|M.Sc Student||Benjamin Joshua|
|Subject||Advanced Control and Simulation of Rapid Thermal Processes|
|Department||Department of Chemical Engineering||Supervisor||Professor Daniel Lewin|
The fast heating of silicon wafers in Rapid Thermal Processes (RTP) result in several temperature control problems: (1) The heating is achieved through radiation, which imposes a non-linear heat transfer mechanism, (2) the silicon properties change significantly with temperature, and (3) the properties of the chamber itself change from wafer to wafer due to temperature changes in the reactor and emissivity variations due to substance deposition.
Temperature control using conventional linear feedback controllers is inadequate since variations of the system dynamics between one wafer and the next may significantly reduce controller performance. Furthermore, conventional tuning methods for feedback controller do not comply non-linear systems.
The goal of this work was to provide control solution for the system using existing hardware and feedback control devices.
The proposed solution is an iterative learning controller (ILC). The ILC identifies changes in the system dynamic behavior and modifies the control action accordingly. The required performances of the system are expressed in terms of temperature overshoot and set point temperature settling time. The ILC is applied utilizing a unique algorithm, which ensures improved performance with each iteration, along with guaranteed robust stability. The algorithm inputs are the previous wafer processing performance, and the current dynamic behavior of the system, as estimated in the previous iteration. The control action is modified from run-to-run by updating the feedback controller tuning parameters.
An ILC algorithm is developed and used with existing system hardware, thus simplifying the implementation of the solution on an operating system. Experimental results are presented that demonstrate the efficiency of the proposed solution.