Enhanced performance of in-plane transition metal dichalcogenides monolayers by configuring local atomic structures

The intrinsic activity of in-plane chalcogen atoms plays a significant role in the catalytic performance of transition metal dichalcogenides (TMDs). A rational modulation of the local configurations is essential to activating the in-plane chalcogen atoms but restricted by the high energy barrier to break the in-plane TM-X (X = chalcogen) bonds. Here, we theoretically design and experimentally realize the tuning of local configurations. The electron transfer capacity of local configurations is used to screen suitable TMDs materials for hydrogen evolution reaction (HER). Among various configurations, the triangular-shape cobalt atom cluster with a central sulfur vacancy (3CoMo-VS) renders the distinct electrocatalytic performance of MoS2 with much reduced overpotential and Tafel slope. The present study sheds light on deeper understanding of atomic-scale local configuration in TMDs and a methodology to boost the intrinsic activity of chalcogen atoms.

M oS 2 is a promising candidate to replace Pt for electrocatalytic hydrogen evolution reaction (HER) due to its environmental friendly and low cost characteristics [1][2][3] . While increasing the conductivity via forming heterojunction bilayer with a conductive substrate can promote the overall catalytic performance [4][5][6] , the performance of pristine MoS 2 is restricted by the density of active sites [7][8][9][10] . The pursuit to maximize MoS 2 utility inspires researchers to explore various ways to rouse the activity of inert sulfurs in the MoS 2 basal plane. For example, edge-site engineering 9,11 , phase transformation 12,13 , amorphization 14,15 , and in-plane doping/vacancy modifications [16][17][18][19][20] have been reported. Notably, changing the local configurations 18,[21][22][23][24] by introducing atomic defects (doping or vacancy) is preferable as defected MoS 2 exhibits better stability compared to transformed 1T′ phase 25 and amorphous MoS 2 10 . However, the reported activity of in-plane sulfur enhanced by local configuration modification is still far from that of Pt-based catalysts 26 . This is because the large energy are required to break the in-plane Mo-S bonds. In fact, few types of atomic local configurations have been realized in the basal plane of MoS 2 [27][28][29][30] , so the tuning ability of local configurations is quite limited so far 16 . Hence, in order to improve the intrinsic activity of in-plane sulfur atoms, it is essential to understand the intrinsic correlation and explore new methodologies to enrich stable and highly efficient local configurations.
Herein, we conducts both computational and experimental investigations in order to establish a correlation between local configuration and the electrocatalytic activity of monolayer MoS 2 . A group of stable local configurations with non-noble period-IV single atom or clusters (Co, Fe, V, and Cr) accompanying additional sulfur vacancy in the in-plane domain of MoS 2 have been attained. Given the correlation between binding strength and local configurations electronegativity, the activity of in-plane sulfur can be regulated by electron transfer capacity of local configurations. The peculiar triangular-shape Co atom cluster surrounding one sulfur vacancy configuration (viz., 3Co Mo −V S ) is identified by both calculation and experiments to be most efficient to activate the inert sulfur sites. Correspondingly, a distinct enhancement in HER activity is achieved (η 10 : 75 mV and Tafel: 57 mV dec −1 ), exhibiting the highest intrinsic HER activity among MoS 2 materials. The microcell HER measurements show a volcano-like relationship between content of specific local configuration and activity, which corroborates the optimized concentration of 3Co Mo −V S . Therefore, as demonstrated in the present work, it is possible to further activate the in-plane sulfur sites by rational engineering of the local configurations. This result may provide a route to unleash the electrocatalytic potential of TMD materials for efficient hydrogen generation in acidic solutions.

Results
Design efficient and stable local configurations. The activation of the basal plane in TMDs have been extensively studied to achieve the stable structure and enhance their catalytic activity [31][32][33] . Sulfur vacancy (Vs) on the surface is an electron donor and can induce a localized gap state in MoS 2 . Below a critical carrier density, the transport of donor states is governed by nearest-neighbor hopping at high temperatures and variable-range hopping (VRH) at low temperatures 23,34-36 . Regional charge states around a defect structure are suggested to make an important contribution to regulating the catalytic activity. Based on the above analysis, we believe that it is reasonable to monitor the defects induced Bader charge fluctuation, and the H adsorption to define the active sites through DFT computational screening.
