Bridging the gap between atomically thin semiconductors and metal leads

Atomically thin transition metal dichalcogenide semiconductors (TMDSCs), similar in geometry to graphene, possess unique properties expanding the two-dimensional (2D) materials landscape. However, unlike graphene, the electrical performance of TMDSC field-effect transistors (FETs), particularly carrier mobility at cryogenic temperatures, is limited, hindering the exploration of quantum transport properties and practical applications. A major obstacle is the inherent difficulty in electrically interfacing atomically thin TMDSCs with metal leads due to undesired metal-semiconductor interface barriers, including tunnel and Schottky barriers. These barriers arise from Fermi-level pinning effects and Schottky barriers caused by defect-induced gap states at the interface, dramatically suppressing carrier-injection efficiency and increasing contact resistance. Existing methods to minimize these barriers, such as doping treatments, microscale phase transformations, and van der Waals (vdW) contacts with inserted tunnel barrier layers or graphene/soft-landed metals, have limitations. VdW contacts, while offering a certain level of improvement, still suffer from reduced charge carrier injection due to the large vdW gap acting as a tunnel barrier. Efficient carrier injection requires effective orbital overlap or hybridization between transition metal and electrode atoms; however, edge contacts, while offering this advantage, often suffer from Fermi-level pinning and Schottky barriers at low temperatures. The study of quantum transport behaviors in TMDSCs at low temperatures is crucial because these behaviors are perturbed at room temperatures. Therefore, improving low-temperature contact performance is essential for advancing the field.

The past decade has seen significant efforts to improve TMDSC contacts, falling into two categories: direct metallization via doping or phase transformations, and van der Waals (vdW) contacts. Direct metallization methods, while effective in some cases, often introduce other challenges or limitations. VdW contacts utilize materials like hexagonal boron nitride (h-BN) or graphene as interface layers, aiming to reduce direct contact between the metal and TMDSC. Though they reduce the Schottky barrier, the vdW gap itself can still impede efficient charge transport. Graphene leads have shown promise in eliminating Schottky barriers in few-layer MoS2 FETs, but the vdW gap still limits charge carrier injection. Edge contacts offer the benefit of orbital hybridization, but typically suffer from Fermi level pinning and Schottky barriers at low temperatures. This limitation greatly affects the study of quantum transport phenomena, which are heavily influenced by temperature and phonon scattering.

To address these challenges, the researchers developed a local bonding distortion (LBD) strategy for creating highly efficient electrical junctions with atomically thin TMDSCs. This strategy involves using soft oxygen plasma treatment to induce a nanoscale trigonal-prismatic-to-octahedral coordination change in the metal-TMDSC interface, creating a semi-metallic bridge that facilitates nearly barrier-free electrical contacts. The process was performed using reactive ion etching (RIE) in two geometries: edge contact and top contact. The edge-contact geometry involved precise etching to expose the TMDSC edges, followed by the oxygen plasma treatment. Top-contact geometry involved removing the top BN layer to expose the TMDSC surface before the treatment. Atomic-resolution cross-section electron microscopy and Raman spectroscopy were employed to confirm the LBD, showing a nanoscale structural change in the metal-TMDSC interface. The electrical performance of FETs fabricated using the LBD contacts was characterized at both room and cryogenic temperatures. Two-probe and four-probe measurements were used to extract contact resistance, and temperature-dependent measurements determined field-effect mobility. Hall effect and Shubnikov-de Haas (SdH) oscillation measurements were also conducted to determine Hall mobility and quantum mobility. The researchers used 3L-MoS2 and 5L-WSe2 as model systems to study both edge and top contact geometries and employed techniques like focused ion beam (FIB) for sample preparation, aberration-corrected scanning transmission electron microscopy (ACSTEM) for atomic-resolution imaging, electron energy-loss spectroscopy (EELS) for elemental analysis, and density functional theory (DFT) calculations for electronic structure analysis. The precise RIE conditions including RF power, frequency, gas composition, pressure, substrate bias and duration are also specified in the paper.

The LBD strategy resulted in significant performance improvements in TMDSC FETs. In 3L-MoS2 FETs with LBD edge contacts, the contact resistance was reduced to around 90 Ωμm at 2 K, approaching the quantum limit. The field-effect mobility reached 23,700 cm²V⁻¹s⁻¹ at 2 K, demonstrating robust ohmic behavior at both room and cryogenic temperatures. The 1L-MoS2 devices exhibited high-quality quantum Hall detection, high field-effect mobility, and Hall mobilities at cryogenic temperatures, highlighting the effectiveness of the LBD method even for monolayer TMDSCs. The 5L-WSe2 FETs, using the top-contact geometry, displayed a contact resistance of 700 Ωμm and a high field-effect mobility of 358,000 cm²V⁻¹s⁻¹ at 0.3 K, along with prominent SdH oscillations, indicative of high quantum mobility. Atomic-resolution imaging confirmed the formation of a semi-metallic LBD region (around 1nm wide) at the metal-TMDSC interface, showing direct bonding between transition metal and electrode atoms with a distance much shorter than the typical vdW gap. This indicated strong orbital overlap and hybridization, explaining the observed improved carrier injection. EELS analysis confirmed the presence of oxygen in the distorted region consistent with the proposed oxygen-substitution-induced bond rearrangement mechanism. The LBD structure, identified as a mixture of 1T and 1T' octahedral derivatives in WSe2, was structurally seamless, without dangling bonds or vacancies. The method's generality was demonstrated through successful application to both MoS2 and WSe2, and its robustness was shown through consistent device performance across different samples and long-term storage. Transmission line experiments corroborated the low contact resistance measurements. The polarity of LBD-interfaced FETs was shown to be contact-metal dependent, showcasing the flexibility of the method.

The success of the LBD contact strategy stems from the octahedral distortion of TMDSCs, which creates semi-metallic behavior with dispersed energy states around its work function. Unlike a standard H-T interface with a mismatch in work functions, the LBD interface allows for facile coupling with varying work functions of different metal leads, enhancing carrier injection efficiency. The observed contact resistance dependence on carrier density and temperature is attributed to the screening of impurity scattering by charge carriers and the reduced phonon scattering at lower temperatures. The high-quality electrical contacts achieved through LBD facilitate the investigation of unconventional quantum transport properties in TMDSCs previously constrained by poor contact quality at low temperatures. The method’s compatibility with standard cleanroom processes makes it promising for scalable integration in vdW devices.

This study presents a novel LBD contact strategy for creating high-quality electrical junctions between metal leads and TMDSCs, achieving low contact resistance, high mobility, and excellent cryogenic performance. The atomic-scale understanding of the LBD mechanism, along with its demonstrated versatility and scalability, makes it a significant advancement for future TMDSC-based device development and the exploration of quantum phenomena in 2D materials. Future research could explore the optimization of plasma treatment parameters and the integration of LBD contacts into complex vdW heterostructures for enhanced functionalities.

While the LBD method shows significant improvements, there is a certain level of device-to-device variation, particularly in WSe2, possibly due to the heterogeneous nature of the LBD. The study primarily focuses on specific TMDSCs (MoS2 and WSe2) and the generalizability to other 2D materials may need further investigation. The contact resistance, though significantly reduced, is still not entirely eliminated, potentially hinting at further optimization possibilities.