A method for detecting finished integrated circuit products

Abstract

The present invention relates to the technical field of integrated circuit detection, and specifically to a method for detecting finished integrated circuits; obtaining layered bare die images at various points; classifying the layered bare die images, wherein the layered bare die image categories include good product images and defective images; storing all defective images of each process subsequence on a production line to which the finished integrated circuit to be detected belongs into a finished product detection database, and screening out feature position data to be verified; dynamically adjusting the number of scanning electron microscopes and the shooting angles of each scanning electron microscope according to the feature position data to be verified in the finished product detection database; performing a similarity comparison analysis between the image scanning results of the scanning electron microscope and the defective images corresponding to the feature position data to be verified in the finished product detection database, and the monitoring center arranges repair measures for unqualified finished integrated circuits according to the analysis results; and significantly improving the detection efficiency on the basis of improving the coverage and depth of defect detection.

Description

Technical Field

[0001] The present invention relates to the technical field of integrated circuit detection, in particular to a method for detecting finished integrated circuit products. Background Art

[0002] With the development of integrated circuit technology, the manufacturing process of integrated circuits is becoming more and more advanced, more and more components are integrated on integrated circuits, and the circuits are becoming more and more complex. In addition, people's requirements for the reliability of integrated circuits are also getting higher and higher. In order to ensure that integrated circuits can still meet the reliability requirements while becoming more and more complex, it is necessary to test the finished integrated circuits.

[0003] In the comparative document CN114594370A "Integrated Circuit Failure Detection Method and Device", the defective area in the integrated circuit is detected by changing the charging effect of the integrated circuit to be detected when the detection probe contacts the integrated circuit to be detected, and the electrical parameters of the integrated circuit to be detected are obtained. This method requires all metal wires in the integrated circuit to be detected. Although it meets the requirements of detection accuracy, the detection efficiency is slow.

[0004] In the comparative document CN116596905A “A method for detecting surface defects of integrated circuit chips”, the pixels in the optimal preset window are processed to make the pixels in the defective area more obvious, and an enhanced image is obtained, so that the details in the image are clearer. Then, the enhanced image is segmented to accurately obtain the defective area in the image. Although the requirements of detection efficiency and detection accuracy are met, this method only detects the surface of the integrated circuit chip, but cannot detect the defects of the various layered chips inside the finished integrated circuit, and cannot meet the requirements of the detection depth of the finished integrated circuit.

[0005] With the improvement of nano-scale manufacturing technology of integrated circuits, there are more than a dozen circuit layers connecting various devices in integrated circuits, and they are packaged by stacking bare chips, welding wire connections or vertical interconnections. When detecting defective areas inside integrated circuits, traditional integrated circuit finished product testing methods need to grind multi-layer stacked integrated circuits to the layer that needs to be tested to determine whether the integrated circuit is reliable. When there are many layers stacked on the layer that needs to be tested, this test method requires grinding off each layer above the layer to be tested, resulting in low detection efficiency. How to improve detection efficiency while ensuring the coverage and depth of defect detection is a problem that we urgently need to solve. Now a method for detecting integrated circuit finished products is proposed. Summary of the invention

[0006] In order to solve the above technical problems, the object of the present invention is to provide a method for detecting an integrated circuit product, comprising the following steps:

[0007] Step S1: obtaining process information of the production line to which the integrated circuit product to be inspected belongs, setting image monitoring points according to the process information, and obtaining layered bare die images at each point;

[0008] Step S2: performing binary comparison between the layered die images on each process subsequence and the images of good layered die that have been strictly inspected, and determining the classification of the layered die images according to the binary comparison results, wherein the layered die image categories include good images and defective images;

[0009] Step S3: storing all defect images of each process subsequence on the production line to which the integrated circuit finished product to be inspected belongs into the finished product inspection database, and extracting features of defect areas of the defect images of each process subsequence in the finished product inspection database to obtain feature position data of each defect image;

[0010] Step S4: obtaining the real-time occurrence frequency of all feature position data in the finished product inspection database, and filtering the feature position data to be verified from all feature position data according to the real-time occurrence frequency;

[0011] Step S5: setting a multi-angle SEM integrated circuit finished product inspection unit, and dynamically adjusting the number of scanning electron microscopes in the multi-angle SEM integrated circuit finished product inspection unit and the shooting angle of each scanning electron microscope according to the feature position data to be verified in the finished product inspection database;

[0012] Step S6: Set up a monitoring center to perform similarity comparison analysis between the image scanning results of the scanning electron microscope and the defect images corresponding to the feature position data to be verified in the finished product inspection database, and send the analysis results to the monitoring center. The monitoring center arranges repair measures for the unqualified integrated circuit products according to the analysis results.

