Electrochemical kinetics as a function of transition metal dichalcogenide thickness

: Two-dimensional transition metal dichalcogenides are used as electroactive materials for electrochemical and electrocatalytic applications. However, it remains unclear how transition metal dichalcogenide thickness influences the electrochemical response measured at its surface. We use scanning electrochemical cell microscopy to assess the electrochemical response of different thicknesses of bottom-contacted MoS 2 , MoSe 2 , WS 2 , and WSe 2 towards the simple outer-sphere redox couple [Ru(NH 3 ) 6 ] 3+/2+ with submicron spatial resolution. A detailed analysis, coupling mass transport and electrochemical kinetics, reveals that the electrochemical response can be described using an electron tunneling barrier, which scales with the band gap of the two-dimensional transition metal dichalcogenide. Our results suggest that interpretation of the electrochemical and electrocatalytic responses on transition metal dichalcogenide-covered electrodes should account for the through-layer electron transport kinetics.

The electrochemical response of bulk TMDCs has previously been shown to differ significantly between two distinct surface types: the basal plane and the edge plane [17,18].Outer-sphere redox mediators, such as hexammineruthenium [Ru(NH3)6] 3+/2+ , which we consider here, showed enhanced electron transfer kinetics on the edge plane compared to the basal plane surfaces [17,18].Similar enhancement of the HER kinetics on the edge planes of MoS2 crystal compared to its basal plane has also been observed [19].
The preparation of TMDCs in their 2D form typically yields heterogeneous samples with a variable flake size and variable flake geometry.The number of TMDC layers stacked strongly affects the electronic structure and the electrochemistry of the stacked material [7,20,21].Isolation and electrochemical investigation of specific TMDC thicknesses remain a significant challenge, primarily because it is difficult to prepare homogeneous mono-and few-layer flakes larger than tens of microns in lateral size.This challenge has recently been overcome by a gold-mediated exfoliation of TMDCs, capable of producing millimeter-to-centimeter sized monolayer flakes, combined with localized electrochemical methods [20,22,23].
Scanning electrochemical probe microscopy techniques are particularly well suited to isolate the electrochemical activity of nano-and micro-scale domains [20,24,25].Scanning electrochemical cell microscopy (SECCM) allows electrochemical mapping with a spatial resolution of tens to hundreds of nanometers using a nanodroplet-electrochemical cell located at the end of a nanopipettes [26][27][28].
A key advantage of a SECCM approach is the localized measurement at the area of interest, without electrochemical activation of the entire sample surface by its exposure to the electrolyte, as is the case in macroscale electrochemistry or scanning electrochemical microscopy.Furthermore, SECCM can be combined with complementary microscopy approaches, such as atomic force microscopy (AFM), to correlate the electrochemical activity with surface morphology.Unwin et al. have previously studied the electrocatalytic activity of bulk TMDCs using SECCM [29,30].
These studies investigated HER on the surface of top-contacted TMDCs (MoS2 and WS2), which showed an enhanced electrocatalytic activity at the edge plane compared to the basal plane.Takahashi et al. recently used SECCM with an impressive 20 nm resolution to characterize a few-layer MoS2, as well as heterostructures of MoS2 and WS2, confirming an increased HER activity at the edge planes of MoS2 but showing no significant differences in HER activity on different number of TMDCs layers [31].
We have previously measured the electrochemical and photoelectrochemical response of 2D MoS2 on an insulating substrate towards reduction of the [Ru(NH3)6] 3+/2+ redox couple using a microdroplet electrochemical cell [32].This showed the significant effect of the number of layers (comparing monolayer, 3-4 layers, 5-10 layers, and bulk) on the electrochemical response, which was also sensitive to illumination.However, the spatial resolution was limited to the size of the microdroplets (20-40 μm), which made measurements on individual layers challenging.In addition, the electrical contact to the MoS2 was made directly at the top layer (top-contacted), which we will show to be a key difference with respect to our current results.
The location of the electrical contact made to the 2D TMDCs strongly influences the electrochemical response we observe.In a top-contact configuration, the electrical contact is made to the top layer of the layered 2D material stack, usually using a conductive adhesive and a metallic wire.In this configuration, only lateral electron transport along the layers occurs due to the high in-plane electrical conductivity through the basal plane [33].In a bottom-contact configuration, electrical contact is established through the supporting metallic substrate on which the 2D layers are immobilized.In this configuration, electron transport occurs between the metallic substrate and the top 2D layer, where the electrochemical reaction takes place.
Therefore, in a bottom-contacted sample, the interlayer electron transport can influence the electrochemical response, as we observe in our studies here.
The HER response of monolayer (1L), bilayer (2L), and trilayer (3L) MoS2 in a bottom-contact configuration has recently been reported [25]; these results revealed that the electron transport between the conductive substrate and 2D TMDC could be the rate-limiting step in HER electrocatalysis.Another work on bottom-contacted MoS2 suggests that HER electrocatalysis relies on electron hopping between the MoS2 layers [21].However, a recent report of HER on bottom-contacted MoS2 did not observe any significant dependence of the electrocatalytic response on the thickness [31].Therefore, a clear scientific question arises about how do the layer thickness of TMDCs and the bottom-contact affect the electrochemical response [34].HER is an inner sphere multistep reaction whose overall rate on TMDC electrodes is proposed to depend on chemical steps involving surface-adsorbate intermediates.Therefore, a well-defined outer-sphere redox couple, sensitive only to the density of electronic states, is better suited to assess the dependence of the electrochemical response on the number of TMDC layers.
To that end, we isolated the electrochemical response from the [Ru(NH3)6] 3+/2+ outer-sphere redox couple on monolayer, bilayer, and trilayer MoS2, MoSe2, WS2, and WSe2, immobilized on an Au substrate.As shown in Figure 1A, a Park Systems NX10 instrument was used to perform both optical characterization, determination of surface morphology by AFM, and the SECCM measurements.We use a SECCM probe with ca.800 nm spatial resolution to perform the electrochemical mapping, targeting TMDC layers of different thicknesses, as shown in Figure 1B.The electrical bottom contact is established through the Au electrode substrate, as shown in Figure 1C.The localized electrochemical response was then correlated with the surface morphology, characterized by optical microscopy and AFM.We used the Gerischer formalism of electron transfer coupled with the diffusional transport of [Ru(NH3)6] 3+/2+ in the SECCM probe to simulate the electrochemical response.We extract the electron transfer kinetics of the [Ru(NH3)6] 3+/2+ redox couple on the different TMDC thicknesses by fitting simulations to experimental data.

