Growth of Low-Defect WSe₂ Film via High-Purity van der Waals Crystal Precursor

Abstract

Two-dimensional (2D) semiconducting transition metal dichalcogenides (TMDs) exhibit exceptional electrical and optical properties, empowering their promising prospects for future nanoelectronics. Despite major advances in n-type 2D semiconductors, the field has yet to synthesize high-mobility p-type 2D TMDs, in particular WSe₂, and systematically query the influence of defects. In this study, we unveil the pivotal role of substitutional impurity defects vis-à-vis the precursor used and growth method employed in defining the quality of 2D p-type WSe ₂ . Density functional theory calculations suggest the adverse effect of Fe-, Co-, Ni- and Si-substituted W impurity defects on the mobility of WSe₂, whereas defects such as O-, S-substituted Se and Mo-substituted W pose negligible impact. Guided by the theory, we pinpoint van der Waals (vdW) crystals, commonly used in mechanical exfoliation, as the optimal precursor, and develop a facile vdW crystal physical vapor deposition (PVD) method to grow high-purity monolayer 2D WSe ₂ film (VPVD-WSe₂) that is continuous across a centimeter scale. A suite of spectroscopies confirms the markedly reduced defect density of the as-synthesized WSe ₂ compared to those by typical chemical vapor deposition methods, and by PVD with commercial or hydrothermal precursors. Scanning tunneling microscopy further evidence the ultralow substitutional impurity defect density of VPVD-WSe₂, greatly outperforming the control samples and approaching the mechanically exfoliated counterparts. The VPVD-WSe ₂ based field-effect transistors exhibit notable electrical performance with record-high field-effect hole mobility up to 112 cm² V^–1s^–1 at room temperature, exceeding the best-reported monolayer WSe₂ synthesized by chemical vapor deposition and rivaling the mechanically exfoliated 2D WSe₂ flakes.

Two-dimensional (2D) semiconducting transition metal dichalcogenides (TMDs) demonstrate excellent electrical properties, such as high carrier mobility, tunable bandgap, and superb electrostatic control, rendering them promising candidates to extend Moore’s Law for next-generation electronics. − Among 2D TMDs, WSe₂ stands out owing to the ambipolar characteristics, empowering its versatility across multiple electrical and optoelectronic applications with complementary functions. − Nevertheless, the mobility of scalable 2D p-type WSe₂ monolayers, prevalently fabricated by the chemical vapor deposition (CVD) method, still lags behind the n-type 2D semiconductors , and falls short of the theoretical limit. −

Point defects have been widely believed to constrain the electrical properties of 2D TMDs. − However, controversy persists over the influence of defects on the mobility. For example, charge trapping Se vacancies were regarded as the primary cause of the low WSe₂ mobility. This notion was then challenged by scanning tunneling microscopy (STM), revealing that oxygen tends to overwhelmingly passivate Se vacancies to form oxygen-substituted Se (O_Se) defects in WSe₂. − Subsequent theoretical studies suggested that such defects do not introduce in-gap states and may not affect the WSe₂ mobility; − meanwhile, Xiao et al. inferred that filling chalcogen vacancies with oxygen should in general improve the mobility of 2D TMDs. The field was further puzzled by recently discovered impurity-substituted W (Im_W) defects, albeit lacking in-depth mechanistic insights. It is therefore highly desired, but not yet realized, to explore how specific types of defects jeopardize the mobility of 2D WSe₂.

Physical vapor deposition (PVD), an emerging technique for crafting high-quality 2D thin materials by directly sublimating the precursors, may well serve this purpose. ,− The commercial PVD precursors usually manifest a maximum purity of about 99.8%. , At high temperatures, the impurities are volatilized together with the precursors, leading to the formation of substitutional impurity defects. This allows us to explicitly probe the influence of different types of defects on 2D WSe₂ mobility.

