Method and system for obtaining polarization aberration compensation parameters, and lithographic machine

Abstract

The application provides a kind of acquisition method, system and photolithography machine of polarization aberration compensation parameter, it is related to polarization illumination technical field.The method includes obtaining the Mueller matrix sequence and the environment parameter sequence corresponding to photolithography machine system, the time sequence of the Mueller matrix sequence and the environment parameter sequence is aligned;The Mueller matrix sequence and the environment parameter sequence are respectively extracted and the features of the same time sequence are combined to obtain a feature tensor sequence;Determine the target compensation parameter according to the feature tensor sequence, the target compensation parameter is used to compensate the polarization aberration of the photolithography machine system.This method can improve the accuracy of acquiring polarization aberration compensation parameters, thereby further improving the overall compensation accuracy and imaging accuracy of the system.

Description

Technical Field

[0001] The present invention relates to the field of polarized illumination technology, and in particular to a method and system for obtaining polarization aberration compensation parameters and a photolithography machine. Background Art

[0002] In advanced lithography, improving pattern resolution and overlay accuracy are the core drivers driving the continuous scaling of integrated circuit process nodes. Currently, high numerical aperture (NA) and polarized illumination are key technologies for improving lithography tool resolution. When the NA reaches 0.85 or above, polarization aberration cannot be ignored. Polarization aberration refers to the change in the polarization state of polarized light at the exit pupil plane after passing through an optical system. It is primarily caused by the varying transmission characteristics of components within the optical system for light waves of different polarization states, such as intrinsic birefringence and stress birefringence of optical materials, and mask absorption effects. When polarization aberration is significant, it can severely reduce the contrast and resolution of lithographic patterns, particularly affecting overlay accuracy. Therefore, polarization compensation is necessary for the optical system. Currently, lithography equipment typically models the measured Mueller matrix through methods such as Fourier expansion and Zernike polynomial fitting, extracting polarization aberration parameters for subsequent system compensation.

[0003] Under actual working conditions, optical components may be affected by factors such as thermal drift and stress evolution, exhibiting significant nonlinear or time-varying behavior. The Mueller matrix is ​​essentially a linear system response model. Existing methods for obtaining compensation parameters are difficult to accurately characterize these dynamic changes, so the accuracy of the obtained polarization aberration compensation parameters is not high. Summary of the Invention

[0004] The present invention provides a method, system, and lithography machine for acquiring polarization aberration compensation parameters. These methods address the low accuracy of polarization aberration compensation in existing technologies. By fusing multi-source data consisting of a Mueller matrix and environmental parameters and extracting features to obtain a characteristic tensor sequence, the target compensation parameters are determined based on the characteristic tensor sequence. This method improves the accuracy of acquiring polarization aberration compensation parameters, thereby further enhancing the overall compensation accuracy and imaging precision of the system.

[0005] The present invention provides a method for obtaining polarization aberration compensation parameters, comprising the following steps:

[0006] Obtaining a Mueller matrix sequence and an environmental parameter sequence corresponding to a lithography system, wherein the timing of the Mueller matrix sequence and the environmental parameter sequence are aligned, the Mueller matrix sequence is used to characterize changes in polarized light transmission in the lithography system, and the environmental parameter sequence is used to characterize changes in environmental parameters that affect the stability of the lithography system;

[0007] Extracting features from the Mueller matrix sequence and the environmental parameter sequence respectively and combining features of the same time series to obtain a feature tensor sequence;

[0008] Target compensation parameters are determined according to the characteristic tensor sequence, and the target compensation parameters are used to compensate for the polarization aberration of the lithography system.

[0009] According to a method for obtaining polarization aberration compensation parameters provided by the present invention, the environmental parameter sequence includes at least one of a temperature sequence, an air pressure sequence and a vibration sequence. The temperature sequence is used to characterize the temperature change of the lithography machine system, the air pressure sequence is used to characterize the air pressure change of the lithography machine system, and the vibration sequence is used to characterize the vibration acceleration change of the moving parts and / or optical parts of the lithography machine system.

[0010] According to the acquisition of polarization aberration compensation parameters provided by the present invention, the environmental parameter sequence includes the temperature sequence, the pressure sequence, and the vibration sequence, and the feature extraction of the Mueller matrix sequence and the environmental parameter sequence is performed respectively, and the features of the same time sequence are combined to obtain a feature tensor sequence, including:

[0011] Performing principal component analysis and dimensionality reduction processing on the Mueller matrix sequence to obtain a Mueller matrix eigenvector sequence after dimensionality reduction;

[0012] performing low-pass filtering on the temperature sequence and the pressure sequence respectively to obtain a filtered temperature sequence and a filtered pressure sequence;

[0013] Performing sliding window aggregation on the vibration sequence to obtain an aggregated acceleration sequence;

[0014] The reduced-dimensional Mueller matrix eigenvector sequence, the filtered temperature sequence, the filtered pressure sequence, and the aggregated acceleration sequence are normalized, and then the features of the same time series are spliced ​​to obtain the feature tensor sequence.

[0015] According to a method for obtaining polarization aberration compensation parameters provided by the present invention, the Mueller matrix sequence includes at least one 4×4 Mueller matrix, and the principal component analysis dimensionality reduction processing is performed on the Mueller matrix sequence to obtain a Mueller matrix eigenvector sequence after dimensionality reduction, including:

[0016] Expand each 4×4 Mueller matrix in the at least one 4×4 Mueller matrix into a 16-dimensional row vector in a row-first manner to obtain at least one 16-dimensional row vector;

[0017] The at least one 16-dimensional row vector is stacked to form a data matrix, and the first K principal components of the data matrix are extracted to obtain the eigenvector sequence of the Mueller matrix after dimensionality reduction.

[0018] According to the application, the temperature sequence and the air pressure sequence correspond to a first sampling period, the temperature sequence and the air pressure sequence are respectively subjected to low-pass filtering to obtain a filtered temperature sequence and a filtered air pressure sequence, and the method comprises the following steps.

[0019] The temperature sequence is subjected to moving average filtering through a first sliding window to obtain the filtered temperature sequence.

[0020] The air pressure sequence is subjected to moving average filtering through a second sliding window to obtain the filtered air pressure sequence, wherein the time length of the first sliding window and the second sliding window is greater than or equal to two first sampling periods, and the moving average filtering comprises taking the mean value of the values of at least two sampling points in the window as the value of the current sampling point.

[0021] The second sampling period corresponding to the vibration sequence is less than the first sampling period, and a third sliding window is a non-overlapping sliding window, the vibration sequence is subjected to sliding window aggregation to obtain an aggregated acceleration sequence, and the method comprises the following steps.

[0022] The time length of the third sliding window is determined to be equal to the first sampling period.

[0023] The vibration sequence is subjected to sliding aggregation through the third sliding window to obtain processed vibration acceleration data, and the sliding aggregation comprises taking the mean value and the variance of the acceleration data in the window as the value of the current sampling point.

[0024] The processed vibration acceleration data is subjected to time sequence alignment according to the first sampling period to obtain the aggregated acceleration sequence.

[0025] According to the application, the target compensation parameter corresponding to the feature tensor sequence is determined, and the method comprises the following steps.

[0026] The feature tensor sequence is input into a trained compensation parameter prediction model to output the target compensation parameter, the trained compensation parameter prediction model is obtained by inputting a measured Mueller matrix sequence and an environmental parameter sequence into an initial compensation parameter prediction model, and optimizing the model parameters according to the mean square error between the predicted compensation parameter output by the model and the target imaging parameter.

[0027] According to the application, the trained compensation parameter prediction model comprises a long short-term memory neural network model and a full connection layer, the full connection layer is used for fusing time sequence features output by the long short-term memory neural network model and generating target compensation parameters, and the target compensation parameters comprise at least one of a polarizer angle compensation amount and a fast axis azimuth angle compensation amount of a quarter wave plate.

