Self-adaptive spectrum adjustment method, device and equipment for wafer detection UV light source

Abstract

The invention provides a self-adaptive spectrum adjustment method, device and equipment for a wafer detection UV light source. The method comprises the following steps: acquiring first spectrum data for performing preliminary spectrum detection on a target wafer; variational mode decomposition processing is carried out on the first spectral data, and spectral characteristic parameters of the first spectral data are determined; processing the spectral characteristic parameters through the refined composite multi-scale weighted multi-dimensional characteristic entropy, and determining spectral characteristic vectors of the spectral characteristic parameters; determining a mapping relation between the spectral features and the wafer material features by analyzing the spectral feature vectors; and carrying out adaptive adjustment on the UV light source spectrum according to the mapping relation. Through the implementation of the scheme of the invention, the spectrum of the UV light source is adaptively adjusted according to the mapping relation between the spectral characteristics and the wafer material characteristics, so that the spectral characteristics of the light source can be ensured to be matched with wafer detection requirements, unnecessary spectrum redundancy or missing is avoided, the spectrum utilization rate is improved, the detection time is shortened, and the detection efficiency is remarkably improved.

Description

Technical Field

[0001] The present application relates to the technical field of light source control, and in particular to a method, device and equipment for adaptively adjusting the spectrum of a UV light source for wafer inspection. Background Art

[0002] In the field of semiconductor manufacturing, wafer inspection is a key link in ensuring product quality. As an important tool for wafer inspection, the spectral characteristics of UV light sources have an important impact on the accuracy and reliability of the inspection results. In the existing technology, the spectrum of UV light sources is usually fixed, which is difficult to adapt to the changes in different wafer materials and inspection requirements. Although some technologies have attempted to adjust the light source, these methods often lack precise control and adaptability of spectral characteristics. Summary of the invention

[0003] The present application provides a method, device and equipment for adaptive spectrum adjustment of UV light source for wafer inspection, which is used to solve the problem in related technologies that the spectrum of UV light source cannot be dynamically adjusted according to the material characteristics of wafer.

[0004] The first aspect of the present application provides an adaptive spectrum adjustment method for a wafer inspection UV light source, the adaptive spectrum adjustment method for a wafer inspection UV light source comprising: Acquire first spectrum data for preliminary spectrum detection of a target wafer; performing variational mode decomposition processing on the first spectral data to determine spectral characteristic parameters of the first spectral data; Processing the spectral characteristic parameters by refining composite multi-scale weighted multi-dimensional feature entropy to determine a spectral characteristic vector of the spectral characteristic parameters; Determine a mapping relationship between spectral features and wafer material properties by analyzing the spectral feature vector; The spectrum of the UV light source is adaptively adjusted according to the mapping relationship.

[0005] Optionally, in a first implementation of the first aspect of the present application, the step of performing variational mode decomposition processing on the first spectral data to determine spectral characteristic parameters of the first spectral data includes: Decomposing the first spectral data into N modal components by a variational modal decomposition algorithm; wherein N is an integer greater than or equal to 1; Analyzing each of the modal components to determine the modal characteristic parameters of each of the modal components; The modal characteristic parameters of the N modal components are integrated to determine the spectral characteristic parameters of the first spectral data.

[0006] Optionally, in a second implementation of the first aspect of the present application, the step of processing the spectral characteristic parameters by refining the composite multi-scale weighted multi-dimensional feature entropy to determine the spectral feature vector of the spectral characteristic parameters includes: Obtaining multi-scale spectral features corresponding to the spectral characteristic parameters by determining entropy values ​​of the spectral characteristic parameters at different scales; Obtaining weighted spectral features corresponding to the spectral characteristic parameters by weighting the multi-scale spectral features; The spectral feature vector of the spectral characteristic parameter is determined by extracting the weighted spectral feature.

[0007] Optionally, in a third implementation of the first aspect of the present application, the step of determining a mapping relationship between spectral features and wafer material properties by analyzing the spectral feature vector includes: Generating a spectral feature cluster by performing cluster analysis on the spectral feature vector; wherein the spectral feature cluster is a cluster of wafer materials with similar spectral features; Performing statistical analysis on the spectral feature vectors corresponding to the spectral feature clusters to determine the wafer material properties corresponding to the spectral feature clusters; According to the spectral feature cluster and the corresponding wafer material characteristics, a mapping relationship between the spectral features and the wafer material characteristics is established.

[0008] Optionally, in a fourth implementation of the first aspect of the present application, the step of adaptively adjusting the spectrum of the UV light source according to the mapping relationship includes: Determining the spectrum requirement of the target wafer according to the mapping relationship; Preliminarily adjusting the spectrum of the UV light source according to the spectrum requirement; Acquire second spectrum data of the UV light source after preliminary adjustment; determining, according to the second spectral data and the mapping relationship, whether the second spectral data meets the spectral requirement of the target wafer; If the second spectrum data does not meet the spectrum requirement, the spectrum of the UV light source is adjusted for a second time until the second spectrum data meets the spectrum requirement.

