CN115132599A - Defect detection method - Google Patents

Defect detection method Download PDF

Info

Publication number
CN115132599A
CN115132599A CN202110328165.1A CN202110328165A CN115132599A CN 115132599 A CN115132599 A CN 115132599A CN 202110328165 A CN202110328165 A CN 202110328165A CN 115132599 A CN115132599 A CN 115132599A
Authority
CN
China
Prior art keywords
defect
test
detection
pictures
initial
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN202110328165.1A
Other languages
Chinese (zh)
Inventor
范蕤
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Semiconductor Manufacturing International Shanghai Corp
Semiconductor Manufacturing International Beijing Corp
Original Assignee
Semiconductor Manufacturing International Shanghai Corp
Semiconductor Manufacturing International Beijing Corp
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Semiconductor Manufacturing International Shanghai Corp, Semiconductor Manufacturing International Beijing Corp filed Critical Semiconductor Manufacturing International Shanghai Corp
Priority to CN202110328165.1A priority Critical patent/CN115132599A/en
Publication of CN115132599A publication Critical patent/CN115132599A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • H—ELECTRICITY
    • H01—ELECTRIC ELEMENTS
    • H01L—SEMICONDUCTOR DEVICES NOT COVERED BY CLASS H10
    • H01L22/00—Testing or measuring during manufacture or treatment; Reliability measurements, i.e. testing of parts without further processing to modify the parts as such; Structural arrangements therefor
    • H01L22/10—Measuring as part of the manufacturing process
    • H01L22/12—Measuring as part of the manufacturing process for structural parameters, e.g. thickness, line width, refractive index, temperature, warp, bond strength, defects, optical inspection, electrical measurement of structural dimensions, metallurgic measurement of diffusions
    • H—ELECTRICITY
    • H01—ELECTRIC ELEMENTS
    • H01L—SEMICONDUCTOR DEVICES NOT COVERED BY CLASS H10
    • H01L22/00—Testing or measuring during manufacture or treatment; Reliability measurements, i.e. testing of parts without further processing to modify the parts as such; Structural arrangements therefor
    • H01L22/20—Sequence of activities consisting of a plurality of measurements, corrections, marking or sorting steps
    • H01L22/24—Optical enhancement of defects or not directly visible states, e.g. selective electrolytic deposition, bubbles in liquids, light emission, colour change
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
    • Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
    • Y02P90/30—Computing systems specially adapted for manufacturing

Landscapes

  • Engineering & Computer Science (AREA)
  • Manufacturing & Machinery (AREA)
  • Computer Hardware Design (AREA)
  • Microelectronics & Electronic Packaging (AREA)
  • Power Engineering (AREA)
  • Testing Or Measuring Of Semiconductors Or The Like (AREA)
  • Investigating Materials By The Use Of Optical Means Adapted For Particular Applications (AREA)

Abstract

A method of defect detection, comprising: acquiring a defect database, wherein the defect database comprises a plurality of defect data; providing a plurality of wafers to be detected; shooting each wafer to be detected from different angles to obtain a plurality of pictures to be detected of each wafer to be detected; acquiring a plurality of initial detection pictures from a plurality of pictures to be detected of each wafer to be detected; splicing a plurality of initial detection pictures of each wafer to be detected to form a detection picture; and detecting the detection picture according to the defect detection library to obtain a detection result. All pixel information of the selected picture is fully utilized, all characteristics of the picture are completely reserved, and the final detection accuracy is effectively improved.

