Analysis of closed loop control and sensor for a reactive sputtering drum coater

The reactive sputtering process is characterized by a hysteresis of reactive gas concentration and reactive gas flow [Schiller, Thin Solid Films 118, 255 (1984); Affinito and Parsons, J. Vac. Sci. Technol A. 2, 1275 (1984)]. The precise control of the reactive sputtering process requires maintaining operation at points on the hysteresis that ensures the desired high sputter flux and deposited thin film stoichiometry. The hysteresis is highly nonlinear in these preferred operating regimes. The practical challenge of meeting this in a reactive sputtering batch coating system requires a control system that can compensate for changes in reaction rate for various sputtered metals, reactive gas sticking coefficients, and system pumping speed. Closed loop control algorithms that rapidly bring the reactive sputtering system to the desired hysteresis steady state operating point are desired for multilayer applications such as thin film interference filters that require many target starts (and stops). Conventional methods of starting the reactive sputtering process such as temporally ramping target power, voltage or current and target shutters are undesirable for precision thin film interference filters. We present a reactive gas sensor and analysis of the proportional-integral-derivative closed loop control algorithm for an industrial reactive sputtering batch drum coater. The key enabler of the control algorithms is a reactive gas sensor employed on these drum coaters that exhibits a temporal response of several hundred milliseconds. We discuss how elements from control theory applied to the reactive sputtering process can determine what points on the hysteresis will be the most difficult to control and the limits to implementation for an industrial system. © 2004 American Vacuum Society.