Non-contact electrostatic chuck controller and system

Abstract

The application discloses a kind of non-contact electrostatic chuck controller and system, including controller main body, active release and monitoring unit, upper layer decision and communication terminal, the controller main body includes: adsorption voltage module, the direct-current high voltage output by adsorption voltage module provides adsorption voltage for electrostatic chuck electrode;Wide spectrum capacitance detection module is used to generate sweep frequency sinusoidal signal V_test, generates small amplitude sinusoidal alternating current test signal V_test (t), is superimposed on direct-current high voltage, then is output to chuck electrode;Multiple harmonic response analysis unit receives and collects the current signal of equivalent capacitance flowing through chuck-wafer interface;The active release and monitoring unit includes real-time monitoring wafer surface potential V_s (t), and the probe preamplifier thereof has locking amplification function, receives reference signal from controller, extracts and neutralizes the surface potential response component V_sig (t) related to means. Significantly improve the reliability, efficiency and safety of wafer transmission in advanced process.

Description

Technical Field

[0001] This invention belongs to the field of electrostatic chuck technology, specifically, it relates to a non-contact electrostatic chuck controller and system. Background Technology

[0002] In semiconductor manufacturing processes, electrostatic chucks (ESCs) are widely used in etching, deposition, and ion implantation processes to achieve non-mechanical fixation of wafers. Their working principle involves applying a high-voltage DC voltage to the chuck electrodes, generating an electrostatic field between the chuck's dielectric layer and the wafer, thus creating an adsorption force. However, as semiconductor technology nodes advance towards 5nm and below, wafers become increasingly thinner, and materials become increasingly low-k dielectrics, traditional ESC control faces increasingly severe challenges during the release phase.

[0003] Current electrostatic chuck release technology mainly employs the reverse voltage method, which theoretically neutralizes adsorbed charges by rapidly switching voltage polarity. However, this technology has inherent drawbacks: First, the semiconductor wafer and the dielectric layer of the chuck may trap charges or experience charge injection. Even when the external voltage is zero, these residual charges will still generate continuous and unstable adsorption forces, leading to delayed release, adhesion, or sudden bouncing of the workpiece, affecting the reliability and positioning accuracy of the robotic arm. Second, at the moment of release, due to the presence of residual forces, the workpiece may undergo micro-sliding or friction with the chuck surface. Simultaneously, the polarity reversal may generate a strong electric field, causing tiny particles adsorbed on the chuck surface to become charged and fly towards the back of the workpiece, causing contamination. Third, applying a reverse high voltage may momentarily cause electrical stress shocks (ESD risk) to sensitive circuits or devices inside the workpiece. Furthermore, the release speed depends on the RC circuit of charge dissipation; for workpieces with high resistivity or chuck materials, the release process is relatively slow.

[0004] To address these issues, the industry has begun exploring non-contact active electrostatic discharge methods. Among existing technologies, KR1027348790000B1 discloses a solution that determines adsorption force by calculating the capacitance of an electrostatic chuck; JP2008179484A and JPH08153774A disclose solutions for non-contact electrostatic neutralization using UV photoionization gas and a charge neutralization gun, respectively. However, these existing technologies remain at the level of passive monitoring and simple open-loop control: that is, the release state is determined by capacitance detection; if no release occurs, the UV lamp or ion gun is activated, and the process ends when the surface potential falls below a threshold. This "segmented" control logic cannot distinguish between free charges and deep-well bound charges, lacks a refined understanding of the release process, poses ESD and stress damage risks for ultra-thin wafers or sensitive devices, and cannot adaptively adjust the neutralization strategy according to different process conditions and material states. The release efficiency and stability are insufficient to meet the requirements of advanced processes. Summary of the Invention

[0005] To address the shortcomings of existing technologies, the present invention aims to provide a non-contact electrostatic chuck controller and system.

[0006] To achieve the aforementioned objectives, the technical solution adopted by this invention includes: A non-contact electrostatic chuck controller includes a controller body, an active release and monitoring unit, and an upper-level decision-making and communication terminal. The controller body includes: The adsorption voltage module outputs a high DC voltage, which first passes through a low-pass filter (LPF, 21) to filter out power supply ripple and prevent it from interfering with the measurement. Then, the input AC or DC power is converted into a precisely settable bipolar DC high voltage (±200V to ±5kV) to provide adsorption voltage for the electrostatic chuck electrodes. Unlike traditional fixed-frequency capacitance measurements, this wide-spectrum capacitance detection module includes a programmable frequency synthesizer and a high-voltage isolation coupling unit. The programmable frequency synthesizer generates a swept-frequency sinusoidal signal V_test, capable of producing a small-amplitude sinusoidal AC test signal V_test(t) with a frequency sweep from f_min to f_max (e.g., 100Hz to 2MHz). This signal is fed into the high-voltage isolation coupling unit, which contains an LC resonant circuit. The coupling capacitor Cc must be a high-voltage capacitor with a withstand voltage higher than the maximum adsorption voltage; the coupling inductor Lc resonates with Cc at the measurement frequency to reduce signal loss and present a low-impedance path to the DC high voltage. V_test is superimposed onto the DC high voltage through this unit and output to the chuck electrode. The multi-harmonic response analysis unit receives and acquires the current signal flowing through the equivalent capacitance of the chuck-wafer interface. This unit uses Fast Fourier Transform (FFT) to decompose the time-domain signal into the real and imaginary parts of the impedance at each frequency, constructing a dielectric relaxation spectrum (DRS). By analyzing the position and intensity of characteristic peaks in the spectrum, the free charge density Q_free and bound charge density Q_trap at the interface are calculated in real time.

