Image defect classification system and method
By using the FEM conversion module and the automatic defect classification system, wafer defect images under different PWQ conditions are merged into the FEM table. Combined with lithography process parameters, the images are stored and automatically classified, which solves the problem of low efficiency in defect image classification in semiconductor manufacturing and achieves the effects of rapid classification and reduced workload for engineers.
Patent Information
- Application Number
- CN202511066271.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-07-30
- Publication Date
- 2025-11-25
AI Technical Summary
In existing technologies, the classification efficiency of defect images in semiconductor manufacturing is low, the workload for engineers is heavy, and it is difficult to meet the rapid classification requirements of massive defect images under advanced process nodes.
The FEM conversion module is used to merge wafer defect images under different PWQ conditions into the FEM table, and the encoding and automatic defect classification module is used to achieve rapid classification. The defect images are obtained by measuring or scanning with EP5 and SEM equipment, and stored and classified in combination with the exposure energy and focal length of the photolithography process.
It enables rapid classification of defective images, reduces the workload of engineers, improves classification efficiency, and simplifies the judgment process.
Smart Images
Figure CN121010806A_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present application relates to the field of semiconductor integrated circuit manufacturing, in particular to an image defect classification system. The present application also relates to an image defect classification method. BACKGROUND
[0002] In semiconductor manufacturing, as the process node is continuously reduced, the number of defect images required for process window verification (PWQ) is increasing, so it is necessary to classify a large number of defect images. The existing image defect classification is realized by manual classification of engineers, that is, manual classification is performed according to the wafer map of the defects measured under the PWQ condition. Since the number of defect images to be classified is huge under the advanced process node, the classification efficiency is low and the burden on engineers is heavy. SUMMARY
[0003] The technical problem to be solved by the present application is to provide an image defect classification system that can realize fast classification of PWQ defect images and reduce the classification workload. To this end, the present application also provides an image defect classification method.
[0004] To solve the above technical problems, the image defect classification system provided by the present application comprises:
[0005] The FEM conversion module combines the defect pictures of the wafers under different PWQ conditions into a FEM table, and each grid point of the FEM table stores zero or more defect pictures.
[0006] Further improvement is that the FEM conversion module further comprises a function of encoding the FEM table, and one FEM table corresponds to one encoding.
[0007] Further improvement is that it further comprises a FEM table selection module for selecting the corresponding FEM table according to the encoding.
[0008] Further improvement is that each defect picture is obtained by measuring the wafer by an EP5 machine or scanning the wafer by a SEM machine.
[0009] Further improvement is that each PWQ condition includes the exposure energy and focal length of the lithography process, and each defect picture on the wafer under each PWQ condition is stored in the corresponding grid point of the corresponding FEM table according to the exposure energy and the interval.
[0010] Further improvement is that it further comprises:
[0011] The module for automatically classifying each of the defect images of each of the grid points is used to select the corresponding defect image in the grid point, and to automatically compare and classify the selected defect image and a reference image.
[0012] To solve the above technical problems, the image defect classification method provided by the present application comprises:
[0013] The FEM conversion is performed to combine each defect image of the wafer under different PWQ conditions into an FEM table, and each grid point of the FEM table stores zero or more defect images.
[0014] Further improvement is that the method further comprises:
[0015] The FEM table is encoded, and one FEM table corresponds to one encoding.
[0016] Further improvement is that the method further comprises:
[0017] After the FEM conversion is completed, the corresponding FEM table is selected according to the encoding.
[0018] Further improvement is that each defect image is obtained by measuring the wafer by an EP5 machine or scanning the wafer by a SEM machine.
[0019] Further improvement is that each PWQ condition includes exposure energy and focal length of a photolithography process, and each defect image on the wafer under each PWQ condition is stored in the corresponding grid point of the corresponding FEM table according to the exposure energy and the focal length.
[0020] Further improvement is that after the FEM table is selected, the method further comprises:
[0021] Each defect image in the grid point is selected, and the selected defect image and a reference image are automatically compared to realize automatic defect classification.
[0022] The FEM conversion module provided by the present application combines each defect image of the wafer under various PWQ conditions into a corresponding FEM table, the FEM table is arranged according to exposure energy and focal length of a photolithography process, and each grid point stores multiple corresponding defect images. Compared with multiple wafer images, the present application can realize storage of each defect image in multiple wafer images through one FEM table, and finally realize rapid classification of PWQ defect images and reduce the burden of classification work.
