Electrostatic chuck voltage filtering method, controller, and electrostatic chuck system

Abstract

This application discloses an electrostatic chuck voltage filtering method, controller, and electrostatic chuck system. The method includes: acquiring voltage sampling data of the electrostatic chuck; determining the target operating condition type of the electrostatic chuck based on the voltage sampling data; adjusting a preset Kalman filter based on the target operating condition type to obtain an adjusted Kalman filter; and filtering the voltage sampling data based on the adjusted Kalman filter. This application can identify the target operating condition type of the electrostatic chuck and adaptively adjust the filtering parameters of the Kalman filter based on the target operating condition type, so that the driving voltage output by the Kalman filter quickly approaches or matches the actual driving voltage, thereby controlling the operation of the electrostatic chuck in a timely, fast, and high-precision manner, ensuring that the electrostatic chuck can operate reliably, stably, and with high precision.

Description

Technical Field

[0001] This application relates to the field of semiconductor technology, and in particular to an electrostatic chuck voltage filtering method, controller, and electrostatic chuck system. Background Technology

[0002] Related technologies can output a drive voltage to an electrostatic chuck, enabling it to firmly hold the wafer for processing by ion etching equipment. The operating environment of an electrostatic chuck is complex. To drive the chuck more reliably, the technology needs to sample the drive voltage feedback from the chuck, filter it, and then reliably control the chuck based on the filtered voltage. However, using fixed filtering parameters to control the chuck under all operating conditions can easily lead to slow response or excessive noise, preventing the chuck from reliably holding the wafer. Summary of the Invention

[0003] One objective of this application is to provide an electrostatic chuck voltage filtering method, controller, and electrostatic chuck system, which improves upon related technologies that use fixed filtering parameters to control electrostatic chucks, resulting in the electrostatic chucks being unable to adapt to different working conditions and operate reliably.

[0004] In a first aspect, embodiments of this application provide an electrostatic chuck voltage filtering method, comprising: acquiring voltage sampling data of the electrostatic chuck; determining the target operating condition type of the electrostatic chuck based on the voltage sampling data; adjusting a preset Kalman filter based on the target operating condition type to obtain an adjusted Kalman filter; and filtering the voltage sampling data based on the adjusted Kalman filter.

[0005] Optionally, determining the target operating condition type of the electrostatic chuck based on voltage sampling data includes: determining an operating condition feature set based on voltage sampling data, wherein the operating condition feature set is used to represent the operating condition of the electrostatic chuck; obtaining a preset operating condition identification model; inputting the operating condition feature set into the operating condition identification model so that the operating condition identification model performs an operating condition identification operation based on the operating condition feature set to obtain the target operating condition type of the electrostatic chuck.

[0006] Optionally, the working condition identification model is configured with a working condition lookup table, which includes multiple working condition types and a feature description set corresponding to each working condition type. The working condition feature set is input into the working condition identification model so that the working condition identification model performs a working condition identification operation based on the working condition feature set to obtain the target working condition type of the working condition where the electrostatic chuck is located. This includes: determining the feature description set that matches the working condition feature set in the working condition lookup table, and setting the working condition type corresponding to the feature description set as the target working condition type.

[0007] Optionally, adjusting a preset Kalman filter based on the target operating condition type to obtain an adjusted Kalman filter includes: adjusting the Kalman gain of the Kalman filter based on the target operating condition type to obtain an adjusted Kalman gain, and adjusting the Kalman filter based on the adjusted Kalman gain to obtain the adjusted Kalman filter.

[0008] Optionally, adjusting the Kalman gain of the Kalman filter based on the target operating condition type to obtain the adjusted Kalman gain includes: increasing the Kalman gain of the Kalman filter based on a preset first gain adjustment model in response to the target operating condition type being a voltage step type, to obtain the adjusted Kalman gain; decreasing the Kalman gain of the Kalman filter based on a preset second gain adjustment model in response to the target operating condition type being a voltage steady-state type, to obtain the adjusted Kalman gain; setting the Kalman gain of the Kalman filter to a first gain value based on a preset third gain adjustment model in response to the target operating condition type being an arc discharge type, to obtain the adjusted Kalman gain; and setting the Kalman gain of the Kalman filter to a second gain value based on a preset fourth gain adjustment model in response to the target operating condition type being a random noise type, to obtain the adjusted Kalman gain.

[0009] Optionally, the first gain adjustment model is configured with a first noise covariance function and a second noise covariance function. Improving the Kalman gain of the Kalman filter based on the preset first gain adjustment model to obtain the adjusted Kalman gain includes: obtaining the voltage change rate of the electrostatic chuck; the first noise covariance function is an increasing function with the voltage change rate as the independent variable and the process noise covariance as the dependent variable; the second noise covariance function is a decreasing function with the voltage change rate as the independent variable and the observation noise covariance as the dependent variable; determining the first process noise covariance based on the first noise covariance function and the voltage change rate; determining the first observation noise covariance based on the second noise covariance function and the voltage change rate; and adjusting the Kalman gain of the Kalman filter based on the first process noise covariance and the first observation noise covariance to obtain the adjusted Kalman gain.

[0010] Optionally, the second gain adjustment model is configured with a third noise covariance function and a fourth noise covariance function. Based on the preset second gain adjustment model, the Kalman gain of the Kalman filter is reduced to obtain the adjusted Kalman gain. This includes: obtaining the steady-state voltage value of the electrostatic chuck; the third noise covariance function is a decreasing function with the steady-state voltage value as the independent variable and the process noise covariance as the dependent variable; the fourth noise covariance function is an increasing function with the steady-state voltage value as the independent variable and the observation noise covariance as the dependent variable; determining the second process noise covariance based on the third noise covariance function and the steady-state voltage value; determining the second observation noise covariance based on the fourth noise covariance function and the steady-state voltage value; and adjusting the Kalman gain of the Kalman filter based on the second process noise covariance and the second observation noise covariance to obtain the adjusted Kalman gain.

[0011] Optionally, the third gain adjustment model is configured with a fifth noise covariance function and a sixth noise covariance function. Based on the preset third gain adjustment model, the Kalman gain of the Kalman filter is set to a first gain value to obtain the adjusted Kalman gain. This includes: determining the third process noise covariance based on the fifth noise covariance function, determining the third observation noise covariance based on the sixth noise covariance function, and controlling the Kalman gain of the Kalman filter to be adjusted in a direction that approaches a preset minimum value based on the third process noise covariance and the third observation noise covariance to obtain the first gain value. The first gain value is used as the adjusted Kalman gain.

[0012] In a second aspect, embodiments of this application provide a controller, including a memory and a processor. The memory is connected to the processor, and the processor is used to execute one or more computer programs stored in the memory. When the processor executes one or more computer programs, it causes the controller to implement the above-described electrostatic chuck voltage filtering method.

[0013] In a third aspect, embodiments of this application provide an electrostatic chuck system, including an electrostatic chuck and the aforementioned controller, wherein the controller is electrically connected to the electrostatic chuck and is used to control the operation of the electrostatic chuck.

[0014] The embodiments of this application can achieve the following technical effects: The electrostatic chuck voltage filtering method provided in the embodiments of this application can identify the target working condition type of the electrostatic chuck, and adaptively adjust the filtering parameters of the Kalman filter based on the target working condition type, so that the driving voltage output by the Kalman filter is closer to or consistent with the real driving voltage, thereby controlling the operation of the electrostatic chuck in a timely, fast and high-precision manner, and ensuring that the electrostatic chuck can work reliably, stably and with high precision. Attached Figure Description

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

[0016] Figure 1 A schematic diagram of the circuit structure of an electrostatic chuck control system provided in an embodiment of this application;

[0017] Figure 2 A schematic flowchart illustrating an electrostatic chuck voltage filtering method provided in an embodiment of this application;

[0018] Figure 3 This is a schematic diagram illustrating the identification of target working condition types based on an artificial intelligence model, provided in an embodiment of this application.

[0019] Figure 4 This is a schematic diagram illustrating the identification of target working condition types based on a large language model, provided in an embodiment of this application.

[0020] Figure 5 This is a schematic diagram showing the voltage sampling data after filtering using a Kalman filter based on related technologies.

[0021] Figure 6 This is a schematic diagram showing the voltage sampling data after being filtered using the filtering method provided in this embodiment of the application.

[0022] Figure 7 This is a schematic diagram of the structure of an electrostatic chuck voltage filtering device provided in an embodiment of this application;

[0023] Figure 8 This is a schematic diagram of the structure of a controller provided in an embodiment of this application. Detailed Implementation

[0024] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are for illustrative purposes only and are not intended to limit the scope of this application. All other embodiments obtained by those skilled in the art based on the embodiments in this application without inventive effort are within the scope of protection of this application.

[0025] It should be noted that, unless there is a conflict, the various features in the embodiments of this application can be combined with each other, all of which are within the protection scope of this application. Furthermore, although functional modules are divided in the device schematic diagram and a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in a different order than the module division in the device or the order in the flowchart. Moreover, the terms "first," "second," and "third" used in this application do not limit the data or execution order, but only distinguish identical or similar items with essentially the same function and effect.

[0026] In implementing the embodiments of this application, the inventors discovered that related technologies transmit a driving voltage to the electrostatic chuck through an electrostatic chuck control system. A higher driving voltage can drive the electrostatic chuck to firmly hold the wafer, preventing wafer movement during etching and thus improving process accuracy. The driving voltage output by the electrostatic chuck control system needs to be monitored so that subsequent stages can perform adsorption force control, temperature management, and heat conduction operations based on the monitored driving voltage. The process for monitoring the driving voltage is as follows: The driving voltage output by the electrostatic chuck control system is much larger than the range of the analog-to-digital converter (ADC). The driving voltage output by the electrostatic chuck control system needs to be compressed by inputting a high-impedance voltage divider or voltage reduction network to reduce the high driving voltage to a range that subsequent circuits can handle. Then, the ADC samples the reduced driving voltage to obtain voltage sampling data. The voltage sampling data is then filtered using a filter to obtain filtered voltage sampling data. Related operations are then performed based on the filtered voltage sampling data.

[0027] Although related technologies have enabled the monitoring of the drive voltage of electrostatic chucks and the establishment of feedback mechanisms to ensure stable operation, the operating conditions of electrostatic chucks are complex, such as high voltage and rapid voltage changes. Therefore, the electrostatic chucks provided by these technologies still have the following technical problems:

[0028] 1) Response lag: During voltage rise, power-on instant, polarity switching or sudden change of set value, the filter responds slowly to rapidly changing voltage signals and cannot reflect the real voltage in time. This causes the electrostatic chuck control system to make decisions based on the lag voltage signal, which in turn leads to the phenomenon of lag control of the electrostatic chuck or the electrostatic chuck overshooting or oscillation.

[0029] 2) High noise and interference: Electrostatic chucks operate in complex environments such as high voltage, strong electric field, rarefied gas, arc discharge and process switching. The signal chain of electrostatic chucks is easily affected by various noise sources such as electromagnetic interference, arc discharge pulses, and switching jumps. The analog-to-digital converter is prone to sampling a large number of high-frequency glitches, random spikes or clusters of interference signals during the signal sampling stage, which seriously weakens the purity of voltage sampling data.

[0030] 3) Decreased measurement accuracy: High noise reduces the signal-to-noise ratio (SNR) and the effective number of bits of the analog-to-digital converter in the electrostatic chuck control system, causing the electrostatic chuck control system to make decisions based on low-quality voltage sampling data, which can easily lead to misjudgment or false triggering of protection actions.

[0031] 4) Poor adaptability to operating conditions: The filters provided by related technologies usually use fixed filtering architecture and fixed filtering parameters for filtering, which cannot distinguish between different stages or operating conditions such as startup, ramp-up, steady state, discharge, and arc. The requirements for "fast response" and "stable and high precision" vary greatly in different stages. The filters provided by related technologies cannot take both into account under relevant operating conditions, and are prone to switching back and forth between slow response and excessive noise.

