Semiconductor wafer film pasting quality monitoring system and method based on artificial intelligence

Abstract

The invention 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, and the system comprises a substrate surface inspection module, a film pasting monitoring analysis module, a film pasting quality inspection module, a film pasting performance comprehensive evaluation module and a supervision terminal. The surface of the semiconductor wafer is scanned through the substrate surface inspection module, the surface quality of the semiconductor wafer is judged, and the semiconductor wafer is subjected to film pasting through the film pasting equipment when the film pasting preparation signal is generated. The film pasting monitoring and analyzing module monitors and analyzes the film pasting process of the semiconductor wafer to identify abnormities in the film pasting process, the film pasting quality of the semiconductor wafer is accurately judged through analysis after film pasting is completed, and the film pasting performance of the film pasting equipment is analyzed through the film pasting performance comprehensive evaluation module. And when the performance inferior table signal is generated, the film pasting equipment is checked and maintained, so that the management difficulty of management personnel is remarkably reduced.

Description

Technical Field

[0001] The present invention relates to the field of semiconductor wafer processing supervision technology, and in particular to a semiconductor wafer film quality monitoring system and method based on artificial intelligence. Background Art

[0002] Semiconductor wafers, also known as silicon wafers or integrated circuit wafers, are round sheets made of silicon materials through purification, cutting, polishing and other processes. They are carriers used in the production of integrated circuits and are widely used in the manufacture of various semiconductor devices. Semiconductor wafer lamination is one of the key steps in the semiconductor manufacturing process, and its quality directly affects the performance and reliability of subsequent chips. At present, when supervising the filming of semiconductor wafers, it is difficult to accurately feedback the surface condition of the semiconductor wafers and judge the necessity of filming before filming. It is also impossible to analyze abnormalities in the filming process and accurately evaluate the filming quality after filming is completed. It is also impossible to reasonably analyze and accurately feedback the filming performance status, which increases the management difficulty of managers and is not conducive to ensuring the filming effect and efficiency. In view of the above technical defects, a solution is now proposed. Summary of the invention

[0003] The purpose of the present invention is to provide a semiconductor wafer film quality monitoring system and method based on artificial intelligence, which solves the problem that the prior art is difficult to achieve effective monitoring and feedback of semiconductor wafers before, during and after film pasting, and cannot reasonably analyze and accurately feedback the film pasting performance status, which increases the management difficulty of management personnel and is not conducive to ensuring the film pasting effect and efficiency.

[0004] To achieve the above object, the present invention provides the following technical solutions: Semiconductor wafer film quality monitoring system based on artificial intelligence, including substrate surface inspection module, film monitoring and analysis module, film quality inspection module, film performance comprehensive evaluation module and supervision terminal; Before film lamination of semiconductor wafers, the substrate surface inspection module scans the surface of the semiconductor wafer to collect a substrate image, and judges the surface quality of the semiconductor wafer through the substrate image, and accordingly generates a film lamination preparation signal or a to-be-rejected signal of the corresponding semiconductor wafer; When the film pasting preparation signal is generated, the film pasting equipment is used to paste the film on the semiconductor wafer, and the film pasting monitoring and analysis module monitors the film pasting process of the semiconductor wafer, generates a film pasting warning signal or a film pasting qualified signal through analysis, and sends the film pasting warning signal or the film pasting qualified signal to the supervision terminal; After the film pasting is completed, the film pasting quality inspection module will analyze the die pasting quality status 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 supervision terminal; The comprehensive evaluation module for film pasting performance is used to set the detection period, analyze the film pasting performance of the semiconductor wafer during the detection period, generate a good performance signal or a bad performance signal through the analysis, and send the good performance signal or the bad performance signal to the supervision terminal.

