Artificial intelligence-based semiconductor wafer film pasting quality monitoring system and method

Abstract

The application belongs to the technical field of semiconductor wafer processing supervision, and particularly relates to a semiconductor wafer film pasting quality monitoring system and method based on artificial intelligence, wherein the system comprises a substrate surface inspection module, a film pasting monitoring and analysis module, a film pasting quality inspection module, a film pasting performance comprehensive evaluation module and a supervision terminal; the substrate surface inspection module is used for scanning the surface of the semiconductor wafer and judging the surface quality of the semiconductor wafer; the film pasting device is used for pasting the film on the semiconductor wafer when the film pasting preparation signal is generated; the film pasting monitoring and analysis module is used for monitoring and analyzing the film pasting process of the semiconductor wafer to identify the abnormality in the film pasting process; the film pasting quality of the semiconductor wafer is accurately judged after the film pasting is completed; and the film pasting performance of the film pasting device is analyzed by the film pasting performance comprehensive evaluation module; the film pasting device is inspected and repaired when the performance inferiority signal is generated, so that the management difficulty of the management personnel is significantly reduced.

Description

Technical Field

[0001] This invention relates to the field of semiconductor wafer processing supervision technology, specifically to a semiconductor wafer film bonding quality monitoring system and method based on artificial intelligence. Background Technology

[0002] Semiconductor wafers, also known as silicon wafers or integrated circuit wafers, are circular thin films made from silicon material through processes such as purification, cutting, and polishing. They are the carriers used in the production of integrated circuits and are widely used in the manufacture of various semiconductor devices. The application of film to semiconductor wafers is one of the key steps in the semiconductor manufacturing process, and its quality directly affects the performance and reliability of subsequent chips. Currently, when supervising the application of film to semiconductor wafers, it is difficult to accurately reflect the surface condition of the semiconductor wafer and determine the necessity of film application before film application. Furthermore, it is impossible to conduct anomaly analysis during the film application process and accurately assess the film application quality after completion. Additionally, it is impossible to reasonably analyze and accurately reflect the performance status of the film application, which increases the management difficulty for managers and is not conducive to ensuring the film application effect and efficiency. To address the aforementioned technical shortcomings, a solution is proposed. Summary of the Invention

[0003] The purpose of this invention is to provide an artificial intelligence-based semiconductor wafer film bonding quality monitoring system and method, which solves the problems of existing technologies that are difficult to effectively monitor and provide feedback before, during, and after semiconductor wafer film bonding, and cannot reasonably analyze and accurately provide feedback on the film bonding performance, which increases the management difficulty for managers and is not conducive to ensuring the film bonding effect and efficiency.

[0004] To achieve the above objectives, the present invention provides the following technical solution: The artificial intelligence-based semiconductor wafer film bonding quality monitoring system includes a substrate surface inspection module, a film bonding monitoring and analysis module, a film bonding quality inspection module, a film bonding performance comprehensive evaluation module, and a monitoring terminal. Before applying the film to the semiconductor wafer, the substrate surface inspection module scans the surface of the semiconductor wafer to acquire a substrate image. The surface quality of the semiconductor wafer is judged by the substrate image, and a corresponding semiconductor wafer film application preparation signal or rejection signal is generated accordingly. When the film application preparation signal is generated, the film is applied to the semiconductor wafer by the film application equipment. The film application monitoring and analysis module monitors the film application process of the semiconductor wafer, and generates a film application warning signal or a film application qualified signal by analysis, and sends the film application warning signal or film application qualified signal to the monitoring terminal. After the film is applied, the film application quality inspection module will analyze the film application quality of the semiconductor wafer, generate a normal inspection signal or an abnormal inspection signal through analysis, and send the normal inspection signal or the abnormal inspection signal to the monitoring terminal. The film application performance comprehensive evaluation module is used to set the detection period, analyze the film application performance of semiconductor wafers during the detection period, generate a good performance signal or a poor performance signal through analysis, and send the good performance signal or poor performance signal to the monitoring terminal.

[0005] Furthermore, the specific analysis process of the substrate surface inspection module is as follows: Several detection areas are set on the surface of a semiconductor wafer. The protrusions, depressions and scratches on the surface of the semiconductor wafer are identified by the substrate image. If the corresponding detection area involves a protrusion, depression or scratch, the corresponding detection area is marked as a damaged area. Furthermore, the substrate image is processed for grayscale, and the actual grayscale value of the corresponding detection area is collected. The actual grayscale value is compared with the corresponding preset grayscale threshold. If the actual grayscale value exceeds the preset grayscale threshold, the corresponding detection area is marked as a non-clean area. The number of damaged areas and non-clean areas on the semiconductor wafer are obtained and marked as damage detection value and non-clean detection value, respectively. The damage detection value and non-clean detection value are compared with the preset damage detection threshold and the preset non-clean detection threshold, respectively. If the damage detection value or non-clean detection value exceeds the corresponding preset threshold, a rejection signal for the corresponding semiconductor wafer is generated.

