量子进化算法在极紫外多层膜表征和设计中的应用

摘要

According to the shortcomings in the characterization and design of EUV multilayer with the genetic algo-rithm (GA), which are large population, slow convergence and poor accuracy, the application of real-code quantum evolutionary algorithm (QEA) in the characterization and design of EUV is demonstrated. The fitting of grazing inci-dence X-ray reflection (GIXR) of EUV multilayer and design of broadband EUV multilayer have been carried out based on the real-code QEA, and the results are compared with that obtained based on real-code GA. The results demonstrate that the real-code QEA which has a smaller population can supply the parameters of multilayer with better convergence and accuracy in the fitting of GIXR,and the deviation of thickness of multilayer can be lower than 0.1nm. Comparing with GA,the real-code QEA which use a smaller population can also supply a good coating design with a similar convergence and accuracy, and the reflection band of designed multilayer is 13~15nm, which has a reflectivity of 25%. As a result, the related researches demonstrated the potential value of practical application of real-code QEA in the characterization and design of EUV multilayer, and the foundation has been laid for the researches in the next-step.%针对EUV多层膜的表征和设计过程中普遍采用的遗传算法（GA）存在的大种群和求解效率低的问题，本文将实数编码量子进化算法（QEA）应用于EUV多层膜的表征和宽光谱EUV多层膜的设计过程中。分别采用实数编码GA和QEA对理论仿真的EUV多层膜的X射线掠入射反射谱进行反演拟合和宽光谱EUV多层膜的反射谱进行设计，进而将QEA在EUV多层膜表征和设计方面的性能进行对比分析。分析结果表明，QEA在多层膜的表征方面具有小种群和反演求解效率高的优点，多层膜的膜厚拟合精度可达±0.1nm；在多层膜设计方面，量子进化算法同样具有小种群的优势，同时求解效率较GA算法接近，设计的多层膜反射率带宽为13~15nm，反射率可达25%。相关工作展现出QEA算法在多层膜研发方面的应用价值，为进一步工作打下基础。