[0009] Optionally, in a fifth implementation of the first aspect of the present application, the method further includes: Comparing the second spectral data with the spectral requirement to determine corresponding spectral deviation data; Acquire a spectral deviation feature vector by performing feature extraction on the spectral deviation data; Inputting the spectral deviation feature vector into a machine learning model, and generating an optimal spectral adjustment strategy through model calculation; The spectrum of the UV light source is adjusted according to the optimal spectrum adjustment strategy.

[0010] Optionally, in a sixth implementation of the first aspect of the present application, the method further includes: Generate a fuzzy spectral feature vector by fuzzifying the spectral feature vector; Formulate fuzzy rules according to the mapping relationship; Comparing the fuzzy spectrum feature vector with the fuzzy rule to determine the fuzzy control output; Converting the fuzzy control output into a spectrum adjustment parameter by defuzzification processing; The spectrum of the UV light source is adjusted according to the spectrum adjustment parameter.

[0011] A second aspect of the present application provides an adaptive spectrum adjustment device for a wafer inspection UV light source, the adaptive spectrum adjustment device for a wafer inspection UV light source comprising: An acquisition module, used for acquiring first spectrum data for preliminary spectrum detection of a target wafer; a first determining module, configured to perform variational mode decomposition processing on the first spectral data to determine spectral characteristic parameters of the first spectral data; A second determination module is used to process the spectral characteristic parameters by refining the composite multi-scale weighted multi-dimensional feature entropy to determine the spectral characteristic vector of the spectral characteristic parameters; A third determination module is used to determine the mapping relationship between the spectral feature and the wafer material property by analyzing the spectral feature vector; The control module is used to adaptively adjust the spectrum of the UV light source according to the mapping relationship.

[0012] A third aspect of an embodiment of the present application provides an electronic device, comprising a memory and a processor, wherein the processor is used to execute a computer program stored on the memory, and when the processor executes the computer program, it implements each step of the method for adaptive spectrum adjustment of UV light source for wafer inspection provided in the first aspect of the embodiment of the present application.

[0013] The fourth aspect of the embodiments of the present application provides a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, each step of the method for adaptive spectrum adjustment of UV light source for wafer inspection provided in the first aspect of the embodiments of the present application is implemented.

[0014] In summary, according to the method, device and equipment for adaptive spectrum adjustment of UV light source for wafer inspection provided by the scheme of the present application, the first spectrum data for preliminary spectrum detection of the target wafer is obtained; the first spectrum data is subjected to variational mode decomposition processing to determine the spectrum characteristic parameters of the first spectrum data; the spectrum characteristic parameters are processed by refined composite multi-scale weighted multi-dimensional feature entropy to determine the spectrum characteristic vector of the spectrum characteristic parameters; the mapping relationship between the spectrum characteristics and the wafer material characteristics is determined by analyzing the spectrum characteristic vector; the spectrum of the UV light source is adaptively adjusted according to the mapping relationship. Through the implementation of the scheme of the present application, the spectrum of the UV light source is adaptively adjusted according to the mapping relationship between the spectrum characteristics and the wafer material characteristics, which can ensure that the spectrum characteristics of the light source match the wafer detection requirements, avoid unnecessary spectrum redundancy or loss, improve spectrum utilization, shorten detection time, and thus significantly improve detection efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 A schematic diagram of a process flow of a method for adaptively adjusting the spectrum of a UV light source for wafer inspection provided in an embodiment of the present application; Figure 2 A schematic diagram of a program module of a device for adaptively adjusting the spectrum of a UV light source for wafer inspection provided in an embodiment of the present application; Figure 3 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0016] In order to make the purpose, features, and advantages of the invention of this application more obvious and easy to understand, the technical solutions in the embodiments of this application will be clearly and completely described below in conjunction with the drawings in the embodiments of this application. Obviously, the described embodiments are only part of the embodiments of this application, not all of them. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of this application.

[0017] In order to solve the problem in the related art that the spectrum of the UV light source cannot be dynamically adjusted according to the characteristics of the wafer material, the embodiment of the present application provides an adaptive spectrum adjustment method for the UV light source for wafer inspection, such as Figure 1 The schematic diagram of the process flow of the adaptive spectrum adjustment method of the wafer inspection UV light source provided in this embodiment includes the following steps: Step 110: Acquire first spectrum data for preliminary spectrum detection of a target wafer.

[0018] Specifically, in this embodiment, a target wafer is selected and placed on a detection platform to ensure that the wafer surface is clean and has no obvious defects. The selection of the target wafer should be based on actual production needs, and can be wafers of different materials, different processes or different batches. The target wafer is preliminarily irradiated with a UV light source, and the spectral range of the UV light source should cover the wavelength range required for wafer detection. During the irradiation process, the stability and uniformity of the light source are maintained to ensure that the acquired spectral data is representative. A spectrometer is used to collect the reflectance spectral data of the wafer, and the resolution of the spectrometer should be high enough to capture subtle changes in the spectral signal. During the acquisition process, the wavelength range, intensity and other information of the spectral data are recorded to form preliminary first spectral data.

[0019] Step 120: Perform variational mode decomposition processing on the first spectral data to determine spectral characteristic parameters of the first spectral data.

