光学薄膜椭圆偏振数据反演的优化算法  

Optimization algorithm for ellipsometry data inversion of optical film

作　　者：童晟飞[1] 王正忆[1] 陈星[1] 

机构地区：[1]浙江大学理学部物理系,浙江杭州310027

出　　处：《物理实验》2012年第9期22-28,共7页Physics Experimentation

摘　　要：运用C++语言通过变步长循环的优化算法开发数据处理程序,利用2组数据确定了多周期真实厚度.通过计算SiO2和ZrO薄膜的折射率和单周期及多周期厚度,并与其他方法的计算结果进行对比分析,发现此方法计算结果可以达到较高的精度.对于有吸收的薄膜,利用量子力学原理计算粒子出现在某一点的概率,通过蒙特卡罗模拟的方式来测量粒子的位置,从而对粒子进行刷新.通过基于量子行为粒子群优化和模拟退火的混合算法设计的程序进行计算,可很快跳出局部极小收敛达到全局最优点,在保证精度的同时达到较快的收敛速度.An iterative C＋＋ program based on optimization algorithm with variable-step cycles was introduced, which could use only two sets of data to determine the true thickness of more than one period. The refractive index and thickness of one or more periods of the SiOz and ZrO thin film were calculated. The results were more accurate than that of other solutions. For absorbing film, the probability of a particle appearing at some point was calculated according to the principles of quantum mechanics, and the particle position was determined by means of Monte Carlo simulation. Thus the particle was refreshed. The procedure based on quantum-behaved particle swarm optimization and simulated annealing hybrid algorithm was designed, which could quickly jump out of local minimum to ensure the accuracy and a faster convergence speed.

关 键 词：光学薄膜 变步长循环 粒子群算法 模拟退火算法 

分 类 号：O484.41[理学—固体物理]