Experimental
Hexaammineruthenium (III) chloride ([Ru(NH3)6]Cl3) was purchased from JMC Corporation (South Korea).Potassium chloride (KCl, ≥ 99.5%) was obtained from VWR Chemicals (USA).All chemicals were used as received.Distilled water with a resistivity of 18 MΩ cm was used to prepare the solution of 10 mM KCl and 1.0 mM [Ru(NH3)6] 3+ .All the procedures were carried out at room temperature.
Samples of the transition metal dichalcogenides (MX2, M = Mo or W and X = S or Se) were prepared via mechanical exfoliation of bulk crystals on Au substrates, as previously reported [20].The bottom-contact electrical connection to the TMDCs was established via the gold substrate.
Optical, AFM, and SECCM measurements were acquired on a Park NX10 (Park Systems, South Korea).The AFM images were obtained in a non-contact mode (NCM) with a PPP-NCHR cantilever type (force constant = 42 N/m, resonance frequency = 330 kHz, Nanosensors).
The SECCM measurements were performed using the ParkSystem NX10 system.The SECCM probe was aligned to the sample area on the surface using the optics and LSVs were then recorded at a preprogrammed squared grid, with individual measurement spots spaced at 1.5 µm.A hopping mode was used in which the probe was approached vertically towards the sample surface at a speed of 0.3 µm/s and a potential of -0.8 V until contact between the nanopipette droplet and the surface was established.The contact was detected as the appearance of a double layer charging current and immediate [Ru(NH3)6] 3+ reduction, with a threshold of 5 pA.At this point, the vertical movement of the probe was stopped immediately, and LSVs were recorded in the potential range of + 0.2 V to -0.8 V at a speed of 1 V/s.After completing the measurement, the nanopipette was retracted and moved to the next sample point on the grid Simulations of the nanoelectrochemical cell at the end of the SECCM probe were performed in COMSOL Multiphysics (COMSOL 5.2, Sweden).See supplementary information (SI-4) for further details.