[0032] In summary, in voltage monitoring of electrostatic chuck control systems, relying solely on the traditional process of "voltage reduction → sampling → fixed filtering" is insufficient to simultaneously meet the two key requirements of "high real-time response" and "steady-state high-precision output," resulting in low working accuracy and poor reliability of the electrostatic chuck.

[0033] In view of this, the electrostatic chuck voltage filtering method provided in the embodiments of this application at least solves the following technical problems:

[0034] 1. Filter response lag problem: Traditional Kalman filtering or average filtering algorithms still operate in steady-state model when the voltage rises rapidly, the set value changes, or the system starts or stops. This results in the output voltage signal responding slowly to the actual voltage changes and failing to reflect the dynamic characteristics of the system in a timely manner, causing control delay or overshoot.

[0035] 2. Outlier sensitivity and glitches:

[0036] 3. In high-voltage systems such as aging processes, power supply drives, and electrostatic chuck control, voltage signals are often affected by transient noise such as arc discharge and spike interference. Traditional filtering algorithms struggle to distinguish between real signals and glitches, easily leading to abnormal shifts or even divergence in filtering results, thus reducing system stability.

[0037] 4. Lack of adaptive capability: Related technologies usually rely on fixed parameters and a single mode, and cannot automatically switch filtering strategies according to system start-up and shutdown, changes in target setpoints, or measurement fluctuations, resulting in an inability to balance response speed and filtering accuracy under different operating conditions.

[0038] 5. Slow filter convergence after parameter changes: When the set voltage changes, the internal state of the Kalman filter is not reset in time, causing the new filtering result to be affected by the old state, resulting in a "delayed tracking" phenomenon, which affects the accuracy and stability of voltage control.

[0039] Therefore, the electrostatic chuck voltage filtering method provided in this application embodiment can identify the target operating condition type of the electrostatic chuck and adaptively adjust the filtering parameters of the Kalman filter based on the target operating condition type, so that the driving voltage output by the Kalman filter is closer to or consistent with the real driving voltage. This is beneficial for the electrostatic chuck control system to control the operation of the electrostatic chuck in a timely, fast and high-precision manner, and ensure that the electrostatic chuck can work reliably, stably and with high precision.

[0040] The following embodiments of this application provide an electrostatic chuck control system. Please refer to... Figure 1 The electrostatic chuck control system 100 includes a high-voltage power supply 11, an electrostatic chuck 12, a voltage divider circuit 13, an analog-to-digital converter 14, and a controller 15.

[0041] The high-voltage power supply device 11 is electrically connected to the electrostatic chuck 12 and is used to output a drive voltage to the electrostatic chuck 12 so that the electrostatic chuck 12 can operate based on the drive voltage.

[0042] The electrostatic chuck 12 is used to firmly hold the wafer so that the ion etching machine can etch the wafer.

[0043] Voltage divider circuit 13 is electrically connected to high-voltage power supply device 11 and electrostatic chuck 12 respectively, and is used to sample and reduce the driving voltage transmitted from high-voltage power supply device 11 to electrostatic chuck 12 to obtain a voltage sampling signal. Voltage divider circuit 13 can be a high-impedance voltage divider or voltage reduction network, etc.

[0044] The analog-to-digital converter 14 is electrically connected to the voltage divider circuit 13 and is used to perform analog-to-digital conversion on the voltage sampling signal to obtain voltage sampling data. The voltage divider circuit 13 can reduce the voltage of a large driving voltage, so that the voltage sampling signal falls within the range of the analog-to-digital converter 14. The analog-to-digital converter 14 does not truncate the voltage sampling signal and can reliably and accurately perform analog-to-digital conversion on the complete voltage sampling signal.

[0045] The controller 15 is electrically connected to the analog-to-digital converter 14 and the high-voltage power supply 11. The controller 15 is equipped with a Kalman filter (KF), which can be an extended Kalman filter or a standard Kalman filter. The controller 15 uses the Kalman filter to filter the voltage sampling data, obtaining filtered voltage sampling data. This filtered voltage sampling data reflects the voltage applied to the electrostatic chuck 12 in real time. Then, based on the filtered voltage sampling data, the controller 15 controls the high-voltage power supply 11 to adjust the operating state of the electrostatic chuck.

[0046] The following embodiments of this application provide a method for filtering the voltage of an electrostatic chuck. Please refer to... Figure 2 The embodiment of this application implements the electrostatic chuck voltage filtering method through steps S21 to S24, as detailed below:

[0047] Step S21: Obtain voltage sampling data of the electrostatic chuck.

[0048] The voltage sampling data is obtained by sampling the drive voltage of the electrostatic chuck, which is used to drive the electrostatic chuck. The voltage sampling data includes multiple voltage sample values ​​arranged in sampling order. The expression for the voltage sampling data is... , This is the i-th voltage sample value.

[0049] Step S22: Determine the target operating condition type of the electrostatic chuck based on the voltage sampling data.

[0050] The target operating condition type refers to the current operating condition of the electrostatic chuck. Operating condition types include voltage step type, voltage steady-state type, arc discharge type, random noise type, normal discharge type, and operating condition transition type. The voltage step type is when the driving voltage jumps from one steady-state value to another in a very short time, without a gradual transition. The voltage steady-state type is when the driving voltage fluctuates within a certain range, and the fluctuation range within this range is less than or equal to a preset fluctuation threshold, for example, 0.5%. The arc discharge type is when the driving voltage drops suddenly and is accompanied by a large current spike. The random noise type is when the driving voltage fluctuates irregularly and with a large voltage amplitude near the steady state. The normal discharge type is when the driving voltage decreases in a controllable step manner at a stable rate. The operating condition transition type is an intermediate state between a steady voltage state and a step voltage state, in which the voltage gradually changes. For example, in some embodiments, the driving voltage applied to the electrostatic chuck changes from a steady voltage state to a step voltage state. The steady voltage state can be a state corresponding to a low voltage level (low voltage amplitude) or a high voltage level (high voltage amplitude). The operating condition type corresponding to the process of the driving voltage changing from the steady voltage state to the step voltage state is the operating condition transition type. In some embodiments, the driving voltage applied to the electrostatic chuck changes from a step voltage state to a steady voltage state. For example, after the driving voltage completes the step change, it begins to enter a steady state. The operating condition type corresponding to the process from the completion of the step change to voltage stabilization is the operating condition transition type.

[0051] It is understood that, in addition to the voltage step type, voltage steady state type, arc discharge type, random noise type, normal discharge type, and operating condition transition type listed in the embodiments of this application, those skilled in the art can design corresponding operating condition types in combination with process characteristics and the actual working conditions of the electrostatic chuck, which will not be elaborated here.

[0052] This embodiment of the application determines the target operating condition type of the electrostatic chuck based on voltage sampling data through steps S221 to S223, as detailed below:

[0053] Step S221: Determine the operating condition feature set based on voltage sampling data.

[0054] The operating condition feature set is used to represent the operating conditions of the electrostatic chuck. The operating condition feature set can describe the characteristics of the operating conditions of the electrostatic chuck from multiple dimensions.

[0055] ① Voltage change rate:

[0056] In some embodiments, the operating condition feature set includes the voltage change rate. This application embodiment determines the voltage change rate based on voltage sampling data. For example, this application embodiment calculates the voltage change rate according to Formula 1, as shown below:

[0057] Formula 1

[0058] in, Let be the rate of change of voltage at time k. Let be the driving voltage at time k. The driving voltage at time k-1, The sampling period is specified. This application embodiment can reflect the operating condition of the electrostatic chuck from the voltage change rate. Please refer to Table 1. This application embodiment, in conjunction with Table 1, describes the characteristics of voltage step type, voltage steady-state type, arc discharge type, random noise type, normal discharge type, and operating condition transition type in terms of voltage change rate, as shown in Table 1:

[0059] Table 1

[0060]

[0061] In Table 1, the time unit is ms, and t is the holding time of the driving voltage within the corresponding value range.

[0062] As shown in Table 1, in voltage step type operating conditions, the voltage change rate is large and the time to maintain a large value is long. For example, |d1| is the absolute value of the voltage change rate, the voltage change rate falls within [3000, 30000], and the holding time t falls within [50, 500].

[0063] In steady-state voltage conditions, the rate of voltage change is very small and the time it remains at a very small value is short. For example, the rate of voltage change falls within [0,50] and the holding time t falls within [0,30].

[0064] In arc discharge type operating conditions, the voltage change rate peaks more often within a certain time range, meaning that the driving voltage changes very drastically within that time range. For example, the voltage change rate peaks ≥ 3 times within a 500ms time range.

[0065] In random noise conditions, the voltage change rate is large and the duration of the large value is short, and the number of peak values ​​within a certain time range is small. That is, the driving voltage undergoes instantaneous changes but the duration is very short. For example, the voltage change rate falls within [1000,+∞), the holding time t falls within [0,30], and the number of times it occurs within 500ms is ≤2.

[0066] In normal discharge conditions, the voltage change rate is negative and has a large absolute value, and the time to maintain a large absolute value is relatively long. For example, the voltage change rate falls within [-100000, -30000), and the holding time t falls within [50, 500].

[0067] In the transitional operating condition, compared to other operating condition types, the voltage change rate and the holding time are moderate. For example, the voltage change rate falls within [50, 100] and the holding time t falls within [50, 100].

[0068] ② Rate of change of voltage change:

[0069] In some embodiments, the operating condition feature set includes the rate of change of voltage change. This application embodiment determines the rate of change of voltage change (i.e., voltage acceleration) based on voltage sampling data. For example, this application embodiment calculates the rate of change of voltage change according to Formula 2, as shown below:

[0070] Formula 2

[0071] in, Let be the rate of change of the voltage at time k. This represents the voltage change rate at time k-1. This embodiment of the application can reflect the operating condition of the electrostatic chuck from the rate of change of voltage change. Please refer to Table 2. This embodiment of the application, in conjunction with Table 2, describes the characteristics of the rate of change of voltage change for voltage step type, steady-state type, arc discharge type, random noise type, normal discharge type, and transitional operating condition type, as shown in Table 2:

[0072] Table 2

[0073]

[0074] As shown in Table 2, in voltage step type operating conditions, the rate of change of voltage change peaks at the start or end point. For example, |d2| is the absolute value of the rate of change of voltage change, and the rate of change of voltage change is greater than 300.

[0075] In steady-state voltage conditions, the rate of voltage change is very small; for example, the rate of change of voltage is less than 30%.

[0076] In arc discharge type operating conditions, the rate of change of voltage changes peaks frequently within a certain time range, and the values ​​are large, but the duration is short. For example, the rate of change of voltage changes is greater than 300, the holding time t is less than 30ms, and the number of peaks within a time range of 500ms is ≥3.

[0077] In random noise conditions, the rate of change of voltage is large and remains at a large value for a short period of time, and the number of times a peak occurs within a certain time range is small. For example, the rate of change of voltage is greater than 300, the holding time t is less than 30ms, and the number of times it occurs within 500ms is ≤2.

[0078] In normal discharge conditions, the absolute value of the rate of change of voltage is large at the initial moment but small in the subsequent stage. For example, the rate of change of voltage is greater than 500 at the initial moment but less than 50 in the subsequent stage.

[0079] In the transitional operating condition, the voltage change rate is moderate compared to other operating condition types; for example, the absolute value of the voltage change rate falls within (30, 50).