[0020] Specifically, in this embodiment, the preliminary spectral data is preprocessed, including smoothing and normalization, to remove noise and outliers and improve data quality. The preprocessed spectral data is decomposed using the variational mode decomposition (VMD) algorithm. The VMD algorithm determines the spectral characteristic parameters of each modal component, including center frequency, bandwidth, and amplitude, by decomposing the spectral data into multiple modal components. The center frequency reflects the main frequency component of the modal component, the bandwidth reflects the frequency range of the modal component, and the amplitude reflects the intensity of the modal component. These spectral characteristic parameters are an important basis for subsequent spectral feature extraction and analysis, and can provide key information about the characteristics of wafer materials.

[0021] In an optional implementation of the present embodiment, the step of performing variational modal decomposition processing on the first spectral data to determine the spectral characteristic parameters of the first spectral data includes: decomposing the first spectral data into N modal components by a variational modal decomposition algorithm; analyzing each modal component to determine the modal characteristic parameters of each modal component; integrating the modal characteristic parameters of the N modal components to determine the spectral characteristic parameters of the first spectral data.

[0022] Specifically, in this embodiment, in wafer inspection, the first spectral data obtained usually contains rich spectral information, but this information may be masked by noise and complex background signals. In order to effectively extract spectral features, the variational mode decomposition (VMD) algorithm can be used to decompose the first spectral data. In this process, the VMD algorithm uses the alternating direction multiplier method (ADMM) to solve this variational problem. After the decomposition is completed, the first spectral data will be finely divided into several modal components. Different modal components represent the manifestation of the first spectral data in different frequency bandwidths and have unique physical meanings and characteristics. During the analysis process, the modal characteristic parameters of each modal component are determined, including the center frequency, bandwidth, amplitude, etc., and the characteristic parameters of each modal component are integrated to form a parameter set that can fully describe the spectral characteristics of the first spectral data. When integrating, it is necessary to consider the importance of each modal component and its proportion in the overall data to ensure that the integrated spectral characteristic parameters can accurately reflect the comprehensive characteristics of the original spectral data.

[0023] It should be noted that the VMD algorithm decomposes data by constructing a mathematical model. The objective function of the model is as follows: , Where K represents the number of modal components of the target decomposition, k is the index of the modal component, and t represents the time variable. is the kth modal component, is the center frequency of the kth modal component, is a constraint parameter used to balance the bandwidth between modal components, is the first spectrum data, is a smoothing parameter used to control the smoothness of the modal component, is the time step used to calculate the continuity of the modal components.

[0024] It should be noted that the bandwidth constraint item: , This term is used to limit the bandwidth of each modal component by penalizing the modal component The high frequency change is achieved. Used to balance the relationship between bandwidth constraints and reconstruction errors. In spectral data processing, this helps to highlight important spectral peaks while suppressing high-frequency noise.

[0025] Smoothness constraint: , This term ensures the continuity and smoothness of each modal component over time t. is a smoothing parameter that adjusts the strength of the smoothing, and is the time step, which represents the interval between consecutive time points. In spectral data, this helps to maintain the shape of the spectral peak and prevent excessive frequency jumps.

[0026] Reconstruct the error term: , This is the calculation of the original data The sum of all modal components The difference between them is squared and integrated to ensure that the sum of all modal components can accurately reconstruct the first spectrum data.

[0027] Step 130: Process the spectral characteristic parameters by refining the composite multi-scale weighted multi-dimensional feature entropy to determine the spectral characteristic vector of the spectral characteristic parameters.

[0028] Specifically, in this embodiment, the spectral characteristic parameters of each modal component are subjected to multi-scale analysis, and the spectral features are extracted using the refined composite multi-scale weighted multidimensional feature entropy (RCMWMFE) algorithm. The RCMWMFE algorithm captures the complex characteristics of the spectral signal by calculating the entropy values ​​of the spectral characteristic parameters at different scales. The entropy values ​​of each modal component at different scales are calculated to form a multi-scale entropy value matrix. The entropy value can be calculated using methods such as Shannon entropy, Renyi entropy or Tsallis entropy, and a suitable entropy definition is selected according to actual needs. The multi-scale entropy value matrix is ​​weighted, and the standard deviation is used as a weight to perform weighted fusion on the entropy values ​​at different scales. The weighted processing can highlight the main features, suppress the secondary features, and improve the representativeness of the feature vector. The weighted entropy values ​​are integrated into a spectral feature vector, which can fully describe the complex characteristics of the spectral signal. The dimension of the spectral feature vector is determined according to actual needs, and can contain feature information of multiple modal components.

[0029] In an optional implementation of the present embodiment, the step of determining the spectral feature vector of the spectral characteristic parameters by refining the composite multi-scale weighted multi-dimensional feature entropy to process the spectral characteristic parameters includes: obtaining the multi-scale spectral features corresponding to the spectral characteristic parameters by determining the entropy values ​​of the spectral characteristic parameters at different scales; obtaining the weighted spectral features corresponding to the spectral characteristic parameters by weighted processing the multi-scale spectral features; and determining the spectral feature vector of the spectral characteristic parameters by feature extraction of the weighted spectral features.

