Self-adaptive regulation and control method and device for brightness of wafer detection light source

Abstract

The invention relates to the technical field of light source control, and discloses a self-adaptive regulation and control method and device for wafer detection light source brightness, and the method comprises the steps: carrying out the preliminary irradiation of a wafer through a preset annular light source, collecting initial image data, carrying out the digital simulation according to the initial detection data, and generating a wafer detection feedback model; analyzing a light source irradiation demand according to the detection feedback model, generating a gradient execution scheme, sequentially configuring light source parameters according to the gradient scheme, carrying out multiple rounds of detection on a wafer, collecting data, feeding detection data back to the model, carrying out parameter adjustment, carrying out expanded light source detection demand analysis on the adjusted model, and generating an expanded detection scheme. According to the method, by automatically adjusting the brightness and configuration of the light source, efficient and accurate wafer surface defect detection is achieved, the detection precision and efficiency are improved, the method adapts to the detection requirements of different wafers, and the problem that in the prior art, due to the fact that the light source used for detection is not comprehensive enough, the defect is not detected is solved.

Description

Technical Field

[0001] The present invention relates to the technical field of light source control, and in particular to a method and device for adaptively controlling the brightness of a wafer inspection light source. Background Art

[0002] As the core material in semiconductor manufacturing, the surface quality of wafers directly affects the performance and reliability of chips. There may be various defects on the surface of wafers, including tiny cracks, scratches, contaminants, etc. These defects are often difficult to detect through traditional detection methods. The light source is one of the most critical elements in wafer surface detection. Different types of light sources (such as ring light sources, laser light sources, polarized light sources, etc.) will produce different reflection characteristics when irradiating the wafer surface, thereby revealing different defect information. Parameters such as light source brightness, angle and distribution need to be adjusted according to the actual state of the wafer to achieve the best detection effect.

[0003] In the actual wafer inspection process, the adjustment of the light source brightness is usually based on experience, which may result in some potential defects not being detected or excessive reflection causing overexposure of the image. Traditional methods cannot automatically adjust the light source brightness according to the real-time reflection of the wafer surface, and it is difficult to meet the detection needs of different types of defects. Summary of the invention

[0004] The purpose of the present invention is to provide a method and device for adaptively controlling the brightness of a wafer inspection light source, aiming to solve the problem in the prior art that the light source used for inspection is not comprehensive enough, resulting in undetected defects.

[0005] The present invention is implemented in this way. In a first aspect, the present invention provides a method for adaptively controlling the brightness of a wafer inspection light source, comprising: Using a pre-set annular light source to illuminate the wafer to be inspected in a surrounding state with an initial light source, and collecting image data of the wafer to be inspected under the illumination of the initial light source in the surrounding state, so as to obtain initial inspection data of the wafer to be inspected; Digitally simulating the wafer to be inspected according to the initial inspection data of the wafer to be inspected to obtain the wafer inspection feedback model, and performing a gradient scheme analysis of subsequent light source illumination requirements on the ring light source based on the wafer inspection feedback model to obtain a light source illumination gradient execution scheme; According to the light source illumination gradient execution scheme, the light source parameters of the annular light source are sequentially configured, so that the annular light source sequentially applies the detection light source illumination to the wafer to be inspected, thereby obtaining the gradient detection data of the wafer to be inspected under each round of detection light source illumination; Substituting the gradient detection data of the wafer to be inspected under each round of detection light source into the wafer detection feedback model to adjust the model parameters of the wafer detection feedback model; An extended light source detection demand analysis is performed based on the wafer detection feedback model after model parameter adjustment to obtain an extended light source detection scheme for the wafer to be inspected, and the wafer to be inspected is inspected based on the extended light source detection scheme to obtain extended detection data and substitute it into the wafer detection feedback model.

[0006] In a second aspect, the present invention provides an adaptive control device for the brightness of a wafer inspection light source, which is used to implement an adaptive control method for the brightness of a wafer inspection light source as described in any one of the first aspects, including: An initial detection module is used to irradiate the wafer to be detected with an initial light source in a surrounding state through a pre-set annular light source, and collect image data of the wafer to be detected under the irradiation of the initial light source in the surrounding state, so as to obtain initial detection data of the wafer to be detected; A scheme analysis module, used for digitally simulating the wafer to be inspected according to the initial inspection data of the wafer to be inspected to obtain the wafer inspection feedback model, and performing a gradient scheme analysis of the subsequent light source illumination requirements on the annular light source based on the wafer inspection feedback model to obtain a light source illumination gradient execution scheme; A gradient detection module, used to sequentially configure light source parameters of the annular light source according to the light source illumination gradient execution scheme, so that the annular light source sequentially applies detection light source illumination to the wafer to be inspected, thereby obtaining gradient detection data of the wafer to be inspected under each round of detection light source illumination; A data feedback module, used for substituting the gradient detection data of the wafer to be detected under each round of detection light source into the wafer detection feedback model, so as to adjust the model parameters of the wafer detection feedback model; The extended detection module is used to perform an extended light source detection demand analysis based on the wafer detection feedback model after model parameter adjustment to obtain an extended light source detection solution for the wafer to be detected, and to detect the wafer to be detected according to the extended light source detection solution to obtain extended detection data and substitute it into the wafer detection feedback model.

[0007] The present invention provides a method for adaptively controlling the brightness of a wafer inspection light source, which has the following beneficial effects: The present invention performs preliminary irradiation on the wafer with a preset annular light source and collects initial image data, performs digital simulation based on the initial detection data, generates a wafer detection feedback model, analyzes the light source irradiation requirements based on the detection feedback model, generates a gradient execution plan, configures the light source parameters in sequence according to the gradient plan, performs multiple rounds of detection on the wafer and collects data, feeds the detection data back to the model, performs parameter adjustment, performs expanded light source detection demand analysis on the adjusted model, generates an expanded detection plan, and performs final detection. The method achieves efficient and accurate wafer surface defect detection by automatically adjusting the light source brightness and configuration, improves detection accuracy and efficiency, adapts to the detection requirements of different wafers, and solves the problem in the prior art that the light source used for detection is not comprehensive enough, resulting in undetected defects. BRIEF DESCRIPTION OF THE DRAWINGS

[0008] Figure 1 It is a schematic diagram of the steps of a method for adaptively controlling the brightness of a wafer inspection light source provided by an embodiment of the present invention; Figure 2 It is a structural schematic diagram of a device for adaptively controlling the brightness of a wafer inspection light source provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0009] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0010] The implementation of the present invention is described in detail below in conjunction with specific embodiments.

[0011] Reference Figure 1 , Figure 2 As shown, a preferred embodiment of the present invention is provided.

[0012] In a first aspect, the present invention provides a method for adaptively controlling the brightness of a wafer inspection light source, comprising: S1: irradiating a wafer to be inspected with an initial light source in a surrounding state through a pre-set annular light source, and collecting image data of the wafer to be inspected under the irradiation of the initial light source in the surrounding state, so as to obtain initial inspection data of the wafer to be inspected; S2: digitally simulating the wafer to be inspected according to the initial inspection data of the wafer to be inspected to obtain the wafer inspection feedback model, and performing a gradient scheme analysis of subsequent light source illumination requirements on the ring light source based on the wafer inspection feedback model to obtain a light source illumination gradient execution scheme; S3: configuring the light source parameters of the annular light source in sequence according to the light source illumination gradient execution scheme, so that the annular light source applies the detection light source illumination to the wafer to be inspected in sequence, thereby obtaining the gradient detection data of the wafer to be inspected under each round of detection light source illumination; S4: Substituting the gradient detection data of the wafer to be inspected under each round of detection light source into the wafer inspection feedback model to adjust the model parameters of the wafer inspection feedback model; S5: Perform an extended light source detection demand analysis based on the wafer detection feedback model after model parameter adjustment to obtain an extended light source detection solution for the wafer to be detected, and detect the wafer to be detected according to the extended light source detection solution to obtain extended detection data and substitute it into the wafer detection feedback model.

[0013] Specifically, in step S1 of the embodiment provided by the present invention, a ring light source is pre-installed and set. This light source usually surrounds the wafer to form a uniform lighting environment. The brightness, angle, position, etc. of the ring light source can be adjusted as needed to ensure that the entire surface of the wafer can be covered. The wafer to be inspected is placed within the illumination range of the light source to ensure that the surface of the wafer can be evenly illuminated by the ring light source. When the ring light source illuminates the wafer, a surrounding illumination method is adopted, that is, the light source is evenly distributed around the wafer to ensure that the light intensity at each angle is basically consistent.

[0014] It can be understood that the configuration of the ring light source can provide uniform lighting, avoiding the shadows and uneven illumination problems that may be caused by traditional light sources, thereby improving the quality of image acquisition and ensuring the comprehensiveness of image data. The ring light source can ensure that the wafer surface is evenly illuminated at all angles, reducing the omission or misjudgment of defects due to insufficient local lighting. Compared with traditional point light sources, the ring light source can illuminate the wafer surface from multiple angles, which helps to improve the brightness uniformity of the preliminary image and reduce the impact on detection.