[0006] Furthermore, if neither the damage detection value nor the non-clean detection value exceeds the corresponding preset threshold, the damaged area and the non-clean area are marked as poor quality areas; if a cluster of poor quality areas is obtained, the number of poor quality areas in the corresponding cluster is marked as the poor quality cluster detection value, and the poor quality cluster detection value is compared with the preset poor quality cluster detection threshold. If the poor quality cluster detection value exceeds the preset poor quality cluster detection threshold, the corresponding cluster is marked as a high-cluster area. The number of high-polymerization areas in the semiconductor wafer is obtained and marked as high-polymerization monitoring values, and the highest value of inferior polymerization detection value is marked as inferior polymerization amplitude value. The substrate evaluation value is obtained by numerically calculating the damage detection value, non-net detection value, high-polymerization monitoring value and inferior polymerization amplitude value. The substrate evaluation value is compared with the preset substrate evaluation threshold. If the substrate evaluation value exceeds the preset substrate evaluation threshold, a rejection signal for the corresponding semiconductor wafer is generated. If the substrate evaluation value does not exceed the preset substrate evaluation threshold, a film-coating preparation signal for the corresponding semiconductor wafer is generated.

[0007] Furthermore, the specific analysis process of the film application monitoring and analysis module is as follows: The film application temperature, film application pressure, and film application speed are collected during the film application process. The deviation of the film application temperature from the preset suitable film application temperature is marked as the film application temperature value, the deviation of the film application pressure from the preset suitable film application pressure is marked as the film application pressure value, and the deviation of the film application speed from the preset suitable film application speed is marked as the film application speed value. The film application monitoring and analysis value is obtained by numerically calculating the film application temperature, film application pressure, and film application speed. The film application monitoring and analysis value is then compared with the preset film application monitoring and analysis threshold. If the film application monitoring and analysis value exceeds the preset film application monitoring and analysis threshold, a film application warning signal is generated; if the film application monitoring and analysis value does not exceed the preset film application monitoring and analysis threshold, a film application qualified signal is generated.

[0008] Furthermore, the specific analysis process of the film application quality inspection module is as follows: The actual film-attached area on the corresponding semiconductor wafer is collected, and the actual film-attached area is overlapped with the standard film-attached area to obtain the film-attached position deviation value. The film-attached position deviation value is compared with the preset film-attached position deviation threshold. If the film-attached position deviation value exceeds the preset film-attached position deviation threshold, an inspection abnormality signal is generated. If the film application position deviation value does not exceed the preset film application position deviation threshold, then bubbles and wrinkles on the semiconductor wafer surface are identified. Bubble defects with a volume exceeding the preset bubble volume threshold are marked as high-impact bubbles, and wrinkles with a volume exceeding the preset wrinkle volume threshold are marked as high-impact wrinkles. If high-impact bubbles or high-impact wrinkles exist on the corresponding semiconductor wafer, then an inspection anomaly signal is generated.

[0009] Furthermore, if there are no high-impact bubbles or high-impact wrinkles on the corresponding semiconductor wafer, the ratio of the area and value covered by bubble defects and wrinkle defects on the semiconductor wafer to the total area of ​​the actual film-coated area is used to calculate the defect coverage ratio. The film application quality inspection value is obtained by weighted summation of the film application position deviation value and the defect coverage ratio value. The film application quality inspection value is then compared with the preset film application quality inspection threshold. If the film application quality inspection value exceeds the preset film application quality inspection threshold, an abnormal inspection signal is generated; if the film application quality inspection value does not exceed the preset film application quality inspection threshold, a normal inspection signal is generated.

[0010] Furthermore, the specific analysis process of the comprehensive performance evaluation module for screen protectors includes: The number of times the film application warning signal is generated during the detection period is obtained and the ratio of the film application time of the film application equipment during the detection period is calculated to obtain the film application warning frequency value. The ratio of the number of semiconductor wafers corresponding to the abnormal signal during the detection period to the total number of semiconductor wafers that have completed film application during the detection period is calculated to obtain the film application abnormality status value. The performance value of the screen protector is obtained by numerically calculating the screen protector warning frequency value and the screen protector anomaly detection frequency value. The performance value of the screen protector is then compared with a preset screen protector performance threshold. If the performance value of the screen protector exceeds the preset screen protector performance threshold, a poor performance signal is generated; if the performance value of the screen protector does not exceed the preset screen protector performance threshold, a good performance signal is generated.