[0030] Specifically, in the present embodiment, in wafer detection, spectral analysis is an important technical means. By analyzing the spectral response of the wafer at different wavelengths, its internal structure and defect information can be obtained. Multi-scale spectral feature extraction is to analyze the spectral data at different scales to obtain more comprehensive feature information. For example, a multi-scale entropy algorithm can be used to calculate the entropy value at different scales to evaluate the complexity and irregularity of the spectral signal. By weighting the multi-scale spectral features, important features can be highlighted and unimportant features can be suppressed, thereby improving the representativeness and discrimination of the spectral features. In wafer detection, the spectral features at different scales have different sensitivities to defects, and the detection capability of defects can be enhanced by weighted processing. For example, the spectral features at different scales can be weighted according to the size of the multi-scale entropy value. The scale with a larger entropy value indicates that the spectral signal has higher complexity and irregularity at this scale, so a larger weight should be given. By extracting features from weighted spectral features, high-dimensional spectral data can be reduced to low-dimensional feature vectors, retaining the main spectral feature information, and facilitating subsequent analysis and application. In wafer inspection, feature extraction can be performed using methods such as principal component analysis (PCA), linear discriminant analysis (LDA), or independent component analysis (ICA). For example, PCA can be used to calculate the covariance matrix of weighted spectral features, solve the eigenvalues ​​and eigenvectors of the covariance matrix, and select the eigenvector with the largest eigenvalue as the principal component, thereby forming a spectral feature vector.

[0031] Step 140: Determine the mapping relationship between the spectral features and the wafer material characteristics by analyzing the spectral feature vector.

[0032] Specifically, in this embodiment, after obtaining the spectral feature vector, it is deeply analyzed to determine the mapping relationship between the spectral feature and the wafer material properties. By establishing the connection between the spectral feature and the material properties, the behavior of the wafer under different spectral conditions can be better understood. For example, some spectral features may correspond to characteristics such as defects, impurities or crystal structure of the wafer. Through the accumulation and analysis of a large amount of data, an accurate mapping model is established to provide guidance for the adaptive adjustment of the spectrum.

[0033] In an optional implementation of the present embodiment, the step of determining the mapping relationship between spectral features and wafer material characteristics by analyzing spectral feature vectors includes: generating spectral feature clusters by performing cluster analysis on spectral feature vectors; wherein the spectral feature clusters are clusters of wafer materials with similar spectral features; performing statistical analysis on spectral feature vectors corresponding to the spectral feature clusters to determine wafer material characteristics corresponding to the spectral feature clusters; and establishing a mapping relationship between spectral features and wafer material characteristics based on the spectral feature clusters and the corresponding wafer material characteristics.

[0034] Specifically, in the present embodiment, the spectral feature vector is obtained by extracting and processing the spectral data of the wafer material, and contains the spectral response information of the wafer material at different wavelengths. Through cluster analysis, wafer materials with similar spectral features can be classified into one category to form a spectral feature cluster. Cluster analysis can use a variety of algorithms, such as K-Means algorithm, hierarchical clustering algorithm, etc. These algorithms divide the spectral feature vectors into different clusters by calculating the similarity or distance between the spectral feature vectors.

[0035] Perform statistical analysis on the spectral feature vectors corresponding to the generated spectral feature clusters to determine the wafer material characteristics. Statistical analysis can include calculating the mean, variance, standard deviation and other statistics of the spectral feature vectors, which can reflect the central tendency and dispersion of the spectral response of the wafer material. For example, the mean can represent the average spectral response intensity of the wafer material at different wavelengths, and the variance can reflect the fluctuation of the spectral response. By analyzing these statistics, the characteristics of the wafer material, such as the purity of the material, the type of defects, etc., can be further determined.

[0036] According to the spectral feature clusters and the corresponding wafer material characteristics, a mapping relationship between spectral features and wafer material characteristics is established. This mapping relationship can be achieved through machine learning algorithms, such as support vector machines (SVM), decision trees, random forests, etc. These algorithms can use spectral feature vectors as input and wafer material characteristics as output, and learn the mapping relationship between spectral features and wafer material characteristics through training. For example, the SVM algorithm separates spectral feature vectors of different categories by finding an optimal hyperplane, thereby achieving classification and prediction of wafer material characteristics. Once the mapping relationship between spectral features and wafer material characteristics is established, the material characteristics of unknown wafer materials can be predicted by analyzing the spectral features, providing strong support for wafer inspection and quality control.

[0037] It should be noted that the cluster analysis formula is expressed as: , in, is the kth cluster, representing a class of wafer materials with similar spectral characteristics. is the cluster center of the kth cluster, representing the mean of all spectral feature vectors in the cluster, is the spectral feature vector, representing the spectral feature of the i-th sample, is the weight parameter, It is represented as the parameter to minimize the objective function, which means selecting the cluster that minimizes the sum of the squared distances between the samples in the cluster and the cluster center. is the clustering objective function, which is used to minimize the sum of squares of the distances between samples within a cluster and the cluster center. This term introduces a weight parameter , which is used to balance the relative importance of the objective function in the process of automatically adjusting the spectrum. By adding this term, the dynamic changes of spectral features can be considered in the clustering process, thus better adapting to the actual needs in wafer inspection.

