CN118608316A - A semiconductor device image data detection system and method based on artificial intelligence - Google Patents

Abstract

The invention discloses a semiconductor device image data detection system and method based on artificial intelligence, and belongs to the technical field of semiconductor device detection. Wherein the system comprises: the historical data analysis module is used for analyzing historical detection data of the semiconductor device; the image data acquisition module is used for acquiring image data to perform various detections to obtain defect type detection data; a defect processing strategy module for determining a defect processing strategy according to the defect type detection data; the defect reason positioning module is used for outputting a defect reason sequence according to the occurrence probability of the defect reason; determining defect reasons corresponding to the defects; and the defect influence evaluation module is used for evaluating the influence of the defects on the production and manufacturing process of the semiconductor device according to the defect type detection data, the defect processing strategy, the defect reasons and the defect reason types. The invention can realize defect detection and evaluate the production progress by analyzing and detecting the image data of the semiconductor device.

Description

A semiconductor device image data detection system and method based on artificial intelligence

Technical Field

The present invention relates to the field of semiconductor device detection, and in particular to an artificial intelligence-based semiconductor device image data detection system and method.

Background Art

The semiconductor industry is one of the fastest growing industries in the world and is widely used in aviation, aerospace, medical, automotive, electronics and other fields. The semiconductor device production process mainly consists of four parts, namely wafer manufacturing, wafer testing, chip packaging and post-packaging testing. The current defect detection in the semiconductor device production process includes acquiring wafer image data for various defect detections, electrical testing after wafer processing, packaging defect detection, and so on. The existing technology is unable to quickly locate defects in the production process of semiconductor devices and evaluate the impact of defect location results on production progress. Therefore, the present invention provides a semiconductor device image data detection system and method based on artificial intelligence.

Summary of the invention

The present invention aims to solve at least one of the technical problems existing in the prior art. To this end, the present invention proposes a semiconductor device image data detection system and method based on artificial intelligence, which can detect defects of semiconductor devices, locate the causes of defects and evaluate production progress.

On the one hand, an embodiment of the present invention provides an artificial intelligence-based semiconductor device image data detection system, including: a historical data analysis module, used to analyze historical detection data of image detection items in a semiconductor device manufacturing process, to obtain each defect type and the defect cause corresponding to the defect type and the probability of occurrence of the defect cause; to classify each defect cause according to whether equipment maintenance is required and whether the equipment maintenance time needs to be modified, to obtain a first defect cause type, a second defect cause type and a third defect cause type; an image data acquisition module, used to acquire first image data of a semiconductor device manufacturing process, and perform various detections based on the first image data to obtain defect type detection data; the defect type detection data includes: defect id, defect type, defect The module comprises a defect handling strategy module, which is used to determine the defect handling strategy according to the defect type detection data; the defect handling strategy is to set the handling strategy according to the defect type and defect severity, including rework and scrapping; a defect cause location module, which is used to sort the defect causes in the defect cause set according to the defect cause occurrence probability, and output the defect cause sequence; and in response to the defect cause information fed back by the technician, determine the defect cause corresponding to each defect ID; a defect impact assessment module, which is used to assess the impact of defects on the semiconductor device production process according to the defect type detection data, the defect handling strategy, the defect cause and the defect cause type, and output the assessment result.

According to some embodiments of the present invention, the defect impact assessment module includes: a production information acquisition unit for acquiring the production efficiency of each node of semiconductor production, the rework time corresponding to each production node of each image detection item, the equipment repair time and equipment regular maintenance cycle corresponding to each defect cause; a production time prediction unit for obtaining the production time estimate according to the defect type detection data and the defect handling strategy :

in, is the number of wafers, is the wafer processing efficiency of the i-th node, A is the total number of wafer processing nodes, is the number of semiconductor devices, is the processing efficiency of the jth node after wafer cutting, B is the total number of processing nodes after wafer cutting, M is the number of defects detected by the first image data, is the defect of the kth defect The processing time corresponding to the defect handling strategy, For the Defect type The defect cause location time and the corresponding equipment repair and maintenance time, The time required for each regular maintenance of the equipment, It is the initial value of the equipment's regular maintenance cycle; and, the defect data of defects occurring in the same object under test are merged and processed, and the processing time corresponding to the defect processing strategy of the defect is adjusted according to the merged processing result.

