Anti-disturbance Switching Control Of Silicon Single Crystal Growth Processes

Semiconductor silicon monocrystal materials are widely used in the manufacture of integrated circuits as a fundamental and critical material.In order to satisfy the needs for higher production efficiency,lower production costs,increasing chip integration and shrinking minimum line widths,the semiconductor materials industry is in urgent need of larger diameter,better quality silicon single crystal wafers.The silicon single crystal growth process is a multi-field,multi-phase coupled complex non-linear process,and switching control theory is very effective in dealing with non-linear and complex control systems.The control domain of the silicon single crystal growth process is appropriately partitioned into multiple sub-domains,and appropriate equilibrium points are determined in each sub-domain,so that the silicon single crystal growth switching model has high accuracy near each equilibrium point,thus improving the control accuracy of the silicon single crystal growth process,which is the main research content of this paper.In this paper,we study the medium diameter growth phase of the silicon single crystal growth process,build a growth mechanism model and design a silicon single crystal growth process controller using an anti-disturbance switching control strategy to achieve effective control of the thermal field and crystal diameter.To this end,the research in this paper is as follows:1.Under certain conditions,the growth process of silicon single crystals by direct drawing is a solid-liquid transformation of silicon material from melt to crystal.Starting from the growth mechanism,a non-linear model of silicon single crystal growth with heater power and crystal lifting speed as control inputs and silicon melt temperature and crystal diameter in the crucible as control outputs is established by analysing each key variable and based on theories such as conservation of energy transfer,conservation of mass and curved lunar surface geometry,and then a small perturbation linearisation theory is used to obtain a silicon single crystal growth switching model,and finally a zero-order retainer is used to obtaining a discrete switching model for silicon single crystal growth;2.An event-triggered model predictive control strategy is used to control the silicon single crystal growth process.Firstly,a model prediction strategy is used to give the controller solution equation for the current moment through Lyapunov stability analysis,and then a rolling optimisation strategy is used to obtain the optimal controller for all moments.Then,an event-triggered strategy is introduced to build an event-triggered mechanism based on silicon melt temperature and silicon single crystal diameter deviation with sampling times to reduce the computational effort in the control process,improve the control real-time,and achieve efficient control of the silicon single crystal growth process.Finally,using the switching system,the controller solution is carried out in the subsystem to improve the control accuracy of the whole system,making the controller efficient in the whole system;3.State feedback is used for the control of the silicon single crystal growth process.Firstly,considering the multiple disturbances in industrial production,for harmonic disturbances,an interference observer is used for estimation.Based on the system state and disturbance estimates,a feed-forward feedback controller is designed to achieve control of the growth process.Then,the finite-time stable,finite-time bounded and finite-time H∞ performance conditions for the silicon single crystal growth process are then given using multi-Lyapunov theory,the mean residence time method,and linear matrix inequalities.Finally,model errors in the modelling process are then considered and uncertainty terms are introduced to improve the robustness of the controller;4.Output feedback is used to control the silicon single crystal growth process.Firstly,some states that cannot be directly measured in the silicon single crystal growth process are considered.Based on the measured output values,a state observer is designed to estimate the states of the system that are not directly measurable,an interference observer is designed using the state estimates to estimate the harmonic interference in the industrial process,and an anti-interference controller is designed using the interference estimates.Then,the finite-time bounded and H∞performance conditions for the silicon single crystal growth process are given by multi-Lyapunov theory,the mean residence time method,and linear matrix inequalities.Finally,the model uncertainty term is introduced to improve the reliability and robustness of the designed controller with respect to the modelling errors in the modelling process.