Description

Defect detection method
Technical Field
The invention relates to the technical field of semiconductor manufacturing, in particular to a defect detection method.
Background
Semiconductor integrated circuit chips are manufactured by batch processing, and a large number of various types of semiconductor devices are formed on the same substrate and connected to each other to have a complete electronic function. Wherein defects generated in any step may cause failure of circuit fabrication. Therefore, in the manufacturing process, it is often necessary to detect and analyze defects of the manufacturing structure in each step, to find out the cause of the defects, and to eliminate the defects. However, with the rapid development of Ultra Large Scale Integration (ULSI), the Integration level of the chip is higher and higher, the size of the device is smaller and smaller, and accordingly, the size of the defect generated in the process of manufacturing the device enough to affect the flatness rate of the device is also smaller and smaller, which puts higher requirements on the defect detection of the semiconductor device.
However, the existing defect detection methods have low accuracy.
Disclosure of Invention
The invention aims to provide a defect detection method, which can effectively improve the precision of defect detection.
In order to solve the above problems, the present invention provides a defect detection method, including: acquiring a defect database, wherein the defect database comprises a plurality of defect data; providing a plurality of wafers to be detected; shooting each wafer to be detected from different angles to obtain a plurality of pictures to be detected of each wafer to be detected; acquiring a plurality of initial detection pictures from a plurality of pictures to be detected of each wafer to be detected; splicing a plurality of initial detection pictures of each wafer to be detected to form a detection picture; and detecting the detection picture according to the defect detection library to obtain a detection result.
Optionally, the method for acquiring the defect database includes: providing a plurality of training wafers, wherein each training wafer has a plurality of defects; training and learning each training wafer to obtain an initial defect database, wherein the initial defect database comprises a plurality of initial defect data; providing a plurality of groups of test wafers, wherein each group of test wafers comprises a plurality of test wafers; and testing and learning the initial defect database according to the plurality of test wafer groups to obtain the defect database.
Optionally, the method for performing training learning on each training wafer to obtain an initial defect database includes: shooting each training wafer from different angles to obtain a plurality of pictures to be trained of each training wafer; acquiring a plurality of initial training pictures from a plurality of pictures to be trained of each training wafer; splicing a plurality of initial training pictures of each training wafer to form training pictures; carrying out image recognition processing on each training picture to obtain initial defect data; the initial defect database is composed of a number of the initial defect data.
Optionally, the method for obtaining a plurality of initial training pictures from a plurality of to-be-trained pictures of each training wafer includes: acquiring the clearest shooting angle of the defect images in the training wafers of the same type according to an empirical rule; and taking the picture to be trained shot by the shooting angle as the initial training picture.
Optionally, the method for obtaining the defect database by performing test learning on the initial defect database according to the plurality of test wafer groups includes: shooting each test wafer from different angles to obtain a plurality of pictures to be tested of each test wafer; obtaining a plurality of initial test pictures from a plurality of pictures to be tested of each test wafer; splicing a plurality of initial test pictures of each test wafer to form a test picture; carrying out image recognition processing on each test picture to obtain test data; acquiring real data of each test picture; detecting the test data according to the initial defect database, and acquiring judgment data from a plurality of initial defect data; when the judgment data and the real data exceed the preset range, the corresponding test wafer is trained and learned again; after the test of one group of the test wafer groups is finished, calculating the test accuracy of the group of the test wafer groups; and when the test accuracy of the test wafer group is lower than a set threshold, performing test learning of the next test wafer group until the test accuracy of the test wafer group reaches the set threshold so as to obtain the defect database and keep the learning model with the best accuracy in the test learning process.
Optionally, the method for obtaining a plurality of initial test pictures from a plurality of pictures to be tested of each test wafer includes: acquiring the clearest shooting angle of the defect images in the same type of test wafers according to an empirical rule; and taking the picture to be tested shot by the shooting angle as the initial test picture.
Optionally, the method for obtaining a plurality of initial detection pictures from a plurality of pictures to be detected includes: acquiring the clearest shooting angle of the defect images in the wafers to be detected in the same type according to experience rules; and taking the picture to be detected shot by the shooting angle as the initial detection picture.
Optionally, the same types include: one or more of the same lot, the same process step, and the same wafer.