[0011] Furthermore, the film application performance comprehensive evaluation module is connected to the operation and maintenance diagnosis and analysis module. The film application performance comprehensive evaluation module sends the performance degradation signal to the operation and maintenance diagnosis and analysis module. When the operation and maintenance diagnosis and analysis module receives the performance degradation signal, it analyzes the operation and maintenance status of the film application equipment during the detection period. Through analysis, it generates an operation and maintenance diagnosis pass signal or an operation and maintenance diagnosis early warning signal, and sends the operation and maintenance diagnosis pass signal or operation and maintenance diagnosis early warning signal to the monitoring terminal. When the monitoring terminal receives the operation and maintenance diagnosis early warning signal, it issues a corresponding warning.

[0012] Furthermore, the specific analysis process of the operation and maintenance diagnosis and analysis module is as follows: The start and end times of the operation and maintenance of the film application equipment are obtained. The time difference between the end time and the start time is calculated to obtain the maintenance duration. The number of maintenance durations that do not exceed the preset maintenance duration threshold during the detection period is marked as the maintenance over-condition value. The average of all maintenance durations during the detection period is calculated to obtain the maintenance performance value. The interval between the start time and the end time of the previous operation and maintenance of the film application equipment is marked as the inter-dimension duration. The number of inter-dimension durations that exceed the preset inter-dimension duration threshold during the detection period is marked as the inter-dimension over-detection value. The average of all inter-dimension durations during the detection period is calculated to obtain the inter-dimension performance value. The operation and maintenance (O&M) diagnostic value is obtained by numerically calculating the over-condition value, performance value, inter-dimensional over-detection value, and inter-dimensional performance value. The O&M diagnostic value is then compared with a preset O&M diagnostic threshold. If the O&M diagnostic value exceeds the preset O&M diagnostic threshold, an O&M diagnostic warning signal is generated; if the O&M diagnostic value does not exceed the preset O&M diagnostic threshold, an O&M diagnostic pass signal is generated.

[0013] Furthermore, this invention also proposes an artificial intelligence-based method for monitoring the quality of semiconductor wafer film bonding, comprising the following steps: Step 1: Before applying the film to the semiconductor wafer, scan the surface of the semiconductor wafer to acquire a substrate image; Step 2: Determine the surface quality of the semiconductor wafer through the substrate image, and generate the corresponding semiconductor wafer film preparation signal or rejection signal accordingly. Step 3: When generating the film application preparation signal, the film application equipment applies film to the semiconductor wafer and monitors the film application process of the semiconductor wafer. The film application warning signal or film application qualified signal is generated by analysis. Step 4: After the film is applied, the quality of the film application on the semiconductor wafer will be analyzed, and a normal or abnormal inspection signal will be generated through the analysis. Step 5: Analyze the film application performance of the semiconductor wafer during the testing period using the film application performance comprehensive evaluation module, and generate a performance indicator signal or a performance indicator signal based on the analysis.

[0014] Compared with the prior art, the beneficial effects of the present invention are: 1. In this invention, the substrate surface inspection module scans the surface of the semiconductor wafer and judges its surface quality before film application. When a film application preparation signal is generated, the film application equipment applies the film to the semiconductor wafer. The film application monitoring and analysis module monitors and analyzes the film application process of the semiconductor wafer to identify film application abnormalities. After film application is completed, the analysis is used to accurately judge the film application quality of the semiconductor wafer. Furthermore, the film application performance comprehensive evaluation module analyzes the film application performance of the film application equipment. When a performance defect signal is generated, the film application equipment is inspected and repaired, significantly reducing the management difficulty for managers. 2. In this invention, the performance degradation signal is sent to the operation and maintenance diagnosis and analysis module through the film application performance comprehensive evaluation module. When the operation and maintenance diagnosis and analysis module receives the performance degradation signal, it analyzes the operation and maintenance status of the film application equipment during the detection period. It can reasonably analyze the correlation between poor operation performance of the film application equipment and operation and maintenance performance. When generating operation and maintenance diagnosis and early warning signals, it strengthens the subsequent operation and maintenance supervision of the film application equipment operation and maintenance personnel, ensures the timeliness and effectiveness of subsequent operation and maintenance, and helps to improve the operation performance of the film application equipment. It has a high degree of intelligence. Attached Figure Description

[0015] To facilitate understanding by those skilled in the art, the present invention will be further described below with reference to the accompanying drawings; Figure 1 This is a system block diagram of Embodiment 1 of the present invention; Figure 2 This is a system block diagram of Embodiment 2 of the present invention; Figure 3 This is a flowchart of the method in Embodiment 3 of the present invention. Detailed Implementation