[0007] Its core algorithm is based on a modified Cole-Cole model: ε*(ω) = ε_inf + (ε_s - ε_inf) / (1 + (iωτ)^(1-α)) + σ_dc / (iωε_0). By fitting the measured spectrum, the relaxation time τ and the distribution parameter α are obtained, and the charge state is deduced. The real part ε' and imaginary part ε of the complex permittivity ε*(ω) are obtained by frequency sweeping. The measured ε*(ω) data are fitted using a nonlinear least squares method (such as the Levenberg-Marquardt algorithm). The goal is to find a set of parameters {ε_inf, ε_s, τ, α, σ_dc} that minimizes the error between the model curve and the measured curve. Free charge density Q_free: directly related to DC conductivity σ_dc. It is estimated using the formula Q_free = (σ_dc * k_B * T) / (μ * q), where k_B is the Boltzmann constant, T is the temperature (provided by a temperature sensor within the cavity), μ is the carrier mobility (a known material constant), and q is the elementary charge. Alternatively, a more engineering-oriented approach is to establish a calibration curve for σ_dc versus Q_free. The bound charge density Q_trap is related to the relaxation strength Δε = ε_s - ε_inf. Stronger polarization relaxation at the interface indicates a greater amount of bound charge. It can be estimated using Q_trap ≈ q * N_t * Δε, where N_t is the effective charge trap density, which needs to be pre-calibrated using specialized deep-level transient spectroscopy (DLTS) testing. Signal detection and processing path: The current signal returning from the chuck electrode contains both DC and AC components. This signal first enters a current-to-voltage converter (24) (transimpedance amplifier) ​​to convert the current signal into a voltage signal.

[0008] The signal then enters a programmable bandpass filter bank (25). The center frequency of this filter bank is synchronously tracked by a sweep signal from f_min to f_max, and its function is to filter out plasma noise (typically broadband noise) and other interference in the process chamber to the greatest extent possible.

[0009] The filtered clean signal enters the multi-harmonic response analysis unit (26). The core of this unit is a high-speed analog-to-digital converter (ADC) and an FPGA (Field Programmable Gate Array). The FPGA internally implements digital down-conversion and FFT, and calculates the real part Z' and imaginary part Z of the impedance at each frequency point in real time.

[0010] The calculated spectrum data is sent to the upper-level decision and communication terminal (3) via a high-speed serial bus (such as PCIe or LVDS) for model fitting and solution.

[0011] The active release and monitoring unit includes: A tunable excimer UV light source, driven by a UV light source controller, can generate narrowband deep ultraviolet light with a continuously tunable wavelength around 172nm (e.g., 170nm - 175nm). By fine-tuning the wavelength, the photoionization threshold of charge traps on the chuck or wafer surface can be precisely matched under different process conditions (such as after different material deposition), achieving selective and highly efficient charge neutralization. Modulated charge neutralization gun: Driven by a charge gun controller, its output is no longer a simple high-frequency, high-voltage waveform, but a composite waveform modulated by amplitude and frequency. The modulation frequency f_mod can dynamically change within the range of 1Hz to 10kHz; High-sensitivity surface potentiometer: Employing the vibration capacitance method, it monitors the wafer surface potential V_s(t) in real time with a resolution better than 1mV. Its probe preamplifier has a lock-in amplification function, which can receive reference signals from the controller and specifically extract the surface potential response component V_sig(t) related to the neutralization method (UV light intensity modulation frequency or ion gun bias modulation frequency). The upper-layer decision-making and communication terminal includes: The host computer and real-time controller employ a heterogeneous computing architecture (e.g., a multi-core ARM processor paired with an FPGA; the ARM processor handles non-real-time tasks such as communication with the host computer, data logging, and background model updates. It runs a complete charge neutralization physics model and periodically (e.g., every few seconds or minutes) recalibrates and optimizes the model coefficients using historical data accumulated over a period of time. The optimized coefficient table (LUT) is downloaded to the FPGA, which handles all real-time tasks. Internally, it implements: high-speed reading of ADC data, calculation of the current V_s and C_air, nanosecond-level prediction of future states, and for optimal control problems within a prediction window, if the model is sufficiently simplified, it can be solved directly analytically or using a lookup table method). The terminal stores and executes a release algorithm based on Model Predictive Control (MPC).