Here, we ascertain the role of various defects on the mobility of WSe₂, underscoring the importance of controlling the substitutional impurity defects for synthesizing high-mobility 2D materials. Density function theory (DFT) calculations suggest that the in-gap states introduced by Fe-, Co-, and Ni-substituted impurity defects impose an adverse effect on the charge mobility of WSe₂. On the contrary, defects such as O-, S-substituted Se and Mo-substituted W exert no such impact. The theoretical prediction guides us to develop a van der Waals (vdW) crystal PVD (VPVD) method for the growth of high-purity continuous monolayer p-type WSe₂ film at centimeter scale. The as-synthesized WSe₂ manifests a drastically reduced substitutional impurity defect density of 8 × 10¹⁰ cm^–2, as evidenced by a suite of microscopies and spectroscopies. We then showcase our VPVD-WSe₂ on field-effect transistors (FETs). Consistent with the ultralow substitutional impurity defect density, the VPVD-WSe₂ based devices exhibit remarkable electrical performance with a hole mobility of 112 cm²V^–1s^–1 at room temperature, among the highest for WSe₂ monolayers grown by vapor deposition.

Selection and Characterizations of Precursors

The theoretical predictions (Figure S1 and discussions) motivated us to manipulate substitutional impurity defects for high-mobility WSe₂. We posit that this can be realized using the PVD technique since the precursors are sublimated and redeposited, and the defect density of the yielded products may be highly dependent on the quality of precursors (Figure a). To test this postulation, we adopted three WSe₂ precursors with distinct grades – that is, commercially available, − hydrothermally synthesized, and chemical vapor transportation (CVT)-grown vdW crystal sources (Figure S2a; details in the Methods). The commercial and hydrothermal samples appeared gray and dark (Figure S2b, c), while the vdW crystal sample possessed a highly glossy crystalline appearance (Figure S 2d). Scanning electron microscopy (SEM) confirmed the morphological disparity among the three precursors: the commercial and hydrothermal powders exhibited irregularly flaky and nanoflower-like morphologies, respectively (Figure S2e, f), whereas the vdW crystal presented a highly crystalline hexagonal structure (Figure S2g).

1.

Schematic illustration of WSe₂ monolayer growth and characterization of precursors. (a) PVD growth of WSe₂ monolayers using precursors with distinct grades. (b) GD-MS, (c) XRD and (d) ESR spectra of commercial, hydrothermal and vdW crystal WSe₂ precursors.

We then conducted elemental analysis to quantitatively differentiate the grades of the precursors. Glow discharge mass spectrometry (GD-MS) revealed the presence of impurities including Mo, S, Si, Fe, Ni, and Co. Glow Discharge Mass Spectrometry (GD-MS) is an advanced analytical technique to assess the composition of solid samples, especially for identifying trace elements and impurities. By integrating glow discharge ionization with mass spectrometry, it enables comprehensive analysis of elemental composition. The concentrations of these impurities in CVT-grown vdW crystals were observed to be several orders of magnitude lower than those in the other two control samples (Figure b). We further carried out X-ray diffraction (XRD) through multiple samplings, which evidenced the existence of impurity phases (marked with asterisk) in both the commercial and hydrothermal precursors. Conversely, the XRD spectrum acquired on vdW crystals aligned well with the standard WSe₂ pattern (PDF#87–2418), suggesting the negligible presence of impurity phases (Figure c and Figure S3). Additionally, electron spin resonance (ESR), capable of discerning the defects, indicated that both commercial and hydrothermal powders manifested a peak at g = 2.0, while no obvious peak was detected on vdW crystals (Figure d). This signifies the charge state linked to Se vacancies in the source materials, and the increased peak intensity implies a denser Se vacancy population. We then mechanically exfoliated WSe₂ monolayers (ME-WSe₂) from the CVT-grown bulk vdW crystal precursor to reflect its level of defect (Figure S4). The room temperature Raman spectroscopy of the ME-WSe₂ monolayer showed strong E/A_1g and 2LA­(M) peaks at 250 and 259 cm^–1, respectively, both of which can be assigned to WSe₂ (Figure S4b). The room temperature Photoluminescence (PL) emission suggested a narrow fwhm of 21 nm (Figure S4c). These observations reaffirm the exceptional quality of the as-grown vdW crystals.

