Wafer inspection apparatus, method, system and readable storage medium

Abstract

The application discloses a wafer detection equipment, a detection method, a system and a readable storage medium, wherein the detection equipment comprises: a material support for carrying an adsorbed wafer to be detected; an optical machine detection device located above the material support and used for collecting a wafer image of the wafer to be detected, the optical machine detection device comprising a camera, a light source, a laser, a cylindrical mirror and an objective lens; a host computer in communication connection with the optical machine detection device and used for processing the wafer image collected by the optical machine detection device; a motion shaft fixedly connected with the material support and used for moving the material support; and a bearing platform in sliding connection with the motion shaft and used for limiting the motion of the motion shaft on the bearing platform. The wafer detection equipment is used for real-time focusing, the problem of imaging blur caused by movement to different positions is solved, the image collection quality is improved, and the detection accuracy is improved. An AI model trained by collecting defects is used for replacing manual work, and the problem of manual rejudgment and classification of defects is solved.

Description

Technical Field

[0001] This invention relates to the fields of image processing and equipment manufacturing technology, and more specifically, to a wafer inspection device, inspection method, system, and readable storage medium. Background Technology

[0002] In actual industrial production, wafer front-side processes such as etching and development are affected by various factors, resulting in various defects such as dirt, foreign matter, etching residue, and scratches. These defects seriously affect chip quality and cause significant losses. Therefore, defect detection and defect classification play an important role in the production process.

[0003] Currently, wafer defect detection equipment on the market is based on optical systems to acquire images, analyze the images to detect defects, and then manually review and classify the defects after detection. This has the disadvantages of decreased image quality leading to blurred images that affect the accuracy of detection, and the need for manual review greatly increases the workload of personnel. Summary of the Invention

[0004] The purpose of this invention is to provide a wafer inspection device, inspection method, system, and readable storage medium. By using the wafer inspection device to focus in real time, the problem of blurred imaging when moving to different positions is solved, the quality of the acquired images is improved, and the inspection accuracy is enhanced. By collecting defect data to train an AI model, machine operation is used to replace manual operation, solving the problem of manual defect re-judgment and classification.

[0005] The first aspect of the present invention provides a wafer inspection device, the wafer inspection device comprising:

[0006] Material stage, used to hold and adsorb the wafer to be tested;

[0007] An optomechanical inspection device, located above the material stage, is used to acquire wafer images of the wafer under test. The optomechanical inspection device includes a camera, a light source, a laser, a telescope, and an objective lens. The camera is used to capture images, the light source is used to provide illumination, the laser is used to assist the camera in image focusing, the telescope is used to support the light emitted by the light source and the laser to illuminate the surface of the wafer under test, and the objective lens is used to adjust the magnification of the camera.

[0008] The host computer is communicatively connected to the optomechanical inspection device and is used to process the wafer images acquired by the optomechanical inspection device.

[0009] A motion shaft is fixedly connected to the material platform and is used to drive the material platform to move.

[0010] A support platform, wherein the motion axis is slidably connected to the support platform, and is used to limit the motion axis to move on the support platform.

[0011] In this solution, the supporting platform is a marble platform.

[0012] In this solution, the marble platform is provided with a motion path for the motion axis to move.

[0013] In this scheme, the wafer to be tested is adsorbed onto the material carrier by vacuum adsorption.

[0014] A second aspect of the present invention also provides a wafer inspection method, the method comprising the following steps:

[0015] Acquire the location and stitched image of each die on the wafer under test;

[0016] A complete image of the wafer is obtained by stitching together each grain on the wafer under test using a feature point matching template.

[0017] Based on the differential template, 2N standard grain images are identified on the wafer under test, and an adaptive differential template suitable for the wafer under test is constructed based on the 2N standard grain images.

[0018] Based on the adaptive differential template, information on all defect regions on the wafer under test is obtained;

[0019] Based on the defect area information and the preset control standards, the defect areas that do not meet the control standards are filtered out.

