CN113077462A - Wafer defect classification method, device, system, electronic equipment and storage medium - Google Patents
Wafer defect classification method, device, system, electronic equipment and storage medium Download PDFInfo
- Publication number
- CN113077462A CN113077462A CN202110477848.3A CN202110477848A CN113077462A CN 113077462 A CN113077462 A CN 113077462A CN 202110477848 A CN202110477848 A CN 202110477848A CN 113077462 A CN113077462 A CN 113077462A
- Authority
- CN
- China
- Prior art keywords
- wafer defect
- defect
- classified
- wafer
- reference sample
- 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
Links
- 230000007547 defect Effects 0.000 title claims abstract description 570
- 238000000034 method Methods 0.000 title claims abstract description 50
- 238000003860 storage Methods 0.000 title claims abstract description 13
- 239000013074 reference sample Substances 0.000 claims abstract description 131
- 235000012431 wafers Nutrition 0.000 claims description 392
- 238000001514 detection method Methods 0.000 claims description 20
- 238000004422 calculation algorithm Methods 0.000 claims description 17
- 238000004891 communication Methods 0.000 claims description 12
- 238000012545 processing Methods 0.000 claims description 9
- 238000004590 computer program Methods 0.000 claims description 3
- 239000000126 substance Substances 0.000 claims description 2
- 238000012549 training Methods 0.000 abstract description 17
- 238000004519 manufacturing process Methods 0.000 abstract description 12
- 239000000523 sample Substances 0.000 abstract description 8
- 238000004364 calculation method Methods 0.000 abstract description 5
- 238000013527 convolutional neural network Methods 0.000 description 18
- 238000013145 classification model Methods 0.000 description 12
- 230000008569 process Effects 0.000 description 10
- 230000003287 optical effect Effects 0.000 description 7
- 238000007689 inspection Methods 0.000 description 6
- 238000010586 diagram Methods 0.000 description 4
- 238000002372 labelling Methods 0.000 description 4
- 239000004065 semiconductor Substances 0.000 description 3
- 238000013528 artificial neural network Methods 0.000 description 2
- 238000007635 classification algorithm Methods 0.000 description 2
- 238000003066 decision tree Methods 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000001914 filtration Methods 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 230000010354 integration Effects 0.000 description 2
- 238000005498 polishing Methods 0.000 description 2
- 238000007781 pre-processing Methods 0.000 description 2
- 230000000644 propagated effect Effects 0.000 description 2
- 238000012706 support-vector machine Methods 0.000 description 2
- 238000013459 approach Methods 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 238000005229 chemical vapour deposition Methods 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 239000000203 mixture Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 239000013307 optical fiber Substances 0.000 description 1
- 238000003909 pattern recognition Methods 0.000 description 1
- 230000002093 peripheral effect Effects 0.000 description 1
- 238000000206 photolithography Methods 0.000 description 1
- 230000004044 response Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0004—Industrial image inspection
- G06T7/001—Industrial image inspection using an image reference approach
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
- G06F18/2155—Generating training patterns; Bootstrap methods, e.g. bagging or boosting characterised by the incorporation of unlabelled data, e.g. multiple instance learning [MIL], semi-supervised techniques using expectation-maximisation [EM] or naïve labelling
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2413—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on distances to training or reference patterns
- G06F18/24147—Distances to closest patterns, e.g. nearest neighbour classification
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20084—Artificial neural networks [ANN]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30108—Industrial image inspection
- G06T2207/30148—Semiconductor; IC; Wafer
Abstract
The invention relates to a wafer defect classification method, which comprises the steps of carrying out unsupervised learning clustering on a plurality of wafer defect patterns to be classified to obtain a plurality of groups of wafer defect groups, selecting a corresponding number of wafer defect patterns to be classified from each group to be matched with a defect pattern reference sample, and judging all wafer defect patterns to be classified in a corresponding group to be wafer defects of corresponding classes if the wafer defect patterns to be classified are matched with the defect pattern reference sample. The classification method of the invention does not need a large amount of training sample data and model training, greatly reduces time resources and calculation resources of equipment, reduces production cost to a certain extent and improves production efficiency. Correspondingly, the invention also provides a wafer defect classification device, a wafer defect classification system and a computer-readable storage medium.
Description
Technical Field
The invention relates to the technical field of semiconductor manufacturing, in particular to a wafer defect classification method, a device, a system, electronic equipment and a computer readable storage medium based on unsupervised learning.
Background
In semiconductor wafer fabrication, chemical vapor deposition, optical development, and chemical mechanical polishing (cmp) can cause defects on the wafer surface during the pulling, slicing, lapping, polishing, layer building, photolithography, doping, thermal processing, probing, and dicing … … processes, which directly affect the lifetime and reliability of the wafer. Wafer defect identification is generally divided into two steps, defect detection and defect classification. The traditional defect detection algorithm selects a difference area between a pattern to be detected and a reference pattern as a pattern defect by comparing the pattern to be detected and the reference pattern. The traditional defect classification algorithm needs to extract features from a wafer pattern first, and then the extracted features are used as input of a classifier, so that pattern recognition is performed. The above method mainly has the following 2 problems: 1) local features extracted by a traditional defect detection algorithm cannot effectively represent different types of defects in a specific problem; 2) the traditional defect detection classification algorithm is weak in robustness, and the whole recognition model needs to be redesigned according to new problems.
With the development of machine learning technology, various recognizers are widely used for wafer map defect recognition and achieve better effects, for example, a supervised recognition classification method: a back-propagation network (BPN) [6], a Generalized Regression Neural Network (GRNN) [7], a Support Vector Machine (SVM) [8-9], a randomized generalized regression network (randomised generalized regression network) [10], a K-neighbor algorithm (KNN) [11], a decision tree (CART4.5) [12], a Gaussian mixture model [13], a multi-feature decision tree integration [14], a mean value standard integration method [15], and the like; and the feature extraction method based on deep learning comprises the following steps: convolutional Neural Networks (CNN). Because the convolutional neural network effectively solves the problems of pattern classification and target identification in different fields by introducing a convolutional kernel structure, the convolutional neural network is one of the wafer defect classification methods which are commonly used at present.
However, the following problems still exist when the convolutional neural network model, such as the CNN algorithm, is widely applied to the field of wafer defect detection and classification:
1) the amount of sample data to train the model is difficult to achieve. In the process of chip defect identification of the convolutional neural network, a convolutional neural network model needs to be trained first, so that a model capable of performing defect identification is obtained. However, during training, a large number of defect patterns and corresponding defect types are required, and the sample size of the data set typically amounts to tens or even hundreds of thousands. However, in actual production, due to the limited number of defect pictures generated on the line and the particularity of the wafer defect detection problem, the defect number sample size of the training model is difficult to reach ten thousand levels, and each factory does not share respective defect data and generates different types of defects, so that the characteristic of large requirement of deep convolutional neural network training data is difficult to meet. If the number of defect patterns used for training is insufficient, the accuracy of the model is affected.