[0013] Furthermore, the process of obtaining process information of the production line to which the integrated circuit finished product to be inspected belongs, setting image monitoring points according to the process information, and obtaining the layered bare die image at each point includes:

[0014] Acquire the process flow characteristics of the production line to which the integrated circuit finished product to be tested belongs, extract process information according to the process flow characteristics, and split the production line to which the integrated circuit finished product to be tested belongs according to the process information into a plurality of process sub-sequences;

[0015] Image monitoring points are arranged on each process subsequence, and the image monitoring points are used to obtain layered bare die images on the process subsequence to which the image monitoring points belong.

[0016] Furthermore, the process information includes: mask making, wafer preparation, mask transfer, and the use of chemical reactions and physical deposition and other process technologies to form different layers of circuits on the wafer, including conductive layers, insulating layers, and connecting layers, etc. The process includes multiple repeated lithography, etching, and deposition steps, and wafer cutting and packaging.

[0017] Furthermore, the process of performing binary comparison on the layered bare die images on each process subsequence and the layered bare die images of good products that have been strictly inspected, and determining the classification of the layered bare die images according to the binary comparison results includes:

[0018] Obtaining the images of good layered bare dies that have been strictly inspected in each process subsequence, and performing binary comparison between the layered bare dies images on each process subsequence and the images of good layered bare dies corresponding to each process subsequence;

[0019] Subtracting the pixel values ​​at corresponding positions of the layered bare die image and the corresponding good layered bare die image to obtain a difference value, converting the difference value into a pixel value of a binary image to generate a binary image;

[0020] Setting a preset pixel threshold and two pixel point traversal pointers, wherein the two pixel point traversal pointers start traversing from the first pixel point and the last pixel point of the binary image area at the same time;

[0021] Marking a pixel point region in the binary image whose pixel value is greater than a preset pixel threshold as a defect region, and marking a layered bare die image that generates the binary image as a defect image;

[0022] The pixel point area in the binary image whose pixel value is less than or equal to the preset pixel threshold is marked as a normal area, and the layered bare die image generated by the binary image is marked as a good image.

[0023] Furthermore, all defect images of each process subsequence on the production line to which the integrated circuit finished product to be inspected belongs are stored in the finished product inspection database, and feature extraction is performed on the defect area of ​​the defect image of each process subsequence in the finished product inspection database to obtain feature position data of each defect image, including:

[0024] The pixel point in the central part of the binary image generated by the defect image is set as the central pixel; a two-dimensional coordinate system is established with the central pixel as the origin, all pixel points in the binary image are mapped to the two-dimensional coordinate system, a pixel point coordinate set located in the defect area in the two-dimensional coordinate system is obtained, and feature position data is generated according to the pixel point coordinate set.

[0025] Furthermore, the process of obtaining the real-time occurrence frequency of all feature position data in the finished product inspection database and selecting the feature position data to be verified from all feature position data according to the real-time occurrence frequency includes:

[0026] Acquire feature position data of all defect images in a finished product inspection database, perform statistical analysis on the feature position data in the finished product inspection database, filter the feature position data according to the number of times each feature position data is generated, set a preset number threshold, and mark feature position data whose number of generation is greater than the preset number threshold as first feature position data;

[0027] Obtain a generation timestamp corresponding to each time the first feature position data is generated in the finished product inspection database, and obtain a historical occurrence frequency based on the generation timestamp and the number of times the first feature position data is generated in the finished product inspection database, and obtain an interval between the generation timestamp corresponding to the last time the first feature position data was generated in the finished product inspection database and the current moment, and obtain an updated real-time occurrence frequency based on the interval and the historical occurrence frequency;

[0028] Setting a real-time occurrence frequency threshold, comparing the real-time occurrence frequency of the first feature position data with the real-time occurrence frequency threshold, and re-marking the first feature position data whose real-time occurrence frequency is less than the real-time occurrence frequency threshold as feature position data;

[0029] The first feature position data whose real-time occurrence frequency is greater than or equal to the real-time occurrence frequency threshold is marked as feature position data to be verified.

