Single-layer transition metal dichalcogenides (TMDCs) based on MoS₂ have garnered a lot of attention in both fundamental and applied research over the last several decades. A classic example of such an MoS₂-based TMDC is Co-promoted MoS₂ which is very important for many globally-relevant applications involving optoelectronics, heterogeneous catalysis, and electrocatalysis. For instance, catalysts based on Co-promoted MoS₂, a transition metal dichalcogenide formed by sulfiding mixed Co-Mo nanoparticles, are used for reducing global SO_x emissions via hydroprocessing in petroleum refineries [1], mixed alcohol synthesis [2], selective olefin hydrogenation [3], and selective mercaptan synthesis [4, 5]. Co-promoted MoS₂ is also widely used as a noble-metal-free electrocatalyst for the hydrogen evolution reaction (HER) [6] and Li-ion battery electrodes [7]. Particularly, its 1T metastable counterpart, has been of great relevance to van der Waals heterostructure-based optoelectronic devices due to its room temperature ferromagnetism [8–11].

In the laboratory, the 1T and 1T’ analogues of pristine and promoted MoS₂ can be synthesized by various strategies. One of the techniques involves using alkali metals like Li, Na, or K during the MoS₂ synthesis. The alkali metals intercalate in the MoS₂ van der Waals gap and stabilize the 1T and 1T’ phases, thus lowering the bandgap for the metastable phase formation with Mo in an octahedral coordination environment [12–17]. This technique has been used to synthesize and explore applications of Na- and K-doped single-layer MoS₂ and Co-promoted MoS₂ for gas-phase and HER catalysis [5, 18]. Intercalation with electron-donating aromatic amines has also proven to be successful in inducing the phase transformation from 1H to 1T [19]. An alternative synthesis strategy for the 1T or 1T’ MoS₂ phase involves using a sulfur-deficient environment during the synthesis by lowering the chemical potential of sulfur [15, 20]. Additionally, using argon ion bombardment of 2H MoS₂ has been shown to generate local 1T mosaic structures [21]. The sulfur vacancy may also be generated by doping with a foreign metal atom of a different formal charge, for instance, Ru or substitutionally co-doping Ni and Co, to generate the 1T phase from the 2H phase [22, 23]. Similarly, oxygen doping has also been used as a technique to generate sulfur vacancies, and thereby, induce the formation of the 1T phase [24]. Electrochemical incorporation of S vacancies to induce a 2H–1T phase transition has also been reported [25]. Chalcogen vacancies in other Mo TMDCs have also been shown to induce a phase transition to the 1T and 1T’ counterpart [26]. These experiments have generated a lot of excitement in the scientific community, because 1T polymorphs of MoS₂ have been shown to have high hydrogen evolution activity due to its metallic nature, unlike the 1H counterpart, and have been explored for applications in energy production and storage devices, in addition to applications in spintronics due to room temperature ferromagnetism [27–30]. Especially, many metastable 1T and 1T’ phases of Co-doped MoS₂ have been demonstrated to be very competitive noble metal-free alternatives for green hydrogen production via water splitting, which has generated a lot of interest in the synthesis of 1T and 1T’ counterparts of Co-promoted MoS₂ [6, 31, 32].

Based on these experimental findings, we hypothesize that it should be possible to directly synthesize 1T Co-promoted MoS₂ slabs from metallic Co and Mo nanoparticles by maintaining highly reducing environments during the synthesis, e.g., through a combination of hydrogen gas and the sulfiding agent. Use of hydrogen during the crystallization is expected to generate S-vacanies on the edges and the basal plane. According to the recent work of Jin et. al. [20], formation of edge S defects and edge sulfur saturations of less than 50% can kinetically favor the formation of the 1T phase by layer sliding. Additionally, the ease of hydrogen dissociation on under-coordinated and metallic Mo [33, 34] during the synthesis may even allow for some stabilization of the Co-promoted MoS₂ slabs through hydride intercalation. Based on the works of Mom et. al. and Grønborg et. al. [33, 34], such reducing conditions can be achieved by using background pressures of hydrogen approaching several bars, typically not possible in an ultra-high vacuum (UHV) setup. Furthermore, using a lower temperature for sulfidation may arrest the transition to the 2H phase. For instance, heating K-promoted MoS₂ to 650 K and above, under high gas pressures has been observed to induce irreversible transition to the 1H phase [5]. Burkhanov et. al. have shown that hydrogen intercalation achieved using H plasma under vacuum environments leads to lattice expansion of MoS₂ similar to that from the 1H to 1T phase transition [35].