[0080] ③ Voltage residual:

[0081] In some embodiments, the operating condition feature set includes voltage residuals. This application embodiment determines the voltage residuals based on voltage sampling data. For example, this application embodiment calculates the voltage residuals according to Formula 3, as shown below:

[0082] Formula 3

[0083] in, Let be the voltage residual at time k. Let be the driving voltage at time k. This is the optimal estimate at time k output via the Kalman filter. This embodiment of the application can reflect the operating condition of the electrostatic chuck from the voltage residual. Please refer to Table 3. This embodiment of the application, in conjunction with Table 3, illustrates the characteristics of voltage step type, voltage steady-state type, arc discharge type, random noise type, normal discharge type, and operating condition transition type in terms of voltage residual, as shown in Table 3:

[0084] Table 3

[0085]

[0086] As shown in Table 3, in the voltage step type operating condition, the voltage residual value is of medium magnitude compared to other operating condition types. For example, the voltage residual falls within [50, 150].

[0087] In steady-state voltage conditions, the voltage residual is smaller compared to other conditions; for example, the voltage residual falls within [0,50].

[0088] In the case of arc discharge, the difference between multiple consecutive voltage residuals and the mean residual is greater than 3 standard deviations σ, and the voltage residuals fall within [150, +∞].

[0089] In random noise conditions, the difference between the voltage residual and the mean residual is greater than 3 standard deviations σ, and the voltage residual falls within [150, +∞].

[0090] In normal discharge conditions, the voltage residual is of moderate magnitude compared to other conditions; for example, the voltage residual falls within [50, 150].

[0091] In the transitional operating conditions, the voltage residual shows an upward trend and falls within [30, 300].

[0092] ④ Variance of voltage residual:

[0093] In some embodiments, the operating condition feature set includes the variance of the voltage residual. This application embodiment determines the variance of the voltage residual based on voltage sampling data. For example, this application embodiment calculates the variance of the voltage residual according to Formula 4, as shown below:

[0094] Formula 4

[0095] in, Let be the variance of the voltage residual at time k. Let N be the mean residual of the voltage residual, N be the total number of voltage residuals captured in a sliding statistical window (i.e., the window length), k be the position index of the current sliding statistical window on the sample chain corresponding to the voltage residual (i.e., the k-th sliding statistical window), and i be the index of a single voltage residual within the current sliding statistical window. This embodiment of the application can reflect the operating condition of the electrostatic chuck from the variance of the voltage residual. Please refer to Table 4. This embodiment of the application, in conjunction with Table 4, illustrates the characteristics of the variance of the voltage residual for voltage step type, voltage steady-state type, arc discharge type, random noise type, normal discharge type, and operating condition transition type, as shown in Table 4:

[0096] Table 4

[0097]

[0098] As shown in Table 4, in voltage step-type operating conditions, the variance of the voltage residual is of moderate magnitude compared to other operating conditions. For example, the variance of the voltage residual falls within

[50] . 2 150 2 ]Inside.

[0099] In steady-state voltage conditions, the variance of the voltage residual is smaller compared to other conditions; for example, the variance of the voltage residual falls within [0, 50]. 2 ]Inside.

[0100] In arc discharge type operating conditions, the variance of the voltage residual is relatively large compared to other operating conditions. For example, the variance of the voltage residual falls within

[200] . 2 ,+∞]inside.

[0101] In random noise conditions, the variance of the voltage residual is moderately low compared to other conditions; for example, the variance of the voltage residual falls within

[50] . 2 100 2 ]Inside.

[0102] Under normal discharge conditions, the variance of the voltage residual is moderately high compared to other operating conditions; for example, the variance of the voltage residual falls within

[100] . 2 200 2 ]Inside.

[0103] In transitional operating conditions, the variance of the voltage residual shows an upward trend; for example, the variance of the voltage residual falls within

[30] . 2 300 2 ]Inside.

[0104] ⑤ Residual entropy of voltage residual:

[0105] In some embodiments, the operating condition feature set includes the residual entropy of the voltage residual. This application embodiment determines the residual entropy of the voltage residual based on voltage sampling data. For example, this application embodiment calculates the residual entropy of the voltage residual according to Formula 5, as shown below:

[0106] Formula 5

[0107] in, Let the residual entropy be the voltage residual at time k. Let be the probability that the voltage residual falls into the j-th amplitude interval within the k-th sliding statistical window, with a value ranging from [0,1]. Here, M is a custom coefficient, representing the maximum value of j. In this embodiment, by performing a weighted logarithmic summation on all voltage residuals, all voltage residuals can be compressed into a single scalar residual entropy H(k), which is used to identify the uncertainty and dispersion of the voltage residual distribution within that time window.

[0108] This application embodiment can reflect the operating condition of the electrostatic chuck from the residual entropy of the voltage residual. Referring to Table 5, this application embodiment, in conjunction with Table 5, illustrates the characteristics of voltage step type, voltage steady-state type, arc discharge type, random noise type, normal discharge type, and operating condition transition type in terms of the residual entropy of the voltage residual, as shown in Table 5:

[0109] Table 5

[0110]

[0111] As shown in Table 5, in voltage step type operating conditions, the residual entropy of the voltage residual is moderate compared to other operating condition types. For example, the residual entropy of the voltage residual falls within [0.45, 0.7].

[0112] In steady-state voltage conditions, the residual entropy of the voltage residual is smaller compared to other conditions; for example, the residual entropy of the voltage residual falls within (0, 0.2).

[0113] In arc discharge type operating conditions, the residual entropy of the voltage residual is relatively large compared to other operating conditions. For example, the residual entropy of the voltage residual falls within [0.45,1].

[0114] In random noise conditions, the residual entropy of the voltage residual is moderate compared to other conditions; for example, the residual entropy of the voltage residual falls within [0.45, 0.7].

[0115] In normal discharge conditions, the residual entropy of the voltage residual is moderate compared to other conditions; for example, the residual entropy of the voltage residual falls within [0.55, 0.8].

[0116] In transitional operating conditions, the residual entropy of the voltage residual is moderate, for example, the residual entropy of the voltage residual falls within [0.45, 0.7].

[0117] As described above, the operating condition feature set can include one or more operating condition features. The operating condition feature can be any one of the following: voltage change rate, voltage change rate, voltage residual, voltage residual variance, and voltage residual entropy.

[0118] In some embodiments, this application can utilize a working condition feature to identify the target working condition type of the electrostatic chuck. For example, the working condition feature is the voltage change rate. This application utilizes the fact that voltage step type working conditions and voltage steady-state type working conditions exhibit different characteristics in terms of voltage change rate, and combines the voltage change rate to identify the target working condition type of the electrostatic chuck.

[0119] In other embodiments, different operating conditions exhibit different behaviors across multiple dimensions. Embodiments of this application can combine two or more operating condition characteristics to jointly describe the behavior of an operating condition. For example, combining Tables 1 to 5, it can be seen that the operating condition characteristic sets for voltage step type, voltage steady-state type, arc discharge type, random noise type, normal discharge type, and operating condition transition type are shown in Table 6.

[0120] Table 6

[0121]

[0122] As shown in Table 6, embodiments of this application can describe the same type of operating condition from multiple dimensions. For example, embodiments of this application select {A1, A2, A3, A4, A5} to describe the voltage step type operating condition, {B1, B2, B3, B4, B5} to describe the voltage steady-state type operating condition, {C1, C2, C3, C4, C5} to describe the arc discharge type operating condition, {D1, D2, D3, D4, D5} to describe the random noise type operating condition, {E1, E2, E3, E4, E5} to describe the normal discharge type operating condition, and {F1, F2, F3, F4, F5} to describe the operating condition transition type operating condition.

[0123] This application embodiment is based on a set of operating condition features, which describes the same type of operating condition from multiple dimensions. This is beneficial for accurately and reliably determining the operating condition of the electrostatic chuck, so as to adaptively adjust the filtering parameters of the Kalman filter according to the operating condition. This helps to obtain a more reliable and accurate driving voltage, and ultimately helps the electrostatic chuck control system make accurate and scientific control decisions based on the feedback driving voltage. This is beneficial for improving the working stability, response timeliness and working accuracy of the electrostatic chuck.

[0124] Step S222: Obtain the preset working condition recognition model.

[0125] The working condition identification model is a model for identifying the target working condition type of the electrostatic chuck.

[0126] In some embodiments, the working condition identification model can be an artificial intelligence model. In this application embodiment, the working condition identification model constructed based on artificial intelligence algorithm is used to identify the working condition of the electrostatic chuck.

[0127] The embodiments of this application can be trained based on artificial intelligence algorithms and training sample data to obtain a working condition recognition model. Artificial intelligence algorithms include neural network algorithms, deep learning algorithms, and so on.

[0128] In other embodiments, the working condition recognition model is a large language model or a multimodal large model. In the embodiments of this application, the general large language model or the general multimodal large model can be fine-tuned and trained to obtain a fine-tuned large language model or a multimodal large model.

[0129] In other embodiments, the operating condition identification model is configured with an operating condition lookup table. In this embodiment, the operating condition lookup table is used to identify the operating condition of the electrostatic chuck.

[0130] Step S223: Input the working condition feature set into the working condition identification model so that the working condition identification model can perform working condition identification operation based on the working condition feature set to obtain the target working condition type of the working condition where the electrostatic chuck is located.

[0131] This application embodiment can utilize various forms of working condition identification models to identify the working condition of the electrostatic chuck, as shown below:

[0132] 1) Artificial intelligence model.

[0133] Please see Figure 3 The artificial intelligence model 300 includes an input layer 31, a processing layer 32, and an output layer 33. The input layer 31 is used to concatenate multiple working condition features from the working condition feature set to obtain concatenated features. The processing layer 32 can be a fully connected layer used to process and analyze the concatenated features. The output layer 33 is configured with various working condition labels, each label corresponding to a working condition type. The output layer 33 can map the output of the processing layer 32 to the corresponding working condition label to output the corresponding target working condition type.

[0134] 2) Large language model or multimodal large model.

[0135] Please see Figure 4 In this embodiment, prompt words for identifying working condition types are constructed, and each working condition feature of the working condition feature set is converted into structured text that matches a large language model or a multimodal large model. The structured text and prompt words are input into the large language model or the multimodal large model, so that the large language model or the multimodal large model outputs the target working condition type of the electrostatic chuck based on the structured text under the prompt words.

[0136] 3) Operating condition query table.

[0137] The operating condition lookup table includes multiple operating condition types and a feature description set corresponding to each operating condition type. The feature description set is used to describe the operating condition type from multiple dimensions. The feature description set includes multiple baseline description information. Each baseline description information describes the operating condition of the electrostatic chuck from one dimension. The baseline description information is constructed in advance by the designer based on the performance of each operating condition type.

[0138] Please refer to Table 6. The working condition query table is shown in Table 6. The working condition query table includes 6 feature description sets, namely the first feature description set T1, the second feature description set T2, the third feature description set T3, the fourth feature description set T4, the fifth feature description set T5, and the sixth feature description set T6.

[0139] The first feature description set T1 is used to describe voltage step-type operating conditions. T1 includes five reference description information sets: A1, A2, A3, A4, and A5. For example, as shown in Table 1, reference description information A1 describes that the voltage change rate needs to remain within a large range for an extended period. As shown in Table 2, reference description information A2 describes that the voltage change rate needs to be greater than 300, and so on.

[0140] The second feature description set T2 is used to describe the voltage steady-state type of operating conditions. The second feature description set T2 includes five reference description information, namely B1, B2, B3, B4, and B5.

[0141] The third feature description set T3 is used to describe the working conditions of the arc discharge type. The third feature description set T3 includes 5 reference description information, namely C1, C2, C3, C4 and C5.

[0142] The fourth feature description set T4 is used to describe the operating conditions of random noise type. The fourth feature description set T4 includes 5 reference description information, namely D1, D2, D3, D4 and D5.