To study the TM and synergistic effect of Vs on sulfur sites, we have considered six configurations including TM atoms (TM: Co, V, Fe, and Cr; TM amounts from 1 to 3) with or without Vs (structures see in Supplementary Fig. 1 and Supplementary Note 1) that are set as models to screen stable catalytic structures through DFT calculations. The hydrogen adsorption free energy (ΔG H ) is an effective descriptor to predict the activity for various catalyst systems 37 . The ideal value of ΔG H is 0 eV, which corresponds to a thermoneutral state of the adsorbed atomic hydrogen and efficient proton/electron transfer and hydrogen release 1 . The correspondingly calculated ΔG H of monolayer MoS 2 with varied local configurations are further exhibited in Supplementary Fig. 2 Fig. 3), the 3Co Mo −Vs, 3Fe Mo −Vs, 1V Mo and 1Cr Mo are identified as the most stable structures in the different possible TM-introduced MoS 2 (Fig. 1a). Together with the activity (value of ΔG H ), the 3Co Mo −Vs is expected to be the potential structure with both good stability and high activity. The predicted HER activity of MoS 2 with different local configurations following the trend 3Co Mo −Vs > 1V Mo > 3Fe Mo −Vs > 1Cr Mo (Fig. 1b). This trend remains the same with solvation correction, as demonstrated by our calculations with the implicit solvation model ( Supplementary Fig. 4). The hydrogen adsorption free energy (ΔG H ) on basal plane of intact MoS 2 is far away from the optimal value. After tuning by local configurations, the ΔG H value of −0.085 eV comparable to that of Pt 38 , is achieved due to the much stronger bonding strength in S atoms with the assistance of 3Co Mo −Vs, which surpasses predicted activity of edge sites 8 . As expected, different configurations induce varied activity; all the structures with co-existence of Vs and TM Mo atoms synergistically tunes the ΔG H when compared to single one (Supplementary Table 1). In addition, the monolayers are used instead of porous 3D materials, minimizing the double layer effect induced by porosity. It is supported by CV curves in non-Faradaic region (blue rectangular in Supplementary Fig. 5) with nearly no hysteresis loop.
It is important to reveal the underlying mechanism of enhanced catalytic activity due to the local configuration. The above analysis indicates that the defects (TM substitution and Svacancy) and H adsorption could induce charge fluctuation of the regional structure due to electron delocalization of MoS 2 . In principle, the catalytic activity depends on the charge transfer capacity before and after H adsorption. To identify the effective catalytic structure, we show the nearest and the next-neighbor atoms which possibly induce a charge fluctuation in HER (Supplementary Fig. 6 and Supplementary Note 2). First, the nearest metals (nMo and doped (3-n)TM, n = 0, 1, 2) and the adsorption S1 atom have relatively large change in charge ( Fig. 1d-f, Supplementary Figs. 7-10). As a result, we consider (3-n)TM-S-nMo as the first-order catalytic structure that is comprised of TM substitutes, adsorption S atom, and the nearest Mo atoms. In contrat, the change in charge for the next-neighbor S and Mo atoms is relatively small. So they are considered as the second-order structure as the distance from the adsorption site is large. As a result, it is reasonable to assume the charge regulation of the second-order catalytic structure has a negligible effect on that of the first-order one. The radial distributions of charge distribution are presented in Supplementary Figs. 7-10. Therefore, in our study, we calculate the total charge difference of adsorption S atom and the nearest metals to depict the charge transfer capacity to S-H bonds. The amount of charge transfer of local configuration (namely, atoms to induce the charge transfer includes: nearest nMo, doped (3-n)TM, n = 0, 1, 2 and adsorption S1 atom) is linearly correlated with ΔG H . This result indicates a charge regulation effect by the local configuration on HER activity (Fig. 1c). The linear correlation indicates that charge transfer capacity induced by varied local configurations are mainly delocalized in the first-order catalytic structure instead of on individual sulfur atoms. We found that, a charge difference around 0.07e − (which corresponds to ΔG H = 0 eV) should correspond to a high HER catalytic activity.
Realization and characterizations of local configurations. In light of the superior activity induced by the predicted local configurations, we employ the chemical vapor deposition (CVD) method to synthesize several monolayer MoS 2 samples with various in-plane local configurations (Methods). The optical images of Co, Fe, Cr, and V-containing MoS 2 monolayers are shown in Supplementary Fig. 11a-d. Raman spectra confirm that all the as-prepared samples preserve the lattice structure of MoS 2 ( Supplementary Fig. 12), as seen from the characteristic A 1g mode at~401 cm −1 and the E 2g 1 mode at~381 cm −1 observed in pristine MoS 2 monolayer 39 . In addition, the Raman mappings indicate homogenous elemental distribution ( Supplementary   Fig. 13). Atomic force microscopy (AFM) measurements further confirm that the as-prepared MoS 2 domains are monolayers ( Fig. 2a-d) with a thickness ranging between 0.7 and 0.9 nm. As for the the system with small doping concentration, the peak-shift is ascribed to the dopant induced Fermi level movement 40 . However, the shifts of X-ray photoelectron spectra in both Mo 3d and S 2p are very small (below 0.3 eV, see Supplementary Fig. 14), likely due to the low dopant concentrations. Therefore the minor peak shift cannot justify if the dopants incorporate into the MoS 2 lattice. More evidence is provided by the high-resolution spectra of TM 2p ( Supplementary Fig. 15), which show clearly the formation of metal-sulfur bonds in all samples and support the substitutional dopants within the MoS 2 lattice.
The annular dark-field (ADF) scanning transmission electron microscopy (STEM) imaging and electron energy loss spectroscopy (EELS) are used to further confirm the local atomic configurations.   typical Mo atoms but similar to S 2 columns due to the nature of the STEM imaging 27 . The line intensity profile of the single isolated dopant site and comparison to the STEM simulation ( Supplementary Fig. 16) confirm that the TM atoms successfully doped into the lattice rather than the adatoms on the surface of MoS 2 . A more careful inspection reveals two main types of local configurations as predicted by previous theoretical calculations: one TM atom substitutes the Mo site forming an isolated single TM site, as marked by the white circles in all four TM-containing MoS 2 monolayers (Fig. 2g, h, k, l); three TM atoms forming a triangular cluster with a connecting central sulfur vacancy, named as 3TM Mo −Vs, as highlighted by green circles in Co and Fecontaining system (Fig. 2e, f, i, j). However, we find the 3TM Mo −Vs are the dominating configuration in the Co-and Fe-containing MoS 2 monolayers. The single isolated TM sites account for very small ratio (<10%) compared to the 3TM Mo −Vs (Supplementary Figs. 17, 20c, and 21c; Supplementary Notes 3, 6, and 7), which corroborates that the latter are the dominating causes towards the HER activity.
On the other hand, Cr and V form predominantly isolated single TM sites in the MoS 2 lattice. This is due to the different formation energy of the two types of local configurations with different TM atoms. The single atom EELS measurements on the TM in each image further confirm the chemical identity of the corresponding introduced element, as recognized by the sharp L edges of Co, Fe, V, and Cr, respectively, offering strong evidence of the presence of TM atoms and the consistence of the predicted local configurations. The reference EELS spectra taken away from the dopant site ( Supplementary Fig. 18 and Supplementary Note 4) confirms the observed sharp peaks in the spectra of Fig. 2 are not an artifact during the collection at the dopant site.
Electrochemical test of MoS 2 with local configurations. To verify the predicted HER activity of in-plane sulfur modulated by designed local configurations, the effects of 3Co Mo −Vs, 3Fe Mo −Vs, 1Cr Mo , and 1V Mo on HER catalytic activity are examined using a three-electrode electrochemical cell in an electrolyte containing 0.5 M H 2 SO 4 . Data are compared to pristine MoS 2 and commercial Pt/C. Before LSV tests, electrochemical activation was implemented. Stable CVs of configured MoS 2 samples after electrochemical activation process (black curves in Supplementary Fig. 19 and Supplementary Note 5) indicate no phase change during activation. Figure 3a shows linear-sweep voltammograms (LSV) in the cathodic direction after the correction of ohmic potential drop (i.e., iR), where the currents are normalized to the electrode geometric area. It is seen that the pristine MoS 2 with an overpotential of 317 mV at 10 mA cm −2 (η 10 ) shows an inferior HER activity than those of configured MoS 2 . The optical image of pristine MoS 2 as shown in Supplementary Fig. 22 shows the similar edge length with that of configured MoS 2 , excluding the edge effect on different activity. Both the 3Fe Mo −Vs and the 1Cr Mo samples exhibit values of η 10 over 200 mV. On the contrary, the 3Co Mo −Vs and the 1V Mo show significantly reduced η 10 values down to below 150 mV. In particular, the 3Co Mo −Vs has a lowest η 10 value of only 75 mV. The first cycles of CV for HER were operated to verify the structural stability of introduced basal configurations after activation ( Supplementary Fig. 19). The trend of activity shown in the first CVs of configured MoS 2 exhibits the same as that of LSV curves. Note that the Co atoms in 3Co Mo −Vs configuration has an over 90% occupancy among the total Co atoms. Hence, even though a minor content (<0.2 at%) of 1Co configuration is detected, we believe the 3Co Mo −Vs configuration plays the   dominating role in the catalyst activity. This is also in concise with the calculation that ΔG H of 3Co Mo −Vs is closest to 0 eV (Supplementary Table 1). The corresponding Tafel plots show the same trend with that of η 10 (Fig. 3b). The 3Co Mo −Vs configuration sharply reduces the Tafel slope from 175 mV dec −1 in pristine MoS 2 to 57 mV dec −1 . And the 1V Mo sample gives an acceptable value of 68 mV dec −1 . In comparison, the 3Fe Mo −Vs and 1Cr Mo configurations have little effect to the Tafel slope. Therefore, the lowered Tafel slopes of the 3Co Mo −Vs and 1V Mo with a fast discharge process of protons 41 (Supplementary Note 8), may reflect a strengthened capability to adsorb H. The Faradic efficiency was determined from the produced H 2 characterized quantitatively by gas chromatography. As shown in Supplementary Fig. 23, the 3Co Mo −Vs sample exhibits >98% efficiency over the time scale of the measurement, confirming the H 2 as the dominating product during the whole electrolysis process. From the overall comparison (Fig. 3c), we can conclude that the 3Co Mo −Vs configuration, with synergistic triangular Co clusters surrounding one Vs in the center, renders MoS 2 monolayer the best HER catalytic performance among all the configured MoS 2 samples 16,42-48 . Supplementary Table 2 provides an extensive comparison to other TMDs and non-noble metal catalysts in their electrocatalysis of HER. The performance of our MoS 2 monolayer with 3Co Mo −Vs configuration exceeds all the pure TMD monolayer catalysts, and also compete with other non-noble metal catalysts.
The turnover frequency (TOF) per sulfur is calculated in order to correlate the intrinsic activity per sulfur atom with the local configuration (Supplementary Note 9). Each sulfur site tuned by 3Co Mo −Vs or 1V Mo possesses much higher efficiency than that in 3Fe Mo −Vs and 1Cr Mo samples with increased value of TOFs. Compared to other configurations, the 3Co Mo −Vs and 1V Mo samples demonstrate the most appropriate tuning on the charge transfer capacity of local configuration.
Electrochemical impedance spectroscopy with fitted circuit models ( Supplementary Fig. 24) shows significantly decreased charge-transfer resistances (R ct ) for the 3Co Mo −Vs (30.3 Ω) and the 1V Mo (47.9 Ω) samples, as compared to those of 3Fe Mo −Vs (91.5 Ω) and 1Cr Mo (100.3 Ω), indicating a facilitated charge transfer between the S and protons in electrolyte (Fig. 3e). In addition, the 3Co Mo −Vs sample exhibits an extraordinary longterm operation durability with small changes in potential ( Fig. 3f and Supplementary Fig. 25). Hence, we may conclude that the 3Co Mo −Vs configuration is efficient for HER during the whole cycling process.
Microcell measurements. In addition to the effect of local configuration type, it is expected that the concentration of such local configurations can also influence the amount of active sulfur sites. In order to prove the concentration effect, the on-chip electrochemical micro-devices are fabricated from a set of 3Co Mo −Vs samples with local probe test, as shown in Fig. 4. Figure 4a, b show the three-electrode setup for the electrochemical measurements (more details are shown in Supplementary Fig. 26). In the first step, a controlled linear I-V scan is done on the PMMA layer ( Supplementary Fig. 27 Fig. 4c-e. We can clearly see that an optimal Co concentration, corresponding to η 10 < 100 mV, should be around 3.8 at%, which translates to the 3Co Mo −Vs concentration of~1.2 at%. The excessive increase in 3Co Mo −Vs concentration results in performance decline. Possible reasons for this are deteriorated surface stability 16 and superfluous lattice distortion 18 . This concentration effect also corroborates the key contribution of 3Co Mo −Vs rather than 1Co Mo to the HER catalyst activity enhancement. In addition, the microcell HER measurements are implemented on 3Co Mo −Vs, 3Fe Mo −Vs, 1V Mo , and 1Cr Mo samples with similar defect concentrations ( Supplementary Fig. 28), showing the similar trend with that of three-electrode measurements (Fig. 3a).

Discussion
We have promoted the per-site electrochemical activity of inplane sulfur sites of MoS 2 monolayer via tuning H-S bonding strength, which can be understood by a hypothetical model of activating the inert sulfur atom into an open valence state. That activation can be correlated with the charge transfer capacity of local configuration. This is realized by forming various local configurations of transition metal atom or clusters (Co, Fe, V, and Cr) and compensative sulfur vacancy (Vs), which are confirmed by STEM images. In particular, the in-plane sulfur atoms modulated by 3Co Mo −Vs configuration render the most active MoS 2 -based HER electrocatalyst in acidic medium to date (an overpotential η 10 of 75 mV). The optimized 3Co Mo −Vs configuration is also verified by the systematic DFT calculations. In addition, the suitable Co concentration for the HER performance is achieved by the in-situ probe measurements of microcell. Our work highlights the potency of local configuration engineering in boosting the in-plane electrocatalytic activity of MoS 2 , as well as possibly other 2D TMD monolayers.

Methods
Synthesis of configured and pristine MoS 2 . Pure MoS 2 and configured MoS 2 were synthesized by CVD method using MoO 3 and sulfur (Sigma) as the precursor. For different TM-doped MoS 2 , V 2 O 5 , CrCl 3 , Fe 2 O 3 , and Co 3 O 4 were used as the corresponding TM sources. The synthesis was conducted using a quartz-tube single-zone furnace (1-inch diameter) in a temperature range from 550 to 650°C. Specifically, for the growth of pure MoS 2 , a quartz boat containing 10 mg MoO 3 powder was put in the center of the tube, and the SiO 2 /Si substrate was placed on top of the quartz boat with the front side facing down. Another quartz boat containing 0.5 g sulfur powder was put upstream. The temperature ramped up to 700°C in 15 min, and was maintained at the peak temperature for 5 min to 10 min.
During the reaction, a constant 80 sccm Ar flow was used as the carrier gas. After the reaction, the furnace cooled down naturally to room temperature. Structural characterizations. Room temperature Raman measurements were performed using a WITEC alpha 300 R Confocal Raman system with an excitation laser of 532 nm. The Raman system was pre-calibrated based on the Raman peak of crystalline Si at 520 cm −1 . The laser power was kept below 1 mW to avoid sample heating. The TEM samples of the 3Co Mo −Vs, 3Fe Mo −Vs, 1V Mo , and 1Cr Mo structured MoS 2 were prepared as follows. A layer of poly (methyl methacrylate) (PMMA) was spin-coated on the sample surface with a thickness of~1 µm, and then baked in an electric oven at 180°C for 3 min. Afterwards, the substrates were immersed in a NaOH solution (1 M) overnight to dissolve the SiO 2 layer. After liftoff, the MoS 2 samples were washed with DI water for several cycles. Then the monolayer samples were fished by a TEM grid (Quantifoil Mo grid). The obtained TEM specimen were dried naturally in ambient environment, and then dipped into high-purity acetone overnight to remove the PMMA layers. The STEM investigation was performed at room temperature on an aberration-corrected Nion UltraSTEM-100 and a JEOL 2100 F with a cold field-emission gun and an aberration corrector (the DELTA-corrector), both operating at 60 kV.
Electrochemical measurements. PMMA methylbenzene was uniformly spun on the SiO 2 /Si substrates deposited with monolayer MoS 2 . After baking at 100°C for 5 min, the PMMA film covered substrates were immersed in a 5 M KOH solution. As a result of the etching effect by KOH, the monolayer MoS 2 samples with the PMMA film were detached from the SiO 2 /Si substrate. Then, the obtained monolayer MoS 2 /PMMA films were washed in DI water and overlaid on the glassy carbon rotating disk electrode (RDE). After the thorough evaporation of DI water between the RDE electrode and the MoS 2 /PMMA films, the PMMA films were further removed by dipping into acetone. As a result, the glassy carbon RDE electrode covered by monolayer MoS 2 were obtained 45,46 .
For the electrochemical measurements, a standard three-electrode cell consisting of the glassy carbon RDE as the working electrode, a graphite carbon counter electrode and a saturated calomel reference electrode (SCE) was used. The electrolyte solution was 0.5 M H 2 SO 4 . An electrochemical workstation (CHI760) coupled with a RDE system (AFMSRCE3529, Pine Research Instrumentation, USA) was used to control the cell. The potential versus the reversible hydrogen electrode (RHE) was calculated according to E RHE = E SCE + E°S CE (0.2412) + 0.059 × pH. Before HER test, the catalysts went through an electrochemical activation process by cyclic voltammetry scanning in the same electrolyte (0.5 M H 2 SO 4 ) with a scan rate of 100 mV s −1 in the potential range of 0.1 to −0.29 V (vs. RHE). Linear sweep voltammetry (LSV) measurements were conducted with a scan rate of 2 mV s −1 under 1500 rpm. The current vs. potential plots were corrected by 90% ohmic compensation. The electrochemical impedance spectroscopy (EIS) were obtained in the same three-electrode configuration in the frequency range of 100 KHz to 0.1 Hz and at an applied current of 10 mA cm −2 . For the stability assessment, polarization data were measured in the beginning and after 1000 CV sweeps (−0.2 and +0.2 V vs. RHE, scan rate: 50 mV s −1 ). In addition, the constantcurrent (10 mA cm −2 ) measurements were also implemented to evaluate the stability of potential. The on-chip electrochemical measurements were carried out following the previous report 11 . Briefly, structured-MoS 2 monolayers were transferred onto SiO 2 (300 nm)/Si substrate with pre-made gold fingers by PMMA assisted wet transfer method. Monolayers were further patterned into domains by e-beam lithography and 5-s treatment in nitrogen plasma. Contacts between gold fingers and monolayers were made via e-beam lithography and gold deposition processes. Microcell reaction windows were made by e-beam lithography on 1-μm-thick spincoated PMMA layer. During measurements, gold fingers connecting configured MoS 2 monolayers, Ag/AgCl encapsulated by Luggin capillary and carbon rod were used as working, reference and counter electrodes respectively. In all, 5 μl of 0.5 M H 2 SO 4 (degassed with Ar bubbling for 10 min) was used as electrolyte for each test. The scan rate for the LSV tests were 10 mV/s. Computational methods. The density functional theory (DFT) calculations were perfomed using the Vienna Ab initio simulation package 49,50 . The generalized gradient approximation with the Perdew−Burke−Ernzerhof exchange−correlation fuctional and a 450-eV cutoff for the plane-wave basis set are employed 51 . The projector-augmented plane wave was adopted to describe the electron−ion interactions 52 . The em piracal dispersions of Grimme (DFT-D2) was applied to account for the long-range van der Waals interacions 53 . All calculations were spin-polarized and the convergence threshold was set to be 10 −4 eV in energy and 0.01 eV/Å. The k-point sampling of the Brillouin zone was obtained using a 4 × 4 × 1 by the Monkhorst−Pack scheme. In addition, a 5 × 5 × 1 supercell was also used to confirm the sufficiency of 4 × 4 × 1 supercell (Supplementary Fig. 29). In the electronic structure calculation, denser k-points (8 × 8 × 1) were used for better accuracy. The vacuum slab of 15 Å was inserted in the z-direction for suface isolation to eliminate periodic interaction. The free energy of the adorbed state was calculated as where ΔE H* is the hydrogen chemisorption energy, and ΔE ZPE is the difference of the zero point energy between the adsorbed state and the gas phase. Considering the fact the vibriation entropy of H* in the adsorbed state is very small, the entropy of 1/2 H 2 adsorption can be approximated as ΔS H ≈ −1/2S 0 H 2 , where S 0 H 2 is the entropy of H 2 in the gas phase at the standard conditions.

Data availability
All relevant data are available from the authors.