[0012] A non-contact electrostatic chuck control system includes the following steps: Step A: Process Stage - Real-time Dielectric State Modeling Throughout the wafer adsorption and normal process execution (such as etching and deposition), the controller is not statically outputting data. A wide-spectrum capacitance detection module periodically performs non-invasive dielectric state scans with a low duty cycle. The controller compares the acquired dielectric relaxation spectrum with a baseline health spectrum stored in the database, monitoring in real time for dielectric layer characteristic drift caused by plasma radiation or process temperature. This stage primarily outputs two key state variables: the interface adhesion coefficient C_contact (determined by the high-frequency capacitance value) and the charge trap density trend D_trap(t) (determined by the low-frequency drift of the relaxation spectrum). Step B: Receiving Release Command and Initial State Assessment The communication terminal receives the release command from the host computer. The system enters the pre-release evaluation phase, which lasts for T_eval (e.g., 200ms). During this phase: The adsorption voltage module reduces the adsorption high voltage to 0V according to a preset zero-voltage ramp. The wide-spectrum capacitance detection module performs a full-spectrum scan to acquire the dielectric relaxation spectrum at zero voltage. A surface potentiometer simultaneously measures the initial residual potential V_initial on the wafer surface. Based on the relaxation spectrum analysis of Q_free and Q_trap, and V_initial, the release difficulty index R_challenge of the current interface is calculated. R_challenge is a weighted sum: R_challenge = k1 * |V_initial| + k2 * Q_trap + k3 * (1 / f_peak) Where f_peak is the characteristic frequency corresponding to the bound charge in the relaxation spectrum. The lower the f_peak, the more deeply the charge is bound and the more difficult it is to neutralize. k1, k2, and k3 are empirical weighting coefficients. Step C: Adaptive release path planning and execution, including: C1. Enter the release process. Based on the initial state S0 = {V_initial, Q_free, Q_trap, f_peak} obtained in step B, generate an initial control sequence U0. U0 includes whether to activate UV, whether to activate the ion gun, and their initial parameters. This initial sequence can be generated based on a pre-trained policy network or rule base, but it is not the final solution; C2. Start executing the control sequence U_t (at time t, for example, only the ion gun is turned on, with a frequency of f_mod). C3, the total output potential V_s(t) of the surface potentiometer. The wide-spectrum capacitance detection module does not perform a full-spectrum scan at this moment, but switches to a high-frequency monitoring mode (e.g., fixed at 1MHz) to quickly track changes in the interface air gap and output C_air(t); Lock-in amplifier operation: One lock-in amplifier uses the modulation frequency f_mod of the ion gun as a reference to extract the potential change component V_ion_response(t) caused by the ion gun from V_s(t); if the UV lamp is turned on and modulated, the other lock-in amplifier uses the UV modulation frequency as a reference to extract V_uv_response(t). C4. Includes a simplified charge neutralization physics model for real-time calculation: dV_s / dt = -α(f_mod)* V_s(t) - β(λ, I) * Q_trap(t) + γ * I_ion(t). Where α(f_mod) is the ion gun neutralization coefficient, a function of the modulation frequency f_mod. Its functional form can be obtained through prior offline calibration; β(λ, I) is the UV photo-assisted excitation coefficient, a function of wavelength λ and light intensity I; the γ * I_ion(t) term represents the small charging current that the ion gun may introduce, and γ is the coupling coefficient; the change of Q_trap(t) is indirectly estimated by dV_uv_response / dt; in actual operation, the error between the model prediction value V_s_pred and the actual measurement value V_s_meas will be fine-tuned in real time using a Kalman filter or recursive least squares (RLS) to make it more consistent with the specific operating conditions of the current chamber; The controller uses the current state S_t and the current control parameter U_t, substitutes them into the above model, and predicts the trajectories of V_s and C_air within a short time window (e.g., 50ms). The controller solves an optimization problem: within a prediction window, find the control parameters U_{t+1} for the next time step (which may include fine-tuning f_mod, λ, and turning an actuator on / off) to minimize a cost function. The cost function can be designed as: J = w1 * |V_s(T)|^2 + w2 * T_release + w3 * ∫(P_uv + P_gun) dt; Where T_release is the total release time, P_uv is the instantaneous power of the UV light source, P_gun is the instantaneous power of the charge neutralization gun, the first term |V_s(T)|^2 requires a low final potential, the second term T_release requires a short release time, the third term ∫(P_uv + P_gun) dt requires low total energy consumption, and w1, w2, w3 are weights; C5. Within each control cycle (e.g., 1-5ms), repeat steps 2-4 to achieve a rolling closed loop of execution-observation-prediction-optimization. The process terminates when both V_s(t) and C_air(t) simultaneously meet the release condition.

[0013] Step D: Confirmation and Reporting of Release Completion The system ultimately confirms that the wafer has been completely released (potential < 1mV, capacitance value stable at the background value without wafer). The controller records the characteristic data of this release process (including R_challenge, actual parameter sequence, time, etc.) and uploads it to the host computer for preventive maintenance of the equipment and continuous optimization of process formulations.