[0016] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0017] Example 1: As Figure 1 As shown, the semiconductor wafer film bonding quality monitoring system based on artificial intelligence proposed in this invention includes a substrate surface inspection module, a film bonding monitoring and analysis module, a film bonding quality inspection module, a film bonding performance comprehensive evaluation module, and a monitoring terminal. Before applying the film to the semiconductor wafer, the substrate surface inspection module scans the surface of the semiconductor wafer to acquire a substrate image. The surface quality of the semiconductor wafer is determined based on the substrate image, and a film application preparation signal or a rejection signal is generated accordingly. The rejection signal is sent to the monitoring terminal to reject the corresponding semiconductor wafer (i.e., the wafer is not applied with the film). This avoids unnecessary time waste and ensures the stability and efficiency of subsequent film application operations. The specific analysis process of the substrate surface inspection module is as follows: Several detection areas are set on the surface of a semiconductor wafer. The protrusions, depressions and scratches on the surface of the semiconductor wafer are identified by the substrate image. If the corresponding detection area involves a protrusion, depression or scratch, the corresponding detection area is marked as a damaged area. Furthermore, the substrate image is processed for grayscale, and the actual grayscale value of the corresponding detection area is collected. The actual grayscale value is compared with the corresponding preset grayscale threshold. If the actual grayscale value exceeds the preset grayscale threshold, it indicates that the corresponding detection area is likely to have surface contamination, and the corresponding detection area is marked as a non-clean area. The number of damaged areas and unclean areas on the semiconductor wafer are obtained and marked as damage detection value and unclean detection value, respectively. The damage detection value and unclean detection value are compared with the preset damage detection threshold and the preset unclean detection threshold, respectively. If the damage detection value or unclean detection value exceeds the corresponding preset threshold, it indicates that the surface quality of the corresponding semiconductor wafer is poor, and a rejection signal for the corresponding semiconductor wafer is generated.

[0018] Furthermore, if neither the damage detection value nor the non-clean detection value exceeds the corresponding preset threshold, the damaged area and the non-clean area are marked as poor areas; if a cluster of poor areas is obtained (i.e., several poor areas are connected together), the number of poor areas in the corresponding cluster is marked as the poor cluster detection value, and the poor cluster detection value is compared with the preset poor cluster detection threshold. If the poor cluster detection value exceeds the preset poor cluster detection threshold, the corresponding cluster is marked as a high cluster area. The number of high-polymerization regions in the semiconductor wafer is obtained and marked as high-polymerization monitoring values, and the worst-quality polymerization detection value with the largest value is marked as the worst-quality polymerization amplitude value; Through formula The substrate evaluation value XN is obtained by numerically calculating the damage detection value RF, the non-net detection value RS, the high polymerization monitoring value RY, and the poor polymerization amplitude value RP; where sg1, sg2, sg3, and sg4 are preset proportional coefficients, sg3 > sg4 > sg1 > sg2 > 0; and the larger the value of the substrate evaluation value XN, the worse the surface quality of the corresponding semiconductor wafer. The substrate evaluation value XN is compared with the preset substrate evaluation threshold. If the substrate evaluation value XN exceeds the preset substrate evaluation threshold, it indicates that the surface quality of the corresponding semiconductor wafer is poor, and a rejection signal for the corresponding semiconductor wafer is generated. If the substrate evaluation value XN does not exceed the preset substrate evaluation threshold, it indicates that the surface quality of the corresponding semiconductor wafer is good, and a film-coating preparation signal for the corresponding semiconductor wafer is generated.

[0019] When generating the film application preparation signal, the film application equipment applies film to the semiconductor wafer. The film application monitoring and analysis module monitors the film application process of the semiconductor wafer, and generates a film application warning signal or a film application qualification signal through analysis. This signal is then sent to the monitoring terminal. Upon receiving the film application warning signal, the monitoring terminal issues an alert to remind management personnel to take timely control measures, thereby ensuring the film application effect and stability. The specific analysis process of the film application monitoring and analysis module is as follows: The film application temperature, film application pressure, and film application speed are collected during the film application process. The deviation of the film application temperature from the preset suitable film application temperature is marked as the film application temperature value, the deviation of the film application pressure from the preset suitable film application pressure is marked as the film application pressure value, and the deviation of the film application speed from the preset suitable film application speed is marked as the film application speed value. Through formula The film application temperature value TP, film application pressure value TX, and film application speed value TS are numerically calculated to obtain the film application monitoring and analysis value TL; where my1, my2, and my3 are preset proportional coefficients with values ​​greater than zero, and the larger the value of the film application monitoring and analysis value TL, the worse the overall performance of the film application at the corresponding time. The screen protector monitoring and analysis value TL is compared with the preset screen protector monitoring and analysis threshold. If the screen protector monitoring and analysis value TL exceeds the preset screen protector monitoring and analysis threshold, it indicates that the screen protector performance at the corresponding time is poor overall, and a screen protector warning signal is generated. If the screen protector monitoring and analysis value TL does not exceed the preset screen protector monitoring and analysis threshold, it indicates that the screen protector performance at the corresponding time is good overall, and a screen protector qualified signal is generated.