[0020] The filtered defect area information is input into a trained AI classification model for classification to obtain wafer inspection results.

[0021] This plan also includes:

[0022] A positioning template for positioning grains is constructed based on shape matching.

[0023] Establish the feature point matching template, and then stitch together the grain image based on the overlapping feature points;

[0024] Scan the wafer under test and construct a differential template based on the images of N standard grains;

[0025] The card control standard is constructed based on the stitched image to establish a defect type mapping, wherein the defect type includes defect area, defect size, defect brightness and darkness and defect shape;

[0026] An AI classification model is obtained by constructing a classification model based on the mapped defect data and training it.

[0027] This solution also includes acquiring defect data to train the AI ​​classification model.

[0028] A third aspect of the present invention provides a wafer inspection system, including a memory and a processor, wherein the memory includes a wafer inspection method program, and the wafer inspection method program, when executed by the processor, performs the following steps:

[0029] Each grain on the wafer under test is located based on a feature point matching template.

[0030] Based on the differential template, 2N standard grain images are identified on the wafer under test, and an adaptive differential template suitable for the wafer under test is constructed based on the 2N standard grain images.

[0031] Based on the adaptive differential template, information on all defect regions on the wafer under test is obtained;

[0032] Based on the defect area information and the preset control standards, the defect areas that do not meet the control standards are filtered out.

[0033] The filtered defect area information is input into a trained AI classification model for classification to obtain wafer inspection results.

[0034] This solution also includes the following steps:

[0035] Establish the feature point matching template, and then stitch the image together based on the overlapping feature points;

[0036] Scan the wafer under test and construct a differential template based on the images of N standard grains;

[0037] The card control standard is constructed based on the stitched image to establish a defect type mapping, wherein the defect type includes defect area, defect size, defect brightness and darkness and defect shape;

[0038] An AI classification model is obtained by constructing a classification model based on the mapped defect data and training it.

[0039] A fourth aspect of the present invention provides a computer-readable storage medium comprising a wafer inspection method program for a machine, wherein when executed by a processor, the wafer inspection method program implements the steps of a wafer inspection method as described in any of the preceding claims.

[0040] This invention discloses a wafer inspection device, inspection method, system, and readable storage medium. By using the wafer inspection device to focus in real time, it solves the problem of blurred imaging when moving to different positions, improves the quality of acquired images, and enhances inspection accuracy. By collecting defect data to train an AI model, it uses machine operation to replace manual operation, solving the problem of manual defect re-judgment and classification. Attached Figure Description

[0041] Figure 1 A flowchart of a wafer inspection method according to the present invention is shown;

[0042] Figure 2 A flowchart of a wafer inspection method according to the present invention is shown;

[0043] Figure 3 A block diagram of a wafer inspection system according to the present invention is shown;

[0044] Figure 4 A schematic diagram of the structure of a wafer inspection device according to the present invention is shown;

[0045] Figures 5a-5b A schematic diagram of a blurred image of a wafer inspection is shown;

[0046] Figure 6 A schematic diagram of the motion path of a wafer inspection device according to the present invention is shown;

[0047] Figures 7a-7b A clear, schematic image of wafer inspection is shown. Detailed Implementation

[0048] To better understand the above-mentioned objectives, features, and advantages of the present invention, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be noted that, unless otherwise specified, the embodiments and features described in these embodiments can be combined with each other.

[0049] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and therefore the scope of protection of the invention is not limited to the specific embodiments disclosed below.

[0050] Figure 1 A flowchart of a wafer inspection method according to this application is shown.

[0051] like Figure 1 As shown, this application discloses a wafer inspection method, including the following steps:

[0052] S102, acquire the location of each die on the wafer under test and the stitched image;

[0053] S104, based on feature point matching templates, stitch together each grain on the wafer under test to obtain a complete image of the wafer;

[0054] S106, Identify 2N standard grain images on the wafer under test based on the differential template, and construct an adaptive differential template suitable for the wafer under test based on the 2N standard grain images;

[0055] S108, Based on the adaptive differential template, obtain information on all defect regions on the wafer under test;

[0056] S110, Based on the defect area information and the preset control standard, filter out defect areas that do not meet the control standard.

