Facile p‑Doping via Ambient Thermal Annealing Enhances Electrical and Photo‑Switching Performance of Ambipolar WSe₂ FETs

Abstract

We demonstrate that a simple thermal anneal in ambient air introduces oxygen‑based p‑doping in ambipolar WSe₂ field‑effect transistors (FETs). The process forms a thin WO₃ layer on the channel, shifting the transfer curve toward positive gate bias and increasing hole mobility from 0.13 to 1.3 cm²V⁻¹s⁻¹ while reducing electron mobility from 5.5 to 0.69 cm²V⁻¹s⁻¹. More importantly, the photo‑switching rise and decay times drop by over 600× and 170×, respectively, eliminating long‑lasting photoconductivity. X‑ray photoelectron, Raman, and photoluminescence spectroscopy confirm WO₃ formation and identify interface‑induced non‑radiative recombination sites that accelerate carrier recombination. These findings provide a scalable route to tailor the electrical and optoelectronic properties of 2D WSe₂ devices.

Background

Two‑dimensional (2D) materials, particularly transition‑metal dichalcogenides (TMDs) such as WSe₂, are celebrated for their intrinsic bandgap, high mobility, and ambipolar behavior, making them prime candidates for next‑generation transistors, sensors, and photodetectors. While graphene’s lack of a bandgap limits its versatility, WSe₂ offers both electron and hole conduction, excellent mechanical flexibility, and robust durability. However, achieving optimal performance requires controlled doping to adjust carrier type and contact resistance. Ambient thermal annealing has emerged as a straightforward method to oxidize WSe₂, forming a WO₃ surface layer that acts as a p‑dopant. Prior work has highlighted improved hole mobility (up to 83 cm²V⁻¹s⁻¹ on h‑BN substrates) but the impact on photonic behavior remains underexplored. This study systematically evaluates how ambient annealing alters both electrical transport and photo‑switching in ambipolar WSe₂ FETs, employing XPS, Raman, and PL to elucidate the underlying mechanisms.

Methods

Monolayer and bilayer WSe₂ flakes were mechanically exfoliated from bulk crystals and transferred onto 270‑nm SiO₂ on heavily doped p++ Si wafers, serving as back gates. Flake thicknesses were measured by AFM (NX 10). Electrode patterns were defined by electron‑beam lithography (JSM‑6510) on PMMA 495K resist, followed by Ti (30 nm) deposition via e‑beam evaporation (KVE‑2004L). Devices were annealed on a hot plate at 200 °C for 1 h in ambient air; control samples were annealed at 200 °C for 1 h under 4.5 × 10⁻⁴ Torr. XPS (AXIS SUPRA), Raman (XperRamn 200), and PL were performed with a 532 nm laser. Electrical measurements used a Janis probe station and Keithley 4200‑SCS; photo‑response tests employed a 405 nm laser (11 mW cm⁻²) at V_GS=0 V and V_DS=10 V.

Results and Discussion

Device Characterization
Figure 1a shows a representative WSe₂ flake and device. The AFM cross‑section confirms a bilayer thickness (~1.2 nm). Raman spectra exhibit characteristic peaks at 245 cm⁻¹ (E¹_2g/A_1g) and 308 cm⁻¹ (B¹_2g), indicating high crystal quality. The transfer curve (Figure 1e) displays classic ambipolar behavior with the n↔p crossover at V_GS≈−15 V.

Electrical Impact of Ambient Annealing
After annealing, the n↔p transition shifts to −5 V, and the I_DS at negative V_GS increases while it decreases at positive V_GS (Figure 2a). Mobility calculations (μ=dI_DS/dV_GS·L/(WC_iV_DS)) reveal a tenfold rise in hole mobility (0.13→1.3 cm²V⁻¹s⁻¹) and a sevenfold drop in electron mobility (5.5→0.69 cm²V⁻¹s⁻¹). Contour plots (Figure 2b) further confirm the positive‑bias shift. In contrast, vacuum annealing only improves both carrier types, underscoring the role of atmospheric oxygen in p‑doping.

Photo‑Switching Enhancement
Ambient annealing reduces the rise and decay times from 92.2 s and 57.6 s to <0.15 s and 0.33 s, a reduction of 610× and 170×, respectively (Figure 3a,b). Vacuum‑annealed devices show only modest changes, confirming that WO₃ formation drives the acceleration. The improved recombination is attributed to interface‑induced traps that facilitate non‑radiative pathways.

Electrical‑Photo Response vs. Gate Bias
Photo‑switching behavior varies strongly with V_GS (Figure 4). At V_GS>V_n↔p, annealing markedly shortens decay times while maintaining high photocurrent. Near V_n↔p, the response remains rapid regardless of annealing. For V_GS≪V_n↔p, both photocurrent and decay time increase, yet the long‑lived component shortens, reflecting an enhanced recombination landscape.

XPS, Raman, and PL Insights
XPS after 250 °C anneals shows rising W⁶⁺ peaks (35.5 eV, 37.8 eV) while Se peaks remain unchanged, confirming WO₃ formation (Figure 5a,b). Raman spectra after 500 °C annealing display new peaks at 712 cm⁻¹ and 806 cm⁻¹, matching WO₃ modes (Figure 6a,b). PL measurements of monolayer flakes reveal a modest bandgap increase from 1.60 eV to 1.61 eV and a pronounced intensity quench with longer annealing (Figure 6c,d), consistent with enhanced non‑radiative recombination at the WO₃/WSe₂ interface.

Conclusions

Ambient thermal annealing at 200 °C for 1 h effectively introduces a WO₃ surface layer, providing p‑doping that shifts the transfer curve, boosts hole mobility, and dramatically accelerates photo‑switching in ambipolar WSe₂ FETs. XPS, Raman, and PL analyses confirm the WO₃ formation and identify interface‑induced non‑radiative sites as the key to the improved performance. This facile, scalable approach offers a practical route to tailor 2D TMD devices for high‑speed electronic and optoelectronic applications.

Availability of Data and Materials

All data are fully available without restriction.

Abbreviations

2D

Two‑dimensional

AFM

Atomic force microscopy

FET

Field‑effect transistor

PL

Photoluminescence

TMDs

Transition metal dichalcogenides

XPS

X‑ray photoelectron spectroscopy