2) A large number of data sets naturally bring labeling difficulty, and according to statistics, the time of about 2 to 3 seconds is needed for marking a single object type in a single pattern, but the data sets in practical application often contain thousands of pictures, and the whole labeling process becomes extremely long. Especially when fine-grained classification and multi-label classification tasks are involved, the labeling cost increases exponentially with the target number and the recognizable difficulty.
3) Because the types of defects generated by different factories are different, corresponding classifiers are obtained by training according to the data sets of known defect types, and the classifiers have no universality, that is, different factories need to construct different data sets, so that different classifiers are trained to classify the defects, and the production and manufacturing cost is further increased.
4) The whole deep neural network architecture training process is not easy to fit, and a large amount of time resources and calculation resources are needed.
Based on the above problems, at present, convolutional neural network models, such as CNN algorithm, cannot be widely applied to the field of wafer defect detection and classification. In view of the above, in order to avoid the acquisition and model training involving a large number of samples, the present disclosure proposes a new wafer defect classification method.
Disclosure of Invention
It is an object of the present invention to provide a method, an apparatus, a system, an electronic device and a storage medium for classifying wafer defects to alleviate or partly solve the above technical problem, thereby avoiding the acquisition and model training involving large amounts of samples.
In a first aspect of the present invention, a wafer defect classification method based on unsupervised learning is provided, which includes the steps of:
obtaining a plurality of wafer defect patterns to be classified;
carrying out unsupervised learning clustering on the wafer defect patterns to be classified to obtain a plurality of groups of wafer defect groups to be classified;
matching the wafer defect patterns to be classified with a preset threshold number in each group with at least one pre-stored wafer defect pattern reference sample, and if the matching is successful, marking corresponding defect type identifications of the wafer defect groups to be classified where the wafer defect patterns to be classified are located, wherein the defect type identifications are defect type identifications of the wafer defect pattern reference samples matched currently;
each wafer defect pattern reference sample is labeled with a defect type identifier in advance, and at least one wafer defect pattern reference sample is labeled with the same defect type identifier.
In some exemplary embodiments of the invention, the method further comprises: and if the matching is unsuccessful, automatically storing the current wafer defect pattern to be classified as a new wafer defect pattern reference sample corresponding to the new wafer defect.
In some exemplary embodiments of the invention, the method further comprises: and generating a custom type identifier corresponding to the new wafer defect pattern reference sample according to a pre-stored defect type identifier custom mode.
In some exemplary embodiments of the present invention, the wafer defect category of the wafer defect group to be classified where the current wafer defect pattern to be classified that is not successfully matched is currently located is marked as the custom category identifier.
In some exemplary embodiments of the invention, the unsupervised clustering model used in unsupervised learning clustering comprises a clustering algorithm model.
In a second aspect of the present invention, an unsupervised learning-based wafer defect classification apparatus is provided, the apparatus comprising:
the data acquisition module is used for acquiring a plurality of wafer defect patterns to be classified;
the data processing module is used for carrying out unsupervised learning clustering on the plurality of wafer defect patterns to be classified acquired by the data acquisition model to obtain a plurality of groups of wafer defect groups to be classified;
the defect classification module is used for matching the wafer defect patterns to be classified with a preset threshold number in each group of wafer defect groups to be classified with at least one pre-stored wafer defect pattern reference sample, and if the matching is successful, marking corresponding defect type identifications for the wafer defect pattern group where the wafer defect patterns to be classified are located, wherein the defect type identifications are the defect type identifications of the wafer defect pattern reference samples matched currently;
each wafer defect pattern reference sample is labeled with a defect type identifier in advance, and at least one wafer defect pattern reference sample is labeled with the same defect type identifier.
In some exemplary embodiments of the invention, the apparatus further comprises: and the self-defining module is used for automatically storing the current wafer defect pattern to be classified as a new wafer defect pattern reference sample corresponding to the new wafer defect when the defect classification module is unsuccessfully matched.
In some exemplary embodiments of the present invention, the customization module is further configured to generate a customized category identifier corresponding to the new wafer defect pattern reference sample according to a pre-stored defect category identifier customization manner.
In some exemplary embodiments of the present invention, the defect classification module is further configured to mark the custom class identifier for a wafer defect class of a wafer defect group to be classified where the current wafer defect pattern to be classified is not successfully matched.
Further, in some of the above exemplary embodiments of the present invention, the unsupervised clustering model employed in unsupervised learning clustering includes a clustering algorithm model.
A third aspect of the present invention provides a wafer defect sorting system, comprising:
the wafer defect detection device is used for detecting a plurality of wafers to obtain a plurality of wafer defect patterns to be classified;
the wafer defect classification device in at least one of the above exemplary embodiments is configured to obtain a plurality of wafer defect patterns to be classified from the wafer defect detection device, and perform unsupervised learning clustering to obtain a plurality of groups of wafer defect groups to be classified;
then, matching the wafer defect patterns to be classified with a preset threshold number in each group with at least one pre-stored wafer defect pattern reference sample, and if the matching is successful, marking corresponding defect type identifications of the wafer defect group to be classified where the wafer defect patterns to be currently classified are located, wherein the defect type identifications are defect type identifications of the wafer defect pattern reference samples which are currently matched;
each wafer defect pattern reference sample is marked with a defect type identifier in advance, and at least one wafer defect pattern reference sample is marked with the same defect type identifier.
A fourth aspect of the present invention provides an electronic device comprising at least one processor, at least one memory, a communication interface, and a bus;
the processor, the memory and the communication interface complete mutual communication through the bus; the memory is used for storing a program for executing the method of any of the above exemplary embodiments; the processor is configured to execute programs stored in the memory.
A fifth aspect of the present invention provides a computer-readable storage medium, in which a computer program is stored, which, when being executed by a processor, carries out the steps of the method of any one of the above-mentioned exemplary embodiments.
Advantageous technical effects
The wafer defect classification method is based on an Unsupervised learning (Unsupervised learning) method to group (or cluster) wafer defect patterns to be classified to obtain a plurality of groups of wafer defect groups to be classified; then, selecting a corresponding number of wafer defect patterns to be classified from each group to be matched with at least one wafer defect pattern reference sample which is classified in advance, if the wafer defect pattern reference samples of the corresponding types are matched, marking all the wafer defect patterns to be classified in the corresponding groups with defect category identifications corresponding to the matched wafer defect pattern reference samples, acquiring and manually marking the number of tens of thousands of training samples in the whole classification process, and performing model training.