[0143] The fifth feature description set T5 is used to describe the operating conditions of the normal discharge type. The fifth feature description set T5 includes five reference description information, namely E1, E2, E3, E4 and E5.

[0144] The sixth feature description set T6 is used to describe the working conditions of the working condition transition type. The sixth feature description set T6 includes 5 reference description information, namely F1, F2, F3, F4 and F5.

[0145] This application embodiment generates a working condition feature set through voltage sampling data. The multi-dimensional working condition feature set is used to jointly constrain and determine the target working condition type of the electrostatic chuck. This approach is accurate and reliable. Furthermore, this application embodiment inputs the working condition feature set into the working condition identification model, which can automatically identify the target working condition type of the electrostatic chuck without manual intervention, thus improving the efficiency of working condition identification.

[0146] Specifically, in order to obtain the target operating condition type of the electrostatic chuck, this embodiment of the application refines step S223 through steps S2231 and S2232, inputting the operating condition feature set into the operating condition identification model, so that the operating condition identification model performs an operating condition identification operation based on the operating condition feature set to obtain the target operating condition type of the electrostatic chuck, as shown below:

[0147] Step S2231: Determine the feature description set that matches the working condition feature set in the working condition query table.

[0148] The operating condition feature set includes a variety of operating condition features. For example, in this application embodiment, voltage change rate, voltage change rate variation, voltage residual, voltage residual variance, and voltage residual entropy are obtained based on voltage sampling data. The operating condition feature is any one of voltage change rate, voltage change rate variation, voltage residual, voltage residual variance, and voltage residual entropy.

[0149] In some embodiments, the operating condition lookup table is configured with multiple reference value ranges. Determining the feature description set that matches the operating condition feature set in the operating condition lookup table includes the following steps: finding the reference value range corresponding to each operating condition feature in the operating condition lookup table, combining the reference value ranges of all operating condition features to obtain a real-time range set, and traversing the operating condition lookup table to find the feature description set that matches the real-time range set.

[0150] For example, regarding the rate of change of voltage, the reference value range for the voltage step type operating condition is [3000, 30000], and the reference value range for the voltage steady-state type operating condition is [0, 50]. Regarding the rate of change of voltage variation, the reference value range for the voltage steady-state type operating condition is (0, 30), and the reference value range for the voltage step type operating condition is (300, +∞).

[0151] In other embodiments, determining the feature description set that matches the operating condition feature set in the operating condition query table includes the following steps: generating feature description information based on the operating condition features, combining the feature description information of all operating condition features to obtain a change description set, and traversing the operating condition query table to find the feature description set that matches the change description set.

[0152] Feature description information is used to describe changes in operating condition characteristics.

[0153] In some embodiments, generating feature description information based on operating condition characteristics includes the following steps: Given that the operating condition characteristic is the voltage change rate, generating first feature description information based on the characteristics of the voltage change rate changing over time. As shown in Table 1, when the first feature description information W1 is that the voltage change rate of the electrostatic chuck is 5000 and lasts for 400 ms, this first feature description information W1 matches the reference description information A1 under the voltage step type. When the first feature description information W1 is that the voltage change rate of the electrostatic chuck is 30 and lasts for 20 ms, this first feature description information W1 matches the reference description information B1 under the voltage steady-state type.

[0154] In some embodiments, generating feature description information based on operating condition characteristics includes the following steps: Given that the operating condition characteristic is the rate of change of voltage change, generating second feature description information based on the characteristics of the rate of change of voltage change over time. As shown in Table 2, when the absolute value of the rate of change of voltage change of the electrostatic chuck is 400, the second feature description information W2 matches the reference description information A2 under the voltage step type, or the reference description information C2 under the arc discharge type, or the reference description information D2 under the random noise type. When the absolute value of the rate of change of voltage change of the electrostatic chuck is 10, the second feature description information W2 matches the reference description information B2 under the voltage steady-state type.

[0155] In some embodiments, generating feature description information based on operating condition characteristics includes the following steps: Given that the operating condition characteristic is voltage residual, generating third feature description information based on the time-varying characteristics of the voltage residual. As shown in Table 3, when the third feature description information W3 is the voltage residual of the electrostatic chuck at 80, this third feature description information W3 matches the reference description information A3 under the voltage step type, or matches the reference description information E3 under the normal discharge type, or matches the reference description information F3 under the operating condition transition type. When the third feature description information W3 is the voltage residual of the electrostatic chuck at 32, this third feature description information W3 matches the reference description information B3 under the voltage steady-state type, or matches the reference description information F3 under the operating condition transition type.

[0156] In some embodiments, generating feature description information based on operating condition characteristics includes the following steps: Given that the operating condition characteristic is the variance of the voltage residual, generating fourth feature description information based on the characteristics of the variation of the voltage residual variance over time. As shown in Table 4, when the fourth feature description information W4 is the variance of the voltage residual of the electrostatic chuck being 125... 2 When the fourth feature description information W4 matches the reference description information A4 under the voltage step type, or the reference description information E4 under the normal discharge type, or the reference description information F4 under the operating condition transition type. When the fourth feature description information W4 is the variance of the voltage residual of the electrostatic chuck, which is 35... 2 At that time, the fourth feature description information W4 matches the reference description information B4 under the voltage steady state type or the reference description information F4 under the operating condition transition type.

[0157] In some embodiments, generating feature description information based on operating condition characteristics includes the following steps: Given that the operating condition characteristic is the residual entropy of the voltage residual, generating fifth feature description information based on the time-varying characteristics of the residual entropy of the voltage residual. As shown in Table 5, when the fifth feature description information W5 is the residual entropy of the voltage residual of the electrostatic chuck being 0.6, this fifth feature description information W5 matches the reference description information A5 under the voltage step type, or matches the reference description information C5 under the arc discharge type, or matches the reference description information D5 under the random noise type, or matches the reference description information E5 under the normal discharge type, or matches the reference description information F5 under the operating condition transition type. When the fifth feature description information W5 is the residual entropy of the voltage residual of the electrostatic chuck being 0.12, this fifth feature description information W5 matches the reference description information B5 under the voltage steady-state type.

[0158] In this embodiment of the application, the first feature description information W1, the second feature description information W2, the third feature description information W3, the fourth feature description information W4, and the fifth feature description information W5 are combined to obtain the change description set {W1,W2,W3,W4,W5}.

[0159] In this embodiment of the application, the feature description set matching the change description set is traversed in the working condition lookup table. For example, this embodiment compares the feature description information of each working condition feature in the change description set {W1, W2, W3, W4, W5} with the baseline description information of the same working condition feature under each working condition type. If the baseline description information of all working condition features under the working condition type is consistent with the corresponding feature description information of the change description set, then this embodiment of the application determines the feature description set under that working condition type.

[0160] For example, when the first feature description information W1 is that the voltage change rate of the electrostatic chuck is 30 and lasts for 20ms, the second feature description information W2 is that the absolute value of the voltage change rate of the electrostatic chuck is 10, the third feature description information W3 is that the voltage residual of the electrostatic chuck is 32, and the fourth feature description information W4 is that the variance of the voltage residual of the electrostatic chuck is 35. 2 When the residual entropy of the voltage residual of the electrostatic chuck is 0.12, the embodiments of this application perform a matching operation in the working condition lookup table based on the above-mentioned feature description information to obtain the following multiple change description sets: {B1,B2,B3,B4,B5}, {B1,B2,B3,F4,B5}, {B1,B2,F3,B4,B5}, {B1,B2,F3,F4,B5}.

[0161] As shown in Table 6, the operating condition lookup table includes 6 feature description sets. Among them, the change description set {B1,B2,B3,B4,B5} matches the second feature description set T2. Apart from that, no other feature description set in the operating condition lookup table matches the change description sets {B1,B2,B3,F4,B5}, {B1,B2,F3,B4,B5}, or {B1,B2,F3,F4,B5}. Therefore, only the second feature description set T2 matches the change description set {B1,B2,B3,B4,B5} in the operating condition lookup table. Moreover, the operating condition type of the second feature description set T2 is the voltage steady-state type. Therefore, the voltage steady-state type is taken as the target operating condition type.

[0162] Step S2233: Set the working condition type corresponding to the feature description set as the target working condition type.

[0163] If the first feature description set of the voltage step type operating condition matches the change description set, then the embodiment of this application selects the voltage step type as the target operating condition type. Similarly, if the second feature description set of the voltage steady-state type operating condition matches the change description set, then the embodiment of this application selects the voltage steady-state type as the target operating condition type, and so on, which will not be elaborated here.

[0164] The embodiments of this application adopt a lookup table method, which can efficiently and reliably identify the target working condition type of the electrostatic chuck without consuming computing power or time to build a complex inference model network.

[0165] Step S23: Adjust the preset Kalman filter based on the target operating condition type to obtain the adjusted Kalman filter.

[0166] Kalman filters can be either standard Kalman filters or extended Kalman filters. The extended Kalman filter locally linearizes the nonlinear state and observation equations through a first-order Taylor expansion and applies the recursive estimation framework of the standard Kalman filter to achieve optimal state estimation for nonlinear Gaussian systems, making it suitable for applications involving electrostatic chucks.

[0167] The expression for the extended Kalman filter includes a prediction step and an update step. The prediction step involves Equations 6 and 7, as detailed below:

[0168] Formula Six

[0169] Formula 7

[0170] in, This is the prior estimate at time k. This is the optimal estimate (i.e., the posterior estimate) at time k-1. Let be the prior covariance at time k. Let be the posterior covariance at time k-1. Let be the process noise covariance at time k.

[0171] Prior estimate at time k For the optimal estimate system of the Kalman filter at time k based on time k-1, the uncalibrated prediction is derived through the state transition function.

[0172] The optimal estimate at time k-1 This is the optimal state estimate obtained by calibrating the Kalman filter at time k-1 by combining historical predictions and current observations.

[0173] Prior covariance at time k To measure the prior estimate at time k The covariance matrix of the error between the system and the true state.

[0174] The posterior covariance at time k To measure the optimal estimate at time k The covariance matrix of the estimation error between the system and the actual state.

[0175] Process noise covariance at time k A quantitative description of the intensity and distribution of process noise, such as process noise covariance. This includes process noise variance of voltage and process noise variance of voltage change rate.

[0176] The update steps include formulas eight, nine, and ten, as detailed below:

[0177] Formula 8

[0178] Formula Nine

[0179] Formula 10

[0180] in, The Kalman gain at time k is... This is the optimal estimate at time k. This is the prior estimate at time k. To observe the noise covariance, For the system observations at time k, Let be the posterior covariance at time k. Let be the prior covariance at time k.

[0181] System observations at time k This represents the actual data vector sampled at time k, i.e., the sampled voltage values. This is the actual input to the extended Kalman filter.

[0182] Observation noise covariance This is a quantitative description of the intensity and distribution of observation noise, which is the sum of errors in the entire observation link. Observation noise includes analog-to-digital sampling noise, voltage divider resistor temperature drift error, electromagnetic interference noise, buffer amplifier noise, etc.

[0183] Kalman gain at time k These are adaptive weighting coefficients used to assign the prior estimate at time k. and system observations at time k The proportion of the final optimal estimate.

[0184] The optimal estimate at time k To obtain the optimal approximation of the true state of the system, the optimal estimate at time k is... It is the core output of the extended Kalman filter (i.e., the filtered voltage sample value).

[0185] The posterior covariance at time k The optimal estimate for quantizing time k. The error.

[0186] As mentioned above, the optimal estimate is constrained by the Kalman gain. The optimal estimate, as the feedback input of the electrostatic chuck system (i.e., the filtered voltage sample value), affects the control strategy of the electrostatic chuck system. Therefore, in this embodiment, by adjusting the Kalman gain, the optimal estimate can be adjusted, thereby affecting the predetermined control strategy of the electrostatic chuck system.