[0014] Compared with the prior art, the advantages of the present invention include: (1) The present invention provides a non-contact electrostatic chuck controller and system. Traditional technologies only passively activate the neutralization device after release failure. However, the present invention periodically performs non-invasive dielectric state scanning during the process through a wide-spectrum capacitance detection module to construct a dielectric relaxation spectrum (DRS) and monitor the dielectric layer characteristic drift caused by plasma radiation or process temperature in real time. More importantly, the system can calculate the free charge density Q_free and the bound charge density Q_trap at the interface based on the modified Cole-Cole model, and can predict the release difficulty index R_challenge before the release command is issued. This predictive ability enables the system to plan the optimal release strategy in advance for different charge states, rather than passively dealing with the issue after the wafer sticks together, fundamentally improving the reliability and controllability of the release process. (2) The non-contact electrostatic chuck controller and system provided by this invention abandons the traditional "three-stage" fixed classification logic and replaces it with a unified control framework based on a real-time state machine. The system has a built-in charge neutralization physical model. In each control cycle (1-5ms), it predicts the trajectories of V_s and C_air in the future window based on the current state S_t and control parameters U_t, and optimizes the next control parameters in real time by minimizing the cost function J, which includes the final residual charge energy, total release time, and total energy consumption. This mechanism enables the system to adaptively generate the optimal "release formula" for various process conditions and material states, achieving true closed-loop intelligent control. (3) The present invention provides a non-contact electrostatic chuck controller and system. Through multi-dimensional innovations such as wide-spectrum dielectric analysis, model predictive control, lock-in amplification decoupling, and online self-learning, the present invention elevates the electrostatic chuck release control from "simple open loop" to a new level of "intelligent closed loop", which significantly improves the reliability, efficiency and safety of wafer transfer in advanced processes. Attached Figure Description

[0015] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0016] Figure 1 This is a flowchart illustrating a non-contact electrostatic chuck controller and system according to the present invention. Detailed Implementation

[0017] In view of the shortcomings of the prior art, the inventors of this invention, through long-term research and extensive practice, have proposed the technical solution of this invention. The technical solution, its implementation process, and principles will be further explained below with reference to the accompanying drawings and specific implementation examples in the embodiments of this application.

[0018] It should be noted that the embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and should not be construed as limiting the present invention. The described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, the present invention covers any substitutions, modifications, equivalent methods and solutions made on the spirit, principles and scope of the present invention as defined by the claims. All other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] In the description of this application, the terms "first," "second," "third," and similar terms do not indicate any order, quantity, or importance, but are merely used to distinguish different components. Similarly, the terms "one" or "a" do not indicate a quantity limitation, but rather indicate the presence of at least one. Terms such as "include" or "contain" mean that the elements or objects preceding "include" cover the elements or objects listed following "include" or their equivalents, and do not exclude other elements or objects. Terms such as "connected" or "linked" are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect.

[0020] In the description of this application, the terms center, up, down, front, back, left, right, vertical, horizontal, top, bottom, inside, outside, etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are used only for the convenience of describing this application and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this application. In addition, when using positional terms such as sides, outer side, top and bottom, etc., it should be understood that they are used only for the convenience of understanding and description, taking into account that the structure may be oriented to other positions.

[0021] In the description of this application, unless otherwise expressly specified and limited, the technical or scientific terms used shall have the ordinary meaning understood by a person with ordinary skills in the art to which this application pertains. Terms such as installation, connection, and linking shall be interpreted broadly, for example, they may refer to fixed connection, detachable connection, mating connection, or integral connection. For a person skilled in the art, the specific meaning of the above terms in this application may be understood according to the specific circumstances.

[0022] The present invention aims to introduce and explain the structural composition of a non-contact electrostatic chuck controller and system, as well as the cooperation relationship between the various components. Unless otherwise specified, the dimensions, materials, and manufacturing processes of the various components suitable for the non-contact electrostatic chuck controller and system in the present invention can be selected according to specific circumstances, and no special limitations or explanations are made here.

[0023] Furthermore, to provide the public with a better understanding of the present invention, certain specific details are described in detail in the following description of the invention. However, those skilled in the art will fully understand the invention even without these detailed descriptions.

[0024] Please see Figure 1 A non-contact electrostatic chuck controller includes a controller body, an active release and monitoring unit, and an upper-level decision-making and communication terminal. The controller body includes: The adsorption voltage module outputs a high DC voltage, which first passes through a low-pass filter (LPF, 21) to filter out power supply ripple and prevent it from interfering with the measurement. Then, the input AC or DC power is converted into a precisely settable bipolar DC high voltage (±200V to ±5kV) to provide adsorption voltage for the electrostatic chuck electrodes. Unlike traditional fixed-frequency capacitance measurements, this wide-spectrum capacitance detection module includes a programmable frequency synthesizer and a high-voltage isolation coupling unit. The programmable frequency synthesizer generates a swept-frequency sinusoidal signal V_test, capable of producing a small-amplitude sinusoidal AC test signal V_test(t) with a frequency sweep from f_min to f_max (e.g., 100Hz to 2MHz). This signal is fed into the high-voltage isolation coupling unit, which contains an LC resonant circuit. The coupling capacitor Cc must be a high-voltage capacitor with a withstand voltage higher than the maximum adsorption voltage; the coupling inductor Lc resonates with Cc at the measurement frequency to reduce signal loss and present a low-impedance path to the DC high voltage. V_test is superimposed onto the DC high voltage through this unit and output to the chuck electrode. The multi-harmonic response analysis unit receives and acquires the current signal flowing through the equivalent capacitance of the chuck-wafer interface. This unit uses Fast Fourier Transform (FFT) to decompose the time-domain signal into the real and imaginary parts of the impedance at each frequency, constructing a dielectric relaxation spectrum (DRS). By analyzing the position and intensity of characteristic peaks in the spectrum, the free charge density Q_free and bound charge density Q_trap at the interface are calculated in real time.