[0028] According to the application, the target compensation parameters comprise a polarizer angle compensation amount and a fast axis azimuth angle compensation amount of a quarter wave plate, and the method further comprises:

[0029] Adjusting a polarizer angle and a fast axis azimuth angle of a quarter wave plate in the polarization state generator according to the target compensation parameters.

[0030] According to the application, the acquisition of the polarization aberration compensation parameters comprises the following steps:

[0031] Acquiring a plurality of groups of Stokes vectors at each sampling point according to a first sampling period, and performing least square fitting on the plurality of groups of Stokes vectors to obtain a Mueller matrix corresponding to each sampling point, and the Mueller matrices corresponding to a plurality of sampling points constitute the Mueller matrix sequence;

[0032] Synchronously acquiring a temperature parameter and an air pressure parameter corresponding to each sampling point according to the first sampling period, and the temperature parameters and the air pressure parameters corresponding to a plurality of sampling points constitute the temperature sequence and the air pressure sequence;

[0033] Acquiring a plurality of vibration acceleration parameters according to a second sampling period, and the plurality of vibration acceleration parameters constitute the vibration sequence, and the second sampling period is less than the first sampling period.

[0034] The application further provides an acquisition system of polarization aberration compensation parameters, comprising the following modules:

[0035] A data acquisition module is configured to acquire a Mueller matrix sequence and an environmental parameter sequence corresponding to a lithography machine system, the time sequences of the Mueller matrix sequence and the environmental parameter sequence are aligned, the Mueller matrix sequence is used for representing changes in transmission of polarized light in the lithography machine system, and the environmental parameter sequence is used for representing changes in environmental parameters that have an impact on the stability of the lithography machine system;

[0036] A feature acquisition module is configured to extract features from the Mueller matrix sequence and the environmental parameter sequence respectively and combine features of the same time sequence to obtain a feature tensor sequence;

[0037] A parameter prediction module is configured to determine a target compensation parameter based on the feature tensor sequence, the target compensation parameter being used to compensate for polarization aberration of the lithography system.

[0038] The application further provides an electronic device, including a memory, a processor, and a computer program stored in the memory and running on the processor, and the processor implements the method for acquiring the polarization aberration compensation parameter according to any one of the above when executing the computer program.

[0039] The application further provides a non-transitory computer-readable storage medium, which stores a computer program, and the computer program is executed by a processor to implement the method for acquiring the polarization aberration compensation parameter according to any one of the above.

[0040] The application further provides a computer program product, which includes a computer program, and the computer program is executed by a processor to implement the method for acquiring the polarization aberration compensation parameter according to any one of the above.

[0041] The application provides the method and system for acquiring the polarization aberration compensation parameter and the lithography machine, the feature tensor sequence is obtained by fusing the multi-source data composed of the Mueller matrix and the environmental parameter and extracting features, and the target compensation parameter is determined according to the feature tensor sequence. The accuracy of acquiring the polarization aberration compensation parameter can be improved, and the overall compensation accuracy and imaging precision of the system are further improved. BRIEF DESCRIPTION OF DRAWINGS

[0042] In order to more clearly illustrate the technical solutions of the present application or the prior art, the following will briefly introduce the drawings needed in the embodiments or prior art description. Obviously, the drawings in the following description are some embodiments of the present application, and those skilled in the art can obtain other drawings according to these drawings without creative labor.

[0043] Figure 1 is a flowchart of the method for acquiring the polarization aberration compensation parameter provided by the application.

[0044] Figure 2 is a flowchart of the method for acquiring the Mueller matrix sequence and the environmental parameter sequence corresponding to the lithography system provided by the application.

[0045] Figure 3 is a structural diagram of the lithography system provided by the application.

[0046] Figure 4 is a flowchart of the method for acquiring the feature tensor sequence provided by the application.

[0047] Figure 5 is a schematic diagram of the feature extraction of the Mueller matrix sequence and the environmental parameter sequence provided by the application.

[0048] Figure 6 It is a schematic diagram of the overall flow of the method for obtaining polarization aberration compensation parameters provided by the present invention.

[0049] Figure 7 It is a structural schematic diagram of the polarization aberration compensation system provided by the present invention.

[0050] Figure 8 It is a schematic diagram of the physical structure of the electronic device provided by the present invention. DETAILED DESCRIPTION

[0051] To make the objectives, technical solutions, and advantages of the present invention more clear, the technical solutions of the present invention will be clearly and completely described below in conjunction with the accompanying drawings. Obviously, the embodiments described are only some of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts shall fall within the scope of protection of the present invention.

[0052] In advanced lithography, improving pattern resolution and overlay accuracy are the core drivers driving the continuous scaling of integrated circuit process nodes. Currently, high numerical aperture (NA) and polarized illumination are key technologies for improving lithography tool resolution. When the NA reaches 0.85 or above, polarization aberration cannot be ignored. Polarization aberration refers to the change in the polarization state of polarized light at the exit pupil plane after it passes through an optical system. It is primarily caused by the varying transmission characteristics of components within the optical system for light waves of different polarization states, such as intrinsic birefringence and stress birefringence of optical materials, and mask absorption effects. When polarization aberration is significant, it can severely impact the contrast and resolution of lithographic patterns, particularly affecting overlay accuracy. Therefore, polarization compensation is necessary for the optical system. Currently, lithography tools primarily use the Mueller matrix modeling method to characterize and compensate for the polarization transmission behavior of optical systems. The Mueller matrix is ​​a mathematical tool that describes the propagation and changes of polarized light in an optical system. It is a 4×4 real number matrix that represents the effect of an optical component or system on a beam of arbitrary polarization state. Because the Mueller matrix describes the linear transition from incident polarization to outgoing polarization, it can be used to model and compensate lithography illumination systems. In current applications, lithography equipment typically uses methods such as Fourier expansion and Zernike polynomial fitting to model the measured Mueller matrix and extract polarization aberration parameters for subsequent system compensation.

[0053] However, under actual working conditions, optical components may be affected by factors such as thermal drift and stress evolution, exhibiting significant nonlinear or time-varying behavior. The Mueller matrix is ​​essentially a linear system response model. Existing methods for obtaining compensation parameters are difficult to accurately characterize these dynamic changes. Therefore, the accuracy of the obtained polarization aberration compensation parameters is not high.

[0054] In light of this, an embodiment of the present invention provides a method for acquiring polarization aberration compensation parameters. The method comprises acquiring a Mueller matrix sequence and an environmental parameter sequence corresponding to a lithography system, wherein the timing of the Mueller matrix sequence and the environmental parameter sequence are aligned; performing feature extraction on each of the Mueller matrix sequence and the environmental parameter sequence, and combining features of the same timing to obtain a feature tensor sequence; and determining target compensation parameters based on the feature tensor sequence. The target compensation parameters are used to compensate for the polarization aberration of the lithography system. This method can improve the accuracy of acquiring polarization aberration compensation parameters, thereby further improving the overall compensation accuracy and imaging precision of the system.

[0055] The technical solutions in the embodiments of the present invention will be described below with reference to the accompanying drawings in the embodiments of the present invention.

[0056] Figure 1 1 is a flow chart of the method for obtaining polarization aberration compensation parameters provided by the present invention. The method for obtaining polarization aberration compensation parameters can be applied to a lithography system. Figure 1 As shown, the method may include the following steps 101 to 103:

[0057] Step 101: Obtain a Mueller matrix sequence and an environmental parameter sequence corresponding to the lithography system, wherein the timing of the Mueller matrix sequence and the environmental parameter sequence are aligned. The Mueller matrix sequence is used to characterize changes in the transmission of polarized light in the lithography system, and the environmental parameter sequence is used to characterize changes in environmental parameters that affect the stability of the lithography system.