[0020] After the film is applied, the film application quality inspection module analyzes the film application quality of the semiconductor wafers. This analysis generates either a normal or abnormal inspection signal, which is then sent to the monitoring terminal. This allows for reasonable analysis and accurate judgment of the film application quality for each semiconductor wafer, facilitating subsequent handling measures and providing information support for subsequent film application performance evaluation and analysis, ensuring the accuracy of the analysis results. The specific analysis process of the film application quality inspection module is as follows: The actual film-attached area on the corresponding semiconductor wafer is collected, and the overlap between the actual film-attached area and the standard film-attached area is detected to obtain the film-attached position deviation value. It should be noted that the smaller the overlap between the actual film-attached area and the standard film-attached area, the less accurate the film-attached position for the corresponding semiconductor wafer, and the larger the value of the film-attached position deviation value. The film application position deviation value is compared with the preset film application position deviation threshold. If the film application position deviation value exceeds the preset film application position deviation threshold, it indicates that the film application position for the corresponding semiconductor wafer is not accurate, indicating that the film application performance for the corresponding semiconductor wafer is poor, and an inspection abnormality signal is generated. If the film application position deviation value does not exceed the preset film application position deviation threshold, then bubbles and wrinkles on the semiconductor wafer surface are identified. Bubble defects with a volume exceeding the preset bubble volume threshold are marked as high-impact bubbles, and wrinkles with a volume exceeding the preset wrinkle volume threshold are marked as high-impact wrinkles. If high-impact bubbles or high-impact wrinkles exist on the corresponding semiconductor wafer, it indicates that the film application performance for the corresponding semiconductor wafer is poor, and an inspection anomaly signal is generated.

[0021] Furthermore, if there are no high-impact bubbles or high-impact wrinkles on the corresponding semiconductor wafer, the ratio of the area and value covered by bubble defects and wrinkle defects on the semiconductor wafer to the total area of ​​the actual film-coated area is used to calculate the defect coverage ratio. The film application quality inspection value NK is calculated by weighting and summing the film application position deviation value NS and the defect coverage ratio value NF using the formula NK=kp1*NS+kp2*NF. Here, kp1 and kp2 are preset weight coefficients with values ​​greater than zero. Furthermore, the larger the value of the film application quality inspection value NK, the worse the overall film application performance for the corresponding semiconductor wafer. The film quality inspection value NK is compared with the preset film quality inspection threshold. If the film quality inspection value NK exceeds the preset film quality inspection threshold, it indicates that the overall film performance of the corresponding semiconductor wafer is poor, and an inspection abnormal signal is generated. If the film quality inspection value NK does not exceed the preset film quality inspection threshold, it indicates that the overall film performance of the corresponding semiconductor wafer is good, and an inspection normal signal is generated.

[0022] The comprehensive performance evaluation module for film application is used to set the testing period, preferably 25 days. It analyzes the film application performance on semiconductor wafers during this period, generating either a superior or inferior performance signal, which is then sent to the monitoring terminal. Upon receiving an inferior performance signal, the monitoring terminal issues a corresponding warning to remind management personnel to promptly inspect and repair the film application equipment, ensuring subsequent film application performance. This significantly reduces the management burden and demonstrates a high degree of automation. The specific analysis process of the comprehensive performance evaluation module for film application is as follows: The number of times the film application warning signal is generated during the detection period is obtained and the ratio of the film application time of the film application equipment during the detection period is calculated to obtain the film application warning frequency value. The ratio of the number of semiconductor wafers corresponding to the abnormal signal during the detection period to the total number of semiconductor wafers that have completed film application during the detection period is calculated to obtain the film application abnormality status value. The film application performance value FX is obtained by numerically calculating the film application warning frequency value WY and the film application anomaly detection value WF using the formula FX=hu1*WY+hu2*WF. Here, hu1 and hu2 are preset weighting coefficients with values ​​greater than zero. Furthermore, the larger the value of the film application performance value FX, the worse the overall film application performance of the film application equipment is during the detection period. The film application performance value FX is compared with the preset film application performance threshold. If the film application performance value FX exceeds the preset film application performance threshold, it indicates that the overall film application performance of the film application equipment is poor during the testing period, and a poor performance signal is generated. If the film application performance value FX does not exceed the preset film application performance threshold, it indicates that the overall film application performance of the film application equipment is good during the testing period, and a good performance signal is generated.

