声子
NMOS逻辑
材料科学
声子散射
绝缘体上的硅
晶体管
凝聚态物理
硅
散射
半导体
光电子学
电子
光学
物理
电压
量子力学
作者
Jae Sik Jin,Joon Sik Lee
出处
期刊:Journal of heat transfer
[ASME International]
日期:2009-06-22
卷期号:131 (9)
被引量:9
摘要
First, the electron-phonon interaction model, which has recently been developed by authors for thermal predictions within the silicon devices in micro/nanoscales, is verified through the comparison with the experimental measurement of average temperature rise in the channel region of a silicon-on-insulator (SOI) transistor. The effect of the silicon layer thickness of the SOI transistor on phonon thermal characteristics is also investigated. It is found that the thickness effect on the peak temperature of the optical phonon mode in the hot spot region is negligible due to its very low group velocity. Thus the acoustic phonons in a specific frequency band, which has the highest scattering rate with the optical phonons, experience relatively less reduction in the peak temperature as the silicon layer thickness increases. Second, the electron-phonon interaction model is applied to the transient thermal transport simulation during the electrostatic discharge (ESD) event in an n-type metal-oxide-semiconductor (NMOS) transistor. The evolution of the peak temperature in the hot spot region during the ESD event is simulated and compared with that obtained by the previous full phonon dispersion model, which treats the electron-phonon scattering as a volumetric heat source. The results show that the lower group velocity acoustic phonon modes (i.e., higher frequency) and optical mode of negligible group velocity acquire high energy density from electrons during the ESD event, which might cause the devices melting problem. The heat transfer rates by individual phonon modes are also examined, and it is found that the key parameter to determine the phonon heat transfer rate during the ESD event is the product of the phonon specific heat and the scattering rates with higher energy density phonons in the hot spot region.
科研通智能强力驱动
Strongly Powered by AbleSci AI