Optionally, the method for detecting the detection picture according to the defect detection library and obtaining the detection result includes: carrying out image identification processing on each detection picture to obtain detection data; respectively comparing the detection data with a plurality of defect data in the defect database; and within a certain contrast error range, when the detection data is the same as any defect data, acquiring the corresponding defect data, and taking the defect data as the detection result.
Optionally, the method for splicing a plurality of initial detection pictures to form a detection picture includes: and butting a plurality of initial detection pictures along a first direction or a second direction to form the detection pictures, wherein the first direction is vertical to the second direction.
Optionally, the learning model used in the training learning and the test learning processes includes a convolutional neural network model.
Compared with the prior art, the technical scheme of the invention has the following advantages:
in the detection method of the technical scheme, a plurality of initial detection pictures are obtained from a plurality of pictures to be detected of each wafer to be detected; splicing a plurality of initial detection pictures of each wafer to be detected to form a detection picture; and detecting the detection picture according to the defect detection library to obtain a detection result. All pixel information of the selected picture is fully utilized, all characteristics of the picture are completely reserved, and the final detection accuracy is effectively improved.
Further, the method for acquiring a plurality of initial detection pictures from a plurality of pictures to be detected comprises the following steps: acquiring the clearest shooting angle of the defect images in the wafers to be detected in the same type according to experience rules; and taking the picture to be detected shot by the shooting angle as the initial detection picture, so that the pseudo pixel interference of other pictures to be detected is reduced, and the final detection accuracy is effectively improved.
Drawings
FIG. 1 is a diagram of a picture processing architecture for a defect detection method;
FIG. 2 is a diagram of an alternative defect detection method;
fig. 3 to 15 are schematic structural diagrams of steps of a defect detection method according to an embodiment of the invention.
Detailed Description
As described in the background, the existing defect detection methods have low accuracy. The following detailed description will be made in conjunction with the accompanying drawings.
The existing wafer defect detection process comprises two types, one method is as follows: the method comprises the steps of mixing a plurality of pictures to be detected, which are shot from multiple angles, according to a random pixel proportion, and further obtaining a detected picture (as shown in figure 1). Although the method fully utilizes all pixel information, some unnatural false pixel information can be introduced, so that the final detection judgment is influenced, and the detection accuracy is reduced.
The other method comprises the following steps: a plurality of pictures to be detected, which are shot from multiple angles, are spliced according to a method of randomly intercepting parts, so that the detected pictures are obtained (as shown in figure 2). The method cannot guarantee that all defective positions of the picture are intercepted, and further final detection and judgment can be influenced, so that the detection accuracy is reduced.
On the basis, the invention provides a defect detection method, which comprises the steps of obtaining a plurality of initial detection pictures from a plurality of pictures to be detected; splicing a plurality of initial detection pictures to form detection pictures; and detecting the detection picture according to the defect detection library to obtain a detection result. All pixel information of the selected picture is fully utilized, all characteristics of the picture are completely reserved, and the final detection accuracy is effectively improved.
In order to make the aforementioned objects, features and advantages of the present invention comprehensible, embodiments accompanied with figures are described in detail below.
Fig. 3 to fig. 15 are schematic structural diagrams illustrating a process of forming a semiconductor structure according to an embodiment of the present invention.
In this embodiment, before performing defect detection on a wafer, a defect database is first acquired, where the defect database includes a plurality of defect data, and the defect of the wafer is determined according to the defect data stored in the defect database, so as to provide a final detection result. Please refer to fig. 3 to fig. 11 for a specific process of acquiring the defect database.
In this embodiment, the defect database is obtained in two stages, where the first stage is to form an initial defect database through training and learning; and the second stage is to obtain the defect database by further testing and verifying the initial defect database. Please refer to fig. 3 to fig. 6 for the process of training learning.
Referring to fig. 3, a plurality of training wafers 100 are provided, each of the training wafers 100 having a plurality of defects 101 thereon.
In the present embodiment, a sample is provided for the defect data in the defect database by the defect 101 on the training wafer 100.
Referring to fig. 4, each of the training wafers 100 is photographed from different angles, and a plurality of to-be-trained pictures 102 of each of the training wafers 100 are obtained.
