Multi-Parameter Coordinated Control for High-Voltage Power Supply of Exposure Machine

In the lithography process of exposure machines, the high-voltage power supply needs to simultaneously control multiple parameters such as output voltage (accuracy ±0.1%), output current (accuracy ±0.5%), module temperature (<60℃), load impedance matching (reflected power <5%), and exposure speed-related voltage (dynamically adjusted with exposure speed). There is a strong coupling between parameters. For example, the increase in exposure speed will lead to an increase in load current, which in turn causes an increase in module temperature. If only the voltage is controlled, the current will overshoot; if only the temperature is controlled, the voltage accuracy will be sacrificed. The traditional "single-parameter PID control" is prone to mutual interference between parameters, leading to unstable power supply output and poor lithography pattern consistency (line width range exceeding 5nm).
Multi-parameter coordinated control needs to build a system of "coupling modeling - decoupling control - dynamic optimization": at the coupling modeling level, based on mechanism analysis and data-driven, a multi-parameter coupling mathematical model is established to quantify the influence coefficients between parameters. For example, for every 1℃ increase in temperature, the output voltage deviation increases by 0.02%; for every 1A increase in load current, the reflected power increases by 0.8%. A simulation model is built through MATLAB/Simulink to simulate the parameter change trend under different working conditions; at the decoupling control level, the "multi-variable decoupling PID" algorithm is adopted. By introducing a decoupling compensator, the coupling between parameters is eliminated, and the multi-variable system is transformed into multiple independent single-variable systems. For example, a temperature compensation term is added to the voltage control loop to real-time correct the impact of temperature on voltage, and a load impedance compensation term is added to the current control loop to avoid current fluctuation caused by impedance change; at the dynamic optimization level, the model predictive control (MPC) algorithm is adopted. With "optimal lithography accuracy" as the goal (line width deviation <1nm), the deviation between the actual value and the target value of each parameter is real-time monitored, and the optimal control quantity is calculated through rolling optimization. The control parameters are updated every 1ms to ensure that each parameter can still be stabilized within the target range when the working condition changes (e.g., exposure speed increases from 100mm/s to 200mm/s).
After applying this scheme in a 14nm process exposure machine, the output voltage accuracy increased from ±0.15% to ±0.08%, the current accuracy increased from ±0.6% to ±0.3%, the module temperature was stabilized at 55℃±2℃, the reflected power was <3%, and the lithography line width range was reduced from 5.2nm to 1.3nm, significantly improving pattern consistency. At the same time, the dynamic response time of the power supply was shortened from 10ms to 3ms, meeting the needs of rapid process switching.