[0025] Its core algorithm is based on a modified Cole-Cole model: ε*(ω) = ε_inf + (ε_s - ε_inf) / (1 + (iωτ)^(1-α)) + σ_dc / (iωε_0). By fitting the measured spectrum, the relaxation time τ and the distribution parameter α are obtained, and the charge state is deduced. The real part ε' and imaginary part ε of the complex permittivity ε*(ω) are obtained by frequency sweeping. The measured ε*(ω) data are fitted using a nonlinear least squares method (such as the Levenberg-Marquardt algorithm). The goal is to find a set of parameters {ε_inf, ε_s, τ, α, σ_dc} that minimizes the error between the model curve and the measured curve. Free charge density Q_free: directly related to DC conductivity σ_dc. It is estimated using the formula Q_free = (σ_dc * k_B * T) / (μ * q), where k_B is the Boltzmann constant, T is the temperature (provided by a temperature sensor within the cavity), μ is the carrier mobility (a known material constant), and q is the elementary charge. Alternatively, a more engineering-oriented approach is to establish a calibration curve for σ_dc versus Q_free. The bound charge density Q_trap is related to the relaxation strength Δε = ε_s - ε_inf. Stronger polarization relaxation at the interface indicates a greater amount of bound charge. It can be estimated using Q_trap ≈ q * N_t * Δε, where N_t is the effective charge trap density, which needs to be pre-calibrated using specialized deep-level transient spectroscopy (DLTS) testing. Signal detection and processing path: The current signal returning from the chuck electrode contains both DC and AC components. This signal first enters a current-to-voltage converter (24) (transimpedance amplifier) ​​to convert the current signal into a voltage signal.

[0026] The signal then enters a programmable bandpass filter bank (25). The center frequency of this filter bank is synchronously tracked by a sweep signal from f_min to f_max, and its function is to filter out plasma noise (typically broadband noise) and other interference in the process chamber to the greatest extent possible.

[0027] The filtered clean signal enters the multi-harmonic response analysis unit (26). The core of this unit is a high-speed analog-to-digital converter (ADC) and an FPGA (Field Programmable Gate Array). The FPGA internally implements digital down-conversion and FFT, and calculates the real part Z' and imaginary part Z of the impedance at each frequency point in real time.

[0028] The calculated spectrum data is sent to the upper-level decision and communication terminal (3) via a high-speed serial bus (such as PCIe or LVDS) for model fitting and solution.

[0029] The active release and monitoring unit includes: A tunable excimer UV light source, driven by a UV light source controller, can generate narrowband deep ultraviolet light with a continuously tunable wavelength around 172nm (e.g., 170nm - 175nm). By fine-tuning the wavelength, the photoionization threshold of charge traps on the chuck or wafer surface can be precisely matched under different process conditions (such as after different material deposition), achieving selective and highly efficient charge neutralization. Modulated charge neutralization gun: Driven by a charge gun controller, its output is no longer a simple high-frequency, high-voltage waveform, but a composite waveform modulated by amplitude and frequency. The modulation frequency f_mod can dynamically change within the range of 1Hz to 10kHz; High-sensitivity surface potentiometer: Employing the vibration capacitance method, it monitors the wafer surface potential V_s(t) in real time with a resolution better than 1mV. Its probe preamplifier has a lock-in amplification function, which can receive reference signals from the controller and specifically extract the surface potential response component V_sig(t) related to the neutralization method (UV light intensity modulation frequency or ion gun bias modulation frequency). The upper-layer decision-making and communication terminal includes: The host computer and real-time controller employ a heterogeneous computing architecture (e.g., a multi-core ARM processor paired with an FPGA; the ARM processor handles non-real-time tasks such as communication with the host computer, data logging, and background model updates. It runs a complete charge neutralization physics model and periodically (e.g., every few seconds or minutes) recalibrates and optimizes the model coefficients using historical data accumulated over a period of time. The optimized coefficient table (LUT) is downloaded to the FPGA, which handles all real-time tasks. Internally, it implements: high-speed reading of ADC data, calculation of the current V_s and C_air, nanosecond-level prediction of future states, and for optimal control problems within a prediction window, if the model is sufficiently simplified, it can be solved directly analytically or using a lookup table method). The terminal stores and executes a release algorithm based on Model Predictive Control (MPC).