According to some embodiments of the present invention, the defect The processing time corresponding to the defect handling strategy is:

in, When the defect handling strategy is rework, the node that needs to be reworked is located according to the defect type and the required rework time is determined according to the reworked node. When the defect handling strategy is scrapping, the current node is determined according to the defect type as a node before or after wafer cutting, thereby determining the number of reprocessing and the time required for reprocessing. is a function that determines the defect handling strategy based on the defect type and defect severity of the current defect. They respectively indicate that the defect handling strategies are rework or scrapping.

According to some embodiments of the present invention, the defect The defect cause location time and the corresponding equipment repair and maintenance time are:

in, The defect cause location time is determined based on the time when the technicians feedback the defect cause information. is a function that determines the equipment repair time based on the defect cause of the current defect, The additional equipment maintenance time after resetting the equipment regular maintenance cycle according to the cause of the current defect, The defect cause identified for the current defect, Defect cause sets for the first, second and third defect cause types respectively.

According to some embodiments of the present invention, the system includes: a defect impact severity level module, which is used to set a number of thresholds with different numerical values and a number of defect impact severity levels, and associate the thresholds with the defect impact severity levels one by one, so that the larger the threshold, the higher the defect impact severity level associated with the threshold; calculate the time difference between a preset production time and the estimated production time, and determine the defect impact severity level based on the comparison result of the time difference with the threshold.

According to some embodiments of the present invention, the image data acquisition module is used to perform various inspections based on the first image data to obtain defect type detection data; wherein the various inspections include: wafer defect detection, wire bond defect detection, and package defect detection.

According to some embodiments of the present invention, the system further includes: a second image data detection module, which is used to obtain finished product image data of the processed semiconductor device and second image data of the semiconductor device after being transported to the place of use, and to perform second defect detection on the finished product image data and the second image data respectively to obtain second defect detection data; and to determine the defect type and defect rate caused during transportation according to the second defect detection data. A third image data detection module is used to obtain the first semi-processed image data after each production and processing link is completed and the second semi-processed image data transported to the next production and processing node, and to obtain the defect results before and after transportation through defect detection, and to obtain the defect rate and defect type of the transportation process of the node.

According to some embodiments of the present invention, the defect impact assessment module is used to calculate the second production time estimate according to the defect rate of the transportation process. :

in, is the defect rate, An estimate of the current production time for the preset.

According to some embodiments of the present invention, the second image data detection module is used to perform binarization image processing and morphological processing on the image data and obtain detection features of the image, and judge the target defects and defect locations in each image data based on the detection features and preset feature detection standards.

The system of the embodiment of the present invention includes at least the following beneficial effects: the embodiment of the present invention obtains various types of historical image defect detection information in the process of semiconductor device manufacturing, statistically analyzes the defect causes corresponding to various types of defects, obtains the probability of occurrence of various types of defect causes, and determines whether equipment maintenance is required to overcome the defect causes according to whether the defect causes are systematic or accidental. In the case of equipment maintenance, the equipment maintenance time required for each defect cause is obtained. The defect causes are classified according to whether equipment maintenance is required and whether the equipment regular maintenance cycle needs to be adjusted to evaluate their impact on the production process. The embodiment of the present invention also generates a defect cause recommendation sequence through the statistically obtained probability of occurrence of various types of defect causes. When a defect occurs, the defect type is determined and the defect cause recommendation sequence is provided to the technician to improve the efficiency of the technician in locating the defect cause. The embodiment of the present invention determines the corresponding defect handling strategy according to the current defect type and defect severity through a preset defect handling strategy. The embodiment of the present invention statistically evaluates the impact of the defect handling strategy in the semiconductor manufacturing process on the production process. The embodiment of the present invention obtains the image data information of the current semiconductor device production and manufacturing to evaluate the impact of the image detection results on the production progress, which is convenient for the manufacturer to adjust the production plan and supply plan.