[0057] S112, The filtered defect area information is input into the trained AI classification model for classification to obtain the wafer inspection result.

[0058] It should be noted that, in this embodiment, the detection phase process includes: "location—assembly—construction of adaptive differential template—defect detection—defect control—defect analysis." Before the detection phase begins, a preparatory phase is also included, specifically, as follows: Figure 2 As shown, the steps in the preparatory stage include:

[0059] S202, a positioning template for positioning grains is constructed based on shape matching;

[0060] S204, Establish the feature point matching template, and then stitch together the grain image based on the overlapping feature points;

[0061] S206, Scan the wafer to be tested and construct a differential template based on the images of N standard grains;

[0062] S208, The card control standard is constructed based on the stitched image to establish a defect type mapping, wherein the defect type includes defect area, defect size, defect brightness and darkness and defect shape;

[0063] S210, an AI classification model is obtained by constructing a classification model based on the mapped defect data and training it.

[0064] Furthermore, in this embodiment, the preparatory stage process includes: "constructing a positioning template—constructing a stitching template—constructing a differential template—constructing a control standard—constructing an AI classification model." Specifically, during wafer manufacturing, the patterns on the die are generally relatively stable, but the stability of colors and other aspects is poor. Therefore, shape matching is more stable than feature-based matching. In this application, shape matching is used to create a positioning template to locate the die. For some large-sized power devices, the camera's field of view cannot simultaneously meet the requirements of detection accuracy and field of view. Therefore, it is necessary to stitch the captured images. This application uses a feature point matching-based stitching method to stitch the images together to form complete images of the die. Therefore, it is necessary to first establish a feature point matching template. Then, based on the overlapping feature points, the image is stitched together, the entire wafer is scanned, and the same wafer is stitched together into an image. N (10-100) standard grain images on the wafer are randomly obtained. Based on the N standard grain images, an average difference template is constructed. On a stitched standard grain, a defect type mapping is established according to the defect area, defect size, defect brightness, and defect shape to construct the control standard. Then, a classification model is constructed based on the mapped defect data and trained to obtain an AI classification model. In the traditional difference template detection method, the defect category cannot be subdivided. In order to meet the needs of process analysis, in addition to the traditional detection method, this application also proposes to collect defect data to train the AI ​​classification model. The specific training process is not described in this application.

[0065] Furthermore, during inspection, after locating each wafer to be tested, the feature point matching template established in the preparatory stage is used to stitch together the grain images to obtain a complete image of all grains on the wafer. Based on the pre-constructed differential template, images of 2N suspected standard grains are selected on the wafer. An adaptive differential template suitable for this wafer is dynamically constructed using the selected 2N images. Based on the adaptive differential template constructed in the preparatory stage, all grains on this wafer are compared to obtain all defect area information. Based on the control standards constructed in the preparatory stage, the defect areas detected in the inspection stage are mapped to the corresponding defect types, and defect areas that do not meet the corresponding control standards are filtered out. Based on the AI ​​classification model constructed in the preparatory stage, the acquired defect images are classified to provide process analysis to obtain wafer inspection results.

[0066] Figure 3 A block diagram of a wafer inspection system according to the present invention is shown.

[0067] like Figure 3 As shown, this invention discloses a wafer inspection system, including a memory and a processor. The memory includes a wafer inspection method program, which, when executed by the processor, performs the following steps:

[0068] Each grain on the wafer under test is located based on a feature point matching template.

[0069] Based on the differential template, 2N standard grain images are identified on the wafer under test, and an adaptive differential template suitable for the wafer under test is constructed based on the 2N standard grain images.

[0070] Based on the adaptive differential template, information on all defect regions on the wafer under test is obtained;

[0071] Based on the defect area information and the preset control standards, the defect areas that do not meet the control standards are filtered out.