[0015] More specifically, a high-resolution camera or optical inspection equipment is used to ensure that the tiny structures and defects on the wafer surface can be accurately captured. Under the illumination of a ring-shaped light source, image data of the wafer surface is collected to obtain detailed image information containing various areas of the wafer. These image data usually include surface texture, defects, blemishes and other features of the wafer. The collected image data is transmitted to a processing system for subsequent image analysis and processing. Due to the uniformity of the light source and sufficient illumination of the wafer surface, the image acquisition process can obtain high-quality, high-contrast images, thereby improving the detection accuracy of wafer defects. The design of the ring-shaped light source ensures illumination of every corner of the wafer surface, avoids detection blind spots caused by blind angles of illumination, and ensures that image data of all areas of the wafer surface are effectively collected.

[0016] More specifically, the collected image data is stored as raw image data files, usually in the format of image files (such as TIFF, JPEG, PNG, etc.) or image data sets (such as matrix data), etc. These image data will serve as the basic data for subsequent detection and analysis. Image processing, defect identification and other steps will be performed later. Based on the collected image data, the image quality is evaluated to check whether there are problems such as blur and noise, so as to decide whether it is necessary to adjust the light source parameters or re-collect data. The image data collection at this stage provides basic data for subsequent defect detection and light source optimization. Subsequent steps will further optimize the light source irradiation and detection methods based on these initial data. The collected initial data can not only be directly used for subsequent defect analysis, but also provide strong support for establishing a model of wafer surface features.

[0017] It can be understood that through the initial irradiation of the ring light source and the image data collection, the technical effect of the entire step is: through the uniform illumination of the ring light source, the lighting conditions during image collection are ensured to be uniform, and the impact of shadows and uneven lighting on the image is minimized, thereby improving the quality and reliability of the detected image. The ring light source ensures that every area of ​​the wafer can be fully illuminated, thereby avoiding the blind spots that may exist in traditional light sources. The collection of preliminary image data is the basis for all subsequent light source adjustments and defect detection, ensuring the accuracy and efficiency of subsequent steps.

[0018] Specifically, in step S2 of the embodiment provided by the present invention, the image data collected by the preliminary ring light source irradiation is used as input for analysis by the digital simulation program. At this time, the image data may contain information such as defects, textures, and material unevenness on the wafer surface. Advanced simulation software or algorithms (such as finite element analysis, optical simulation, etc.) are used to model the wafer surface according to the input initial detection data to generate a virtual wafer surface model. Based on the wafer surface model, the image response under different light source irradiation conditions is simulated to obtain a wafer detection feedback model. This model shows the response characteristics of the wafer under different lighting conditions, including factors such as visibility, brightness, and contrast of defects. The virtual model generated by digital simulation can accurately reflect the real physical properties of the wafer, making subsequent light source adjustment and defect detection more accurate. Simulating detection feedback under different light source conditions helps to evaluate which areas of defects are difficult to identify under existing lighting conditions, thereby providing a basis for subsequent light source adjustments. The wafer detection feedback model provides a theoretical basis for formulating subsequent light source control plans, which can effectively optimize light source parameters and avoid unnecessary light waste.

[0019] More specifically, based on the generated wafer inspection feedback model, the illumination requirements of different areas on the wafer surface are analyzed. For example, some areas may require stronger light due to the nature of the surface structure or defects, while other areas may only require weaker light to avoid overexposure or over-illumination. According to the regional requirements fed back by the model, a gradient scheme for light source illumination is formulated. This gradient scheme will specify in detail the light source intensity, illumination angle and other parameters required for each area. Specifically, the light intensity required for different areas may be different, and it needs to be optimized by adjusting the brightness of the light source. The illumination angle of different areas can be fine-tuned according to the different defect types. For example, some defects are more obvious under oblique illumination. By simulating the impact of different gradient schemes on the wafer, the visibility and detection effect of wafer defects under each scheme are evaluated, and finally the optimal light source illumination gradient execution scheme is determined.

[0020] It is understandable that through precise analysis and gradient adjustment of the lighting requirements of different areas of the wafer, the problem of over-irradiation or uneven lighting caused by global uniform light source illumination can be avoided, ensuring that each area receives the most suitable lighting. Certain small or weak defects are more easily found under specific lighting conditions. By adjusting the gradient light source illumination scheme, the detection sensitivity of these tiny defects can be effectively improved. By flexibly adjusting the light source illumination intensity and angle, overexposure of some areas can be avoided, or insufficient lighting in other areas can be avoided, thereby maximizing the accuracy of defect detection.

[0021] More specifically, based on the previous analysis of the light source gradient scheme, a specific light source irradiation gradient execution plan is formed. The plan should list in detail the parameters such as the light source intensity, angle, irradiation time, etc. required for each area, and provide implementation details. The gradient execution plan is applied in the actual detection environment, and its effect is verified through experiments. If it is found that there are still undetected defects in some areas, it may be necessary to further adjust the light source irradiation plan. After several rounds of adjustment and optimization, the final gradient execution plan is confirmed to ensure that defects in each area can be accurately detected while avoiding false alarms and missed alarms. Through the execution of precise gradient light source schemes, the inspection of wafer surfaces can be completed quickly and efficiently, and it can adapt to the surface characteristics and defect types of different wafers. By optimizing the light source irradiation method, the detection accuracy of different defect types and different areas has been significantly improved, ensuring that all defects (regardless of size or type) can be effectively captured.

[0022] It can be understood that through the analysis and adjustment of the gradient light source scheme, each wafer area can be properly illuminated according to actual needs, avoiding the problem of insufficient or excessive illumination caused by unified light source illumination in traditional methods. Digital simulation and feedback model optimization under different lighting conditions enable tiny and complex defects to be clearly presented under appropriate lighting conditions, thereby improving the accuracy of detection. This method can dynamically adjust the light source parameters according to actual detection needs, adapt to the detection requirements of different types of wafers and defects, and has strong flexibility and adaptability. Through preliminary model simulation and gradient scheme analysis, unnecessary detection steps are reduced, ultimately improving the efficiency and accuracy of the entire detection process.

[0023] Specifically, in step S3 of the embodiment provided by the present invention, according to the determined gradient light source illumination scheme, the illumination parameters of the light source are gradually adjusted, mainly including: light source intensity (brightness): adjusting the brightness of the light source according to the needs of different areas to ensure that the wafer surface in different areas is most appropriately illuminated; illumination angle: adjusting the angle between the light source and the wafer surface to ensure the illumination effect of different areas, especially the effect of revealing surface defects; illumination time: adjusting the duration of light source illumination according to the required exposure time to ensure that each area is adequately illuminated without overexposure or underexposure.

[0024] More specifically, the annular light source is adjusted successively according to the set gradient scheme and irradiates the wafer surface in turn. During each round of light source irradiation, the parameters of the light source will change according to the gradient scheme to ensure that each area receives different irradiation conditions. After each round of light source irradiation, a detection device (such as an optical imaging system or a camera) is used to capture the light signal reflected or transmitted by the wafer surface to obtain image data or other detection data of the wafer under different irradiation conditions. These data will reflect the response of the wafer surface under different light sources, including defect information, surface unevenness, reflectivity, etc. The light source parameters are gradually adjusted and optimized according to the needs of each area, ensuring that the wafer surface of each area is irradiated and detected under the best conditions, thereby obtaining more accurate detection data. Through multiple irradiations under different light source conditions, different defect types and manifestations on the wafer surface can be covered, thereby increasing the comprehensiveness of defect detection. By precisely adjusting the intensity, angle and time of the light source, it can adapt to the detection needs of different wafers and different defect types, thereby realizing a personalized customized detection process.

[0025] More specifically, after each light source irradiation, the detection system (such as a CCD camera or optical sensor) will collect light signals reflected or transmitted from the wafer surface in real time. These signals will be converted into image data or other forms of feedback data. The collected image or light signal data will be preliminarily processed and calibrated to remove noise and enhance key information such as image contrast and brightness to ensure that defects are clearly visible. The data after each round of irradiation will be recorded and stored for subsequent analysis and comparison. The parameters of each irradiation (such as light source intensity, angle, and time) will be stored together with the corresponding detection data to form a complete gradient detection data set.

[0026] More specifically, through data collection under multiple gradient illumination schemes, multi-dimensional detection data can be obtained, including reflection, transmission, contrast and other information under different illumination conditions. These data can provide a comprehensive reference for subsequent analysis. By gradually adjusting the light source parameters, it can ensure that the key information of the wafer surface can be extracted to the greatest extent in each round of illumination, improving the quality of defect identification. Data under different light sources can help identify different types of defects, such as tiny cracks, surface contamination, non-uniform areas, etc., and can more accurately classify and locate wafer surface defects.

[0027] More specifically, by comparing the data collected under different light source conditions, we analyze which illumination schemes have more obvious defects on the wafer surface, and which areas may have missed detection or false detection. Based on the results of data analysis, we further adjust the light source illumination gradient, improve the distribution of light source intensity, angle or time, and optimize the detection effect. Through experiments and data feedback, we continuously improve the detection process to achieve the best detection results. We use the data collected under each round of gradient light source illumination, combined with defect recognition algorithms (such as image processing, machine learning, etc.) to identify and locate defects, and further extract the defect characteristics of the wafer surface.