Further, when the matching is unsuccessful, that is, the corresponding wafer defect pattern reference sample is not matched, the wafer defect pattern to be classified is used as a new wafer defect pattern reference sample, so that the reference samples in the database are enriched, and even if the wafer defect patterns of the same type appear, the corresponding wafer defect pattern reference sample can be directly matched, so that the corresponding defect category identification can be directly marked according to the new wafer defect pattern reference sample, that is, the defect type can be identified, model training is not required to be performed on all the wafer defect patterns to be classified again, thus further saving the calculation resources and time resources and improving the classification efficiency.
Furthermore, the wafer defect pattern to be classified which is unsuccessfully matched is used as a new wafer defect pattern reference sample, and a corresponding self-defined wafer defect type is generated for the new wafer defect pattern reference sample according to a pre-stored self-defined mode, so that the new wafer defect type can be added into the database in a self-defined mode without manually adding the new wafer defect type, and the corresponding wafer defect pattern reference sample is not required to be manually added for the new wafer defect type, namely, the newly added wafer defect pattern reference sample is not required to be manually labeled with the corresponding wafer defect type, and a model is not required to be retrained, so that the labor intensity of workers is greatly reduced, and the production efficiency is improved.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below. Throughout the drawings, like elements or portions are generally identified by like reference numerals. In the drawings, elements or portions are not necessarily drawn to scale. It is obvious that the drawings in the following description are some embodiments of the invention, and that for a person skilled in the art, other drawings can be derived from them without inventive exercise.
Fig. 1 is a flowchart of a wafer defect classification method according to an exemplary embodiment of the present invention;
fig. 2 is a schematic structural diagram of an embodiment of a wafer defect sorting apparatus according to an exemplary embodiment of the invention;
fig. 3 is a schematic structural diagram of an electronic device according to an exemplary embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention. It is to be understood that the embodiments described are only a few embodiments of the present invention, and not all embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
Herein, suffixes such as "module", "part", or "unit" used to denote elements are used only for facilitating the description of the present invention, and have no specific meaning in itself. Thus, "module", "component" or "unit" may be used mixedly.
The term is defined as:
"wafer defect pattern": in a semiconductor manufacturing process, a defect Inspection apparatus, such as an Automated Optical Inspection (AOI), an X-ray Inspection, a Scanning Electron Microscope (SEM), etc., is generally used to scan/inspect each wafer, so as to obtain a wafer pattern and a pattern of various defects on the wafer (including characteristic parameters such as shapes, positions, and sizes of the defects). Since different processes are involved in the whole manufacturing process, and different types of defects are formed on the wafer by the different processes, patterns of various types of defects are obtained after scanning, and the patterns of the defects obtained after the wafer is scanned by the inspection equipment are collectively referred to as wafer defect patterns.
"defect class identification": as is well known, there are various types (or types) of wafer defects, and therefore, in order to distinguish different types of wafer defects, different identifications need to be defined for the various wafer defects, and such identifications are defect type identifications in this document. For example, a defect type name (e.g., an industry-common short circuit defect, open circuit defect, or island defect, etc.), or a code (e.g., a numerical number, an alphabetical code, or a combination of numerical and alphabetical codes) for uniquely identifying the defect type.
Example one
Referring to fig. 1, a flowchart of a wafer defect classification method according to an exemplary embodiment of the invention is shown. Specifically, the wafer defect classification method of the present exemplary embodiment includes the steps of:
s101, obtaining a plurality of wafer defect patterns to be classified.
In some embodiments, the defect pattern of the wafer to be classified is obtained by scanning each wafer through various defect inspection apparatuses, such as SEM, in advance, and accordingly, a plurality of defect patterns of the wafer to be classified can be directly obtained from the defect inspection apparatuses. For example, a plurality of wafer defect patterns to be classified may be acquired from the defect detecting device through wired communication or wireless communication, that is, the acquired wafer defect set to be classified includes: a1, A2, A3, A4, A5. cndot. AN.
S103, carrying out unsupervised learning clustering on the plurality of wafer defect patterns to be classified acquired in the step S101 to obtain a plurality of groups of wafer defect groups to be classified.
As is well known, there are various types of wafer defects, such as short-circuit defects, open-circuit defects, island defects, etc., and each wafer defect pattern corresponding to the same type of wafer defect has the same or similar characteristics, so in some embodiments, an unsupervised learning method is used to cluster (or group) the acquired wafer defect patterns to be classified, so as to divide the same or similar wafer defect patterns into the same group. Specifically, an unsupervised clustering model such as a clustering algorithm may be used for clustering/grouping. For example, clustering the wafer defect set to be classified by using a clustering algorithm to obtain a plurality of wafer defect groups to be classified: group 1{ A1, A4, A6 · }, group 2{ A2, A5, A7 · }, group 3{ A3, A10, A11 · }.
In some embodiments, to improve the accuracy of clustering, before performing step S103, each wafer defect pattern to be classified acquired in step S101 may be preprocessed, specifically, the preprocessing includes filtering, sharpening, histogram equalization, super-resolution, brightness and contrast adjustment, and the like.
S105, selecting wafer defect patterns to be classified with a preset threshold quantity from each group of wafer defect groups to be classified, matching the wafer defect patterns with at least one pre-stored wafer defect pattern reference sample, if the wafer defect patterns are matched, executing the step S107, otherwise, executing the step S109.
In some embodiments, a wafer defect pattern reference sample library (hereinafter referred to as a reference sample library) may be pre-constructed for storing at least one defect pattern reference sample corresponding to each type of currently existing or known wafer defect, that is, a plurality of wafer defect pattern reference samples (hereinafter referred to as reference samples) are pre-stored in the reference sample library, and each reference sample is pre-marked with a corresponding defect type identifier. Specifically, a professional may mark/label each reference sample in the reference sample library with a corresponding defect type identifier (such as a defect type name, or a code identifying various defect types), and at least one reference sample is labeled with the same defect type identifier, that is, each wafer defect corresponds to at least one wafer defect pattern reference sample. For example, the reference sample library has: the reference samples C1, C2, C3. cndot. Cm are labeled with a first defect type identifier, i.e., corresponding to a first type of wafer defect; the reference samples Cm +1, Cm +2, Cm +3, · · Cm + p are labeled with a second defect type identifier, i.e., corresponding to a second type of wafer defect; the reference samples Cm + p +1, Cm + p +2, Cm + p +3,. Cm + p + q are labeled with a third defect type identifier, i.e., corresponding to a third wafer defect, a total of predicted X wafer defects, and each wafer defect corresponds to one defect type identifier.
In some embodiments, a pre-trained image classification model is used to complete the matching between the wafer defect pattern to be classified and the wafer defect pattern reference sample. Specifically, the training of the image classification model is obtained by learning according to at least one pre-stored wafer defect pattern reference sample.
