Chip production process supervision system and method based on Internet of Things

Through the Internet of Things-based chip production process supervision system, combined with data acquisition and feature vector analysis, the problem of lack of in-depth data mining in chip production is solved, the identification of hidden dangers and defects and the positioning of production line failures is realized, and the supervision efficiency and detection are improved. Accuracy.

CN118983241AActive Publication Date: 2024-11-19JIANGSU MANWANG SEMICON TECH CO LTD
View PDF 9 Cites 0 Cited by

Patent Information

Application Number
CN202411012810.9
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-26
Publication Date
2024-11-19
Estimated Expiration
2044-07-26

AI Technical Summary

Technical Problem

There is a lack of in-depth data mining of defective chips during the production process of existing chips, and it is impossible to determine whether there are hidden dangers and defects, which affects fault discovery and production line supervision efficiency.

Method used

Using the Internet of Things chip production process supervision system, through data acquisition, surface defect image analysis and electrical property defect detection, feature vectors are generated for correlation analysis, to judge the correlation influence between surface defects and electrical defects, and to make double defect judgments , and combine production batch data for fault analysis and feedback.

Benefits of technology

It realizes deep correlation mining of chip defects, determines hidden dangers and defects, and locates chip production line failures, which improves the detection accuracy of the regulatory system and the accuracy of fault analysis.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118983241A_ABST
    Figure CN118983241A_ABST
Patent Text Reader

Abstract

The invention discloses a chip production process supervision system and method based on the Internet of Things, and belongs to the technical field of data supervision. The method comprises the following steps: collecting chip production process data, transmitting the data to a database, carrying out label endowing operation on warehousing data, and constructing a corresponding data set; calling chip data of the basic production architecture, carrying out surface defect image data analysis and electrical property defect analysis on each chip, obtaining an analysis result, and extracting corresponding feature data to generate a corresponding feature defect vector; carrying out correlation defect analysis by combining the surface defect feature vector and the electrical defect feature vector, judging the correlation influence degree between the surface defect and the electrical defect of the current chip according to an analysis result, and carrying out double defect judgment on the current chip according to a judgment result; and performing fault analysis on a chip production flow stage by combining a chip defect analysis result of the current production batch, and performing manual end data feedback on a flow fault analysis result and a chip defect analysis result.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of data supervision, and in particular to a chip production process supervision system and method based on the Internet of Things. Background Art

[0002] Chip production process supervision is an indispensable part of semiconductor manufacturing. Its core purpose is to ensure product quality, performance and reliability while meeting market demand, regulatory requirements and corporate strategic goals;

[0003] In the current environment, the supervision of chip production is relatively strict. Chip manufacturers determine the benign condition of chips through precise observation and testing before chips leave the factory, and eliminate defective chips. However, this model is not perfect. In the existing environment, the practicality of chips can be guaranteed to a certain extent by testing and screening new products, but there is a lack of data mining for defective chips. It is impossible to determine whether the abnormal operation of defective chips is entirely caused by the observed defects. Therefore, it is impossible to determine whether there are hidden defects, which is not conducive to fault discovery of the entire chip production line. Summary of the invention

[0004] The purpose of the present invention is to provide a chip production process monitoring system and method based on the Internet of Things to solve the problems raised in the above background technology.

[0005] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0006] A chip production process supervision method based on the Internet of Things, the method comprising the following steps:

[0007] S100, collecting data from each chip production process segment, transmitting the data to the database, assigning attribute labels to each data entry, and constructing a corresponding chip data set;

[0008] S200, retrieving data of chips that have completed the basic production architecture, performing surface defect image data analysis and electrical property defect analysis on each chip, obtaining defect analysis results of each chip and extracting corresponding feature data to generate feature defect vectors corresponding to each chip;

[0009] S300, performing correlation defect analysis based on the surface defect feature vector and the electrical defect feature vector of each chip, judging the correlation influence between the surface defect and the electrical defect of the current chip according to the analysis result, and performing a double defect judgment on the current chip according to the judgment result; the double defect judgment is to judge whether the current chip has hidden defects according to the correlation analysis;

[0010] S400, performing fault analysis on the chip production process stage in combination with the chip defect analysis results of the current production batch, and performing manual end data feedback on the process fault analysis results and the chip defect analysis results.

[0011] The S100 collects data from each chip production process segment, transmits it to the database, and assigns attribute labels to each data entry. The specific steps of constructing the corresponding chip data set are as follows:

[0012] S101, inductive sensing equipment is installed in each process section of the chip production process line, and when the chip wafer is detected, the time point, the sequence data of the current chip wafer and the production operation type data are recorded; the sequence data is the position data of the current chip wafer in the current production line; the data recorded by each chip wafer in each process section on the production line are integrated and aggregated and transmitted to the database;

[0013] S102. When chip data is stored in the database, the storage order data is used as the head tag to search for in the chip database, and sub-clue tags are generated by extending the head tag; each sub-clue tag is composed of the production operation type data of the chip and the execution time period of the corresponding generation operation type; the head tag of each chip data is used as the head node, and the corresponding sub-clue tag is used as the associated sub-node to generate the corresponding element node, and the element nodes of each chip in the current batch are integrated to generate the data set of the corresponding batch of chip data.