[0187] The process of adjusting a preset Kalman filter based on the target operating condition type to obtain an adjusted Kalman filter includes the following steps: adjusting the Kalman gain of the Kalman filter based on the target operating condition type to obtain the adjusted Kalman gain; and adjusting the Kalman filter based on the adjusted Kalman gain to obtain the adjusted Kalman filter.

[0188] Based on the target operating condition type, the embodiments of this application adaptively adjust the Kalman gain corresponding to the target operating condition type, so that the adjusted Kalman filter can adapt to the current operating condition of the electrostatic chuck. The adjusted Kalman filter outputs accurate and reliable voltage sampling values, which helps to improve the real-time performance, accuracy and reliability of controlling the electrostatic chuck, thereby improving the process precision.

[0189] From Equations 7 and 8, we can see that the Kalman gain... Mainly affected by process noise covariance and observation noise covariance The impact of Kalman gain. Covariance of process noise There is a positive correlation, that is, under the premise that other factors remain unchanged, the process noise covariance The larger the value, the higher the Kalman gain. The larger the value, the greater the process noise covariance. The smaller the value, the higher the Kalman gain. The smaller.

[0190] Kalman gain Also with observation noise covariance A negative correlation is formed, meaning that, assuming other factors remain constant, the observed noise covariance... The larger the value, the higher the Kalman gain. The smaller the value, the lower the observation noise covariance. The smaller the value, the higher the Kalman gain. The larger.

[0191] 1) Adjustment of Kalman gain under voltage step type operating conditions.

[0192] In some embodiments, adjusting the Kalman gain of the Kalman filter based on the target operating condition type to obtain the adjusted Kalman gain includes the following steps: in response to the target operating condition type being a voltage step type, increasing the Kalman gain of the Kalman filter based on a preset first gain adjustment model to obtain the adjusted Kalman gain.

[0193] The first gain adjustment model is a model that can improve the Kalman gain under voltage step-type operating conditions. Under voltage step-type operating conditions, the driving voltage of the electrostatic chuck experiences instantaneous jumps within a short period of time. In order to improve the voltage tracking accuracy under voltage step-type operating conditions, the embodiments of this application configure the first gain adjustment model as a model with voltage change rate as input, so that the Kalman gain output by the first gain adjustment model can rapidly jump with the voltage change rate.

[0194] The first gain adjustment model is configured with a first noise covariance function and a second noise covariance function. In this embodiment, steps J1 to J4 are used to increase the Kalman gain of the Kalman filter based on the preset first gain adjustment model to obtain the adjusted Kalman gain, as detailed below:

[0195] Step J1: Obtain the voltage change rate of the electrostatic chuck.

[0196] Step J2: Determine the first process noise covariance based on the first noise covariance function and the voltage change rate.

[0197] The first noise covariance function is an increasing function with the voltage change rate as the independent variable and the process noise covariance as the dependent variable. When the voltage change rate is used as the input to the first noise covariance function, the first noise covariance function increases the process noise covariance and outputs the increased first process noise covariance.

[0198] Step J3: Determine the first observation noise covariance based on the second noise covariance function and the voltage change rate.

[0199] The second noise covariance function is a decreasing function with the voltage change rate as the independent variable and the observation noise covariance as the dependent variable. When the voltage change rate is used as the input to the second noise covariance function, the second noise covariance function reduces the observation noise covariance and outputs the reduced first observation noise covariance.

[0200] Step J4: Based on the first process noise covariance and the first observation noise covariance, adjust the Kalman gain of the Kalman filter to obtain the adjusted Kalman gain.

[0201] This application embodiment obtains the adjusted Kalman gain by combining the first process noise covariance and the first observation noise covariance using Formulas 7 and 8. For example, this application embodiment provides the following formulas to illustrate the process of adjusting the Kalman gain, as follows:

[0202] Formula Eleven

[0203] Formula 12

[0204] Formula Thirteen

[0205] The noise covariance during the first iteration of the Kalman filter. , , , All are custom coefficients. The first candidate Kalman gain at time k is obtained by the Kalman filter based on the first process noise covariance and the first observation noise covariance. For the first process noise covariance, The first observation noise covariance, For Kalman gain, Let be the rate of change of voltage at time k. The observation noise covariance for the first iteration of the Kalman filter, where, Greater than or equal to the natural number 1.

[0206] As can be seen from Formula 11, the process noise covariance in this embodiment is adjusted according to the voltage change rate. The larger the voltage change rate, the larger the first process noise covariance is obtained, and the smaller the voltage change rate, the smaller the first process noise covariance is obtained.

[0207] As can be seen from Formula 12, the observation noise covariance in this embodiment is adjusted according to the voltage change rate. The larger the voltage change rate, the smaller the first observation noise covariance is obtained, and the smaller the voltage change rate, the larger the first observation noise covariance is obtained.

[0208] As can be seen from Formulas 7 and 8, when the noise covariance of the first process increases and the noise covariance of the first observation decreases, the first candidate Kalman gain is larger and the first candidate Kalman gain increases faster, which is conducive to accelerating the improvement of Kalman gain, so that the filtered driving voltage can quickly respond to voltage step type operating conditions.

[0209] It is understood that, in some embodiments, the embodiments of this application may directly use the first candidate Kalman gain at time k. The Kalman gain is set to the adjusted value. In other embodiments, such as Equation 13, to improve response speed, the embodiments of this application require a rapid increase in the Kalman gain. Therefore, the embodiments of this application may combine coefficients... Gain of the first candidate Kalman Further improvements are made to obtain the adjusted Kalman gain.

[0210] As mentioned earlier, under voltage step-type operating conditions, the driving voltage of the electrostatic chuck experiences a transient jump within a short period of time. The embodiments of this application not only improve the process noise covariance but also simultaneously reduce the observation noise covariance, and further enhance the fusion coefficients. In order to quickly improve the Kalman gain. As can be seen from Formula 9, after improving the Kalman gain in this embodiment, it is equivalent to allocating more weight to the actual sampled observation data (i.e., By reducing the weight of prior estimates, the embodiments of this application place greater trust in the voltage sampling data obtained from actual sampling. This enables the filtered voltage sampling data to track the step changes of the driving voltage in a timely manner, approximating the actual driving voltage to the greatest extent possible, and improving the response speed and real-time performance of the electrostatic chuck control system.

[0211] 2) Adjustment of Kalman gain under steady-state voltage conditions.

[0212] In some embodiments, adjusting the Kalman gain of the Kalman filter based on the target operating condition type to obtain the adjusted Kalman gain includes the following steps: in response to the target operating condition type being a voltage steady-state type, reducing the Kalman gain of the Kalman filter based on a preset second gain adjustment model to obtain the adjusted Kalman gain.

[0213] The second gain adjustment model is a model that can reduce the Kalman gain under steady-state voltage conditions. Under steady-state voltage conditions, the driving voltage of the electrostatic chuck is relatively stable. In order to improve the voltage feedback accuracy and noise reduction capability under steady-state voltage conditions, this embodiment configures the second gain adjustment model as a model with steady-state voltage value as input, so that the Kalman gain output by the second gain adjustment model becomes stable due to the constraint of the steady-state voltage value.

[0214] The second gain adjustment model is configured with a third noise covariance function and a fourth noise covariance function. In this embodiment, through steps L1 to L4, the Kalman gain of the Kalman filter is reduced based on the preset second gain adjustment model to obtain the adjusted Kalman gain, as shown below:

[0215] Step L1: Obtain the steady-state voltage value of the electrostatic chuck.

[0216] The steady-state voltage value reflects the stability of the voltage sampling data. For example, in this embodiment, the steady-state voltage value is obtained according to Formula Fourteen, as shown below:

[0217] Formula Fourteen

[0218] Let be the steady-state voltage value at time k. Let be the voltage residual at time i. Let be the voltage residual at time i-1. For custom coefficients, N is the total number of voltage residuals captured in a sliding statistical window (i.e., the window length), k is the position index of the current sliding statistical window, and i is the index of a single voltage residual within the current sliding statistical window.

[0219] Step L2: Determine the second process noise covariance based on the third noise covariance function and the steady-state voltage value.

[0220] The third noise covariance function is a decreasing function with the steady-state voltage value as the independent variable and the process noise covariance as the dependent variable. When the steady-state voltage value is used as the input to the third noise covariance function, the function reduces the process noise covariance and outputs the reduced second process noise covariance.

[0221] Step L3: Determine the second observation noise covariance based on the fourth noise covariance function and the steady-state voltage value.

[0222] The fourth noise covariance function is an increasing function with the steady-state voltage value as the independent variable and the observed noise covariance as the dependent variable. When the steady-state voltage value is used as the input to the fourth noise covariance function, the function increases the observed noise covariance and outputs the increased second observed noise covariance.

[0223] Step L4: Based on the second process noise covariance and the second observation noise covariance, adjust the Kalman gain of the Kalman filter to obtain the adjusted Kalman gain.

[0224] This application embodiment obtains the adjusted Kalman gain by combining the second process noise covariance and the second observation noise covariance using formulas seven and eight. For example, this application embodiment provides the following formulas to illustrate the process of adjusting the Kalman gain, as follows:

[0225] Formula Fifteen

[0226] Formula Sixteen

[0227] Formula 17

[0228] The noise covariance during the first iteration of the Kalman filter. , , , All are custom coefficients. The second candidate Kalman gain at time k is obtained by the Kalman filter based on the second process noise covariance and the second observation noise covariance. For the noise covariance of the second process, For the second observation noise covariance, For Kalman gain, Let be the steady-state voltage value at time k. The observation noise covariance for the first iteration of the Kalman filter.

[0229] As can be seen from Formula 15, the embodiment of this application adjusts the process noise covariance according to the steady-state voltage value. The larger the steady-state voltage value, the smaller the obtained second process noise covariance; the smaller the steady-state voltage value, the larger the obtained second process noise covariance.

[0230] As can be seen from Formula 16, in this embodiment of the application, the observation noise covariance is adjusted according to the steady-state voltage value. The larger the steady-state voltage value, the larger the second observation noise covariance is obtained; the smaller the steady-state voltage value, the smaller the second observation noise covariance is obtained.

[0231] In this embodiment, the second candidate Kalman gain can be directly set to the adjusted Kalman gain. As can be seen from Formulas 7 and 8, when the second process noise covariance decreases and the second observation noise covariance increases, the smaller the second candidate Kalman gain is and the faster the second candidate Kalman gain decreases, which is beneficial to accelerate the reduction of Kalman gain, so that the filtered driving voltage can quickly and accurately reflect the voltage change under the steady-state voltage condition.

[0232] It is understood that, in some embodiments, the embodiments of this application may directly use the second candidate Kalman gain at time k. The Kalman gain is set to the adjusted value. In other embodiments, such as Equation 17, to improve noise reduction capability and stability, the embodiments of this application require a rapid reduction of the Kalman gain. Therefore, the embodiments of this application can combine coefficients... Correct candidate Kalman gain Further reductions are made to obtain the adjusted Kalman gain.

[0233] As mentioned earlier, under electrically steady-state operating conditions, the driving voltage of the electrostatic chuck is relatively stable. The embodiments of this application not only reduce the process noise covariance but also simultaneously increase the observation noise covariance, and further improve the fusion coefficient. In order to quickly reduce the Kalman gain. As can be seen from Formula 9, reducing the Kalman gain in this embodiment is equivalent to reducing the amount of observation data sampled from the actual data (i.e., ...). By adjusting the weight allocation of the prior estimate, the proportion of the prior estimate in the optimal estimate is increased. In this case, the prior estimate is trusted more in the embodiments of this application. This can filter out the interference of noise on the real driving voltage, so that the filtered voltage sampling data can approximate the real driving voltage to the greatest extent and improve the stability of the electrostatic chuck control system in controlling the electrostatic chuck.