In order to test our hypothesis, we have sulfided a precursor containing mixed Co-Mo nanoparticles supported on Au(111) under high-pressure sulfo-reductive conditions and observed in situ the crystallization process using the ReactorSTM setup. At temperatures of up to ∼500 K, a disordered CoMoS_x phase is observed to form which then crystallizes into single-layer 1T Co-promoted MoS₂ when the temperature is raised to 600 K. Sulfiding an identical CoMo precursor using H₂S under vacuum results in the formation of 1H Co-promoted MoS₂ slabs which do not transform into the 1T counterpart when exposed to the same sulfo-reductive conditions, suggesting that reducing conditions are necessary during the crystallization step to form the 1T phase. XPS spectra acquired post synthesis show that the 1T Co-promoted MoS₂ slabs are largely sulfur-deficient, suggesting the role of sulfur vacancies in metastable phase formation. Thus, we demonstrate that the 1T Co-promoted MoS₂ can be readily synthesized and stabilized by using a lower sulfidation temperature and sufficiently sulfo-reductive conditions, without the need for any additional intercalating agents. Furthermore, the ability to observe the formation of a metastable 1T Co-MoS₂ phase in situ (i.e., while it happens) with atomic resolution under 1 bar of gas pressure, elevated temperature, and aggressive chemical conditions, is the main novelty of this work, as it has remained an experimental challenge to do so for several decades. Additionally, the experimental observation that elevating the gas pressure to attain sufficiently reducing conditions to favor the formation of 1T Co-MoS₂ without the need of an intercalating agent is a very important finding that provides us a fundamental understanding of the 1T phase formation process and is also an additional novelty. To the best of our knowledge, the experimental work presented in this paper has not been carried out elsewhere in the past.

We first prepare identical precursors containing Co-Mo mixed nanoparticles supported on a clean Au(111) substrate by sequential physical vapor deposition (PVD) of 0.2 monolayers (ML) Mo metal, followed by 0.2 ML Co metal at room temperature. The exact synthesis procedure is detailed in the Materials and Methods section and in the Supplementary Materials.

First, as a control experiment, one of the identical Co-Mo precursors was sulfided at 650 K under 2 × 10⁻⁶ mbar of H₂S and cooled to room temperature according to the recipe detailed in the Materials and Methods section, in order to replicate the vacuum synthesis recipe of single layer 1H Co-promoted MoS₂ slabs on Au(111) [48]. After this step, the sample was loaded into the ReactorSTM operated in the UHV mode. Figure 1 shows a large-scale STM image thus obtained. In Figure 1, we observe the formation of atomically-flat hexagonal slabs which are identified as single-layer 1H Co-promoted MoS₂ slabs. Under the sulfur-rich conditions used in this experiment, pristine MoS₂ slabs are known to display only the Mo-terminated edges due to the thermodynamic stability of the Mo-termination and hence, adopt a triangular shape [49]. The substitution of Mo atoms by Co along the S-terminated edges leads to the thermodynamic stability of the S-termination as well, thus, driving the formation of hexagon-shaped slabs such as those in Figure 1 [33]. The Co-promoted MoS₂ slabs have a bright outline along their periphery while their basal planes are imaged relatively darker. The darker basal plane is attributed to the semi-conducting nature of 1H Co-promoted MoS₂ while the bright edge features are attributed to the 1D metallic edge states called BRIM sites [33]. Additionally, we also observe the formation of the single-layer 2D CoS₂ phase as a byproduct, with the characteristic hexagonal moiré structure which is 7 × 7 Au atoms (1.95 nm × 1.95 nm) wide. The 2D CoS₂ phase is also observed to form between the Co-promoted MoS₂ slabs on the Au(111) steps. These observations are in excellent agreement with our recent work on 2D CoS₂ slabs supported on Au(111) [46]. Furthermore, the observation of the complete conversion of the Co and Mo nanoparticles into the respective TMDC phases after the sulfidation process is in agreement with the prior experimental works on the sulfidation of Co and Mo nanoparticles under controlled vacuum conditions [33, 49, 50].

In order to test our hypothesis on the formation of 1T Co-promoted MoS₂ slabs under highly reducing conditions, a freshly prepared identical Co-Mo precursor was sulfided using high-pressure sulfo-reductive conditions within the ReactorSTM. For this purpose, a sulfo-reductive gas mixture (9:1 H₂:CH₃SH) was used to pressurize the ReactorSTM to 1 bar. Thereafter, the temperature was raised from room temperature to 520 K at a rate of 2 K/min. Figure 2A shows a large-scale STM image obtained in situ during the sulfidation under the sulfo-reductive gas mixture. We observed the formation of two phases: islands with a somewhat hexagonal shape and disordered arrangement of bright protrusions and large atomically-flat islands with an ordered hexagonal pattern, both of which are seen in Figures 2A, B.

The somewhat hexagon-shaped islands are identified as those of poorly crystalline molybdenum sulfide (MoS_x). Due to the presence of a sufficient number of diffusing cobalt atoms, we expect that any MoS_x phase formed will likely be doped with Co atoms, and hence, we denote this phase as CoMoS_x. The CoMoS_x islands are measured to be 2.8 Å high (see heightline, Figure 2B), matching closely with the measured height of single-layer Co-promoted MoS₂ from previous experimental reports [33]. The CoMoS_x islands are likely an intermediate phase towards the formation of crystalline Co-promoted MoS₂. No further changes are observed in this disordered CoMoS_x phase up to 67 min into the scanning in the sulfo-reductive gas environment at 520 K (see Figures 2C, D), suggesting that the temperature is insufficient for any crystallization to occur.

The atomically-flat slabs with a hexagonal moiré pattern on their basal plane are identified as 2D CoS₂ islands. These slabs were also observed to form in our control experiment (see Figure 1). The 2D CoS₂ also retains the characteristic hexagonal moiré structure indicating that the basal plane of 2D CoS₂ does not undergo significant structural changes under the sulfo-reductive gas mixture used. Additionally, the formation of this moiré structure shows that the 2D CoS₂ islands are crystalline in nature and that the sulfidation of Co nanoparticles is already complete during the process of bringing the ReactorSTM to the operating conditions. This rapid spread and reaction of Co is attributed to the cluster diffusion behavior of Co nanoparticles in the presence of S adatoms which leads to rapid growth of 2D CoS₂ [50].

To speed up the process of CoMoS_x crystallization, we further increased the temperature to 603 K at the rate of 2 K/min while maintaining the sulfo-reductive reaction gases and thermally stabilize the STM before commencing the scanning in situ. Figure 3A shows a large-scale STM image obtained in situ at 603 K at t = 121 min. Ten minutesafter attaining 603K, the edges of the CoMoS_x islands are observed to appear slightly brighter in the STM images, as can be seen in Figure 3B. Over the next 35 min, the CoMoS_x slabs are observed to crystallize from the step edges towards the interior (see Figures 3B–F). The crystallization is evident from the formation of an ordered hexagonal lattice. The new crystalline phase has a hexagonal arrangement of bright protrusions which are 0.33 ± 0.01 nm apart (see Figure 3F; height profile in Figure 3G). This matches closely with the basal plane S-S distance of single-layer MoS₂ and Co-promoted MoS₂ [49]. Furthermore, the basal plane protrusions are imaged bright with respect to gold, unlike the 1H Co-promoted MoS₂ in Figure 1, suggesting that this phase has a metallic nature. The crystalline Co-promoted MoS₂ formed under sulfo-reductive conditions (see Figures 3B–D), however, does not have a bright BRIM along the edges, unlike the 1H Co-promoted MoS₂ slabs in the control experiment (see Figure 1). All characteristics of this new crystalline phase formed in the sulfo-reductive environment agree very well with those reported for a metastable 1T-MoS₂ phase in which Mo is in an octahedral coordination environment [15, 28]. Therefore, we identify this new phase as that of single-layer 1T Co-promoted MoS₂. Figures 3B–D also show that the crystalline phase is not resolved well along one of the crystallographic directions especially along the slow scan direction (top to down), giving the appearance of lines. We attribute this to tip assymetry arising from interaction with the gases, especially CH₃SH at the elevated temperature and pressure used. Such effects were also observed by Mom et. al. under similar gas conditions when using similar as-cut Pt-Ir tips [34].

One may expect that employing a high partial pressure of hydrogen during the cyrstallization favors the formation of a large number of sulfur vacancies along the Co-MoS₂ edges and the basal plane. Recent theoretical work [20] has shown that increasing the number of sulfur vacancies can make translational layer sliding to form the 1T configuration more kinetically favorable in pristine MoS₂ slabs. While it may also be argued that the electron donation from Co also aids in the formation of 1T Co-promoted MoS₂, merely having the Co alone does not yield the formation of the 1T phase under the vacuum sulfidation-based control experiment. We also consider the possibility that using a different sulfiding agent may also aid in the formation of the 1T phase as we have used H₂S for our control experiment and CH₃SH for the high-pressure sulfidation as the sulfiding agents. Research carried out in the past has shown that the effect of sulfiding agents on the chemical potential of sulfur is minor for small thiol-containing molecules like H₂S and CH₃SH, as they all readily disssociate on the gold surface and form adsorbed–SH while aromatic (such as thiophenes) and doubly substituted thiols (like dimethyl sulfide) lead to a lower chemical potential of sulfur [33], [51–53]. Additionally, previous experiments using a variety of H₂:H₂S mixtures under vacuum pressures for MoS₂ synthesis did not lead to the formation of the 1T phase [49]. Therefore, the elevated H₂ gas pressure used in our experiment to form a sufficient number of sulfur vacancies remains as a major factor that assists in the formation of 1T Co-promoted MoS₂.