[0023] Example 2: Figure 2 As shown, the difference between this embodiment and Embodiment 1 is that the film application performance comprehensive evaluation module is connected to the operation and maintenance diagnosis and analysis module. The film application performance comprehensive evaluation module sends the performance degradation signal to the operation and maintenance diagnosis and analysis module. When the operation and maintenance diagnosis and analysis module receives the performance degradation signal, it analyzes the operation and maintenance status of the film application equipment during the detection period and generates an operation and maintenance diagnosis qualified signal or an operation and maintenance diagnosis early warning signal through analysis. Furthermore, the system sends either a qualified or early warning signal for the maintenance diagnosis to the monitoring terminal. Upon receiving the early warning signal, the monitoring terminal issues a corresponding alert, enabling reasonable analysis of the correlation between poor performance of the screen protector application equipment and maintenance practices. When management personnel receive the early warning signal, they can strengthen supervision of the screen protector application equipment maintenance personnel to ensure timely and effective maintenance, thereby improving the overall performance of the application equipment. The specific analysis process of the maintenance diagnosis analysis module is as follows: The start and end times of the operation and maintenance of the film application equipment are obtained. The time difference between the end time and the start time is calculated to obtain the maintenance duration. The number of maintenance durations that do not exceed the preset maintenance duration threshold during the detection period is marked as the maintenance over-condition value. The average of all maintenance durations during the detection period is calculated to obtain the maintenance performance value. The interval between the start time and the end time of the previous operation and maintenance of the film application equipment is marked as the inter-dimension duration. The number of inter-dimension durations that exceed the preset inter-dimension duration threshold during the detection period is marked as the inter-dimension over-detection value. The average of all inter-dimension durations during the detection period is calculated to obtain the inter-dimension performance value. The maintenance diagnosis value LP is obtained by numerically calculating the inter-dimensional over-condition value LN, inter-dimensional performance value LS, inter-dimensional over-detection value LR, and inter-dimensional performance value LW using the formula LP=(a1*LN+a2 / (LS+1)+a3*LR+a4*LW) / 2. Among them, a1, a2, a3, and a4 are preset proportional coefficients with values ​​greater than zero. Furthermore, the larger the value of the maintenance diagnosis value LP, the worse the overall maintenance performance of the film-applying equipment is during the testing period. The maintenance and repair (OPR) diagnostic value LP is compared with the preset OPR diagnostic threshold. If the OPR diagnostic value LP exceeds the preset OPR diagnostic threshold, it indicates that the overall OPR performance of the screen protector application equipment was poor during the testing period, and the poor performance of the screen protector application equipment was highly correlated with the OPR performance. In this case, an OPR diagnostic warning signal is generated. If the OPR diagnostic value LP does not exceed the preset OPR diagnostic threshold, it indicates that the overall OPR performance of the screen protector application equipment was good during the testing period. In this case, an OPR diagnostic pass signal is generated.

[0024] Example 3: Figure 3 As shown, the difference between this embodiment and Embodiments 1 and 2 is that the semiconductor wafer film bonding quality monitoring method based on artificial intelligence proposed in this invention includes the following steps: Step 1: Before applying the film to the semiconductor wafer, scan the surface of the semiconductor wafer to acquire a substrate image; Step 2: Determine the surface quality of the semiconductor wafer through the substrate image, and generate the corresponding semiconductor wafer film preparation signal or rejection signal accordingly. Step 3: When generating the film application preparation signal, the film application equipment applies film to the semiconductor wafer and monitors the film application process of the semiconductor wafer. The film application warning signal or film application qualified signal is generated by analysis. Step 4: After the film is applied, the quality of the film application on the semiconductor wafer will be analyzed, and a normal or abnormal inspection signal will be generated through the analysis. Step 5: Analyze the film application performance of the semiconductor wafer during the testing period using the film application performance comprehensive evaluation module, and generate a performance indicator signal or a performance indicator signal based on the analysis.

[0025] The working principle of this invention is as follows: Before applying the film to the semiconductor wafer, the substrate surface inspection module scans the surface of the semiconductor wafer to acquire an image and determine its surface quality. When a rejection signal is generated, the corresponding semiconductor wafer is not coated, avoiding unnecessary time waste and ensuring the stability and efficiency of subsequent coating operations. When a coating preparation signal is generated, the coating equipment applies the film to the semiconductor wafer. The coating monitoring and analysis module monitors and analyzes the coating process, reminding managers to take timely control measures when a coating warning signal is generated. After coating is completed, the coating quality inspection module analyzes the coating quality of the semiconductor wafer to accurately determine the coating quality for each wafer. Furthermore, the coating performance comprehensive evaluation module analyzes the coating performance of the semiconductor wafer during the testing period. When a poor performance signal is generated, the coating equipment is inspected and repaired to ensure subsequent coating performance, significantly reducing the management difficulty and demonstrating a high degree of intelligence.