[0005] Furthermore, the specific analysis process of the substrate surface inspection module is as follows: A plurality of detection areas are set on the surface of the semiconductor wafer, and the protrusions, depressions and scratches on the surface of the semiconductor wafer are identified through the substrate image. If the corresponding detection area involves a protrusion, depression or scratch, the corresponding detection area is marked as a damaged area; The substrate image is gray-scaled, the actual gray-scale value of the corresponding detection area is collected, and the actual gray-scale value is numerically compared with the corresponding preset gray-scale threshold. If the actual gray-scale value exceeds the preset gray-scale threshold, the corresponding detection area is marked as a non-clean area; The number of damaged areas and the number of non-clean areas on the semiconductor wafer are obtained and marked as damage detection values ​​and non-net detection values, respectively. The damage detection values ​​and non-net detection values ​​are numerically compared with the preset damage detection threshold and the preset non-net detection threshold, respectively. If the damage detection value or the non-net detection value exceeds the corresponding preset threshold, a signal for the corresponding semiconductor wafer to be rejected is generated.

[0006] Furthermore, if both the damage detection value and the non-clean detection value do not exceed the corresponding preset threshold, the damaged area and the non-clean area are marked as poor quality areas; the clustering area of ​​the poor quality areas is obtained, the number of poor quality areas in the corresponding clustering area is marked as a poor quality clustering detection value, the poor quality clustering detection value is numerically compared with the preset poor quality clustering detection threshold, and if the poor quality clustering detection value exceeds the preset poor quality clustering detection threshold, the corresponding clustering area is marked as a high-clustering area; The number of high-poly cluster areas in the semiconductor wafer is obtained and marked as high-poly cluster monitoring values, and the low-quality poly cluster detection value with the largest numerical value is marked as the low-quality poly cluster amplitude value; the substrate evaluation value is obtained by numerically calculating the damage detection value, the non-net detection value, the high-poly cluster monitoring value and the low-quality poly cluster amplitude value, and the substrate evaluation value is numerically compared with the preset substrate evaluation threshold value. If the substrate evaluation value exceeds the preset substrate evaluation threshold value, a signal for rejection of the corresponding semiconductor wafer is generated; if the substrate evaluation value does not exceed the preset substrate evaluation threshold value, a film preparation signal for the corresponding semiconductor wafer is generated.

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

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

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

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

[0011] Furthermore, the comprehensive film performance evaluation module is communicatively connected to the operation and maintenance diagnosis and analysis module. The comprehensive film performance evaluation module sends a poor performance signal to the operation and maintenance diagnosis and analysis module. When the operation and maintenance diagnosis and analysis module receives the poor performance signal, it analyzes the operation and maintenance status of the film 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 the operation and maintenance diagnosis early warning signal to the supervision terminal. When the supervision terminal receives the operation and maintenance diagnosis early warning signal, it issues a corresponding early warning.

[0012] Furthermore, the specific analysis process of the operation and maintenance diagnosis and analysis module is as follows: The start time and end time of the operation and maintenance of the film laminating equipment are obtained, and the time difference between the end time and the start time is calculated to obtain the maintenance and testing time. The number of maintenance and testing times that do not exceed the preset maintenance and testing time threshold during the detection period is marked as the maintenance and testing excess value, and the average of all maintenance and testing times during the detection period is calculated to obtain the maintenance and testing performance value. The interval between the start time and the end time of the last operation and maintenance of the film-sticking equipment is marked as the inter-dimensional time, the number of inter-dimensional time that exceeds the preset inter-dimensional time threshold during the detection period is marked as the inter-dimensional excess detection value, and the average of all inter-dimensional time within the detection period is calculated to obtain the inter-dimensional performance value; The operation and maintenance diagnosis value is obtained by numerically calculating the maintenance and measurement excess value, maintenance and measurement performance value, inter-maintenance over-inspection value and inter-maintenance performance value, and the operation and maintenance diagnosis value is numerically compared with the preset operation and maintenance diagnosis threshold. 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.