Synthesis and Characterizations of High-Purity WSe₂

The high quality of CVT-grown vdW crystals drove us to synthesize large-area, low-defect-density WSe₂ monolayers through PVD. We chose the reverse-flow PVD system (Figure S5) to prevent the undesired crystal nucleation with precise monolayer thickness control, and realized the growth of centimeter-scale continuous monolayer WSe₂ polycrystalline film on sapphire using the vdW crystal precursor (VPVD-WSe₂, Figure a). The VPVD-WSe₂ film was first assessed by optical microscopy, PL, and Raman intensity mapping over a 100 μm × 100 μm area, all of which indicated the exceptional uniformity of the sample (Figures b, c and Figure S6). Raman line scan with a step size of 0.02 cm also confirmed the homogeneity of the VPVD-WSe₂ monolayer film: the WSe₂ characteristic peak intensity remained largely unchanged across a distance of 1 cm (Figure d). Atomic force microscopy (AFM) determined the thickness of the VPVD-WSe₂ film to be 0.75 nm (Figure e).

2.

Characterizations of WSe₂ monolayers. (a) Photograph, (b) optical microscopy and (c) PL mapping of a continuous monolayer VPVD-WSe₂ film (scale bar, 20 μm). (d) Color-coded Raman peak profiles across a distance of 1 cm on a VPVD-WSe₂ film. (e) AFM image of the VPVD-WSe₂ monolayer film grown on sapphire substrate (scale bar, 3 μm). (f) Raman, (g) selected PL spectra and (h) statistical distribution of PL FWMH for CPVD-WSe₂, HPVD-WSe₂, CVD-WSe₂ and VPVD-WSe₂ monolayers on SiO₂/Si. (i) PL fwhm of PVD-grown, CVD-grown ,, and ME-WSe₂ − reported in the literature.

We further examined the quality of the VPVD-WSe₂ flakes, benchmarked by commercial (CPVD-WSe₂), hydrothermal (HPVD-WSe₂) and conventional CVD (CVD-WSe₂) synthesized WSe₂. All the samples were monolayers as confirmed by AFM (Figure S7). Both the Raman and PL spectra showed WSe₂ peaks with considerably narrower fwhm on VPVD-WSe₂ than the other three control samples (Figure f, g). Moreover, Raman and PL mappings revealed the nonuniform WSe₂ peak intensity distributions on CPVD-WSe₂, HPVD-WSe₂ and CVD-WSe₂ in sharp contrast to VPVD-WSe₂ (Figure S8e–l). We then assessed the defect levels in the as-prepared WSe₂ monolayers by quantitatively analyzing one hundred room temperature PL spectra that were collected from multiple flakes. The statistics indicate an average fwhm of 24 nm for VPVD-WSe₂, significantly lower than that of CPVD-WSe₂ (34 nm), HPVD-WSe₂ (35 nm), CVD-WSe₂ (32 nm) and other up-to-date CVD-grown WSe₂, and approaching the ME-WSe₂ (Figure h, i). These findings affirm the strong dependence of PVD-grown WSe₂ on the quality of their precursors, and exhibit the notably high quality of VPVD-WSe₂ with extremely low defect density.

