Identification and control of deposition processes

Abstract

The electrochemical deposition process is defined as the production of a coating on a surface from an aqueous solution composed of several substances. Electrochemical deposition processes are characterized by strong nonlinearity, large complexity and disturbances. Therefore, improving production quality requires the identification of a reasonably accurate model which should be found from data in a reasonable amount of time and with a reasonable computational effort. This identification makes it possible to predict the behavior of unmeasured signals and design a control algorithm to meet the demands of consumers. This thesis addresses the identification and control of the deposition processes. A model for an electrochemical cell that takes into account both electrode interfaces and the activity of ions participating in the deposition process is developed and a method for taking into account uncompensated resistance is proposed. Identifiability of two models, the conventional model and the developed model, is investigated under step and sweep form of applied voltage. It is proven that conventional electrochemical cell model can be identified uniquely using a series of step voltage experiments or in a single linear sweep voltammetry experiment on the basis of the measurements of cell current. The Zakai filtering and pathwise filtering methods are applied to a nonlinear in the parameters electrochemical cell model to estimate the electrode kinetics and mass-transfer parameters of the copper electrodeposition process. In the case of known parameters the feedforward controllers that force the concentration at the boundary to follow the desired reference concentration are designed for the deposition processes. The adaptive boundary concentration control problem for the electrochemical cell with simultaneous parameter identification is solved using the Zakai filtering method. Using such a control, depletion in industrial applications, such as copper deposition baths, can be avoided. An identification method for identifying kinetic parameters and a time-varying mixed potential process of the nonlinear electroless nickel plating model is proposed. The method converts the original nonlinear time-varying identification problem into a time-invariant quadratic optimization problem solvable by conventional least squares.