[0072] The filtered defect area information is input into a trained AI classification model for classification to obtain wafer inspection results.

[0073] It should be noted that, in this embodiment, the detection phase process includes: "location—assembly—construction of adaptive differential template—defect detection—defect control—defect analysis." Before the detection phase begins, a preparatory phase is also included, specifically, as follows: Figure 2 As shown, the steps in the preparatory stage include:

[0074] A positioning template for positioning grains is constructed based on shape matching.

[0075] Establish the feature point matching template, and then stitch together the grain image based on the overlapping feature points;

[0076] Scan the wafer under test and construct a differential template based on the images of N standard grains;

[0077] The card control standard is constructed based on the stitched image to establish a defect type mapping, wherein the defect type includes defect area, defect size, defect brightness and darkness and defect shape;

[0078] An AI classification model is obtained by constructing a classification model based on the mapped defect data and training it.

[0079] Furthermore, in this embodiment, the preparatory stage process includes: "constructing a positioning template—constructing a stitching template—constructing a differential template—constructing a control standard—constructing an AI classification model." Specifically, during wafer manufacturing, the patterns on the die are generally relatively stable, but the stability of colors and other aspects is poor. Therefore, shape matching is more stable than feature-based matching. In this application, shape matching is used to create a positioning template to locate the die. For some large-sized power devices, the camera's field of view cannot simultaneously meet the requirements of detection accuracy and field of view. Therefore, it is necessary to stitch the captured images. This application uses a feature point matching-based stitching method to stitch the images together to form complete images of the die. Therefore, it is necessary to first establish a feature point matching template. Then, based on the overlapping feature points, the image is stitched together, the entire wafer is scanned, and the same wafer is stitched together into an image. N (10-100) standard grain images on the wafer are randomly obtained. Based on the N standard grain images, an average difference template is constructed. On a stitched standard grain, a defect type mapping is established according to the defect area, defect size, defect brightness, and defect shape to construct the control standard. Then, a classification model is constructed based on the mapped defect data and trained to obtain an AI classification model. In the traditional difference template detection method, the defect category cannot be subdivided. In order to meet the needs of process analysis, in addition to the traditional detection method, this application also proposes to collect defect data to train the AI ​​classification model. The specific training process is not described in this application.

[0080] Furthermore, during inspection, after locating each wafer to be tested, the feature point matching template established in the preparatory stage is used to stitch together the grain images to obtain a complete image of all grains on the wafer. Based on the pre-constructed differential template, images of 2N suspected standard grains are selected on the wafer. An adaptive differential template suitable for this wafer is dynamically constructed using the selected 2N images. Based on the adaptive differential template constructed in the preparatory stage, all grains on this wafer are compared to obtain all defect area information. Based on the control standards constructed in the preparatory stage, the defect areas detected in the inspection stage are mapped to the corresponding defect types, and defect areas that do not meet the corresponding control standards are filtered out. Based on the AI ​​classification model constructed in the preparatory stage, the acquired defect images are classified to provide process analysis to obtain wafer inspection results.

[0081] Figure 4 A schematic diagram of the structure of a wafer inspection device according to the present invention is shown.

[0082] Currently available wafer inspection equipment does not acquire images based on the optimal focal point. Due to the unevenness of the stage and the inherent high-order differences in the wafer pattern, the vertical axis of the wafer will be inconsistent at different positions. The vertical axis is the Z-axis of the inspection coordinate system, meaning the optimal focal point is the same at different positions. When the vertical axis height fluctuation range exceeds the depth of field of the optical system, the quality of the acquired image degrades, the image becomes blurry, and this affects the accuracy of the inspection. Figures 5a-5b As shown, where, Figure 5b This is a schematic diagram of a blurry image, illustrating the blurriness problem in images acquired using existing technologies.

