Cu Electrodeposition on Nanostructured MoS₂ and WS₂ and Implications for HER Active Site Determination

MoS₂ and WS₂ films were prepared by plasma-enhanced atomic layer deposition (PEALD) in an Oxford Instrument FlexALTM ALD reactor on either electrochemical quartz crystal microbalance (EQCM) gold electrodes or glassy carbon electrodes (Carbon-Vitreous-3000C-Foil-VC000550, Goodfellow Cambridge Ltd. UK).^24,25 The PE-ALD process was based on a combination of a metal organic precursor bis(tert-butylimido) bis(dimethylamido) molybdenum ([(N^tBu)₂(NMe₂)₂Mo]) or [(N^tBu)₂(NMe₂)₂W] and H₂S + Ar or H₂S + H₂ + Ar plasma as a co-reactant.^24,25 WS₂ and MoS₂ films with different morphologies and number of ALD cycles, i.e. different thicknesses, are tested in this work.

Electrochemical characterization was performed in a three-electrode configuration cell with MoS₂ or WS₂ deposited glassy carbon plates (22 × 22 × 2 mm³) as the working electrode, Pt foil as the counter electrode (Pt contamination under the same conditions was excluded by XPS measurements in our previous research²⁶) and a saturated calomel electrode (SCE, calibrated with a value of +0.269 V vs reversible hydrogen electrode (RHE)) as the reference electrode. Hydrogen evolution measurements were carried out in an Ar-saturated 0.1 M H₂SO₄ solution and Cu UPD was performed in an Ar-saturated 2 mM CuSO₄ in 0.1 M H₂SO₄.

XAS was performed at the 061D-1 (HXMA) beamline of the Canadian Light Source (CLS). Samples were mounted in an N₂ protected holder and measured at a grazing incidence angle of 0.2° in fluorescence detection mode by using a 32 element Canberra array detector. X-ray Near Edge Spectroscopy (XANES) and Extended X-ray Absorption Fine Structure (EXAFS) spectroscopy were collected at the Cu K-edge (8.98 keV) and recorded by co-addition of 5 scans. The code used for XANES modeling was FDMNES.²⁷ The structural model for metallic Cu was obtained from Crystallography Open Database (COD ID: 1512504).²⁸ Structural models of Cu bonding on MoS₂ were obtained from DFT calculations, which are discussed in detail in Part V of the SI.

DFT calculations were carried out using the Vienna Ab-initio Simulation Package (VASP), a periodic plane wave DFT code which includes the interactions between the core and valence electrons via the Projector Augmented Wave (PAW) method.^29–31 The generalized gradient approximation (GGA) with the Perdew–Burke–Ernzerhof (PBE) functional was used for the calculation of the electronic exchange-correlation potential.³² Wave functions were expanded in a plane wave basis with a high energy cutoff of 600 eV and the convergence criterion was set to 10−6 eV between two ionic steps for the self-consistency process. The long-range Van der Waals interactions were incorporated in the DFT calculations through the DFT-D3 scheme by Grimme, which adds a semi-empirical dispersion potential to the conventional Kohn–Sham DFT energy.³³

Figure 1 displays the potential dynamic profiles of different MoS₂ films during Cu UPD and overpotential deposition (OPD). For comparison, cyclic voltammetry (CV) curves of both the bare glassy carbon (GC) substrate and MoS₂ film are given in Fig. 1a. From these data, we can conclude that the cathodic reduction of Cu²⁺ on MoS₂ starts at more positive potentials than bulk Cu deposition on bare GC. The Nernst potential of ${E}_{C{u}^{2+}/Cu}$ in this system (2 mM CuSO₄) is +0.26 V vs RHE while the UPD on MoS₂ shifts around +100 mV. A series of cyclic voltammograms with varying potential scan ranges were recorded (Fig. 1b). The data in Fig. 1 show the transition from Cu UPD to OPD with decreasing deposition potential. The complex loop between 0 and 0.2 V vs RHE as shown in Fig. 1b might be due to intercalation of Cu, which is reported in previous work.^34,35 The same phenomenon is also observed for WS₂ (Fig. 1d). Thus, Cu can be deposited on MoS₂ and WS₂ at potentials more positive than the Nernstian potential, or in other words Cu UPD occurs on MoS₂ and WS₂.