TMDC thickness characterization
Mechanical exfoliation was used to isolate high-quality 2D flakes of MoS2, MoSe2, WS2, and WSe2 on an Au substrate [20].Regions of the Au substrate coated with TMDC flakes of a different number of layers could be distinguished optically, as shown for MoSe2 in Figure 2A.The optical contrast and color of the flakes change with the different number of layers of MoSe2 [36].Similar results were obtained for MoS2, WS2, and WSe2, as reported in the Supporting Information (Figures S1-S3).
The optical images were therefore used to locate surface regions containing various TMDCs thicknesses (1L-3L) prior to SECCM characterization.AFM was used to corroborate the layer assignment from the step-height profiles obtained after the SECCM measurement, as shown in Figure 2B.Step-height profiles of 1L/2L and 2L/3L areas displayed in Figure 2C and 2D yield monolayer thickness of ca.0.81 nm and 0.78 nm, in agreement with previous monolayer MoS2 estimates from AFM (~0.85 nm) [37].The step-height from Au to 1L could not be clearly determined due to the roughness of the gold surface.AFM step-height profiles for the MoS2, MoSe2, and WSe2 samples are displayed in Figure S1 -S3.The measured layer thicknesses agree with previous studies of different TMDCs and support our optical analysis [20,[38][39][40].

Local electrochemical measurements
SECCM imaging was carried out using a single-barreled nanopipette probe of approximately 800 nm diameter.An Ag/AgCl wire immersed in the electrolyte solution was used as a reference electrode.Linear sweep voltammograms (LSVs) for the reduction of [Ru(NH3)6] 3+ to [Ru(NH3)6] 2+ were obtained at a regular grid of sample points spaced at 1.5 m intervals on all four TMDC materials (MoS2, MoSe2, WS2, WSe2); multiple measurements were carried out on each surface type (Au, 1L, 2L, and 3L).The AFM images shown in Figure 2B, which were obtained after the SECCM measurements, reveal salt residues left at points of contact between the SECCM droplet and the surface after water evaporation.
The droplet/surface contact area was estimated from the AFM images by measuring the diameter of the droplet residue; these values are summarized in Figure S4.The droplet size is determined by the pipette aperture, with a small variability expected due to the nature of the pipette pulling process.A new pipette was used for each experiment, thus, the average droplet size was slightly different for each of the TMDC materials investigated and estimated to be 0.81 ± 0.05 m, 0.73 ± 0.05 m and 0.80 ± 0.09 m for MoSe2, WS2, and WSe2, respectively, as shown in Figure S4.The average droplet size of MoS2 was found to be considerably larger at approximately 1.30 ± 0.11 m in diameter.The larger droplet size observed on MoS2 is due to a larger pipette aperture for this experiment (determined from the diffusionlimited current on MoS2, see section SI-6).The droplet size displayed only a narrow dispersion with the TMDC thickness, as seen in Figure S4 histograms, while the variance observed is consistent with the wetting behaviour of TMDCs surfaces after exposure to air [41,42].Grid points sampled on bare gold surfaces (see MoSe2 and WSe2 samples in Figure S6 -S7) exhibited the largest droplet contact size, which can be attributed to the higher hydrophilicity of Au in comparison to TMDC [42,43].
The salt residues were used to assign each LSV to a specific surface type: bare gold, 1L, 2L, 3L, and layer boundaries.Figure 3A shows an AFM image of a WS2 sample recorded after the SECCM measurements, where the sampled areas are color-coded to reflect the corresponding surface type, as identified by optical microscopy and AFM.The same protocol was used for the assignment of LSV curves of MoSe2 and WSe2, as shown in Figures S6 -S7.The relative spacing (1.5 µm) of the SECCM point grid was identical for all four TMDCs.Due to the greater wetting on MoS2, a larger fraction of the sample became covered by salt residues, which rendered the AFM analysis of the surface type unreliable.We therefore employed optical microscopy to assign the surface type as detailed in Figure S8.
The above procedure resulted in multiple LSV curves obtained at each surface type (Au, 1L, 2L, and 3L), thus generating a representative dataset for further analysis.
Both AFM and optical microscopy indicate that several LSVs measurements were obtained at boundaries between different layer thicknesses (such as Au-1L, 1L-2L, Au-2L, 1L-3L boundaries).However, the dearth of data from the boundaries prevents us from drawing any strong conclusions and we leave the analysis of these types of measurement to future work.The AFM imaging also revealed various forms of defects and surface irregularities, such as cracks in the 2D flakes and defects in the underlying Au substrate, as shown in Figure S5-S7.As these defects are likely to affect the electrochemical response, the corresponding LSVs curves were also discounted from the analysis of the electrochemical response as a function of the TMDC thickness, details on the excluded points are given in Table S1.
The TMDC thickness (1L, 2L, 3L) for each point used in our LSV analysis is displayed in Figure 3B for WS2 and in Figure S6 -S8 for MoSe2, WSe2, and MoS2.
The LSVs corresponding to 1L, 2L, and 3L WS2 are shown as black traces in Figure 3C, 3D, and 3E, respectively, while the LSVs on MoS2, MoSe2, and WSe2 are shown in the Figures S9-S12.Figures S9 -S12 also show every LSV measurement on TMDCs (including the measurements at defect sites) and LSV measurements once sites with defects have been excluded.Both LSV datasets show the dependence of the electrochemical behavior on the TMDC layer thickness.However, the LSVs collected at the defect sites are noticeably different to those at the pristine basal plane and resemble the Au surface response (Figure S16).The defective areas generally show faster apparent kinetics.One therefore expects macroscopic measurements over both the pristine and defective areas would also yield enhanced kinetics, subject to the defect density.

