Integrating First-Principles-Based Non-Fourier Thermal Analysis Into Nanoscale Device Simulation

Abstract

Thermal analysis is an essential component of semiconductor device simulation for device design and thermal management. The prevalent approach of device thermal analysis uses Fourier-law-based heat diffusion equation (HDE). However, Fourier’s law is known to fail when the characteristic length is smaller than the phonon mean free path (MFP), resulting in a significant underestimation of local temperature rise. In this study, we implement non-Fourier thermal analysis on nanoscale devices using the first-principles-based nongray Boltzmann transport equation (BTE). A 3-D structure of nanoscale silicon-based FinFET is adopted as a case study. Non-Fourier effects are considered in both thermal generation and transport processes. In the thermal generation process, we use first-principles methods to investigate the selective electron–phonon energy transfer process and obtain the mode-level phonon generation rates. In the thermal transport process, we solve the nongray phonon BTE to determine the temperature distribution of the devices, in which the material-dependent phonon properties are calculated by first-principles methods. Through comparisons with HDE and previous models, we demonstrate the considerable impact of non-Fourier effects on temperature rise and electrical performance, highlighting the significance of incorporating non-Fourier thermal analysis into nanoscale device simulations. Our method also shows good agreement with experimental temperature measurements, which can be readily extended to a variety of devices and operating conditions.