[0014] The specific steps of S200 for retrieving data of chips that have completed the basic production architecture, performing surface defect image data analysis and electrical property defect analysis on each chip, obtaining defect analysis results of each chip and extracting corresponding feature data to generate feature defect vectors corresponding to each chip are as follows:

[0015] S201, retrieve data from a database for chips that have completed a basic production architecture; the chips that have completed a basic production architecture refer to chips that have completed basic production type operations of a product and are facing a packaging step; determine the production data of each chip in each production type operation of the production line based on the retrieved data, and perform surface defect image data detection and electrical property defect detection on the chip in combination with the generated data of each chip; extract surface images of each chip, and input the extracted image data into a surface defect image analysis model; conduct a power-on experiment on each chip, obtain the electrical operation data of the corresponding chip based on the experimental data, and input the operation data into a corresponding electrical property defect analysis model; the surface defect image analysis model is a defect analysis model constructed by a corresponding chip staff using a big data model combined with a large number of chip defect surface data images and defect type data corresponding to the corresponding defect image feature data; the electrical property defect analysis model is an electrical defect analysis model constructed by a chip staff through electrical experiments on a large number of chips, obtaining the operation data of each electrical property of the chip, fitting the corresponding electrical property operation curve by combining time data, and analyzing the curve evolution;

[0016] S202, perform defect analysis output on each chip according to the surface defect image analysis model and the electrical property defect analysis model, obtain the corresponding chip surface defect type data and defect image coordinate data and the corresponding chip electrical defect type and defect time curve data; perform feature data conversion on the surface defect data and electrical defect data of each chip respectively, and obtain feature vector data of the corresponding defect; the conversion calculation formulas are respectively in and are the surface defect feature vector and electrical defect feature vector of the corresponding chip respectively; s n is the area of ​​defects on each surface of the chip; s all is the total area of ​​defects on the chip surface; is the defect type data corresponding to each surface defect of the chip; α is the surface defect type; n is the number of surface defects of the chip; The vector constructed by the coordinates of the two endpoints with the largest distance in each surface defect image of the chip; y m is the curve amplitude of each electrical defect of the chip; all is the total amplitude of the curve of chip electrical defects; is the curve type data corresponding to each electrical defect of the chip; β is the electrical defect type; m is the number of electrical defects of the chip; It is a vector constructed by the coordinates of the points in two time periods of each defect curve segment in the chip electrical defect curve; i and u are constants; since there may be multiple surface defects and electrical defects in the chip, the area ratio and amplitude ratio are introduced as balance coefficients to perform vector transformation calculations on each defect.

[0017] The S300 combines the surface defect feature vector and the electrical defect feature vector of each chip to perform correlation defect analysis, and determines the correlation influence between the surface defect and the electrical defect of the current chip according to the analysis result. The specific steps of performing double defect determination on the current chip according to the determination result are as follows:

[0018] S301, respectively, mapping the surface defect feature vectors and electrical defect vectors of the corresponding chip in the same plane to obtain the angle data between each surface defect vector and each electrical defect vector; combining the angle data to analyze the correlation effect of each surface defect on each electrical defect, the calculation formula is: in is the influence of each surface defect on each electrical defect; is the mapping angle between each surface defect vector and each electrical defect phasor; if It is judged that the corresponding surface defect has a strong correlation effect on the corresponding electrical defect; if It is judged that the corresponding surface defect has a weak correlation effect on the corresponding electrical defect; where G re It is the system default value;

[0019] S302, based on the impact analysis of different surface defects on various electrical defects, the comprehensive surface defect data of the chip is subjected to correlation impact analysis on the comprehensive electrical defect data, and the calculation formula is: ComG is the correlation influence of the comprehensive surface defect data of the chip on the comprehensive electrical defect data; The angle is constructed after the surface defect feature vector and the electrical defect feature vector of the chip are mapped in the same plane; is the number of influences of each surface defect on each electrical defect that has a strong correlation effect on the corresponding electrical defect; n*m is the total number of influences of each surface defect on each electrical defect;

[0020] S303, combining the comprehensive surface defect data of the chip with the comprehensive electrical defect data to perform a correlation impact analysis result, and make a double defect judgment on the chip; if ComG <ComG re , then it is judged that the current chip has hidden defects; if ComG ≥ ComG re , it is judged that the current chip does not have hidden defects; among which ComG re It is the system preset value. Since the electrical operation defects of the chip are generally caused by the manufacturing defects of the chip itself, if the influence of the currently observable surface defects on the electrical defects is less than the calculation threshold, it means that the defects that affect the electrical defects of the current chip are not only the observable surface defects, so the current chip has hidden defects.