[0013] Furthermore, the present invention also proposes a semiconductor wafer film quality monitoring method based on artificial intelligence, comprising the following steps: Step 1: Before laminating the semiconductor wafer, scan the surface of the semiconductor wafer to collect a substrate image; Step 2: judging the surface quality of the semiconductor wafer through the substrate image, and generating a film-mounting preparation signal or a to-be-removed signal of the corresponding semiconductor wafer accordingly; Step 3: When the film-mounting preparation signal is generated, the film-mounting equipment mounts the film on the semiconductor wafer, monitors the film-mounting process of the semiconductor wafer, and generates a film-mounting warning signal or a film-mounting qualified signal through analysis; Step 4: After the film is attached, the die attach quality of the semiconductor wafer is analyzed, and a normal inspection signal or an abnormal inspection signal is generated through the analysis; Step 5: Analyze the film bonding performance of the semiconductor wafer during the test period through the comprehensive film bonding performance evaluation module, and generate a good performance signal or a bad performance signal through the analysis.

[0014] Compared with the prior art, the present invention has the following beneficial effects: 1. In the present invention, the substrate surface inspection module is used to scan the surface of the semiconductor wafer before film pasting and judge the surface quality of the semiconductor wafer. When the film pasting preparation signal is generated, the film pasting equipment is used to paste the film on the semiconductor wafer. The film pasting monitoring and analysis module monitors and analyzes the film pasting process of the semiconductor wafer to identify film pasting abnormalities. After the film pasting is completed, the film pasting quality of the semiconductor wafer is accurately judged through analysis. The film pasting performance comprehensive evaluation module analyzes the film pasting performance of the film pasting equipment. When the performance poor table signal is generated, the film pasting equipment is inspected and repaired, which significantly reduces the management difficulty of the management personnel; 2. In the present invention, the performance-poor signal is sent to the operation and maintenance diagnosis and analysis module through the comprehensive evaluation module of the film performance. When the operation and maintenance diagnosis and analysis module receives the performance-poor signal, it analyzes the operation and maintenance status of the film-sticking equipment during the detection period, and can reasonably analyze the correlation between the poor operating performance of the film-sticking equipment and the operation and maintenance performance. When the operation and maintenance diagnosis and early warning signal is generated, the subsequent operation and maintenance supervision of the operation and maintenance personnel of the film-sticking equipment is strengthened to ensure the timeliness and effect of the subsequent operation and maintenance, which is conducive to improving the operating performance of the film-sticking equipment and has a high degree of intelligence. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to facilitate understanding by those skilled in the art, the present invention is further described below in conjunction with 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 flow chart of the method of Embodiment 3 of the present invention. DETAILED DESCRIPTION

[0016] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0017] Embodiment 1: Figure 1 As shown, the semiconductor wafer film quality monitoring system based on artificial intelligence proposed in the present invention includes a substrate surface inspection module, a film monitoring and analysis module, a film quality inspection module, a film performance comprehensive evaluation module and a supervision terminal; Before laminating the semiconductor wafer, the substrate surface inspection module scans the surface of the semiconductor wafer to collect the substrate image, and judges the surface quality of the semiconductor wafer through the substrate image, and generates the laminating preparation signal or the to-be-rejected signal of the corresponding semiconductor wafer accordingly, and sends the to-be-rejected signal to the supervision terminal to reject the corresponding semiconductor wafer (i.e., not laminating the corresponding semiconductor wafer), which can avoid unnecessary time waste and ensure the stability and efficiency of the subsequent laminating operation; the specific analysis process of the substrate surface inspection module is as follows: A plurality of detection areas are set on the surface of the semiconductor wafer, and the protrusions, depressions and scratches on the surface of the semiconductor wafer are identified through the substrate image. If the corresponding detection area involves a protrusion, depression or scratch, the corresponding detection area is marked as a damaged area; The substrate image is gray-scaled, and the actual gray-scale value of the corresponding detection area is collected, and the actual gray-scale value is numerically compared with the corresponding preset gray-scale threshold. If the actual gray-scale value exceeds the preset gray-scale threshold, it indicates that the corresponding detection area is likely to be contaminated on the surface, and the corresponding detection area is marked as a non-clean area. The number of damaged areas and the number of non-clean areas on the semiconductor wafer are obtained and marked as damage detection values ​​and non-net detection values, respectively. The damage detection values ​​and non-net detection values ​​are numerically compared with the preset damage detection threshold and the preset non-net detection threshold, respectively. If the damage detection value or the non-net detection value exceeds the corresponding preset threshold, indicating that the surface quality of the corresponding semiconductor wafer is poor, a signal to be rejected for the corresponding semiconductor wafer is generated.