[0038] The statistical analysis formula is expressed as: , , in, The mean of the kth cluster represents the average value of all spectral feature vectors in the cluster. The variance of the kth cluster represents the average of the sum of squared deviations of all spectral feature vectors in the cluster from the mean. This term introduces a weight parameter , which is used to balance the relative importance of the objective function in the process of automatically adjusting the spectrum. By adding this term, the dynamic changes of spectral features can be considered when calculating the mean, thereby better reflecting the characteristics of the wafer material. This term introduces a weight parameter δ to balance the relative importance of the objective function in the automatic spectrum adjustment process. By adding this term, the dynamic changes of spectral features can be considered when calculating the variance, thereby better evaluating the stability of wafer material characteristics.

[0039] Step 150: adaptively adjust the spectrum of the UV light source according to the mapping relationship.

[0040] Specifically, in this embodiment, the spectrum of the UV light source is adaptively adjusted according to the established mapping relationship. The spectrum characteristics of the UV light source, such as the center frequency, bandwidth, and amplitude, are adjusted to match the wafer material characteristics. This adaptive adjustment can improve the accuracy and efficiency of wafer detection and optimize the detection process. For example, for wafers with specific defect characteristics, the detection capability of these defects can be enhanced by adjusting the spectrum of the UV light source, thereby improving the sensitivity and reliability of detection.

[0041] In an optional implementation of the present embodiment, the step of adaptively adjusting the spectrum of the UV light source according to the mapping relationship includes: determining the spectrum requirement of the target wafer according to the mapping relationship; preliminarily adjusting the spectrum of the UV light source according to the spectrum requirement; obtaining second spectrum data of the UV light source after the preliminary adjustment; judging whether the second spectrum data meets the spectrum requirement of the target wafer according to the second spectrum data and the mapping relationship; if the second spectrum data does not meet the spectrum requirement, performing a second adjustment on the spectrum of the UV light source until the second spectrum data meets the spectrum requirement.

[0042] Specifically, in this embodiment, the spectral requirements of the target wafer are first determined based on the established mapping relationship. The spectral requirements refer to the specific requirements for the spectral characteristics (such as wavelength range, intensity distribution, etc.) of the UV light source in a specific application scenario according to the material characteristics of the target wafer and the purpose of detection, so as to ensure that the UV light source can provide a spectrum that matches the material characteristics of the target wafer, thereby achieving effective detection and analysis. For example, in wafer detection, if the material characteristic of the target wafer is high-purity silicon, then the spectral requirement may be a spectrum with high intensity within a specific wavelength range in order to better stimulate the spectral characteristics of the silicon material. Then, the wavelength and intensity of the UV light source are adjusted according to the spectral requirements so that the spectrum it outputs is closer to the spectrum required by the target wafer. For example, if the target wafer requires a spectrum within the wavelength range of 300nm to 400nm, then the output wavelength of the UV light source needs to be adjusted within this range, and the intensity also needs to be adjusted accordingly. After preliminary adjustment, the second spectral data of the UV light source is obtained through a spectrometer and other equipment. Whether the spectrum of the UV light source meets the spectrum requirement of the target wafer is judged according to the second spectrum data. If there is a deviation between the second spectrum data and the spectrum requirement of the target wafer, the spectrum of the UV light source needs to be adjusted for the second time. Through cyclic adjustment and judgment, the spectrum output of the UV light source finally meets the spectrum requirement of the target wafer.

[0043] In an optional implementation of the present embodiment, the second spectral data is compared with the spectral requirement to determine the corresponding spectral deviation data; a spectral deviation feature vector is obtained by performing feature extraction on the spectral deviation data; the spectral deviation feature vector is input into a machine learning model, and an optimal spectral adjustment strategy is generated through model calculation; and the spectrum of the UV light source is adjusted according to the optimal spectral adjustment strategy.

[0044] Specifically, in this embodiment, the process of comparing the second spectral data with the spectral demand to determine the spectral deviation data first requires a detailed analysis of the second spectral data, and by comparing with the preset spectral demand, the difference between the two, i.e., the spectral deviation data, is calculated. This comparison process can use a variety of methods, such as Euclidean distance, Mahalanobis distance, etc. For example, the Euclidean distance can be obtained by calculating the square root of the sum of the squares of the difference between the two spectral data at each wavelength. In this way, the difference between the second spectral data and the spectral demand can be quantified to determine the spectral deviation data. Feature extraction can reduce the high-dimensional spectral deviation data to a low-dimensional feature space and extract the most representative features. For example, PCA calculates the covariance matrix of the spectral deviation data and finds the direction with the largest variance in the data, i.e., the principal component, so that the data is projected onto these principal components to obtain the spectral deviation feature vector. This process not only reduces the dimension of the data, but also retains the main information of the data, providing effective features for subsequent machine learning model input. The machine learning model can be a support vector machine (SVM), a neural network (NN), a decision tree (DT), etc. These models can predict the optimal spectral adjustment strategy based on the input eigenvector by learning the relationship between the spectral deviation feature vector and the spectral adjustment strategy. For example, the neural network can learn complex nonlinear relationships through multiple layers of neurons and activation functions, thereby more accurately predicting the optimal spectral adjustment strategy. The optimal spectral adjustment strategy generated by the model is applied to the control system of the UV light source to accurately adjust the spectrum of the UV light source. For example, if the optimal spectral adjustment strategy indicates that the spectral intensity of a specific wavelength needs to be increased, the control system of the UV light source will adjust the corresponding parameters, such as current, voltage, etc., to achieve an increase in spectral intensity. In this way, it can be ensured that the spectral output of the UV light source meets the spectral requirements of the target wafer, thereby improving the accuracy and efficiency of wafer inspection.

