ion beam is focused and directed into the third vacuum chamber for mass selection. By using the Birmingham Time-of-Flight mass filter 1 , the mass distribution of clusters can be monitored in real time. When the desired cluster size distribution within the sampled ion beam is achieved, the high voltages applied to the deflector in the centre of the Ion Optics chamber are switched to deflection mode, so that the positively ionised fraction of the cluster beam is deflected horizontally towards the deposition chamber. Glassy carbon (GC) stubs (5 mm X 5 mm X 3 mm, mirror finish) are mounted on a carousel, which can rotate and also translate vertically. The rotation speed of the carousel and its vertical motion are carefully controlled to ensure an even cluster distribution on the substrates. For all the samples, an average cluster spacing of 2.5 nm was targeted. According to this cluster spacing and the mass spectra, the mass loadings are 1.28 µg/cm 2 , 3.45 µg/cm 2 , and 4.25 µg/cm 2 for Ni, MoS2 and hybrid Ni-MoS 2 nanoclusters, respectively. A high voltage bias is applied to the carousel in order to control the impact energy of the clusters landing on the support.  S1. Schematic of the Cluster-Beam system (top view). It consists of four sections: magnetron sputtering, ion optics, mass filter and cluster deposition. Note that in the experiments described the mass filter is only used for cluster size monitoring, not for deposition. The clusters are instead deposited directly onto substrates in the chamber shown at the top of the figure.

Hybrid Ni-MoS 2 nanocluster composition calculation
The HAADF intensity (I) of two kinds of elements (A and B) follows the relationship I A /I B = (Z A /Z B ) 1.46 , in which Z is the atomic number, for our microscope calibration 2 . Thus the intensity relationship between Mo, S and Ni can be listed as below: Since the single atom intensity of Mo is much higher than that of S and Ni, we assume the atoms most easily visible in STEM images are Mo. According to the intensity line profile shown in Figure 4, the number of Mo columns intersected by the line and the number of Mo atoms in each column can be obtained. Once all the surface area of the nanocluster is scanned by such lines, the total number of Mo atoms (N Mo ) is obtained. The number of S atoms (N s ) in this nanocluster can also be derived on the assumption that the ratio of Mo atoms to S atoms is 1:2. Therefore the composition of the nanocluster can be regarded as: (MoS 2 ) N Mo Ni x . Now the integrated intensity of the whole nanocluster (I cluster ) can be given by: Since the intensity of the whole nanocluster can be measured from STEM image, against using single Mo atom as the standard, the value of x can be given by combining equations (1) and (2): Then the nanocluster composition is revealed. By this method, the compositions of six nanoclusters of varying size were calculated and are listed in Table 1 (see manuscript). It can be seen that the ratio of Ni varies substantially, but broadly speaking the ratio tends to go up with increasing nanocluster size. Figure S3. High-resolution XPS spectra of Mo 3d (left) and S 2p (right) for fresh (top) and 14 h air exposed (bottom) (MoS 2 ) 300 nanoclusters. Labels: raw spectra (solid black), cumulative peak fit (solid red), Mo 4+ 3d 5/2 (solid green), Mo 4+ 3d 3/2 (dashed green), Mo 6+ 3d 5/2 (solid orange), Mo 6+ 3d 3/2 (dashed orange), S 2p 3/2 (solid blue) and S 2p 1/2 (dashed blue). Figure S4. High-resolution XPS spectra of Ni 2p fresh (top) and 14 h air exposed (bottom) Ni nanoclusters. Labels: raw spectra (solid black), cumulative peak fit (solid red), Ni 0 2p 3/2 peak deconvolution (solid blue), Ni 2+ (NiO) 2p 3/2 peak deconvolution (solid green) and Ni 2+ [Ni(OH) 2 ] 2p 3/2 peak deconvolution (solid orange). Number of peaks, FWHM, peak positions and relative area percentages used from Biesinger et al. 3

Surface coverage estimation: Randles-Sevčik equation
The surface coverage of the deposited nanoclusters on the sample can be estimated by means of the irreversible Randles-Sevčik expression 4 : Where ห‫ܫ‬ ,ுாோ ห stands for the current intensity achieved at the diffusion decay peak observed in the HER LSV scans, n for the number of electrons transferred in the redox process (in the case of the reaction 2 ‫ܪ‬ ሺሻ ା 2 ݁ ି ⟶ ‫ܪ‬ ଶ ሺሻ ݊ ൌ 2), ߙ for the electron transfer coefficient (assumed to be ߙ ൌ 0.5), ‫ܣ‬ ௧௩ for the electrochemically active surface area of the nanoclusters, ‫ܥ‬ for the bulk concentration of the electroactive species in solution (in our case ሾ‫ܪ‬ ା ሿ ൎ 2 ൈ 10 ି ‫ݏ݈݉‬ ܿ݉ ିଷ ), ‫ܦ‬ ு శ for the proton diffusion coefficient ‫ܦ(‬ ு శ ൎ 1 ൈ 10 ିସ ܿ݉ ଶ ‫ݏ‬ ିଵ ) 5 and ߭ for the scan rate.
As the slope of the linear regression equals the terms that multiply ߭ ଵ/ଶ in the irreversible Randles-Sevčik expression, the rearrangement of eqn. 1 will enable us to obtain the value of ‫ܣ‬ The results obtained are summarised in Table S1:  Table S1. Compilation of linear regression fit parameters, the calculated nanoclusters electrochemically active surface area ‫ܣ(‬ ௧௩ ) and the nanoclusters surface coverage ‫ܣ(‬ ௧௩ / ‫ܣ‬ ), calculated as the ratio of the electrochemically active surface area and the glassy carbon geometric area.

