Nb Doping and Alloying of 2D WS2 by Atomic Layer Deposition for 2D Transition Metal Dichalcogenide Transistors and HER Electrocatalysts

We utilize plasma-enhanced atomic layer deposition to synthesize two-dimensional Nb-doped WS2 and NbxW1–xSy alloys to expand the range of properties and improve the performance of 2D transition metal dichalcogenides for electronics and catalysis. Using a supercycle deposition process, films are prepared with compositions spanning the range from WS2 to NbS3. While the W-rich films form crystalline disulfides, the Nb-rich films form amorphous trisulfides. Through tuning the composition of the films, the electrical resistivity is reduced by 4 orders of magnitude compared to pure ALD-grown WS2. To produce Nb-doped WS2 films, we developed a separate ABC-type supercycle process in which a W precursor pulse precedes the Nb precursor pulse, thereby reducing the minimum Nb content of the film by a factor of 3 while maintaining a uniform distribution of the Nb dopant. Initial results are presented on the electrical and electrocatalytic performances of the films. Promisingly, the NbxW1–xSy films of 10 nm thickness and composition x ≈ 0.08 are p-type semiconductors and have a low contact resistivity of (8 ± 1) × 102 Ω cm to Pd/Au contacts, demonstrating their potential use in contact engineering of 2D TMD transistors.

XPS peak fitting was performed using Thermo Fisher Scientific Avantage software.
For the peak shape, a 30% Lorentzian/Gaussian mix was used.Background subtraction was done using the Avantage "smart background" algorithm which constrains the background to not be higher than the data.For fitting of doublets, the area ratio of the two peaks was constrained to the appropriate value for the orbital quantum number.Figure S1 shows the peak fitting results for the samples in the composition series, and the peak parameters are reported in Table S1.The data is shown as black dots, the fitted peaks in blue, the fitted peak envelope in red and the residual in yellow.The peak parameters are given in Table S1.II.The elemental abundance of Nb, W, S and H was derived from the spectra above by fitting, and the results are tabulated below.Units are thin film units (TFU) = 10 15 atoms per cm 2 .Statistical errors are 5% for H, 0.3 TFU for S, 0.1 TFU for Nb, and 1% for W except for sample K27 where it is 20%.On top of the statistical errors, the systematic errors are 7% for ERD (same deviation for H in all samples) and 2% for RBS (same deviation for all elements except H in all samples).The elemental abundance of Nb, W and S was derived from the spectra above by simulation.The hydrogen amounts are obtained by comparison of the peak contents with that of a reference sample of LPCVD-grown silicon nitride.

Raman measurements on low Nb content alloys
In the Raman spectrum of Nb 0.08 W 0.92 S 2 , no new peaks are observed compared to the spectrum of the pure WS 2 sample.The new peak observed in the spectrum of the sample 1Nb/4W around 391 cm -1 (see main text Figure 4b) is not observed either at a low Nb content of 8%.The main difference between the WS 2 and Nb 0.08 W 0.92 S 2 spectra is a reduced intensity of the resonance Raman peaks, most notably the 2LA(M) peak at 350 cm -1 .In contrast, the (nonresonant) A 1g peak at 410 cm -1 is of similar intensity in both spectra.V was used to extract the carrier mobility in each case, and values of 0.003 cm 2 /Vs and 0.66 cm 2 /Vs were determined for the WS 2 and Nb 0.08 W 0.92 S 2 samples, respectively.

Tafel plots and methodology
Tafel slopes were estimated by plotting the numerical derivative of the Tafel plot as a function of the logarithmic current density and identifying the linear part of the Tafel slope by finding the current density where the numerical derivative is (approximately) constant.For the NbS 3 sample, no linear region was found such that a Tafel slope could not be extracted.

Figure S1 :
Figure S1: Peak fitting of the XPS elemental scans of the composition series samples.

Figure S2 :
Figure S2: Comparison between XPS results before and after correction of the W quantification as discussed in the main text.

Figure S3 :
Figure S3: XPS scans for 3 samples prepared using a 4W/1ABC/4W process where the Nb dosetime during the ABC cycle was set to 10, 15 and 20 seconds.This data is represented in numerical form in TableII.

Figure S5 :Figure S6 :Figure S7 :
Figure S5: Raman spectra of pure WS 2 and Nb 0.08 W 0.92 S 2 grown by ALD.For the preparation of the alloy sample, a 4W/1ABC/4W supercycle was used.

Figure S8 :
Figure S8: First 5 cyclic voltammetry sweeps of the samples presented in the main text.The fifth sweep was used to determine the Tafel slope and η 10 overpotential.

Figure S9 :
Figure S9: Tafel plots and extraction of Tafel slope values from the linear part of the Tafel plot.

Figure S11 :
Figure S11: AFM rms roughness values of Nb x W 1-x S y films on SiO2 and glassy carbon substrates show similar trends in surface roughness with alloy composition on both substrates.

Table S1 :
Peak parameters of the XPS peak fits in Figure S1.