We then sought to evaluate the quality of the as-prepared WSe₂ at atomic level. Scanning transmission electron microscopy (STEM) scans over large (30 nm × 30 nm) and small (5 × 5 nm²) areas were first performed. The STEM images in Figures S9 and 10 showed reduced cluster defects in VPVD-WSe₂ monolayers, while CPVD-WSe₂ and HPVD-WSe₂ manifested abundant Se defects, including cluster defects. We further used scanning tunneling microscopy (STM) to accurately probe the type and density of defects, which is beyond the capability of STEM. To avoid cross-contamination during the transfer process, each WSe₂ monolayers were directly grown on highly oriented pyrolytic graphite (HOPG) substrates. HOPG substrates were employed as the growth substrate in light of its high conductivity, which allows for enhanced STM imaging. The defect levels in the as-grown WSe₂ monolayers on HOPG were comparable to those grown on other substrates (for example, SiO₂/Si and sapphire), as examined by the room-temperature PL measurements (Figure g and Figure S11). The 50 nm × 50 nm STM images evidenced a substantially denser population of structural defects on CVD-WSe₂, CPVD-WSe₂ and HPVD-WSe₂ compared to VPVD-WSe₂, in line with previous spectroscopic observations (Figure a-c and Figure S12). The structural defects can be further classified into three types – that is, oxygen-substituted bottom Se (O_Se bottom), oxygen-substituted top Se (O_Se top) and Im_W, as featured in the magnified STM images and dI/dV spectra (Figure d and Figure S13). , Other defects, such as S_Se, exhibited a minimal presence (Figure a-c and Figure S12), thereby posing negligible impact on hole mobility, consistent with our theoretical predictions. Statistical analysis (Figure e and Supplementary Table 2) unveiled an O_Se(top) defect density of 1.8 × 10¹² cm^–2 on VPVD-WSe₂, 1 order of magnitude lower than that of CPVD-WSe₂ (2.8 × 10¹³ cm^–2), and of the same order of magnitude as HPVD (4.4 × 10¹² cm^–2) and CVD-WSe₂ (4.6 × 10¹² cm^–2). The O_Se(bottom) defect density of 1.76 × 10¹² cm^–2 on VPVD-WSe₂, 1 order of magnitude lower than that of CPVD-WSe₂ (2.17 × 10¹³ cm^–2), and of the same order of magnitude as HPVD (3.4 × 10¹² cm^–2) and CVD-WSe₂ (1.96 × 10¹² cm^–2). Likewise, the Im_W defect density of VPVD-WSe₂ (8 × 10¹⁰ cm^–2) was 2 orders of magnitude lower than that of CPVD-WSe₂ (4.12 × 10¹² cm^–2), HPVD- WSe₂ (5.6 × 10¹² cm^–2) and 1 order of magnitude lower than that of CVD-WSe₂ (8 × 10¹¹ cm^–2). To rigorously evaluate the overall defect density, STM scans were conducted on eight additional, randomly selected 50 nm × 50 nm areas of multiple VPVD-WSe₂ monolayers (Figure S14a–h). The average defect density of VPVD-WSe₂, including O_Se bottom, O_Se top and Im_W, were calculated to be (4 ± 0.5) × 10¹² cm^–2 (Figure S14i), comparable to those synthesized by metal–organic CVD (MOCVD) and mechanical exfoliation that typically yield low-defect-density WSe₂ (Figure f). ,, The WSe₂ mechanically exfoliated from our bulk crystals exhibited 7.5 × 10¹⁰ cm^–2 O_Se defect density and 3.75 × 10¹⁰ cm^–2 Im_W defect density (Figure S15). The difference in the defect densities between the VPVD-WSe₂ and ME-WSe₂ is likely caused by the VPVD process. High-resolution X-ray photoelectron spectroscopy (XPS) was then performed to identify the impurity elements. Fe, Co, and Ni 2p peaks were exclusively observed on CPVD-WSe₂, whereas no such peaks were detected on VPVD-WSe₂ (Figure S16). The obvious presence of Fe, Co, and Ni impurities in CPVD-WSe₂ and their low concentrations in VPVD-WSe₂ agree well with GD-MS results of the corresponding precursors (Figure b). The VPVD-WSe₂ also matches with the theory-predicted defect profile of a high-quality WSe₂ (Figure S1), primed for ensuing electrical measurements.

3.

Defect analysis by STM. (a–c) STM images (scale bar, 10 nm) of (a) CPVD-WSe₂ (bias voltage (V) = 2 V, current (I) = 40 pA), (b) CVD-WSe₂ (V = 1.3 V, I = 8 pA) and (c) VPVD-WSe₂ (V = 2.2 V, I = 10 pA). (d) Magnified STM images of WSe₂ monolayers showing O_se top, O_se bottom and Im_W defects (scale bar, 1 nm), and the corresponding side-view atomic structures. (e) Comparison of O_se and Im_W defect densities in HPVD-, CPVD-, CVD- and VPVD-WSe₂. (f) Comparison of O_se and Im_W defect densities in VPVD-, ME- , and MOCVD-WSe₂.