[0083] like Figure 4 As shown, the present invention discloses a wafer inspection device 40, characterized in that the wafer inspection device comprises:

[0084] Material stage 41 is used to hold and adsorb the wafer to be tested;

[0085] The optomechanical inspection device 42, located above the material stage 41, is used to acquire wafer images of the wafer under test. The optomechanical inspection device 42 includes a camera 421, a light source 422, a laser 423, a telescope 424, and an objective lens 425. The camera 421 is used to capture images, the light source 422 is used to provide illumination, the laser 423 is used to assist the camera 421 in image focusing, the telescope 424 is used to support the light emitted by the light source 422 and the laser 423 to illuminate the surface of the wafer under test, and the objective lens 425 is used to adjust the magnification of the camera 421.

[0086] The host 43 is communicatively connected to the optomechanical inspection device 42 and is used to process the wafer image acquired by the optomechanical inspection device 42;

[0087] The motion shaft 44 is fixedly connected to the material platform 41 and is used to drive the material platform 41 to move.

[0088] The support platform 45 is slidably connected to the motion shaft 44, which is used to limit the movement of the motion shaft 44 on the support platform 45.

[0089] It should be noted that the supporting platform 45 is a marble platform, which can provide shock absorption during motion photography, such as... Figure 6 As shown, the marble platform has a motion path for the motion axis 44 to move. The wafer to be tested is adsorbed onto the material carrier by vacuum adsorption. In specific operation, the wafer to be tested is placed on the material carrier 41 and adsorbed by vacuum. Figure 6The motion path indicated by the wide-body arrow sets the motion axis 44, causing it to move to the detection position. The laser focuses, and the Z-axis is adjusted in real time. The optomechanical detection device 42 acquires an image of the wafer. The camera 421 captures the image, the light source 422 provides illumination, and the laser 423 assists the camera 421 in focusing. The telescope 424 supports the light emitted by the light source 422 and the laser 423 onto the surface of the wafer under test, and also supports the light emitted by the laser 423 onto the wafer surface and reflected back to the optical path system. The objective lens 425 adjusts the magnification of the camera 421, supporting different magnifications according to different detection requirements, such as... Figures 7a-7b As shown, they are respectively Figure 6 Clear images were obtained from points pm and PN.

[0090] A fourth aspect of the present invention provides a computer-readable storage medium comprising a wafer inspection method program, wherein when executed by a processor, the wafer inspection method program implements the steps of a wafer inspection method as described in any of the preceding claims.

[0091] This invention discloses a wafer inspection device, inspection method, system, and readable storage medium. By using the wafer inspection device to focus in real time, it solves the problem of blurred imaging when moving to different positions, improves the quality of acquired images, and enhances inspection accuracy. By collecting defect data to train an AI model, it uses machine operation to replace manual operation, solving the problem of manual defect re-judgment and classification.

[0092] In the several embodiments provided in this application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are merely illustrative. For example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods, such as: multiple units or components can be combined, or integrated into another system, or some features can be ignored or not executed. In addition, the coupling, direct coupling, or communication connection between the various components shown or discussed can be through some interfaces, and the indirect coupling or communication connection between devices or units can be electrical, mechanical, or other forms.

[0093] The units described above as separate components may or may not be physically separate. The components shown as units may or may not be physical units. They may be located in one place or distributed across multiple network units. Some or all of the units may be selected to achieve the purpose of this embodiment according to actual needs.

[0094] In addition, in the various embodiments of the present invention, each functional unit can be integrated into one processing unit, or each unit can be a separate unit, or two or more units can be integrated into one unit; the integrated unit can be implemented in hardware or in the form of hardware plus software functional units.

[0095] Those skilled in the art will understand that all or part of the steps of the above method embodiments can be implemented by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When the program is executed, it performs the steps of the above method embodiments. The aforementioned storage medium includes various media capable of storing program code, such as mobile storage devices, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0096] Alternatively, if the integrated units of this invention are implemented as software functional modules and sold or used as independent products, they can also be stored in a computer-readable storage medium. Based on this understanding, the technical solutions of the embodiments of this invention, or the parts that contribute to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as mobile storage devices, ROM, RAM, magnetic disks, or optical disks.