[0018] Furthermore, if both the damage detection value and the non-clean detection value do not exceed the corresponding preset threshold, the damaged area and the non-clean area are marked as poor quality areas; the clustering area of ​​poor quality areas (i.e., several poor quality areas are connected together) is obtained, the number of poor quality areas in the corresponding clustering area is marked as a poor quality clustering detection value, the poor quality clustering detection value is numerically compared with the preset poor quality clustering detection threshold, and if the poor quality clustering detection value exceeds the preset poor quality clustering detection threshold, the corresponding clustering area is marked as a high-clustering area; The number of high-polygon areas in the semiconductor wafer is obtained and marked as a high-polygon monitoring value, and the low-quality polygon detection value with the largest value is marked as a low-quality polygon amplitude value; By formula The damage detection value RF, the non-net detection value RS, the high-poly monitoring value RY and the low-quality poly amplitude value RP are numerically calculated to obtain the substrate evaluation value XN; wherein 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 numerically 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 preparation signal for the corresponding semiconductor wafer is generated.

[0019] When the film pasting preparation signal is generated, the film pasting equipment is used to paste the film on the semiconductor wafer. The film pasting monitoring and analysis module monitors the film pasting process of the semiconductor wafer, generates a film pasting warning signal or a film pasting qualified signal through analysis, and sends the film pasting warning signal or the film pasting qualified signal to the supervision terminal. When the supervision terminal receives the film pasting warning signal, it issues an early warning to remind the management personnel to take corresponding control measures in time, so as to ensure the film pasting effect and film pasting stability. The specific analysis process of the die pasting monitoring and analysis module is as follows: The film pasting temperature, film pasting pressure and film pasting speed of the die pasting process are collected, and the deviation value of the film pasting temperature compared with the preset film pasting suitable temperature is marked as the film pasting temperature value, the deviation value of the film pasting pressure compared with the preset film pasting suitable pressure is marked as the film pasting pressure value, and the deviation value of the film pasting speed compared with the preset film pasting suitable speed is marked as the film pasting speed value; By formula The film pasting temperature value TP, the film pasting pressure value TX and the film pasting speed value TS are numerically calculated to obtain the film pasting monitoring and analysis value TL; wherein my1, my2 and my3 are preset proportional coefficients with values ​​greater than zero, and the larger the value of the film pasting monitoring and analysis value TL is, the worse the film pasting execution performance at the corresponding moment is overall; The film pasting monitoring and analysis value TL is numerically compared with the preset film pasting monitoring and analysis threshold. If the film pasting monitoring and analysis value TL exceeds the preset film pasting monitoring and analysis threshold, it indicates that the film pasting execution performance at the corresponding moment is generally poor, and a film pasting warning signal is generated; if the film pasting monitoring and analysis value TL does not exceed the preset film pasting monitoring and analysis threshold, it indicates that the film pasting execution performance at the corresponding moment is generally good, and a film pasting qualified signal is generated.

[0020] After the film is pasted, the film quality inspection module will analyze the 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 supervision terminal, which can reasonably analyze and accurately judge the film quality of each semiconductor wafer, facilitate the subsequent disposal measures for the corresponding semiconductor wafer, and provide information support for the subsequent film performance evaluation and analysis process to ensure the accuracy of its analysis results; the specific analysis process of the film quality inspection module is as follows: The actual film pasting area on the corresponding semiconductor wafer is collected, and the actual film pasting area is overlapped with the standard film pasting area to obtain the film pasting position deviation value. It should be noted that the smaller the overlap between the actual film pasting area and the standard film pasting area, the less accurate the film pasting position for the corresponding semiconductor wafer is, and the larger the film pasting position deviation value is; Compare the film pasting position deviation value with the preset film pasting position deviation threshold value. If the film pasting position deviation value exceeds the preset film pasting position deviation threshold value, it indicates that the film pasting position for the corresponding semiconductor wafer is not accurate, indicating that the film pasting performance for the corresponding semiconductor wafer is poor, and an inspection abnormality signal is generated; If the film position deviation value does not exceed the preset film position deviation threshold, the bubble and wrinkle defects on the surface of the semiconductor wafer are identified, and the bubble defects whose volume exceeds the preset bubble volume threshold are marked as high-impact bubbles, and the wrinkle defects whose volume exceeds 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, indicating that the film performance for the corresponding semiconductor wafer is poor, an inspection abnormality signal is generated.