Electrical Performance of WSe₂ Based FETs

The theoretical calculations and characterizations suggest that the mobility of our VPVD-WSe₂ may be exceptional. To test this hypothesis, we fabricated VPVD-WSe₂ based FETs by transferring the centimeter-scale continuous monolayer WSe₂ film onto a SiO₂/Si substrate. Figure a showed the configuration of a FET with Pt/Au as contact electrodes (20 nm Pt followed by 20 nm Au) and 270 nm SiO₂ as gate dielectrics. A 0.8 mm × 0.7 mm optical image illustrated the representative array of back-gated FET devices and the zoomed-in image showed a typical FET with a channel length/width (L_ch/W) of 3 μm/12 μm (Figure b). The output curves (drain-to-source current, I _d , versus drain-to-source voltage, V _ds ) of a typical VPVD-WSe₂ FET showed linear characteristics (Figure c). The transfer curves (I _d versus gate-to-source voltage, V _gs ) of one hundred sixty-five randomly selected VPVD-WSe₂ FETs at V _ds = −1 V exhibited a p-type semiconductor behavior (Figure d), with a maximum hole field-effect mobility (μ _FE ) of 112 cm²V^–1s^–1 at a stable on/off ratio (I _on /I _off ) of 1.4 × 10⁸ at room temperature (Figure e). The temperature-dependent transfer curve of a typical VPVD-WSe₂ FET showed that the mobility increased with decreasing temperature, achieving a hole mobility of 175 cm²V^–1s^–1 at 15 K (Figure S17). Statistics of the tested FETs signified the notable electrical properties of the VPVD-WSe₂ film – that is, a hole μ _FE of 69 ± 11.3 cm²V^–1s^–1, an I _on /I _off of (3.4 ± 2.0) × 10⁸, and an on-state current (I _on ) of 9 ± 3.1 μA/μm (Figure f and Figure S18). We then benchmarked the electrical performance of VPVD-WSe₂ against CPVD-WSe₂, HPVD-WSe₂ and CVD-WSe₂. The average hole μ _FE of HPVD-WSe₂, CPVD-WSe₂ and CVD-WSe₂ based FETs were measured to be 0.31, 14, and 29 cm²V^–1s^–1, respectively, significantly lower than that of VPVD-WSe₂ (Figure g and Figure S19). The contact resistance values of CPVD-WSe₂, HPVD-WSe₂, CVD-WSe₂ and VPVD-WSe₂ FETs with Pt contacts were extracted by the transfer length method (TLM) to exclude the effects of the electrodes. The contact resistances for the HPVD-WSe₂, CPVD-WSe₂, CVD-WSe₂ and VPVD-WSe₂ FETs were 20, 8.5, 13.9, and 8 kΩ μm at n _2D of 3.9 × 10¹² cm^–2, 3.7 × 10¹² cm^–2, 6.6 × 10¹² cm^–2 and 5.5 × 10¹² cm^–2, respectively (Figure S20). As a result, the enhanced mobility in VPVD-WSe₂ can be predominantly attributed to the reduction in Im_W defect density. The poor electrical performance of the three control samples originates from their high-defect-density precursors, in line with DFT calculations. Moreover, the low defect density of the VPVD-WSe₂ significantly reduced the hysteresis of the FET, indicating its low border traps and interface states (Figure S21). To eliminate the influence of the dielectric substrate and ensure a fair comparison, we have fabricated the CVD-WSe₂ devices on the same 270 nm SiO₂ gate dielectric. The statistical analysis (Figure S22) of the hysteresis revealed that the VPVD-WSe₂ FETs still exhibit an order-of-magnitude lower hysteresis than the other types of WSe₂ devices. Figure h and Supplementary Table 3 summarized the μ _FE of state-of-the-art 2D WSe₂ based FETs as a function of I _on /I _off . Our VPVD-WSe₂ not only represents the best among PVD- and CVD-grown WSe₂, but also performs comparably with (or better than) the mechanically exfoliated single/multilayer WSe₂, albeit with higher O_Se concentration (Figure f). These results again corroborate that the WSe₂ mobility can be selectively mediated by certain types of substitutional impurity defects, consistent with the theoretical predictions.

4.