[0021] The specific steps of performing fault analysis on the chip production process stage in combination with the chip defect analysis results of the current production batch and providing manual data feedback of the chip defect analysis results and the process fault analysis results are as follows:

[0022] S401, perform fault analysis on the chip production process stage by analyzing the proportion of chips with hidden defects in the current batch of chips, and the calculation formula is: Where Q is the ratio of the number of chips with hidden dangers in the corresponding batch; M(ht) is the number of chips with hidden dangers in the current batch; M all is the number of chips in the current batch; if Q re , then conduct a fault analysis on the production operation that caused the surface defects of the chip; if Q ≥ Q re , it is recommended to stop the production line and conduct a comprehensive inspection;

[0023] S402, the chip defect analysis results and the process failure analysis results are fed back and displayed through the manual supervision port.

[0024] A chip production process monitoring system based on the Internet of Things, the system includes a chip data storage module, a defect feature vector generation module, a defect correlation analysis module and a fault feedback analysis module;

[0025] The chip data storage module collects data from each chip production process segment, transmits it to the database, assigns attribute labels to each data storage, and constructs a corresponding chip data set; the defect feature vector generation module retrieves data from chips that have completed the basic production architecture, performs surface defect image data analysis and electrical property defect analysis on each chip, obtains the defect analysis results of each chip, and extracts the corresponding feature data to generate a feature defect vector corresponding to each chip; the defect association analysis module performs an association defect analysis based on the surface defect feature vector and the electrical defect feature vector of each chip, determines the association influence between the surface defect and the electrical defect of the current chip based on the analysis results, and performs a double defect judgment on the current chip based on the judgment results; the double defect judgment is to determine whether the current chip has hidden defects based on the association analysis; the fault feedback analysis module performs a fault analysis on the chip production process stage based on the chip defect analysis results of the current production batch, and performs manual data feedback on the process fault analysis results and the chip defect analysis results.

[0026] ​The chip data storage module includes a chip data acquisition unit and a data storage unit; the chip data acquisition unit performs inductive sensing equipment in each process section of the chip production process line, and records the time point, the sequence data of the current chip wafer and the production operation type data when the chip wafer is detected; the sequence data is the position data of the current chip wafer in the current production line; the data recorded by each chip wafer in each process section on the production line are integrated and aggregated and transmitted to the database; when the chip data is stored in the data storage unit, the storage sequence data is used as the header tag to retrieve the chip library, and the sub-clue tag is generated by extending the header tag; each sub-clue tag is composed of the production operation type data of the chip and the execution time period of the corresponding generated operation type; the header tag of each chip data is used as the head node, and the corresponding sub-clue tag is used as the associated sub-node to generate the corresponding element node, and the element nodes of each chip in the current batch are integrated to generate the data set of the corresponding batch of chip data.

[0027] The defect feature vector generation module includes a defect detection unit and a defect feature vector construction unit; the defect detection unit retrieves data from a database for chips that have completed the basic production architecture; determines the production data of each chip in each production type operation of the production line based on the retrieved data, and performs surface defect image data detection and electrical property defect detection on the chip in combination with the generated data of each chip; extracts the surface image of each chip, and inputs the extracted image data into a surface defect image analysis model; conducts a power-on experiment on each chip, obtains the electrical operation data of the corresponding chip based on the experimental data, and inputs the operation data into the corresponding electrical property defect analysis model; the defect feature vector construction unit performs defect analysis output on each chip based on the surface defect image analysis model and the electrical property defect analysis model, obtains the surface defect type data and defect image coordinate data of the corresponding chip, and the electrical defect type and defect time curve data of the corresponding chip; performs feature data conversion on the surface defect data and electrical defect data of each chip respectively, and obtains feature vector data of the corresponding defect.

[0028] The defect correlation analysis module includes a defect correlation analysis unit, a comprehensive correlation analysis unit and a double judgment unit; the defect correlation analysis unit respectively maps the surface defect feature vectors of the corresponding chip with the electrical defect vector in the same plane to obtain the angle data between each surface defect vector and each electrical defect vector; analyzes the correlation influence of each surface defect on each electrical defect in combination with the angle data; judges the correlation influence of the corresponding surface defect on the corresponding electrical defect according to the analysis result; the comprehensive correlation analysis unit analyzes the influence of different surface defects on each electrical defect, and performs correlation influence analysis on the comprehensive surface defect data of the chip and the comprehensive electrical defect data; the double judgment unit performs correlation influence analysis results on the comprehensive electrical defect data in combination with the comprehensive surface defect data of the chip, and performs double defect judgment on the chip.

[0029] The fault feedback analysis module includes a production process fault analysis unit and a data feedback unit; the production process fault analysis unit performs fault analysis on the chip production process stage by analyzing the proportion of chips with hidden defects in the current batch of chips; the data feedback unit feedbacks and displays the chip defect analysis results and the process fault analysis results through the manual supervision port.

[0030] Compared with the prior art, the beneficial effects achieved by the present invention are as follows: the present invention realizes data collection, marking, defect detection, defect correlation analysis and fault feedback analysis in the chip production process through a multi-module unit; further correlation analysis is performed on the defects to the extent of conventional chip defect detection, so as to determine the correlation influence of each surface defect on its electrical operation, and conduct in-depth correlation mining between the surface defects and electrical defects of each chip, so as to obtain whether the corresponding chip has hidden defects based on the cause-effect situation between the defects; by determining the hidden defects and combining the chip production batch data, the fault location analysis of the chip production line is carried out, and fault analysis suggestions are given in the corresponding situation. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0032] Figure 1 It is a structural schematic diagram of a chip production process monitoring system based on the Internet of Things of the present invention. DETAILED DESCRIPTION

[0033] 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.

