基于数字微流控的芯片冷却模型与路径优化

The Chip Cooling Model and Route Optimization with Digital Microfluidics

使用微流控技术实现集成芯片冷却被认为是扩展摩尔定律生命力的一个方向。高密度热点的散热是制约芯片三维集成的瓶颈。热点通常具有不稳定的空间和时间分布性,是影响电子系统工作性能和导致故障的主要原因,如何针对热点准确冷却成为芯片热管理的关键。基于介电润湿原理,采用数字滴液微流控技术,提出芯片热点自适应冷却方法。构建双平面热点冷却器件的三维模型,分析了冷却热点的工作原理及特性;构建了单平面热点冷却器件的电容模型,并从电学角度深入分析了它的工作原理;设计了一种具有互锁齿结构的电极,从而保证了液滴的连续运动,针对热点动态分布提出滴液路径优化冷却,最后采用蚁群算法验证2个液滴遍历所有可能热点的最短路径。

Using microfluidic technology to achieve integrated chip cooling is becoming a promising method to extend Moore law effective period. The thermal management is always critical for 3D integrated circuit design. Hot spots due to spatially non-uniform heat flux in integrated circuits can cause physical stress that further reduces reliability. The critical point for chip cooling is to use microfluidic cooling accurately on the hot spots. First, based on electro-wetting on dielectric, the paper presents an adaptive chip cooling technique using the digital microfluidics. Then, a two-plans 3D chip cooling model has been given with its working principle and characteristics. And single plan chip cooling model is presented, including its capacitance performance and models. Moreover, the dentate electrode is designed to achieve droplet continuing movement. Next, the ant colony optimization is adopted to get optimal route during electrode moving. Last, the experiments demonstrate the adaptive chip cooling technique proposed in this paper is effective and efficiency.