[0030] A non-contact electrostatic chuck control system includes the following steps: Step A: Process Stage - Real-time Dielectric State Modeling Throughout the wafer adsorption and normal process execution (such as etching and deposition), the controller is not statically outputting data. A wide-spectrum capacitance detection module periodically performs non-invasive dielectric state scans with a low duty cycle. The controller compares the acquired dielectric relaxation spectrum with a baseline health spectrum stored in the database, monitoring in real time for dielectric layer characteristic drift caused by plasma radiation or process temperature. This stage primarily outputs two key state variables: the interface adhesion coefficient C_contact (determined by the high-frequency capacitance value) and the charge trap density trend D_trap(t) (determined by the low-frequency drift of the relaxation spectrum). Step B: Receiving Release Command and Initial State Assessment The communication terminal receives the release command from the host computer. The system enters the pre-release evaluation phase, which lasts for T_eval (e.g., 200ms). During this phase: The adsorption voltage module reduces the adsorption high voltage to 0V according to a preset zero-voltage ramp. The wide-spectrum capacitance detection module performs a full-spectrum scan to acquire the dielectric relaxation spectrum at zero voltage. A surface potentiometer simultaneously measures the initial residual potential V_initial on the wafer surface. Based on the relaxation spectrum analysis of Q_free and Q_trap, and V_initial, the release difficulty index R_challenge of the current interface is calculated. R_challenge is a weighted sum: R_challenge = k1 * |V_initial| + k2 * Q_trap + k3 * (1 / f_peak) Where f_peak is the characteristic frequency corresponding to the bound charge in the relaxation spectrum. The lower the f_peak, the more deeply the charge is bound and the more difficult it is to neutralize. k1, k2, and k3 are empirical weighting coefficients. Step C: Adaptive release path planning and execution, including: C1. Enter the release process. Based on the initial state S0 = {V_initial, Q_free, Q_trap, f_peak} obtained in step B, generate an initial control sequence U0. U0 includes whether to activate UV, whether to activate the ion gun, and their initial parameters. This initial sequence can be generated based on a pre-trained policy network or rule base, but it is not the final solution; C2. Start executing the control sequence U_t (at time t, for example, only the ion gun is turned on, with a frequency of f_mod). C3, the total output potential V_s(t) of the surface potentiometer. The wide-spectrum capacitance detection module does not perform a full-spectrum scan at this moment, but switches to a high-frequency monitoring mode (e.g., fixed at 1MHz) to quickly track changes in the interface air gap and output C_air(t); Lock-in amplifier operation: One lock-in amplifier uses the modulation frequency f_mod of the ion gun as a reference to extract the potential change component V_ion_response(t) caused by the ion gun from V_s(t); if the UV lamp is turned on and modulated, the other lock-in amplifier uses the UV modulation frequency as a reference to extract V_uv_response(t). C4. Includes a simplified charge neutralization physics model for real-time calculation: dV_s / dt = -α(f_mod)* V_s(t) - β(λ, I) * Q_trap(t) + γ * I_ion(t). Where α(f_mod) is the ion gun neutralization coefficient, a function of the modulation frequency f_mod. Its functional form can be obtained through prior offline calibration; β(λ, I) is the UV photo-assisted excitation coefficient, a function of wavelength λ and light intensity I; the γ * I_ion(t) term represents the small charging current that the ion gun may introduce, and γ is the coupling coefficient; the change of Q_trap(t) is indirectly estimated by dV_uv_response / dt; in actual operation, the error between the model prediction value V_s_pred and the actual measurement value V_s_meas will be fine-tuned in real time using a Kalman filter or recursive least squares (RLS) to make it more consistent with the specific operating conditions of the current chamber; The controller uses the current state S_t and the current control parameter U_t, substitutes them into the above model, and predicts the trajectories of V_s and C_air within a short time window (e.g., 50ms). The controller solves an optimization problem: within a prediction window, find the control parameters U_{t+1} for the next time step (which may include fine-tuning f_mod, λ, and turning an actuator on / off) to minimize a cost function. The cost function can be designed as: J = w1 * |V_s(T)|^2 + w2 * T_release + w3 * ∫(P_uv + P_gun) dt; Where T_release is the total release time, P_uv is the instantaneous power of the UV light source, P_gun is the instantaneous power of the charge neutralization gun, the first term |V_s(T)|^2 requires a low final potential, the second term T_release requires a short release time, the third term ∫(P_uv + P_gun) dt requires low total energy consumption, and w1, w2, w3 are weights; C5. Within each control cycle (e.g., 1-5ms), repeat steps 2-4 to achieve a rolling closed loop of execution-observation-prediction-optimization. The process terminates when both V_s(t) and C_air(t) simultaneously meet the release condition.

[0031] Step D: Confirmation and Reporting of Release Completion The system ultimately confirms that the wafer has been completely released (potential < 1mV, capacitance value stable at the background value without wafer). The controller records the characteristic data of this release process (including R_challenge, actual parameter sequence, time, etc.) and uploads it to the host computer for preventive maintenance of the equipment and continuous optimization of process formulations.