[0026] The above formulas are all dimensionless numerical calculations. These formulas are derived from software simulations using collected data to obtain the most recent real-world results. The preset parameters in the formulas are set by those skilled in the art according to actual conditions. The preferred embodiments of the present invention disclosed above are merely illustrative of the invention. The preferred embodiments do not describe all details exhaustively, nor do they limit the invention to specific implementations. Obviously, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention. The invention is limited only by the claims and their full scope and equivalents.

Claims

1. A semiconductor wafer film bonding quality monitoring system based on artificial intelligence, characterized in that, It includes a substrate surface inspection module, a film application monitoring and analysis module, a film application quality inspection module, a film application performance comprehensive evaluation module, and a monitoring terminal; Before applying the film to the semiconductor wafer, the substrate surface inspection module scans the surface of the semiconductor wafer to acquire a substrate image. The surface quality of the semiconductor wafer is judged by the substrate image, and a film application preparation signal or a rejection signal is generated accordingly. The rejection signal is sent to the monitoring terminal to reject the corresponding semiconductor wafer. When generating the film application preparation signal, the film is applied to the semiconductor wafer by the film application equipment. The film application monitoring and analysis module monitors the film application process of the semiconductor wafer, and generates a film application warning signal or a film application qualified signal by analysis. The film application warning signal or the film application qualified signal is sent to the monitoring terminal. When the monitoring terminal receives the film application warning signal, it issues a warning. After the film is applied, the film application quality inspection module will analyze the film application quality of the semiconductor wafer, generate a normal inspection signal or an abnormal inspection signal through analysis, and send the normal inspection signal or an abnormal inspection signal to the monitoring terminal to take subsequent handling measures for the corresponding semiconductor wafer, and provide information support for the subsequent film application performance evaluation and analysis process. The film coating performance comprehensive evaluation module is used to set the testing period, analyze the film coating performance of semiconductor wafers during the testing period, generate a good performance signal or a poor performance signal through analysis, and send the good performance signal or poor performance signal to the monitoring terminal. When the monitoring terminal receives the poor performance signal, it issues a corresponding warning.

2. The semiconductor wafer lamination quality monitoring system based on artificial intelligence according to claim 1, characterized in that, The specific analysis process of the substrate surface inspection module is as follows: Several detection areas are set on the surface of a semiconductor wafer. The protrusions, depressions and scratches on the surface of the semiconductor wafer are identified by the substrate image. If the corresponding detection area involves a protrusion, depression or scratch, the corresponding detection area is marked as a damaged area. Furthermore, grayscale processing is performed on the substrate image to collect the actual grayscale value of the corresponding detection area. If the actual grayscale value exceeds the preset grayscale threshold, the corresponding detection area is marked as a non-clean area. The number of damaged areas and non-clean areas on the semiconductor wafer are obtained and marked as damage detection value and non-clean detection value, respectively. If the damage detection value or non-clean detection value exceeds the corresponding preset threshold, a rejection signal for the corresponding semiconductor wafer is generated.

3. The artificial intelligence-based semiconductor wafer film bonding quality monitoring system according to claim 2, characterized in that, If neither the damage detection value nor the non-clean detection value exceeds the corresponding preset threshold, the damaged area and the non-clean area are marked as poor quality areas; if a cluster of poor quality areas is obtained, the number of poor quality areas in the corresponding cluster is marked as the poor quality cluster detection value. The poor quality cluster detection value is compared with the preset poor quality cluster detection threshold. If the poor quality cluster detection value exceeds the preset poor quality cluster detection threshold, the corresponding cluster is marked as a high-cluster area. The number of high-aggregation areas in the semiconductor wafer is obtained and marked as high-aggregation monitoring value, and the worst-quality aggregation detection value with the largest value is marked as the worst-quality aggregation amplitude value; the substrate evaluation value is obtained by numerically calculating the damage detection value, non-net detection value, high-aggregation monitoring value and worst-quality aggregation amplitude value. If the substrate evaluation value exceeds the preset substrate evaluation threshold, a rejection signal for the corresponding semiconductor wafer is generated. If the substrate evaluation value does not exceed the preset substrate evaluation threshold, a corresponding semiconductor wafer film preparation signal is generated.

4. The artificial intelligence-based semiconductor wafer film bonding quality monitoring system according to claim 1, characterized in that, The specific analysis process of the template monitoring and analysis module is as follows: The film application temperature, film application pressure, and film application speed are collected during the film application process. The deviation of the film application temperature from the preset suitable film application temperature is marked as the film application temperature value, the deviation of the film application pressure from the preset suitable film application pressure is marked as the film application pressure value, and the deviation of the film application speed from the preset suitable film application speed is marked as the film application speed value. The film application monitoring and analysis value is obtained by numerically calculating the film application temperature value, film application pressure value, and film application speed value. If the film application monitoring and analysis value exceeds the preset film application monitoring and analysis threshold, a film application warning signal is generated. If the screen protector monitoring and analysis value does not exceed the preset screen protector monitoring and analysis threshold, a screen protector qualification signal will be generated.