On the other hand, an embodiment of the present invention provides a semiconductor device image data detection method based on artificial intelligence, comprising the following steps: S100, analyzing historical detection data of image detection items in a semiconductor device manufacturing process to obtain each defect type and the defect cause corresponding to the defect type and the probability of occurrence of the defect cause; classifying each defect cause according to whether equipment maintenance is required and whether the equipment maintenance time needs to be modified to obtain a first defect cause type, a second defect cause type and a third defect cause type; S200, acquiring first image data of a semiconductor device manufacturing process, and performing various detections based on the first image data to obtain defect type detection data; the defect type detection data includes: defect id, defect S300, determining a defect handling strategy based on the defect type detection data; the defect handling strategy is to set a handling strategy based on the defect type and defect severity, including rework and scrapping; S400, sorting the defect causes in the defect cause set according to the defect cause occurrence probability, and outputting a defect cause sequence; and in response to the defect cause information fed back by the technician, determining the defect cause corresponding to each defect ID; S500, evaluating the impact of defects on the semiconductor device production and manufacturing process based on the defect type detection data, the defect handling strategy, the defect cause and the defect cause type, and outputting the evaluation result.

The method of the embodiment of the present invention includes at least the following beneficial effects: the embodiment of the present invention obtains various types of historical image defect detection information in the process of semiconductor device manufacturing, statistically analyzes the defect causes corresponding to various types of defects, obtains the probability of occurrence of various types of defect causes, and determines whether equipment maintenance is required to overcome the defect causes according to whether the defect causes are systematic or accidental. In the case of equipment maintenance, the equipment maintenance time required for each defect cause is obtained. The defect causes are classified according to whether equipment maintenance is required and whether the equipment regular maintenance cycle needs to be adjusted to evaluate their impact on the production process. The embodiment of the present invention also generates a defect cause recommendation sequence through the statistically obtained probability of occurrence of various types of defect causes. When a defect occurs, the defect type is determined and the defect cause recommendation sequence is provided to the technician to improve the efficiency of the technician in locating the defect cause. The embodiment of the present invention determines the corresponding defect handling strategy according to the current defect type and defect severity through a preset defect handling strategy. The embodiment of the present invention statistically evaluates the impact of the defect handling strategy in the semiconductor manufacturing process on the production process. The embodiment of the present invention obtains the image data information of the current semiconductor device production and manufacturing to evaluate the impact of the image detection results on the production progress, which is convenient for the manufacturer to adjust the production plan and supply plan.

Additional aspects and advantages of the present invention will be given in part in the following description and in part will be obvious from the following description, or will be learned through practice of the present invention.

BRIEF DESCRIPTION OF THE DRAWINGS

The above and/or additional aspects and advantages of the present invention will become apparent and easily understood from the description of the embodiments in conjunction with the following drawings, in which:

FIG1 is a schematic block diagram of modules of a system according to an embodiment of the present invention;

FIG. 2 is a schematic flow chart of a method according to an embodiment of the present invention.

Reference numerals:

Historical data analysis module 100, image data acquisition module 200, defect handling strategy module 300, defect cause location module 400, defect impact assessment module 500.

DETAILED DESCRIPTION

Embodiments of the present invention are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and cannot be understood as limiting the present invention.

In the description of the present invention, "several" means one or more, "more" means two or more, "greater than", "less than", "exceed" etc. are understood as not including the number itself, and "above", "below", "within" etc. are understood as including the number itself. If there is a description of "first" or "second", it is only used for the purpose of distinguishing the technical features, and cannot be understood as indicating or implying the relative importance or implicitly indicating the number of the indicated technical features or implicitly indicating the order of the indicated technical features.

The embodiment of the present invention is applicable to image detection of semiconductor devices. The embodiment of the present invention obtains various types of historical image defect detection information for image defect detection during the manufacturing process of semiconductor devices, statistically analyzes the defect causes corresponding to various types of defects, obtains the probability of occurrence of various types of defect causes, and determines whether equipment maintenance is required to overcome the defect causes according to whether the defect causes are systematic or accidental. In the case of equipment maintenance, the equipment maintenance time required for each defect cause is obtained. In particular, the overcoming of individual defect causes requires adjustment of the regular maintenance time of the equipment. Therefore, the embodiment of the present invention classifies the defect causes according to whether equipment maintenance is required and whether the regular maintenance cycle of the equipment needs to be adjusted, so as to evaluate its impact on the production process. The embodiment of the present invention also generates a defect cause recommendation sequence by statistically obtaining the probability of occurrence of various types of defect causes. When a defect occurs, while determining the defect type, the defect cause recommendation sequence is provided to the technician to improve the efficiency of the technician in locating the defect cause, and at the same time, the type of the current defect cause is determined according to the defect cause information fed back by the technician. The embodiment of the present invention further determines the corresponding defect handling strategy according to the current defect type and defect severity through a preset defect handling strategy. The embodiment of the present invention statistically evaluates the impact of defect handling strategies in the semiconductor manufacturing process on production and manufacturing. The embodiment of the present invention obtains image data information of current semiconductor device manufacturing to evaluate the impact of image detection results on production progress, so as to facilitate manufacturers to adjust production plans and supply plans.