[0034] See also Figure 1 , the present invention provides a technical solution:

[0035] A chip production process supervision method based on the Internet of Things, the method comprising the following steps:

[0036] S100, collecting data from each chip production process segment, transmitting the data to the database, assigning attribute labels to each data entry, and constructing a corresponding chip data set;

[0037] S200, retrieving data of chips that have completed the basic production architecture, performing surface defect image data analysis and electrical property defect analysis on each chip, obtaining defect analysis results of each chip and extracting corresponding feature data to generate feature defect vectors corresponding to each chip;

[0038] S300, performing correlation defect analysis based on the surface defect feature vector and the electrical defect feature vector of each chip, judging the correlation influence between the surface defect and the electrical defect of the current chip according to the analysis result, and performing a double defect judgment on the current chip according to the judgment result; the double defect judgment is to judge whether the current chip has hidden defects according to the correlation analysis;

[0039] S400, performing fault analysis on the chip production process stage in combination with the chip defect analysis results of the current production batch, and performing manual end data feedback on the process fault analysis results and the chip defect analysis results.

[0040] The S100 collects data from each chip production process segment, transmits it to the database, and assigns attribute labels to each data entry. The specific steps of constructing the corresponding chip data set are as follows:

[0041] S101, inductive sensing equipment is installed in each process section of the chip production process line, and when the chip wafer is detected, the time point, the sequence data of the current chip wafer and the production operation type data are recorded; the sequence data is the position data of the current chip wafer in the current production line; the data recorded by each chip wafer in each process section on the production line are integrated and aggregated and transmitted to the database;

[0042] S102. When chip data is stored in the database, the storage order data is used as the head tag to search for in the chip database, and sub-clue tags are generated by extending the head tag; each sub-clue tag is composed of the production operation type data of the chip and the execution time period of the corresponding generation operation type; the head tag of each chip data is used as the head node, and the corresponding sub-clue tag is used as the associated sub-node to generate the corresponding element node, and the element nodes of each chip in the current batch are integrated to generate the data set of the corresponding batch of chip data.

[0043] The specific steps of S200 for retrieving data of chips that have completed the basic production architecture, performing surface defect image data analysis and electrical property defect analysis on each chip, obtaining defect analysis results of each chip and extracting corresponding feature data to generate feature defect vectors corresponding to each chip are as follows:

[0044] S201, retrieve data from the database for chips that have completed the basic production architecture; determine the production data of each chip in each production type operation of the production line based on the retrieved data, and perform surface defect image data detection and electrical property defect detection on the chip in combination with the generated data of each chip; extract the surface image of each chip and input the extracted image data into the surface defect image analysis model; conduct a power-on experiment on each chip, obtain the electrical operation data of the corresponding chip based on the experimental data, and input the operation data into the corresponding electrical property defect analysis model;

[0045] S202, perform defect analysis output on each chip according to the surface defect image analysis model and the electrical property defect analysis model, obtain the corresponding chip surface defect type data and defect image coordinate data and the corresponding chip electrical defect type and defect time curve data; perform feature data conversion on the surface defect data and electrical defect data of each chip respectively, and obtain feature vector data of the corresponding defect; the conversion calculation formulas are respectively in and are the surface defect feature vector and electrical defect feature vector of the corresponding chip respectively; s n is the area of ​​defects on each surface of the chip; s all is the total area of ​​defects on the chip surface; is the defect type data corresponding to each surface defect of the chip; α is the surface defect type; n is the number of surface defects of the chip; The vector constructed by the coordinates of the two endpoints with the largest distance in each surface defect image of the chip; y m is the curve amplitude of each electrical defect of the chip; all is the total amplitude of the curve of chip electrical defects; is the curve type data corresponding to each electrical defect of the chip; β is the electrical defect type; m is the number of electrical defects of the chip; A vector constructed from the coordinates of two time period points of each defect curve segment in the chip electrical defect curve; i and u are constants.

[0046] The S300 combines the surface defect feature vector and the electrical defect feature vector of each chip to perform correlation defect analysis, and determines the correlation influence between the surface defect and the electrical defect of the current chip according to the analysis result. The specific steps of performing double defect determination on the current chip according to the determination result are as follows:

[0047] S301, respectively, mapping the surface defect feature vectors and electrical defect vectors of the corresponding chip in the same plane to obtain the angle data between each surface defect vector and each electrical defect vector; combining the angle data to analyze the correlation effect of each surface defect on each electrical defect, the calculation formula is: in is the influence of each surface defect on each electrical defect; is the mapping angle between each surface defect vector and each electrical defect phasor; if It is judged that the corresponding surface defect has a strong correlation effect on the corresponding electrical defect; if It is judged that the corresponding surface defect has a weak correlation effect on the corresponding electrical defect; where G re It is the system default value;

