Direct imaging of dopant and impurity distributions in 2D MoS$_2$

Molybdenum disulfide (MoS$_2$) nanosheet is a two-dimensional material with high electron mobility and with high potential for applications in catalysis and electronics. We synthesized MoS$_2$ nanosheets using a one-pot wet-chemical synthesis route with and without Re-doping. Atom probe tomography revealed that 3.8 at.% Re is homogeneously distributed within the Re-doped sheets. Other impurities are found also integrated within the material: light elements including C, N, O, and Na, locally enriched up to 0.1 at.%, as well as heavy elements such as V and W. Analysis of the non-doped sample reveals that the W and V likely originate from the Mo precursor.

This is commonly gained through (scanning) transmission electron microscopy-energydispersive X-ray spectroscopy ((S)TEM-EDS) and X-ray photoelectron spectroscopy (XPS) to detect dopants and impurities. [22,23] Atom probe tomography (APT) represents an attractive alternative. APT is a burgeoning characterization technique allowing for mapping the elemental distribution in nanostructured materials with a unique combination of 3D capability, subnanometer spatial resolution, and ~10 ppm-level detection sensitivity for all elements irrespective of their mass. [24,25] APT shows great potential for nanomaterial characterization especially for dopant/impurity analysis.
Chemical vapor deposition (CVD) has been used to deposit single 2D layer or only a few stacked layers, which can be used in electronic application. [26] Less well-controlled assemblies of 2D layers synthesized by wet-chemistry have been used for catalytic applications. [27][28][29] Here, we studied wet-chemically self-assembled porous MoS2. This material exhibits a more complex morphology compared to a simple 2D layer and could in principle be used for both electronic and catalytic applications.
First, we studied the intentional doping of MoS2 with Re by using APT. We also reveal unintentional doping by impurity elements present during the synthesis and originating from the metal-precursor or the solvent. Both heavy elements, such as V and W, and light elements such as C, N, and O were reported to act as electron acceptor (p-type dopant) and Na was reported to be an n-type dopant, up to a few atomic percent of all these elements are detected in the nanosheets. Underpinned by the challenge associated with measuring low quantities of these impurity elements, the influence of the presence of both dopant and impurity elements on the activity of the material have been too often neglected. This tends to restrict strategies to optimize the materials performance to being empirical rather than guided by physics or chemistry.
We first synthesized Re-doped MoS2 nanosheets by using the one-pot wet-chemical method outlined by Xia et al. [19] Ammonium molybdate ((NH4)6Mo7O24) and thiourea (NH2CSNH2) precursors are dissolved along with ammonium perrhenate (NH4ReO4 for Re-doped MoS2) in distilled water and heated at 200 o C for 20h. The collected powder was then thoroughly rinsed with distilled water to remove excess impurities. The details are described in the Supporting Information.   Figure S1a) shows that Mo, S, and Re are apparently rather homogeneously distributed within the 2D assembled nanosheets. The details on peaks in EDS spectrum is described in Figure S1b. The introduced Re induces locally the formation of the 1T phase in 2H phase of MoS2 as shown in top view of high resolution TEM images (Figure 1b and c). The transition from 2H to 1T for the Re-doped MoS2 is explained by crystal field theory. [14,18,19] The Fermi level of Re-doped MoS2 is heavily shifted towards the conduction band minimum, and electrons are delocalized, donating electrons. These free electrons fill stable d-orbitals in the 1T-phase rather than those of 2H-phase according to the Hund's rule; therefore, the structure becomes more stable with Re content. [19] In order to map the distribution of Re in 3D and to verify the presence of impurity elements in the Re-MoS2 sheets, we performed APT. We used co-electrodeposition of a Ni-film to embed freestanding Re-MoS2 nanosheets, applying the method described in Ref. [30] , so as to enable the preparation of a sharp specimens (<100 nm at the apex). A similar approach had previously been used for nanoparticles [31][32][33][34][35][36][37][38][39][40][41] , but never for 2D materials.
Scanning electron micrographs obtained from the surface of the Ni/Re-MoS2 co-electroplated sample are displayed in Figure S2a and S2b. Following cross-sectional focused-ion beam milling (FIB), regions with a dark contrast appear in the micrographs in Figure S2c and S2d.
These regions are ascribed to assemblies of Re-MoS2 nanosheets encapsulated in the Ni-matrix, and, importantly, no noticeable voids can be detected. Site-specific APT specimen preparation was performed near one of these regions using the protocol outlined in Ref. [42] Experimental results from the Re-MoS2 APT measurement are presented in Figure S3. clearly are partition into the nanosheets. As expected, Re-related peaks were not found in the Ni/MoS2 by APT (see Figure S4) indicating that Re only originates from the Re precursor.
A cuboidal region-of-interest of the Re-MoS2 region viewed along z-axis from the acquired 3D atom map (Figure 2c) reveals that Na atoms are distributed along the MoS2 nanosheets. Na is not detected in the electroplated Ni (see Figure S5). Therefore, Na atoms are stemmed from the Mo precursor reagent ((NH4)6Mo7O24, Sigma Aldrich) as it contains 0.01 wt.% of Na by the chemical specification sheet. Across the nanosheets, approx. 0.1 at.% of Na is detected and segregated within Re-MoS2, forming clusters of approx. 4-5 nm in size (see Supporting   Information Table S1). From the density functional theroy (DFT) calculations, Na doping was reported to enhance charge transfer of the MoS2 by donating electrons causing, i.e. n-type doping. [43,44] Na is shown here incorporated within the nanosheets, and this is also rationalized by the excellent intercalation ability of MoS2 for alkali ions with a high capacity and stability reported for metal-ion batteries. [45] Furthermore, it has been reported that Na atoms are detected in exfoliated-geological and CVD-grown MoS2 nanosheets [46] as well as CVD-grown WS2 nanosheets. [47]  Besides undesired impurity of Na, unexpected heavy elements of V and W are also detected in both Re-doped and non-doped MoS2 nanosheets. Detailed mass spectra ranges are presented in Figure S6. Strong peaks are detected for 51 V + and all the natural isotopes of W + (see Table S2).