Electrical performance of WSe₂ based FETs and benchmarks. (a) Schematic of a monolayer WSe₂ based FET. (b) Optical microscopy image of the fabricated back-gated FET arrays (scale bar, 1 mm). Inset: zoomed-in view of a FET device (scale bar, 20 μm). (c) Output characteristics of a typical VPVD-WSe₂ FET. (d) Transfer curves of one hundred sixty-five VPVD-WSe₂ FETs at V _ds = −1 V. (e) Transfer curve for record-high hole μ _FE of 112 cm²V^–1s^–1 at an I _on /I _off of 1.4 × 10⁸ and V _ds = −1 V. (f) Statistical distribution of hole μ _FE , I _on /I _off and I _on for the tested one hundred sixty-six VPVD-WSe₂ FETs. (g) Transfer curves of CPVD-WSe₂, HPVD-WSe₂ and VPVD-WSe₂ based FETs at V _ds = −1 V. (h) Summary of μ _FE in PVD, [, − CVD, ,,,,,− and ME synthesized WSe₂ ,,,, with respect to I _on /I _off .

Conclusion

To conclude, we demonstrated that the origin of low mobility in WSe₂ can be traced to specific types of defects engendered in the crystal during growth, which can be further traced to the intrinsic impurities in the precursors used. Guided by theory, we developed a vdW crystal PVD method to minimize the defect density for centimeter-scale continuous monolayer WSe₂ film. Our protocol establishes a new benchmark for vapor deposition-synthesized WSe₂ FETs, and delivers comparable electrical performance to incumbent mechanical exfoliation techniques. In a broader context, this work is envisaged to extend to other 2D TMDs thin films and holds promise for the development of 2D complementary logic circuits, contingent upon further advancements in scalable production.

Methods

Preparation of Precursors

All chemicals and reagents, including the commercial WSe₂ powder (99.8% purity), were purchased from Alfa Aesar and used as received without further purification. The hydrothermal WSe₂ precursor was synthesized using the one-step solvothermal process. NaBH₄, Se, and NaWO₄ were first used as raw materials and mixed with 60 mL dimethylformamide (DMF). The solution was then mixed with 10 mL deionized water (Millipore, 18.2 MΩ cm), transferred into a Teflon liner loaded in a stainless-steel autoclave and kept at 200 °C for 48 h.

VdW crystals were synthesized by CVT using iodine as the transport agent. W (99.999%) and Se (99.999%) powders in a ratio of 1:2 were first mixed with iodine (99.999%), and sealed in an evacuated (10^–3 Torr) quartz ampule. The ampule was then placed into a tube furnace and heated to 950 °C at a rate of 1 °C min^–1 and held for 120 h. Given that large bulk crystals are difficult to volatilize, we avoided the long reaction times used in previous reports for large bulk crystal preparation. Instead, we chose a shorter reaction time (120 h) and utilized small-sized WSe₂ vdW crystals (Figure S 1d) as the precursor in our PVD process. Finally, the ampule was cooled down to room temperature. For the synthesis of Fe doped WSe₂ vdW crystals, Fe (99.999%) was used as dopant. For the synthesis of the 0.1, 0.5, and 2% Fe doped WSe₂ bulk crystals, Fe, W and Se powders were mixed in molar ratio (0.001:1:2, 0.005:1:2 and 0.02:1:2, respectively).

PVD Growth of WSe₂ Monolayers

WSe₂ monolayers and films were prepared by a reverse-flow PVD system. First, different types of precursors (2g) were loaded into the center of the quartz tube. A piece of sapphire (HOPG, SiO₂/Si) substrate was then placed in the downstream heating zone with a distance of 12 cm. The growth process was performed under ambient pressure. Prior to heating, the quartz tube was purged with 2000 sccm Ar for 2 min. Subsequently, the tube furnace was ramped up to 1150 °C within 50 min with an Ar reverse flow of 50 sccm (from the substrate to the precursor). The Ar flow was switched to forward (from precursor to substrate) immediately after the temperature reached 1150 °C and kept for 1 min to grow single-crystal flakes and 6 min to synthesize the continuous film. Finally, the tube furnace was cooled down naturally. Noted that all the growth variables except for the precursor quality were kept the same during the process.

