基于LS-SVM的光刻过程R2R预测控制方法

针对光刻过程非线性、时变和产品质量不易在线测量的特性,提出了一种基于最小二乘支持向量机预测模型和微粒群滚动优化方法的批次控制预测控制器。通过历史批次样本数据构建光刻过程的最小二乘支持向量机预测模型,解决了复杂光刻过程难以建立精确数学模型的难题,提高了预测模型的精度。通过预测误差的反馈校正和微粒群滚动优化算法求解最优控制律,提高了控制精度。性能分析结果表明,与指数加权移动平均方法及非线性模型预测控制方法相比较,批次控制预测控制器控制器减小了不同批次关键尺寸输出的差异,显著降低了关键尺寸输出的均方根误差,有效抑制了过程扰动影响。

LS-SVM Based R2R Predictive Control for Lithography Process

For the lithography process characteristics of nonlinear,time-varying and not being in-situ measured easily,a run-to-run(R2R) predictive controller named support vector machine predictive(SVMP) R2R of lithography process was proposed based on least squares support vector machine(LS-SVM) predictive model and particle swarm optimization(PSO) receding optimization algorithm.The LS-SVM predictive model constructed by sample data of historical batches solved the difficult problem of constructing accurate mathematical model of the lithography process and improved the prediction accuracy.The optimal control law achieved from feedback correction of prediction errors and PSO receding optimization algorithm improved the control precision.The performance analysis results illustrate that the critical dimension(CD) variation in various runs of products is reduced,and the root mean squared error for CD is brought down substantially,the disturbance is rejected effectively by SVMP R2R controller compared with methods of exponentially weighted moving average(EWMA) and nonlinear model predictive control(NMPC).