These elements are not found in the APT mass spectrum of the Ni matrix, i.e. where no Re-MoS2 are detected (see Figure S5), whereas they are also observed in the non-doped MoS2 (see Figure S7). V and W hence most likely come from the Mo precursor. According to the chemical specification sheet of the Mo precursor reagent ((NH4)6Mo7O24, Sigma Aldrich), it contains 0.001 wt.% of "heavy metals" impurities. Despite significant attention to extract high-purity Mo, the separation of Mo from V and W is reported to be very difficult, since they have similar chemical properties [48][49][50] and are known impurities within Mo. [51] Therefore, V and W likely remain alongside with Mo.
We performed X-ray photoelectron spectroscopy (XPS) on the undoped and Re-doped MoS2 samples to look for trace amounts of V, W, and Na (see Figure S10 and S11). No other impurities except C, N, and O were found due to the detection limit of XPS. Addou et al. used inductively coupled plasma mass spectrometry (ICP-MS) and time-of-flight mass spectrometry (TOF-MS) on CVD-grown MoS2 nanosheets and found V and W within MoS2. [46] Likewise, they did not observe the corresponding peaks of these elements by XPS.
DFT calculation for V0.08Mo0.92S2 showed increased electronic properties, such as 40 times increase in the in-plane conductivity and 20 times higher carrier concentration than MoS2. This change in properties led to higher catalytic activity for the hydrogen evolution reaction. [54] However, the bulk atomic compositions of V and W in the Re-MoS2 are only 419 and 206 ppm, respectively. The position of W is assigned to the same hexagonal parent structure of MoS2 sharing the metal sites with Mo [52,53] and V doping is also considered as intralayer doping in MoS2. [53] The bandgap of the semiconducting phase (2H structure) of VS2 (1.87 eV) [55] and WS2 (1.91 eV) [56] are similar to that of MoS2 (1.78 eV) [56] . Moreover, VS2, WS2, and MoS2 lattice constants are 0.317, 0.318, and 0.318 nm, respectively, that are very close to each other. [57] The effect of the presence of V and W in such low compositions here could have only a limited effect on the electronic properties of MoS2, yet this would need to be confirmed by complementary local electronic-and catalytic property measurement.  The atomic concentration ratio of S to Mo in the Re-doped MoS2 nanosheets are 1.7 which is lower than the expected stoichiometry of MoS2. It has been reported that single-layer MoS2 oxidizes under ambient condition [58] as S vacancies are formed through oxidation spontaneously followed by O substitution process. [59] Using DFT calculation, the enthalpies of each reaction step were calculated to be -0.49 eV for S vacancy formation and -0.39 eV for O saturation, which implies that the oxidation is hence thermodynamically favorable. [60] Moreover, it is well known that surface of MoS2 naturally oxidizes on surface. [61,62] This reaction results in a complex molecular structure of MoS(2-x)Ox. [63] Peaks pertaining to MoSO + molecular ions are detected (see Figure S3b) and the 1D profile across the reconstructed MoS2 nanosheets clearly show that the presence of O is not limited to the surface but rather through the MoS2 assemblies as shown in Figure 4a. As a result, the stoichiometric ratio of S+O to Mo gives 1.9. Some S vacancies are expected since S vacancies are reported to be the most abundant defects in MoS2 [64] and Re doping, which pushes the MoS2 Fermi level close to its conduction band, could result in favorable formation of S vacancies. [65] Furthermore, it is also possible that some of the S is lost during field evaporation, as has been observed for some covalently bonded materials [66] , but the correlation histograms calculated here do not give indications of such specific losses as shown in Figure S9. respectively. The incorporation of C in MoS2 is reported to facilitate the transfer of the photogenerated electrons and holes that makes the photocatalytic reaction more efficient enhancing photocatalytic reaction [67] likewise the N incorporation improves the electronic conductivity of MoS2 showing outstanding performance for hydrogen evolution reaction. [68] However, it was not possible to detect such elements locally at relatively low concentration with surface science spectroscopy. In contrast, our approach of using APT and TEM allows to obtain ppm-level chemistry information with high spatial resolution for 2D materials.
In conclusion, direct imaging of the as-synthesized Re-doped MoS2 nanosheets is done using TEM and APT. The Re dopant is shown to have been incorporated within the MoS2 structure, with an average composition of 3.8 at.%, and we clearly observe other impurities from heavy (V, W) to light element (C, N, O, Na), with Na showing a tendency for clustering. Our results indicate that elements from the precursor were incorporated into the nanosheets during its synthesis. The chemistry of the MoS2 synthesized using wet-chemical method is much more complex than expected and often reported. Controlling the synthesis environment is crucial to avoid contamination. We expect that our approach, amenable to other nanomaterials, will help understanding the role of dopant and impurity elements on the activity of MoS2 nanosheets.