[0058] It should be noted that there are many methods for obtaining the Mueller matrix sequence and environmental parameter sequence corresponding to the lithography system. For example, the Mueller matrix and environmental parameters can be obtained by collecting them at multiple sampling points according to a fixed sampling period, or by receiving information transmitted by other devices. The present invention does not limit the method for obtaining the Mueller matrix sequence and environmental parameter sequence corresponding to the lithography system.

[0059] The Mueller matrix can be collected by the lithography system, and the environmental parameters can be collected by a data acquisition module. The data acquisition module can include various types of sensors, and the type of sensor is determined by the environmental parameters to be collected. For example, the environmental reference may include temperature parameters, air pressure parameters, etc.

[0060] Step 102: extract features from the Mueller matrix sequence and the environmental parameter sequence respectively and combine features of the same time series to obtain a feature tensor sequence.

[0061] It should be noted that there are many ways to extract features from the Mueller matrix sequence and the environmental parameter sequence, such as filtering, aggregation, and dimensionality reduction. The method for combining the features of the same time series with the extracted features may be feature concatenation or fusion. The present invention does not limit the method for extracting features from the Mueller matrix sequence and the environmental parameter sequence and combining the features of the same time series.

[0062] Step 103: Determine target compensation parameters according to the characteristic tensor sequence, where the target compensation parameters are used to compensate for the polarization aberration of the lithography system.

[0063] It should be noted that there are many ways to determine the target compensation parameters based on the feature tensor sequence. For example, the feature tensor sequence can be input into a trained prediction model, or calculated based on the feature tensor sequence and a preset algorithm. The present invention does not limit the method for determining the target compensation parameters based on the feature tensor sequence.

[0064] It is understandable that the existing methods for obtaining polarization aberration compensation parameters are difficult to capture the nonlinear response mechanism, because the Mueller matrix is ​​essentially a linear system response model, which is suitable for modeling optical systems under steady-state conditions. However, under actual working conditions, optical elements may be affected by factors such as thermal drift and stress evolution, showing significant nonlinear or time-varying behavior, and existing modeling methods are difficult to accurately characterize these dynamic changes. The present invention provides a method for synchronously acquiring a Mueller matrix sequence and an environmental parameter sequence according to a fixed sampling period, then performing feature extraction and combination to obtain a feature tensor sequence, and then determining the target compensation parameters based on the feature tensor sequence. This method integrates the nonlinear response into the Mueller matrix to determine the target compensation parameters, which can improve the accuracy of obtaining the polarization aberration compensation parameters, thereby further improving the overall compensation accuracy and imaging accuracy of the system.

[0065] In some embodiments, the environmental parameter sequence may include at least one of a temperature sequence, an air pressure sequence, and a vibration sequence. The temperature sequence is used to characterize the temperature changes of the lithography machine system, the air pressure sequence is used to characterize the air pressure changes of the lithography machine system, and the vibration sequence is used to characterize the vibration acceleration changes of the moving parts and / or optical parts of the lithography machine system.

[0066] It should be noted that the temperature, pressure, and vibration acceleration changes of the lithography system have been determined to have a significant impact on the stability of the lithography system, specifically, on polarization aberration. Therefore, at least one of the temperature, pressure, and vibration sequences can be used as an environmental parameter sequence to capture nonlinear response characteristics. Vibration acceleration monitoring of the lithography system primarily targets moving parts (wafer stage, mask stage) and / or core optical components (projection objective, illumination system), requiring high-frequency sampling and time-series modeling to eliminate their impact on polarization aberration. Therefore, the vibration acceleration can include vibration data from at least one of the components: the wafer stage, mask stage, projection objective, or illumination system. This can improve the accuracy of obtaining polarization aberration compensation parameters, thereby further enhancing the overall compensation accuracy and imaging precision of the system.

[0067] Figure 2 Schematic diagram of the process of obtaining the Mueller matrix sequence and environmental parameter sequence corresponding to the lithography system provided by the present invention. Figure 2 As shown, the acquisition of the Mueller matrix sequence and the environmental parameter sequence corresponding to the lithography system may include:

[0068] Step 201: Acquire multiple groups of Stokes vectors at each sampling point according to a first sampling period, and perform least squares fitting on the multiple groups of Stokes vectors to obtain a Mueller matrix corresponding to each sampling point. The Mueller matrices corresponding to multiple sampling points constitute the Mueller matrix sequence.

[0069] It should be noted that the Mueller matrix sequence can be obtained using the original lithography system. That is, multiple sets of Stokes vectors are obtained at each sampling point corresponding to a fixed sampling period, i.e., the first sampling period, and the multiple sets of Stokes vectors are fitted with least squares to obtain the Mueller matrix corresponding to each sampling point.

[0070] Exemplarily, the lithography system may include an illumination device, a polarization state generator and other equipment.

[0071] Step 202: synchronously acquiring the temperature parameter and the air pressure parameter corresponding to each sampling point according to the first sampling period. The temperature parameters and the air pressure parameters corresponding to the plurality of sampling points constitute the temperature sequence and the air pressure sequence.

[0072] Step 203: Acquire multiple vibration acceleration parameters according to a second sampling period, where the multiple vibration acceleration parameters constitute the vibration sequence, and the second sampling period is shorter than the first sampling period.

[0073] It should be noted that the environmental parameter sequence includes a temperature sequence, an air pressure sequence, and a vibration sequence. To obtain the environmental parameter sequence, corresponding sensor modules can be configured in the lithography system, such as temperature sensors, air pressure sensors, and acceleration sensors. Environmental parameters are collected simultaneously with the Mueller matrix to obtain the environmental parameter sequence. The sampling period for the temperature and air pressure sequences is the same as that for the Mueller matrix, while the sampling frequency for vibration acceleration is higher, resulting in a shorter sampling period than that for the Mueller matrix.

[0074] Figure 3 Schematic diagram of the structure of the lithography system provided by the present invention. Figure 3 As shown in FIG, the lithography system is also a system for collecting Mueller matrix sequences and environmental parameter sequences, and mainly includes the following modules: illumination and lithography module, data acquisition module, polarization state analyzer (PSA) and charge-coupled device (CCD) detector.

[0075] Among them, the illumination and photolithography module includes: an illumination device: generates excimer laser; a polarization state generator (PSG): composed of a polarizer and a quarter-wave plate, used to generate a linearly polarized light beam; a projection objective: used to shrink the circuit pattern on the mask and transfer it to the wafer surface; a collimator: used to adjust the light beam output by the projection objective into parallel light, ensuring uniform energy distribution and consistent angle in the optical path.

[0076] The data acquisition module includes: temperature, air pressure, and acceleration sensors, which are used to measure the temperature, air pressure, and vibration acceleration of the lithography system respectively;

[0077] Polarization state analyzers and charge-coupled device detectors are used to detect the parameters of the Stokes vector. By measuring multiple sets of Stokes vectors and performing fitting calculations, the Mueller matrix at the current moment can be obtained.

[0078] It can be understood that obtaining the Mueller matrix and environmental parameters at each sampling point according to a fixed sampling period lays the foundation for subsequent feature extraction and combination, and also provides conditions for obtaining accurate target compensation parameters.

[0079] Figure 4 Schematic diagram of the method for obtaining a feature tensor sequence provided by the present invention. Figure 4 As shown, the environmental parameter sequence includes the temperature sequence, the air pressure sequence, and the vibration sequence. The feature extraction of the Mueller matrix sequence and the environmental parameter sequence is performed separately and the features of the same time series are combined to obtain a feature tensor sequence, which may include:

[0080] Step 301: performing principal component analysis dimensionality reduction processing on the Mueller matrix sequence to obtain a Mueller matrix eigenvector sequence after dimensionality reduction.

[0081] It should be noted that the principal component analysis (PCA) dimensionality reduction method can be used. That is, the Mueller matrix is ​​a 4×4 real matrix (16 elements), but there are strong physical constraints and correlations between the actual elements. Therefore, PCA (principal component analysis) can be used to reduce the dimensionality and remove redundant information.