1 , an embodiment of the present invention provides an artificial intelligence-based semiconductor device image data detection system, comprising:

The historical data analysis module 100 is used to analyze the historical detection data of the image detection items in the semiconductor device manufacturing process to obtain each defect type and the defect cause corresponding to the defect type and the probability of occurrence of the defect cause; each defect cause is classified according to whether equipment maintenance is required and whether the equipment maintenance time needs to be modified to obtain a first defect cause type, a second defect cause type and a third defect cause type. In this embodiment, each defect cause is classified considering the different operations required to overcome each defect cause. For example, when the defect cause is a human error in software and equipment operation, the equipment does not need to be repaired. For another example, a single wafer defect is a defect that occurs randomly and has a high probability of occurring in the lithography area. Its probability of occurrence is related to the operating time of the machine, so it is necessary to optimize the regular maintenance time of the equipment.

The image data acquisition module 200 is used to acquire the first image data of the semiconductor device manufacturing process, and perform various tests based on the first image data to obtain defect type detection data; the defect type detection data includes: defect id, defect type, defect quantity, defect severity, defect cause set and defect cause occurrence probability. The defect detection method involved in this embodiment includes: an automatic optical detection system and a scanning electron microscope detection system. The automatic optical detection system is based on optical principles. The main method is to illuminate the target to be measured by designing an illumination system (divided into bright field, dark field, transmission field and other imaging methods), use the imaging system to image the object to be measured, and convert it into a digital image signal through an image sensor (CMOS/CCD). The upper computer system performs image analysis to realize defect detection. The scanning electron microscope detection system bombards the surface of the sample to be measured by gathering an extremely narrow electron beam with extremely high energy, and collects the particles generated by the interaction between the scanning beam and the material point by point, and obtains various physical information from it.

The defect handling strategy module 300 is used to determine the defect handling strategy according to the defect type detection data; the defect handling strategy is to set the handling strategy according to the defect type and defect severity, including rework and scrapping.

The defect cause location module 400 is used to sort the defect causes in the defect cause set according to the defect cause occurrence probability and output the defect cause sequence; and in response to the defect cause information fed back by the technician, determine the defect cause corresponding to each defect ID.

The defect impact assessment module 500 is used to assess the impact of defects on the semiconductor device manufacturing process based on defect type detection data, defect handling strategy, defect cause and defect cause type, and output the assessment result.

In some embodiments, the defect impact assessment module 500 includes: a production information acquisition unit for acquiring the production efficiency of each node of semiconductor production, the rework time corresponding to each production node of each image detection item, the equipment repair time and equipment regular maintenance cycle corresponding to each defect cause; a production time prediction unit for obtaining the production time estimate according to the defect type detection data and the defect handling strategy :

in, is the number of wafers, is the wafer processing efficiency of the i-th node, A is the total number of wafer processing nodes, is the number of semiconductor devices, is the processing efficiency of the jth node after wafer cutting, B is the total number of processing nodes after wafer cutting, M is the number of defects detected by the first image data, is the defect of the kth defect The processing time corresponding to the defect handling strategy, For the Defect type The defect cause location time and the corresponding equipment repair and maintenance time, The time required for each regular maintenance of the equipment, It is the initial value of the equipment's regular maintenance cycle; and, the defect data of defects occurring in the same object under test are merged and processed, and the processing time corresponding to the defect processing strategy of the defect is adjusted according to the merged processing result.

In some embodiments, the defect The processing time corresponding to the defect handling strategy is:

in, When the defect handling strategy is rework, the node that needs to be reworked is located according to the defect type and the required rework time is determined according to the reworked node. When the defect handling strategy is scrapping, the current node is determined according to the defect type as a node before or after wafer cutting, thereby determining the number of reprocessing and the time required for reprocessing. is a function that determines the defect handling strategy based on the defect type and defect severity of the current defect. They respectively indicate that the defect handling strategy is rework or scrapping. In this embodiment, the function for determining the rework time is a function model determined by statistically analyzing the mean of the rework time required for rework of each defect type; the defect handling strategy of this embodiment determines the required reprocessing time by determining the processing node according to the defect type. For example, if it is a defect at a wafer processing node, the number of wafers that need to be reprocessed after scrapping is one wafer. If it is a packaging process, the number of wafers that need to be reprocessed is determined based on the number of semiconductor devices corresponding to one wafer, thereby calculating the reprocessing time.