[0028] More specifically, data comparison and analysis under different light source conditions can help accurately identify different types of defects, especially those tiny defects or complex defects that are not easy to find under conventional light sources. Through the feedback of gradient light source irradiation data, self-optimization and dynamic adjustment can be achieved, so that the light source configuration can more accurately adapt to the surface characteristics and defect types of different wafers, thereby improving the overall detection capability. Through rounds of gradient irradiation and multiple data comparisons, the occurrence of missed detections and false detections can be effectively reduced, thereby improving the reliability and accuracy of detection.

[0029] It is understandable that through multiple rounds of irradiation and data collection under different light source conditions, the wafer surface can be fully covered without missing any potential defect areas, ensuring that all defects can be discovered. The ring light source gradually adjusts the light source parameters according to the gradient execution plan, accurately controls the light source irradiation, avoids unnecessary overexposure or insufficient irradiation, and improves the accuracy of wafer defect detection. Through repeated light source adjustment and data feedback, the detection process is continuously optimized, the detection efficiency is improved, and the best detection effect is ensured. Based on the gradient data under different light source irradiation conditions, various defects can be identified more clearly and accurately, especially the detection of small or complex defects.

[0030] Specifically, in step S4 of the embodiment provided by the present invention, from the previous step, by adjusting the light source parameters one by one and performing irradiation, the wafer surface detection data (such as reflectivity, transmission image, optical image, etc.) under different light source conditions are obtained, and the acquired gradient detection data is subjected to necessary preprocessing, including noise removal, image enhancement (such as contrast adjustment, smoothing), standardization, etc. This process ensures that the data input to the model is of high quality and consistent, and the processed gradient detection data is input into the wafer detection feedback model as the input feature of the model. At this time, the gradient data is not only the original information of each round of detection, but also includes information related to the environmental conditions and detection parameters under each round of light source irradiation. The preprocessed gradient detection data ensures the quality of the model input data, thereby improving the effect of subsequent model training and adjustment. By substituting the data under multiple rounds of light source irradiation into the feedback model, multi-dimensional and multi-angle information is provided, which helps to more comprehensively characterize the defect characteristics of the wafer.

[0031] More specifically, the wafer inspection feedback model is usually initialized based on some preliminary model parameters (such as thresholds, weights, biases, etc.) to form an initial inspection state. This process can be based on historical data or preliminary assumptions. The input gradient inspection data is used for model prediction to obtain the defect probability or defect type prediction value for each area. The model prediction results are compared with the actual inspection results (if there is ground truth data), and the error (such as the deviation between the predicted defect location and the actual defect location, the false detection rate, the missed detection rate, etc.) is calculated. Based on the error value, the model parameters are adjusted through the back propagation algorithm (in the deep learning model) or other optimization algorithms (such as gradient descent, genetic algorithm, etc.). The update process usually includes adjusting the model parameters such as weights, biases, thresholds, etc. to reduce the prediction error.

[0032] More specifically, under multiple rounds of light source conditions, the model will continue to learn and adjust itself. Through repeated iterative optimization, the model's recognition accuracy of wafer surface defects will be gradually improved. The adjusted model needs to be verified by new data to confirm that the adjusted parameters can significantly improve the accuracy of defect detection. If errors are still found, continue to iteratively adjust the model parameters until the ideal detection effect is obtained. By substituting actual gradient detection data and combining error back propagation, the model can adaptively optimize its own parameters. With each round of adjustment, the model's detection capability will continue to improve. Through fine parameter adjustment, the model can more accurately identify defects under different lighting conditions and gradually improve its defect recognition capabilities in complex scenarios. With the continuous input of gradient data, the model can continuously adapt to different types of wafers, different defect types, different light source conditions and other changing factors, making the detection system have stronger dynamic adaptability.

[0033] More specifically, after each model adjustment, the new test results can be used as data feedback again for a second or even multiple rounds of learning. The input of each round of feedback data can further optimize the model and make it more accurate. Through cross-validation and other techniques, the generalization ability of the model is ensured not to be affected by overfitting. This process helps to improve the robustness of the model under different wafers and different light source conditions. After multiple rounds of iteration and optimization, the final model will be comprehensively evaluated based on multidimensional data (including data under different light source conditions) to test its performance in a real production environment, such as defect detection rate, missed detection rate, and false detection rate. Through repeated iterations and parameter adjustments, the recognition accuracy of the model will continue to improve, and ultimately high-precision defect detection can be achieved in a complex process environment. Through multiple rounds of gradient data feedback, the model can adapt to different process changes and handle wafer defects of different types and sizes, so that it can maintain good stability and adaptability in a constantly changing production environment. With the improvement of model accuracy, the detection system can identify defects on the wafer in a shorter time, reducing the detection time and the risk of misjudgment in production, thereby improving overall production efficiency.

[0034] It can be understood that by inputting the gradient detection data under each round of light source illumination into the feedback model and adjusting the model parameters, the recognition accuracy of wafer defects can be effectively improved, especially in complex situations with large changes in lighting conditions. The feedback model has the ability of self-learning and optimization by continuously absorbing gradient data and adjusting parameters, and can adapt to new process requirements or environmental changes. Through gradient data and parameter adjustment, the model can not only identify conventional defects, but also handle complex and minor defects, thereby improving the comprehensiveness and accuracy of detection. The optimized model can identify defects in real time and efficiently in the production line, reduce missed detections and false detections, and improve the overall production quality and efficiency. In summary, substituting gradient detection data into the wafer detection feedback model and adjusting the model parameters can improve the accuracy and efficiency of defect detection, optimize the detection process, and achieve dynamic adaptation, ultimately improving the accuracy and production efficiency of wafer surface defect detection.

[0035] Specifically, in step S5 of the embodiment provided by the present invention, based on the parameter-adjusted wafer detection feedback model, the detection effect of the existing light source conditions on the wafer surface defects is analyzed, which includes the evaluation of indicators such as defect detection accuracy, missed detection rate, and false detection rate under different light source conditions. By modeling the optical properties of the wafer (such as reflectivity, transmittance, gloss, etc.), it is determined which light source parameters (such as wavelength, light intensity, angle, etc.) have a greater impact on the identification of wafer defects. This step helps to find out the deficiencies of the current light source configuration and determine the light source characteristics that need to be expanded.

[0036] More specifically, based on multiple dimensions such as defect type, wafer material, and surface treatment process, we conduct a demand analysis for expanding the light source, and identify the types of light sources that need to be added and their corresponding parameters. For example, we may need to add light sources with specific wavelengths and different illumination angles, or a light source configuration for multi-band joint detection. Ultimately, we design a complete detection solution that can cover different light source conditions. Through demand analysis, we can fully understand the detection bottlenecks of the current light source conditions and provide a clear direction for the subsequent solution design. We determine the appropriate light source parameters for different defect types and different characteristics of wafer materials, and design a highly adaptable expanded light source detection solution.

[0037] More specifically, based on the results of the demand analysis, light sources with different wavelengths, different intensities, and different incident angles are selected for combination. These light sources include ultraviolet light, infrared light, white light, and other types. The specific selection of light sources needs to be combined with the material, defect type, and production process requirements of the wafer. The layout of the light source is designed according to the size and shape of the wafer to ensure that the light source can evenly illuminate the wafer surface. Usually, the distribution of the light source (such as annular, radial, etc.) and the change in the irradiation angle are considered to capture the detailed information of different areas on the wafer surface. According to the selected light source configuration, the corresponding detector (such as photodiode, CCD sensor, etc.) is designed to capture the light signal reflected from the wafer surface. The number and position of the detector should match the light source to obtain the best detection results. In the expansion of the light source solution, it may be necessary to synchronize or work in turns with multiple light sources to ensure that each light source can fully detect defects on the wafer surface at different time periods and different positions. At this time, the working frequency and synchronization mechanism of the light source need to be coordinated to avoid interference.

[0038] More specifically, by designing a variety of different types of light sources and layout schemes, the visibility of wafer surface defects can be improved, especially defects that are not easy to detect under conventional light sources (such as tiny cracks, surface scratches, etc.) will be more accurately identified. By precisely designing the light source layout, full coverage of all areas of the wafer surface is guaranteed, reducing the problem of missed detection due to uneven light source coverage.

[0039] More specifically, according to the extended light source detection scheme, different light sources are started, and the wafers are inspected one by one or synchronously. In each round of detection, the wavelength, intensity and incident angle of the light source are adjusted according to the designed light source parameters. In each round of detection, a detector is used to collect light signal data reflected from the wafer surface. These data usually include information such as light intensity, reflection image, transmission image, etc. The collected data is preprocessed, including denoising, standardization, contrast adjustment, etc. to ensure data quality. The extended data of each round of detection is stored in a database, and the corresponding light source conditions (such as light source type, wavelength, incident angle, etc.) are marked. By using the extended light source, multi-dimensional optical detection data (such as reflection and transmission data under different light sources) is obtained, which provides rich information for subsequent defect identification. The collected extended data can effectively supplement the defect characteristics that may be missed under conventional light sources, and enhance the comprehensive recognition capability of the entire detection system.