[0032] In this way, Using two lock-in amplifiers, with the ion gun modulation frequency and the UV light modulation frequency as references respectively, the response components V_ion_response(t) and V_uv_response(t) induced by different neutralization methods are extracted from the total surface potential V_s(t). This coherent demodulation technique can extract weak but effective signals against a strong plasma noise background, evaluate the contribution and efficiency of different neutralization methods in real time, and provide high-quality feedback input for model predictive control.

[0033] Employing a heterogeneous computing architecture, the ARM processor handles background model updates, using historical data to fine-tune the coefficients α(f_mod), β(λ, I), etc., of the charge neutralization physics model in real time through Kalman filters or recursive least squares methods, making it more closely reflect the specific operating conditions of the current chamber; the FPGA is responsible for nanosecond-level real-time prediction and control. This "offline calibration + online update" mechanism ensures that the model accuracy continuously improves as the system operates, adapting to equipment aging and process drift. The cost function J comprehensively considers the final residual charge energy (|V_s(T)|). 2 The release time (T_release) and energy consumption (∫(P_uv+P_gun)dt) are calculated and balanced using weighting coefficients w1, w2, and w3. Process engineers can adjust the weights according to specific needs: for critical processes, w1 can be increased to ensure complete charge neutralization; for non-critical processes, w2 can be increased to improve capacity; and for processes with strict thermal budgets, w3 can be increased. 3 Extend equipment lifespan. This flexibility allows the system to adapt to the differentiated needs of various process scenarios; By precisely matching the photoionization threshold of charge traps in different materials using a tunable excimer UV light source (continuously adjustable from 170nm to 175nm), selective and highly efficient charge neutralization is achieved. Furthermore, by modulating the amplitude and frequency of a modulated charge neutralization gun, the ion gun perturbation is guided to resonate with the characteristic response frequency of the interface charge, achieving precise charge neutralization with minimal energy input. This "spectral matching" and "resonance excitation" mechanism fundamentally eliminates the risk of ESD damage to sensitive devices, making it particularly suitable for advanced processes of 5nm and below. The system records characteristic data for each release process (including the release difficulty index R_challenge, the actual execution parameter sequence, and the time consumed), and uploads it to the host computer. The long-term accumulated dielectric relaxation spectrum and release difficulty index provide a quantitative basis for assessing the aging state of the electrostatic chuck and predicting its remaining lifespan, realizing an upgrade from preventive maintenance to predictive maintenance.

[0034] It should be understood that the above embodiments are only for illustrating the technical concept and features of the present invention, and are intended to enable those skilled in the art to understand the content of the present invention and implement it accordingly. It should not be considered that the specific implementation of the present invention is limited to these descriptions. For those skilled in the art, several simple deductions or substitutions can be made without departing from the concept of the present invention. All equivalent changes or modifications made in accordance with the spirit and essence of the present invention should be covered within the protection scope of the present invention.

Claims

1. A non-contact electrostatic chuck controller, characterized in that: It includes the controller body, active release and monitoring unit, and upper-level decision-making and communication terminal. The controller body includes: The adsorption voltage module outputs a high DC voltage, which first passes through a low-pass filter to remove power ripple, and then converts the input AC or DC power into a precisely settable bipolar DC high voltage to provide adsorption voltage for the electrostatic chuck electrodes. The wide-spectrum capacitance detection module is used to generate a swept-frequency sinusoidal signal V_test, which produces a small-amplitude sinusoidal AC test signal V_test(t), which is superimposed on the DC high voltage and then output to the chuck electrode. The multi-harmonic response analysis unit receives and acquires the current signal flowing through the equivalent capacitance of the chuck-wafer interface; The active release and monitoring unit includes: Tunable excimer UV light source generates continuously tunable narrowband deep ultraviolet light, and precisely matches the photoionization threshold of the charge traps on the chuck or wafer surface under different process conditions by fine-tuning the wavelength. The modulated charge neutralization gun outputs a composite waveform that has undergone amplitude modulation and frequency modulation. A high-sensitivity surface potentiometer monitors the wafer surface potential V_s(t) in real time. Its probe preamplifier has a lock-in amplification function, receives a reference signal from the controller, and extracts the surface potential response component V_sig(t) related to the neutralization method. The upper-layer decision-making and communication terminal includes: The host computer and real-time controller are composed of a multi-core ARM processor and an FPGA. The ARM processor is responsible for non-real-time tasks, and the FPGA is responsible for all real-time tasks. It also stores and executes a release algorithm based on model predictive control.

2. The non-contact electrostatic chuck controller according to claim 1, characterized in that: The wide-spectrum capacitance detection module includes a programmable frequency synthesizer and a high-voltage isolation coupling unit. The programmable frequency synthesizer generates a swept-frequency sinusoidal signal V_test, which produces a small-amplitude sinusoidal AC test signal V_test(t). The small-amplitude sinusoidal AC test signal V_test(t) is sent to the high-voltage isolation coupling unit, which has an internal LC resonant circuit. The small-amplitude sinusoidal AC test signal V_test is superimposed on the DC high voltage through the high-voltage isolation coupling unit and output to the chuck electrode.