[0082] Step 302: performing low-pass filtering on the temperature sequence and the pressure sequence respectively to obtain a filtered temperature sequence and a filtered pressure sequence.

[0083] It should be noted that a low-pass filtering method can be used, that is, low-pass filtering can be performed on parameters with a relatively large time scale of change, such as temperature and air pressure, to smooth out some short-term fluctuations and noise and reduce interference from small fluctuations.

[0084] Step 303: performing sliding window aggregation on the vibration sequence to obtain an aggregated acceleration sequence.

[0085] It should be noted that a sliding aggregation method can be used, that is, window aggregation is performed on high-frequency vibration acceleration information with a relatively small change time scale to align the time scale.

[0086] Step 304: normalize the reduced-dimensional Mueller matrix eigenvector sequence, the filtered temperature sequence, the filtered pressure sequence, and the aggregated acceleration sequence, and then concatenate features of the same time series to obtain the feature tensor sequence.

[0087] It should be noted that after obtaining the reduced-dimensional Mueller matrix eigenvector sequence, the filtered temperature sequence, the filtered pressure sequence, and the aggregated acceleration sequence, the processed features can be normalized and the features of the same time series can be combined into a three-dimensional input tensor:

[0088] ;

[0089] Among them, the feature tensor sequence at each time point contains:

[0090] ;

[0091] In the above formula, is the eigenvector of the Mueller matrix after dimensionality reduction, 、 It is the temperature and pressure value after low-pass smoothing filtering. and are the mean and variance values ​​of the acceleration after aggregation.

[0092] Understandably, existing methods ignore component coupling effects. Specifically, they independently adjust optical components for different aberration components (such as depolarization and retardation), neglecting the coupling effects of different components and lacking the ability to adaptively handle complex situations. The method for obtaining a feature tensor sequence, provided by the present invention, fuses multiple features to form a feature tensor sequence, which can improve the accuracy of the obtained polarization aberration compensation parameters.

[0093] Furthermore, the Mueller matrix sequence includes at least one 4×4 Mueller matrix, and performing principal component analysis dimensionality reduction processing on the Mueller matrix sequence to obtain a sequence of Mueller matrix eigenvectors after dimensionality reduction may include: expanding each 4×4 Mueller matrix in the at least one 4×4 Mueller matrix into a 16-dimensional row vector in a row-first manner to obtain at least one 16-dimensional row vector; stacking the at least one 16-dimensional row vector to form a data matrix, and extracting the first K principal components of the data matrix to obtain the sequence of Mueller matrix eigenvectors after dimensionality reduction.

[0094] It should be noted that the dimensionality reduction process of the Mueller matrix may include the following steps: (1) Each 4×4 Mueller matrix M(t n ) is expanded row-first into a 16-dimensional row vector m(t n ),m(t n )=[M 11 (t n ),M 12 (t n ),…,M 14 (t n ),M 21 (t n ),…,M 44 (t n )]∈R 1×16 ; (2) The data matrix X is formed by stacking the expansion vectors at N moments vertically:

[0095] ,in represents the field of real numbers, Represents a real number matrix with N rows and 16 columns. Use the centralized data matrix and the projection matrix V obtained in the training phase K ∈R 16×K By multiplying them, we get the reduced-dimensional Mueller matrix eigenvector sequence, as shown in the following formula. We can see that the original 16-dimensional eigenvector is reduced to K dimensions.

[0096] .

[0097] It can be understood that the Mueller matrix dimensionality reduction method can extract key features and improve the accuracy of the obtained polarization aberration compensation parameters.

[0098] For example, during the model training phase, the process used may be as follows:

[0099] 1. Data centering, let the original training data set be X∈R N×16 , where N is the number of samples and each row is a 16-dimensional Mueller matrix expansion vector. Each column (feature dimension) is centered:

[0100] Centralized data ,

[0101] where μ = [μ1,μ2,...,μ 16 ] T is the mean vector of each feature dimension.

[0102] ,

[0103] Among them 1 N is an N×1 column vector whose elements are all 1, express The element in the i-th row and j-th column of the matrix.

[0104] 2. Calculate the covariance matrix of the centralized data, C∈R 16×16 .

[0105] ; in, express The matrix obtained after matrix centering is, express The transpose of the matrix obtained by centering the matrix.

[0106] Perform eigendecomposition on C and obtain the eigenvalues ​​λ1≥λ2≥…≥λ 16 and the corresponding eigenvectors v1,v2,…,v 16 ;

[0107] 3. Select the eigenvectors corresponding to the first K largest eigenvalues ​​and construct the projection matrix V K ∈R 16×K :

[0108] V K =[v1,v2,…,v k ].

[0109] 4. Project the centralized data to reduce the dimension and obtain the feature matrix Y after dimension reduction train ∈R N×K .

[0110] .

[0111] The data matrix after dimensionality reduction is used for model training to obtain a trained compensation parameter prediction model.

[0112] In the prediction stage, new sample dimensionality reduction is performed. For new sample x∈R 1×16 , follow these steps:

[0113] 1. Centering: Use the mean μ during the training phase for centering:

[0114] New sample data after centralization ,in represents the sequence of Mueller matrix eigenvectors obtained in the prediction phase;

[0115] 2. Projection: obtained through training Dimensionality reduction:

[0116] New sample data after dimensionality reduction ;

[0117] 3. Output y new ∈R 1×K is the feature vector after dimensionality reduction.

[0118] The method of dimensionality reduction of the Mueller matrix can extract key features and improve the accuracy of the obtained polarization aberration compensation parameters.

[0119] Furthermore, the sampling periods corresponding to the temperature sequence and the pressure sequence are both first sampling periods, and the low-pass filtering of the temperature sequence and the pressure sequence to obtain the filtered temperature sequence and the filtered pressure sequence may include: performing moving average filtering on the temperature sequence through a first sliding window to obtain the filtered temperature sequence; performing moving average filtering on the pressure sequence through a second sliding window to obtain the filtered pressure sequence, wherein the time lengths of the first sliding window and the second sliding window are greater than or equal to two first sampling periods, and the moving average filtering includes taking the average of the values ​​of at least two sampling points in the window as the value of the current sampling point.

[0120] Exemplarily, a moving average filter with a sliding window size of W1 and W2 is applied to the temperature sequence and the pressure sequence, that is, the filtered temperature sequence is obtained by calculating the average value of the data corresponding to W1 sampling points before each current sampling point, and the filtered pressure sequence is obtained by calculating the average value of the data corresponding to W2 sampling points before each current sampling point, and W1 and W2 are different.

[0121] Exemplarily, W1 and W2 can be the same W, and average filtering processing is performed on the measured temperature and air pressure sequence sliding. A suitable window size W is set, and the average value of W data in the sliding window corresponding to the sampling point at the current moment is calculated as the filtered value at the current moment. Through the above method, short-term fluctuations can be smoothed out, and the influence of high-frequency noise can be reduced.

[0122] It can be understood that the average filtering processing of the measured temperature and air pressure can extract key features and improve the accuracy of the acquired polarization aberration compensation parameters.

[0123] Further, the sliding window aggregation of the vibration sequence to obtain the aggregated acceleration sequence can include: determining that the time length of a third sliding window is equal to the first sampling period; performing sliding aggregation on the vibration sequence through the third sliding window, that is, calculating the mean and variance of the acceleration data in each sliding window to obtain processed vibration acceleration data; and then time-synchronously aligning the processed vibration acceleration data according to the first sampling period to obtain the aggregated acceleration sequence.

[0124] It should be noted that the acceleration sensor original signal can be aggregated in a sliding window to calculate the acceleration mean and variance in the first sampling period ΔT time window.

[0125] It can be understood that the sliding aggregation processing of the measured vibration acceleration data can extract key features and improve the accuracy of the acquired polarization aberration compensation parameters.