[0045] In an optional implementation of the present embodiment, a fuzzy spectral feature vector is generated by fuzzifying the spectral feature vector; a fuzzy rule is formulated according to the mapping relationship; the fuzzy spectral feature vector is compared with the fuzzy rule to determine the fuzzy control output; the fuzzy control output is converted into a spectral adjustment parameter through defuzzification; and the spectrum of the UV light source is adjusted according to the spectral adjustment parameter.

[0046] Specifically, in this embodiment, the fuzzy spectral feature vector is generated by fuzzifying the spectral feature vector. Fuzzification is to convert the precise spectral feature vector into a fuzzy set. The spectral feature vector can be expressed as The goal of fuzzy processing is to transform each feature parameter Convert to its corresponding fuzzy set, fuzzy set membership function It can be expressed in the form of triangle, trapezoid or Gaussian function. For example, using Gaussian function as the fuzzy membership function, it can be expressed as: , in, is the characteristic parameter The central value of is the standard deviation of the fuzzy set. Through this formula, each characteristic parameter can be Convert it to its fuzzy value to generate a fuzzy spectral feature vector , fuzzy rules define the relationship between the fuzzy spectrum feature vector and the output spectrum adjustment parameters. Usually, fuzzy rules can be expressed in an "if-then" form, for example: "If is high and is low, then the output spectrum adjustment parameters These rules can be formulated by expert knowledge or data-driven methods. In the data-driven method, historical data and machine learning algorithms can be used to automatically generate fuzzy rules. Then, the fuzzy spectral feature vector is compared with the fuzzy rules to determine the fuzzy control output. The comparison process is to compare the fuzzy spectral feature vector Match the rules in the fuzzy rule base and calculate the activation of each rule. Assuming there are M fuzzy rules, the activation of each rule can be calculated by the intersection of fuzzy sets. For example, for the jth fuzzy rule, its activation It can be expressed as: , By calculating the activation degree of each rule, the fuzzy control output can be determined Defuzzification is the process of converting the fuzzy control output Converted into an accurate spectral adjustment parameter Y. Common defuzzification methods include the centroid method, the maximum membership method, etc. The centroid method determines the exact value by calculating the centroid of the fuzzy set, and its formula can be expressed as: , in, is the member function of the fuzzy control output. Through this formula, the accurate spectrum adjustment parameter Y can be calculated. Finally, the spectrum of the UV light source is adjusted according to the spectrum adjustment parameter. The spectrum adjustment parameter Y can include specific parameters such as the intensity, wavelength, and bandwidth of the light source. By adjusting the control parameters such as the current and temperature of the UV light source, the output spectrum of the light source can be adjusted in real time.

[0047] According to the adaptive spectrum adjustment method of wafer inspection UV light source provided by the scheme of the present application, the first spectrum data for preliminary spectrum inspection of the target wafer is obtained; the first spectrum data is subjected to variational mode decomposition processing to determine the spectrum characteristic parameters of the first spectrum data; the spectrum characteristic parameters are processed by refined composite multi-scale weighted multi-dimensional feature entropy to determine the spectrum characteristic vector of the spectrum characteristic parameters; the mapping relationship between the spectrum characteristics and the wafer material characteristics is determined by analyzing the spectrum characteristic vector; the spectrum of the UV light source is adaptively adjusted according to the mapping relationship. Through the implementation of the scheme of the present application, the spectrum of the UV light source is adaptively adjusted according to the mapping relationship between the spectrum characteristics and the wafer material characteristics, which can ensure that the spectrum characteristics of the light source match the wafer inspection requirements, avoid unnecessary spectrum redundancy or loss, improve spectrum utilization, shorten the inspection time, and thus significantly improve the inspection efficiency.

[0048] Figure 2 The present application provides an adaptive spectrum adjustment device for a UV light source for wafer inspection, which can be used to implement the adaptive spectrum adjustment method for a UV light source for wafer inspection in the above-mentioned embodiment. Figure 2 As shown, the adaptive spectrum adjustment device of the wafer inspection UV light source mainly includes: An acquisition module 10 is used to acquire first spectrum data for preliminary spectrum detection of a target wafer; A first determination module 20, configured to perform variational mode decomposition processing on the first spectral data to determine spectral characteristic parameters of the first spectral data; A second determination module 30 is used to process the spectral characteristic parameters by refining the composite multi-scale weighted multi-dimensional feature entropy to determine the spectral characteristic vector of the spectral characteristic parameters; A third determination module 40 is used to determine a mapping relationship between spectral features and wafer material properties by analyzing the spectral feature vector; The control module 50 is used to adaptively adjust the spectrum of the UV light source according to the mapping relationship.