CVD Growth of WSe₂ Monolayers

WSe₂ monolayers were grown by the conventional CVD method using WO₃ (Sigma-Aldrich, 99.90%) and Se (Sigma-Aldrich, 99.99%) powders as the precursors. The center heating zone was heated to 900 °C, and the Se at the upstream was maintained at 270 °C. A piece of sapphire (HOPG, SiO₂/Si) was placed at the downstream with an Ar/H₂ flow at 30 Torr. The growth continued for 15 min, followed by natural cooling to room temperature.

Material Characterization

The ESR spectra of precursors were recorded on a Bruker X-band A200 spectrometer (9.4 GHz, 77 K). The impurity elemental concentration analysis of precursors was conducted using GD-MS (Nu instruments, Astrum). The morphology and thickness of WSe₂ monolayers were determined by AFM (Bruker Dimension Icon system). Raman and PL spectra were acquired using the WITec confocal spectrometer with a laser excitation wavelength of 532 nm. XRD was performed using the Bruker D8 Advance diffractometer with a Cu–Kα radiation. Different types of precursors, each weighing 2g, were used for the XRD analysis. The morphology of the precursors was examined by SEM (FEI Quanta 600). The atomic structure and defects were characterized by the FEI Titan Themis Z STEM equipped with a double Cs (spherical aberration) corrector operating at 80 kV. XPS was operated at an (Omicron) UHV system (10–10 mbar) with an Al K^α X-ray source (1486.6 eV) operating at 15 kV. High-resolution spectra were collected at a constant energy of 15 eV. All the types of WSe₂ were measured under identical experimental conditions. Each spectrum was acquired over 9 h to improve the signal-to-noise ratio.

STM and dI/dV Measurements

The STM measurements of the CPVD-WSe₂, CVD-WSe₂ and VPVD-WSe₂ samples were performed under ultrahigh vacuum (UHV) conditions at a temperature of 77 K using a commercial Omicron LT-SPM instrument. The STM measurements of the HPVD-WSe₂ were performed at temperature of 10K using commercial closed-cycle Infinity Omicron SPM instrument. The STM measurements of the bulk WSe₂ were performed at room temperature using four-probe Omicron instrument. All STM images were collected in the constant-current mode using a chemically etched tungsten tip that was prepared by repeated indentation into a clean Au (111) surface and calibrated with respect to the Shockley state of the Au (111) surface. The bias voltage values refer to the sample with respect to the STM tip. Prior to STM measurements, the WSe₂ /HOPG samples were annealed at 500 K for 2 h under UHV conditions, while ME-WSe₂ was produced by in situ exfoliation of the bulk WSe₂ crystals in the UHV conditions. The point dI/dV were measured using a lock-in amplifier with a bias modulation voltage of 20 meV at a frequency of 815 Hz. The number of defects can be counted based on their topographic appearance. For each STM image, the defect density was calculated as the number of observed defects divided by the corresponding scanned area.

DFT Simulations

The DFT calculations were performed using the Vienna Ab initio Simulation Package (VASP) [ − with the projector augmented wave (PAW) , method and the Perdew–Burke–Ernzerhof (PBE) exchange-correlation functional. A k-point mesh of 2 × 2 × 1 was used for the Brillouin zone integration, and the cutoff energy for the plane-wave basis was set to 500 eV. To account for the long-range van der Waals (vdW) interactions, which are crucial for accurately describing structural properties of layered materials like WSe₂, we employed the DFT-D3 dispersion correction [ − method in our DFT calculations. The electronic self-consistent field iterations were converged with a strict energy tolerance of 1 × 10^–6 eV. For ionic relaxations, the convergence criterion was set to a maximum Hellmann–Feynman force of 0.02 eV/Å on each atom. A vacuum layer of 15 Å was introduced along the z-direction to minimize the interaction between periodic images of the doped WSe₂ monolayers. The atomic structure models of the doped WSe₂ monolayers were visualized using the VESTA software package, a tool for visualizing three-dimensional structural models based on crystallographic data. The preparation of input files for DFT calculations, as well as the analysis and visualization of the computed results, were performed using the VASPKIT program.