[0030] Furthermore, the calculation formula for the real-time occurrence frequency is:

[0031]

[0032] Among them, C v represents the historical frequency, C K It represents the real-time occurrence frequency, T represents the generation timestamp corresponding to each time the first feature position data is generated in the finished product inspection database; N represents the number of times the feature position data is generated in the finished product inspection database; t represents the interval between the generation timestamp corresponding to the last time the first feature position data was generated in the finished product inspection database and the current moment.

[0033] Furthermore, a multi-angle SEM integrated circuit finished product inspection unit is set, and a process of dynamically adjusting the number of scanning electron microscopes in the multi-angle SEM integrated circuit finished product inspection unit and the shooting angle of each scanning electron microscope according to the feature position data to be verified in the finished product inspection database includes:

[0034] The SEM integrated circuit finished product inspection unit is composed of a plurality of scanning electron microscopes, which are used to scan each layered bare chip inside the integrated circuit finished product through an electron beam, and generate an image scanning result through secondary electrons and reflected electron signals generated by the interaction between the electron beam and the layered bare chip;

[0035] The SEM integrated circuit finished product inspection unit obtains the number K of feature position data to be verified in the finished product inspection database, and determines the number K of scanning electron microscopes to be inspected according to the number K of feature position data to be verified, and determines the shooting angles corresponding to the K scanning electron microscopes according to the K feature position data to be verified, and associates the image scanning result of the scanning electron microscope with the feature position data to be verified for determining the shooting angle of the scanning electron microscope;

[0036] The shooting angle indicates the number of layered bare die layers inspected inside the integrated circuit finished product and the specific locations on the layered bare die inspected.

[0037] Furthermore, a monitoring center is set up to compare and analyze the image scanning results of the scanning electron microscope with the defect image corresponding to the feature position data to be verified in the finished product inspection database, and the process of sending the analysis results to the monitoring center includes:

[0038] Obtain K image scanning results of K scanning electron microscopes, and convert the defect images corresponding to the K feature position data to be verified in the finished product inspection database and the K image scanning results into binary code formats respectively;

[0039] Generate K image scanning result hash codes and K defect image hash codes according to the K image scanning results and K defect images converted into binary code format, set a similarity threshold, and compare the K image scanning result hash codes with their associated defect image hash codes for similarity;

[0040] If the similarity between the hash code of the image scanning result and the hash code of the defect image is greater than the similarity threshold, the image scanning result is marked as unqualified, and a detection unqualified signal is generated according to the image scanning result and sent to the monitoring center;

[0041] If the similarity between the hash code of the image scanning result and the hash code of the defect image is less than or equal to the similarity threshold, the image scanning result is marked as qualified, and a qualified detection signal is generated according to the image scanning result and sent to the monitoring center.

[0042] Furthermore, the process of the monitoring center arranging repair measures for the unqualified integrated circuit finished products according to the analysis results includes:

[0043] After receiving the unqualified detection signal, the monitoring center obtains an unqualified image scanning result according to the unqualified detection signal, determines the specific location of the defective area inside the integrated circuit product according to the unqualified image scanning result, and arranges repair measures according to the specific location of the defective area inside the integrated circuit product.

[0044] Compared with the prior art, the present invention has the following beneficial effects:

[0045] 1. The present invention installs image monitoring points on the production line of the finished integrated circuit to be inspected, thereby obtaining the probability of hidden defects that may exist in each layered bare chip inside the finished integrated circuit to be inspected during the production process, and saves the historical defect images that appeared in the production line of the finished integrated circuit to be inspected into the finished product inspection database. The SEM integrated circuit finished product inspection unit dynamically adjusts the number of scanning electron microscopes and the shooting angles of the scanning electron microscopes in the unit according to the finished product inspection database, and uses multiple scanning electron microscopes to shoot and inspect the finished integrated circuit at different angles, thereby more comprehensively covering and analyzing each area on each layer of the circuit board inside the integrated circuit, improving the coverage and depth of defect detection, and increasing the accuracy and reliability of defect detection.

