【摘要】 器件稳健性设计本质上是一个多个目标的优化问题 .将实验设计和响应表面方法相结合可用来满足减少所需的 TCAD模拟次数的强烈需求 .然而对一些非线性问题 ,RSM模型的误差可能会大到影响设计结果的有效性 .为找到可行空间 ,提出了一种考虑 RSM模型误差时 ,面向目标的多代设计方法 .通过在由当前代中的 RSM模型结果所得到的有希望空间中增加设计 ,使得模型误差减小并得到可行空间 .对 FIBMOS例子的结果显示本方法是有效的 .更多还原

【Abstract】 Device robust-design is inherently a multiple-objective optimization problem.Using design of experiments (DoE) combined with response surface methodology (RSM) can satisfy the great incentive to reduce the number of technology CAD(TCAD) simulations that need to be performed.However,the errors of RSM models might be large enough to diminish the validity of the results for some nonlinear problems.To find the feasible design space,a new method with objectives-oriented design in generations that takes the errors of RSM model into account is presented.After the augment design of experiments in promising space according to the results of RSM model in current generation,the feasible space will be emerging as the model errors deceasing.The results on FIBMOS examples show that the methodology is efficient.更多还原