Wafer defect detection method and device, electronic equipment and storage medium
By performing anomaly detection and feature extraction on wafer surface images, and combining feature matching and multi-level filtering of noise and defect signals, the problem of poor defect detection accuracy in the downstream process of aluminum wire is solved, and efficient defect signal identification and screening is achieved.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- GUANGZHOU ZENGXIN TECH CO LTD
- Filing Date
- 2026-03-27
- Publication Date
- 2026-06-23
AI Technical Summary
In the current technology for the downstream processes of aluminum wire, the signal intensities of grain signals and defect signals are similar, resulting in poor accuracy of defect detection and difficulty in effectively distinguishing noise signals from real defect signals.
The raw image is obtained by scanning the wafer surface. Anomaly detection and differential imaging are performed to determine the abnormal signal points and their candidate detection areas. Image feature information is extracted, and feature matching and threshold control are performed by combining the standard feature information of noise signal and defect signal. NEF filtering is used for further screening to identify the real defect signal points.
It improves the accuracy of defect signal detection, reduces the probability of missed and false detections, ensures the retention of valid defect signals, and enhances production yield and cost control capabilities.
Smart Images
Figure CN122265243A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of semiconductor technology, and more specifically, to a wafer defect detection method, apparatus, electronic device, and storage medium. Background Technology
[0002] In semiconductor integrated circuit manufacturing, wafer defect detection is one of the key steps to ensure chip yield. Especially in the back-end processes of aluminum line (AL line), as device size continues to shrink and integration density continues to increase, higher demands are placed on the ability to detect minute defects. Common defect types include particle contamination and circuit bridging, which can lead to electrical short circuits, open circuits, or decreased reliability, seriously affecting product performance and yield.
[0003] Typically, in the later stages of aluminum wire interconnects, after annealing and other post-processing, grain signals are generated. Since these signals are very similar to the actual defect signals to be detected, they can interfere with the detection of defect signals.
[0004] Currently, defect detection is often performed using a method based on signal strength value filtering.
[0005] Because the signal intensities of grain signals and defect signals are quite similar, the accuracy of the above-mentioned defect detection is poor. Summary of the Invention
[0006] The purpose of this application is to address the shortcomings of the prior art by providing a wafer defect detection method, apparatus, electronic device, and storage medium to improve the detection accuracy of wafer defect signals.
[0007] To achieve the above objectives, the technical solutions adopted in the embodiments of this application are as follows: In a first aspect, embodiments of this application provide a wafer defect detection method, including: Scan the surface of the wafer to be inspected to obtain the original image of the wafer; Anomaly detection is performed on the original wafer image to identify at least one abnormal signal point, and candidate detection region image blocks corresponding to each abnormal signal point are determined based on each abnormal signal point. For each candidate detection region image block corresponding to an abnormal signal point, image features are extracted to obtain the feature information corresponding to each abnormal signal point. Based on the feature information corresponding to each abnormal signal point, the defect signal points in the wafer to be detected are identified.
[0008] Optionally, the step of performing anomaly detection on the original wafer image, identifying at least one anomalous signal point, and determining candidate detection region image blocks corresponding to each anomalous signal point, includes: Differential imaging processing is performed on the original wafer image to determine at least one abnormal signal point and the location information of each abnormal signal point; Based on the location information of each abnormal signal point, a local image region of a preset size is cropped from the original wafer image with each abnormal signal point as the center, to obtain the candidate detection region image block corresponding to each abnormal signal point.
[0009] Optionally, the step of extracting image features from the candidate detection region image blocks corresponding to each anomalous signal point to obtain feature information corresponding to each anomalous signal point includes: Based on the candidate detection region image blocks corresponding to the abnormal signal points, extract one or more of the following features corresponding to the candidate detection region image blocks: morphological features, intensity features, contrast, correlation, and texture features. The features corresponding to the candidate detection region image blocks are used as the feature information corresponding to the abnormal signal points.