[0126] Figure 5 is a schematic diagram provided by the present application for feature extraction of Mueller matrix sequence and environmental parameter sequence. As shown in Figure 5 , the sampling periods of the Mueller matrix sequence, the temperature sequence and the air pressure sequence are all T, the sampling period of the vibration sequence is much smaller than T, the sliding window W1 corresponding to the temperature sequence is equal to three sampling periods, the sliding window W2 corresponding to the air pressure sequence is equal to two sampling periods, the aggregation window W3 corresponding to the vibration sequence is W3, and the result of sliding aggregation is to aggregate multiple values in the W3 window into a mean and variance, the time length of W3 is equal to the sampling period T, and all points in the period T are aggregated to obtain a sampling value.

[0127] In some embodiments, the determining the corresponding target compensation parameter according to the feature tensor sequence can include: inputting the feature tensor sequence into a trained compensation parameter prediction model, and outputting the target compensation parameter, wherein the trained compensation parameter prediction model is obtained by inputting a measured Mueller matrix sequence and an environmental parameter sequence into an initial compensation parameter prediction model, and optimizing model parameters according to a mean square error between a predicted compensation parameter output by the model and a target imaging parameter corresponding to a measured imaging parameter.

[0128] It should be noted that the existing method for obtaining a polarization aberration compensation parameter has strong dependence on a model, and the degree of automation, that is, the method usually depends on a pre-defined mode expansion (such as Fourier or Zernike decomposition) to represent the characteristics of the Mueller matrix. Such a method depends on a parameter structure selected manually, and the modeling process depends on domain experience and lacks adaptive learning ability, and it is difficult to cover complex polarization distortion modes.

[0129] The present application inputs a measured Mueller matrix sequence and an environmental parameter sequence into an initial compensation parameter prediction model, and optimizes model parameters according to a mean square error between a predicted compensation parameter output by the model and a target imaging parameter corresponding to a measured imaging parameter to obtain a trained compensation parameter prediction model. The feature tensor sequence is input into the trained compensation parameter prediction model, and the target compensation parameter is output.

[0130] It can be understood that the joint modeling and collaborative optimization of multiple adjustment parameters affecting the polarization performance are realized in the training process, and then the system coupling effect caused by independent adjustment of each compensation unit in the traditional scheme is avoided, the overall compensation accuracy and imaging precision are improved, and good system integration capability and engineering feasibility are possessed.

[0131] Further, the trained compensation parameter prediction model includes a long short-term memory neural network model and a fully connected layer, the fully connected layer is used to fuse time sequence features output by the initial compensation parameter prediction model and generate a target compensation parameter, and the target compensation parameter includes at least one of a polarization plate angle compensation amount and a fast axis azimuth angle compensation amount of a quarter wave plate.

[0132] It should be noted that the long short-term memory network (Long Short-Term Memory, LSTM) is a special recurrent neural network (RNN), which can automatically mine key features from data by virtue of its strong self-learning ability, thereby effectively overcoming the dependence on artificial experience of the existing method, and realizing efficient representation of complex polarization characteristics. The features are extracted by the LSTM model, and the features are fused by the fully connected layer, and finally the compensation parameter of the polarization controller is output, and the compensation parameter can be at least one of the angle compensation amount of the polarization plate and the fast axis azimuth angle compensation amount of the quarter wave plate.

[0133] In addition, a feedback verification module can be used, that is, a polarization controller is used to compensate the polarization parameters of the illumination system according to the output of the machine learning model. The effect of compensation is fed back to the long short-term memory neural network model, and the model parameters are optimized.

[0134] It can be understood that the prior art generally uses an end-to-end global optimization method, while the application uses a long short-term memory neural network model (Long Short-Term Memory, LSTM) as a data-driven modeling framework, which can realize joint modeling and collaborative optimization of multiple adjustment parameters (such as polarization plate angle, wave plate phase delay, etc.) that affect the polarization performance in the training process, thereby avoiding the system coupling effect caused by independent adjustment of each compensation unit in the traditional scheme, improving the overall compensation accuracy and imaging precision, and having good system integration capability and engineering feasibility.

[0135] In some embodiments, the target compensation parameters include a polarization plate angle compensation amount and a fast-axis azimuth angle compensation amount of a quarter-wave plate, and the method can further include adjusting the polarization plate angle and the fast-axis azimuth angle of the quarter-wave plate in the polarization state generator according to the target compensation parameters.

[0136] It should be noted that after obtaining the polarization plate angle compensation amount and the fast-axis azimuth angle compensation amount of the quarter-wave plate, the polarization plate angle and the fast-axis azimuth angle of the quarter-wave plate in the polarization state generator can be adjusted by the polarization plate angle compensation amount and the fast-axis azimuth angle compensation amount of the quarter-wave plate to realize polarization aberration compensation.

[0137] It can be understood that in order to overcome the limitations of the existing polarization aberration compensation method based on Mueller matrix in adaptive feature extraction, nonlinear response modeling and collaborative optimization of coupling effect, the application proposes a polarization aberration compensation method based on long short-term memory neural network (Long Short-Term Memory, LSTM). The method acquires a time series data set of Mueller matrix parameters, lithography system environment state parameters (such as temperature, air pressure, vibration acceleration), etc., inputs the trained LSTM model, the trained LSTM model has learned the variation law of polarized light in the optical system, and performs polarization compensation on the results obtained by calculation simulation through a feedback system, which can improve the accuracy of obtaining polarization aberration compensation parameters, thereby reducing polarization aberration and improving imaging quality.

[0138] Furthermore, the present invention utilizes a dynamic polarization aberration compensation system architecture based on LSTM and the Mueller matrix. This system utilizes a comprehensive technology chain consisting of Mueller matrix sequence acquisition, environmental sensor data fusion, PCA dimensionality reduction preprocessing, LSTM time series modeling, and closed-loop compensation control. This system utilizes a multi-source data fusion method and sliding window aggregation technology based on the "Mueller matrix-environmental parameters" approach. This improves the accuracy of obtaining polarization aberration compensation parameters, thereby further enhancing the overall compensation accuracy and imaging precision of the system.

[0139] The following describes an exemplary application of an embodiment of the present invention in a practical application scenario.

[0140] The lithography process of a lithography machine can be briefly divided into several stages: the simulation design phase, the exposure preparation phase, and the exposure phase. During the simulation design phase, the lithography process is simulated using simulation software to predict imaging results and optimize parameters (including illumination mode, mask design, polarization configuration, etc.). In particular, while the parameters derived from computational simulation take polarization aberration into account, the results generally cannot completely eliminate its influence, necessitating polarization compensation based on this. Polarization compensation is performed during the exposure preparation phase. During this phase, the polarization controller (such as a rotating wave plate or micromirror array) is adjusted to compensate for aberration based on the newly measured Mueller matrix or a preset model. During the exposure phase, exposure is typically performed according to the computational simulation design and the parameters determined for polarization aberration compensation, and polarization state adjustment is generally not performed. The present invention is applied during the exposure preparation phase. Using an LSTM model, it extracts the characteristics of the Mueller matrix sequence and the temperature, pressure, and vibration sequence of the lithography system to predict the compensation parameters of the polarization generator (PSG). Feedforward compensation is then performed to fine-tune the polarization parameters through a feedback system.

[0141] Specifically, the ideal parameters of the design simulation stage are obtained in the simulation design stage, including: polarizer angle compensation amount sequence {θ0(t0),θ0(t1),...,θ0(t n )}, and the fast axis azimuth sequence of the quarter-wave plate {φ0(t0),φ0(t1),...,φ0(t n )}; the imaging parameter error sequence is the target critical dimension (Critical Dimension, CD) error sequence {CD0(t0),CD0(t1),...,CD0(t n )}, (where t n Indicates the nth measurement moment, ΔT=t n -t n-1 is a fixed sampling interval). Exemplarily, the imaging parameter error sequence may be a mean square error sequence between the predicted imaging linewidth and the measured imaging linewidth.