In some embodiments, the defect The defect cause location time and the corresponding equipment repair and maintenance time are:

in, The defect cause location time is determined based on the time when the technicians feedback the defect cause information. is a function that determines the equipment repair time based on the defect cause of the current defect, The additional equipment maintenance time after resetting the equipment regular maintenance cycle according to the cause of the current defect, The defect cause identified for the current defect, Defect cause sets for the first, second and third defect cause types respectively.

In some embodiments, the system of the embodiment of the present invention includes: a defect impact severity level module, which is used to set a number of thresholds with different numerical values and a number of defect impact severity levels, and associate the thresholds with the defect impact severity levels one by one, so that the larger the threshold, the higher the defect impact severity level associated with the threshold; calculate the time difference between the preset production time and the estimated production time, and determine the defect impact severity level based on the comparison result of the time difference with the threshold.

In some embodiments, the image data acquisition module 200 is used to perform various tests based on the first image data to obtain defect type detection data; wherein the various tests include: wafer defect detection, wire bond defect detection, package defect detection, etc.

In some embodiments, the system of the embodiment of the present invention also includes: a second image data detection module, which is used to obtain the finished product image data of the processed semiconductor device and the second image data of the semiconductor device after being transported to the place of use, and perform a second defect detection on the finished product image data and the second image data respectively to obtain second defect detection data; determine the defect type and defect rate caused during the transportation process according to the second defect detection data. The defect detection method involved in this embodiment includes a deep learning method (CNN-based image recognition method), which has high performance in image classification and target detection, and can greatly improve the recognition rate of irregular defects. This embodiment can adjust the production and processing order plan by determining the defect rate caused by the transportation process, and by comparing the main defect types caused by the transportation process with the main defect types caused by the production and manufacturing process, determine the links where defects found during use may occur.

In particular, in some embodiments, the system of the embodiment of the present invention further includes: a third image data detection module, which is used to obtain the first semi-processed image data after each production and processing link is completed and the second semi-processed image data transported to the next production and processing node, and obtain the defect results before and after transportation through defect detection, and obtain the defect rate and defect type of the node transportation process. This embodiment provides a basis for locating the defect occurrence link for subsequent processing detection by counting the main defect types caused by each transportation link.

In some embodiments, the defect impact assessment module 500 is used to calculate the second production time estimate according to the defect rate of the transportation process. :

in, is the defect rate, An estimate of the current production time for the preset.

In some embodiments, the second image data detection module is used to perform binarization image processing and morphological processing on the image data and obtain detection features of the image, and determine the target defects and defect locations in each image data based on the detection features and preset feature detection standards.

Corresponding to the above-mentioned embodiments, the present invention also provides a method embodiment. As for the method embodiment, since it basically corresponds to the system embodiment, the relevant parts can be referred to the partial description of the system embodiment.

2 , an embodiment of the present invention provides a semiconductor device image data detection method based on artificial intelligence, comprising the following steps:

S100. Analyze historical inspection data of image inspection items in the semiconductor device manufacturing process to obtain each defect type and the defect cause corresponding to the defect type and the probability of occurrence of the defect cause; classify each defect cause according to whether equipment maintenance is required and whether the equipment maintenance time needs to be modified to obtain a first defect cause type, a second defect cause type and a third defect cause type.

S200, obtaining first image data of a semiconductor device manufacturing process, and performing various tests based on the first image data to obtain defect type detection data; the defect type detection data includes: defect ID, defect type, defect quantity, defect severity, defect cause set and defect cause occurrence probability.

S300, determining a defect handling strategy according to defect type detection data; the defect handling strategy is to set a handling strategy according to the defect type and defect severity, including rework and scrapping.

S400, sorting the defect causes in the defect cause set according to the defect cause occurrence probability, and outputting a defect cause sequence; and determining the defect cause corresponding to each defect ID in response to the defect cause information fed back by the technician.

S500 , evaluating the impact of defects on the semiconductor device manufacturing process based on defect type detection data, defect handling strategy, defect cause and defect cause type, and outputting evaluation results.