In some embodiments, the wafer defect pattern reference sample (i.e., reference sample) is selected by the following principles: an image classification model, such as a CNN algorithm, can extract classified key feature parameters (such as features and attribute information of defects themselves, such as optical signal intensity of defects in a pattern, positions of defects, sizes of defects, and the like) and weight relations thereof from the reference sample, so that the influence of different backgrounds (i.e., patterns around the defects) can be minimized, that is, the trained image classification model can classify the defect patterns of a wafer to be classified based on the defects themselves in the pattern, while neglecting surrounding patterns, that is, without being interfered by the background, for example, after the CNN learns each reference sample in the reference sample library, an image classification model capable of classifying based on the defects itself is obtained. Therefore, when the selected wafer defect pattern to be classified is learned, the CNN algorithm can extract the classification key feature parameters and the weight relationship thereof of each wafer defect pattern to be classified, and then match the classification key feature parameters and the weight relationship thereof with the key feature parameters and the weight relationship of the reference sample, thereby obtaining a corresponding matching result. That is, in the embodiment, the defects in the defect pattern of the wafer to be classified are classified without considering the patterns around the defects, so that the influence of the patterns around the defects or different backgrounds on defect classification is avoided, and the accuracy of clustering is improved. For example, the wafer defect patterns to be classified are subjected to CNN learning to extract classification key feature parameters and weight relations, so as to minimize the influence of different backgrounds (such as patterns around the defect).
In some embodiments, as described above, since the classification key feature parameters and the weight relationship thereof of each reference sample are known, and the trained image classification model may be used to learn the wafer defect pattern to be classified to extract the classification key feature parameters and the weight relationship thereof, the extracted classification key feature parameters and the weight relationship of the wafer defect pattern to be classified may be matched with the classification key feature parameters and the weight relationship thereof of the reference sample, once the matching is successful (for example, a matching score may be obtained according to a matching degree between the classification key feature parameters and the weight relationship thereof, and if the matching score is obtained or the matching degree exceeds a preset threshold, the matching is considered to be successful), the matching is stopped, and step S107 is performed, otherwise, step S109 is performed.
Preferably, a corresponding wafer defect pattern reference sample is stored for each wafer defect, that is, each wafer defect pattern in the reference sample library corresponds to a defect type identifier. For example, in the reference sample library: the reference sample C1 is labeled with a first defect type identifier X1 corresponding to the first wafer defect; the reference sample Ci is labeled with a second defect type identifier X2 corresponding to the second wafer defect; the reference sample Cj is labeled with a third defect type identifier X3 ·correspondingto a third wafer defect.
In some embodiments, the preset threshold is typically 2, that is, two wafer defect patterns to be classified are selected from each group of wafer defect groups to be classified to match with the pre-stored wafer defect pattern reference samples. Specifically, two wafer defect patterns to be classified can be randomly selected from each group of wafer defect groups to be classified, for example, a4 and a6 are randomly selected from { a1, a4, a6 · } in the group 1, and then are matched with at least one wafer defect pattern reference sample corresponding to each wafer defect in advance; similarly, randomly selecting A5 and A7 from the group 2{ A2, A5, A7 · } to match with at least one wafer defect pattern reference sample corresponding to each wafer defect; randomly selecting A3 and A11 from the group 3{ A3, A10, A11- }, and matching the at least one wafer defect pattern reference sample corresponding to each wafer defect in advance.
Of course, the preset threshold may also be adjusted according to actual needs, for example, 1 or 3 frames are selected.
In some embodiments, in order to improve the accuracy of defect classification and further refine the wafer defect pattern to be classified, the selection rule of the wafer defect pattern to be classified may use the data of the clustering cluster center (a KNN algorithm may be used), and random selection may be performed.
In other embodiments, since the reference sample library stores reference samples corresponding to a plurality of wafer defects, when one wafer defect sample to be classified is matched with a reference sample, the current wafer defect pattern to be classified and at least one reference sample corresponding to each wafer defect in the reference sample library can be matched in a traversal manner, and if one reference sample is matched, step S107 is executed; of course, if no corresponding reference sample is matched after matching the reference sample with all wafer defect patterns, step S109 is executed.
S107, marking corresponding defect category identifications of all the wafer defect patterns to be classified in the wafer defect group to be classified where the current wafer defect patterns to be classified are successfully matched.
In some embodiments, since similar wafer defect patterns to be classified have been classified into one wafer defect group through unsupervised learning clustering in step S103, accordingly, when any one or more wafer defect patterns to be classified in each group are matched with the corresponding wafer defect pattern reference sample, i.e. find the corresponding defect type identifier (i.e. find the corresponding wafer defect type), all the wafer defect patterns to be classified in this group naturally correspond to the same defect type, that is, the defect types of all the wafer defect patterns to be classified in the group are the same as the defect types corresponding to the matched wafer defect pattern reference samples, and correspondingly, the same defect type identifiers are marked on all the wafer defect patterns to be classified in the group, and the defect type identifier is the defect type identifier corresponding to the wafer defect pattern reference sample matched currently. For example, a6 is randomly selected from the group 1, and is matched with each wafer defect pattern reference sample C1, C2, C3 · · in the reference sample library, and is matched with the corresponding wafer defect pattern reference sample Cm +3, so that all the wafer defect patterns to be classified in the group 1 are correspondingly marked with the second defect type identifier X2, that is, all the wafer defect patterns to be classified in the group 1 are judged as the second wafer defect; similarly, a2 is randomly selected from the group 2, and is matched with each wafer defect pattern reference sample C1, C2, C3 · · · in the reference sample library, and is matched with the corresponding wafer defect pattern reference sample Cm + p +2, so that all the wafer defect patterns to be classified in the group 2 are correspondingly determined as the third wafer defect, and the corresponding third defect type identifier X3 · · · is labeled.
In the wafer defect classification method according to the exemplary embodiment, after a plurality of wafer defect patterns to be classified are clustered through unsupervised learning clustering, the wafer defect types of all the wafer defect patterns to be classified in the corresponding group (the wafer defect types and the wafer defect types in the present document refer to the types of defects on wafers) can be identified only by selecting a corresponding number of the wafer defect patterns to be classified from each group to complete matching operation, so that corresponding calculation is not required for all the wafer defect patterns to be classified, for example, all the wafer defect patterns to be classified are input into a classifier to be classified for classification, a large amount of calculation resources and time resources are saved, and the efficiency of wafer defect type identification is improved.
S109, storing the unsuccessfully matched current wafer defect pattern to be classified as a new wafer defect pattern reference sample corresponding to a new type of wafer defects, and generating a custom type identifier corresponding to the new defect pattern reference sample according to a pre-stored defect type identifier custom mode.
