An alternative structure for next generation regulatory controllers and scale-up of copper(indium gallium)selenide thin film co-evaporative physical vapor deposition process.

摘要

Process control systems have three key performance attributes: Set-point tracking (T)---ability to cause the process output to follow set-point changes rapidly and faithfully; Disturbance rejection (D)---ability to counteract the effects of external disturbances; and Robustness ( R)---ability to remain stable and perform well in the face of inevitable plant/model mismatch. A controller whose tuning constants are related directly to these performance attributes will have definite advantages over other controllers. However, the popular PID controller, even though simple, has an intrinsic structure that results in a complicated, hence non-transparent, relationship between its tuning parameters and the three controller performance attributes, limiting the controller's achievable performance and making tuning arguably more complex than necessary. In order to overcome the weaknesses of the PID controller, we have developed an alternative regulatory controller (the RTDA 1 controller) having the following salient features: it requires precisely the same information that is required for tuning PID controllers; its tuning parameters are directly related to the three key controller attributes of R, T, and D (an auxiliary fourth parameter, influences the overall controller aggressiveness (A)); all four tuning parameters are normalized to lie between 0 and 1; and the magnitude of a tuning parameter is related to performance aggressiveness, where the higher magnitudes signify conservative performance in the attribute of interest. In addition, the proposed predictive controller is not any more complicated to implement, in either software or hardware, than the PID controller.; In order to study how the choices of various RTDA controller parameter values jointly and individually affect closed-loop stability, a theoretical robust stability analysis is performed. The results of this analysis are subsequently used to develop systematic strategies for choosing the RTDA controller parameters that provide the best possible trade-off between robust stability and performance. The design and implementation of the RTDA controller in practice are illustrated experimentally using two processes: a lab-scale four-tank process with time delay and a pilot-scale physical vapor deposition process with nonlinear dynamics. These experiments demonstrate the RTDA controller's improved performance over PID controllers. The RTDA control scheme is also extended to integrating and open-loop unstable processes.; A pilot-scale co-evaporative physical vapor deposition (PVD) process for manufacture of copper indium gallium diselenide (Cu(InGa)Se2) thin films is chosen to validate the proposed RTDA controller experimentally, since robust control of film thickness and composition set-points for long deposition times cannot be achieved without effective base regulatory control. However, unlike film thickness and composition set-points that can be achieved with proper process control, achieving film thickness uniformity across large area substrates is a process design issue. To achieve good process performance, the process design issues are addressed first, and then the regulatory controller design is improved.; The work presented in Part II of this thesis is focused mainly on the evaporation source design. Such a study requires not only the detailed knowledge of the evaporation source temperature profile, but also accurate estimation of nozzle flow properties (effusion rates and vapor flux distribution). A three-dimensional first-principles electro-thermal model of the source is developed using the COM-SOL Multiphysics'(TM) finite-element method, and the Direct Simulation Monte Carlo (DSMC) technique is employed to predict accurately the nozzle flow properties for any given nozzle geometry and evaporant. These models are validated experimentally, and subsequently used to design evaporation sources that not only achieve the targeted film thickness uniformity, but also max