Claims

1. A wafer inspection method, characterized in that, The method includes the following steps: Acquire the location and stitched image of each die on the wafer under test; A complete image of the wafer is obtained by stitching together each grain on the wafer under test using a feature point matching template. Based on the differential template, 2N standard grain images are identified on the wafer under test, and an adaptive differential template suitable for the wafer under test is constructed based on the 2N standard grain images. Based on the adaptive differential template, information on all defect regions on the wafer under test is obtained; Based on the defect area information and the preset control standards, the defect areas that do not meet the control standards are filtered out. The filtered defect area information is input into a trained AI classification model for classification to obtain wafer inspection results.

2. The wafer inspection method according to claim 1, characterized in that, Also includes: A positioning template for positioning grains is constructed based on shape matching. Establish the feature point matching template, and then stitch together the grain image based on the overlapping feature points; Scan the wafer under test and construct a differential template based on the images of N standard grains; The card control standard is constructed based on the stitched image to establish a defect type mapping, wherein the defect type includes defect area, defect size, defect brightness and darkness and defect shape; An AI classification model is obtained by constructing a classification model based on the mapped defect data and training it.

3. The wafer inspection method according to claim 2, characterized in that, It also includes acquiring defect data to train the AI ​​classification model.

4. A wafer inspection apparatus using the wafer inspection method of claim 1, characterized in that, The wafer inspection equipment includes: Material stage, used to hold and adsorb the wafer to be tested; An optomechanical inspection device, located above the material stage, is used to acquire wafer images of the wafer under test. The optomechanical inspection device includes a camera, a light source, a laser, a telescope, and an objective lens. The camera is used to capture images, the light source is used to provide illumination, the laser is used to assist the camera in image focusing, the telescope is used to support the light emitted by the light source and the laser to illuminate the surface of the wafer under test, and the objective lens is used to adjust the magnification of the camera. The host computer is communicatively connected to the optomechanical inspection device and is used to process the wafer images acquired by the optomechanical inspection device. A motion shaft is fixedly connected to the material platform and is used to drive the material platform to move. A support platform, wherein the motion axis is slidably connected to the support platform, and is used to limit the motion axis to move on the support platform.

5. A wafer inspection device according to claim 4, characterized in that, The supporting platform is a marble platform.

6. The wafer inspection equipment according to claim 5, characterized in that, The marble platform is provided with a motion path for the motion axis to move.

7. A wafer inspection device according to claim 4, characterized in that, The wafer to be tested is adsorbed onto the material carrier by vacuum adsorption.

8. A wafer inspection system, characterized in that, The system includes a memory and a processor. The memory contains a wafer inspection method program, which, when executed by the processor, performs the following steps: Each grain on the wafer under test is located based on a feature point matching template. Based on the differential template, 2N standard grain images are identified on the wafer under test, and an adaptive differential template suitable for the wafer under test is constructed based on the 2N standard grain images. Based on the adaptive differential template, information on all defect regions on the wafer under test is obtained; Based on the defect area information and the preset control standards, the defect areas that do not meet the control standards are filtered out. The filtered defect area information is input into a trained AI classification model for classification to obtain wafer inspection results.

9. A wafer inspection system according to claim 8, characterized in that, It also includes implementing the following steps: Establish the feature point matching template, and then stitch the image together based on the overlapping feature points; Scan the wafer under test and construct a differential template based on the images of N standard grains; The card control standard is constructed based on the stitched image to establish a defect type mapping, wherein the defect type includes defect area, defect size, defect brightness and darkness and defect shape; An AI classification model is obtained by constructing a classification model based on the mapped defect data and training it.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes a wafer inspection method program, which, when executed by a processor, implements the steps of a wafer inspection method as described in any one of claims 1 to 3.

Images