Fractional-Order Variable-Gain Super-Twisting Control With Application to Wafer Stages of Photolithography Systems

Zhian Kuang, Liting Sun, Huijun Gao, Masayoshi Tomizuka2020 · DOI

TL;DR

This paper introduces a novel control strategy, the Fractional-Order Variable-Gain Super-Twisting Algorithm (PFVSTA), to significantly improve the precision and speed of wafer stages in photolithography systems. By combining a fractional-order sliding surface with a variable-gain super-twisting approach, the PFVSTA effectively mitigates chattering, enhances state dynamics, and robustly compensates for disturbances and uncertainties, leading to superior tracking performance.

What Problem Does This Paper Address?

Photolithography systems are critical for semiconductor manufacturing, requiring extremely precise movement of wafer stages to achieve nanometer-level overlay accuracy. However, wafer stages are subject to various challenges that degrade performance:

  • Chattering: Traditional sliding mode control (SMC) methods suffer from chattering, a high-frequency oscillation that reduces precision.
  • Disturbances and Uncertainties: External vibrations, force ripples, and cable compliance introduce unpredictable disturbances. Furthermore, the wafer stage's dynamics can be position-dependent and influenced by factors like acceleration, making precise control difficult.
  • "Water-Bed Effect": This phenomenon describes how suppressing disturbances in one area can amplify them in another, posing a challenge for controllers.
  • Balancing Performance: Achieving fast response during rapid movements and high precision during steady scanning phases simultaneously is a complex task.

Existing control methods, like PID and standard SMC, often struggle to address all these issues effectively, necessitating a more advanced control strategy.

Methods

The paper proposes a Practical Fractional-Order Variable-Gain Super-Twisting Algorithm (PFVSTA), which builds upon the super-twisting algorithm (STA) and sliding mode control (SMC) principles. The key novelties are:

  • Fractional-Order Sliding Surface (FSS): Instead of a traditional integer-order sliding surface, a fractional-order surface is designed. This incorporates fractional calculus terms, which are shown to accelerate state convergence and reduce overshoot, leading to faster and smoother dynamics on the sliding surface.
  • Variable-Gain Super-Twisting Algorithm (VGSTA): The control gains are made dependent on the system's state (specifically, velocity). This "variable-gain" approach allows for better adaptation to changing system dynamics and disturbances, particularly during high-acceleration phases.
  • Equivalent Control and Switching Control: The controller is structured with an equivalent control term (for nominal system behavior) and a switching control term (for robustness). The equivalent control term is designed to incorporate feedforward elements, further enhancing performance by anticipating system behavior.
  • Stability Analysis: The paper provides a theoretical stability analysis using Lyapunov functions to prove that the proposed controller can drive the system states to the desired region, ensuring performance guarantees.

Datasets

Not discussed in the paper. The evaluation relies on simulations and experiments conducted on a real wafer scanning stage testbed.

Key Insights and Ideas

  • Fractional Calculus for Smoother Dynamics: The introduction of fractional-order terms in the sliding surface is a clever way to improve the transient response (faster convergence, less overshoot) without significantly increasing the number of tunable parameters.
  • Variable Gains are Crucial for Acceleration: The paper strongly emphasizes that variable gains are essential for handling the high acceleration phases common in wafer stage operation. Constant gains struggle to adapt, leading to larger errors.
  • Feedforward is Not Just an Add-on: The inclusion of a feedforward term within the equivalent control is shown to be vital for achieving excellent performance, not just for robustness but also for improving the overall tracking accuracy.
  • Interplay Between Phases: The analysis highlights that errors in acceleration/deceleration phases can negatively impact performance in the critical scanning phases, underscoring the need for robust control across all operational stages.

Key Results

  • Simulation Performance: Simulations demonstrated that the PFVSTA achieved significantly smaller tracking errors compared to conventional controllers (CGSTA, VGSTA, IFVSTA) across various scenarios, including acceleration phases.
  • Experimental Validation: Experiments on a real wafer stage testbed confirmed the simulation results. The PFVSTA consistently exhibited the smallest RMS and maximum errors in idle, scanning, and acceleration/deceleration phases, even under increased acceleration demands.
  • Robustness: The PFVSTA showed superior robustness to disturbances, maintaining high precision where other controllers experienced noticeable error increases.
  • Reduced Chattering: The PFVSTA also demonstrated a significant reduction in chattering, contributing to smoother and more precise motion.
  • Parameter Tuning: The paper notes that the PFVSTA has a comparable number of tunable parameters to existing methods, making it practical for implementation.

Limitations and Future Work

  • Limitations:
    • The theoretical analysis relies on assumptions about bounded uncertainties and Lipschitz continuity of velocity, which are standard but might not perfectly capture all real-world complexities.
  • While the paper demonstrates effectiveness, the "water-bed effect" is mentioned as a persistent challenge in wafer stage control, and the extent to which PFVSTA fully mitigates it across all scenarios might require further investigation.
  • Future Work:
    • The paper suggests that further investigation into the relationship between variable gains and the region of convergence could lead to even better performance.
  • The authors implicitly suggest that the PFVSTA's success in handling acceleration phases could be further leveraged for more aggressive scanning trajectories.
  • While the paper focuses on a 1D stage, extending the PFVSTA to multi-dimensional wafer stages would be a natural next step.

Summary updated

Abstract

Abstract In this paper, a novel fractional-order variable-gain super-twisting control (FVSTC) scheme is proposed and applied to improve the tracking performance of wafer stages in the photolithography systems. The FVSTC overcomes the drawbacks of the super-twisting control (STC) such as slow response speed and incomplete compensation to disturbances. First, to improve the dynamics of the states on the sliding surface, a fractional-order sliding surface is designed. Moreover, to improve the dynamics of the sliding mode variable, an equivalent-control-based method is utilized, and a switching controller based on a variable-gain super-twisting algorithm is deployed. Via such designed schemes, the proposed controller is robust against external disturbances and model uncertainties. Stability proof of the closed-loop system is provided. Numerical simulations to track a sinusoidal signal and experiments on a wafer stage testbed are conducted. The results show that the proposed FVSTC scheme can achieve much better tracking performance than conventional methods.
BibTeX
@misc{kuang2020fractional,
  author = {Zhian Kuang and Liting Sun and Huijun Gao and Masayoshi Tomizuka},
  title = {{Fractional-Order Variable-Gain Super-Twisting Control With Application to Wafer Stages of Photolithography Systems}},
  year = {2020},
  doi = {10.1115/isfa2020-9635},
}
Figure