[0023] The FEM table can be numbered, and each grid point of the FEM table can be automatically classified. The numbering and automatic classification can further accelerate the judgment speed of the engineers on the classification of the defect image. BRIEF DESCRIPTION OF DRAWINGS
[0024] The application will be further described in detail below in combination with the drawings and specific embodiments:
[0025] Figure 1 FIG. 1 is a structural schematic diagram of a FEM conversion module of an image defect classification system according to an embodiment of the application;
[0026] Figure 2 FIG. 2 is a schematic diagram of a PWQ wafer map and a FEM table in FEM conversion of the FEM conversion module of the image defect classification system according to the embodiment of the application;
[0027] Figure 3 FIG. 3 is an operation interface diagram of the image defect classification system according to the embodiment of the application. DETAILED DESCRIPTION
[0028] As shown in FIG. 1, which is a structural schematic diagram of a FEM conversion module of an image defect classification system according to an embodiment of the application; the image defect classification system according to the embodiment of the application comprises: Figure 1 The FEM conversion module 101 combines each defect image 105 of a wafer under different PWQ conditions into a FEM table 103. Each grid point 104 of the FEM table 103 stores zero or more defect images 105.
[0029] In the FEM conversion module 101, each PWQ wafer map 102 represents each defect image 105 of a wafer under each PWQ condition. Figure 1 In the embodiment of the application, each defect image 105 is obtained by measuring the wafer by an EP5 machine or scanning the wafer by a SEM machine.
[0030] Each PWQ condition includes exposure energy and focal length of a photolithography process. Each defect image 105 on the wafer under each PWQ condition is stored in the corresponding grid point 104 of the corresponding FEM table 103 according to the exposure energy and the focal length.
[0031] In order to clearly understand the FEM conversion module 101 according to the embodiment of the application, further description will be made below in combination with FIG. 2, which is a schematic diagram of a PWQ wafer map and a FEM table in FEM conversion of the FEM conversion module of the image defect classification system according to the embodiment of the application.
[0032] Figure 2 Figure 2 Figure 2 In the diagram, the FEM table is individually labeled 103a, and the two PWQ wafer images are labeled 102a and 102b, respectively. It can be seen that PWQ wafer image 102a has 27 defect images, but it is not easy to classify each defect image from PWQ wafer images 102a and 102b. In this embodiment, the FEM conversion module 101 can merge the defect images in PWQ wafer images 102a and 102b into the same FEM table 103a. Each grid point in FEM table 103a is individually labeled 104a. Each grid point 104a is divided according to the exposure energy and focal length in the photolithography process, and during merging, each defect image is also placed into the corresponding grid point 104a in a one-to-one correspondence between exposure energy and focal length. Figure 2 The diagram shows that 54 of the grid points 104a each contain one of the defect images 105a, while the other grid points 104a do not contain one. If more PWQ wafer images need to be merged, the number of defect images 105a placed in some of the grid points 104a may be greater than one. Figure 2 The image also shows a time photograph of one of the defective images 105a.
[0033] In this embodiment of the invention, the FEM conversion module 101 further includes the function of encoding the FEM table 103, with one FEM table 103 corresponding to one encoding. Thus, the image defect classification system of this embodiment can obtain multiple FEM tables 103 as needed.
[0034] In this embodiment of the invention, it further includes: an FEM table selection module, used to select the corresponding FEM table 103 according to the encoding.
[0035] like Figure 3 The figure shown is an operation interface diagram of the image defect classification system according to an embodiment of the present invention. Figure 3 The image shows the operation module 201 corresponding to the FEM table selection module. The operation module 201 has various selectable codes. Figure 3 The document further lists the encoding when the value "300" is selected.
[0036] When the code with a value of 300 is selected, the FEM table corresponding to the marker 103b will be displayed.
[0037] In this embodiment of the invention, it further includes:
[0038] The module for automatically classifying each of the defect pictures 105 of each of the grids 104 is used to select the corresponding defect picture 105 in the grid 104, and automatically compare and classify the selected defect picture 105 and the reference picture.
[0039] As shown in the FEM table 103b, the selected grid 104b displays 279 defects, and one defect corresponds to one defect picture 105. Figure 3 Figure 3 In the defect list 202, one defect can be selected, such as the defect corresponding to the mark 204, so that the defect picture corresponding to the mark 105b can be called up. In the embodiment of the present application, the defect picture 105b and the reference picture 203 can be automatically compared and classified, and the automatic comparison and classification can be realized by the existing automatic defect classification (ADC) module.