[0010] Optionally, identifying defect signal points in the wafer to be inspected based on the feature information corresponding to each abnormal signal point includes: Based on the characteristic information corresponding to each abnormal signal point and the standard characteristic information of noise signals and defect signals recorded in the database, each abnormal signal point is identified and classified. Noise signal points are initially screened out from each abnormal signal point to obtain each candidate defect signal point. Each candidate defect signal point is further screened, and based on the screening results, the defect signal points in the wafer to be detected are obtained.
[0011] Optionally, based on the feature information corresponding to each abnormal signal point and the standard feature information of noise signals and defect signals recorded in the database, each abnormal signal point is identified and classified. Noise signal points are initially screened out from each abnormal signal point to obtain each candidate defect signal point, including: The feature information corresponding to each abnormal signal point is matched with the standard feature information of the noise signal and the standard feature information of the defect signal, respectively. Based on the matching results, the abnormal signal point corresponding to the noise signal is determined. From the abnormal signal points, the abnormal signal points corresponding to each noise signal are screened out to obtain each candidate defect signal point.
[0012] Optionally, the step of performing secondary screening on each candidate defect signal point and obtaining the defect signal points in the wafer to be inspected based on the screening results includes: Based on the preset signal filtering threshold and noise equivalent flux, each candidate defect signal point is screened a second time to obtain the defect signal points in the wafer to be detected.
[0013] Optionally, it also includes: Based on the defect signal point identification results, the standard feature information recorded in the database is optimized.
[0014] Secondly, embodiments of this application also provide a wafer defect detection device, including: an acquisition module, a determination module, a processing module, and an identification module; The acquisition module is used to scan the surface of the wafer to be inspected and acquire the original image of the wafer; The determining module is used to perform anomaly detection on the original wafer image, determine at least one abnormal signal point, and determine the candidate detection region image block corresponding to each abnormal signal point based on each abnormal signal point. The processing module is used to extract image features from the candidate detection region image blocks corresponding to each abnormal signal point to obtain the feature information corresponding to each abnormal signal point. The identification module is used to identify defect signal points in the wafer to be inspected based on the feature information corresponding to each abnormal signal point.
[0015] Optionally, the determining module is specifically used to perform differential imaging processing on the original wafer image to determine at least one abnormal signal point and determine the location information of each abnormal signal point; Based on the location information of each abnormal signal point, a local image region of a preset size is cropped from the original wafer image with each abnormal signal point as the center, to obtain the candidate detection region image block corresponding to each abnormal signal point.
[0016] Optionally, the processing module is specifically used to extract one or more of the following features corresponding to the candidate detection region image block corresponding to the abnormal signal point: morphological features, intensity features, contrast, correlation and texture features. The features corresponding to the candidate detection region image blocks are used as the feature information corresponding to the abnormal signal points.
[0017] Optionally, the identification module is specifically used to identify and classify each abnormal signal point based on the feature information corresponding to each abnormal signal point and the standard feature information of noise signals and defect signals recorded in the database, and to initially screen out noise signal points from each abnormal signal point to obtain each candidate defect signal point. Each candidate defect signal point is further screened, and based on the screening results, the defect signal points in the wafer to be detected are obtained.
[0018] Optionally, the identification module is specifically used to perform feature matching processing on the feature information corresponding to each abnormal signal point with the standard feature information of the noise signal and the standard feature information of the defect signal, and determine the abnormal signal point corresponding to the noise signal based on the matching result; From the abnormal signal points, the abnormal signal points corresponding to each noise signal are screened out to obtain each candidate defect signal point.
[0019] Optionally, the identification module is specifically used to perform secondary screening on each candidate defect signal point according to a preset signal screening threshold and noise equivalent flux to obtain the defect signal points in the wafer to be detected.
[0020] Optionally, the processing module is further configured to optimize the standard feature information recorded in the database based on the defect signal point identification results.
[0021] Thirdly, embodiments of this application provide an electronic device, including: a processor, a storage medium, and a bus. The storage medium stores machine-readable instructions executable by the processor. When the electronic device is running, the processor communicates with the storage medium via the bus, and the processor executes the machine-readable instructions to implement the wafer defect detection method provided in the first aspect.
[0022] Fourthly, embodiments of this application provide a computer-readable storage medium storing a computer program that, when executed by a processor, performs the wafer defect detection method as provided in the first aspect.