In some embodiments, when the wafer defect pattern to be classified is not matched with the corresponding wafer defect pattern reference sample, it indicates that the reference sample library does not store the wafer defect pattern reference sample corresponding to the type of wafer defect, therefore, the wafer defect pattern to be classified which is not successfully matched is automatically stored as a new wafer defect pattern reference sample corresponding to a new type of wafer defect, and a customized type identifier corresponding to the new wafer defect pattern is generated according to a pre-stored defect type identifier customized manner, i.e. automatic identification of the wafer defect type is performed, so as to distinguish the wafer defect pattern from other types of wafer defects in the reference sample library, and there is no need to manually add a new wafer defect reference sample, and manually perform defect type labeling, so that the corresponding wafer defect group to be classified can be identified directly according to the new wafer defect reference sample, the efficiency of wafer defect classification is further improved. For example, the wafer defect pattern Cm + p to be classified in the group 2 that is not successfully matched is automatically added to the reference sample library to serve as a new defect pattern reference sample corresponding to a new type of wafer defect, and a pre-stored defect type identifier self-defining manner is adopted to automatically generate a self-defined type identifier Y1 corresponding to the new wafer defect pattern reference sample.
In some embodiments, in order to make the defect type identifier have universality (e.g., compatible with other systems or devices), the pre-stored defect type identifier can be a digital code for identifying a wafer defect, for example, an automatically generated digital code combining the grouping number corresponding to the to-be-classified wafer defect pattern and the wafer production lot, time, random check code, etc., and of course, the type of the wafer defect corresponding to the digital code is different from the types of other wafer defects existing in the reference sample library. Of course, other customization methods may be used, such as customization based on the common characteristics or similar characteristics of the groups of the wafer defect patterns to be classified.
Of course, in other embodiments, when the matching in step S105 is unsuccessful, and the current wafer defect pattern to be classified which is not successfully matched is used as the reference sample of the new wafer defect pattern, the user may be prompted to label the reference sample of the new wafer defect pattern with the corresponding defect type identifier by a message prompt or the like instead of generating the corresponding customized type identifier in a pre-stored customized manner of the defect type identifier, and then the corresponding customized defect type identifier is generated in response to the operation instruction input by the user. Of course, the defect type identifier generated according to the operation instruction input by the user may be customized by the user, may also be a defect type known in the industry, and may also adopt the above encoding mode.
And S110, marking the wafer defect type of the wafer defect group where the current to-be-classified wafer defect pattern which is not successful is located as the custom type identifier generated in the step S109.
In some embodiments, since the current wafer defect pattern to be classified that is not successfully matched currently is used as the reference sample of the new wafer defect pattern, and a new wafer defect type is customized for the current wafer defect pattern to be classified, as described above, Y1, the wafer defect types of all the wafer defect patterns to be classified in the current group of the wafer defect pattern to be classified can all be marked as the defect type corresponding to the reference sample of the new defect pattern, that is, the wafer defect types of all the wafer defect patterns to be classified in the group can be determined as the new type of wafer defect corresponding to the customized type identifier. For example, the automatically generated digital code Y1 (identifying a defect type) is marked for all wafer defect patterns to be classified within the group.
Example two
Fig. 2 is a schematic structural diagram of a wafer defect classification apparatus according to an exemplary embodiment of the invention. Specifically, the wafer defect classification apparatus of the present exemplary embodiment includes:
the data acquisition module 11 is used for acquiring a plurality of wafer defect patterns to be classified;
the data processing module 12 is configured to perform unsupervised learning clustering on the plurality of wafer defect patterns to be classified acquired by the data acquisition module 11 to obtain a plurality of wafer defect groups to be classified;
the defect classification module 13 is configured to match, with at least one wafer defect pattern reference sample prestored, wafer defect patterns to be classified of a preset threshold number in each group of wafer defect groups to be classified obtained from the data processing module 12, and if the matching is successful, mark corresponding defect category identifiers on all the wafer defect patterns to be classified in the wafer defect group where the current wafer defect patterns to be classified which are successfully matched are located; specifically, the marked defect type identifier is a defect type identifier corresponding to the matched wafer defect pattern reference sample.
In some embodiments, the defect pattern of the wafer to be classified is obtained by scanning each wafer through various defect detection devices, such as SEM, in advance, and accordingly, the data obtaining module 11 may directly obtain a plurality of defect patterns of the wafer to be classified from the defect detection devices in a wireless or wired manner.
In some embodiments, in order to improve the accuracy of clustering, the data processing module 12 further has a preprocessing function, and specifically, the data processing module may preprocess the wafer defect pattern to be classified by filtering, sharpening, histogram equalization, super-resolution, brightness and contrast adjustment, and the like.
In some embodiments, the data processing module 12 clusters (or groups) the acquired wafer defect patterns to be classified by using an unsupervised learning method, so as to divide the same or similar wafer defect patterns into the same group. Specifically, an unsupervised clustering model such as a clustering algorithm may be used for clustering/grouping.
In some embodiments, the defect classification module 13 specifically includes: a wafer defect pattern reference sample library (hereinafter referred to as a reference sample library) for pre-storing at least one defect pattern reference sample corresponding to each type of currently existing or known wafer defect, that is, a plurality of wafer defect pattern reference samples (hereinafter referred to as reference samples) are pre-stored in the reference sample library, and each reference sample is marked with a corresponding defect type identifier in advance; specifically, a professional may mark/label each reference sample in the reference sample library with a corresponding defect type identifier (such as a defect type name, or a code identifying various defect types), and at least one reference sample is labeled with the same defect type identifier, that is, each wafer defect corresponds to at least one wafer defect pattern reference sample. Preferably, a corresponding wafer defect pattern reference sample is stored for each wafer defect, that is, each wafer defect pattern in the reference sample library corresponds to a defect type identifier.
In some embodiments, the defect classification module 13 further includes: the image classification system comprises a model training unit, an image selection unit and an image classification unit, wherein the model training unit is used for training an image classification model according to sample references in the sample reference library; specifically, the image classification model may employ a CNN algorithm; the image selecting unit is used for selecting wafer defect patterns to be classified with a preset threshold quantity from each group of wafer defect groups to be classified; the image classification unit is used for carrying out image classification on the selected wafer defect pattern to be classified based on the trained image classification model, namely matching the selected wafer defect pattern to be classified with at least one wafer defect pattern reference sample which is prestored.
In some embodiments, each wafer defect pattern reference sample (i.e. reference sample) in the reference sample library is selected according to the following principle: an image classification model, such as a CNN algorithm, can extract classified key feature parameters (such as features and attribute information of defects themselves, such as optical signal intensity of defects in a pattern, positions of defects, sizes of defects, and the like) and weight relations thereof from the reference sample, so that the influence of different backgrounds (i.e., patterns around the defects) can be minimized, that is, the trained image classification model can classify the defect patterns of a wafer to be classified based on the defects themselves in the pattern, while neglecting surrounding patterns, that is, without being interfered by the background, for example, after the CNN learns each reference sample in the reference sample library, an image classification model capable of classifying based on the defects itself is obtained.