[0142] Figure 6FIG. 1 is a schematic diagram of the overall flow of the method for obtaining polarization aberration compensation parameters provided by the present invention. Figure 6 As shown, the measurement data sequence of the exposure preparation stage is obtained, including:

[0143] Polarization state analysis unit (polarization state analyzer PSA and charge coupled device CCD detector): measures the Stokes vector parameters with a fixed period ΔT and calculates the measured Mueller matrix sequence {M(t0),M(t1),...,M(t n )}.

[0144] Environmental monitoring unit: measures the temperature sequence {T(t0), T(t1), ..., T(t n )} and the pressure sequence {P(t0),P(t1),...,P(t n )}, since the vibration frequency is relatively high, the original acceleration sequence {a(τ0),...,a(τ m )}.

[0145] Then the acquired temperature sequence is low-pass filtered, the acquired pressure sequence is low-pass filtered, the acquired vibration sequence is sliding aggregated, and the acquired Mueller matrix sequence is PCA-dimensionality reduced to obtain a feature tensor sequence.

[0146] The obtained feature tensor sequence is then input into the LSTM model, and the target compensation parameters are output through the fully connected layer of the LSTM model. The target compensation parameters include the polarizer angle compensation and the fast axis azimuth compensation of the quarter-wave plate .

[0147] The processed parameters are input into the trained LSTM neural network for prediction and output as compensation sequence through the fully connected layer network. .

[0148] Closed-loop timing compensation control includes: for each time t_{n+1}, parameter compensation is as follows, and then the polarization controller is used to adjust the polarizer angle and the fast axis azimuth of the quarter-wave plate.

[0149]

[0150] in, and It is an ideal parameter in the design simulation stage. and is the polarization aberration compensation parameter output by the LSTM neural network.

[0151] in addition, Figure 6The steps in the dotted box are to simulate the transformation of the exposure system, which is used to back-propagate the error between the measured critical dimension (CD) and the simulated ideal CD to the and In

[15] , the LSTM model is trained. The loss function is the mean square sum of the simulated target CD and the actual measured CD.

[0152] The system also includes a machine learning prediction module: it mainly includes an LSTM model to extract features and a fully connected layer to fuse the features, and finally outputs the compensation parameters of the polarization controller, mainly the angle of the polarizer and the fast axis azimuth of the quarter-wave plate.

[0153] The system can also include a feedback verification module: mainly a polarization controller that compensates the polarization parameters of the illumination system based on the output of the machine learning model.

[0154] As can be understood, to overcome the limitations of existing Mueller matrix-based polarization aberration compensation methods in adaptive feature extraction, nonlinear response modeling, and coordinated optimization of coupling effects, the present invention proposes a polarization aberration compensation method based on a long short-term memory (LSTM) neural network. This method constructs a time series dataset based on Mueller matrix parameters and lithography system state parameters (such as temperature, air pressure, and vibration acceleration) to train an LSTM model to learn how polarized light varies in the optical system. Polarization compensation is then applied to the results obtained through computational simulations through a feedback system, thereby reducing polarization aberration and improving imaging quality.

[0155] It offers the following advantages: Adaptive feature learning: The Long Short-Term Memory (LSTM) network is a special type of recurrent neural network (RNN). Leveraging its powerful autonomous learning capabilities, it can automatically and deeply mine key features from data, effectively overcoming the reliance of existing methods on manual experience and enabling efficient characterization of complex polarization properties. It also possesses nonlinear mapping and time series modeling capabilities: Although the Mueller matrix itself is a linear transformation model, components in a lithography system produce nonlinear responses under certain conditions, which are reflected in the measured Mueller matrix. Therefore, by training the Mueller matrix sequence at different times into the LSTM model, it is possible to learn the nonlinear mapping of the lithography system. End-to-end global optimization: As a data-driven modeling framework, the LSTM model enables joint modeling and coordinated optimization of multiple adjustment parameters that affect polarization performance (such as polarizer angle and waveplate phase delay) during training. This avoids the system coupling effects caused by the independent adjustment of each compensation unit in traditional solutions, improves overall compensation accuracy and imaging precision, and offers excellent system integration capabilities and engineering feasibility.

[0156] Based on the aforementioned embodiments, an embodiment of the present invention provides a system for acquiring polarization aberration compensation parameters. The modules included in the system, and the units included in each module, can be implemented by a processor; of course, they can also be implemented by a specific logic circuit. During implementation, the processor can be a central processing unit (CPU), a microprocessor (MPU), a digital signal processor (DSP), or a field programmable gate array (FPGA), etc.

[0157] The polarization aberration compensation parameter system provided by the present invention is described below. The polarization aberration compensation parameter acquisition system described below and the polarization aberration compensation parameter acquisition method described above can be referred to in correspondence with each other.

[0158] Figure 7 Schematic diagram of the structure of the polarization aberration compensation system provided by the present invention. Figure 7 As shown, the system 400 includes a data acquisition module 401, a feature acquisition module 402 and a parameter prediction module 403, wherein:

[0159] Data acquisition module 401, configured to acquire a Mueller matrix sequence and an environmental parameter sequence corresponding to a lithography system, wherein the timing of the Mueller matrix sequence and the environmental parameter sequence are aligned. The Mueller matrix sequence is used to characterize changes in polarized light transmission in the lithography system, and the environmental parameter sequence is used to characterize changes in environmental parameters that affect the stability of the lithography system.

[0160] A feature acquisition module 402 is configured to extract features from the Mueller matrix sequence and the environmental parameter sequence respectively and combine features of the same time series to obtain a feature tensor sequence;

[0161] The parameter prediction module 403 is used to determine target compensation parameters according to the characteristic tensor sequence, where the target compensation parameters are used to compensate for the polarization aberration of the lithography system.

[0162] In some embodiments, the environmental parameter sequence includes at least one of a temperature sequence, an air pressure sequence, and a vibration sequence. The temperature sequence is used to characterize the temperature changes of the lithography system, the air pressure sequence is used to characterize the air pressure changes of the lithography system, and the vibration sequence is used to characterize the vibration acceleration changes of the moving parts and / or optical parts of the lithography system.

[0163] In some embodiments, the environmental parameter sequence includes the temperature sequence, the air pressure sequence, and the vibration sequence, and the feature acquisition module 402 includes:

[0164] A dimensionality reduction processing unit, configured to perform principal component analysis dimensionality reduction processing on the Mueller matrix sequence to obtain a sequence of eigenvectors of the Mueller matrix after dimensionality reduction;

[0165] a filtering processing unit, configured to perform low-pass filtering on the temperature sequence and the air pressure sequence respectively to obtain a filtered temperature sequence and a filtered air pressure sequence;

[0166] a window aggregation unit, configured to perform sliding window aggregation on the vibration sequence to obtain an aggregated acceleration sequence;

[0167] A feature combination unit is used to normalize the reduced-dimensional Mueller matrix eigenvector sequence, the filtered temperature sequence, the filtered pressure sequence, and the aggregated acceleration sequence, and then splice the features of the same time series to obtain the feature tensor sequence.

[0168] In some embodiments, the Mueller matrix sequence includes at least one 4×4 Mueller matrix, and the dimensionality reduction processing unit is specifically used to: expand each 4×4 Mueller matrix in the at least one 4×4 Mueller matrix into a 16-dimensional row vector in a row-first manner to obtain at least one 16-dimensional row vector; stack the at least one 16-dimensional row vector to form a data matrix, and extract the first K principal components of the data matrix to obtain the reduced Mueller matrix eigenvector sequence.

[0169] In some embodiments, the sampling periods corresponding to the temperature sequence and the pressure sequence are both first sampling periods, and the filtering processing unit is specifically used to: perform moving average filtering on the temperature sequence through a first sliding window to obtain the filtered temperature sequence; perform moving average filtering on the pressure sequence through a second sliding window to obtain the filtered pressure sequence, wherein the time lengths of the first sliding window and the second sliding window are greater than or equal to two first sampling periods, and the moving average filtering includes taking the average of the values ​​of at least two sampling points in the window as the value of the current sampling point.