[0040] More specifically, the multi-dimensional detection data obtained from the extended light source detection is input into the adjusted wafer detection feedback model. At this time, the model will receive extended data including information such as light source type, wavelength, and incident angle. The model identifies and classifies defects based on the extended detection data and outputs defect information for each detection area. These results may include defect type, defect location, defect size, etc. Based on the defect identification results fed back by the extended detection data, the model can further self-optimize and adjust. For example, if the detection result under a certain light source condition is not ideal, the model can adjust the relevant parameters based on the feedback to improve the accuracy of the next round of detection. By substituting the extended light source data into the model, it can help the model better understand the defect characteristics under different light source conditions, thereby improving the accuracy and comprehensiveness of defect identification. The model can continuously adjust and optimize itself through feedback learning, adapt to a variety of light source configurations, and improve robustness and stability in different production environments.

[0041] It is understandable that by expanding the light source detection scheme, the model can comprehensively collect defect data on the wafer surface, including tiny defects that are difficult to detect under different light source conditions, thereby significantly improving the detection accuracy. The combination of expanding the light source detection scheme and the model feedback mechanism enables the detection system to flexibly adapt to different types of wafers, different defects and different production environments. Through a variety of light source configurations and data feedback, the system can more comprehensively identify various types of defects, especially those that are difficult to find under conventional light sources, such as surface microcracks, changes in transparency, etc. In summary, by expanding the light source detection demand analysis, designing an expanded light source detection scheme and combining it with the wafer detection feedback model, the accuracy, comprehensiveness and adaptability of wafer detection can be greatly improved, the detection efficiency of the production line can be optimized, and missed detection and false detection can be reduced.

[0042] The present invention provides a method for adaptively controlling the brightness of a wafer inspection light source, which has the following beneficial effects: The present invention performs preliminary irradiation on the wafer with a preset annular light source and collects initial image data, performs digital simulation based on the initial detection data, generates a wafer detection feedback model, analyzes the light source irradiation requirements based on the detection feedback model, generates a gradient execution plan, configures the light source parameters in sequence according to the gradient plan, performs multiple rounds of detection on the wafer and collects data, feeds the detection data back to the model, performs parameter adjustment, performs expanded light source detection demand analysis on the adjusted model, generates an expanded detection plan, and performs final detection. The method achieves efficient and accurate wafer surface defect detection by automatically adjusting the light source brightness and configuration, improves detection accuracy and efficiency, adapts to the detection requirements of different wafers, and solves the problem in the prior art that the light source used for detection is not comprehensive enough, resulting in undetected defects.

[0043] Preferably, the steps of irradiating the wafer to be inspected with an initial light source in a surrounding state by a pre-set annular light source, and collecting image data of the wafer to be inspected under the irradiation of the initial light source in the surrounding state to obtain initial inspection data of the wafer to be inspected include: S11: configuring initial brightness parameters and initial orientation parameters for each light source unit in a preset annular light source, so that each light source unit in the annular light source is in an initial irradiation state, and each light source unit in the initial irradiation state performs initial light source irradiation in a surrounding state on the wafer to be inspected; S12: Parameters of the camera position and the camera mode are configured for a plurality of pre-set camera modules, so that each of the camera modules collects image data of the wafer to be inspected, so as to obtain a detection image of the wafer to be inspected at each of the camera positions obtained by the camera modules in the camera mode, and each of the detection images together constitutes the initial detection data of the wafer to be inspected.

[0044] Specifically, according to the characteristics of the wafer to be inspected (such as material, surface reflection characteristics, etc.), the initial brightness parameters of each light source unit in the ring light source are adjusted. This can be achieved by adjusting the output power of each light source unit to ensure that the brightness of the light source can illuminate the surface of the wafer without causing overexposure. The illumination angle of each light source unit is adjusted so that the light source evenly illuminates the wafer surface from different angles to avoid shadows and dead angles caused by illumination from a single angle of light source, thereby improving the visibility of defects. Each light source unit is configured to be in an initial illumination state to ensure that the light source can illuminate the wafer evenly and continuously to form an initial light source illumination condition in a surrounding shape. By accurately setting the brightness and orientation of the light source unit, it can be ensured that each area of ​​the wafer surface is evenly illuminated, minimizing the problem of missed detection caused by shadows and uneven illumination. By adjusting the brightness and angle of the light source, the illumination method can be flexibly adjusted according to different wafer materials and surface optical characteristics to improve the effectiveness of defect identification.

[0045] More specifically, the installation position of the camera module is determined to ensure that the wafer can be fully photographed from multiple angles. According to the layout of the annular light source, camera modules are set at different positions (for example, inside and outside the annular light source, different shooting angles, etc.) to ensure full coverage of the wafer surface. The working mode of each camera module is set according to the detection requirements, usually including: resolution, focus mode, exposure time, image acquisition rate, etc. Different camera modes can help capture defects of different sizes and characteristics, and improve the diversity and comprehensiveness of data. Under the configuration of different camera modules, ensure that they synchronously or take turns to collect image data of the wafer surface at different angles and in different modes. At this time, the working parameters of all camera modules must be coordinated to avoid mutual interference. By reasonably arranging the positions of the camera modules, ensure that images are collected from multiple perspectives to fully reflect the surface conditions of the wafer. Through precise camera mode parameter settings, ensure that each camera module obtains clear and high-quality image data under appropriate exposure, resolution and focus, providing a more reliable image basis for subsequent analysis.

[0046] More specifically, according to the set camera position and mode, multiple camera modules will simultaneously or sequentially capture images of the wafer surface, and each camera module will obtain image data at the corresponding position and in a specific mode. The collected image data will undergo necessary preprocessing, such as denoising, image enhancement, brightness balancing, and other operations to ensure that the image is clear and the details are discernible. The detection image data from different camera modules are combined to form complete initial detection data. This initial detection data includes multi-angle and multi-viewing angle image information of the wafer surface at different angles and under different light sources, which comprehensively reflects the state of the wafer surface. By accurately configuring the parameters of the camera module, it is ensured that each camera module can obtain clear and high-resolution images, thereby improving the accuracy of defect recognition. Through the collaboration of multiple camera modules, it is ensured that the image data covers all areas of the wafer surface, providing rich information for subsequent defect analysis.

[0047] More specifically, the image data obtained from different positions and under different light sources are summarized to form preliminary inspection data, which will be used as input for subsequent defect detection, defect classification and model optimization. The initial inspection data is preliminarily analyzed to identify possible surface defects and flawed areas. This information can provide a reference for subsequent defect diagnosis and repair processes. By summarizing multi-angle and multi-mode image data, the initial inspection data contains comprehensive information on the wafer surface, which can provide a more valuable foundation for subsequent analysis. The high quality and comprehensiveness of the initial inspection data enable the subsequent defect recognition model to more accurately identify and classify tiny defects on the wafer surface.

[0048] It is understandable that by setting up a ring light source, the wafer surface can be illuminated evenly and in all directions, reducing the problem of missed detection caused by uneven illumination of the light source. The collaborative work of multiple camera modules can fully cover the wafer surface, not only avoiding blind spots, but also improving the accuracy of defect recognition through different modes. The collected multi-dimensional image data can provide more information for subsequent defect analysis and ensure that the system can adapt to different types of wafers and defects. Due to the high quality and comprehensiveness of the initial detection data, the subsequent defect detection system can use these data for more accurate defect analysis and judgment, thereby improving the overall detection efficiency and accuracy.

[0049] Preferably, the steps of digitally simulating the wafer to be inspected according to the initial inspection data of the wafer to be inspected to obtain the wafer inspection feedback model, and performing a gradient scheme analysis of subsequent light source illumination requirements on the ring light source based on the wafer inspection feedback model to obtain a light source illumination gradient execution scheme include: S21: parsing the wafer basic information and wafer detection information of the initial detection data of the wafer to be detected to obtain the wafer basic information and wafer detection information of the wafer to be detected; wherein the wafer basic information is used to describe the wafer specifications and wafer shape of the wafer to be detected, and the wafer detection information is used to describe the detection status displayed on the wafer surface of the wafer to be detected; S22: digitally simulating the wafer to be detected according to the wafer basic information of the wafer to be detected, so as to obtain a wafer detection feedback model for performing digital twin simulation feedback on the wafer to be detected; wherein the wafer detection feedback model includes a wafer substrate simulation part and a wafer surface simulation part; S23: Substituting the wafer detection information into the wafer detection feedback model, so as to adjust the model parameters of the wafer surface simulation part of the wafer detection feedback model according to the wafer detection information, thereby performing digital simulation feedback on the wafer surface detection status of the wafer to be detected through the wafer surface simulation part after the model parameter adjustment; S24: Analyze the wafer quality at each specific location on the wafer surface simulation part of the wafer detection feedback model to obtain the detection quality feature distribution of the wafer detection feedback model, and extract and annotate the defect features of the wafer detection feedback model according to the detection quality feature distribution to obtain a defect annotation set on the wafer detection feedback model; wherein the defect annotation set is used to provide digital feedback on the defects on the wafer to be detected on the wafer detection feedback model; S25: analyzing the light source detection effect of each specific position on the wafer surface simulation part of the wafer detection feedback model based on the initial light source irradiation, so as to obtain the light source detection defect type adaptation range corresponding to the detection quality characteristic distribution of the wafer detection feedback model; S26: Performing difference analysis on the adaptability range of the light source defect detection type according to the preset standard of defect type to be detected, so as to obtain the adaptability range of several types of light source defect detection types to be executed subsequently, and performing working parameter analysis and execution gradient arrangement of the ring light source illumination requirements according to the adaptability range of several types of light source defect detection types to be executed subsequently, so as to obtain the light source illumination gradient execution plan of the ring light source corresponding to the adaptability range of several types of light source defect detection types to be executed subsequently.