3. A non-contact electrostatic chuck controller according to claim 2, characterized in that: The multi-harmonic response analysis unit also includes using fast Fourier transform to decompose the time-domain signal into the real and imaginary parts of the impedance at each frequency to construct a dielectric relaxation spectrum; by analyzing the position and intensity of the characteristic peaks in the spectrum, the free charge density and bound charge density at the interface are calculated in real time.

4. A non-contact electrostatic chuck controller according to claim 3, characterized in that: The free charge density Q_free and the bound charge density Q_trap are based on a modified Cole-Cole model: ε*(ω) = ε_inf + (ε_s - ε_inf) / (1 + (iωτ)^(1-α)) + σ_dc / (iωε_0), by fitting the measured spectrum, the relaxation time τ and the distribution parameter α are obtained, and then the charge state is deduced.

5. A non-contact electrostatic chuck control system, employing the system described in any one of claims 1-4, characterized in that: Includes the following steps: Step A: Compare the acquired dielectric relaxation spectrum with the baseline health spectrum stored in the database, monitor the dielectric layer characteristic drift caused by plasma radiation or process temperature in real time, and output two state variables: interface adhesion coefficient C_contact and charge trap density trend D_trap(t). Step B: The communication terminal receives the release command from the host computer and enters the pre-release evaluation phase, which lasts for T_eval. During this phase: The adsorption voltage module reduces the adsorption high voltage to 0V according to a preset zero-voltage ramp. The wide-spectrum capacitance detection module performs a full-spectrum scan to acquire the dielectric relaxation spectrum at zero voltage. A surface potentiometer simultaneously measures the initial residual potential V_initial on the wafer surface. Based on the free charge density Q_free and bound charge density Q_trap obtained from the relaxation spectrum, as well as the initial residual potential V_initial, the dynamic programming engine calculates the release difficulty index R_challenge of the current interface. R_challenge is a weighted sum: R_challenge = k1 * |V_initial| + k2 * Q_trap + k3 * (1 / f_peak), Where f_peak is the characteristic frequency of the bound charge in the relaxation spectrum, and k1, k2, k3 are empirical weighting coefficients. Step C: Adaptive release path planning and execution, including: C1. Enter the release process. Based on the initial state S0 = {V_initial, Q_free, Q_trap,f_peak} obtained in step B, generate an initial control sequence U0. U0 contains whether to start UV, whether to start ion gun, and their initial parameters. C2. Begin executing the control sequence U_t; C3, the total potential output of the surface potentiometer is V_s(t); C4. Built-in charge neutralization physical model: dV_s / dt = -α(f_mod) * V_s(t) - β(λ, I) * Q_trap(t) + γ * I_ion(t), Where α(f_mod) is the ion gun neutralization coefficient, a function of the modulation frequency f_mod; β(λ, I) is the UV photo-assisted excitation coefficient, a function of wavelength λ and light intensity I; the γ * I_ion(t) term represents the small charging current that the ion gun may introduce, and γ is the coupling coefficient; the change in Q_trap(t) is indirectly estimated by dV_uv_response / dt; The controller uses the current state S_t and the current control parameter U_t, substitutes them into the above charge neutralization physical model, and predicts the trajectories of V_s and C_air within a short time window in the future; The controller solves an optimization problem: within the prediction window, find the control parameters U_{t+1} for the next time step that minimize a cost function. C5. Repeat steps C2-C4 within each control cycle to achieve a rolling closed loop of execution-observation-prediction-optimization; the process terminates when both V_s(t) and C_air(t) satisfy the release condition. Step D: Finally, confirm that the wafer has been completely released. The controller records the characteristic data of this release process and uploads it to the host computer.

6. A non-contact electrostatic chuck control system according to claim 5, characterized in that: The cost function is: J = w1 * |V_s(T)|^2 + w2 * T_release + w3 * ∫(P_uv + P_gun) dt, Where T_release is the total release time, P_uv is the instantaneous power of the UV light source, P_gun is the instantaneous power of the charge neutralization gun, the first term |V_s(T)|^2 requires a low final potential, the second term T_release requires a short release time, the third term ∫(P_uv + P_gun) dt requires low total energy consumption, and w1, w2, w3 are weights.

7. A non-contact electrostatic chuck control system according to any one of claims 5 or 6, characterized in that: In step C3, the wide-spectrum capacitance detection module switches to high-frequency monitoring mode to quickly track changes in the interface air gap and outputs C_air(t). A lock-in amplifier is used to extract the potential change component V_ion_response(t) caused by the ion gun from V_s(t) with reference to the modulation frequency f_mod of the ion gun; if the UV lamp is turned on and modulated, another lock-in amplifier is used to extract V_uv_response(t) with reference to the UV modulation frequency.

8. A computer-readable medium having a computer program stored thereon, characterized in that: When the program is executed by the processor, it performs the steps as described in any one of claims 5-7.

Images