光刻机超精密运动台数据驱动MIMO定结构前馈控制

Data-driven MIMO Fixed-structure Feedforward Control for Ultra-precision Motion Stages of Lithographic Scanners

超精密多输入多输出(Multiple-input multiple-output,MIMO)运动台是集成电路制造装备——光刻机的核心部件之一,其轨迹跟踪性能是保证光刻机产率和分辨率的关键。提出一种高效的数据驱动MIMO定结构前馈控制方法,有效补偿参考轨迹引入的伺服误差,降低各自由度间的耦合,从而提高光刻机超精密运动台对不同轨迹跟踪任务的适应性及跟踪性能。采用多项式对MIMO前馈控制器进行参数化,以保证前馈控制器的固有稳定性及其对变轨迹的鲁棒性。利用脉冲响应实验法无偏估计过程灵敏度函数矩阵的马尔可夫参数,进而建立高效的数据驱动参数优化算法,根据伺服系统的输入-输出数据迭代优化MIMO前馈控制器参数。所提方法无需对光刻机运动台建模,且每次迭代仅需一次轨迹跟踪实验。将所提方法应用于课题组自主研发的光刻机硅片台,2种参考轨迹的跟踪实验结果验证所提方法在有效提高轨迹跟踪性能的同时降低各自由度间的耦合。

The ultra-precision multiple-input multiple-output(MIMO)motion stage is an important mechatronic unit of industrial lithographic scanners for manufacturing integrated circuits,and its excellent tracking performance is the key to ensure the throughput and resolution.An efficient data-driven fixed-structure MIMO feedforward control approach is synthesized for the ultra-precision motion stage to reduce the coupling among multiple degrees of freedom and enhance tracking performance in view of varying references.Specifically,the MIMO feedforward controller is parameterized with the polynomial basis functions.As a result,the inherent stability and robustness against varying references are guaranteed for the MIMO feedforward controller.The Markov parameters of the process sensitivity function matrix are unbiasedly estimated by using impulse response experiment.Then,an efficient data-driven parameter optimization algorithm is developed such that the optimal parameters of the MIMO feedforward controller can be iteratively solved based entirely on the measured data.The proposed approach requires no parametric model of the ultra-precision motion stage,and needs only one tracking experiment in each iteration.The proposed approach is applied to a wafer stage of an industrial lithographic scanner.By tracking 2 different references,the experimental results illustrate that the proposed approach effectively improves the tracking performance and reduces the coupling among the multiple degrees of freedom.