[0234] 3) Adjustment of Kalman gain under arc discharge conditions.

[0235] In some embodiments, adjusting the Kalman gain of the Kalman filter based on the target operating condition type to obtain the adjusted Kalman gain includes the following steps: in response to the target operating condition type being an arc discharge type, setting the Kalman gain of the Kalman filter to a first gain value based on a preset third gain adjustment model to obtain the adjusted Kalman gain.

[0236] The third gain adjustment model is a model that can gradually reduce the Kalman gain under arc discharge conditions. Under arc discharge conditions, the driving voltage of the electrostatic chuck exhibits clustered spikes and a sparse distribution. In order to improve the noise reduction capability under arc discharge conditions, the embodiments of this application configure the third gain adjustment model to gradually adjust the Kalman gain to 0.

[0237] The third gain adjustment model is configured with a fifth noise covariance function and a sixth noise covariance function. In this embodiment, through steps M1 to M3, the Kalman gain of the Kalman filter is set to a first gain value based on the preset third gain adjustment model to obtain the adjusted Kalman gain, as shown below:

[0238] Step M1: Determine the noise covariance of the third process based on the fifth noise covariance function.

[0239] Step M2: Determine the third observation noise covariance based on the sixth noise covariance function.

[0240] Step M3: Based on the third process noise covariance and the third observation noise covariance, the Kalman gain of the Kalman filter is adjusted in the direction of approaching the preset minimum value to obtain the first gain value, which is used as the adjusted Kalman gain.

[0241] The preset minimum value is customized by the designer based on engineering experience; for example, the preset minimum value is 0.

[0242] This application embodiment obtains the adjusted Kalman gain by combining the third process noise covariance and the third observation noise covariance using Formulas 7 and 8. For example, this application embodiment provides the following formulas to illustrate the process of adjusting the Kalman gain, as follows:

[0243] Formula 18

[0244] Formula 19

[0245] Formula 20

[0246] The noise covariance during the first iteration of the Kalman filter. , All are custom coefficients, among which, For very small coefficients, for example It can be 0.01 or 0.1, etc. For very large coefficients, for example For example, 10000 or 15000. For the noise covariance of the third process, For the third observation noise covariance, For Kalman gain.

[0247] As can be seen from Formula 18, the embodiments of this application utilize a very small coefficient. By combining the process noise covariance of the first iteration operation to minimize the process noise covariance, the obtained third process noise covariance is very small, which is beneficial to making the Kalman gain very small.

[0248] As can be seen from Formula 19, the embodiments of this application utilize a very large coefficient. By combining the observation noise covariance of the first iteration operation to maximize the observation noise covariance, the obtained third observation noise covariance is very large, which further makes the Kalman gain smaller, and finally makes the adjusted Kalman gain approach 0, that is, the first gain value is 0 or close to 0.

[0249] As can be seen from Formulas 7 and 8, when the noise covariance of the third process decreases and the noise covariance of the third observation increases, the embodiment of this application obtains a smaller Kalman gain (i.e., the first gain value). The smaller Kalman gain can help the Kalman filter suppress noise and improve the voltage filtering effect.

[0250] Under arc discharge conditions, the electrostatic chuck is in an arc discharge state, and the driving voltage of the electrostatic chuck exhibits clustered spikes and a sparse distribution. Therefore, the driving voltage is severely affected by noise interference. To improve noise reduction capability, this embodiment not only reduces the process noise covariance but also simultaneously increases the observation noise covariance to quickly reduce the Kalman gain. According to Equation Nine, reducing the Kalman gain in this embodiment is equivalent to reducing the amount of observation data sampled from the actual data (i.e.,...). By adjusting the weight allocation of the prior estimate, the proportion of the prior estimate in the optimal estimate is increased. In this case, the prior estimate is trusted more in the embodiments of this application. This can filter out the interference of noise on the real driving voltage, so that the filtered voltage sampling data can approximate the real driving voltage to the greatest extent and improve the stability of the electrostatic chuck control system in controlling the electrostatic chuck.

[0251] 4) Adjustment of Kalman gain under random noise conditions.

[0252] In some embodiments, adjusting the Kalman gain of the Kalman filter based on the target operating condition type to obtain the adjusted Kalman gain includes the following steps: in response to the target operating condition type being random noise, adjusting the Kalman gain of the Kalman filter based on a preset fourth gain adjustment model to obtain the adjusted Kalman gain.

[0253] The fourth gain adjustment model is a model that can rapidly reduce the Kalman gain under random noise conditions. Under random noise conditions, the driving voltage of the electrostatic chuck exhibits discontinuous, isolated spikes. In order to improve noise reduction capability under random noise conditions, this embodiment configures the fourth gain adjustment model to rapidly adjust the Kalman gain to 0.

[0254] The fourth gain adjustment model is configured with a seventh noise covariance function. In this embodiment, through steps X1 and X2, the Kalman gain of the Kalman filter is adjusted based on the preset fourth gain adjustment model to obtain the adjusted Kalman gain, as shown below:

[0255] Step X1: Determine the fourth observation noise covariance based on the seventh noise covariance function.

[0256] The seventh noise covariance function is a function of the observation noise covariance of the first iteration operation and the preset maxima coefficient. The preset maxima coefficient can cause the fourth observation noise covariance output by the seventh noise covariance function to tend to the maximum observation noise covariance. The Kalman gain is close to or equal to 0 under the influence of the maximum observation noise covariance. Therefore, the preset maxima coefficient can be regarded as an infinitely large positive coefficient.

[0257] Step X2: Based on the fourth observation noise covariance, control the Kalman gain of the Kalman filter to adjust to a preset minimum value to obtain the adjusted Kalman gain.

[0258] It is understood that, in the embodiments of this application, the Kalman gain of the Kalman filter can be adjusted to a preset minimum value based on the fourth process noise covariance and the fourth observation noise covariance. The fourth process noise covariance is the process noise covariance at time k obtained according to the normal iterative update method.

[0259] The fourth observation noise covariance is very large, sufficient to offset the influence of the fourth process noise covariance obtained normally on the Kalman gain. Therefore, the Kalman gain is mainly affected by the fourth observation noise covariance. Specifically, in this embodiment, the adjusted Kalman gain is obtained by combining the fourth observation noise covariance with formulas 20 and 22. For example, this embodiment provides the following formulas to illustrate the process of adjusting the Kalman gain, as follows:

[0260] Formula 21

[0261] Formula 22

[0262] The observation noise covariance for the first iteration of the Kalman filter. To preset the maximum coefficient, For the fourth observation noise covariance, For Kalman gain.

[0263] As can be seen from Formula 21, the embodiments of this application utilize a very large preset maximum coefficient. By combining the observation noise covariance of the first iteration operation to maximize the observation noise covariance, the obtained fourth observation noise covariance is very large, which is beneficial to make the Kalman gain become very large instantaneously.

[0264] As can be seen from Formula 22, the embodiments of this application utilize a very large preset maximum coefficient. The observation noise covariance is maximized by combining the observation noise covariance of the first iteration operation, so that the Kalman gain approaches or equals 0 instantaneously.

[0265] Under random noise conditions, the driving voltage of the electrostatic chuck is affected by random noise. To improve noise reduction capability, this embodiment can instantly and significantly increase the fourth observation noise covariance, rapidly adjusting the Kalman gain to 0 or close to 0. According to Formula Nine, reducing the Kalman gain in this embodiment is equivalent to reducing the amount of observation data actually sampled (i.e.,...). By adjusting the weight allocation of the prior estimate, the proportion of the prior estimate in the optimal estimate is increased. In this case, the prior estimate is trusted more in the embodiments of this application. This can filter out the interference of noise on the real driving voltage, so that the filtered voltage sampling data can approximate the real driving voltage to the greatest extent and improve the stability of the electrostatic chuck control system in controlling the electrostatic chuck.

[0266] 5) Adjustment of Kalman gain under normal discharge conditions.

[0267] In some embodiments, adjusting the Kalman gain of the Kalman filter based on the target operating condition type to obtain the adjusted Kalman gain includes the following steps: in response to the target operating condition type being a normal discharge type, increasing the Kalman gain of the Kalman filter based on a preset fifth gain adjustment model to obtain the adjusted Kalman gain.

[0268] The fifth gain adjustment model is a model that can improve the Kalman gain under normal discharge conditions. Under normal discharge conditions, the driving voltage of the electrostatic chuck gradually decreases. In order to improve the voltage tracking accuracy under normal discharge conditions, this embodiment configures the fifth gain adjustment model as a model with the voltage change rate as input, so that the Kalman gain output by the fifth gain adjustment model can quickly jump with the voltage change rate.

[0269] The fifth gain adjustment model is configured with an eighth noise covariance function and a ninth noise covariance function. In this embodiment, steps Y1 to Y4 are used to increase the Kalman gain of the Kalman filter based on the preset fifth gain adjustment model, resulting in the adjusted Kalman gain, as detailed below:

[0270] Step Y1: Obtain the voltage change rate of the electrostatic chuck.

[0271] Step Y2: Determine the fifth process noise covariance based on the eighth noise covariance function and the voltage change rate.

[0272] The eighth noise covariance function is an increasing function with the voltage change rate as the independent variable and the process noise covariance as the dependent variable. When the voltage change rate is used as the input to the eighth noise covariance function, the function increases the process noise covariance and outputs the increased fifth process noise covariance.

[0273] Step Y3: Determine the fifth observation noise covariance based on the ninth noise covariance function.

[0274] The ninth noise covariance function is a function of the observation noise covariance of the first iteration operation.

[0275] Step Y4: Based on the fifth process noise covariance and the fifth observation noise covariance, adjust the Kalman gain of the Kalman filter to obtain the adjusted Kalman gain.

[0276] This application embodiment obtains the adjusted Kalman gain by combining the fifth process noise covariance and the fifth observation noise covariance using formulas seven and eight. For example, this application embodiment provides the following formulas to illustrate the process of adjusting the Kalman gain, as follows:

[0277] Formula 23

[0278] Formula 24

[0279] Formula 25

[0280] , All are custom coefficients. Let K be the fifth candidate Kalman gain at time k, obtained by the Kalman filter based on the fifth process noise covariance and the fifth observation noise covariance. For the noise covariance of the fifth process, For the fifth observation noise covariance, For Kalman gain, Let be the rate of change of voltage at time k. The observation noise covariance for the first iteration of the Kalman filter.

[0281] As can be seen from Formula 23, since the driving voltage of the electrostatic chuck gradually decreases in the discharge state, the faster the driving voltage decreases, the greater the noise caused, and the slower the decrease, the smaller the noise caused. Therefore, in this embodiment, the noise covariance of the process is adjusted according to the voltage change rate. The larger the voltage change rate, the larger the fifth process noise covariance is obtained, and the smaller the voltage change rate, the smaller the fifth process noise covariance is obtained.

[0282] As shown in Equation 24, the decrease in the driving voltage of the electrostatic chuck in the discharge state is gradual. To improve the smoothness and stability of noise reduction, the embodiments of this application can directly use the observation noise covariance of the first iteration operation. Set as the fifth observation noise covariance .

[0283] As can be seen from Formulas 7 and 8, when the noise covariance of the fifth process increases, the fifth candidate Kalman gain follows the increase of the noise covariance of the fifth process. In this embodiment, the fifth candidate Kalman gain can be directly set to the adjusted Kalman gain, which is beneficial to make the filtered driving voltage accurately and reliably reflect the voltage change under the normal discharge type of working condition.