Although specific embodiments are described herein, those of ordinary skill in the art will recognize that many other modifications or alternative embodiments are also within the scope of the present disclosure. For example, any of the functions and/or processing capabilities described in conjunction with a particular device or component may be performed by any other device or component. In addition, although various exemplary implementations and architectures have been described according to embodiments of the present disclosure, those of ordinary skill in the art will recognize that many other modifications to the exemplary implementations and architectures described herein are also within the scope of the present disclosure.

It will be appreciated by those skilled in the art that all or some of the steps and systems in the methods disclosed above may be implemented as software, firmware, hardware, and appropriate combinations thereof. Some or all physical components may be implemented as software executed by a processor, such as a central processing unit, a digital signal processor, or a microprocessor, or as hardware, or as an integrated circuit, such as an application-specific integrated circuit. Such software may be distributed on a computer-readable medium, which may include a computer storage medium (or non-transitory medium) and a communication medium (or transient medium). As known to those skilled in the art, the term computer storage medium includes volatile and non-volatile, removable and non-removable media implemented in any method or technology for storing information, such as computer-readable instructions, data structures, program modules, or other data. Computer storage media include, but are not limited to, RAM, ROM, EEPROM, flash memory or other memory technology, CD-ROM, digital versatile disk (DVD) or other optical disk storage, magnetic cassettes, magnetic tapes, disk storage or other magnetic storage devices, or any other medium that can be used to store desired information and can be accessed by a computer. Furthermore, it is well known to those skilled in the art that communication media typically embodies computer readable instructions, data structures, program modules, or other data in a modulated data signal such as a carrier wave or other transport mechanism, and may include any information delivery media.

The above is a specific description of the preferred implementation of the present application, but the present application is not limited to the above-mentioned implementation mode. Technical personnel familiar with the field can also make various equivalent deformations or substitutions without violating the spirit of the present application. These equivalent deformations or substitutions are all included in the scope defined by the claims of the present application.

Claims (10)

1. An artificial intelligence based semiconductor device image data detection system, comprising:

the historical data analysis module is used for analyzing historical detection data of an image detection item in the manufacturing process of the semiconductor device to obtain each defect type and defect reasons corresponding to the defect types and occurrence probability of the defect reasons; classifying the defect reasons according to whether equipment maintenance is required or not and whether equipment maintenance time is required to be modified or not to obtain a first defect reason type, a second defect reason type and a third defect reason type;

The image data acquisition module is used for acquiring first image data in the manufacturing process of the semiconductor device, and performing various detections based on the first image data to obtain defect type detection data; the defect type detection data includes: defect id, defect type, number of defects, defect severity, defect cause set, and probability of occurrence of defect cause;

a defect processing strategy module for determining a defect processing strategy according to the defect type detection data; the defect treatment strategy is set according to the defect type and defect severity, and comprises reworking and scrapping;

The defect cause positioning module is used for sequencing the defect causes in the defect cause set according to the occurrence probability of the defect causes and outputting a defect cause sequence; responding to defect reason information fed back by technicians, and determining defect reasons corresponding to each defect id;

And the defect influence evaluation module is used for evaluating the influence of the defects on the production and manufacturing process of the semiconductor device according to the defect type detection data, the defect processing strategy, the defect reasons and the defect reason types and outputting an evaluation result.

2. The artificial intelligence based semiconductor device image data inspection system of claim 1, wherein the defect impact assessment module comprises:

The production information acquisition unit is used for acquiring the production efficiency of each node in semiconductor production, the reworking time corresponding to the production node corresponding to each image detection item, the equipment maintenance time corresponding to each defect reason and the equipment periodic maintenance period;

a production time prediction unit for obtaining a production time predicted value according to the defect type detection data and the defect processing strategy ：

；

Wherein, For the number of wafers to be used,For the ith node wafer processing efficiency, A is the total number of wafer processing nodes,For the number of semiconductor devices to be used,For the processing efficiency of the jth node after the wafer dicing, B is the total number of the processing nodes after the wafer dicing, M is the number of times of detecting the defect of the first image data,Defects that are the kth defectIs set according to the processing time corresponding to the defect processing strategy,Is the firstDefect type of individual defectDefect cause locating time and corresponding equipment repair and maintenance time,For the time required for each periodic maintenance of the equipment,Periodically maintaining a period initial value for the equipment;

And, carry on the merger processing to the defect data that the defect happens in the identical measured body, carry on the adjustment to the processing time that the defect processing tactics of the defect corresponds according to the merger processing result.

