一种集成热电模块的节能且重力不敏感环路热管用于恶劣环境中高功率移动电子芯片的冷却

Abstract Artificial Intelligence (AI) chips play a pivotal role in mobile intelligent devices including mobile robots, drones, smart electric vehicles, and others. With the rising computing power demands of these devices, thermal management systems for AI chips are facing increasingly crucial challenges, particularly in scenarios requiring reliable cooling under different orientations and high-temperature conditions. Although loop heat pipes (LHPs) overcome limitations of conventional heat pipes, LHPs have persistently failed to meet the thermal management requirements of high-power chips (greater than 500 W) operating in variable orientations. In this work, a thermoelectric cooler (TEC) was introduced to fully cool the LHP compensation chamber and greatly enhanced the condensed liquid return, which fundamentally solved the above problems. The performance of the novel LHP integrated with TEC module in various orientations and different environment temperatures was comprehensively studied through experiments. Additionally, a theoretical model of the coupled system was proposed. Experimental results demonstrate that the minimum total thermal resistance and maximum coefficient of performance (COP) were 0.0978 K/W and 26.8 in gravity-assisted orientation, 0.0997 K/W and 28.0 in horizontal orientation, 0.107 K/W and 21.5 in anti-gravity orientation, while maintaining junction temperatures below 85 °C.