[0049] In an optional implementation of the present embodiment, the first determination module is specifically used to: decompose the first spectral data into N modal components through a variational modal decomposition algorithm; wherein N is an integer greater than or equal to 1; analyze each of the modal components to determine the modal characteristic parameters of each of the modal components; integrate the modal characteristic parameters of the N modal components to determine the spectral characteristic parameters of the first spectral data.

[0050] In an optional implementation of the present embodiment, the second determination module is specifically used to: obtain multi-scale spectral features corresponding to the spectral characteristic parameters by determining the entropy values ​​of the spectral characteristic parameters at different scales; obtain weighted spectral features corresponding to the spectral characteristic parameters by weighted processing the multi-scale spectral features; and determine the spectral feature vector of the spectral characteristic parameters by performing feature extraction on the weighted spectral features.

[0051] In an optional implementation of the present embodiment, the third determination module is specifically used to: generate a spectral feature cluster by performing cluster analysis on the spectral feature vector; wherein the spectral feature cluster is a clustering of wafer materials with similar spectral features; perform statistical analysis on the spectral feature vectors corresponding to the spectral feature cluster to determine the wafer material characteristics corresponding to the spectral feature cluster; and establish a mapping relationship between spectral features and wafer material characteristics based on the spectral feature cluster and the corresponding wafer material characteristics.

[0052] In an optional implementation of the present embodiment, the control module is specifically used to: determine the spectral requirement of the target wafer according to the mapping relationship; preliminarily adjust the spectrum of the UV light source according to the spectral requirement; obtain second spectral data of the UV light source after the preliminary adjustment; determine whether the second spectral data meets the spectral requirement of the target wafer according to the second spectral data and the mapping relationship; if the second spectral data does not meet the spectral requirement, perform a second adjustment on the spectrum of the UV light source until the second spectral data meets the spectral requirement.

[0053] In an optional implementation of the present embodiment, the control module is also used to: compare the second spectral data with the spectral requirement to determine the corresponding spectral deviation data; obtain a spectral deviation feature vector by performing feature extraction on the spectral deviation data; input the spectral deviation feature vector into a machine learning model, and generate an optimal spectral adjustment strategy through model calculation; and adjust the spectrum of the UV light source according to the optimal spectral adjustment strategy.

[0054] In an optional implementation of the present embodiment, the control module is also used to: generate a fuzzy spectral feature vector by fuzzifying the spectral feature vector; formulate a fuzzy rule based on the mapping relationship; compare the fuzzy spectral feature vector with the fuzzy rule to determine the fuzzy control output; convert the fuzzy control output into a spectral adjustment parameter through defuzzification; and adjust the spectrum of the UV light source according to the spectral adjustment parameter.

[0055] According to the adaptive spectrum adjustment device of a wafer detection UV light source provided by the scheme of the present application, the first spectrum data for preliminary spectrum detection of the target wafer is obtained; the first spectrum data is subjected to variational mode decomposition processing to determine the spectrum characteristic parameters of the first spectrum data; the spectrum characteristic parameters are processed by refined composite multi-scale weighted multi-dimensional feature entropy to determine the spectrum characteristic vector of the spectrum characteristic parameters; the mapping relationship between the spectrum characteristics and the wafer material characteristics is determined by analyzing the spectrum characteristic vector; the spectrum of the UV light source is adaptively adjusted according to the mapping relationship. Through the implementation of the scheme of the present application, the spectrum of the UV light source is adaptively adjusted according to the mapping relationship between the spectrum characteristics and the wafer material characteristics, which can ensure that the spectrum characteristics of the light source match the wafer detection requirements, avoid unnecessary spectrum redundancy or loss, improve spectrum utilization, shorten detection time, and thus significantly improve detection efficiency.

[0056] According to the application plan provided Figure 3 An electronic device provided in an embodiment of the present application. The electronic device can be used to implement the adaptive spectrum adjustment method of the wafer inspection UV light source in the aforementioned embodiment, mainly comprising: The memory 301, the processor 302, and the computer program 303 stored in the memory 301 and executable on the processor 302, the memory 301 and the processor 302 are connected by communication. When the processor 302 executes the computer program 303, the adaptive spectrum adjustment method of the UV light source for wafer inspection in the aforementioned embodiment is implemented. The number of processors may be one or more.

[0057] The memory 301 may be a high-speed random access memory (RAM) memory, or a non-volatile memory (non-volatile memory), such as a disk memory. The memory 301 is used to store executable program codes, and the processor 302 is coupled to the memory 301 .

[0058] Furthermore, the present application also provides a computer-readable storage medium, which may be provided in the electronic device in the above embodiments. Figure 3 Memory in the illustrated embodiment.

[0059] The computer readable storage medium stores a computer program, and when the program is executed by the processor, the adaptive spectrum adjustment method of the wafer inspection UV light source in the aforementioned embodiment is implemented. Furthermore, the computer storable medium can also be a U disk, a mobile hard disk, a read-only memory (ROM), a RAM, a magnetic disk or an optical disk, etc., which can store program codes.

[0060] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0061] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions to enable a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk, etc., various media that can store program codes.

[0062] As described above, the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application 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 embodiments of the present application.