[0048] S302, based on the impact analysis of different surface defects on various electrical defects, the comprehensive surface defect data of the chip is subjected to correlation impact analysis on the comprehensive electrical defect data, and the calculation formula is: ComG is the correlation influence of the comprehensive surface defect data of the chip on the comprehensive electrical defect data; The angle is constructed after the surface defect feature vector and the electrical defect feature vector of the chip are mapped in the same plane; is the number of influences of each surface defect on each electrical defect that has a strong correlation effect on the corresponding electrical defect; n*m is the total number of influences of each surface defect on each electrical defect;

[0049] S303, combining the comprehensive surface defect data of the chip with the comprehensive electrical defect data to perform a correlation impact analysis result, and make a double defect judgment on the chip; if ComG <ComG re , then it is judged that the current chip has hidden defects; if ComG ≥ ComG re , it is judged that the current chip does not have hidden defects; among which ComG re This is the system default value.

[0050] The specific steps of performing fault analysis on the chip production process stage in combination with the chip defect analysis results of the current production batch and providing manual data feedback of the chip defect analysis results and the process fault analysis results are as follows:

[0051] S401, perform fault analysis on the chip production process stage by analyzing the proportion of chips with hidden defects in the current batch of chips, and the calculation formula is: Where Q is the ratio of the number of chips with hidden dangers in the corresponding batch; M(ht) is the number of chips with hidden dangers in the current batch; M all is the number of chips in the current batch; if Q re , then conduct a fault analysis on the production operation that caused the surface defects of the chip; if Q ≥ Q re , it is recommended to stop the production line and conduct a comprehensive inspection;

[0052] S402, the chip defect analysis results and the process failure analysis results are fed back and displayed through the manual supervision port.

[0053] A chip production process monitoring system based on the Internet of Things, the system includes a chip data storage module, a defect feature vector generation module, a defect correlation analysis module and a fault feedback analysis module;

[0054] The chip data storage module collects data from each chip production process segment, transmits it to the database, assigns attribute labels to each data storage, and constructs a corresponding chip data set; the defect feature vector generation module retrieves data from chips that have completed the basic production architecture, performs surface defect image data analysis and electrical property defect analysis on each chip, obtains the defect analysis results of each chip, and extracts the corresponding feature data to generate a feature defect vector corresponding to each chip; the defect association analysis module performs an association defect analysis based on the surface defect feature vector and the electrical defect feature vector of each chip, determines the association influence between the surface defect and the electrical defect of the current chip based on the analysis results, and performs a double defect judgment on the current chip based on the judgment results; the double defect judgment is to determine whether the current chip has hidden defects based on the association analysis; the fault feedback analysis module performs a fault analysis on the chip production process stage based on the chip defect analysis results of the current production batch, and performs manual data feedback on the process fault analysis results and the chip defect analysis results.

[0055] ​The chip data storage module includes a chip data acquisition unit and a data storage unit; the chip data acquisition unit performs inductive sensing equipment in each process section of the chip production process line, and records the time point, the sequence data of the current chip wafer and the production operation type data when the chip wafer is detected; the sequence data is the position data of the current chip wafer in the current production line; the data recorded by each chip wafer in each process section on the production line are integrated and aggregated and transmitted to the database; when the chip data is stored in the data storage unit, the storage sequence data is used as the header tag to retrieve the chip library, and the sub-clue tag is generated by extending the header tag; each sub-clue tag is composed of the production operation type data of the chip and the execution time period of the corresponding generated operation type; the header tag of each chip data is used as the head node, and the corresponding sub-clue tag is used as the associated sub-node to generate the corresponding element node, and the element nodes of each chip in the current batch are integrated to generate the data set of the corresponding batch of chip data.

[0056] The defect feature vector generation module includes a defect detection unit and a defect feature vector construction unit; the defect detection unit retrieves data from a database for chips that have completed the basic production architecture; determines the production data of each chip in each production type operation of the production line based on the retrieved data, and performs surface defect image data detection and electrical property defect detection on the chip in combination with the generated data of each chip; extracts the surface image of each chip, and inputs the extracted image data into a surface defect image analysis model; conducts a power-on experiment on each chip, obtains the electrical operation data of the corresponding chip based on the experimental data, and inputs the operation data into the corresponding electrical property defect analysis model; the defect feature vector construction unit performs defect analysis output on each chip based on the surface defect image analysis model and the electrical property defect analysis model, obtains the surface defect type data and defect image coordinate data of the corresponding chip, and the electrical defect type and defect time curve data of the corresponding chip; performs feature data conversion on the surface defect data and electrical defect data of each chip respectively, and obtains feature vector data of the corresponding defect.

[0057] The defect correlation analysis module includes a defect correlation analysis unit, a comprehensive correlation analysis unit and a double judgment unit; the defect correlation analysis unit respectively maps the surface defect feature vectors of the corresponding chip with the electrical defect vector in the same plane to obtain the angle data between each surface defect vector and each electrical defect vector; analyzes the correlation influence of each surface defect on each electrical defect in combination with the angle data; judges the correlation influence of the corresponding surface defect on the corresponding electrical defect according to the analysis result; the comprehensive correlation analysis unit analyzes the influence of different surface defects on each electrical defect, and performs correlation influence analysis on the comprehensive surface defect data of the chip and the comprehensive electrical defect data; the double judgment unit performs correlation influence analysis results on the comprehensive electrical defect data in combination with the comprehensive surface defect data of the chip, and performs double defect judgment on the chip.

