材料科学
微通道
高电子迁移率晶体管
光电子学
宽禁带半导体
氮化镓
晶体管
计算流体力学
电力电子
电压
电气工程
机械
纳米技术
图层(电子)
物理
工程类
作者
Man Prakash Gupta,Ajit K. Vallabhaneni,Satish Kumar
出处
期刊:IEEE Transactions on Components, Packaging and Manufacturing Technology
[Institute of Electrical and Electronics Engineers]
日期:2017-05-15
卷期号:7 (8): 1305-1312
被引量:16
标识
DOI:10.1109/tcpmt.2017.2693399
摘要
AlGaN/GaN high electronic mobility transistors (HEMTs) are highly promising for power and RF electronics due to large bandgap and high breakdown voltage. However, the thermal bottlenecks associated with the elevated temperatures in the device at high power density presents significant barriers in achieving the full potential of these HEMTs. Embedded cooling solutions have been explored for these devices for the efficient thermal management of extreme heat fluxes and hotspots. This involves conductive, convective, and evaporative cooling methods located closer to the electrically active regions of the device. Developing simulation tools that can guide thermal-mechanical-electrical co-design are necessary for implementation of the embedded cooling solutions. We present self-consistent electrothermal models for AlGaN/GaN HEMTs to study the passive and embedded microchannel cooling. In the case of passive cooling, we develop a fully compact electrothermal model that is computationally very efficient and provides high-fidelity results. The model can capture the spatial variation of temperature profile in the device and can predict the temperature-dependent drain current variation among the various fingers in a multifinger HEMT. We then explore the potential of liquid cooling of a multifinger AlGaN/GaN HEMTs using microchannels embedded in the substrate. A 3-D computational fluid dynamics (CFD) model of the device is coupled with the compact dc current model. Our analysis shows that the self-consistency effects on power and temperature remain significant even for the high flow rates in microchannels. Predictive capability of the "standalone" CFD models of AlGaN/GaN HEMTs can be limited unless they are self-consistently coupled with the electrical models.
科研通智能强力驱动
Strongly Powered by AbleSci AI