[0023] The beneficial effects of this application are: This application provides a wafer defect detection method, apparatus, electronic device, and storage medium, comprising: scanning the surface of a wafer to be inspected to acquire an original image of the wafer; performing anomaly detection on the original wafer image to determine at least one anomalous signal point, and determining candidate detection region image blocks corresponding to each anomalous signal point based on each anomalous signal point; performing image feature extraction on the candidate detection region image blocks corresponding to each anomalous signal point to obtain feature information corresponding to each anomalous signal point; and identifying defect signal points in the wafer to be inspected based on the feature information corresponding to each anomalous signal point. This solution determines candidate detection region image blocks corresponding to anomalous signal points through preliminary detection, thereby performing feature extraction on anomalous signal points based on the candidate detection region image blocks, and classifying defect signals and noise signals based on the extracted feature information. This can improve the accuracy of defect signal detection, ensuring the online detection rate while avoiding the loss of valid defect signals.
[0024] Secondly, based on feature information-based classification, this scheme further employs threshold control combined with NEF filtering to filter candidate defect signal points again. Through multi-level filtering, it effectively filters out possible noise signal points and retains as many valid defect signal points as possible.
[0025] This method can reduce the probability of missed or false detections and greatly improve the accuracy of defect signal point detection results. Attached Figure Description
[0026] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.
[0027] Figure 1 A schematic flowchart of the wafer defect detection method provided in the embodiments of this application. Figure 1 ; Figure 2 A schematic flowchart of the wafer defect detection method provided in the embodiments of this application. Figure 2 ; Figure 3 A schematic flowchart of the wafer defect detection method provided in the embodiments of this application. Figure 3 ; Figure 4 A schematic flowchart of the wafer defect detection method provided in the embodiments of this application. Figure 4 ; Figure 5 A schematic flowchart of the wafer defect detection method provided in the embodiments of this application. Figure 5 ; Figure 6 This is a schematic diagram illustrating a classification result provided in an embodiment of this application; Figure 7 This is a schematic diagram of a wafer defect detection device provided in an embodiment of this application; Figure 8 This is a schematic diagram of an electronic device provided in an embodiment of this application. Detailed Implementation
[0028] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. It should be understood that the accompanying drawings in this application are for illustrative and descriptive purposes only and are not intended to limit the scope of protection of this application. Furthermore, it should be understood that the schematic drawings are not drawn to scale. The flowcharts used in this application illustrate operations implemented according to some embodiments of this application. It should be understood that the operations in the flowcharts may not be implemented in sequence, and steps without logical contextual relationships may be reversed or implemented simultaneously. In addition, those skilled in the art, guided by the content of this application, may add one or more other operations to the flowcharts, or remove one or more operations from the flowcharts.
[0029] Furthermore, the described embodiments are merely some, not all, of the embodiments of this application. The components of the embodiments of this application described and illustrated herein can typically be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely to illustrate selected embodiments of the application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.
[0030] It should be noted that the term "comprising" will be used in the embodiments of this application to indicate the presence of the features declared thereafter, but does not exclude the addition of other features.
[0031] In semiconductor integrated circuit manufacturing, wafer defect detection is a crucial step in ensuring chip yield. Especially in aluminum line (AL) back-end processes, as device dimensions continue to shrink and integration density increases, the requirements for detecting minute defects become even higher. Common defect types include particle contamination, bridges, open circuits, residues, and embedded defects. These defects can lead to electrical short circuits, open circuits, or decreased reliability, severely impacting product performance and yield.
[0032] Currently, the industry commonly uses Automated Optical Inspection (AOI) equipment or electron beam inspection systems to scan wafer surface defects. These devices identify potential defect locations by imaging the wafer surface and analyzing abnormal signal points in the images, and generate a defect map for subsequent sampling review and root cause analysis.
[0033] However, in actual testing, besides identifiable signals from actual defects, the physical structural characteristics of the material itself can also trigger similar response signals. For example, after aluminum interconnects undergo annealing and other post-processing, the metal layer forms a polycrystalline structure. The grain boundaries and surface undulations appear as localized high-contrast or high-intensity signal regions in the inspection image, known as grain signals. Similarly, the hillock phenomenon in copper wires also produces similar interference signals.