[0040] The FEM conversion module 101 provided in the embodiment of the present application combines the wafer pictures under various PWQ conditions, i.e., each of the defect pictures 105 of the wafer, into the corresponding FEM table 103, and the FEM table 103 is arranged according to the exposure energy and focus of the photolithography process. Each grid 104 stores a plurality of corresponding defect pictures 105. Compared with the scattered wafer pictures, the embodiment of the present application can realize the storage of each of the defect pictures 105 in the plurality of wafer pictures through one FEM table 103, and finally realize the rapid classification of the PWQ defect pictures, and reduce the classification work burden.
[0041] The embodiment of the present application can also number the FEM table 103, and automatically classify the defect pictures 105 of each grid 104 of the FEM table 103. The numbering and automatic classification can further accelerate the judgment speed of the engineers on the classification of the defect pictures.
[0042] By comparing the PWQ wafer pictures 102a and 102b and the FEM table 103a, it can be seen that in the FEM table 103a, the interface presents the split of Energy and Focus, and the defect code that can be modified and the ADC function are added to accelerate the judgment speed of the engineers.
[0043] The embodiment of the present application can quickly arrange the defects of each condition of the PWQ on the interface, and can classify the defect code (Defect Code). The system can add the automatic defect classification (ADC) function, and can be compatible with the EP5 and Review SEM devices.
[0044] The image defect classification method of the embodiment of the present application comprises:
[0045] The FEM conversion is performed to combine the defect pictures 105 of the wafers under different PWQ conditions into the FEM table 103, and each cell 104 of the FEM table 103 stores one or more of the defect pictures 105.
[0046] In the method of the embodiment of the present application, each of the defect pictures 105 is obtained by measuring the wafer by an EP5 machine or scanning the wafer by a SEM machine.
[0047] Each of the PWQ conditions includes exposure energy and focal length of a photolithography process, and each of the defect pictures 105 on the wafer under each of the PWQ conditions is stored in the corresponding cell 104 of the corresponding FEM table 103 according to the exposure energy and the focal length.
[0048] Each of the FEM tables 103 is encoded, and one FEM table 103 corresponds to one encoding.
[0049] After the FEM conversion is completed, the corresponding FEM table 103 is selected according to the encoding.
[0050] After the FEM table 103 is selected, the method further comprises:
[0051] Each of the defect pictures 105 in the cell 104 is selected, and the selected defect picture 105 is compared with a reference picture to realize automatic defect classification.
[0052] The present application is described in detail above through specific embodiments, but these do not constitute a limitation on the present application. Those skilled in the art can also make many modifications and improvements without departing from the principles of the present application, and these should also be considered as falling within the protection scope of the present application.
Claims
1. An image defect classification system, characterized by, The method comprises: A FEM conversion module, which combines each defect picture of a wafer under different PWQ conditions into a FEM table, each grid point of the FEM table storing zero or more defect pictures.
2. The image defect classification system of claim 1, wherein: The FEM conversion module further comprises a function of encoding the FEM table, one FEM table corresponding to one encoding.
3. The image defect classification system of claim 2, wherein, The method further comprises: A FEM table selection module, which selects a corresponding FEM table according to the encoding.
4. The image defect classification system of claim 2, wherein: Each defect picture is obtained by measuring the wafer by an EP5 machine or scanning the wafer by a SEM machine.
5. The image defect classification system of claim 2, wherein: Each PWQ condition includes exposure energy and focal length of a photolithography process, and each defect picture on the wafer under each PWQ condition is stored in a corresponding grid point of a corresponding FEM table according to the exposure energy and the focal length.
6. The image defect classification system of claim 3, wherein, The method further comprises: A module for automatically classifying each defect picture of each grid point, which selects a corresponding defect picture in the grid point and automatically compares and classifies the selected defect picture with a reference picture.
7. An image defect classification method characterized by, The method comprises: Performing FEM conversion to combine each defect picture of a wafer under different PWQ conditions into a FEM table, each grid point of the FEM table storing zero or more defect pictures.
8. The image defect classification method of claim 7, wherein, The method further comprises: Encoding the FEM table, one FEM table corresponding to one encoding.
9. The image defect classification method of claim 8, wherein, The method further comprises: After completing the FEM conversion, selecting a corresponding FEM table according to the encoding.
10. The image defect classification method of claim 8, wherein: Each defect picture is obtained by measuring the wafer by an EP5 machine or scanning the wafer by a SEM machine.
11. The image defect classification method of claim 8, wherein: Each PWQ condition includes exposure energy and focal length of a photolithography process, and each defect picture on the wafer under each PWQ condition is stored in a corresponding grid point of a corresponding FEM table according to the exposure energy and the focal length.
12. The image defect classification method of claim 9, wherein, After selecting the FEM table, the method further comprises: Selecting each defect picture in the grid point, and automatically comparing the selected defect picture with a reference picture to achieve automatic defect classification.