5. The semiconductor wafer lamination quality monitoring system based on artificial intelligence according to claim 1, characterized in that, The specific analysis process of the screen protector quality inspection module is as follows: The actual film-attached area on the corresponding semiconductor wafer is collected, and the actual film-attached area is overlapped with the standard film-attached area for detection. The film-attached position deviation value is obtained accordingly. If the film-attached position deviation value exceeds the preset film-attached position deviation threshold, an inspection abnormality signal is generated. If the film application position deviation value does not exceed the preset film application position deviation threshold, then bubbles and wrinkles on the semiconductor wafer surface are identified. Bubble defects with a volume exceeding the preset bubble volume threshold are marked as high-impact bubbles, and wrinkles with a volume exceeding the preset wrinkle volume threshold are marked as high-impact wrinkles. If high-impact bubbles or high-impact wrinkles exist on the corresponding semiconductor wafer, then an inspection anomaly signal is generated.

6. The artificial intelligence-based semiconductor wafer film bonding quality monitoring system according to claim 5, characterized in that, If there are no high-impact bubbles or high-impact wrinkles on the corresponding semiconductor wafer, the ratio of the area and value covered by bubble defects and wrinkle defects on the semiconductor wafer to the total area of ​​the actual film-coated area is used to calculate the defect coverage ratio. The film quality inspection value is obtained by weighting and summing the film application position deviation value and the defect coverage ratio value. If the film quality inspection value exceeds the preset film quality inspection threshold, an inspection abnormality signal is generated. If the film application quality inspection value does not exceed the preset film application quality inspection threshold, a normal inspection signal will be generated.

7. The artificial intelligence-based semiconductor wafer film bonding quality monitoring system according to claim 1, characterized in that, The specific analysis process of the comprehensive performance evaluation module for screen protectors includes: The number of times the film application warning signal is generated during the detection period is obtained and the ratio of the film application time of the film application equipment during the detection period is calculated to obtain the film application warning frequency value. The ratio of the number of semiconductor wafers corresponding to the abnormal signal during the detection period to the total number of semiconductor wafers that have completed film application during the detection period is calculated to obtain the film application abnormality status value. The performance value of the screen protector is obtained by numerically calculating the screen protector warning frequency value and the screen protector anomaly detection frequency value. If the screen protector performance value exceeds the preset screen protector performance threshold, a poor performance signal is generated; if the screen protector performance value does not exceed the preset screen protector performance threshold, a good performance signal is generated.

8. The artificial intelligence-based semiconductor wafer film bonding quality monitoring system according to claim 7, characterized in that, The film application performance comprehensive evaluation module communicates with the operation and maintenance diagnosis and analysis module. The film application performance comprehensive evaluation module sends the performance deterioration signal to the operation and maintenance diagnosis and analysis module. When the operation and maintenance diagnosis and analysis module receives the performance deterioration signal, it analyzes the operation and maintenance status of the film application equipment during the detection period, and generates an operation and maintenance diagnosis qualified signal or an operation and maintenance diagnosis early warning signal through analysis, and sends the operation and maintenance diagnosis qualified signal or operation and maintenance diagnosis early warning signal to the monitoring terminal.

9. The artificial intelligence-based semiconductor wafer film bonding quality monitoring system according to claim 8, characterized in that, The specific analysis process of the operation and maintenance diagnosis and analysis module is as follows: The start and end times of the operation and maintenance of the film application equipment are obtained. The time difference between the end time and the start time is calculated to obtain the maintenance duration. The number of maintenance durations that do not exceed the preset maintenance duration threshold during the detection period is marked as the maintenance over-condition value. The average of all maintenance durations during the detection period is calculated to obtain the maintenance performance value. The interval between the start time and the end time of the previous operation and maintenance of the film application equipment is marked as the inter-dimension duration. The number of inter-dimension durations that exceed the preset inter-dimension duration threshold during the detection period is marked as the inter-dimension over-detection value. The average of all inter-dimension durations during the detection period is calculated to obtain the inter-dimension performance value. The operation and maintenance diagnosis value is obtained by numerically calculating the over-condition value, performance value, inter-dimensional over-detection value, and inter-dimensional performance value. If the operation and maintenance diagnosis value exceeds the preset operation and maintenance diagnosis threshold, an operation and maintenance diagnosis warning signal is generated; if the operation and maintenance diagnosis value does not exceed the preset operation and maintenance diagnosis threshold, an operation and maintenance diagnosis qualified signal is generated.

10. A semiconductor wafer film bonding quality monitoring method based on artificial intelligence, characterized in that, The method employs an artificial intelligence-based semiconductor wafer film bonding quality monitoring system as described in any one of claims 1-9.

Images