Zoom In Zoom Out Reset image size

To study the UPD phenomenon of Cu atoms on MoS₂ surfaces in more detail, we employed electrochemical quartz crystal microbalance (EQCM) measurements to quantitatively analyze the mass change during the Cu UPD process.³⁶ Figure 2 presents cyclic voltammograms (CV) and the simultaneously recorded frequency change (Δf) on PE-ALD prepared MoS₂-Au-EQCM electrodes in 2 mM CuSO₄ in 0.1 M H₂SO₄. Upon negative scanning from +0.85 V, the QCM frequency first remains constant and then decreases quickly as the first Cu UPD peak appears at ∼+0.40 V. On the reverse scan, the frequency gradually increases accordingly when the Cu stripping peak approaches ∼+0.50 V. Although there might be drifts in EQCM, the Cu UPD and stripping peaks increase gradually within the first four CV scans and remain similar for the 5th and 6th scan, which is also reflected in the frequency change in EQCM measurements. In addition, the starting frequency (∼0.8 V vs RHE) value becomes larger with increasing scan numbers and remains constant after the 4th scan. This phenomenon can be ascribed to the strong binding between Cu atoms and MoS₂ leading to an incomplete desorption of adsorbed Cu in the initial cycles. However, the absolute frequency change in each cycle (Fig. 2b) remains constant, which suggests that the same amount of Cu has adsorbed. The frequency change (Δf) in EQCM is proportional to the change in mass (Δm) per unit area (A) on the working electrode, given by the Sauerbrey equation:

$$\begin{eqnarray*}&&{\rm{\Delta }}f=-\left(2{f}_{0}^{2}\right){\rm{\Delta }}m/A\sqrt{{\mu }_{q}{\rho }_{q}}\end{eqnarray*}$$

where ${f}_{0}$ is the resonance frequency (Hz), A is the piezo-electrically active crystal area, ${\mu }_{q}$ is the shear modulus of quartz ($2.947\times {10}^{11}\,{\rm{g}}\cdot {{\rm{cm}}}^{-1}\cdot {{\rm{s}}}^{-2}$) and ${\rho }_{q}$ is the density of quartz ($2.648\,{\rm{g}}\,{{\rm{cm}}}^{-3}$).^36–39 Based on this equation, the sensitivity of the EQCM is calibrated to be $13\,\,{\rm{ng}}\,\cdot {{\rm{Hz}}}^{-1}\cdot {{\rm{cm}}}^{-2}.$ Therefore, the adsorbed Cu atoms on the MoS₂ electrode amount to $78\,\pm 7\,{\rm{ng}}\cdot {{\rm{cm}}}^{-2}$ or $7.3\pm 0.6\times {10}^{14}\,{\rm{atoms}}\cdot {{\rm{cm}}}^{-2}$ in the range of +0.31 to +0.85 V vs RHE. Figure 2c, d displays the current density and corresponding frequency change at Cu UPD potential (0.44 V vs RHE) for 100 s. Given the 2-electron process for Cu UPD, the charge calculated via EQCM is 78 ∼ μC cm⁻², which is higher than that obtained by integrating the current density over time (45 ∼ μC cm⁻²). The Cu UPD charge discrepancy between EQCM and recorded current density can be ascribed to the poor conductivity and hydrophobic effects of semiconducting MoS₂ that result in significant deviations of Δf, which greatly lower the sensitivity of EQCM measurements.⁴⁰ In conjunction with electrochemical measurements, XAS measurements were also performed to enable the assignment of structural configurations for Cu deposition on TMDs.