3. The artificial intelligence based semiconductor device image data inspection system of claim 2, wherein the defect is a defectThe processing time corresponding to the defect processing strategy is as follows:

；

Wherein, For the defect processing strategy to be a function of locating the nodes to be reworked according to the defect type and determining the required reworking time according to the reworked nodes,Determining the current node as a node before or after wafer dicing according to the defect type when the defect processing strategy is scrapped, thereby determining the number of reworks and the time required for reworking,To determine the function of the defect handling strategy based on the defect type and defect severity of the current defect,Indicating that the defect handling policy is reworked or scrapped, respectively.

4. The artificial intelligence based semiconductor device image data inspection system of claim 2, wherein the defect is a defectThe defect cause positioning time and the corresponding equipment maintenance time are as follows:

；

Wherein, For the defect cause locating time determined according to the point of time when the technician feeds back the defect cause information,To determine a function of equipment repair time based on the defect cause of the current defect,To reset the increased equipment maintenance time after the equipment periodic maintenance period according to the defect cause of the current defect,The defect cause determined for the current defect,And defect cause sets of the first, second and third defect cause types, respectively.

5. The artificial intelligence based semiconductor device image data inspection system of claim 2, wherein the system comprises:

The defect influence severity level module is used for setting a plurality of thresholds with unequal values and a plurality of defect influence severity levels, and associating the thresholds with the defect influence severity levels one by one, so that the larger the threshold is, the higher the defect influence severity level associated with the threshold is; and calculating a time difference between preset production time and the production time estimated value, and determining the severity level of defect influence according to a comparison result of the time difference and the threshold value.

6. The artificial intelligence based semiconductor device image data detection system of claim 1, wherein the image data acquisition module is configured to perform each detection based on the first image data to obtain defect type detection data; wherein the detecting comprises: wafer defect detection, bonding wire defect detection and packaging defect detection.

7. The artificial intelligence based semiconductor device image data inspection system of claim 1, further comprising:

The second image data detection module is used for acquiring finished product image data of the processed semiconductor device and second image data of the semiconductor device transported to a using place, and respectively carrying out second defect detection on the finished product image data and the second image data to obtain second defect detection data; determining the defect type and defect rate caused in the transportation process according to the second defect detection data;

The third image data detection module is used for acquiring first half machining image data after each production and machining link is finished and second half machining image data transported to a next production and machining node, and obtaining defect results before and after transportation through defect detection to obtain defect rate and defect type in the transportation process of the node.

8. The artificial intelligence based semiconductor device image data inspection system of claim 7, wherein the defect impact assessment module is configured to calculate a second production time estimate based on a defect rate of the transportation process：

；

Wherein, In order for the defect rate to be a high level,And a preset current production time pre-estimated value is obtained.

9. The image data detection system of claim 7, wherein the second image data detection module is configured to perform binarization image processing and morphological processing on the image data and obtain detection features of the image, and determine the target defect and the defect position in each image data based on the detection features and a preset feature detection standard.

10. The image data detection method of the semiconductor device based on artificial intelligence is characterized by comprising the following steps of:

S100, analyzing historical detection data of an image detection item in a semiconductor device manufacturing process to obtain each defect type and defect reasons corresponding to the defect types and occurrence probability of the defect reasons; classifying the defect reasons according to whether equipment maintenance is required or not and whether equipment maintenance time is required to be modified or not to obtain a first defect reason type, a second defect reason type and a third defect reason type;

S200, acquiring first image data in the manufacturing process of the semiconductor device, and performing various detections based on the first image data to obtain defect type detection data; the defect type detection data includes: defect id, defect type, number of defects, defect severity, defect cause set, and probability of occurrence of defect cause;

S300, determining a defect processing strategy according to the defect type detection data; the defect treatment strategy is set according to the defect type and the defect severity, and comprises reworking and scrapping;

S400, sorting the defect reasons in the defect reason set according to the occurrence probability of the defect reasons, and outputting a defect reason sequence; responding to defect reason information fed back by technicians, and determining defect reasons corresponding to each defect id;

S500, evaluating influences of defects on the semiconductor device manufacturing process according to the defect type detection data, the defect processing strategy, the defect reasons and the defect reason types, and outputting evaluation results.