Synopsys QuantumATK T-2022.03 was utilized to calculate phonon-limited mobility, which could be divided into the following three steps. First, geometry optimization is performed with force tolerance of 0.02 eV Å^–1. Second, relaxed structures take 3 × 3 × 1 repetitions, and dynamical matrix of supercells is calculated. Third, the derivative of the Hamiltonian is calculated, the one and dynamical matrix are employed to calculate electron–phonon coupling matrix. Then, mobility could be extracted after electron–phonon coupling calculations. The configurations of linear Combination of Atomic Orbitals (LCAO) calculator include: Perdew–Burke–Ernzerhof generalized gradient approximation (PBE-GGA) exchange-correlation functional; Hartwigsen-Goedecker-Hutter (HGH) pseudopotentials with Tier4 basis sets; 240 Ry mesh cutoff; 8 × 8 × 1 k-point sampling; DFT-D3 dispersion correction; 1 × 10^–6 eV convergence tolerance of self-consistent iteration loop.

FET Fabrication and Electrical Performance Measurement

The WSe₂ films were transferred by the poly­(methyl methacrylate) (PMMA)-assisted wet transfer method. First, PMMA was spin-coated onto the monolayer WSe₂ films. After soaking in deionized water, the free-standing PMMA/WSe₂ stacks were delaminated from substrates and transferred to P²⁺ Si substrates with a 270 nm SiO₂ layer. For CVD-WSe₂ FETs, the monolayer CVD-WSe₂ film was transferred to P²⁺ Si substrates with an 8 nm atomic layer deposition (ALD) deposited HfO₂ layer. To dry the samples and enhance adhesion, the PMMA/WSe₂ stacks were postbaked on the hot plate at 120 °C for 20 min under vacuum. To remove PMMA residue, PMMA/WSe₂ stacks were soaked in 80 °C hot acetone for 30 min. Monolayer VPVD-WSe₂, CPVD-WSe₂, HPVD-WSe₂, and CVD-WSe₂ transistors were fabricated by electron beam lithography (EBL). O₂ (10 mTorr, 100 W) reactive ion gas plasma was applied to define the channel region and source/drain area pattern, followed by electron beam evaporation of 20 nm Pt/20 nm Au and sequent lift off. Electrical performance was measured by Keysight B1500A semiconductor parameter analyzer with a closed-cycle cryogenic probe station and all device measurements were performed under a vacuum environment (∼10^–6 Torr). The mobilities of the monolayer WSe₂ in Figure were calculated by the equation: μ = (dI _d /dV _gs) × [L_ch/(WC_ox V _ds )], where I _d is the drain current, V _gs is the gate voltage, L_ch is the channel length, W is the channel width, C_ox is the capacitance per unit area of the gate dielectric (12.7 nFcm^–2 for 270 nm SiO₂), and V _ds is the voltage between the source/drain.

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsnano.5c21076.

• DFT calculations of hole mobility and band structure for various defects, OM images, AFM images, STEM images, STM images, SEM images of the WSe₂, XRD of the precursors, Raman and PL of the different types of WSe₂, XPS of the different types of WSe₂, Electrical properties of the different types of WSe₂. (PDF)

+.

H.L., N.Y., J.M., and M.T. contributed equally to this work.

H. L., X.L. and L.-J.L. conceived the idea. X.L. and L.-J.L. supervised the project. H.L. and J.H.L. prepared the precursors and synthesized the WSe₂ monolayers. H.L. and Y.W. characterized the WSe₂ samples. N.Y., W.Y.L., L.-Y.H., C.D.Y. and T.C.G. fabricated the device and measured the electrical performance. J.C.M. and S.K. performed the DFT calculations. M.T., H.J., C.S.T., C.Q.Z., and K.P.L. performed STM. G.H., Z.Z.L., and S.H. performed XPS analyses. H.L., X.L., N.Y., S.L., and L.-J.L. wrote the paper. All authors discussed the results and commented on the manuscript.

The authors declare no competing financial interest.

Notice: This manuscript has been authored by UT-Battelle, LLC, under contract DE-AC05–00OR22725 with the US Department of Energy (DOE). The US government retains and the publisher, by accepting the article for publication, acknowledges that the US government retains a nonexclusive, paid-up, irrevocable, worldwide license to publish or reproduce the published form of this manuscript, or allow others to do so, for US government purposes. DOE will provide public access to these results of federally sponsored research in accordance with the DOE Public Access Plan (http://energy.gov/downloads/doe-public-access-plan).