Since one shooting angle cannot completely reflect the characteristics of the defect 101 of the training wafer 100, in this embodiment, each piece of the training wafer 100 is shot from different angles, so as to ensure that the defect information on the training wafer 100 can be acquired more accurately and comprehensively.
In this embodiment, the shooting angles include vertically downward, 45 ° downward above left, and 45 ° downward above right.
Referring to fig. 5, a plurality of initial training pictures 103 are obtained from a plurality of to-be-trained pictures 102 of each training wafer 100; and splicing a plurality of initial training pictures 103 of each training wafer 100 to form a training picture 104.
In this embodiment, the method for obtaining a plurality of initial training pictures 103 from a plurality of pictures to be trained 102 of each training wafer 100 includes: acquiring the clearest shooting angle of the defect images in the training wafers 100 of the same type according to an empirical rule; and taking the picture to be trained 102 shot by the shooting angle as the initial training picture 103.
In this embodiment, the same types include: one or more of the same lot, the same process step, and the same wafer.
Because some shooting angles in all the shooting angles can not clearly reflect the defect information of the training wafer 100, if the to-be-trained pictures 102 which can not clearly reflect the defect information are also selected, the interference of many false pixels is increased, and further, the final detection accuracy is influenced.
The staff knows the best shooting angles corresponding to different types of training wafers 100 through long-time accumulated empirical rules, the pictures shot at the shooting angles can clearly reflect defect information, the pictures to be trained 102 shot at the shooting angles are used as the initial training pictures 103, the pseudo-pixel interference of the rest pictures to be trained 102 can be reduced, and the final detection accuracy is effectively improved.
In this embodiment, the method for splicing the plurality of initial training pictures 103 of each training wafer 100 to form the training picture 104 includes: forming a plurality of initial training pictures 103 along a first direction X to form the training picture 104; in other embodiments, the method for forming the training pictures by splicing a plurality of initial training pictures of each training wafer includes: and forming the training pictures by using the plurality of initial training pictures along a second direction Y, wherein the first direction X is vertical to the second direction Y.
In this embodiment, for example, each training wafer 100 is used to take 3 images 102 to be trained, and 2 of the images are selected as initial training images 103.
In the embodiment, all pixel information of the selected picture is fully utilized, and all characteristics of the picture are completely reserved, so that the final detection accuracy is effectively improved.
Referring to fig. 6, image recognition processing is performed on each of the training pictures 104 to obtain initial defect data; the initial defect database is composed of a number of the initial defect data.
In this embodiment, an image of the training picture 104 is scanned by using an image recognition process, a defect in the training picture 104 is recognized, and the defect is converted into corresponding defect data to be stored.
In this embodiment, the defect data corresponding to each training wafer 100 further includes type information of the training wafer 100.
In this embodiment, the learning model used in the training learning process is a convolutional neural network model.
In this embodiment, the initial defect database is obtained through training and learning, but the reliability of the initial defect database is not verified. Therefore, after the training learning is performed, the initial defect database needs to be tested and learned, and the defect database is further acquired, so as to improve the reliability of the initial defect data. Please refer to fig. 7 to fig. 11 for the process of the test learning.
Referring to fig. 7, a plurality of test wafer groups are provided, each of the test wafer groups includes a plurality of test wafers 200.
In this embodiment, the purpose of providing a plurality of groups of test wafer groups is to calculate the test accuracy of the initial defect database after completing the test learning of a group of test wafer groups, and if the test accuracy reaches a threshold, complete the test learning; when the test accuracy does not reach the threshold, the next group of the test wafer group is used for test learning until the test accuracy reaches the threshold.
The process of the test learning will be described below by taking a group of the test wafers as an example.
Referring to fig. 8, each of the test wafers 200 is photographed from different angles, and a plurality of pictures 201 to be tested of each of the test wafers 200 are obtained.
In this embodiment, the process of acquiring the picture to be tested 201 is consistent with the process of acquiring the picture to be trained 102, which will not be described herein again.
Referring to fig. 9, a plurality of initial test pictures 202 are obtained from a plurality of pictures 201 to be tested of each test wafer 200; and splicing a plurality of initial test pictures 202 of each test wafer 200 to form a test picture 203.
In this embodiment, the processes of obtaining the plurality of initial test pictures 202 and forming the test picture 203 are the same as the processes of obtaining the plurality of initial training pictures 103 and forming the training picture 104, which will not be described herein again.
It should be noted that the shooting angles of the test wafer 200 and the training wafer 100 of the same type should be consistent, that is, the shooting angle corresponding to the initial test picture 202 is consistent with the shooting angle corresponding to the initial training picture 103.