[0284] Under normal discharge conditions, the driving voltage of the electrostatic chuck gradually decreases. The embodiment of this application, by following the voltage change rate and outputting a larger fifth-process noise covariance, can improve the Kalman gain. According to Equation Nine, improving the Kalman gain in this embodiment is equivalent to allocating more weight to the actual sampled observation data (i.e.,...). By reducing the weight of prior estimates, the embodiments of this application place greater trust in the voltage sampling data obtained from actual sampling. This enables the filtered voltage sampling data to track the changes in the driving voltage during the discharge process in a timely manner, approximating the actual driving voltage to the greatest extent possible, and improving the response speed and real-time performance of the electrostatic chuck control system.

[0285] 6) Adjustment of Kalman gain under transient operating conditions.

[0286] In some embodiments, adjusting the Kalman gain of the Kalman filter based on the target operating condition type to obtain the adjusted Kalman gain includes the following steps: in response to the target operating condition type being an operating condition transition type, increasing the Kalman gain of the Kalman filter based on a preset sixth gain adjustment model to obtain the adjusted Kalman gain.

[0287] The sixth gain adjustment model is designed to improve the Kalman gain under transient operating conditions. Under transient operating conditions, the drive voltage of the electrostatic chuck is between the step-state drive voltage and the steady-state drive voltage. To improve voltage tracking accuracy under transient operating conditions, embodiments of this application control the sixth gain adjustment model to increase the Kalman gain.

[0288] The sixth gain adjustment model is configured with a tenth noise covariance function and an eleventh noise covariance function. In this embodiment, steps Z1 to Z3 are used to increase the Kalman gain of the Kalman filter based on the preset sixth gain adjustment model to obtain the adjusted Kalman gain, as detailed below:

[0289] Step Z1: Determine the noise covariance of the sixth process based on the tenth noise covariance function.

[0290] The tenth noise covariance function is a function of the process noise covariance of the first iteration operation and the first transition coefficient, where the first transition coefficient is greater than 1.

[0291] Step Z2: Determine the sixth observation noise covariance based on the eleventh noise covariance function.

[0292] The eleventh noise covariance function is a function of the observation noise covariance of the first iteration and the second transition coefficient, where the second transition coefficient is greater than 1.

[0293] Step Z3: Based on the sixth process noise covariance and the sixth observation noise covariance, adjust the Kalman gain of the Kalman filter to obtain the adjusted Kalman gain.

[0294] This application embodiment obtains the adjusted Kalman gain by combining the noise covariance of the sixth process and the noise covariance of the sixth observation using formulas seven and eight. For example, this application embodiment provides the following formulas to illustrate the process of adjusting the Kalman gain, as follows:

[0295] Formula 26

[0296] Formula 27

[0297] Formula 28

[0298] The first transition coefficient, This is the second transition coefficient. The sixth candidate Kalman gain at time k is obtained by the Kalman filter based on the sixth process noise covariance and the sixth observation noise covariance. For the noise covariance of the sixth process, The sixth observation noise covariance, For Kalman gain, The observation noise covariance for the first iteration of the Kalman filter. The noise covariance during the first iteration of the Kalman filter.

[0299] As shown in Formula 26, under transitional operating conditions, the driving voltage of the electrostatic chuck can range from a smaller value to a larger value, or it can decrease from a larger value to a smaller value. To improve voltage tracking accuracy, this embodiment requires a larger sixth-process noise covariance to increase the Kalman gain.

[0300] As can be seen from Formula 27, in order to obtain a higher Kalman gain, the embodiments of this application require a smaller sixth observation noise covariance.

[0301] As can be seen from Formulas 7 and 8, when the noise covariance of the sixth process increases and the noise covariance of the sixth observation decreases, the sixth candidate Kalman gain increases. In this embodiment, the sixth candidate Kalman gain can be directly set to the adjusted Kalman gain, and thus the adjusted Kalman gain increases accordingly.

[0302] Under transitional operating conditions, the change in driving voltage is continuous and gradual, without steep jumps or other fluctuations, making it relatively controllable. Combining formulas 26 and 27, it can be seen that this embodiment does not require adjusting the sixth process noise covariance or the sixth observation noise covariance based on the voltage change rate. It only needs to configure a fixed first transition coefficient according to formula 26 to increase the sixth process noise covariance and a fixed second transition coefficient to decrease the sixth observation noise covariance. This allows for a fixed increase in the Kalman gain, ensuring that the change pattern of the filtered driving voltage is consistent with or similar to the change pattern of the actual driving voltage, thus improving voltage tracking accuracy.

[0303] As can be seen from Formula 9, after increasing the Kalman gain in the embodiments of this application, it is equivalent to allocating more weight to the actual sampled observation data (i.e., By reducing the weight of prior estimates, the embodiments of this application place greater trust in the voltage sampling data obtained from actual sampling. This enables the filtered voltage sampling data to track the transition changes of the driving voltage in a timely manner, approximating the actual driving voltage to the greatest extent possible, and improving the response speed and real-time performance of the electrostatic chuck control system.

[0304] Step S24: Filter the voltage sampling data based on the adjusted Kalman filter.

[0305] When the Kalman gain of the Kalman filter is adjusted, the Kalman filter can filter the voltage sampling data based on the adjusted Kalman gain, thereby obtaining accurate and reliable voltage sampling data, which is beneficial to improving the control accuracy, reliability and real-time performance of the electrostatic chuck.

[0306] To gain a deeper understanding of the differences between the embodiments of this application and related technologies, the embodiments of this application are combined with... Figure 5 and Figure 6 A detailed explanation is provided below:

[0307] Please see Figure 5 The relevant technology uses fixed filtering parameters to filter the voltage sampling data, where both the process noise covariance and the observation noise covariance remain unchanged. For example... Figure 5As shown, under voltage step-type operating conditions, although related technologies use a Kalman filter to filter the original voltage sampling data (as shown in the first curve S1) to obtain filtered voltage sampling data (as shown in the second curve S2), for ease of identification, in Figure 5 In the diagram, the first curve S1 is the orange curve, and the second curve S2 is the blue curve.

[0308] like Figure 5 As shown, the second curve S2 does not effectively track the first curve S1. There is a large "disengagement region 51" between the second curve S2 and the first curve S1. Therefore, the Kalman filter provided by the related technology has limited response capability to rapidly changing signals, resulting in a significant output lag.

[0309] Please see Figure 6 In this embodiment, the operating condition of the electrostatic chuck is automatically identified as a voltage step type condition through the operating condition feature set. It then immediately increases the process noise covariance Q(k) and decreases the observation noise covariance R(k), thereby improving the Kalman gain K(k). Under this adaptive mechanism, the Kalman filter rapidly increases its confidence in the actual measurement data, achieving millisecond-level tracking and avoiding the response lag problem in related technologies. Figure 6 As shown, the third curve S3 can accurately track the first curve S1 during the step phase and maintain a stable value with low volatility during the steady-state phase.

[0310] Please continue reading. Figure 6 When the driving voltage reaches a steady state, the residual r(k) drops to a low level, and both the variance ρ(k) and the voltage change rate of the residual are close to zero, while the residual entropy H(k) increases, exhibiting typical steady-state Gaussian noise characteristics. This embodiment again determines the operating condition of the electrostatic chuck as a voltage steady-state condition based on a new set of operating condition characteristics. This allows for adaptive reduction of the process noise covariance Q(k) and increase of the observation noise covariance R(k), thereby reducing the Kalman gain K(k) and enabling the Kalman filter to suppress noise. This embodiment significantly reduces the impact of high-frequency glitches, sampling jitter, and plasma background noise, resulting in a smoothness of the steady-state voltage curve exceeding 80%, and a substantial improvement in measurement accuracy and signal-to-noise ratio.

[0311] When voltage sampling data exhibits isolated spikes (such as spikes caused by interference), the residuals instantaneously exceed three times the standard deviation, leading to an immediate increase in the proportion of large residuals. This application's embodiment automatically identifies the operating condition of the electrostatic chuck as a random noise type based on the operating condition feature set, rapidly amplifying the observed noise covariance to quickly bring the Kalman gain K(k) close to 0, thus achieving instantaneous suppression of outliers.

[0312] For operating conditions such as arc discharge and other "clustered spikes", the residual entropy decreases significantly, the proportion of large residuals is high, and the residuals continuously exceed the threshold. The embodiments of this application will significantly amplify the observation noise covariance R(k) and increase the process noise covariance Q(k), which can improve the robustness of the Kalman filter and ensure that the Kalman filter will not diverge due to continuous anomalies.

[0313] This application embodiment performs real-time analysis using mathematical characteristics such as voltage change rate, rate of change of voltage change rate, voltage residual, voltage residual variance, and voltage residual entropy. It can automatically identify various operating conditions of the electrostatic chuck, including voltage step, steady-state maintenance, discharge decay, random noise glitches, arc discharge, and transient states. Based on these conditions, it dynamically adjusts the process noise covariance Q(k), observation noise covariance R(k), and Kalman gain K(k) of the Kalman filter. Therefore, the electrostatic chuck exhibits millisecond-level response capabilities during start-up, shutdown, polarity switching, or rapid changes; achieves strong noise suppression and high-precision output in the steady-state phase; and maintains robustness and stability under noise and arc interference. Ultimately, it simultaneously meets the dual requirements of "real-time performance" and "steady-state accuracy" for electrostatic chucks, resolving the technical contradiction that related technologies cannot simultaneously achieve both.

[0314] In summary, the embodiments of this application achieve at least the following technical effects:

[0315] 1) Real-time analysis is performed using mathematical characteristics such as voltage change rate, rate of change of voltage change rate, voltage residual, voltage residual variance, and voltage residual entropy to accurately identify different operating conditions of the electrostatic chuck.

[0316] 2) During the voltage ramp-up phase or the setting change phase, the embodiments of this application can track the voltage changes in the above two phases with a millisecond-level response, and quickly reflect the real driving voltage.

[0317] 3) Under steady-state voltage conditions, the embodiments of this application can automatically smooth voltage sampling data, significantly suppress noise and glitches, and improve output stability.

[0318] 4) It has strong anti-interference ability against sudden spikes and outliers, and avoids abnormal filtering state or divergence caused by abnormal data.

[0319] 5) Maintain high-precision estimation capability under normal data conditions without sacrificing filtering performance.

[0320] It should be noted that in the above embodiments, there is no necessarily a certain order between the steps. Those skilled in the art can understand from the description of the embodiments of this application that the above steps may have different execution orders in different embodiments, that is, they may be executed in parallel or in turn, etc.

[0321] As another aspect of the embodiments of this application, this application provides an electrostatic chuck voltage filtering device. The electrostatic chuck voltage filtering device can be a software module, which includes several instructions stored in a memory. A processor can access the memory, call the instructions, and execute them to complete the electrostatic chuck voltage filtering method described in the above embodiments.

[0322] In some embodiments, the electrostatic chuck voltage filtering device can also be constructed from hardware components. For example, the electrostatic chuck voltage filtering device can be constructed from one or more chips, which can work in coordination to complete the electrostatic chuck voltage filtering method described in the various embodiments above. As another example, the electrostatic chuck voltage filtering device can also be constructed from various logic devices, such as general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), microcontrollers, ARM (Acorn RISC Machine) or other programmable logic devices, discrete gate or transistor logic, discrete hardware components, or any combination of these components.

[0323] Please see Figure 7 The electrostatic chuck voltage filtering device 700 includes a data acquisition module 71, a working condition determination module 72, a filter adjustment module 73, and a filter processing module 74.

[0324] The data acquisition module 71 is used to acquire voltage sampling data of the electrostatic chuck, the operating condition determination module 72 is used to determine the target operating condition type of the electrostatic chuck based on the voltage sampling data, the filter adjustment module 73 is used to adjust the preset Kalman filter based on the target operating condition type to obtain the adjusted Kalman filter, and the filter processing module 74 is used to filter the voltage sampling data based on the adjusted Kalman filter.

