Exploring optimal TMDC multi-channel GAA-FET architectures at sub-1nm nodes

Abstract

This paper explores the design and optimization of multi-Nanosheet Field-Effect Transistors (mNS-FETs) employing a Transition Metal Dichalcogenide (TMDC) channel, specifically MoS₂, for the 0.7 nm technology node using calibrated Technology Computer-Aided Design (TCAD) simulations. A comprehensive analysis is conducted at both the device and circuit levels, considering various structural parameters such as the number of MoS₂ layers, vertical and lateral nanosheet stacking configurations, and nanosheet widths. To enable more effective structural optimization, the resistance and capacitance components of the device are carefully segmented, providing a detailed framework for design refinements. The results indicate that a trilayer configuration outperforms its monolayer counterpart by reducing external resistance through an increased surface area, making it the preferred option at a 12 nm gate length. This observation also elucidates the advantage of single lateral stacking over double lateral stacking. While vertical stacking increases the effective width for on-current enhancement, excessive stacking compromises switching speed at the same power level, identifying four vertical stack structures as the optimal configuration. Among the evaluated configurations, the trilayer MoS₂ mNS-FET with four vertical stacks, single lateral stacking, and a 17 nm nanosheet width was identified as the optimal structure for the 0.7 nm node. Furthermore, at the circuit level, the effective width is evaluated to ensure compliance with the circuit area constraints of the target technology node. Analyzing the impact of parasitic resistance and capacitance in the Middle-of-Line (MOL) and Back-End-of-Line (BEOL) reveals that time delay can lead to up to a 58 % degradation in inverter circuit performance. By systematically investigating the impact of MoS₂-based mNS-FET structures, this study provides critical insights to guide the future design of TMDC-based mNS-FETs.