[0170] In some embodiments, the second sampling period corresponding to the vibration sequence is smaller than the first sampling period, the third sliding window is a non-overlapping sliding window, and the window aggregation unit is specifically used to: determine that the time length of the third sliding window is equal to the first sampling period; perform sliding aggregation on the vibration sequence through the third sliding window to obtain processed vibration acceleration data, wherein the sliding aggregation includes calculating the mean and variance of the acceleration data within the window as the value of the current sampling point; and perform time alignment on the processed vibration acceleration data according to the first sampling period to obtain the aggregated acceleration sequence.

[0171] In some embodiments, the parameter prediction module 403 is specifically used to: input the feature tensor sequence into a trained compensation parameter prediction model, and output the target compensation parameter. The trained compensation parameter prediction model is obtained by inputting the measured Mueller matrix sequence and the environmental parameter sequence into the initial compensation parameter prediction model, and optimizing the model parameters according to the mean square error between the measured imaging parameters corresponding to the predicted compensation parameters output by the model and the target imaging parameters.

[0172] In some embodiments, the trained compensation parameter prediction model includes a long short-term memory neural network model and a fully connected layer, and the fully connected layer is used to fuse the timing features output by the long short-term memory neural network model and generate target compensation parameters, and the target compensation parameters include at least one of the polarizer angle compensation amount and the quarter-wave plate fast axis azimuth compensation amount.

[0173] In some embodiments, the target compensation parameters include the polarizer angle compensation amount and the quarter-wave plate fast-axis azimuth compensation amount, and the device also includes a compensation implementation module for adjusting the polarizer angle compensation amount and the quarter-wave plate fast-axis azimuth in the polarization state generator according to the target compensation parameters.

[0174] In some embodiments, the data acquisition module 401 includes:

[0175] a first acquisition unit, configured to acquire multiple groups of Stokes vectors at each sampling point according to a first sampling period, and perform least squares fitting on the multiple groups of Stokes vectors to obtain a Mueller matrix corresponding to each sampling point, wherein the Mueller matrices corresponding to the multiple sampling points constitute the Mueller matrix sequence;

[0176] a second acquisition unit, configured to synchronously acquire a temperature parameter and an air pressure parameter corresponding to each sampling point according to the first sampling period, wherein the temperature parameters and the air pressure parameters corresponding to the plurality of sampling points constitute the temperature sequence and the air pressure sequence;

[0177] The third acquisition unit is configured to acquire a plurality of vibration acceleration parameters according to a second sampling period, wherein the plurality of vibration acceleration parameters constitute the vibration sequence, and the second sampling period is shorter than the first sampling period.

[0178] In an embodiment of the present invention, multi-source data consisting of the Mueller matrix and environmental parameters is fused and features are extracted to obtain a characteristic tensor sequence. Target compensation parameters are then determined based on this characteristic tensor sequence. This can improve the accuracy of obtaining polarization aberration compensation parameters, thereby further improving the overall compensation accuracy and imaging precision of the system.

[0179] Figure 8 Schematic diagram of the physical structure of the electronic device provided by the present invention. Figure 8 As shown, the electronic device may include: a processor 510, a communication interface 520, a memory 530, and a communication bus 540, wherein the processor 510, the communication interface 520, and the memory 530 communicate with each other via the communication bus 540. The processor 510 may call logic instructions in the memory 530 to execute a method for obtaining polarization aberration compensation parameters, the method comprising: obtaining a Mueller matrix sequence and an environmental parameter sequence corresponding to a lithography system, wherein the timing of the Mueller matrix sequence and the environmental parameter sequence are aligned, the Mueller matrix sequence is used to characterize changes in polarized light transmitted in the lithography system, and the environmental parameter sequence is used to characterize changes in environmental parameters that affect the stability of the lithography system; performing feature extraction on the Mueller matrix sequence and the environmental parameter sequence respectively, and combining features of the same timing to obtain a feature tensor sequence; and determining target compensation parameters based on the feature tensor sequence, wherein the target compensation parameters are used to compensate for the polarization aberration of the lithography system.

[0180] In addition, the logic instructions in the memory 530 described above can be implemented in the form of software functional units and sold or used as independent products, and can be stored in a computer readable storage medium. Based on such understanding, the technical solutions of the present application essentially or the part that contributes to the prior art or part of the technical solutions can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for making a computer device (which can be a personal computer, a server, or a network device, etc.) execute all or part of the steps of the methods described in various embodiments of the present application. The aforementioned storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a magnetic disk or an optical disk, and various media that can store program codes.

[0181] In another aspect, the present application also provides a computer program product, which comprises a computer program, the computer program can be stored on a non-transitory computer readable storage medium, and the computer program can be executed by a processor to enable a computer to execute the method for obtaining the polarization aberration compensation parameter provided by the above-mentioned methods, the method comprising: obtaining a Mueller matrix sequence and an environmental parameter sequence corresponding to a lithography machine system, the time sequence of the Mueller matrix sequence and the environmental parameter sequence are aligned, the Mueller matrix sequence is used to represent the change of the polarization light in the lithography machine system, and the environmental parameter sequence is used to represent the change of the environmental parameter which has an influence on the stability of the lithography machine system; performing feature extraction on the Mueller matrix sequence and the environmental parameter sequence respectively and combining the features of the same time sequence to obtain a feature tensor sequence; determining a target compensation parameter according to the feature tensor sequence, the target compensation parameter is used to compensate the polarization aberration of the lithography machine system.

[0182] The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the flow or function described in the embodiments of the present application is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium or transferred from one computer-readable storage medium to another, for example, the computer instructions can be transferred from one website, computer, server or data center to another website, computer, server or data center through wired (such as coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (such as infrared, wireless, microwave, etc.) mode. The computer-readable storage medium can be any available medium that a computer can store or a data storage device such as a server, data center, etc. integrated with one or more available media. The available media can be a magnetic medium (for example, a floppy disk, a hard disk, a magnetic tape), an optical medium (for example, a DVD), or a semiconductor medium (for example, a solid state disk (SSD)), etc.

[0183] In another aspect, the present application also provides a non-transitory computer readable storage medium having stored thereon a computer program, which, when executed by a processor, implements a method for obtaining a polarization aberration compensation parameter provided by the above-mentioned methods, the method comprising: obtaining a sequence of Mueller matrices corresponding to a lithography machine system and a sequence of environmental parameters, the time sequence of the sequence of Mueller matrices and the sequence of environmental parameters being aligned, the sequence of Mueller matrices being used to represent changes in the transmission of polarized light in the lithography machine system, and the sequence of environmental parameters being used to represent changes in environmental parameters that have an impact on the stability of the lithography machine system; performing feature extraction on the sequence of Mueller matrices and the sequence of environmental parameters respectively and combining the features of the same time sequence to obtain a sequence of feature tensors; determining a target compensation parameter according to the sequence of feature tensors, the target compensation parameter being used to compensate for the polarization aberration of the lithography machine system.

[0184] The computer-readable storage medium may be any combination of one or more computer-readable media. The computer-readable medium may be a computer-readable signal medium or a computer-readable storage medium. The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples (a non-exhaustive list) of computer-readable storage media include: an electrical connection having one or more conductors, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM) or flash memory, optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In this document, a computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device.

[0185] A computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, which carries computer-readable program code. Such a propagated data signal may take a variety of forms, including, but not limited to, electromagnetic signals, optical signals, or any suitable combination thereof. A computer-readable signal medium may also be any computer-readable medium other than a computer-readable storage medium that can transmit, propagate, or transport a program for use by or in conjunction with an instruction execution system, apparatus, or device.

[0186] Program code embodied on a computer-readable medium may be transmitted using any appropriate medium, including but not limited to wireless, wireline, optical fiber cable, radio frequency (RF), etc., or any suitable combination of the foregoing.