In some embodiments, the defect classification module 13 performs image classification based on the attribute information (such as the optical signal intensity of the defect in the pattern, the position of the defect, the size of the defect, etc.) associated with the defect pattern of the wafer to be classified. In some embodiments, after the selected wafer defect pattern to be classified is input into the defect classification module 13, the defect classification module 13 may extract the classification key features and their weight relationships in each wafer defect pattern to be classified, match the classification key features and their weight relationships with at least one reference sample (known) in the sample library, and mark the defect category identifier corresponding to the reference sample for the wafer defect pattern group where the wafer defect pattern to be currently classified is located once matching is successful.
In some embodiments, the defect classification apparatus further includes: and the self-defining module 14 is configured to, when the defect classification module 13 fails to match, automatically store the current wafer defect pattern to be classified that is not successfully matched as a new wafer defect pattern reference sample corresponding to the new type of wafer defect.
In some embodiments, the customization module 14 is further configured to generate a customized category identifier corresponding to the reference sample of the new wafer defect pattern according to a pre-stored defect category identifier customization manner. In some embodiments, the pre-stored defect type identifier may be a digital code for identifying a wafer defect, for example, a digital code automatically generated by combining the grouping number corresponding to the to-be-classified wafer defect pattern and the wafer production lot, time, random check code, etc., of course, the type of the wafer defect corresponding to the digital code is different from the types of other wafer defects existing in the reference sample library. Of course, other customization approaches may be used.
In some embodiments, the defect classification module 13 is further configured to mark all the wafer defect classes of all the wafer defect patterns to be classified in the wafer defect group to be classified where the current wafer defect pattern to be classified that is not successfully matched is located as the custom class identifier.
Correspondingly, based on the wafer defect classification device, a wafer defect classification system is provided, which comprises at least one defect detection device and the wafer defect classification device, wherein the defect detection device is used for detecting a plurality of wafers to obtain a plurality of wafer defect patterns to be classified; the wafer defect classification device is used for acquiring a plurality of wafer defect patterns to be classified from the wafer defect detection device and carrying out unsupervised learning clustering to obtain a plurality of groups of wafer defect groups to be classified; and then, matching the wafer defect patterns to be classified with a preset threshold number in each group with at least one pre-stored wafer defect pattern reference sample, and if the matching is successful, marking the corresponding defect type identification on the wafer defect group to be classified where the wafer defect patterns to be classified are located, namely marking the defect type of the wafer defect group to be classified (each wafer defect pattern to be classified) as the defect type identification of the wafer defect pattern reference sample matched currently.
EXAMPLE III
Based on the above wafer defect classification apparatus, a third aspect of the present invention further provides a wafer defect classification system, which includes:
the wafer defect detection device is used for detecting a plurality of wafers to obtain a plurality of wafer defect patterns to be classified; specifically, the wafer defect detection device can adopt equipment such as SEM and the like;
at least one wafer defect classification device in the second embodiment may perform data communication (based on wireless communication or wired communication) with the wafer defect detection device to obtain a plurality of wafer defect patterns to be classified from the wafer defect detection device, and perform unsupervised learning clustering to obtain a plurality of wafer defect groups to be classified; and then, matching the wafer defect patterns to be classified with a preset threshold number in each group with at least one pre-stored wafer defect pattern reference sample, if the matching is successful, marking corresponding defect type identifiers on the wafer defect group to be classified where the wafer defect patterns to be classified are located, and then enabling the defect type identifiers to be the defect type identifiers of the wafer defect pattern reference samples which are matched currently.
In some embodiments, each wafer defect pattern reference sample is previously marked with a defect type identifier, and at least one wafer defect pattern reference sample is marked with the same defect type identifier.
Example four
In a fourth aspect of the invention, an electronic device is provided, comprising a memory 502, a processor 501 and a computer program stored on the memory 502 and executable on the processor 501, wherein the processor 501 executes the program to implement the steps of the method as described above. For convenience of explanation, only the parts related to the embodiments of the present specification are shown, and specific technical details are not disclosed, so that reference is made to the method parts of the embodiments of the present specification. The electronic device may be any electronic device including various electronic devices, a PC computer, a network cloud server, and even a mobile phone, a tablet computer, a PDA (Personal Digital Assistant), a POS (Point of Sales), a vehicle-mounted computer, a desktop computer, and the like.
In particular, the electronic device shown in fig. 3 in connection with the solution provided by the embodiments of the present description constitutes a block diagram, and the bus 500 may comprise any number of interconnected buses and bridges linking together various circuits including one or more processors represented by the processor 501 and a memory represented by the memory 502. The bus 500 may also link together various other circuits such as peripherals, voltage regulators, power management circuits, and the like, which are well known in the art, and therefore, will not be described any further herein. A communication interface 503 provides an interface between the bus 500 and the receiver and/or transmitter 504, and the receiver and/or transmitter 504 may be a separate independent receiver or transmitter or may be the same element, such as a transceiver, providing a means for communicating with various other apparatus over a transmission medium. The processor 501 is responsible for managing the bus 500 and general processing, and the memory 502 may be used for storing data used by the processor 501 in performing operations.
Through the above description of the embodiments, those skilled in the art will readily understand that the exemplary embodiments described herein may be implemented by software, or by software in combination with necessary hardware. Therefore, the technical solution according to the embodiments of the present disclosure may be embodied in the form of a software product, which may be stored in a computer-readable storage medium (which may be a CD-ROM, a usb disk, a removable hard disk, etc.) or on a network, and includes several instructions to enable a computing device (which may be a personal computer, a server, or a network device, etc.) to execute the above method according to the embodiments of the present disclosure.
The computer readable storage medium may include a propagated data signal with readable program code embodied therein, for example, in baseband or as part of a carrier wave. Such a propagated data signal may take many forms, including, but not limited to, electro-magnetic, optical, or any suitable combination thereof. A readable storage medium may also be any readable medium that is not a readable storage medium and that can communicate, propagate, or transport a program for use by or in connection with an instruction execution system, apparatus, or device. Program code embodied on a readable storage medium may be transmitted using any appropriate medium, including but not limited to wireless, wireline, optical fiber cable, RF, etc., or any suitable combination of the foregoing.
Program code for carrying out operations for the present disclosure may be written in any combination of one or more programming languages, including an object oriented programming language such as Java, C + + or the like and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The program code may execute entirely on the user's computing device, partly on the user's device, as a stand-alone software package, partly on the user's computing device and partly on a remote computing device, or entirely on the remote computing device or server. In the case of a remote computing device, the remote computing device may be connected to the user computing device through any kind of network, including a Local Area Network (LAN) or a Wide Area Network (WAN), or may be connected to an external computing device (e.g., through the internet using an internet service provider).