Electrochemical kinetics
The LSVs in Figures 3C-3E show that the [Ru(NH3)6] 3+ reduction is more facile on thinner WS2; similar trends are observed for the other three TMDC materials (Figure S10 -S12).To quantify the observed electrochemical response, a finite element simulation implemented by COMSOL Multiphysics was used to describe the transient diffusional transport of [Ru(NH3)6] 3+/2+ in the SECCM probe (details in section SI-5) and the electron transfer kinetics at the sample surface.The geometry of each SECCM probe used, such as the droplet contact area and pipette aperture, were derived from AFM imaging and limiting current calculations, respectively (see section SI-2 and SI-6).
The simulated rate of heterogeneous electron transport (HET) was described by a Gerischer formalism.As theoretically predicted by Feldberg [44] and recently verified experimentally [45], for an ultramicroelectrode with slow kinetics a significant difference exists between the HET rate constant value calculated from the Butler-Volmer and Marcus-Hush/Gerischer formalism.In the Gerischer formalism, the kinetics of HET can be described by considering the overlap between isoenergetic electronic states of the electrode and the redox species in solution (electronic coupling in the Marcus-Hush formalism).Therefore, the kinetics of cathodic electron transfer (  ) is given by the overlap between isoenergetic occupied states at the electrode (()  ) and empty electron acceptor states in the solution (  ) [46].Likewise, the anodic kinetics (  ) is determined by the overlap of isoenergetic filled electron donor states (  ) in the solution and unoccupied states in the electrode (()  ).Therefore, the cathodic and anodic kinetic rates can be written as:

𝜕𝐸
(Eq. 1) where the integration parameter  is the electron energy (in eV), and the prefactor, kmax, is the maximum rate of electron transfer [46].The solution states (  and   ) are described by normalized Gaussian distributions and the electrode states (()  and ()  ) are described by normalized Fermi-Dirac distributions for free electrons in a metal, see supplementary information (SI-7) for more details.
The overlap integrals for anodic and cathodic processes, as described in Eq.1 and Eq.2, were calculated by numerical integration, obtaining two waveforms (for anodic and cathodic processes), which describe the ratio of the overlap of the reacting redox species as a function of the applied potential.The waveforms, displayed in Figure S15, range between 0 and 1 and assume a metallic electrode with flat DOS and reorganization energy (λ) for [Ru(NH3)6] 3+/2+ of 1 eV [45,47,48].Thus, the total current can be expressed as: where n = 1 is the number of electrons in the reaction, F is Faraday's constant, Cred is the concentration of [Ru(NH3)6] 2+ at the electrode surface, and Cox is the concentration of [Ru(NH3)6] 3+ at the electrode surface.
The simulated LSVs were matched to the experimental data recorded on different TMDC thicknesses by using only kmax as the fitting parameter, with E0' set to -0.29 V as determined on a bare gold electrode in SI-8.At the bare gold surface, kmax,Au was determined to be at least 210 4 cm/s, as shown in Figure S16B.Note that kmax values greater than 210 4 cm/s yielded simulated LSV curves that were no longer distinguishable (Figure S16C).Our kmax,Au value is close to the pre-exponential frequency factor for the adiabatic heterogeneous electron transfer dominated by the reorganization energy predicted by the Marcus-Hush theory (10 4 -10 5 cm/s) [49,50] and also to the experimentally determined value (1.110 5 cm/s) obtained for [Ru(NH3)6] 3+/2+ on bare gold [51].
Matching simulation to each individual experimental LSV over the kinetic region studied materials, MoS2, WS2, MoSe2, and WSe2, are summarized in Table 1 and plotted versus the layer number in Figure 5A.
The responses on all four materials are similar and indicate a decrease in   with an increase in layer thickness, as shown in Figure 5A, except for kmax.2L for WSe2, which is significantly larger compared to the other TMDCs studied, although it remains unclear why this is the case.According to the Gerischer model for heterogeneous electron transfer, the prefactor   scales with the electronic transmission coefficient, which represents the probability of the electron transfer.The electron tunneling barrier arising from the TMDC material itself would reduce the transmission coefficient, and so   , as the number of layers increases.Previously an electron tunneling barrier model has been used to evaluate photocurrents obtained for 2D MoS2 in a bottomcontact configuration [52], and has also been used to assess the HER kinetics on mono and few-layer MoS2 [21].Therefore, assuming that tunneling is the dominant process leading to a decrease in   , the  .corresponding to nL TMDC can be described as: where  is the tunneling decay constant, ℎ is the thickness of the TMDCs layer (h = 6.15Å for MoS2, 6.47 Å for MoSe2, 6.16 Å for WS2 and 7.00 Å for WSe2) [7],  is the number of TMDCs layers.Diagram of tunnelling barrier can be found in Figure S22.
The  values for [Ru(NH3)6] 3+ reduction on different layer stacks for MoS2, MoSe2, WS2, and WSe2 obtained from Eq. 6 are reported in Table 1.The four different TMDCs have tunneling decay constants in the range 1.0 -0.4 Å -1 , which vary significantly depending on total thickness.Our β values are consistent for 2D TMDCs, which have previously been estimated to have tunneling decay constants of 0.512 Å -1 [53].Instead, the kinetics are faster for a larger number of layers than would be predicted by a tunneling model with a constant β.
The electronic band structure of TMDCs depends on the number of stacked layers [54][55][56].The band gap of MoS2, MoSe2, WS2, and WSe2 increases with the decreasing layer number (<5L) due to strong quantum confinement in thin layers [56,57].The larger band gap of thinner layers should therefore lead to an increase in the tunneling barrier height, and therefore larger tunneling decay constants (β).
Conversely, the smaller band gap of thicker layers should lead to smaller tunneling decay constants.Diagrams of tunnelling barriers presented Figure S22 illustrate changes of tunnelling barrier with number of layers stacked.Indeed, the derived tunneling decay constants listed in Table 1 agree with this trend.
The  − values describe n-layers of TMDCs acting as a single tunneling barrier, thus  − can be compared with the band gaps reported for n layers of TMDCs (also displayed in Table S5) [7].From Table S5, we observe a correlation who experimentally determined the tunneling barrier height on the different number of layers (1-5L) of MoS2 on gold using conducting AFM [58].The barrier height maximum is observed at 1L, then decreases and stabilizes at >5L, in good agreement with values reported in Table 1 obtained using an outer-sphere redox probe.
The electrochemical response on a few-layer 2D TMDCs, in a bottom-contacted configuration, depends on the electron transport process through the TMDC layers.
This electron transport is strongly influenced by the electronic structure of the 2D TMDCs, with larger band gaps resulting in a larger decrease in kmax with each additional layer, and therefore a higher electron tunneling barrier.However, 2D TMDCs with a fewer number of stacked layers show faster electron transport kinetics due to a narrower tunneling barrier, despite having a larger tunneling decay constant.As a consequence, bottom-contacted TMDC samples with a heterogeneous number of stacked layers will produce an electrochemical response dominated by the monolayer activity at lower overpotentials but with increasing activity of thicker layers at higher overpotentials (assuming there is no diffusional overlap).Our findings suggest that the electrochemical response of TMDC electrode surfaces is profoundly affected by the mode of contact established on electrodes.This has implications for the interpretation of electrochemical and, importantly, electrocatalytic experiments at TMDC nanoarchitectures.Through-layer or within-layer electron transport must be considered when interpreting the performance of TMDC materials in the HER or other reactions of relevance to advanced electrochemical applications.Trends in the electron transfer rates were described by considering an electron tunneling barrier arising from the TMDC itself.This allowed us to extract the apparent tunneling decay constants, β, which range over 1.0 -0.4 Å -1 across all TMDC materials tested; these values are consistent with those reported for tunneling decay constants of 2D TMDCs by other groups [53,58].