Mechanism of Ni oxidation in acidic media
The electrochemical features of Ni oxidation in acidic media have been reported to be dependent on the electrolyte, pH and experimental conditions (hydrodynamic parameters, voltage range, etc) [6][7][8][9] . The reaction mechanism can be summarised as follows 9 , ሾܰ݅ሿ ିଵ ሾܰ݅ሺܱ‫ܪ‬ሻ ଶ ሿ ⇌ ሾܰ݅ሿ ‫ܪܱ2‬ ି ܰ݅ ଶା (10) where the application of sufficiently positive potentials drives the two-electron Ni electro-oxidation in the presence of adsorbed water molecules. Formation of Ni(OH) and Ni(OH + ) intermediates in eqns. 6 and 7 is followed by either electrodissolution (eqn. 8) or the formation and further chemical dissolution of a Ni(OH) 2 anodic layer (eqns. 9,10) by protons provided from the acidic media.
Voltammetric investigations in sulfuric acid (pH≤3) revealed two oxidation peaks at ca. 0.2 V and ca. 0.3-0.5 V vs. NHE, followed by a passivation current extending from 0.7 V vs. NHE upwards 10 . The first peak was reported to be the result of the competition between nickel electrodissolution and Ni(OH) 2 anodic layer formation, the former having the largest contribution, whereas the second peak was attributed to the Ni(OH) 2 layer growth depicted in eqn. 9 7,11 . The dissolution of this nickel hydroxide layer by eqn. 10 is interrupted by the reduction of Ni(OH) 2 phase water content at more positive potentials that yields a Ni x O y -based passive film (eqn. 11).

Elucidation of turnover frequency (TOF)
Estimation of turnover frequency was performed by the deconvolution and integration of the anodic peaks observed after 10 cycles in the 0 to 1.2 V voltage range vs. SCE (0.35 to 1.6 V vs. NHE after calibration). The choice of anodic stripping voltammetry (ASV) for TOF estimation is based on the fundamentals of the technique and the nature of the nanoclusters ASV, unlike other techniques reported in the literature for estimating TOF values (probe molecules adsorption, electrochemical capacitance measurement), allows to discriminate the contribution of the HER inactive basal sites and the HER active edge sites according to their electrochemical stability (anodic peak potential), avoiding active surface area overestimation that would lead to undervalued turnover frequencies. In addition to this, electrochemical features originated by the Ni atoms/clusters allocated in inactive sites can be effectively deconvoluted and withdrawn from the Ni-doped active sites response to enable a rigorous analysis. This analysis does not necessarily guarantee that all the sites electrochemically oxidised are available to the electrolyte due to geometric constraints, but it provides a reasonable estimation of the active sites present in the sample. The deconvolution of the first anodic linear sweep was performed in the 0.6 to 1 V and 0.4 to 0.8 V voltage range vs. NHE for (MoS 2 ) 300 and (Ni-MoS 2 ) 1000 nanoclusters, respectively.
The formula used for calculating the TOF per-site is the following 12 : ‫ܨܱܶ‬ ‫ݎ݁‬ ‫݁ݐ݅ݏ‬ ൌ ‫݈ܽݐܶ‬ ‫݊݁݃ݎ݀ݕ‪ℎ‬‬ ‫݉ܿ/ݏݎ݁ݒ݊ݎݑݐ‬ ଶ ‫ܿ݅ݎݐ݁݉݁݃‬ ‫ܽ݁ݎܽ‬ ‫ݎܾ݁݉ݑܰ‬ ‫݂‬ ‫ݕ݈݈ܽܿ݅݉݁‪ܿℎ‬ݎݐ݈ܿ݁݁‬ ‫݁ݒ݅ݐܿܽ‬ ‫݉ܿ/ݏ݁ݐ݅ݏ‬ ଶ ‫ܿ݅ݎݐ݁݉݁݃‬ ‫ܽ݁ݎܽ‬ The total number of hydrogen turnover events is calculated using the conversion 13 : Calculation of the turnover frequency can also be calculated for the (Ni-MoS 2 ) 1000 hybrid nanoclusters at the (Ni-MoS 2 ) 1000 hydrogen evolution peak half maximum (E= -0.640 V, scan rate = 25 mV s -1 ): The total number of electrochemically MoS 2 active edge sites per geometric area is elucidated as it follows: Assuming that the anodic electrochemical process undergoes the reaction mechanism proposed by Bonde et al. 14 , corresponding to the partial oxidation of sulphur in MoS 2 The integration of the ASV deconvoluted peaks responsible for the MoS 2 active edge sites oxidation will enable to provide an estimation of the TOF.
Peak integration values obtained for first scan ASV:

Normalization of exchange current densities
The exchange current density values obtained from Tafel plot analysis (scan rate: 25mV s -1 ) can be normalized with respect to the density of sites on Pt(111) to compare the values obtained with the ones of transition metals. The expression used is the following 15 : Where ݆ , stands for the normalized exchange current density, ݆ ,ெௌ మ for the experimental MoS 2 exchange current density, ߩ ௧ሺଵଵଵሻ for the density of sites on Pt (111) Table S2. Compilation of peak half maximum current density (݆ ௫ ), peak current density (݆ ), Tafel slope (b), exchange current density (݆ ), normalized exchange current density (݆ , ) and turnover frequency (TOF) values of the doped/undoped MoS 2 nanoclusters samples evaluated.