[0046] 2. When detecting defective areas inside an integrated circuit, the traditional integrated circuit finished product testing method needs to grind the multi-layered integrated circuit to the layer that needs to be tested in order to determine whether the integrated circuit is reliable. When there are many layers stacked on the layer that needs to be tested, this testing method needs to grind off each layer above the layer to be tested, resulting in low testing efficiency. The present invention determines in advance the specific layered positions where the probability of defective areas inside the integrated circuit is relatively high based on the defect images in the finished product detection database, and directly detects the specific layered positions where the probability of defective areas inside the integrated circuit is relatively high through the SEM integrated circuit finished product detection unit. There is no need to grind the integrated circuit to the layer to be tested, and the internal defects can be detected. On the basis of improving the coverage and depth of defect detection, the detection efficiency is significantly improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] Figure 1 This is a schematic diagram of a method for detecting finished integrated circuits according to an embodiment of the present application. DETAILED DESCRIPTION

[0048] The following is a clear and complete description of the technical solutions in the embodiments of the present application in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present application.

[0049] like Figure 1 As shown, a method for testing an integrated circuit finished product comprises the following steps:

[0050] Step S1: obtaining process information of the production line to which the integrated circuit product to be inspected belongs, setting image monitoring points according to the process information, and obtaining layered bare die images at each point;

[0051] Step S2: performing binary comparison between the layered die images on each process subsequence and the images of good layered die that have been strictly inspected, and determining the classification of the layered die images according to the binary comparison results, wherein the layered die image categories include good images and defective images;

[0052] Step S3: storing all defect images of each process subsequence on the production line to which the integrated circuit finished product to be inspected belongs into the finished product inspection database, and extracting features of defect areas of the defect images of each process subsequence in the finished product inspection database to obtain feature position data of each defect image;

[0053] Step S4: obtaining the real-time occurrence frequency of all feature position data in the finished product inspection database, and filtering the feature position data to be verified from all feature position data according to the real-time occurrence frequency;

[0054] Step S5: setting a multi-angle SEM integrated circuit finished product inspection unit, and dynamically adjusting the number of scanning electron microscopes in the multi-angle SEM integrated circuit finished product inspection unit and the shooting angle of each scanning electron microscope according to the feature position data to be verified in the finished product inspection database;

[0055] Step S6: Set up a monitoring center to perform similarity comparison analysis between the image scanning results of the scanning electron microscope and the defect images corresponding to the feature position data to be verified in the finished product inspection database, and send the analysis results to the monitoring center. The monitoring center arranges repair measures for the unqualified integrated circuit products according to the analysis results.

[0056] It should be further explained that, in the specific implementation process, the process information of the production line to which the integrated circuit finished product to be inspected belongs is obtained, image monitoring points are set according to the process information, and the process of obtaining the layered bare die image at each point includes:

[0057] Acquire the process flow characteristics of the production line to which the integrated circuit finished product to be tested belongs, extract process information according to the process flow characteristics, and split the production line to which the integrated circuit finished product to be tested belongs according to the process information into a plurality of process sub-sequences;

[0058] Image monitoring points are arranged on each process subsequence, and the image monitoring points are used to obtain layered bare die images on the process subsequence to which the image monitoring points belong.

[0059] It should be further explained that, in the specific implementation process, the process information includes: mask production, wafer preparation, mask transfer, and the use of chemical reactions and physical deposition processes to form different layers of circuits on the wafer, including conductive layers, insulating layers, and connecting layers, etc. The process includes repeated lithography, etching, and deposition steps, and wafer cutting and packaging.

[0060] It should be further explained that, in the specific implementation process, the layered bare die images on each process subsequence are binary compared with the layered bare die images of strictly inspected good products, and the process of determining the classification of the layered bare die images according to the binary comparison results includes:

[0061] Obtaining the images of good layered bare dies that have been strictly inspected in each process subsequence, and performing binary comparison between the layered bare dies images on each process subsequence and the images of good layered bare dies corresponding to each process subsequence;

[0062] Subtracting the pixel values ​​at corresponding positions of the layered bare die image and the corresponding good layered bare die image to obtain a difference value, converting the difference value into a pixel value of a binary image to generate a binary image;

[0063] Setting a preset pixel threshold and two pixel point traversal pointers, wherein the two pixel point traversal pointers start traversing from the first pixel point and the last pixel point of the binary image area at the same time;

[0064] Marking a pixel point region in the binary image whose pixel value is greater than a preset pixel threshold as a defect region, and marking a layered bare die image that generates the binary image as a defect image;

[0065] The pixel point area in the binary image whose pixel value is less than or equal to the preset pixel threshold is marked as a normal area, and the layered bare die image generated by the binary image is marked as a good image.