[0058] The fault feedback analysis module includes a production process fault analysis unit and a data feedback unit; the production process fault analysis unit performs fault analysis on the chip production process stage by analyzing the proportion of chips with hidden defects in the current batch of chips; the data feedback unit feeds back and displays the chip defect analysis results and process fault analysis results through the manual supervision port;

[0059] In the example:

[0060] At present, a certain chip manufacturing factory is equipped with the chip production supervision system of the present invention. When it produces a batch of chips, it uses inductive sensing equipment in each process section of the chip production process line. When the chip wafer is detected, the time point, the sequence data of the current chip wafer and the production operation type data are recorded; the sequence data is the position data of the current chip wafer in the current production line; the data recorded by each chip wafer in each process section on the production line are integrated and aggregated and transmitted to the database; when the chip data is stored in the warehouse, the storage sequence data is used as the head tag to retrieve the chip library, and the sub-clue tag is generated by extending the head tag; wherein each sub-clue tag is composed of the production operation type data of the chip and the execution time period of the corresponding generated operation type; the head tag of each chip data is used as the head node, and the corresponding sub-clue tag is used as the associated sub-node to generate the corresponding element node, and the element nodes of each chip in the current batch are integrated to generate the data set of the corresponding batch of chip data;

[0061] Retrieve data from the database for chips that have completed the basic production architecture; determine the production data of each chip in each production type operation on the production line based on the retrieved data, and perform surface defect image data detection and electrical property defect detection on the chip in combination with the generated data of each chip; extract the surface image of each chip, and input the extracted image data into the surface defect image analysis model; conduct a power-on experiment on each chip, obtain the electrical operation data of the corresponding chip based on the experimental data, and input the operation data into the corresponding electrical property defect analysis model; perform defect analysis output on each chip based on the surface defect image analysis model and the electrical property defect analysis model, and obtain the corresponding chip surface defect type data and defect image coordinate data as well as the corresponding chip electrical defect type and defect time curve data; perform feature data conversion on the surface defect data and electrical defect data of each chip respectively, and obtain the feature vector data of the corresponding defect; the conversion calculation formulas are as follows: The surface defect feature vectors and electrical defect vectors of the corresponding chips are mapped in the same plane to obtain the angle data between each surface defect vector and each electrical defect vector; the correlation effect of each surface defect on each electrical defect is analyzed in combination with the angle data, and the calculation formula is: According to the calculation results, each chip in the current batch is judged. If It is judged that the corresponding surface defect has a strong correlation effect on the corresponding electrical defect; if It is judged that the corresponding surface defect has a weak correlation effect on the corresponding electrical defect; where G re It is the system default value;

[0062] According to the impact analysis of different surface defects on various electrical defects, the correlation impact analysis of the comprehensive surface defect data of the chip on the comprehensive electrical defect data is performed, and the calculation formula is: According to the calculation results, the chips in the current batch are judged. If ComG <ComG re , then it is judged that the current chip has hidden defects; if ComG ≥ ComG re , it is judged that the current chip does not have hidden defects; among which ComG re It is the system default value;

[0063] By analyzing the proportion of chips with hidden defects in the current batch of chips, the chip production process stage is analyzed for failures. The calculation formula is: Since the current calculation result is Q re , a fault analysis is conducted on the production operation that causes the surface defects of the chip; the chip defect analysis results and process fault analysis results are fed back and displayed through the manual supervision port.

[0064] ​It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0065] Finally, it should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art can still modify the technical solutions described in the aforementioned embodiments or replace some of the technical features therein by equivalents. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A chip production process supervision method based on the Internet of Things, characterized by: The method comprises the following steps: S100, collecting data from each chip production process segment, transmitting the data to the database, assigning attribute labels to each data entry, and constructing a corresponding chip data set; S200, retrieving data of chips that have completed the basic production architecture, performing surface defect image data analysis and electrical property defect analysis on each chip, obtaining defect analysis results of each chip and extracting corresponding feature data to generate feature defect vectors corresponding to each chip; S300, performing correlation defect analysis based on the surface defect feature vector and the electrical defect feature vector of each chip, judging the correlation influence between the surface defect and the electrical defect of the current chip according to the analysis result, and performing a double defect judgment on the current chip according to the judgment result; the double defect judgment is to judge whether the current chip has hidden defects according to the correlation analysis; S400, performing fault analysis on the chip production process stage in combination with the chip defect analysis results of the current production batch, and performing manual end data feedback on the process fault analysis results and the chip defect analysis results.