[0050] Specifically, it includes the wafer's specifications, shape and other geometric features (such as diameter, thickness, curvature, etc.). These data are used to establish a digital model of the wafer to ensure that the simulation is consistent with the shape and size of the actual physical wafer. It involves the detection conditions displayed on the wafer surface (for example: defect type, defect distribution, surface roughness, etc.). This information is used to describe the current state of the wafer surface and provide a reference for subsequent simulation and feedback adjustments. Through accurate analysis of the initial data, it is ensured that the basic information of the subsequent simulation fully reflects the characteristics and current state of the actual wafer, providing a reliable basis for subsequent operations. The wafer's geometric information is combined with the detection conditions to ensure that the simulation process takes into account both the physical characteristics of the wafer and its surface defects.

[0051] More specifically, through digital simulation technology (based on the basic information of the wafer), the digital twin modeling of the wafer to be inspected is performed to generate a wafer inspection feedback model for inspection feedback. The model consists of two main parts: the wafer substrate simulation part simulates the basic physical structure of the wafer, including the material properties and thickness of the wafer; the wafer surface simulation part simulates the morphology, defects and other characteristics of the wafer surface based on the initial inspection data. Through digital twin technology, the physical state and surface defects of the wafer can be accurately reproduced, providing an accurate virtual platform for subsequent light source and defect analysis. The simulated model can provide real-time feedback on changes in the wafer surface, thereby providing data support for the optimization of the light source irradiation strategy.

[0052] More specifically, the actual wafer inspection information collected is substituted into the feedback model, especially the wafer surface simulation part. By adjusting the simulation parameters (such as the distribution and depth of surface defects), the surface condition of the wafer can be more accurately reflected. By adjusting the model parameters in real time, accurate simulation feedback can be made according to the actual wafer surface state (such as defect type and distribution), thereby improving the accuracy of the simulation results. The adjusted model is closer to the actual wafer inspection situation, providing more practical data for subsequent light source optimization and defect detection.

[0053] More specifically, through the surface simulation part of the simulation feedback model, quality analysis is performed on each specific position of the wafer to determine the detection status of each position. According to the results of the wafer quality analysis, defect features are extracted and annotated to form a defect annotation set. This annotation set reflects the specific position, type and severity of defects on the wafer surface. Through the analysis of simulation data and defect annotation, the problem area on the wafer surface can be accurately identified, providing guidance for subsequent light source irradiation strategies and defect repair. Not only can the defect type be qualitatively judged, but the severity of the defect can also be quantitatively analyzed to facilitate subsequent processing.

[0054] More specifically, based on the initial light source irradiation conditions, the light source effect analysis is performed on the wafer surface simulation part in the feedback model to determine the impact of different light source irradiations on wafer surface defects. Combined with the light source detection effect, it is analyzed under different light source irradiation conditions which types of defects can be detected and which defects are not easily discovered under the existing light source conditions. Through the adaptation analysis of the light source and the defect type, the best light source configuration can be selected for each defect type, thereby maximizing the detection efficiency. By analyzing the adaptability of the light source to different defects, the subsequent optimization of the light source irradiation parameters can be guided to improve the defect detection rate.

[0055] More specifically, based on preset defect type standards, a difference analysis of the adaptability range of light source detection defect types is performed to identify which light source irradiation conditions are suitable for a certain defect type and which are not. Based on the analysis results, a gradient scheme for light source irradiation requirements is determined, and for each defect type adaptability range, the corresponding light source irradiation gradient execution scheme is arranged. Through the execution of the gradient scheme, the most appropriate light source irradiation method can be adopted for different defect types to improve the utilization efficiency of the light source. The gradient analysis ensures that each defect type can receive the best light source irradiation, thereby improving the intelligence and accuracy of detection. Through the gradient execution of the light source, unnecessary waste of light source is avoided, while improving the detection efficiency and reducing the potential impact of overexposure on the wafer surface.

[0056] It is understandable that through digital simulation and defect type adaptive analysis, precise light source configuration for different defect types can be achieved, thereby maximizing detection accuracy. The entire process reduces human intervention and improves the automation and intelligence level of the detection process through digital twin simulation, real-time adjustment and gradient execution scheme design. Through multi-angle and multi-mode simulation feedback and optimization analysis, it can ensure that all potential defects on the wafer surface can be accurately detected under different light source conditions. Gradient analysis based on light source requirements ensures efficient use of the light source and reduces unnecessary waste of resources. Overall, this process combines digital simulation with feedback control to make the light source configuration more precise and intelligent, which can significantly improve the effect and efficiency of wafer defect detection.

[0057] Preferably, the step of analyzing the light source detection effect of each specific position on the wafer surface simulation part of the wafer detection feedback model based on the initial light source irradiation to obtain the light source detection defect type adaptation range corresponding to the detection quality characteristic distribution of the wafer detection feedback model includes: S251: dividing the wafer surface simulation part of the wafer detection feedback model into a positioning grid coordinate system to obtain a positioning grid coordinate system for dividing the wafer surface simulation part into a plurality of positioning grids having positioning coordinates; S252: performing grid decomposition processing on the initial light source illumination based on the positioning grid coordinate system to obtain a light source illumination parameter distribution corresponding to the detection quality feature distribution; S253: performing adaptability analysis of detection effects on several preset wafer defect detection types according to the light source illumination parameter distribution, so as to obtain the adaptability of the detection effects of the light source illumination parameter distribution corresponding to various wafer defect detection types; S254: performing a combined analysis on the detection effect adaptability of the light source illumination parameter distribution corresponding to various wafer defect detection types to obtain an adaptability range of the light source detection defect type corresponding to the detection quality feature distribution of the wafer detection feedback model.

[0058] Specifically, the wafer surface simulation part of the wafer detection feedback model is divided to create a positioning grid coordinate system, which divides the wafer surface into several grid areas with positioning coordinates. Each grid area represents a specific position on the wafer surface, which is convenient for the subsequent analysis and evaluation of the light source illumination effect. Through this grid division, each position in the model can be accurately associated with the actual detection effect. By finely dividing the grid coordinate system, the light source detection effect of each grid position can be evaluated separately, thereby improving the accuracy of positioning and analysis. This grid division can help to systematically organize the detection data and provide a basis for the subsequent light source illumination parameter distribution analysis.

[0059] More specifically, based on the above positioning grid coordinate system, the initial light source illumination conditions (including light source intensity, illumination angle, illumination range, etc.) are decomposed into grids, which means that each grid area will have an independent light source illumination parameter allocation. Through this decomposition, you can set different light source illumination conditions for each grid area separately to obtain a complete light source illumination parameter distribution map, in which the light source illumination conditions at each grid position have clear values. The decomposition of light source illumination ensures that each grid area can obtain accurate light source illumination according to the needs of its specific location, which improves the flexibility and accuracy of detection. Different areas may require light source illumination of different intensities. The decomposition process can ensure that the light source illumination of each area is optimized in a targeted manner.

[0060] More specifically, based on the distribution of light source illumination parameters in each grid area, an adaptability analysis is performed on several preset wafer defect detection types. Specifically, for each defect type (for example, surface scratches, cracks, bubbles, etc.), the detection effect under different light source illumination conditions is evaluated. This step is to analyze whether the light source illumination can effectively highlight or detect these defects by comparing the light source illumination parameters with the response model of each defect. The detection effects under specific light source illumination conditions can be analyzed according to different types of defects, which helps to determine the light source configuration that best suits a certain defect type. By performing an adaptability analysis on the detection effect of each defect, the adaptability of different light source illumination conditions to different defect types can be quantified, thereby providing data support for light source configuration.

[0061] More specifically, a combined analysis will be conducted on the adaptability of the detection effect for each wafer defect type. By combining the detection effects of each grid area under different light sources, the overall detection quality feature distribution can be obtained, and then the defect types suitable for each light source illumination condition can be determined. The result of the combined analysis is the "light source detection defect type adaptation range", that is, under specific light source conditions, which defect types can be effectively detected, and which may be missed or insufficiently detected. Through combined analysis, the interaction of multiple light source conditions and defect types can be comprehensively considered to provide an optimal light source configuration solution. The combined analysis can help determine the most suitable light source illumination range for various defects and optimize the overall detection solution.

[0062] More specifically, the results obtained through combined analysis can determine the adaptability range of light source detection defect types corresponding to the detection quality feature distribution of the wafer inspection feedback model. This adaptability range indicates the wafer defect types that can be effectively detected under specific light source irradiation conditions. This adaptability range provides a basis for subsequent light source irradiation execution plans, and can achieve more accurate defect detection. The adaptability range analysis of the light source ensures that each defect type can be effectively detected under a specific light source, avoiding missed detections due to improper light source configuration. By optimizing the adaptability range of the light source, efficient utilization of light source irradiation resources can be achieved, which not only improves detection accuracy but also reduces energy consumption.