[0066] It should be further explained that, in the specific implementation process, all defect images of each process subsequence on the production line to which the integrated circuit finished product to be inspected belongs are stored in the finished product inspection database, and the defect area of ​​the defect image of each process subsequence in the finished product inspection database is extracted, and the process of obtaining the feature position data of each defect image includes:

[0067] The pixel point in the central part of the binary image generated by the defect image is set as the central pixel; a two-dimensional coordinate system is established with the central pixel as the origin, all pixel points in the binary image are mapped to the two-dimensional coordinate system, a pixel point coordinate set located in the defect area in the two-dimensional coordinate system is obtained, and feature position data is generated according to the pixel point coordinate set.

[0068] It should be further explained that, in the specific implementation process, the process of obtaining the real-time occurrence frequency of all feature position data in the finished product inspection database and selecting the feature position data to be verified from all feature position data according to the real-time occurrence frequency includes:

[0069] Acquire feature position data of all defect images in a finished product inspection database, perform statistical analysis on the feature position data in the finished product inspection database, filter the feature position data according to the number of times each feature position data is generated, set a preset number threshold, and mark feature position data whose number of generation is greater than the preset number threshold as first feature position data;

[0070] Obtain a generation timestamp corresponding to each time the first feature position data is generated in the finished product inspection database, and obtain a historical occurrence frequency based on the generation timestamp and the number of times the first feature position data is generated in the finished product inspection database, and obtain an interval between the generation timestamp corresponding to the last time the first feature position data was generated in the finished product inspection database and the current moment, and obtain an updated real-time occurrence frequency based on the interval and the historical occurrence frequency;

[0071] Setting a real-time occurrence frequency threshold, comparing the real-time occurrence frequency of the first feature position data with the real-time occurrence frequency threshold, and re-marking the first feature position data whose real-time occurrence frequency is less than the real-time occurrence frequency threshold as feature position data;

[0072] The first feature position data whose real-time occurrence frequency is greater than or equal to the real-time occurrence frequency threshold is marked as feature position data to be verified.

[0073] It should be further explained that, in the specific implementation process, the calculation formula of the real-time occurrence frequency is:

[0074]

[0075]

[0076] Among them, C v represents the historical frequency, C K It represents the real-time occurrence frequency, T represents the generation timestamp corresponding to each time the first feature position data is generated in the finished product inspection database; N represents the number of times the feature position data is generated in the finished product inspection database; t represents the interval between the generation timestamp corresponding to the last time the first feature position data was generated in the finished product inspection database and the current moment.

[0077] It should be further explained that, in the specific implementation process, a multi-angle SEM integrated circuit finished product inspection unit is set up, and the process of dynamically adjusting the number of scanning electron microscopes in the multi-angle SEM integrated circuit finished product inspection unit and the shooting angle of each scanning electron microscope according to the feature position data to be verified in the finished product inspection database includes:

[0078] The SEM integrated circuit finished product inspection unit is composed of a plurality of scanning electron microscopes, which are used to scan each layered bare chip inside the integrated circuit finished product through an electron beam, and generate an image scanning result through secondary electrons and reflected electron signals generated by the interaction between the electron beam and the layered bare chip;

[0079] The SEM integrated circuit finished product inspection unit obtains the number K of feature position data to be verified in the finished product inspection database, and determines the number K of scanning electron microscopes to be inspected according to the number K of feature position data to be verified, and determines the shooting angles corresponding to the K scanning electron microscopes according to the K feature position data to be verified, and associates the image scanning result of the scanning electron microscope with the feature position data to be verified for determining the shooting angle of the scanning electron microscope;

[0080] The shooting angle indicates the number of layered bare die layers inspected inside the integrated circuit finished product and the specific locations on the layered bare die inspected.

[0081] It should be further explained that, in the specific implementation process, a monitoring center is set up to compare and analyze the image scanning results of the scanning electron microscope with the defect image corresponding to the feature position data to be verified in the finished product inspection database for similarity, and the process of sending the analysis results to the monitoring center includes:

[0082] Obtain K image scanning results of K scanning electron microscopes, and convert the defect images corresponding to the K feature position data to be verified in the finished product inspection database and the K image scanning results into binary code formats respectively;

[0083] Generate K image scanning result hash codes and K defect image hash codes according to the K image scanning results and K defect images converted into binary code format, set a similarity threshold, and compare the K image scanning result hash codes with their associated defect image hash codes for similarity;

[0084] If the similarity between the hash code of the image scanning result and the hash code of the defect image is greater than the similarity threshold, the image scanning result is marked as unqualified, and a detection unqualified signal is generated according to the image scanning result and sent to the monitoring center;

[0085] If the similarity between the hash code of the image scanning result and the hash code of the defect image is less than or equal to the similarity threshold, the image scanning result is marked as qualified, and a qualified detection signal is generated according to the image scanning result and sent to the monitoring center.