2. The chip production process supervision method based on the Internet of Things according to claim 1 is characterized in that: The S100 collects data from each chip production process segment, transmits it to the database, and assigns attribute labels to each data entry. The specific steps of constructing the corresponding chip data set are as follows: S101, inductive sensing equipment is installed in each process section of the chip production process line, and when the chip wafer is detected, the time point, the sequence data of the current chip wafer and the production operation type data are recorded; the sequence data is the position data of the current chip wafer in the current production line; the data recorded by each chip wafer in each process section on the production line are integrated and aggregated and transmitted to the database; S102. When chip data is stored in the database, the storage order data is used as the head tag to search for in the chip database, and sub-clue tags are generated by extending the head tag; each sub-clue tag is composed of the production operation type data of the chip and the execution time period of the corresponding generation operation type; the head tag of each chip data is used as the head node, and the corresponding sub-clue tag is used as the associated sub-node to generate the corresponding element node, and the element nodes of each chip in the current batch are integrated to generate the data set of the corresponding batch of chip data.

3. The chip production process supervision method based on the Internet of Things according to claim 2 is characterized in that: The specific steps of S200 for retrieving data of chips that have completed the basic production architecture, performing surface defect image data analysis and electrical property defect analysis on each chip, obtaining defect analysis results of each chip and extracting corresponding feature data to generate feature defect vectors corresponding to each chip are as follows: S201, retrieve data from the database for chips that have completed the basic production architecture; determine the production data of each chip in each production type operation of the production line based on the retrieved data, and perform surface defect image data detection and electrical property defect detection on the chip in combination with the generated data of each chip; extract the surface image of each chip and input the extracted image data into the surface defect image analysis model; conduct a power-on experiment on each chip, obtain the electrical operation data of the corresponding chip based on the experimental data, and input the operation data into the corresponding electrical property defect analysis model; S202, performing defect analysis output on each chip according to the surface defect image analysis model and the electrical property defect analysis model, obtaining surface defect type data and defect image coordinate data of the corresponding chip and electrical defect type and defect time curve data of the corresponding chip; performing feature data conversion on the surface defect data and electrical defect data of each chip respectively, and obtaining feature vector data of the corresponding defect; The conversion calculation formulas are: in and are the surface defect feature vector and electrical defect feature vector of the corresponding chip respectively; s n is the area of ​​defects on each surface of the chip; s all is the total area of ​​defects on the chip surface; is the defect type data corresponding to each surface defect of the chip; α is the surface defect type; n is the number of surface defects of the chip; The vector constructed by the coordinates of the two endpoints with the largest distance in each surface defect image of the chip; y m is the curve amplitude of each electrical defect of the chip; y all is the total amplitude of the curve of chip electrical defects; is the curve type data corresponding to each electrical defect of the chip; β is the electrical defect type; m is the number of electrical defects of the chip; A vector constructed from the coordinates of two time period points of each defect curve segment in the chip electrical defect curve; i and u are constants.

4. The chip production process supervision method based on the Internet of Things according to claim 3 is characterized in that: The S300 combines the surface defect feature vector and the electrical defect feature vector of each chip to perform correlation defect analysis, and determines the correlation influence between the surface defect and the electrical defect of the current chip according to the analysis result. The specific steps of performing double defect determination on the current chip according to the determination result are as follows: S301, respectively, mapping the surface defect feature vectors and electrical defect vectors of the corresponding chip in the same plane to obtain the angle data between each surface defect vector and each electrical defect vector; combining the angle data to analyze the correlation effect of each surface defect on each electrical defect, the calculation formula is: in is the influence of each surface defect on each electrical defect; is the mapping angle between each surface defect vector and each electrical defect phasor; if It is judged that the corresponding surface defect has a strong correlation effect on the corresponding electrical defect; if It is judged that the corresponding surface defect has a weak correlation effect on the corresponding electrical defect; where G re It is the system default value; S302, based on the impact analysis of different surface defects on various electrical defects, the comprehensive surface defect data of the chip is subjected to correlation impact analysis on the comprehensive electrical defect data, and the calculation formula is: ComG is the correlation influence of the comprehensive surface defect data of the chip on the comprehensive electrical defect data; The angle is constructed after the surface defect feature vector and the electrical defect feature vector of the chip are mapped in the same plane; is the number of influences of each surface defect on each electrical defect for which the corresponding surface defect has a strong correlation influence on the corresponding electrical defect; n*m is the total number of calculated effects of each surface defect on each electrical defect; S303, combining the comprehensive surface defect data of the chip with the comprehensive electrical defect data to perform a correlation impact analysis result, and make a double defect judgment on the chip; if ComG <ComG re , then it is judged that the current chip has hidden defects; if ComG ≥ ComG re , it is judged that the current chip does not have hidden defects; among which ComG re This is the system default value.

5. The chip production process supervision method based on the Internet of Things according to claim 4 is characterized in that: The specific steps of performing fault analysis on the chip production process stage in combination with the chip defect analysis results of the current production batch and providing manual data feedback of the chip defect analysis results and the process fault analysis results are as follows: S401, perform fault analysis on the chip production process stage by analyzing the proportion of chips with hidden defects in the current batch of chips, and the calculation formula is: Where Q is the ratio of the number of chips with hidden dangers in the corresponding batch; M(ht) is the number of chips with hidden dangers in the current batch; M all is the number of chips in the current batch; if Q re , then conduct a fault analysis on the production operation that caused the surface defects of the chip; if Q ≥ Q re , it is recommended to stop the production line and conduct a comprehensive inspection;​ S402, the chip defect analysis results and the process failure analysis results are fed back and displayed through the manual supervision port.