Referring to fig. 10, image recognition processing is performed on each of the test pictures 203 to obtain test data.
In this embodiment, the process of performing the image recognition processing on the test picture 203 is the same as the process of performing the image recognition processing on the training picture 104, and will not be described herein again.
Referring to fig. 11, real data of each of the test pictures 203 is obtained; detecting the test data according to the initial defect database, and acquiring judgment data from a plurality of initial defect data; and when the judgment data and the real data exceed a preset range, replacing the corresponding initial defect data with the real data, and updating the initial defect data into the initial defect database.
As the test learning process is adopted, the real data passes through the result of artificial judgment, the judgment of the initial defect database is verified through artificial judgment, and the reliability of the initial defect data is further verified.
When the artificial judgment is inconsistent with the judgment of the initial defect database, the judgment is regarded as an error of the initial defect database, and the corresponding test wafer 200 is trained and learned again, so that the test accuracy of the initial defect database is improved.
After completing the test learning of a group of test wafers, calculating the test accuracy of the defect database, and if the test accuracy reaches a threshold value, completing the test learning; when the test accuracy does not reach the threshold, the next group of the test wafer groups is used for test learning until the test accuracy reaches the threshold. So far, the acquisition process of the defect database is finished, and the model with the best accuracy in the test learning process is reserved.
In this embodiment, the learning model used in the test learning process is a convolutional neural network model.
Referring to fig. 12, a plurality of wafers 300 to be tested are provided.
In this embodiment, whether the wafer 300 to be detected has defects is unknown, so the wafer 300 to be detected needs to be determined through the obtained defect database, and then the detection result is obtained.
Referring to fig. 13, each wafer 300 to be detected is photographed from different angles, and a plurality of pictures 301 to be detected of each wafer 300 to be detected are obtained.
In this embodiment, the acquiring process of the picture to be detected 301 is consistent with the acquiring process of the picture to be trained 102, which will not be described herein again.
Referring to fig. 14, a plurality of initial inspection pictures 302 are obtained from a plurality of pictures 301 to be inspected of each wafer 300 to be inspected; and splicing the initial detection pictures 302 of each wafer 300 to be detected to form a detection picture 303.
In this embodiment, the processes of obtaining the plurality of initial detection pictures 302 and forming the detection picture 303 are the same as the processes of obtaining the plurality of initial training pictures 103 and forming the training picture 104, which will not be described herein again.
It should be noted that the shooting angles of the wafer 300 to be detected and the training wafer 100 of the same type should be consistent, that is, the shooting angle corresponding to the initial detection picture 302 is consistent with the shooting angle corresponding to the initial training picture 103. This enables subsequent detection to be performed under the same pixel conditions.
Referring to fig. 15, the detection picture 303 is detected according to the defect detection library, and a detection result is obtained.
In this embodiment, the method for detecting the detection picture 303 according to the defect detection library and obtaining the detection result includes: performing image recognition processing on each detection picture 303 to obtain detection data; respectively comparing the detection data with a plurality of defect data in the defect database; and within a certain contrast error range, when the detection data is the same as any defect data, acquiring the corresponding defect data, and taking the defect data as the detection result.
In this embodiment, the process of performing the image recognition processing on the detection picture 303 is the same as the process of performing the image recognition processing on the training picture 104, which will not be described herein again.
In this embodiment, the detection data corresponding to each wafer 300 to be detected further includes type information of the wafer 300 to be detected.
It should be noted that before the detection data is compared with the defect data, it is required to ensure that the type information corresponding to the detection data is consistent with the type information corresponding to the defect data.
In this embodiment, a plurality of initial inspection pictures 302 are obtained from a plurality of pictures 301 to be inspected of each wafer 300 to be inspected; splicing a plurality of initial detection pictures 302 of each wafer 300 to be detected to form a detection picture 303; and detecting the detection picture 303 according to the defect detection library to obtain a detection result. All pixel information of the selected picture is fully utilized, all characteristics of the picture are completely reserved, and the final detection accuracy is effectively improved.
Although the present invention is disclosed above, the present invention is not limited thereto. Various changes and modifications may be effected by one skilled in the art without departing from the spirit and scope of the invention, as defined in the appended claims.