Claims

1. A method for adaptive spectrum adjustment of UV light source for wafer inspection, characterized in that: include: Acquiring first spectrum data for preliminary spectrum detection of a target wafer; performing variational mode decomposition processing on the first spectral data to determine spectral characteristic parameters of the first spectral data; Processing the spectral characteristic parameters by refining composite multi-scale weighted multi-dimensional feature entropy to determine a spectral characteristic vector of the spectral characteristic parameters; Determine a mapping relationship between spectral features and wafer material properties by analyzing the spectral feature vector; The spectrum of the UV light source is adaptively adjusted according to the mapping relationship.

2. The method for adaptive spectrum adjustment of wafer inspection UV light source according to claim 1, characterized in that: The step of performing variational mode decomposition processing on the first spectral data to determine the spectral characteristic parameters of the first spectral data includes: Decomposing the first spectral data into N modal components by a variational modal decomposition algorithm; wherein N is an integer greater than or equal to 1; Analyzing each of the modal components to determine the modal characteristic parameters of each of the modal components; The modal characteristic parameters of the N modal components are integrated to determine the spectral characteristic parameters of the first spectral data.

3. The method for adaptive spectrum adjustment of wafer inspection UV light source according to claim 1, characterized in that: The step of processing the spectral characteristic parameters by refining the composite multi-scale weighted multi-dimensional feature entropy to determine the spectral characteristic vector of the spectral characteristic parameters comprises: Obtaining multi-scale spectral features corresponding to the spectral characteristic parameters by determining entropy values ​​of the spectral characteristic parameters at different scales; Obtaining weighted spectral features corresponding to the spectral characteristic parameters by weighting the multi-scale spectral features; The spectral feature vector of the spectral characteristic parameter is determined by extracting the weighted spectral feature.

4. The method for adaptive spectrum adjustment of wafer inspection UV light source according to claim 1, characterized in that: The step of determining the mapping relationship between the spectral feature and the wafer material property by analyzing the spectral feature vector comprises: Generating a spectral feature cluster by performing cluster analysis on the spectral feature vector; wherein the spectral feature cluster is a cluster of wafer materials with similar spectral features; Performing statistical analysis on the spectral feature vectors corresponding to the spectral feature clusters to determine the wafer material properties corresponding to the spectral feature clusters; According to the spectral feature cluster and the corresponding wafer material characteristics, a mapping relationship between the spectral features and the wafer material characteristics is established.

5. The method for adaptive spectrum adjustment of wafer inspection UV light source according to claim 4, characterized in that: The step of adaptively adjusting the spectrum of the UV light source according to the mapping relationship comprises: Determining the spectrum requirement of the target wafer according to the mapping relationship; Preliminarily adjusting the spectrum of the UV light source according to the spectrum requirement; Acquire second spectrum data of the UV light source after preliminary adjustment; determining, according to the second spectral data and the mapping relationship, whether the second spectral data meets the spectral requirement of the target wafer; If the second spectrum data does not meet the spectrum requirement, the spectrum of the UV light source is adjusted for a second time until the second spectrum data meets the spectrum requirement.

6. The method for adaptive spectrum adjustment of wafer inspection UV light source according to claim 5, characterized in that: The method further comprises: Comparing the second spectral data with the spectral requirement to determine corresponding spectral deviation data; Acquire a spectral deviation feature vector by performing feature extraction on the spectral deviation data; Inputting the spectral deviation feature vector into a machine learning model, and generating an optimal spectral adjustment strategy through model calculation; The spectrum of the UV light source is adjusted according to the optimal spectrum adjustment strategy.

7. The method for adaptive spectrum adjustment of wafer inspection UV light source according to claim 1, characterized in that: The method further comprises: Generate a fuzzy spectral feature vector by fuzzifying the spectral feature vector; Formulate fuzzy rules according to the mapping relationship; Comparing the fuzzy spectrum feature vector with the fuzzy rule to determine the fuzzy control output; Converting the fuzzy control output into a spectrum adjustment parameter by defuzzification processing; The spectrum of the UV light source is adjusted according to the spectrum adjustment parameter.

8. An adaptive spectrum adjustment device for wafer inspection UV light source, characterized in that: The self-adaptive spectrum adjustment device of the wafer inspection UV light source comprises: An acquisition module, used for acquiring first spectrum data for preliminary spectrum detection of a target wafer; a first determining module, configured to perform variational mode decomposition processing on the first spectral data to determine spectral characteristic parameters of the first spectral data; A second determination module is used to process the spectral characteristic parameters by refining the composite multi-scale weighted multi-dimensional feature entropy to determine the spectral characteristic vector of the spectral characteristic parameters; A third determination module is used to determine the mapping relationship between the spectral feature and the wafer material property by analyzing the spectral feature vector; The control module is used to adaptively adjust the spectrum of the UV light source according to the mapping relationship.

9. An electronic device, characterized in that: The device comprises a memory and a processor, wherein: The processor is used to execute the computer program stored in the memory; When the processor executes the computer program, the steps in the method for adaptive spectrum adjustment of a UV light source for wafer inspection described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps in the method for adaptive spectrum adjustment of a UV light source for wafer inspection described in any one of claims 1 to 7 are implemented.

Images