Single-input, single-output run-to-run feedback controller using triple exponentially weighted moving average smoothing for semiconductor manufacturing processes

During the past decade, a variety of ‘run-to-run’ (R2R) control schemes have been proposed and extensively investigated under various assumed models of process dynamics and disturbances whilst applied to the setpoints (recipe) adjustment between batches for the production of semiconductor manufacturing. The R2R control mechanism of this kind represents an interdisciplinary methodology combining response surface modelling, engineering process control (EPC) and statistical process control (SPC), of which the main objective is to fine-tune the recipe so as to maintain the process output of each run as close to the nominal target as possible. In this paper, the single-input single-output (SISO) case is addressed. To overcome the shortcomings in published R2R controllers, an EWMA feedback controller through a strategy of triple smoothing is proposed (in line with the construction of EWMA statistics) which, in the standard R2R practice, compensates process shifts, a deterministic process drift, disturbance of coloured noises, as well as positive autocorrelation (up to the autoregressive order of two) present in the observed response. It is shown via an experimental study that the triple EWMA filter can provide better control performance for R2R applications as compared with those of the control actions of double EWMA and self-tuning. At last, a relevant application to chemical mechanical planarization in semiconductor manufacturing, a critical step in chip fabrication, is used to illustrate the proposed controller.