[0021] Furthermore, if there are no high-impact bubbles and high-impact wrinkles on the corresponding semiconductor wafer, the defect coverage ratio is calculated by ratioing the area and value covered by the bubble defects and wrinkle defects on the semiconductor wafer to the total area of ​​the actual film-mounting area; The film bonding quality inspection value NK is calculated by weighting and summing the film bonding position deviation value NS and the defect coverage ratio value NF through the formula NK=kp1*NS+kp2*NF; wherein, kp1 and kp2 are preset weight coefficients with values ​​greater than zero, and the larger the value of the film bonding quality inspection value NK is, the worse the film bonding performance for the corresponding semiconductor wafer is overall; The film quality inspection value NK is numerically 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 film performance for the corresponding semiconductor wafer is generally poor, and an inspection abnormality signal is generated; if the film quality inspection value NK does not exceed the preset film quality inspection threshold, it indicates that the film performance for the corresponding semiconductor wafer is generally good, and a inspection normal signal is generated.

[0022] The film laminating performance comprehensive evaluation module is used to set the detection period, preferably, the detection period is twenty-five days; the film laminating performance of the semiconductor wafer during the detection period is analyzed, and a good performance signal or a bad performance signal is generated through analysis, and the good performance signal or the bad performance signal is sent to the supervision terminal. When the supervision terminal receives the bad performance signal, it issues a corresponding warning to remind the management personnel to check and repair the film laminating equipment in time to ensure its subsequent film laminating performance, significantly reducing the management difficulty of the management personnel, and having a high degree of intelligence; the specific analysis process of the film laminating performance comprehensive evaluation module is as follows: The number of times the film pasting warning signal is generated during the detection period is obtained and the ratio is calculated with the film pasting operation time of the film pasting equipment during the detection period to obtain the film pasting warning frequency value, and the number of semiconductor wafers corresponding to the inspection abnormality signal during the detection period is calculated with the total number of semiconductor wafers that have completed film pasting during the detection period to obtain the film pasting abnormality proportion value; The film pasting performance value FX is obtained by numerically calculating the film pasting warning frequency value WY and the film pasting detection abnormality value WF through the formula FX=hu1*WY+hu2*WF; wherein hu1 and hu2 are preset weight coefficients with values ​​greater than zero, and the larger the value of the film pasting performance value FX is, the worse the film pasting performance of the film pasting equipment is in general during the detection period; The film pasting performance value FX is numerically compared with the preset film pasting performance threshold. If the film pasting performance value FX exceeds the preset film pasting performance threshold, it indicates that the film pasting performance of the film pasting equipment during the detection period is generally poor, and a poor performance signal is generated; if the film pasting performance value FX does not exceed the preset film pasting performance threshold, it indicates that the film pasting performance of the film pasting equipment during the detection period is generally good, and a good performance signal is generated.

[0023] Embodiment 2: Figure 2 As shown, the difference between this embodiment and the first embodiment is that the film-mounting performance comprehensive evaluation module is connected to the operation and maintenance diagnosis and analysis module in communication, and the film-mounting performance comprehensive evaluation module sends the poor performance table signal to the operation and maintenance diagnosis and analysis module. When the operation and maintenance diagnosis and analysis module receives the poor performance table signal, it analyzes the operation and maintenance status of the film-mounting 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 the operation and maintenance diagnosis qualified signal or operation and maintenance diagnosis warning signal is sent to the supervision terminal. When the supervision terminal receives the operation and maintenance diagnosis warning signal, it issues a corresponding warning, which can reasonably analyze the correlation between the poor operation performance of the film laminating equipment and the operation and maintenance performance. When the management personnel receive the operation and maintenance diagnosis warning signal, they will strengthen the operation and maintenance supervision of the operation and maintenance personnel of the film laminating equipment in the future, ensure the timeliness and effect of subsequent operation and maintenance, and thus help improve the operation performance of the film laminating equipment; the specific analysis process of the operation and maintenance diagnosis analysis module is as follows: The start time and end time of the operation and maintenance of the film laminating equipment are obtained, and the time difference between the end time and the start time is calculated to obtain the maintenance and testing time. The number of maintenance and testing times that do not exceed the preset maintenance and testing time threshold during the detection period is marked as the maintenance and testing excess value, and the average of all maintenance and testing times during the detection period is calculated to obtain the maintenance and testing performance value. The interval between the start time and the end time of the last operation and maintenance of the film-sticking equipment is marked as the inter-dimensional time, the number of inter-dimensional time that exceeds the preset inter-dimensional time threshold during the detection period is marked as the inter-dimensional excess detection value, and the average of all inter-dimensional time within the detection period is calculated to obtain the inter-dimensional performance value; The maintenance and measurement over-condition value LN, the maintenance and measurement performance value LS, the maintenance and measurement over-inspection value LR and the maintenance and measurement performance value LW are numerically calculated by the formula LP=(a1*LN+a2 / (LS+1)+a3*LR+a4*LW) / 2 to obtain the operation and maintenance diagnosis value LP; wherein a1, a2, a3 and a4 are preset proportional coefficients with values ​​greater than zero, and the larger the value of the operation and maintenance diagnosis value LP is, the worse the operation and maintenance performance of the film-sticking equipment is in general during the detection period; The operation and maintenance diagnosis value LP is numerically compared with the preset operation and maintenance diagnosis threshold. If the operation and maintenance diagnosis value LP exceeds the preset operation and maintenance diagnosis threshold, it indicates that the operation and maintenance performance of the film pasting equipment during the detection period is generally poor, and the poor operating performance of the film pasting equipment is highly correlated with the operation and maintenance performance, then an operation and maintenance diagnosis warning signal is generated; if the operation and maintenance diagnosis value LP does not exceed the preset operation and maintenance diagnosis threshold, it indicates that the operation and maintenance performance of the film pasting equipment during the detection period is generally good, then a qualified operation and maintenance diagnosis signal is generated.

[0024] Embodiment 3: Figure 3 As shown, the difference between this embodiment and the first and second embodiments is that the semiconductor wafer film quality monitoring method based on artificial intelligence proposed in the present invention includes the following steps: Step 1: Before laminating the semiconductor wafer, scan the surface of the semiconductor wafer to collect a substrate image; Step 2: judging the surface quality of the semiconductor wafer through the substrate image, and generating a film-mounting preparation signal or a to-be-removed signal of the corresponding semiconductor wafer accordingly; Step 3: When the film-mounting preparation signal is generated, the film-mounting equipment mounts the film on the semiconductor wafer, monitors the film-mounting process of the semiconductor wafer, and generates a film-mounting warning signal or a film-mounting qualified signal through analysis; Step 4: After the film is attached, the die attach quality of the semiconductor wafer is analyzed, and a normal inspection signal or an abnormal inspection signal is generated through the analysis; Step 5: Analyze the film bonding performance of the semiconductor wafer during the test period through the comprehensive film bonding performance evaluation module, and generate a good performance signal or a bad performance signal through the analysis.

[0025] The working principle of the present invention is as follows: when in use, before filming the semiconductor wafer, the surface of the semiconductor wafer is scanned by the substrate surface inspection module to collect the substrate image and judge the surface quality of the semiconductor wafer. When a signal to be rejected is generated, the corresponding semiconductor wafer is not filmed, which can avoid unnecessary time waste and ensure the stability and efficiency of subsequent filming operations. When a filming preparation signal is generated, the semiconductor wafer is filmed by the filming equipment. The filming monitoring and analysis module monitors and analyzes the filming process of the semiconductor wafer. When a filming warning signal is generated, the management personnel are reminded to take corresponding control measures in time. After the filming is completed, the filming quality inspection module is used to analyze the die quality status of the semiconductor wafer to accurately judge the filming quality of each semiconductor wafer. The comprehensive evaluation module of filming performance is used to analyze the filming performance of the semiconductor wafer during the detection period. When a poor performance signal is generated, the filming equipment is inspected and repaired to ensure its subsequent filming performance, which significantly reduces the management difficulty of the management personnel and has a high degree of intelligence.

[0026] The above formulas are all dimensionless and numerical calculations. The formula is a formula obtained by collecting a large amount of data and performing software simulation to obtain the most recent real situation. The preset parameters in the formula are set by technicians in this field according to actual conditions. The preferred embodiments of the present invention disclosed above are only used to help explain the present invention. The preferred embodiments do not describe all the details in detail, nor do they limit the invention to only specific implementation methods. Obviously, many modifications and changes can be made according to the contents of this specification. This specification selects and specifically describes these embodiments in order to better explain the principles and practical applications of the present invention, so that technicians in the relevant technical field can understand and use the present invention well. The present invention is only limited by the claims and their full scope and equivalents.

Claims

1. The semiconductor wafer film quality monitoring system based on artificial intelligence is characterized by: It includes substrate surface inspection module, film pasting monitoring and analysis module, film pasting quality inspection module, film pasting performance comprehensive evaluation module and supervision terminal; Before film lamination of semiconductor wafers, the substrate surface inspection module scans the surface of the semiconductor wafer to collect a substrate image, and judges the surface quality of the semiconductor wafer through the substrate image, and accordingly generates a film lamination preparation signal or a to-be-rejected signal of the corresponding semiconductor wafer; When the film pasting preparation signal is generated, the film pasting equipment is used to paste the film on the semiconductor wafer, and the film pasting monitoring and analysis module monitors the film pasting process of the semiconductor wafer, generates a film pasting warning signal or a film pasting qualified signal through analysis, and sends the film pasting warning signal or the film pasting qualified signal to the supervision terminal; After the film pasting is completed, the film pasting quality inspection module will analyze the die pasting quality status 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 supervision terminal; The comprehensive evaluation module for film pasting performance is used to set the detection period, analyze the film pasting performance of the semiconductor wafer during the detection period, generate a good performance signal or a bad performance signal through the analysis, and send the good performance signal or the bad performance signal to the supervision terminal.

2. The semiconductor wafer film quality monitoring system based on artificial intelligence according to claim 1 is characterized in that: The specific analysis process of the substrate surface inspection module is as follows: A plurality of detection areas are set on the surface of the semiconductor wafer, and the protrusions, depressions and scratches on the surface of the semiconductor wafer are identified through the substrate image. If the corresponding detection area involves a protrusion, depression or scratch, the corresponding detection area is marked as a damaged area; The substrate image is gray-scaled and the actual gray-scale value of the corresponding detection area is collected. If the actual gray-scale value exceeds a preset gray-scale threshold, the corresponding detection area is marked as a non-clean area. The number of damaged areas and the number of non-clean areas on the semiconductor wafer are obtained and marked as damage detection values ​​and non-clean detection values ​​respectively. If the damage detection value or the non-clean detection value exceeds the corresponding preset threshold, a signal to be rejected for the corresponding semiconductor wafer is generated.

3. The semiconductor wafer film quality monitoring system based on artificial intelligence according to claim 2 is characterized in that: If both the damage detection value and the non-clean detection value do not exceed the corresponding preset threshold, the damaged area and the non-clean area are marked as poor quality areas; the clustering area of ​​the poor quality areas is obtained, the number of poor quality areas in the corresponding clustering area is marked as the poor quality clustering detection value, the poor quality clustering detection value is numerically compared with the preset poor quality clustering detection threshold, if the poor quality clustering detection value exceeds the preset poor quality clustering detection threshold, the corresponding clustering area is marked as a high clustering area; The number of high-polygon areas in the semiconductor wafer is obtained and marked as a high-polygon monitoring value, and the low-quality polygon detection value with the largest value is marked as a low-quality polygon amplitude value; a substrate evaluation value is obtained by numerically calculating the damage detection value, the non-net detection value, the high-polygon monitoring value and the low-quality polygon amplitude value, and if the substrate evaluation value exceeds a preset substrate evaluation threshold, a signal to be rejected for the corresponding semiconductor wafer is generated; If the substrate evaluation value does not exceed the preset substrate evaluation threshold, a film attaching preparation signal for the corresponding semiconductor wafer is generated.

4. The semiconductor wafer film quality monitoring system based on artificial intelligence according to claim 1 is characterized in that: The specific analysis process of the mold monitoring and analysis module is as follows: The film pasting temperature, film pasting pressure and film pasting speed of the die pasting process are collected, and the deviation value of the film pasting temperature compared with the preset film pasting suitable temperature is marked as the film pasting temperature value, the deviation value of the film pasting pressure compared with the preset film pasting suitable pressure is marked as the film pasting pressure value, and the deviation value of the film pasting speed compared with the preset film pasting suitable speed is marked as the film pasting speed value; The film pasting monitoring and analysis value is obtained by numerically calculating the film pasting temperature value, the film pasting pressure value and the film pasting speed value. If the film pasting monitoring and analysis value exceeds the preset film pasting monitoring and analysis threshold value, a film pasting warning signal is generated; If the film pasting monitoring analysis value does not exceed the preset film pasting monitoring analysis threshold, a film pasting qualified signal is generated.

5. The semiconductor wafer film quality monitoring system based on artificial intelligence according to claim 1 is characterized in that: The specific analysis process of the film quality inspection module is as follows: The actual film pasting area on the corresponding semiconductor wafer is collected, and the actual film pasting area is overlapped with the standard film pasting area to obtain the film pasting position deviation value. If the film pasting position deviation value exceeds the preset film pasting position deviation threshold, an inspection abnormality signal is generated; If the film position deviation value does not exceed the preset film position deviation threshold, the bubble and wrinkle defects on the surface of the semiconductor wafer are identified, and the bubble defects whose volume exceeds the preset bubble volume threshold are marked as high-impact bubbles, and the wrinkle defects whose volume exceeds 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, an inspection abnormality signal is generated.

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

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

8. The semiconductor wafer film quality monitoring system based on artificial intelligence according to claim 7 is characterized in that: The comprehensive film performance evaluation module is communicatively connected to the operation and maintenance diagnosis and analysis module. The comprehensive film performance evaluation module sends a poor performance signal to the operation and maintenance diagnosis and analysis module. When the operation and maintenance diagnosis and analysis module receives the poor performance signal, it analyzes the operation and maintenance status of the film 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 the operation and maintenance diagnosis early warning signal to the supervision terminal.

9. The semiconductor wafer film quality monitoring system based on artificial intelligence according to claim 8 is characterized in that: The specific analysis process of the operation and maintenance diagnosis and analysis module is as follows: The start time and end time of the operation and maintenance of the film laminating equipment are obtained, and the time difference between the end time and the start time is calculated to obtain the maintenance and testing time. The number of maintenance and testing times that do not exceed the preset maintenance and testing time threshold during the detection period is marked as the maintenance and testing excess value, and the average of all maintenance and testing times during the detection period is calculated to obtain the maintenance and testing performance value. The interval between the start time and the end time of the last operation and maintenance of the film-sticking equipment is marked as the inter-dimensional time, the number of inter-dimensional time that exceeds the preset inter-dimensional time threshold during the detection period is marked as the inter-dimensional excess detection value, and the average of all inter-dimensional time within the detection period is calculated to obtain the inter-dimensional performance value; The operation and maintenance diagnosis value is obtained by numerically calculating the maintenance and measurement excess value, maintenance and measurement performance value, inter-maintenance over-inspection value and inter-maintenance 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 quality monitoring method based on artificial intelligence, characterized in that: The method adopts the artificial intelligence-based semiconductor wafer film quality monitoring system as described in any one of claims 1-9.

Images