Hybrid model based iterative learning control for semiconductor processes with uncetain metrology delay

A hybrid model based iterative learning control method, called HMILC, which combines the strengths of a deterministic model and a stochastic model, is developed. The deterministic model based controller can respond quickly to deterministic changes, such as set-point changes and drifts, while the stochastic model based controller is effective in minimizing the effects due to random inputs, such as non-Gaussian disturbances and metrology lags of the uncertain duration. HMILC has a built-in capability of updating its parameters online. The performance of the proposed HMILC is evaluated in an industrial setting. Results show that the process with HMILC exhibits the best responses to the set-point changes and drifts in the presence of non-Gaussian disturbances and metrology lags of the uncertain delay.