[0086] It should be further explained that, in the specific implementation process, the process in which the monitoring center arranges repair measures for the unqualified integrated circuit finished products according to the analysis results includes:

[0087] After receiving the unqualified detection signal, the monitoring center obtains an unqualified image scanning result according to the unqualified detection signal, determines the specific location of the defective area inside the integrated circuit product according to the unqualified image scanning result, and arranges repair measures according to the specific location of the defective area inside the integrated circuit product.

[0088] The above embodiments are only used to illustrate the technical method of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical method of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical method of the present invention.

Claims

1. A method for testing integrated circuit products, characterized in that: The following steps are involved: Step S1: obtaining process information of the production line to which the integrated circuit product to be inspected belongs, setting image monitoring points according to the process information, and obtaining layered bare die images at each point; Step S2: performing binary comparison between the layered die images on each process subsequence and the images of good layered die that have been strictly inspected, and determining the classification of the layered die images according to the binary comparison results, wherein the layered die image categories include good images and defective images; Step S3: storing all defect images of each process subsequence on the production line to which the integrated circuit finished product to be inspected belongs into the finished product inspection database, and extracting features of defect areas of the defect images of each process subsequence in the finished product inspection database to obtain feature position data of each defect image; Step S4: obtaining the real-time occurrence frequency of all feature position data in the finished product inspection database, and filtering the feature position data to be verified from all feature position data according to the real-time occurrence frequency; Step S5: dynamically adjusting the number of scanning electron microscopes in the multi-angle SEM integrated circuit finished product inspection unit and the shooting angle of each scanning electron microscope according to the feature position data to be verified in the finished product inspection database; Step S6: Setting up a monitoring center, performing a similarity comparison analysis between the image scanning results of the scanning electron microscope and the defect images corresponding to the feature position data to be verified in the finished product inspection database, and sending the analysis results to the monitoring center; the monitoring center arranges repair measures for the unqualified integrated circuit finished products according to the analysis results.

2. The method for detecting an integrated circuit product according to claim 1, characterized in that: The process of obtaining process information of the production line to which the integrated circuit finished product to be inspected belongs, setting image monitoring points according to the process information, and obtaining the layered bare die image at each point includes: Acquire the process flow characteristics of the production line to which the integrated circuit finished product to be tested belongs, extract process information according to the process flow characteristics, and split the production line to which the integrated circuit finished product to be tested belongs according to the process information into a plurality of process sub-sequences; Image monitoring points are arranged on each process subsequence, and the image monitoring points are used to obtain layered bare die images on the process subsequence to which the image monitoring points belong.

3. The method for testing an integrated circuit product according to claim 1, characterized in that: The process of performing binary comparison between the layered die images on each process subsequence and the layered die images of good products that have been strictly inspected, and determining the classification of the layered die images according to the binary comparison results includes: Obtaining the images of good layered bare dies that have been strictly inspected in each process subsequence, and performing binary comparison between the layered bare dies images on each process subsequence and the images of good layered bare dies corresponding to each process subsequence; Subtracting the pixel values ​​at corresponding positions of the layered bare die image and the corresponding good layered bare die image to obtain a difference value, converting the difference value into a pixel value of a binary image to generate a binary image; Setting a preset pixel threshold and two pixel point traversal pointers, wherein the two pixel point traversal pointers start traversing from the first pixel point and the last pixel point of the binary image area at the same time; Marking a pixel point region in the binary image whose pixel value is greater than a preset pixel threshold as a defect region, and marking a layered bare die image that generates the binary image as a defect image; The pixel point area in the binary image whose pixel value is less than or equal to the preset pixel threshold is marked as a normal area, and the layered bare die image generated by the binary image is marked as a good image.

4. The method for testing an integrated circuit product according to claim 1, characterized in that: The process of storing all defect images of each process subsequence on the production line of the integrated circuit finished product to be inspected in the finished product inspection database, and extracting features of the defect areas of the defect images of each process subsequence in the finished product inspection database to obtain feature position data of each defect image includes: The pixel point in the central part of the binary image generated by the defect image is set as the central pixel; a two-dimensional coordinate system is established with the central pixel as the origin, all pixel points in the binary image are mapped to the two-dimensional coordinate system, a pixel point coordinate set located in the defect area in the two-dimensional coordinate system is obtained, and feature position data is generated according to the pixel point coordinate set.