[0034] Signals caused by non-defective structures are called noise defects. Although they do not represent process abnormalities, they are very similar to some real risk defects (such as tiny bridges) in terms of shape, intensity, and size, making it difficult for traditional detection methods to distinguish them effectively.
[0035] Currently, threshold-based signal filtering is commonly used: signal filtering is performed based on the signal value and a set threshold. However, since the signal values of defect signals and noise signals are quite similar, this method is difficult to distinguish, thus severely limiting the defect detection rate and detection accuracy.
[0036] Based on this, the wafer defect detection method provided in this solution first classifies signals by extracting signal features, and initially classifies them into noise signal points and defect signal points. Then, the defect signal points obtained by classification are filtered again based on threshold control and interference event filter (NEF) to finally accurately and effectively detect the real defect signal points.
[0037] Figure 1 A schematic flowchart of the wafer defect detection method provided in the embodiments of this application. Figure 1 The subject executing this method can be a machine or equipment, such as... Figure 1 As shown, the method includes: S101. Scan the surface of the wafer to be inspected to obtain the original image of the wafer.
[0038] Machine tools refer to automated equipment used to perform specific process steps or inspection tasks. They are complex systems integrating hardware, sensors, control software, and algorithms. In this solution, the machine tools are mainly responsible for acquiring images or signals from the wafer surface, and based on the detected images, executing a defect detection process to ultimately detect all real defect signal points on the wafer.
[0039] The wafer to be inspected can be placed on the equipment, and the scanning device in the equipment can perform a full scan of the surface of the wafer to be inspected to obtain the original image of the wafer.
[0040] S102. Perform anomaly detection on the original wafer image, identify at least one anomalous signal point, and determine the candidate detection region image block corresponding to each anomalous signal point based on each anomalous signal point.
[0041] By identifying and detecting anomalies in the original wafer image, at least one abnormal signal point can be determined. Of course, in practical applications, many abnormal signal points will be identified.
[0042] For each anomalous signal point, a candidate detection region image block corresponding to each anomalous signal point can be determined by image region cropping. That is, each candidate detection region image block corresponds to a "suspected defect point" initially marked by the machine equipment.
[0043] The candidate detection region image patches corresponding to all anomalous signal points constitute the patch image dataset.
[0044] The reason for extracting the corresponding candidate detection region image patch for each abnormal signal point is to extract the local abnormal structure into standardized and analyzable small image units, which facilitates subsequent feature extraction and classification processing.
[0045] S103. For the candidate detection region image blocks corresponding to each abnormal signal point, perform image feature extraction to obtain the feature information corresponding to each abnormal signal point.
[0046] In this embodiment, feature extraction can be performed based on the candidate detection region image blocks corresponding to each abnormal signal point and the features set in the feature database to obtain the feature information corresponding to each abnormal signal point.
[0047] By extracting features based on image patches of candidate detection regions, the system only needs to focus on the local context information around the anomalous signal points, avoiding interference from invalid information, thus enabling more accurate extraction of the feature information of anomalous signal points.
[0048] S104. Identify the defect signal points in the wafer to be inspected based on the feature information corresponding to each abnormal signal point.
[0049] Typically, noise signals like Grain signals and real defect signals may both be identified as anomalous signal points during initial detection. Since the signal strengths of noise signals and defect signals are quite similar, this embodiment can identify and classify anomalous signal points based on the extracted feature information, so as to identify real defect signal points from the anomalous signal points and complete the defect signal point detection of the wafer to be inspected.
[0050] Based on the detected defect signal points, adjustments to the process flow and equipment parameters during subsequent wafer manufacturing can be made to avoid the recurrence of similar defects, thereby improving the yield, cost control capabilities, and technology iteration speed of the production line.