[0063] It can be understood that the entire process, from light source irradiation analysis to determination of defect type adaptation range, combines the following technical effects: through detailed grid division and light source parameter decomposition, precise light source adjustment for different wafer areas and defect types is achieved; through fitness analysis of each defect, it is ensured that different types of defects can achieve the best detection effect, avoiding missed detection and false detection; the light source adaptation range after combined analysis can provide clear guidance for subsequent detection, ensuring the efficiency of light source configuration and the reliability of detection quality; and through precise light source irradiation gradient execution plan, the detection efficiency is improved and energy consumption is reduced.

[0064] Preferably, the steps of sequentially configuring the light source parameters of the annular light source according to the light source illumination gradient execution scheme so that the annular light source sequentially applies the detection light source illumination to the wafer to be inspected, thereby obtaining the gradient detection data of the wafer to be inspected under each round of detection light source illumination include: S31: according to the light source illumination gradient execution scheme, sequentially configure the gradient brightness parameters and the gradient orientation parameters of each light source unit in the annular light source, so that each light source unit in the annular light source is in a gradient illumination state, and each light source unit in the gradient illumination state performs gradient light source illumination in a surrounding state on the wafer to be inspected; S32: According to the light source illumination gradient execution plan, the parameters of the camera position and the camera mode are configured for a plurality of pre-set camera modules, so that each of the camera modules collects image data of the wafer to be inspected, so as to obtain a detection image of the wafer to be inspected at each of the camera positions obtained by the camera module in the camera mode, and each of the detection images together constitutes the gradient detection data of the wafer to be inspected.

[0065] Specifically, according to the light source illumination gradient execution plan, the gradient brightness parameters and gradient orientation parameters of each light source unit in the ring light source are configured in turn, and the brightness of each light source unit changes gradually. Light sources of different brightness can be applied at different positions, so that the brightness of each light source unit irradiating the wafer surface forms a gradual state. By adjusting the illumination direction of the light source, it is ensured that the light of each light source unit is directed to different angles to form a gradual illumination angle, thereby providing a surround light source illumination mode. By adjusting the brightness and orientation gradients, uniform illumination of the ring light source on the wafer surface can be achieved, while avoiding errors caused by excessive or weak illumination, ensuring the balance of light source illumination. By adjusting the gradient orientation, the wafer to be inspected can be illuminated at multiple angles, thereby enhancing the detection capability of different areas on the wafer surface, especially for wafers with complex geometric shapes or special defects.

[0066] More specifically, according to the light source illumination gradient implementation plan, the camera positions and camera modes are configured for several pre-set camera modules: the positioning of the camera module will be adjusted according to the position of the wafer to be inspected and the illumination area of ​​the annular light source. The camera module should collect images at multiple angles and positions to ensure full coverage of the surface of the wafer to be inspected. The working mode of the camera module (such as exposure time, focus, imaging mode, etc.) is adjusted according to the different light source illumination states to optimize the image quality under different light source gradients. Each camera module collects image data of the wafer to be inspected according to its camera position and mode, and obtains the inspection image of the wafer at each camera position. All these images together constitute the gradient inspection data of the wafer to be inspected.

[0067] More specifically, by configuring multiple camera modules and adjusting their positions, image data of the wafer to be inspected can be obtained from different perspectives, avoiding blind spots that may be caused by a single angle, thereby ensuring comprehensive detection coverage. By configuring the camera mode according to different light source gradients, the exposure time, focus and other parameters can be adjusted according to the lighting conditions to ensure that images under different light sources can maintain high quality, especially in low or high brightness areas. Multiple camera modules combined with changes in the light source illumination gradient can capture details of the wafer surface under different lighting conditions, especially defect information under different brightness and angles, thereby improving detection accuracy and sensitivity.

[0068] More specifically, the detection images obtained by each camera module together constitute the gradient detection data of the wafer to be inspected. These images contain the surface information of the wafer under different light source conditions and can be used to analyze the quality status of the wafer. After post-processing (such as image stitching, feature extraction, etc.), these images can be further used for defect recognition, defect location, defect classification and other analysis processes. The formation of gradient detection data is based on image data under multi-angle and multi-brightness conditions, so as to ensure that defects in all areas of the wafer surface can be discovered and recorded. By integrating image data from multiple perspectives and different light source illumination states, higher-dimensional wafer defect information can be obtained, providing more basis for subsequent defect analysis and judgment. Through the synthesized gradient detection data, different types of wafer defects, such as tiny scratches, bubbles, cracks, etc., can be efficiently identified, especially in the detection of complex surfaces or tiny defects, which improves the detection accuracy.

[0069] It can be understood that through the brightness and orientation gradient configuration of each light source unit in the ring light source, a multi-angle, gradient illumination mode can be achieved, which is crucial for the detection of complex wafer surfaces and different types of defects. The position and mode configuration of the camera module ensures that high-quality image data can be collected in all areas of the wafer surface, avoiding missed detections caused by a single perspective. The resulting gradient detection data, after being acquired by multiple camera modules from different angles and under different illumination conditions, can fully reflect the surface state of the wafer and improve the accuracy and reliability of defect identification. Through the coordination of precise light source illumination gradients and image acquisition, subtle defects on the wafer surface can be effectively discovered, especially in low-contrast areas, thereby optimizing the sensitivity and accuracy of defect detection.

[0070] Preferably, the step of substituting the gradient detection data of the wafer to be inspected under each round of detection light source into the wafer detection feedback model to adjust the model parameters of the wafer detection feedback model includes: S41: Substituting the gradient detection data of the wafer to be detected under each round of detection light source into the wafer surface simulation part of the wafer detection feedback model, and making the wafer surface simulation part of the wafer detection feedback model perform data allocation processing on the gradient detection data under each round of detection light source, so that the gradient detection data is in a corresponding relationship with each specific position of the wafer surface simulation part; S42: parsing the gradient detection data according to the light source detection defect type adaptation range corresponding to the gradient detection data to obtain the defect detection result fed back by the gradient detection data, and substituting the defect detection result fed back by the gradient detection data into the specific positions of the wafer surface simulation part according to the corresponding relationship between the gradient detection data and the specific positions of the wafer surface simulation part; S43: Adjust the model parameters of each specific position of the wafer surface simulation part according to the defect detection results fed back by the gradient detection data, and perform a comprehensive weight analysis on the model parameters adjusted by the defect detection results fed back by the gradient detection data in each round at each specific position of the wafer surface simulation part, so as to finally determine the model parameters of each specific position of the wafer surface simulation part.

[0071] Specifically, the gradient detection data obtained from the wafer to be inspected under different rounds of detection light sources are substituted into the wafer surface simulation part of the wafer detection feedback model. The wafer surface simulation part is responsible for simulating the surface characteristics of the wafer according to the model. The goal of this step is to ensure that the detection data can accurately correspond to the specific position of the wafer surface simulation part, that is, according to the specific position of each data point in the gradient detection data, it is matched with the corresponding position in the simulation part to ensure the accuracy of data allocation. By corresponding the gradient detection data with the specific position of the wafer surface simulation part, the relationship between each data point and the corresponding position in the simulation model can be ensured, which avoids data offset or incorrect correspondence and improves the accuracy of the model. This step can ensure that the detection data of each point on the wafer surface is accurately processed, which is crucial for subsequent defect analysis and model adjustment.

[0072] More specifically, data analysis is performed based on the adaptive range of the light source defect detection type corresponding to the gradient detection data. This step aims to identify the defect type (such as cracks, bubbles, scratches, etc.) fed back by each gradient detection data point. Based on the analysis result of each data point, the defect detection result of that position is obtained, and these defect detection results are matched one by one with the specific positions of the simulated part of the wafer surface. For example, if a scratch is detected at a certain position, the defect detection result corresponding to that position will be fed back to that position of the wafer surface simulation model. By analyzing the gradient detection data and feeding back the defect information, accurate judgments can be made for the specific defects at each position, which is crucial for defect location and analysis. Since the gradient detection data is obtained under the irradiation of different light sources, it can more comprehensively feedback the defect information of the wafer surface under different conditions, avoiding the missed detection caused by a single light source. This analysis and feedback mechanism can capture tiny defects or subtle changes on the wafer surface, thereby improving the sensitivity and accuracy of defect detection.

[0073] More specifically, according to the defect detection results at each position, the model parameters of the wafer surface simulation part are adjusted. Specifically, if a defect is detected at a certain position, the model parameters at that position will be adjusted accordingly. For the defect detection results fed back by each round of gradient detection data, the impact on different positions of the wafer surface simulation part is comprehensively analyzed. This involves weighting the defect feedback at each position and ultimately determining the optimal model parameters for each position. By adjusting the parameters of the wafer surface simulation part in real time, the model can adjust itself according to actual defect feedback, thereby improving the adaptability and accuracy of the model. This adjustment can ensure that the model can make precise adjustments when facing defects of different types and locations, avoiding the loss of accuracy caused by large-scale adjustments. By weighted analysis of the defect detection results at each position, the interference of a single data can be avoided, and the feedback effect of each data point can be comprehensively considered to ensure the accuracy and reliability of the model parameter adjustment.

[0074] More specifically, through the aforementioned adjustment and weight analysis, the model parameters of each position of the wafer surface simulation part are finally determined. These parameters reflect the state of each point on the wafer surface under specific light source conditions, including defect information and wafer surface characteristics. Through multiple rounds of gradient data feedback and model parameter adjustment, the final model is highly accurate and can better adapt to actual detection needs. Accurate model parameters enable the defect detection system to more efficiently identify and locate various defects on the wafer surface, thereby improving the reliability and efficiency of the detection process.