Zoom In Zoom Out Reset image size

Grazing incidence X-ray absorption spectroscopy (XAS) at the Cu K-edge, including X-ray absorption near-edge structure (XANES) and extended X-ray absorption fine structure (EXAFS), allows for the investigation of the local structural environment and the site occupancy of adsorbed Cu on TMDs.^41,42 As shown in Fig. 3, the similarity in the line shape between the sample of Cu deposition at 0.21 V vs RHE and Cu foil for the XANES edge jump within region "A" suggests that likely there is a Cu species with Cu-Cu direct bonding present in the measured sample. In addition, the clearly resolved edge shift by ∼2 eV (region "B" indicated by the arrow, Fig. 3a) indicates that there exists a second, different Cu species without Cu-Cu direct bonding, which may correspond to Cu atoms directly adsorbed on MoS₂. The theoretically modeled Cu adsorbate system reveals a particle-size induced progressive change in the XANES features (Fig. 3b) for metallic Cu. The smallest particle used for the modeling is a cluster R3.0 Å (cluster size 3.0 Å), containing 12 Cu atoms. Figure 3c compares the sample spectrum and the XANES theoretical modeling based on a R3.0 Å cluster. The similarity in XANES between the two extends from the edge jump up to the white line (A1). In addition, the metallic Cu specific shoulder feature "C" is revealed for the measured sample and the theoretical model at around the same energy position, further suggesting a Cu-Cu direct bonding. Nevertheless, the edge shift in Fig. 3a is also well pronounced comparing with R3.0 Å, confirming the presence of Cu adsorption on MoS₂. Thus, although the Cu deposition potential (0.21 V vs RHE) is slightly negative than UPD condition, XANES based observation suggests two coexisting Cu species in the sample, i.e., metallic Cu–Cu direct bonding, and Cu atoms adsorbed on MoS₂.

Zoom In Zoom Out Reset image size

Using density functional theory (DFT) calculations, we evaluated possible Cu bonding scenarios on MoS₂, namely single Cu monodentate (M1, Fig. 4a), single Cu bidentate (M2, Fig. 4b) and paired Cu monodentate (M3, Fig. 4c) configurations. Fig. 4c presents the comparison of experimental and FEFF (automated program for ab initio multiple scattering calculations of XAS) modeled EXAFS magnitudes of Fourier Transforms (FT), which indicates the existence of M2 and M3 species upon Cu deposition at +0.21 V on MoS₂, with bond distances of 2.36 Å for Cu-S (M2) and 2.56 Å for Cu–Cu (M3) (Table SIV is available online at stacks.iop.org/JES/167/116517/mmedia).²⁷ As the Cu–K edge XAS was recorded for a sample with Cu deposition at a relatively negative potential (<+0.26 V vs RHE), it is reasonable to assume the existence of Cu-Cu direct bonding, indicating the presence of Cu-Cu, which is consistent with XANES analysis shown in Fig. 3. Although the sample was characterized under more negative potentials than Cu UPD, the strong contribution of the Cu bidentate (M2) configuration in the R-space of the Fourier Transformed EXAFS (Figs. 4d, 4e) indicates the bonding mode of Cu UPD on MoS₂, which is also indicated by the edge shift in XANES (Fig. 3). It should be noted that the current EXAFS fitting models are based on Cu adsorption on MoS₂ basal planes. Even though edge sites of TMDs have been identified to be the HER active sites, FEFF modelled XANES of adsorbed Cu on MoS₂ edges (Fig. S16) did not fit well with the experimental data.⁴³ However, DFT derived optimized structures of Cu on S–Mo–S edges (Fig. S15) indicates that Cu-bridge-S, which is structurally similar to bidentate binuclear M2 (Fig. 4b), has the largest (negative) average binding energy (E_ab), which points to the adsorption of Cu atoms on edge or defect sites.⁴⁴ Owing to the ultrathin, sub-monolayer nature of the Cu UPD layers on the MoS₂ and WS₂ samples, the S/N ratio of the data recorded for Cu UPD sample (Cu_MoS₂@0.44 V) did not allow us to unambiguously prove the absence of Cu–Cu direct bonding (Fig. S11). Despite that, we think it is meaningful to show that even under lower deposition potential, we still observe the dominant peak of Cu on S edges (Fig. 4). Based on these data, further operando XAS measurements would be needed to investigate Cu UPD on TMDs in more detail.²⁶