[0187] Computer program code for performing the operations of this specification may be written in one or more programming languages, or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, and C++, as well as conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may execute entirely on the user's computer, partially on the user's computer, as a stand-alone software package, partially on the user's computer and partially on a remote computer, or entirely on the remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., through the Internet using an Internet service provider).

[0188] The system embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units. They may be located in one place or distributed across multiple network units. Some or all of the modules may be selected based on actual needs to achieve the objectives of this embodiment. Persons of ordinary skill in the art will be able to understand and implement the present invention without inventive effort.

[0189] Through the above description of the embodiments, those skilled in the art will clearly understand that each embodiment can be implemented using software plus a necessary general-purpose hardware platform, or of course, hardware. Based on this understanding, the essence of the above technical solution, or the portion that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a magnetic disk, or an optical disk, and includes a number of instructions for causing a computer device (such as a personal computer, server, or network device) to execute the methods described in each embodiment or certain portions of the embodiments.

[0190] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. A method for obtaining polarization aberration compensation parameters, characterized in that: include: Obtaining a Mueller matrix sequence and an environmental parameter sequence corresponding to a lithography system, wherein the timing of the Mueller matrix sequence and the environmental parameter sequence are aligned, the Mueller matrix sequence is used to characterize changes in polarized light transmission in the lithography system, and the environmental parameter sequence is used to characterize changes in environmental parameters that affect the stability of the lithography system; Extracting features from the Mueller matrix sequence and the environmental parameter sequence respectively and combining features of the same time series to obtain a feature tensor sequence; Target compensation parameters are determined according to the characteristic tensor sequence, and the target compensation parameters are used to compensate for the polarization aberration of the lithography system.

2. The method for obtaining polarization aberration compensation parameters according to claim 1, wherein: The environmental parameter sequence includes at least one of a temperature sequence, an air pressure sequence and a vibration sequence. The temperature sequence is used to characterize the temperature changes of the lithography machine system, the air pressure sequence is used to characterize the air pressure changes of the lithography machine system, and the vibration sequence is used to characterize the vibration acceleration changes of the moving parts and / or optical parts of the lithography machine system.

3. The method for obtaining polarization aberration compensation parameters according to claim 2, wherein: The environmental parameter sequence includes the temperature sequence, the air pressure sequence, and the vibration sequence. Feature extraction is performed on the Mueller matrix sequence and the environmental parameter sequence respectively, and features of the same time series are combined to obtain a feature tensor sequence, including: Performing principal component analysis and dimensionality reduction processing on the Mueller matrix sequence to obtain a Mueller matrix eigenvector sequence after dimensionality reduction; performing low-pass filtering on the temperature sequence and the pressure sequence respectively to obtain a filtered temperature sequence and a filtered pressure sequence; Performing sliding window aggregation on the vibration sequence to obtain an aggregated acceleration sequence; The reduced-dimensional Mueller matrix eigenvector sequence, the filtered temperature sequence, the filtered pressure sequence, and the aggregated acceleration sequence are normalized, and then the features of the same time series are spliced ​​to obtain the feature tensor sequence.

4. The method for obtaining polarization aberration compensation parameters according to claim 3, wherein: The Mueller matrix sequence includes at least one 4×4 Mueller matrix, and the principal component analysis dimensionality reduction process is performed on the Mueller matrix sequence to obtain a Mueller matrix eigenvector sequence after dimensionality reduction, including: Expand each 4×4 Mueller matrix in the at least one 4×4 Mueller matrix into a 16-dimensional row vector in a row-first manner to obtain at least one 16-dimensional row vector; The at least one 16-dimensional row vector is stacked to form a data matrix, and the first K principal components of the data matrix are extracted to obtain the eigenvector sequence of the Mueller matrix after dimensionality reduction.

5. The method for obtaining polarization aberration compensation parameters according to claim 3, wherein: The sampling periods corresponding to the temperature sequence and the pressure sequence are both the first sampling period, and the low-pass filtering is performed on the temperature sequence and the pressure sequence to obtain the filtered temperature sequence and the filtered pressure sequence, respectively, including: Performing a moving average filter on the temperature sequence through a first sliding window to obtain the filtered temperature sequence; performing a moving average filter on the air pressure sequence through a second sliding window to obtain the filtered air pressure sequence, wherein the time lengths of the first sliding window and the second sliding window are greater than or equal to two first sampling periods, and the moving average filtering includes taking the average of the values ​​of at least two sampling points within the window as the value of the current sampling point; The second sampling period corresponding to the vibration sequence is smaller than the first sampling period, the third sliding window is a non-overlapping sliding window, and the sliding window aggregation of the vibration sequence to obtain the aggregated acceleration sequence includes: Determine that the time length of the third sliding window is equal to the first sampling period; performing sliding aggregation on the vibration sequence through the third sliding window to obtain processed vibration acceleration data, wherein the sliding aggregation includes calculating the mean and variance of the acceleration data within the window as the value of the current sampling point; The processed vibration acceleration data is time-series aligned according to the first sampling period to obtain the aggregated acceleration sequence.

6. The method for obtaining polarization aberration compensation parameters according to claim 1, wherein: The determining of corresponding target compensation parameters according to the characteristic tensor sequence includes: The characteristic tensor sequence is input into a trained compensation parameter prediction model to output the target compensation parameter. The trained compensation parameter prediction model is obtained by inputting the measured Mueller matrix sequence and the environmental parameter sequence into the initial compensation parameter prediction model, and optimizing the model parameters according to the mean square error between the measured imaging parameters corresponding to the predicted compensation parameters output by the model and the target imaging parameters.

7. The method for obtaining polarization aberration compensation parameters according to claim 6, wherein: The trained compensation parameter prediction model includes a long short-term memory neural network model and a fully connected layer. The fully connected layer is used to fuse the time series features output by the long short-term memory neural network model and generate the target compensation parameters. The target compensation parameters include at least one of the polarizer angle compensation amount and the quarter-wave plate fast axis azimuth compensation amount.

8. The method for obtaining polarization aberration compensation parameters according to claim 2, wherein: The environmental parameter sequence includes the temperature sequence, the air pressure sequence, and the vibration sequence. The step of obtaining the Mueller matrix sequence and the environmental parameter sequence corresponding to the lithography system includes: Acquire multiple groups of Stokes vectors at each sampling point according to the first sampling period, and perform least squares fitting on the multiple groups of Stokes vectors to obtain a Mueller matrix corresponding to each sampling point, wherein the Mueller matrices corresponding to the multiple sampling points constitute the Mueller matrix sequence; synchronously acquiring the temperature parameter and the air pressure parameter corresponding to each sampling point according to the first sampling period, wherein the temperature parameters and the air pressure parameters corresponding to the plurality of sampling points constitute the temperature sequence and the air pressure sequence; A plurality of vibration acceleration parameters are acquired according to a second sampling period, the plurality of vibration acceleration parameters forming the vibration sequence, and the second sampling period is shorter than the first sampling period.

9. A system for acquiring polarization aberration compensation parameters, characterized in that: include: A data acquisition module, configured to acquire a Mueller matrix sequence and an environmental parameter sequence corresponding to a lithography system, wherein the timing of the Mueller matrix sequence and the environmental parameter sequence are aligned, the Mueller matrix sequence is used to characterize changes in polarized light transmission in the lithography system, and the environmental parameter sequence is used to characterize changes in environmental parameters that affect the stability of the lithography system; A feature acquisition module is used to extract features from the Mueller matrix sequence and the environmental parameter sequence respectively and combine features of the same time series to obtain a feature tensor sequence; A parameter prediction module is used to determine target compensation parameters according to the characteristic tensor sequence, and the target compensation parameters are used to compensate for the polarization aberration of the lithography machine system.

10. A lithography machine comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that: When the processor executes the computer program, the method for obtaining polarization aberration compensation parameters according to any one of claims 1 to 8 is implemented.

Images