[0075] It is understandable that through precise gradient detection data analysis, various types of defects on the wafer surface can be identified and fed back. The model is adjusted according to real-time defect feedback to ensure that the simulation parameters of each detection position are more accurate, thereby enhancing the adaptability of the model. With the support of multiple rounds of detection data under different light source conditions, wafer defects can be comprehensively and accurately identified, thereby improving the robustness and accuracy of the detection system. Comprehensive weight analysis ensures that the adjustment of model parameters is more precise, avoiding unnecessary over-adjustment, and ultimately obtaining optimized wafer surface simulation results, thereby improving the overall detection effect.

[0076] Preferably, the steps of performing an extended light source detection demand analysis according to the wafer detection feedback model after the model parameters are adjusted to obtain an extended light source detection scheme for the wafer to be detected, and detecting the wafer to be detected according to the extended light source detection scheme to obtain extended detection data and substitute it into the wafer detection feedback model include: S51: Acquire performance information of a preset extended light source module; wherein the extended light source module includes a laser light source module and a polarized light source module; S52: performing a fixed-point analysis of potential defects on the wafer inspection feedback model after the model parameters are adjusted to obtain a potential defect detection point of the wafer inspection feedback model; S53: performing an extended light source detection requirement analysis on the potential defect detection points according to the performance information of the extended light source module to obtain an extended light source detection solution for each of the potential defect detection points of the wafer to be detected; S54: configuring parameters of the extended light source module according to the extended light source detection scheme, so that the extended light source module applies a specified form of extended light source to the wafer to be detected, and collects images of potential defect detection points of the wafer to be detected that are subjected to the specified form of extended light source processing through a preset camera module, so as to obtain extended detection data; S55: Substitute the extended detection data into the specific position of the potential defect detection point corresponding to the wafer detection feedback model, and enable the wafer detection feedback model to perform real-time analysis on the extended detection data to adjust the model parameters of the wafer detection feedback model, thereby providing digital feedback on the actual condition of the wafer to be inspected.

[0077] Specifically, the performance information of the pre-set extended light source module is obtained. The extended light source module includes a laser light source module and a polarized light source module. The laser light source module is usually used to focus and locate tiny defects on the wafer surface, such as cracks, tiny scratches, etc. The polarized light source module is used to detect the reflection characteristics of the wafer surface, which can reveal the stress, texture and some more hidden defects inside the material. By obtaining the performance information of the light source module, different types of light sources can be applied in a targeted manner to ensure that the characteristics of each light source can best match the detection target (wafer surface defects) and improve the effectiveness of subsequent detection. The combination of different light source modules enables the system to perform all-round detection of wafers under different optical conditions to better reveal potential defects.

[0078] More specifically, the wafer inspection feedback model that has completed model parameter adjustment is subjected to a fixed-point analysis of potential defects to identify potential defect detection points, which are locations on the wafer surface where defects or flaws may exist. These potential defect areas can be identified through the feedback model, and these areas often require further refined detection. Through fixed-point analysis, possible defect points can be identified from the overall scan, providing a clear direction for subsequent more in-depth inspections. By optimizing the selection of inspection points, indiscriminate inspection of the entire wafer surface can be avoided, saving time and resources, and focusing on areas with more potential defects.

[0079] More specifically, based on the performance information of the extended light source module, an extended light source detection demand analysis is performed on potential defect detection points, and an extended light source detection plan is formulated for each detection point. The plan determines which light source (laser or polarized light) to use and how to apply the light source (such as the angle and intensity of the light source) based on the nature of the detection point and the problem to be solved (such as cracks, stress, reflection characteristics, etc.). This analysis can select appropriate light sources for precise illumination and imaging at different potential defect locations, avoiding the use of redundant light sources and improving the accuracy of defect identification. Customized light source usage plans can adopt different detection strategies for different types of defects (such as surface scratches, tiny cracks, stress points, etc.), thereby improving the comprehensiveness and accuracy of detection.

[0080] More specifically, according to the extended light source detection plan, the parameters of the extended light source module are configured to ensure that it can apply light source processing to the wafer according to the set requirements. For example, the laser light source can irradiate the wafer surface at a specific angle, and the polarized light source can select different polarization angles. The wafer with the extended light source applied is imaged through a preset camera module. After the camera module obtains the image data, it is used for further analysis, especially around potential defect detection points. According to the specific extended light source requirements, adjusting the parameters of the light source module can ensure the optimal light source irradiation method for each potential defect point, thereby improving the sensitivity and accuracy of detection. The cooperation of the camera module ensures that the images captured by the extended light source have sufficient clarity and details, providing reliable data support for subsequent image analysis and defect location.

[0081] More specifically, the extended inspection data (i.e., image data obtained through the extended light source module and camera module) is substituted into the wafer inspection feedback model. These data will correspond to the specific locations of potential defect detection points. The model will perform further real-time analysis of the wafer surface based on these data. Based on the analysis, the model will adjust the parameters again according to the new data, thereby providing digital feedback on the actual condition of the wafer to be inspected. Through real-time analysis and parameter adjustment, the wafer inspection feedback model can self-optimize after each round of inspection and continuously improve its ability to identify wafer defects. Digital feedback allows inspectors to obtain wafer defect information in real time, thereby making decisions and corrections faster, improving the efficiency and quality of the production process.

[0082] It is understandable that by obtaining the performance information of the extended light source module and formulating a light source scheme that meets the needs of potential defect detection, different defects can be detected under optimal conditions. By performing fixed-point analysis of potential defects on the wafer surface, areas where defects may exist can be quickly and accurately located, providing precise targets for subsequent in-depth detection. The extended light source detection scheme for each potential defect point is customized to ensure efficient detection under the most suitable light source conditions. After the extended light source is applied, high-quality images are obtained through the camera module, and then analyzed and fed back in real time through the wafer detection feedback model, and the model parameters are continuously adjusted to improve detection accuracy.

[0083] Reference Figure 2 As shown, in a second aspect, the present invention provides an adaptive control device for the brightness of a wafer inspection light source, which is used to implement an adaptive control method for the brightness of a wafer inspection light source as described in any one of the first aspects, including: An initial detection module is used to irradiate the wafer to be detected with an initial light source in a surrounding state through a pre-set annular light source, and collect image data of the wafer to be detected under the irradiation of the initial light source in the surrounding state, so as to obtain initial detection data of the wafer to be detected; A scheme analysis module, used for digitally simulating the wafer to be inspected according to the initial inspection data of the wafer to be inspected to obtain the wafer inspection feedback model, and performing a gradient scheme analysis of the subsequent light source illumination requirements on the annular light source based on the wafer inspection feedback model to obtain a light source illumination gradient execution scheme; A gradient detection module, used to sequentially configure light source parameters of the annular light source according to the light source illumination gradient execution scheme, so that the annular light source sequentially applies detection light source illumination to the wafer to be inspected, thereby obtaining gradient detection data of the wafer to be inspected under each round of detection light source illumination; A data feedback module, used for substituting the gradient detection data of the wafer to be detected under each round of detection light source into the wafer detection feedback model, so as to adjust the model parameters of the wafer detection feedback model; The extended detection module is used to perform an extended light source detection demand analysis based on the wafer detection feedback model after model parameter adjustment to obtain an extended light source detection solution for the wafer to be detected, and to detect the wafer to be detected according to the extended light source detection solution to obtain extended detection data and substitute it into the wafer detection feedback model.

[0084] In this embodiment, for the specific implementation of each module in the above-mentioned device embodiment, please refer to the above-mentioned method embodiment, which will not be described in detail here.

[0085] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for adaptively controlling the brightness of a wafer inspection light source, characterized in that: include: Using a pre-set annular light source to illuminate the wafer to be inspected in a surrounding state with an initial light source, and collecting image data of the wafer to be inspected under the illumination of the initial light source in the surrounding state, so as to obtain initial inspection data of the wafer to be inspected; Digitally simulating the wafer to be inspected according to the initial inspection data of the wafer to be inspected to obtain a wafer inspection feedback model, and performing a gradient scheme analysis of subsequent light source illumination requirements on the annular light source based on the wafer inspection feedback model to obtain a light source illumination gradient execution scheme; According to the light source illumination gradient execution scheme, the light source parameters of the annular light source are sequentially configured, so that the annular light source sequentially applies the detection light source illumination to the wafer to be inspected, thereby obtaining the gradient detection data of the wafer to be inspected under each round of detection light source illumination; Substituting the gradient detection data of the wafer to be inspected under each round of detection light source into the wafer detection feedback model to adjust the model parameters of the wafer detection feedback model; An extended light source detection demand analysis is performed based on the wafer detection feedback model after model parameter adjustment to obtain an extended light source detection scheme for the wafer to be inspected, and the wafer to be inspected is inspected based on the extended light source detection scheme to obtain extended detection data and substitute it into the wafer detection feedback model.