Claims (11)

1. A method of defect detection, comprising:
acquiring a defect database, wherein the defect database comprises a plurality of defect data;
providing a plurality of wafers to be detected;
shooting each wafer to be detected from different angles to obtain a plurality of pictures to be detected of each wafer to be detected;
acquiring a plurality of initial detection pictures from a plurality of pictures to be detected of each wafer to be detected; splicing a plurality of initial detection pictures of each wafer to be detected to form a detection picture;
and detecting the detection picture according to the defect detection library to obtain a detection result.
2. The defect detection method of claim 1, wherein the method of obtaining the defect database comprises: providing a plurality of training wafers, wherein each training wafer has a plurality of defects; training and learning each training wafer to obtain an initial defect database, wherein the initial defect database comprises a plurality of initial defect data; providing a plurality of groups of test wafers, wherein each group of test wafers comprises a plurality of test wafers; and testing and learning the initial defect database according to the plurality of test wafer groups to obtain the defect database.
3. The method of claim 2, wherein the training learning is performed on each of the training wafers, and the method of obtaining the initial defect database comprises: shooting each training wafer from different angles to obtain a plurality of pictures to be trained of each training wafer; acquiring a plurality of initial training pictures from a plurality of pictures to be trained of each training wafer; splicing a plurality of initial training pictures of each training wafer to form training pictures; carrying out image recognition processing on each training picture to obtain initial defect data; and composing the initial defect database by a plurality of initial defect data.
4. The defect detection method of claim 3, wherein the method for obtaining a plurality of initial training pictures from a plurality of pictures to be trained of each training wafer comprises: acquiring the clearest shooting angle of the defect images in the training wafers of the same type according to an empirical rule; and taking the picture to be trained shot by the shooting angle as the initial training picture.
5. The method of claim 3, wherein the initial defect database is learned from a plurality of test wafer lots, and the obtaining the defect database comprises: shooting each test wafer from different angles to obtain a plurality of pictures to be tested of each test wafer; obtaining a plurality of initial test pictures from a plurality of pictures to be tested of each test wafer; splicing a plurality of initial test pictures of each test wafer to form a test picture; carrying out image recognition processing on each test picture to obtain test data; acquiring real data of each test picture; detecting the test data according to the initial defect database, and acquiring judgment data from a plurality of initial defect data; when the judgment data and the real data exceed the preset range, the corresponding test wafer is trained and learned again; after the test of one group of test wafer groups is finished, calculating the test accuracy of the group of test wafer groups; and when the test accuracy of the test wafer group is lower than a set threshold, performing test learning of the next test wafer group until the test accuracy of the test wafer group reaches the set threshold so as to obtain the defect database and keep the learning model with the best accuracy in the test learning process.
6. The method of defect detection according to claim 5, wherein the method of obtaining a plurality of initial test pictures from a plurality of the pictures to be tested for each of the test wafers comprises: acquiring the clearest shooting angle of the defect images in the same type of test wafers according to an empirical rule; and taking the picture to be tested shot by the shooting angle as the initial test picture.
7. The defect detection method of claim 1, wherein the method for obtaining a plurality of initial inspection pictures from a plurality of pictures to be inspected comprises: acquiring the clearest shooting angle of the defect images in the wafers to be detected in the same type according to experience rules; and taking the picture to be detected shot by the shooting angle as the initial detection picture.
8. The defect detection method of claim 4, 6 or 7, wherein the same type comprises: one or more of the same lot, the same process step, and the same wafer.
9. The method for detecting the defects of claim 1, wherein the method for detecting the detection pictures according to the defect detection library and obtaining the detection results comprises: carrying out image identification processing on each detection picture to obtain detection data; respectively comparing the detection data with a plurality of defect data in the defect database; and within a certain comparison error range, when the detection data is the same as any defect data, acquiring the corresponding defect data, and taking the defect data as the detection result.
10. The defect detection method of claim 1, wherein the step of splicing a plurality of the initial detection pictures to form a detection picture comprises: and butting a plurality of initial detection pictures along a first direction or a second direction to form the detection picture, wherein the first direction is vertical to the second direction.
11. The defect detection method of claim 2, wherein the learning models employed in the training learning and the test learning processes include convolutional neural network models.
CN202110328165.1A 2021-03-26 2021-03-26 Defect detection method Pending CN115132599A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202110328165.1A CN115132599A (en) 2021-03-26 2021-03-26 Defect detection method