We make use of XPS analysis to determine the number of sulfur vacancies and the amount of Co incorporation into the edges of 1T Co-promoted MoS_2. As the 1T phase is metallic, BRIM sites are not present on the edges, making the identification of the Co-substituted S edges from the STM images alone difficult. XPS, on the other hand, can detect and resolve Co present in 1T MoS₂ and CoS₂ phases, albeit in a statistically averaged manner [9, 54]. Typically, the Co present on the S edges of 1H and 1T MoS₂ has a signal resembling metallic Co sulfide but shifted by 0.6–0.8 eV to higher binding energy depending on the metallicity of the support [9, 54, 55]. Additionally, the structural relaxation upon the transformation of the 1T phase into the 2H phase results in a downshift in binding energy of the Mo 3days and S 2p spectra of the 1T phase by ∼0.9 eV with respect to the 1H phase, thus allowing for the identification of the 1T phase [56, 57]. Before measuring the XPS spectra, the sample exposed to high pressure was cooled to room temperature in the sulfo-reductive gas environment and then pumped down to UHV. Thereafter, XPS spectra of Mo 3d, S 2p and Co 2p were acquired and analyzed. The Mo 3d spectrum (see Figure 4A) shows the main Mo 3d_5/2 peak at 282.3 eV which matches well with the reported XPS spectra of 1T-MoS₂ and 1T-Co-promoted MoS₂ [54, 57]. The component at 290.2 eV is attributed to the unconverted MoS_x phase as we did not observe any 1H Co-promoted MoS₂ in the STM images. The Co 2p_3/2 spectrum (see Figure 4B) shows the presence of 2 types of cobalt species, namely, the Co in metallic 2D CoS₂ at 778.1 eV (55%) and the Co present on the edges of MoS₂ at 778.7 eV (45%). The corresponding S 2p spectrum (see Figure 4C) shows an asymmetric peak that consists of contributions from S in 2D CoS₂ and the 1T phase. Furthermore, comparing the overall signals of Co, Mo, and S, the molecuar formula of 1T Co-MoS₂ is determined to be Mo_0.67Co_0.32S_1.89, with 11% sulfur vacancies in the top layer (see Supplementary Material S1.2, Supplementary Table S2). The high number of sulfur vacancies is expected due to the highly reducing environment used during the synthesis in the ReactorSTM. In comparison, the molecular formula of the 1H Co-promoted MoS₂ from the control experiment is determined to be Mo_0.65Co_0.35S_2.05.

As an additional control experiment, the control sample containing 1H Co-promoted MoS₂ slabs grown using 2 × 10⁻⁶ mbar H₂S was loaded into the ReactorSTM and exposed to the high-pressure sulfo-reductive gas mixture (9:1 H₂:CH₃SH, 1 bar) at 603 K for up to 6 h. Figure 5 shows a large-scale STM image obtained in situ after 6 h. Comparing Figures 1, 5, we observe that the morphology of the Co-promoted MoS₂ slabs is largely preserved. Under these gas conditions, we expect that the edges of the Co-promoted MoS₂ slabs are reduced to the same degree as in our high-pressure sullfidation experiment since the gas conditions are identical. In Figure 5, we clearly observe that all the Co-promoted MoS₂ slabs remain in the 1H state and do not transform into the 1T state, as can also be seen in the atom-resolved inset where 1H Co-promoted MoS₂ slabs with BRIM sites are clearly imaged. This shows that the 1T state is accessible at the gas conditions and temperatures used, if and only if, the thermodynamically stable 1H state has not already been formed. A schematic summarizing all our findings is shown in Figure 6.

We have demonstrated using the in situ STM imaging capabilities of the ReactorSTM, that using sufficiently reducing environments during the crystallization phase can favor the formation of metallic 1T Co-promoted MoS₂. On the other hand, if 1H Co-promoted MoS₂ is already formed, then there are large kinetic barriers that hinder the formation of a sufficient number of S vacancies to drive the transformation to the 1T phase through atomic S-layer sliding. These observations with the STM are confirmed and supported very well by the corresponding XPS spectra which show that the 1T Co-MoS₂ slabs have a large number of sulfur vacancies in comparison to the 1H Co-MoS₂ slabs. The formation of pure 1T Co-promoted MoS₂ under sulfo-reductive conditions without an alkali metal intercalator is remarkable as it demonstrates the feasibility for directly synthesizing monolayers of pure 1T or 1H Co-promoted MoS₂. This is of great relevance for fundamental research in optoelectronics, green hydrogen production, and heterogeneous and electro-catalysis.