基于混合优化算法的纳米薄膜参数表征

为了在椭圆偏振测量过程中得到精确的纳米薄膜参数,提出了一种求解纳米薄膜参数的混合优化算法。结合人工神经网络算法反向传播和粒子群算法快速寻优的特点,建立了改进粒子群-神经网络(Improved Particle Swarm Optimization-Neural Network,IPSO-NN)混合优化算法。该算法在较少的迭代次数下具有快速跳出局部最优解的能力,从而快速寻找椭偏方程最优解。文中使用该算法对标称值为(26.7±0.4)nm的硅上二氧化硅纳米薄膜厚度标准样片进行薄膜参数计算。结果表明:采用IPSO-NN混合优化算法计算薄膜厚度时相对误差小于2%,折射率误差小于0.1。同时,文中通过实验对比了传统粒子群算法与IPSO-NN算法,验证了IPSO-NN算法计算薄膜参数时能有效优化迭代次数和寻找最优解的过程,实现快速收敛,提高计算效率。

Characterization of nanofilm parameters based on hybrid optimization algorithm

In order to obtain accurate nano-film parameters in the ellipsometry measurement process, a hybrid optimization algorithm for nano-film parameter data processing was proposed. An Improved Particle Swarm Optimization-Neural Network(IPSO-NN) hybrid optimization algorithm has been proposed, based on the features of artificial neural network algorithm back propagation and particle swarm algorithm for fast optimization. This algorithm has the ability to jump out of the local optimal solution quickly with fewer iterations, so as to quickly find the optimal solution of ellipsometric equation. The algorithm was used to calculate the film parameters of silicon dioxide nano-film thickness standard template with a strandard value of 26.7±0.4 nm in this paper. The results show that the relative error of the film thickness calculation by IPSO-NN hybrid optimization algorithm is less than 2%, and the refractive index error is less than 0.1. At the same time, this paper compares the traditional particle swarm algorithm with the IPSO-NN algorithm through experiments, and verifies that the IPSO-NN algorithm can optimize the number of iterations effectively and the process of finding the optimal solution. This algorithm can achieve rapid convergence and improve the calculation efficiency when calculating the thin film parameters.