5. The method for testing an integrated circuit product according to claim 1, characterized in that: The process of obtaining the real-time occurrence frequency of all feature position data in the finished product inspection database and selecting the feature position data to be verified from all feature position data according to the real-time occurrence frequency includes: Acquire feature position data of all defect images in a finished product inspection database, perform statistical analysis on the feature position data in the finished product inspection database, filter the feature position data according to the number of times each feature position data is generated, set a preset number threshold, and mark feature position data whose number of generation is greater than the preset number threshold as first feature position data; Obtain a generation timestamp corresponding to each time the first feature position data is generated in the finished product inspection database, and obtain a historical occurrence frequency based on the generation timestamp and the number of times the first feature position data is generated in the finished product inspection database, and obtain an interval between the generation timestamp corresponding to the last time the first feature position data was generated in the finished product inspection database and the current moment, and obtain an updated real-time occurrence frequency based on the interval and the historical occurrence frequency; Setting a real-time occurrence frequency threshold, comparing the real-time occurrence frequency of the first feature position data with the real-time occurrence frequency threshold, and re-marking the first feature position data whose real-time occurrence frequency is less than the real-time occurrence frequency threshold as feature position data; The first feature position data whose real-time occurrence frequency is greater than or equal to the real-time occurrence frequency threshold is marked as feature position data to be verified.

6. The method for testing an integrated circuit product according to claim 1, characterized in that: The process of setting a multi-angle SEM integrated circuit finished product inspection unit and dynamically adjusting the number of scanning electron microscopes in the multi-angle SEM integrated circuit finished product inspection unit and the shooting angle of each scanning electron microscope according to the feature position data to be verified in the finished product inspection database includes: The SEM integrated circuit finished product inspection unit is composed of a plurality of scanning electron microscopes, which are used to scan each layered bare chip inside the integrated circuit finished product through an electron beam, and generate an image scanning result through secondary electrons and reflected electron signals generated by the interaction between the electron beam and the layered bare chip; The SEM integrated circuit finished product inspection unit obtains the number K of feature position data to be verified in the finished product inspection database, and determines the number K of scanning electron microscopes to be inspected according to the number K of feature position data to be verified, and determines the shooting angles corresponding to the K scanning electron microscopes according to the K feature position data to be verified, and associates the image scanning result of the scanning electron microscope with the feature position data to be verified for determining the shooting angle of the scanning electron microscope; The shooting angle indicates the number of layered bare die layers inspected inside the finished integrated circuit and the specific locations on the layered bare die inspected.

7. The method for testing an integrated circuit product according to claim 1, characterized in that: The process of setting up a monitoring center, performing similarity comparison analysis between the image scanning results of the scanning electron microscope and the defect image corresponding to the feature position data to be verified in the finished product inspection database, and sending the analysis results to the monitoring center includes: Obtain K image scanning results of K scanning electron microscopes, and convert the defect images corresponding to the K feature position data to be verified in the finished product inspection database and the K image scanning results into binary code formats respectively; Generate K image scanning result hash codes and K defect image hash codes according to the K image scanning results and K defect images converted into binary code format, set a similarity threshold, and compare the K image scanning result hash codes with their associated defect image hash codes for similarity; If the similarity between the hash code of the image scanning result and the hash code of the defect image is greater than the similarity threshold, the image scanning result is marked as unqualified, and a detection unqualified signal is generated according to the image scanning result and sent to the monitoring center; If the similarity between the hash code of the image scanning result and the hash code of the defect image is less than or equal to the similarity threshold, the image scanning result is marked as qualified, and a qualified detection signal is generated according to the image scanning result and sent to the monitoring center.

8. The method for testing an integrated circuit product according to claim 1, characterized in that: The process of the monitoring center arranging repair measures for unqualified integrated circuit finished products according to the analysis results includes: After receiving the unqualified detection signal, the monitoring center obtains an unqualified image scanning result according to the unqualified detection signal, determines the specific location of the defective area inside the integrated circuit product according to the unqualified image scanning result, and arranges repair measures according to the specific location of the defective area inside the integrated circuit product.

Images