Zoom In Zoom Out Reset image size

Recently, the Cu UPD method has been used to measure the number of HER active sites on WS₂, MoS₂ and Pd₃P₂S₈.^22–24 However, whether Cu adsorbs on the same sites, where hydrogen evolution happens, remains an unresolved question. Particularly as shown in Fig. 1, there is no clear H UPD region in MoS₂ and WS₂ making it impossible to derive Q_H. Nevertheless, the Cu UPD charge (Q_Cu) and hydrogen adsorption charge ("Q_H") were compared and adopted in TMDs-based electrocatalysts.^22–24 Here, we use the background capacitive charge (Q_BC) as a background correction to determine the charge which goes to the Cu UPD process (Q_Cu). To correlate the relative Q_Cu with Q_BC, a plot of Q_Cu/Q_BC as a function of deposition potential is presented in Fig. 5a.⁸ Q_Cu is obtained by integrating the Cu stripping charge from the Cu deposition potential (E_Dep.) to +0.67 V vs RHE (corrected for MoS₂ background Q_BC in 0.1 M H₂SO₄), whereas Q_BC is calculated with the same method in the absence of CuSO₄. The Q_Cu/Q_BC ratio reaches a value of ∼2 in a narrow potential window of +0.44–0.46 V vs RHE for MoS₂ (Fig. 5a, Table SI) and WS₂ (Fig. 5b, Table SII). However, we should point out that Q_Cu/Q_BC ratio of ∼2 does not necessarily suggest the same surface density between copper and hydrogen atoms under such conditions.^22,23 First of all, the absence of a clear H UPD signature in MoS₂ and WS₂ leads to a unknown quantitative relation between Cu UPD and H UPD sites making a quantitative evaluation of H adsorption sites via the Cu UPD method impossible. Secondly, other factors such as (i) different accessibility of solvated Cu²⁺ to nanostructured TMD sites compared to H⁺, (ii) Cu cluster formation due to the improper selection of the E_dep. window and scan rate, (iii) variations in crystallinity of TMD leading to presence of different HER active sites, and (iv) difference in nature of HER and Cu UPD sites such as basal plane adsorption of Cu on TMDs, can also influence the validity of Cu UPD method and are difficult to control. All these above aspects imply that the direct transfer of the Cu UPD method from noble metal systems with a clear H UPD phenomenon to TMDs appears to be not appropriate.

Zoom In Zoom Out Reset image size

In order to further evaluate the applicability of the Cu UPD method in relatively comparing different samples based on the number of active sites for the electrocatalytic HER, we measured Q_Cu, the double layer capacitance (C_dl) and the HER current density at −0.4 and −0.6 V vs RHE for a selection of MoS₂ (Figs. 5c, 5d) and WS₂ (Figs. 5e, 5f) samples; see Table SIII for sample IDs and preparation details. The linear sweep voltammetry (LSV) curves of different WS₂ and MoS2 films are presented in Fig. S2, S6 and the current density values at certain overpotentials presented in Fig. 5 reflect their HER performance. Q_Cu is calculated by integrating the CV curve of the samples in 2 mM CuSO₄ in 0.1 M H₂SO₄ from +0.46 to +0.67 V vs RHE corrected for the background charge. The double layer capacitance (C_dl) was extracted by plotting Δj = j_a − j_c (j_a and j_c represent anodic and cathodic current densities, respectively) at a given potential against CV scan rates using the following equation: $\tfrac{{j}_{a}\,-\,{j}_{c}}{2}=\,{C}_{dl}\,\tfrac{dE}{dt}$(Fig. S3). Current densities at both −0.40 and −0.60 V show that WS₂-5 presents the highest current density value. However, C_dl measurements show a significantly higher value for WS₂-1 compared to the other WS₂ films which clearly does not correlate with their HER activities. Therefore, evaluating the ECSA with C_dl in this case would lead to a random correlation with HER activity.^19,45 In contrast, Q_Cu values shown in Fig. 5e, f display the same trend as HER activity at both −0.40 and −0.60 V vs RHE, which suggests that Cu UPD apparently better correlates with HER activity and therefore would be a more reliable method for the determination of ECSA than C_dl. The correlation of Cu UPD (Q_Cu) with HER activity of MoS₂ films is shown in Figs. 5c, 5d. With Cu UPD we can assess the number of Cu UPD adsorption sites which apparently relates linearly to the HER activity of the tested samples allowing a relative comparison of N_AS (Tables I, II). A quantitative relation between N_{Cu UPD} sites and N_AS (HER) cannot be easily established due to the absence of the H UPD phenomenon on the tested nanostructured TMD samples. Further experiments, e.g. in situ scanning tunneling microscopy (STM) on single crystal MoS₂ and WS₂ model samples, would be needed to create a more solid link between N_{Cu UPD} sites and N_AS (HER). Importantly, parameters such as deposition potential window and porosity of the catalysts should be considered when using Cu UPD as a method for ECSA estimation. With that, the relation between N_{Cu UPD} sites and N_AS (HER) can be presented as N_AS (HER) = a * N_{Cu UPD} + b with a being a constant between 0 and 1and b being a constant considering the Cu deposition on basal planes of TMDs and/or Cu cluster formation.