The computer readable medium carries one or more programs which, when executed by a device, cause the computer readable medium to perform the functions of: obtaining a plurality of wafer defect patterns to be classified; carrying out unsupervised learning clustering on the wafer defect patterns to be classified to obtain a plurality of groups of wafer defect groups to be classified; and matching the wafer defect patterns to be classified with a preset threshold number in each group with at least one prestored defect pattern reference sample, and if the matching is successful, marking the wafer defect category of the wafer defect group where the wafer defect patterns to be classified are successfully matched as the defect category identification of the defect pattern reference sample which is matched currently.
Those skilled in the art will appreciate that the modules described above may be distributed in the apparatus according to the description of the embodiments, or may be modified accordingly in one or more apparatuses unique from the embodiments. The modules of the above embodiments may be combined into one module, or further split into multiple sub-modules.
Through the above description of the embodiments, those skilled in the art will clearly understand that the method of the above embodiments can be implemented by software plus a necessary general hardware platform, and certainly can also be implemented by hardware, but in many cases, the former is a better implementation manner. Based on such understanding, the technical solutions of the present invention may be embodied in the form of a software product, which is stored in a storage medium (such as ROM/RAM, magnetic disk, optical disk) and includes instructions for enabling a computer terminal (such as a mobile phone, a computer, a server, or a network device) to execute the method according to the embodiments of the present invention.
It should be noted that, in this document, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other like elements in a process, method, article, or apparatus that comprises the element.
While the present invention has been described with reference to the embodiments shown in the drawings, the present invention is not limited to the embodiments, which are illustrative and not restrictive, and it will be apparent to those skilled in the art that various changes and modifications can be made therein without departing from the spirit and scope of the invention as defined in the appended claims.
Claims (13)
1. A wafer defect classification method based on unsupervised learning is characterized by comprising the following steps:
obtaining a plurality of wafer defect patterns to be classified;
carrying out unsupervised learning clustering on the wafer defect patterns to be classified to obtain a plurality of groups of wafer defect groups to be classified;
matching the wafer defect patterns to be classified with a preset threshold number in each group with at least one pre-stored wafer defect pattern reference sample, and if the matching is successful, marking corresponding defect type identifications of the wafer defect groups to be classified where the wafer defect patterns to be classified are located, wherein the defect type identifications are defect type identifications of the wafer defect pattern reference samples matched currently;
each wafer defect pattern reference sample is labeled with a defect type identifier in advance, and at least one wafer defect pattern reference sample is labeled with the same defect type identifier.
2. The method of claim 1, further comprising: and if the matching is unsuccessful, automatically storing the current wafer defect pattern to be classified as a new wafer defect pattern reference sample corresponding to the new wafer defect.
3. The method of claim 2, further comprising: and generating a custom type identifier corresponding to the new wafer defect pattern reference sample according to a pre-stored defect type identifier custom mode.
4. The method of claim 3, further comprising: and marking the wafer defect category of the wafer defect group to be classified where the matching of the current wafer defect pattern to be classified is unsuccessful as the custom category identification.
5. The method of any one of claims 1 to 4, wherein the unsupervised clustering model used in unsupervised learning clustering comprises a clustering algorithm model.
6. A wafer defect classification device based on unsupervised learning is characterized by comprising:
the data acquisition module is used for acquiring a plurality of wafer defect patterns to be classified;
the data processing module is used for carrying out unsupervised learning clustering on the plurality of wafer defect patterns to be classified acquired by the data acquisition model to obtain a plurality of groups of wafer defect groups to be classified;
the defect classification module is used for matching the wafer defect patterns to be classified with a preset threshold number in each group of wafer defect groups to be classified with at least one pre-stored wafer defect pattern reference sample, and if the matching is successful, marking corresponding defect type identifications for the wafer defect pattern group where the wafer defect patterns to be classified are located, wherein the defect type identifications are the defect type identifications of the wafer defect pattern reference samples matched currently;
each wafer defect pattern reference sample is labeled with a defect type identifier in advance, and at least one wafer defect pattern reference sample is labeled with the same defect type identifier.
7. The apparatus of claim 6, further comprising: and the self-defining module is used for automatically storing the current wafer defect pattern to be classified as a new wafer defect pattern reference sample corresponding to the new wafer defect when the defect classification module is unsuccessfully matched.
8. The apparatus of claim 7, wherein the customization module is further configured to generate a custom class identifier corresponding to the reference sample of the new wafer defect pattern in a pre-stored custom way of the defect class identifier.
9. The apparatus of claim 8, wherein the defect classification module is further configured to mark the custom class identifier for a wafer defect class of a wafer defect group to be classified where the current wafer defect pattern to be classified is not successfully matched.
10. The apparatus according to any one of claims 6 to 9, wherein the unsupervised clustering model used in unsupervised learning clustering comprises a clustering algorithm model.
11. A wafer defect classification system, comprising:
the wafer defect detection device is used for detecting a plurality of wafers to obtain a plurality of wafer defect patterns to be classified;
at least one wafer defect classification device according to any one of claims 6 to 10, configured to obtain a plurality of wafer defect patterns to be classified from the wafer defect detection device, and perform unsupervised learning clustering to obtain a plurality of groups of wafer defect groups to be classified; then, matching the wafer defect patterns to be classified with a preset threshold number in each group with at least one pre-stored wafer defect pattern reference sample, and if the matching is successful, marking corresponding defect type identifications of the wafer defect group to be classified where the wafer defect patterns to be currently classified are located, wherein the defect type identifications are defect type identifications of the wafer defect pattern reference samples which are currently matched;
each wafer defect pattern reference sample is marked with a defect type identifier in advance, and at least one wafer defect pattern reference sample is marked with the same defect type identifier.
12. An electronic device comprising at least one processor, at least one memory, a communication interface, and a bus; wherein the content of the first and second substances,
the processor, the memory and the communication interface complete mutual communication through the bus;
the memory is used for storing a program for executing the method of any one of claims 1 to 5;
the processor is configured to execute programs stored in the memory.