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202110328165.1A CN115132599A (en) 2021-03-26 2021-03-26 Defect detection method

Publications (1)

Publication Number Publication Date
CN115132599A true CN115132599A (en) 2022-09-30

Family

ID=83374845

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202110328165.1A Pending CN115132599A (en) 2021-03-26 2021-03-26 Defect detection method

Country Status (1)

Country Link
CN (1) CN115132599A (en)

Similar Documents

Publication Publication Date Title
US12020417B2 (en) Method and system for classifying defects in wafer using wafer-defect images, based on deep learning
CN109558620B (en) Method for generating inspection plan and system thereof
CN112424826A (en) Pattern grouping method based on machine learning
US20090290784A1 (en) Methods and systems for binning defects detected on a specimen
JP2000208575A (en) Method and system for detecting defect based on features of profile
KR20180095715A (en) Improved defect sensitivity of semiconductor wafer testers using design data with wafer image data
WO2013040063A2 (en) Determining design coordinates for wafer defects
KR100598381B1 (en) in-line typed apparatus for auto wafer-defect classification and control method thereof
US10719655B2 (en) Method and system for quickly diagnosing, classifying, and sampling in-line defects based on CAA pre-diagnosis database
TWI765442B (en) Method for defect level determination and computer readable storage medium thereof
CN113269709A (en) Method for detecting defects in semiconductor wafer and semiconductor wafer defect detection system
TW201731004A (en) Reducing registration and design vicinity induced noise for intra-die inspection
CN118401898A (en) Method and system for defect detection of inspection samples based on machine learning model
CN117751382A (en) Abnormality-based defect inspection method and system
US20160110859A1 (en) Inspection method for contact by die to database
CN116740074B (en) Wafer defect accurate identification method based on machine vision
CN113299572A (en) Chip defect detection method
US12062166B2 (en) Method and system for diagnosing a semiconductor wafer
US7020536B2 (en) Method of building a defect database
JP2768344B2 (en) Pattern inspection apparatus and inspection method thereof
CN115132599A (en) Defect detection method
CN116263559A (en) Method for predicting optical proximity defect on graph to be detected based on machine learning
KR20020041785A (en) Two-dimensional scatter plot technique for defect inspection
EP1048956A2 (en) Method and apparatus for analizing a semiconductor wafer manufacturing process
CN114113142B (en) Defect detection method and device for semiconductor device and electronic equipment

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination