基于热叠加模型的叠层3D多芯片组件芯片热布局优化研究  

Study on Thermal Placement Optimization of Stacked 3D-MCM Based on Thermal Superposition Model

作　　者：梁颖[1] 黄春跃[2] 阎德劲[3] 李天明 

机构地区：[1]成都航空职业技术学院,四川成都610021 [2]桂林电子科技大学机电工程学院,广西桂林541004 [3]中国西南电子技术研究所,四川成都610036 [4]桂林航天高等专科学校科研与设备处,广西桂林541004

出　　处：《电子学报》2009年第11期2520-2524,共5页Acta Electronica Sinica

基　　金：广西自然科学基金(No.0832083);广西制造系统与先进制造技术重点实验室主任基金(No.07109008-011-Z)

摘　　要：叠层三维多芯片组件(3D Multi-Chip Module,MCM)芯片的位置布局直接影响其内部温度场分布,进而影响其可靠性.本文研究了叠层3D-MCM内芯片热布局优化问题,目标是降低芯片最高温度、平均芯片温度场.基于热叠加模型并结合热传导公式,选取芯片的温度作为评价指标,确定出用于3D-MCM热布局优化的适应度函数,采用遗传算法对芯片热布局进行优化,得出了最优芯片热布局方案,总结出了可用于指导叠层3D-MCM芯片热布局设计的热布局规则;采用有限元仿真方法,对所得的热布局优化结果进行验证,结果表明热布局优化结果与仿真实验结果一致,本文所提出的基于热叠加模型的MCM热布局优化算法可实现叠层3D-MCM芯片的热布局优化.In the thermal design of stacked three-dimensional multi-chip module （3D-MCM）,the placement of multiple chips on 3D-MCM substrate has a significant effect on temperature field in the 3D-MCM,thus influences the reliability of the 3D-MCM. The thermal placement optimization of chips in stacked 3D-MCM was studied in this paper, the goal of this work is to decrease temperature and achieve uniform thermal field distribution within stacked 3D-MCM. The average temperature of chips in stacked 3D-MCM was chosen as the evaluation target based on thermal superposition model and equation of heat conduction, and the fitness function for thermal placement optimization was determined. Based on genetic algorithms, chips placement optimization algo- rithm of thermal model was proposed and the optimization chips placement of stacked 3D-MCM was achieved by corresponding optimization program. Based on this work, the chips placement rules which can be used for directing thermal design of stacked 3D- MCM were summarized. To demonstrate the effectiveness of the obtained optimization chips placement, by using ANSYS, finite ele- ment analysis （FEA） was carried out to assess the thermal field distribution of the optimization chips placement in stacked 3D- MCM, the result shows that the thermal field distributions of the optimization chips placement are in good accord with the FEA results. It turns out that the chip placement optimization approach proposed in this work can be effective and robust in providing thermal optimal design of chip placement in stacked 3D-MCM.

关 键 词：叠层三维多芯片组件 遗传算法 热布局优化 热叠加模型 有限元分析 

分 类 号：O224[理学—运筹学与控制论] TN305.94[理学—数学]