Experimental Section
Re-doped and non-doped MoS2 nanosheet synthesis 0.989 g of (NH4)6Mo7O24•4H2O (Sigma-Aldrich (Germany), ACS reagent, 99.98% trace metals basis) and 2.284 g of NH2CSNH2 (Sigma-Aldrich (Germany), ACS reagent, ≥99.0%) were dissolved in 10 ml of distilled water along with or without 0.376 g of NH4ReO4 (Sigma-Aldrich, 99.999% trace metals basis) and a homogeneous transparent solution was prepared. Then, the solution was poured into a Teflon container placed in a stainless-steel autoclave and heated at a fixed temperature of 200 o C. After 20h, the autoclave was cooled down to room temperature and the black powder was collected. Then the powder was repeatedly washed thrice with ethanol and distilled water for each centrifugation and re-dispersion steps. For MoS2 materials, oxygen plasma cleaning for impurities removal of residuals could not be used since the O plasma induces substitutional oxidation of MoS2 resulting MoSO/MoO3 solid solution crystal. [60] Heat treatment in calcination method for removals could result MoC from assynthesized MoS2 [69] and this surface cleaning method could not efficiently remove all C-based molecules. [70] Therefore, for removal of residuals, we choose the water-washing treatment. [71] Finally, the rinsed powder was dried at room temperature.

Sample preparation: co-electrodeposition process
As-synthesized MoS2 nanosheets were electrodeposited within Ni film for embedding nanoparticles according to Kim et al. [72] Nickel sulfate hexahyrdrate (NiSO4 6H2O, Sigma-Aldrich (Germany)), and boric acid (H3BO3, Sigma-Aldrich (Germany)) were dissolved in distilled water. The nanosheets were then dispersed in as-prepared electrolyte solution and poured in a vertical cell for co-electrodeposition process. The vertical cell including a Cu substrate and a Pt-mesh counter electrode was used. A positive bias was applied to Pt electrode and electrodeposition of MoS2 nanosheets and Ni were performed at constant current of -19 mA for 500 sec.

TEM analysis
TEM was performed to investigate morphology and crystal phase of Re-MoS2 using  Titan Themis operated at 300 kV with a Cs-corrector for the image forming lens. Chemical composition was analyzed using EDS in STEM mode (60- 300 Titan Themis operated at 300 kV with a Cs-corrector for the probe).

XPS analysis (ask Olga) APT analysis
Needle-shaped APT specimens were prepared from MoS2/Ni composite film using focused ion beam (FIB) (Helios 600) according to Thompson et al. [42] CAMECA LEAP 5000 XS system in pulsed laser mode at a specimen temperature of 50K was used for nanosheets APT analysis. A laser pulse energy of 80 pJ and a pulse frequency of 125 kHz were set. Data reconstruction was done using the Imago visualization and analysis system (IVAS) 3.8.4 developed by CAMECA instruments. The standard voltage protocol was used for all data-set reconstruction. [73] Supporting Information Supporting Information is available from the Wiley Online Library or from the author.        Figure S6. Sectioned mass spectra of Re-MoS2 result from Figure S3b for (a) a V + peak and (b) W 2+ peaks.   In order to quantify whether Re exhibits a tendency to segregate, we performed a Re-Re nearestneighbor analysis [3] on an individual nanosheet that was first extracted from within a Re-doped MoS2 nanosheets dataset. The distribution was compared with a randomized Re distribution, in which the atomic positions are unchanged by the mass-to-charge ratios are randomly swapped. Figure S8 shows that the experimental Re-Re nearest-neighbor distribution has no significant deviation from the randomized Re distribution. A frequency distribution analysis was also performed, and the corresponding Pearson coefficient (µ), which gives a quantitative assessment of the randomness associated to a a χ 2 -statistical test. [4] The coefficient value lies between 0 and 1, where µ = 0 corresponds to complete randomness and µ = 1 is completely ordered. The calculated µRe value acquired from the Re-MoS2 nanosheets data is 0.044, which qualitatively proves that Re is homogenously distributed within the as-synthesized Re-MoS2.
These analyses were performed within the IVAS 3.8.4 software suite.