2. The method for adaptively controlling the brightness of a wafer inspection light source according to claim 1, characterized in that: The steps of irradiating the wafer to be inspected with an initial light source in a surrounding state by a pre-set annular light source, and collecting image data of the wafer to be inspected under the irradiation of the initial light source in the surrounding state to obtain initial inspection data of the wafer to be inspected include: The initial brightness parameters and initial orientation parameters of each light source unit in the pre-set annular light source are configured so that each light source unit in the annular light source is in an initial irradiation state, and each light source unit in the initial irradiation state performs initial light source irradiation in a surrounding state on the wafer to be inspected; The parameters of the camera position and the camera mode are configured for several pre-set camera modules so that each of the camera modules can collect image data of the wafer to be inspected, so as to obtain a detection image of the wafer to be inspected at each of the camera positions obtained by the camera modules in the camera mode, and each of the detection images together constitutes the initial detection data of the wafer to be inspected.

3. The method for adaptively controlling the brightness of a wafer inspection light source according to claim 1, characterized in that: The steps of digitally simulating the wafer to be inspected according to the initial inspection data of the wafer to be inspected to obtain the wafer inspection feedback model, and performing a gradient scheme analysis of subsequent light source illumination requirements on the ring light source based on the wafer inspection feedback model to obtain a light source illumination gradient execution scheme include: Parsing the wafer basic information and wafer detection information of the initial detection data of the wafer to be detected to obtain the wafer basic information and wafer detection information of the wafer to be detected; wherein the wafer basic information is used to describe the wafer specifications and wafer shape of the wafer to be detected, and the wafer detection information is used to describe the detection status displayed on the wafer surface of the wafer to be detected; Digitally simulating the wafer to be detected according to the wafer basic information of the wafer to be detected to obtain a wafer detection feedback model for performing digital twin simulation feedback on the wafer to be detected; wherein the wafer detection feedback model includes a wafer substrate simulation part and a wafer surface simulation part; Substituting the wafer detection information into the wafer detection feedback model, so as to adjust the model parameters of the wafer surface simulation part of the wafer detection feedback model according to the wafer detection information, thereby performing digital simulation feedback on the wafer surface detection status of the wafer to be detected through the wafer surface simulation part after the model parameter adjustment; Analyzing the wafer quality at each specific location on the wafer surface simulation part of the wafer detection feedback model to obtain the detection quality characteristic distribution of the wafer detection feedback model; Based on the initial light source irradiation, the light source detection effect is analyzed at each specific position on the wafer surface simulation part of the wafer detection feedback model, so as to obtain the light source detection defect type adaptation range corresponding to the detection quality characteristic distribution of the wafer detection feedback model; According to the preset standards for defect types to be detected, a difference analysis is performed on the adaptability range of the light source defect detection type to obtain the adaptability ranges of several light source defect detection types to be executed subsequently, and according to the adaptability ranges of several light source defect detection types to be executed subsequently, the working parameters of the ring light source are analyzed and the execution gradient is arranged according to the light source illumination requirements to obtain the light source illumination gradient execution plan of the ring light source corresponding to the adaptability ranges of several light source defect detection types to be executed subsequently.

4. The method for adaptively controlling the brightness of a wafer inspection light source according to claim 3, characterized in that: The steps of analyzing the light source detection effect of each specific position on the wafer surface simulation part of the wafer detection feedback model based on the initial light source irradiation to obtain the light source detection defect type adaptation range corresponding to the detection quality feature distribution of the wafer detection feedback model include: Dividing the wafer surface simulation part of the wafer detection feedback model into a positioning grid coordinate system to obtain a positioning grid coordinate system for dividing the wafer surface simulation part into a plurality of positioning grids having positioning coordinates; Performing grid decomposition processing on the initial light source illumination based on the positioning grid coordinate system to obtain a light source illumination parameter distribution corresponding to the detection quality feature distribution; Performing adaptability analysis on detection effects of several preset wafer defect detection types according to the light source illumination parameter distribution to obtain the adaptability of detection effects of the light source illumination parameter distribution corresponding to various wafer defect detection types; The adaptability of the detection effects of the light source illumination parameter distribution corresponding to various wafer defect detection types is combined and analyzed to obtain the adaptability range of the light source detection defect types corresponding to the detection quality feature distribution of the wafer detection feedback model.

5. The method for adaptively controlling the brightness of a wafer inspection light source according to claim 1, characterized in that: The steps of sequentially configuring the light source parameters of the annular light source according to the light source illumination gradient execution scheme so that the annular light source sequentially applies the detection light source illumination to the wafer to be inspected, thereby obtaining the gradient detection data of the wafer to be inspected under each round of detection light source illumination include: According to the light source illumination gradient execution scheme, the gradient brightness parameters and the gradient orientation parameters of each light source unit in the annular light source are sequentially configured, so that each light source unit in the annular light source is in a gradient illumination state, and each light source unit in the gradient illumination state performs gradient light source illumination in a surrounding state on the wafer to be inspected; According to the light source illumination gradient execution scheme, the parameters of the camera position and the camera mode are configured for several pre-set camera modules, so that each of the camera modules collects image data of the wafer to be inspected, so as to obtain a detection image of the wafer to be inspected at each of the camera positions obtained by the camera module in the camera mode, and each of the detection images together constitutes the gradient detection data of the wafer to be inspected.

6. The method for adaptively controlling the brightness of a wafer inspection light source as claimed in claim 3, characterized in that: Substituting the gradient detection data of the wafer to be inspected under each round of detection light source into the wafer detection feedback model to adjust the model parameters of the wafer detection feedback model includes: Substituting the gradient detection data of the wafer to be inspected under each round of detection light source into the wafer surface simulation part of the wafer detection feedback model, and allowing the wafer surface simulation part of the wafer detection feedback model to perform data allocation processing on the gradient detection data under each round of detection light source, so that the gradient detection data is in a corresponding relationship with the specific positions of each location of the wafer surface simulation part; The gradient detection data is parsed according to the light source detection defect type adaptation range corresponding to the gradient detection data to obtain the defect detection result fed back by the gradient detection data, and according to the correspondence between the gradient detection data and the specific positions of the wafer surface simulation part, the defect detection result fed back by the gradient detection data is substituted into the specific positions of the wafer surface simulation part; The model parameters of each specific position of the wafer surface simulation part are adjusted according to the defect detection results fed back by the gradient detection data, and a comprehensive weight analysis is performed on the model parameters adjusted by the defect detection results fed back by the gradient detection data in each round at each specific position of the wafer surface simulation part, so as to finally determine the model parameters of each specific position of the wafer surface simulation part.

7. The method for adaptively controlling the brightness of a wafer inspection light source according to claim 1, characterized in that: The steps of performing an extended light source detection demand analysis according to the wafer detection feedback model after the model parameters are adjusted to obtain an extended light source detection scheme for the wafer to be detected, and detecting the wafer to be detected according to the extended light source detection scheme to obtain extended detection data and substitute it into the wafer detection feedback model include: Acquire performance information of a preset extended light source module; wherein the extended light source module includes a laser light source module and a polarized light source module; Performing a fixed-point analysis of potential defects on the wafer inspection feedback model after the model parameters have been adjusted to obtain a potential defect detection point of the wafer inspection feedback model; Performing an extended light source detection demand analysis on the potential defect detection points according to the performance information of the extended light source module to obtain an extended light source detection solution for each of the potential defect detection points of the wafer to be detected; According to the extended light source detection scheme, the extended light source module is configured with parameters so that the extended light source module applies a specified form of extended light source to the wafer to be detected, and an image of a potential defect detection point of the wafer to be detected that is subjected to the specified form of extended light source processing is captured through a preset camera module to obtain extended detection data; The extended detection data is substituted into the specific position of the potential defect detection point corresponding to the wafer detection feedback model, and the wafer detection feedback model is made to perform real-time analysis on the extended detection data so as to adjust the model parameters of the wafer detection feedback model, thereby providing digital feedback on the actual condition of the wafer to be inspected.

8. An adaptive control device for wafer inspection light source brightness, characterized in that: A method for adaptively controlling the brightness of a wafer inspection light source according to any one of claims 1 to 7, comprising: An initial detection module is used to irradiate the wafer to be detected with an initial light source in a surrounding state through a pre-set annular light source, and collect image data of the wafer to be detected under the irradiation of the initial light source in the surrounding state, so as to obtain initial detection data of the wafer to be detected; A scheme analysis module, used for digitally simulating the wafer to be inspected according to the initial inspection data of the wafer to be inspected to obtain the wafer inspection feedback model, and performing a gradient scheme analysis of the subsequent light source illumination requirements on the annular light source based on the wafer inspection feedback model to obtain a light source illumination gradient execution scheme; A gradient detection module, used to sequentially configure light source parameters of the annular light source according to the light source illumination gradient execution scheme, so that the annular light source sequentially applies detection light source illumination to the wafer to be inspected, thereby obtaining gradient detection data of the wafer to be inspected under each round of detection light source illumination; A data feedback module, used for substituting the gradient detection data of the wafer to be detected under each round of detection light source into the wafer detection feedback model, so as to adjust the model parameters of the wafer detection feedback model; The extended detection module is used to perform an extended light source detection demand analysis based on the wafer detection feedback model after model parameter adjustment to obtain an extended light source detection solution for the wafer to be detected, and to detect the wafer to be detected according to the extended light source detection solution to obtain extended detection data and substitute it into the wafer detection feedback model.

Images