[0325] In some embodiments, the operating condition determination module 72 is specifically used to: determine an operating condition feature set based on voltage sampling data, wherein the operating condition feature set is used to represent the operating condition of the electrostatic chuck, obtain a preset operating condition identification model, input the operating condition feature set into the operating condition identification model, so that the operating condition identification model performs an operating condition identification operation based on the operating condition feature set to obtain the target operating condition type of the operating condition of the electrostatic chuck.

[0326] In some embodiments, the working condition identification model is configured with a working condition query table, which includes multiple working condition types and a feature description set corresponding to each working condition type. The working condition determination module 72 is further specifically used to: determine the feature description set that matches the working condition feature set in the working condition query table, and set the working condition type corresponding to the feature description set as the target working condition type.

[0327] In some embodiments, the filter adjustment module 73 is specifically used to: adjust the Kalman gain of the Kalman filter based on the target operating condition type to obtain the adjusted Kalman gain, and adjust the Kalman filter based on the adjusted Kalman gain to obtain the adjusted Kalman filter.

[0328] In some embodiments, the filter adjustment module 73 is further specifically configured to: increase the Kalman gain of the Kalman filter based on a preset first gain adjustment model in response to the target operating condition type being a voltage step type, to obtain an adjusted Kalman gain; decrease the Kalman gain of the Kalman filter based on a preset second gain adjustment model in response to the target operating condition type being a voltage steady-state type, to obtain an adjusted Kalman gain; set the Kalman gain of the Kalman filter to a first gain value based on a preset third gain adjustment model in response to the target operating condition type being an arc discharge type, to obtain an adjusted Kalman gain; and set the Kalman gain of the Kalman filter to a second gain value based on a preset fourth gain adjustment model in response to the target operating condition type being a random noise type, to obtain an adjusted Kalman gain.

[0329] In some embodiments, the first gain adjustment model is configured with a first noise covariance function and a second noise covariance function. The filter adjustment module 73 is further specifically used to: obtain the voltage change rate of the electrostatic chuck; the first noise covariance function is an increasing function with the voltage change rate as the independent variable and the process noise covariance as the dependent variable; the second noise covariance function is a decreasing function with the voltage change rate as the independent variable and the observation noise covariance as the dependent variable; determine the first process noise covariance based on the first noise covariance function and the voltage change rate; determine the first observation noise covariance based on the second noise covariance function and the voltage change rate; and adjust the Kalman gain of the Kalman filter based on the first process noise covariance and the first observation noise covariance to obtain the adjusted Kalman gain.

[0330] In some embodiments, the second gain adjustment model is configured with a third noise covariance function and a fourth noise covariance function. The filter adjustment module 73 is further specifically used to: obtain the steady-state voltage value of the electrostatic chuck; the third noise covariance function is a decreasing function with the steady-state voltage value as the independent variable and the process noise covariance as the dependent variable; the fourth noise covariance function is an increasing function with the steady-state voltage value as the independent variable and the observation noise covariance as the dependent variable; determine the second process noise covariance based on the third noise covariance function and the steady-state voltage value; determine the second observation noise covariance based on the fourth noise covariance function and the steady-state voltage value; and adjust the Kalman gain of the Kalman filter based on the second process noise covariance and the second observation noise covariance to obtain the adjusted Kalman gain.

[0331] In some embodiments, the third gain adjustment model is configured with a fifth noise covariance function and a sixth noise covariance function. The filter adjustment module 73 is further specifically used to: determine the third process noise covariance based on the fifth noise covariance function, determine the third observation noise covariance based on the sixth noise covariance function, and control the Kalman gain of the Kalman filter to be adjusted in the direction of approaching a preset minimum value based on the third process noise covariance and the third observation noise covariance, so as to obtain a first gain value, and the first gain value is used as the adjusted Kalman gain.

[0332] It should be noted that the above-described electrostatic chuck voltage filtering device can execute the electrostatic chuck voltage filtering method provided in the embodiments of this application, and has the corresponding functional modules and beneficial effects of the method. Technical details not described in detail in the embodiments of the electrostatic chuck voltage filtering device can be found in the electrostatic chuck voltage filtering method provided in the embodiments of this application.

[0333] See Figure 8 , Figure 8 This is a schematic diagram of a controller provided in an embodiment of this application. The controller 15 includes one or more processors 151 and a memory 152. The memory 152 is connected to one or more processors 151, for example, via a bus.

[0334] Processor 151 is configured to support the controller in performing the corresponding functions in the methods described in the above method embodiments. The processor may be a central processing unit (CPU), a network processor (NP), a hardware chip, or any combination thereof. The aforementioned hardware chip may be an application-specific integrated circuit (ASIC), a programmable logic device (PLD), or a combination thereof. The aforementioned PLD may be a complex programmable logic device (CPLD), a field-programmable gate array (FPGA), a generic array logic (GAL), or any combination thereof.

[0335] Memory 152 is used to store program code, etc. Memory may include volatile memory (VM), such as random access memory (RAM); memory may also include non-volatile memory (NVM), such as read-only memory (ROM), flash memory, hard disk drive (HDD), or solid-state drive (SSD); memory may also include combinations of the above types of memory.

[0336] The memory 152 can be used to store non-volatile software programs, non-volatile computer-executable programs, and modules, such as the program instructions / modules corresponding to the electrostatic chuck voltage filtering method in the embodiments of this application. The processor executes the various functional applications and data processing of the electrostatic chuck voltage filtering method and electrostatic chuck voltage filtering device by running the non-volatile software programs, instructions, and modules stored in the memory, thereby realizing the functions of each module or unit of the electrostatic chuck voltage filtering method and electrostatic chuck voltage filtering device provided in the above method embodiments.

[0337] The memory 152 may include a program storage area and a data storage area, wherein the program storage area may store the operating system and application programs required for at least one function. The data storage area may store data created based on the use of the electrostatic chuck voltage filter, etc. In some embodiments, the memory may optionally include memory remotely located relative to the processor, which can be connected to the electrostatic chuck voltage filter via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.

[0338] The one or more modules are stored in the memory. When executed by the one or more processors, they perform the electrostatic chuck voltage filtering method in any of the above method embodiments. For example, they perform the method steps described in the above method embodiments to realize the functions of the modules described in the above device embodiments.

[0339] This application also provides a computer-readable storage medium storing a computer program, the computer program including program instructions, which, when executed by a controller, cause the controller to perform the method described in the foregoing embodiments.

[0340] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. The storage medium can be a magnetic disk, optical disk, read-only memory (ROM), or random access memory (RAM), etc.

[0341] The above-disclosed embodiments are merely preferred embodiments of this application and should not be construed as limiting the scope of this application. Therefore, any equivalent variations made in accordance with the claims of this application shall still fall within the scope of this application.

Claims

1. A method for filtering voltage in an electrostatic chuck, characterized in that, include: Obtain the voltage sampling data of the electrostatic chuck; The target operating condition type of the electrostatic chuck is determined based on the voltage sampling data. Adjusting the Kalman gain of the Kalman filter based on the target operating condition type to obtain the adjusted Kalman gain, wherein adjusting the Kalman gain of the Kalman filter based on the target operating condition type to obtain the adjusted Kalman gain includes: responding to the target operating condition type being a voltage step type, increasing the Kalman gain of the Kalman filter based on a preset first gain adjustment model to obtain the adjusted Kalman gain; the first gain adjustment model is configured with a first noise covariance function and a second noise covariance function, and increasing the Kalman gain of the Kalman filter based on the preset first gain adjustment model to obtain the adjusted Kalman gain includes: The voltage change rate of the electrostatic chuck is obtained. The first noise covariance function is an increasing function with the voltage change rate as the independent variable and the process noise covariance as the dependent variable; the second noise covariance function is a decreasing function with the voltage change rate as the independent variable and the observation noise covariance as the dependent variable; a first process noise covariance is determined based on the first noise covariance function and the voltage change rate; a first observation noise covariance is determined based on the second noise covariance function and the voltage change rate; the Kalman gain of the Kalman filter is adjusted based on the first process noise covariance and the first observation noise covariance to obtain the adjusted Kalman gain. The Kalman filter is adjusted based on the adjusted Kalman gain to obtain the adjusted Kalman filter; The voltage sampling data is filtered based on the adjusted Kalman filter.

2. The method according to claim 1, characterized in that, The step of determining the target operating condition type of the electrostatic chuck based on the voltage sampling data includes: A set of operating conditions is determined based on the voltage sampling data, wherein the set of operating conditions is used to represent the operating conditions of the electrostatic chuck. Obtain the preset working condition recognition model; The working condition feature set is input into the working condition identification model so that the working condition identification model performs a working condition identification operation based on the working condition feature set to obtain the target working condition type of the working condition in which the electrostatic chuck is located.

3. The method according to claim 2, characterized in that, The operating condition identification model is configured with an operating condition lookup table, which includes multiple operating condition types and a feature description set corresponding to each operating condition type. The operating condition feature set is input into the operating condition identification model so that the model performs an operating condition identification operation based on the feature set to obtain the target operating condition type of the electrostatic chuck. This includes: Determine the feature description set that matches the working condition feature set in the working condition query table; Set the operating condition type corresponding to the feature description set as the target operating condition type.

4. The method according to claim 1, characterized in that, The step of adjusting the Kalman gain of the Kalman filter based on the target operating condition type to obtain the adjusted Kalman gain further includes: In response to the target operating condition being a voltage steady-state type, the Kalman gain of the Kalman filter is reduced based on a preset second gain adjustment model to obtain the adjusted Kalman gain; In response to the target operating condition being an arc discharge type, the Kalman gain of the Kalman filter is set to the first gain value based on a preset third gain adjustment model, thus obtaining the adjusted Kalman gain. In response to the target operating condition being random noise, the Kalman gain of the Kalman filter is set to the second gain value based on a preset fourth gain adjustment model, thus obtaining the adjusted Kalman gain.

5. The method according to claim 4, characterized in that, The second gain adjustment model is configured with a third noise covariance function and a fourth noise covariance function. The step of reducing the Kalman gain of the Kalman filter based on the preset second gain adjustment model to obtain the adjusted Kalman gain includes: Obtain the steady-state voltage value of the electrostatic chuck; the third noise covariance function is a decreasing function with the steady-state voltage value as the independent variable and the process noise covariance as the dependent variable; the fourth noise covariance function is an increasing function with the steady-state voltage value as the independent variable and the observation noise covariance as the dependent variable. The second process noise covariance is determined based on the third noise covariance function and the steady-state voltage value. The second observation noise covariance is determined based on the fourth noise covariance function and the steady-state voltage value; Based on the second process noise covariance and the second observation noise covariance, the Kalman gain of the Kalman filter is adjusted to obtain the adjusted Kalman gain.

6. The method according to claim 4, characterized in that, The third gain adjustment model is configured with a fifth noise covariance function and a sixth noise covariance function. The third gain adjustment model, based on a preset value, sets the Kalman gain of the Kalman filter to a first gain value to obtain the adjusted Kalman gain, including: The noise covariance of the third process is determined based on the fifth noise covariance function. The third observation noise covariance is determined based on the sixth noise covariance function. Based on the third process noise covariance and the third observation noise covariance, the Kalman gain of the Kalman filter is adjusted in the direction of approaching a preset minimum value to obtain a first gain value, which is used as the adjusted Kalman gain.

7. A controller, characterized in that, The device includes a memory and a processor, the memory being connected to the processor, the processor being configured to execute one or more computer programs stored in the memory, and the processor, when executing the one or more computer programs, causing the controller to implement the electrostatic chuck voltage filtering method as described in any one of claims 1-6.

8. An electrostatic chuck system, characterized in that, include: electrostatic chuck; The controller as described in claim 7 is electrically connected to the electrostatic chuck and is used to control the operation of the electrostatic chuck.

Images