Conclusions
We observe that the prefactor, kmax, decays more slowly than we would expect if each additional layer added an identical electron tunneling barrier.Thus, the previously reported tunneling barrier model for HER on a few-layer MoS2, where tunneling decay constant was unaffected by the number of TMDCs layers [21], is insufficient to explain our experimental data.The correlation observed between band gap and tunneling decay constant suggest that non-exponential decay of rate constants is due to a change in the electronic properties of the tunneling barrier, resulting from band gap changes as a function of the number of stacked layers of TMDCs.Therefore, the variation on the electronic structure of the 2D material affects the electron transport to such an extent that we can observe the effect in our electrochemical measurements.S1.
up to the 1/3 of the diffusion-limited current) by adjusting   allows us to estimate the apparent   on the different layers ( .for nL), as shown in Figure 3C-3E for WS2.The median  .on MoS2, MoSe2, and WSe2 surfaces are shown in Figure S17-S19.The [Ru(NH3)6] 3+ reduction kinetics slows down with the increasing number of layers, as shown in Figure 4.This observation agrees with previous work on the HER at 2D TMDC electrodes [21].The median  .values extracted for all four between  − values and band gap on the different layer thickness of TMDCs.For instance, of the TMDCs we studied, WS2 has the largest band gap and WS2 also showed the largest tunneling decay constant,  − , values.Likewise, of the materials we considered, MoSe2 has the smallest band gap and also shows the smallest  − values.In fact, we observe a strong correlation between the band gap values sourced from literature and our measured tunneling decay constant,  − , values, as shown in Figure 5B.Our findings are supported by recent work by Lee et al.,

SECCM has been used
to quantify the effect of the number of stacked layers of 2D TMDCs (MoS2, MoSe2, WS2, WSe2) on the electrochemical response of the [Ru(NH3)6] 3+/2+ redox couple.Combining SECCM and AFM allowed us to precisely determine the exact nature of the underlying 2D material at each sample point, and to quantify the electrochemical response on the different numbers of layers.There is a significant difference in the electrochemical response on the different numbers of stacked layers, with more facile kinetics observed on thinner stacks of all four materials.The kinetic dependence was assessed by comparison with a finite element simulation that coupled the diffusional mass transport of [Ru(NH3)6] 3+/2 in the nanopipette probe with the Gerischer description of the electrochemical kinetics.We quantified the rate of electron transfer on a different number of stacked layers of four TMDCs by fitting a simulated response to the experimentally measured LSVs by only varying the maximum kinetic rate, kmax.MoS2, MoSe2, WS2, and WSe2 all show a decrease in the electron transfer rate constant as the number of layers increases.

Figure 2 :
Figure 2: A) Optical image of MoSe2 crystals with differing number of layers.B) AFM

Figure 3 :
Figure 3: A) Color-coded surface-type assignment in the AFM image showing the

Figure 5 :
Figure 5: A)   values determined from fitting simulation to experimental data for different layer thickness.Solid points correspond to each   value and dashed lines are displayed to guide the eye.Error bar correspond to   values for 25th and 75th percentiles.B)  − values plotted against band gap for 1-, 2-, and 3-layer MoS2, MoSe2, WS2, and WSe2 [7].Error bar correspond to  − derived from   values

Table 1 :
Median prefactor,   , and tunneling decay constant, β, evolution with the number of TMDCs layers The data in Table1show that the tunneling decay constant, β, decreases with increasing numbers of layers for all four TMDCs.However,  , values do not follow an exponential decay with increased layer thickness as we might expect if each additional layer added an identical electron tunneling barrier, as shown in Figure5A.