Modulating self-heating effects in FinFETs through doping engineering

Abstract

While FinFETs are widely adopted in advanced nodes for high-performance integration, their elevated power density induces pronounced self-heating effects that pose significant challenges to device reliability and performance, underscoring the importance of effective strategies for mitigating self-heating effects. By combining electro-thermal simulations based on the drift-diffusion model with phonon Monte Carlo (MC) simulations, this work elucidates the heat generation mechanisms in FinFETs and accordingly proposes a doping-engineering approach to mitigate self-heating effects. The results indicate a shift in the dominant heat generation mechanism with operating bias. While Joule heat is the primary heat generation mechanism under the saturation bias, the contribution from Thomson heat becomes significant under the typical CMOS operating bias, emerging as the dominant driver of heat generation non-uniformity due to its highly localized distribution in devices with reduced feature sizes. As a result, Thomson heat provides a pronounced contribution, approximately 16%, to the maximum device temperature rise. To address this issue, it has been found that reducing the doping concentration gradient in the device extension region can significantly reduce Thomson heat without decreasing the electrical performance, whereby the peak Thomson heat is reduced by 66% and the hotspot temperature rise is decreased by 13%. These findings provide insights into the self-heating behavior of FinFETs and suggest a potential approach for device-level thermal management.