Asymmetrical contact scaling and measurements in MoS2 FETs

TL;DR

This study investigates contact scaling in MoS2 FETs using asymmetrical contact measurements and simulations, revealing the distinct behaviors of source and drain contacts and the importance of interface quality.

Contribution

It introduces a novel asymmetrical contact measurement method to analyze contact scaling behavior in 2D FETs, providing new insights into charge injection and transfer length.

Findings

Short contact lengths increase variability and reduce drain current.

Transfer length can be as short as 5 nm, depending on interface quality.

Charge injection differs between source and drain contacts.

Abstract

Two-dimensional (2D) materials have great potential for use in future electronics due to their atomically thin nature which withstands short channel effects and thus enables better scalability. Device scaling is the process of reducing all device dimensions to achieve higher device density in a certain chip area. For 2D materials-based transistors, both the channel and contact scalability must be investigated. The channel scalability of 2D materials has been thoroughly investigated, confirming their resilience to short-channel effects. However, systematic studies on contact scalability remain rare and the current understanding of contact scaling in 2D FET is inconsistent and oversimplified. Here we combine physically scaled contacts and asymmetrical contact measurements to investigate the contact scaling behavior in 2D field-effect transistors (FETs). The asymmetrical contact measurements directly compare electron injection with different contact lengths while using the exact same channel, eliminating channel-to-channel variations. Compared to devices with long contact lengths, devices with short contact lengths (scaled contacts) exhibit larger variation, smaller drain currents at high drain-source voltages, and a higher chance of showing early saturation and negative differential resistance. Quantum transport simulations show that the transfer length of Ni-MoS2 contacts can be as short as 5 nm. Our results suggest that charge injection at the source contact is different from injection at the drain side: scaled source contacts can limit the drain current, whereas scaled drain contacts cannot. Furthermore, we clearly identified that the transfer length depends on the quality of the metal-2D interface. The asymmetrical contact measurements proposed here will enable further understanding of contact scaling behavior at various interfaces.