薄膜椭偏测量的自适应混合反演算法

An Adaptive Hybrid Inversion Algorithm for Ellipsometric Measurement of Thin Films

基于梯度的线性反演方法计算效率高,基于随机扰动的模拟退火方法寻找最优解能力强针对薄膜椭偏测量的多极值问题,综合两者的优点,提出一种求解薄膜椭偏测量问题的混合反演算法.模型每次扰动采用线性寻优方法搜寻局部最优解,叠代过程中采用均匀设计的模拟退火方法随机搜寻模型,使该算法有跳出局部最优解的能力,可以在较少的叠代次数内搜寻到全局最优解,从而提高求解薄膜椭偏测量非线性反演方法的计算效率.对反演过程控制参数进行讨论,该算法具有自适应的特点.计算表明,该算法可有效求解薄膜椭偏测量的多极值问题.

A hybrid optimization algorithm is developed for solving ellipsometric measurement problem.This algorithm is based on the high ability of simulated annealing method in seeking optimum solution and the high computation efficiency of simple shape method.It seeks the global minimum solution of objective function and gets proper parameters of optical model.The computation efficiency is higher than that of simulated annealing algorithm.The selection of inversion controlling parameters is discussed to make this algorithm adaptive.