13. A computer-readable storage medium, in which a computer program is stored which, when being executed by a processor, carries out the steps of the method of one of claims 1 to 5.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110477848.3A CN113077462A (en) | 2021-04-30 | 2021-04-30 | Wafer defect classification method, device, system, electronic equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110477848.3A CN113077462A (en) | 2021-04-30 | 2021-04-30 | Wafer defect classification method, device, system, electronic equipment and storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113077462A true CN113077462A (en) | 2021-07-06 |
Family
ID=76616083
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110477848.3A Pending CN113077462A (en) | 2021-04-30 | 2021-04-30 | Wafer defect classification method, device, system, electronic equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113077462A (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113610155A (en) * | 2021-08-06 | 2021-11-05 | 上海众壹云计算科技有限公司 | Wafer defect classification method and device based on similarity comparison model, electronic equipment and storage medium |
| CN113610801A (en) * | 2021-08-05 | 2021-11-05 | 上海众壹云计算科技有限公司 | Defect classification method and device based on minimum unit graph, electronic equipment and storage medium |
| CN115187927A (en) * | 2022-07-27 | 2022-10-14 | 上海志远生态园林工程有限公司 | Remote monitoring management method and system for viewing seat |
| WO2023095505A1 (en) * | 2021-11-24 | 2023-06-01 | 東レエンジニアリング株式会社 | Automatic defect classifier |
| CN116415660A (en) * | 2023-06-12 | 2023-07-11 | 合肥喆塔科技有限公司 | Construction and retrieval method, device and medium based on wafer defect knowledge base |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2005274157A (en) * | 2004-03-22 | 2005-10-06 | Olympus Corp | Defective image inspection device and its method |
| US20100057391A1 (en) * | 2008-08-29 | 2010-03-04 | St Pierre Eric R | Classification of spatial patterns on wafer maps |
| CN102053093A (en) * | 2010-11-08 | 2011-05-11 | 北京大学深圳研究生院 | Method for detecting surface defects of chip cut from wafer surface |
| CN109034262A (en) * | 2018-08-13 | 2018-12-18 | 东北大学 | A kind of batch processing method of X-ray orientation device defect recognition |
| CN109242033A (en) * | 2018-09-21 | 2019-01-18 | 长鑫存储技术有限公司 | Wafer defect method for classifying modes and device, storage medium, electronic equipment |
| CN112069069A (en) * | 2020-09-03 | 2020-12-11 | 平安信托有限责任公司 | Defect automatic positioning analysis method, device and readable storage medium |
| CN112200806A (en) * | 2020-12-03 | 2021-01-08 | 晶芯成(北京)科技有限公司 | Wafer defect analysis method and system |
-
2021
- 2021-04-30 CN CN202110477848.3A patent/CN113077462A/en active Pending
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2005274157A (en) * | 2004-03-22 | 2005-10-06 | Olympus Corp | Defective image inspection device and its method |
| US20100057391A1 (en) * | 2008-08-29 | 2010-03-04 | St Pierre Eric R | Classification of spatial patterns on wafer maps |
| CN102053093A (en) * | 2010-11-08 | 2011-05-11 | 北京大学深圳研究生院 | Method for detecting surface defects of chip cut from wafer surface |
| CN109034262A (en) * | 2018-08-13 | 2018-12-18 | 东北大学 | A kind of batch processing method of X-ray orientation device defect recognition |
| CN109242033A (en) * | 2018-09-21 | 2019-01-18 | 长鑫存储技术有限公司 | Wafer defect method for classifying modes and device, storage medium, electronic equipment |
| CN112069069A (en) * | 2020-09-03 | 2020-12-11 | 平安信托有限责任公司 | Defect automatic positioning analysis method, device and readable storage medium |
| CN112200806A (en) * | 2020-12-03 | 2021-01-08 | 晶芯成(北京)科技有限公司 | Wafer defect analysis method and system |
Non-Patent Citations (3)
| Title |
|---|
| 卢笑蕾;余建波;: "基于混合模型与流形调节的晶圆表面缺陷识别", 计算机集成制造系统, no. 02 * |
| 赵天绪, 马佩军, 郝跃: "划分IC缺陷团的聚类算法", 计算机学报, no. 06 * |
| 赵春东;朱奎锋;苏艳杰;陈俭永;张东洋;陈晓艳;: "一种提高晶粒表面缺陷检测准确性的方法研究", 传感技术学报, no. 07 * |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113610801A (en) * | 2021-08-05 | 2021-11-05 | 上海众壹云计算科技有限公司 | Defect classification method and device based on minimum unit graph, electronic equipment and storage medium |
| CN113610155A (en) * | 2021-08-06 | 2021-11-05 | 上海众壹云计算科技有限公司 | Wafer defect classification method and device based on similarity comparison model, electronic equipment and storage medium |
| WO2023095505A1 (en) * | 2021-11-24 | 2023-06-01 | 東レエンジニアリング株式会社 | Automatic defect classifier |
| CN115187927A (en) * | 2022-07-27 | 2022-10-14 | 上海志远生态园林工程有限公司 | Remote monitoring management method and system for viewing seat |
| CN115187927B (en) * | 2022-07-27 | 2024-03-26 | 上海志远生态园林工程有限公司 | Remote monitoring and management method and system for sightseeing seat |
| CN116415660A (en) * | 2023-06-12 | 2023-07-11 | 合肥喆塔科技有限公司 | Construction and retrieval method, device and medium based on wafer defect knowledge base |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113077462A (en) | Wafer defect classification method, device, system, electronic equipment and storage medium | |
| US10650508B2 (en) | Automatic defect classification without sampling and feature selection | |
| CN112037222B (en) | Automatic updating method and system of neural network model | |
| KR100824217B1 (en) | Hybrid invariant adaptive automatic defect classification | |
| CN113095438B (en) | Wafer defect classification method, device and system thereof, electronic equipment and storage medium | |
| US20130279794A1 (en) | Integration of automatic and manual defect classification | |
| CN110309073B (en) | Method, system and terminal for automatically detecting user interface errors of mobile application program | |
| CN108564102A (en) | Image clustering evaluation of result method and apparatus | |
| US20230029045A1 (en) | Automatic image classification and processing method based on continuous processing structure of multiple artificial intelligence model, and computer program stored in computer-readable recording medium to execute the same | |
| US20220309765A1 (en) | Method and device for optimizing object-class model based on neural network | |
| KR20200045023A (en) | Method and apparatus for recognizing vehicle number based on learning model | |
| JP2017054331A (en) | Image sorting method, method for configuring sorter, and image sorting device | |
| CN111796957A (en) | Transaction abnormal root cause analysis method and system based on application log | |
| CN113128536A (en) | Unsupervised learning method, system, computer device and readable storage medium | |
| CN110009045A (en) | The recognition methods of internet-of-things terminal and device | |
| CN110796210A (en) | Method and device for identifying label information | |
| CN114238659A (en) | Method for intelligently designing network security architecture diagram | |
| CN112819799A (en) | Target defect detection method, device, system, electronic equipment and storage medium | |
| CN113610801B (en) | Defect classification method, device, equipment and storage medium based on minimum unit | |
| CN112967248A (en) | Method, apparatus, medium, and program product for generating defect image sample | |
| CN112036169A (en) | Event recognition model optimization method, device and equipment and readable storage medium | |
| EP3748549B1 (en) | Learning device and learning method | |
| CN116188445A (en) | Product surface defect detection and positioning method and device and terminal equipment | |
| CN115984158A (en) | Defect analysis method and device, electronic equipment and computer readable storage medium | |
| CN113723555A (en) | Abnormal data detection method and device, storage medium and terminal |
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 |