6. A chip production process monitoring system based on the Internet of Things, characterized by: The system includes a chip data storage module, a defect feature vector generation module, a defect association analysis module and a fault feedback analysis module; The chip data storage module collects data from each chip production process segment, transmits it to the database, assigns attribute labels to each data storage, and constructs a corresponding chip data set; the defect feature vector generation module retrieves data from chips that have completed the basic production architecture, performs surface defect image data analysis and electrical property defect analysis on each chip, obtains the defect analysis results of each chip, and extracts the corresponding feature data to generate a feature defect vector corresponding to each chip; The defect correlation analysis module combines the surface defect feature vector and the electrical defect feature vector of each chip to perform correlation defect analysis, judges the correlation influence between the surface defect and the electrical defect of the current chip according to the analysis result, and performs a double defect judgment on the current chip according to the judgment result; the double defect judgment is to judge whether the current chip has hidden defects according to the correlation analysis; The fault feedback analysis module performs fault analysis on the chip production process stage in combination with the chip defect analysis results of the current production batch, and performs manual end data feedback on the process fault analysis results and the chip defect analysis results.

7. The chip production process monitoring system based on the Internet of Things according to claim 6 is characterized in that: The chip data storage module includes a chip data acquisition unit and a data storage unit; the chip data acquisition unit performs inductive sensing equipment in each process section of the chip production process line, and records the time point, the sequence data of the current chip wafer and the production operation type data when the chip wafer is detected; the sequence data is the position data of the current chip wafer in the current production line; The data recorded by each chip wafer in each process segment on the production line are integrated and aggregated and transmitted to the database; when the chip data is entered into the database, the data entry unit uses the entry sequence data as the head tag to retrieve the chip library, and generates a sub-clue tag by extending the head tag; wherein each sub-clue tag is composed of the production operation type data of the chip and the execution time period of the corresponding generated operation type; the head tag of each chip data is used as the head node, and the corresponding sub-clue tag is used as the associated sub-node to generate the corresponding element node, and the element nodes of each chip in the current batch are integrated to generate the data set of the corresponding batch of chip data.

8. The chip production process monitoring system based on the Internet of Things according to claim 7 is characterized in that: The defect feature vector generation module includes a defect detection unit and a defect feature vector construction unit; The defect detection unit retrieves data from the database for chips that have completed the basic production architecture; determines the production data of each chip in each production type operation of the production line based on the retrieved data, and performs surface defect image data detection and electrical property defect detection on the chip in combination with the generated data of each chip; extracts the surface image of each chip, and inputs the extracted image data into the surface defect image analysis model; conducts a power-on experiment on each chip, obtains the electrical operation data of the corresponding chip based on the experimental data, and inputs the operation data into the corresponding electrical property defect analysis model; the defect feature vector construction unit performs defect analysis output on each chip based on the surface defect image analysis model and the electrical property defect analysis model, obtains the corresponding chip surface defect type data and defect image coordinate data and the corresponding chip electrical defect type and defect time curve data; performs feature data conversion on the surface defect data and electrical defect data of each chip respectively, and obtains feature vector data of the corresponding defect.

9. The chip production process monitoring system based on the Internet of Things according to claim 8 is characterized in that: The defect correlation analysis module includes a defect correlation analysis unit, a comprehensive correlation analysis unit and a double judgment unit; The defect correlation analysis unit maps the surface defect feature vectors of the corresponding chip to the electrical defect vectors in the same plane, respectively, to obtain the angle data between each surface defect vector and each electrical defect vector; analyzes the correlation influence of each surface defect on each electrical defect in combination with the angle data; judges the correlation influence of the corresponding surface defect on the corresponding electrical defect based on the analysis results; the comprehensive correlation analysis unit analyzes the influence of different surface defects on each electrical defect, and performs correlation influence analysis on the comprehensive surface defect data of the chip and the comprehensive electrical defect data; the double judgment unit performs correlation influence analysis results on the comprehensive electrical defect data in combination with the comprehensive surface defect data of the chip, and performs double defect judgment on the chip.

10. The chip production process monitoring system based on the Internet of Things according to claim 9 is characterized in that: The fault feedback analysis module includes a production process fault analysis unit and a data feedback unit; the production process fault analysis unit performs fault analysis on the chip production process stage by analyzing the proportion of chips with hidden defects in the current batch of chips; the data feedback unit feedbacks and displays the chip defect analysis results and the process fault analysis results through the manual supervision port.

Citation Information

Patent Citations

  • Production fault management analysis method

    CN108268892A

  • Optical chip production detection system and method based on Internet of Things

    CN116068378A

  • Intelligent monitoring and early warning system for wafer production

    CN116978834A

  • Chip appearance defect automatic detection system based on machine vision

    CN117686516A

  • Control chip automatic data processing system and method based on Internet of Things

    CN117762069A