[0051] In summary, the wafer defect detection method provided in this embodiment includes: scanning the surface of the wafer to be inspected to acquire an original image of the wafer; performing anomaly detection on the original wafer image to determine at least one anomalous signal point, and determining candidate detection region image blocks corresponding to each anomalous signal point based on each anomalous signal point; performing image feature extraction on the candidate detection region image blocks corresponding to each anomalous signal point to obtain feature information corresponding to each anomalous signal point; and identifying defect signal points in the wafer to be inspected based on the feature information corresponding to each anomalous signal point. This solution determines candidate detection region image blocks corresponding to anomalous signal points through preliminary detection, thereby performing feature extraction on anomalous signal points based on the candidate detection region image blocks, and performing defect signal and noise signal classification based on the extracted feature information. This can improve the accuracy of defect signal detection, ensuring the online detection rate while avoiding the loss of effective defect signal points.
[0052] Figure 2 A schematic flowchart of the wafer defect detection method provided in the embodiments of this application. Figure 2 Optionally, in step S102, anomaly detection is performed on the original wafer image to determine at least one anomalous signal point, and based on each anomalous signal point, candidate detection region image blocks corresponding to each anomalous signal point are determined, including: S201. Perform differential imaging processing on the original wafer image to determine at least one abnormal signal point and the location information of each abnormal signal point.
[0053] Typically, there are multiple repeating chip units (dies) on the same wafer. Each die should have a high degree of consistency. Based on the original image of the wafer, the current die can be compared pixel by pixel with the adjacent die (or mirror symmetrical die). If a significant difference occurs at a certain position, it is marked as an abnormal signal point.
[0054] Specifically, image registration can be performed first to align adjacent pixels, and then pixel-by-pixel difference operations can be performed to generate a difference image. A threshold can be set, such as a grayscale difference greater than 50. By performing connected component analysis on the difference image, the region where the abnormal signal is located can be extracted, and the center coordinates of the region can be calculated to obtain the location information of the abnormal signal point.
[0055] Alternatively, the original wafer image can be compared with the expected image generated by the design database. The expected image is a reference image generated by simulating optical proximity effect, etching deviation, etc. The original wafer image and the reference image are differentially processed to find the position that deviates from the design, which is used as the position information of the abnormal signal point.
[0056] S202. Based on the location information of each abnormal signal point, take each abnormal signal point as the center, and extract a local image region of a preset size from the original wafer image to obtain the candidate detection region image block corresponding to each abnormal signal point.
[0057] In some embodiments, the location information of the abnormal signal points can be used as the center to extract an image region of a preset size from the original wafer image as a candidate detection region image block corresponding to each abnormal signal point.
[0058] Typically, the capture window can be square, and the capture is taken from the original wafer image to preserve true texture and intensity information.
[0059] Figure 3 A schematic flowchart of the wafer defect detection method provided in the embodiments of this application. Figure 3 Optionally, in step S103, image feature extraction is performed on the candidate detection region image blocks corresponding to each abnormal signal point to obtain the feature information corresponding to each abnormal signal point, including: S301. Based on the candidate detection region image blocks corresponding to the abnormal signal points, extract one or more of the following features corresponding to the candidate detection region image blocks: morphological features, intensity features, contrast, correlation, and texture features.
[0060] Taking the feature extraction process of a candidate detection region image block corresponding to an abnormal signal point as an example, the morphological features corresponding to the abnormal signal point can be extracted. These morphological features include, but are not limited to, area, shape, and structure. Intensity features, contrast, correlation, morphology, and texture corresponding to the abnormal signal point can also be extracted.
[0061] Optionally, the area feature in the morphological features can be obtained by counting the number of pixels in the candidate detection region image patch; the shape feature can be obtained by calculating the perimeter, roundness, solidity, elongation, or major-minor axis ratio of the candidate detection region image patch. Among them, the perimeter is used to define the boundary length of the abnormal signal point; roundness is used to measure the degree to which the shape is close to a circle; and solidity is used to reflect the degree of concavity.
[0062] Intensity features can be obtained by calculating the average gray level and standard deviation of pixels within the candidate detection region image block. Intensity features are used to reflect the drastic degree of brightness change at anomalous signal points.
[0063] Contrast ratio can be calculated using local contrast ratio methods or gradient-based contrast ratio methods. Contrast ratio reflects the brightness difference between anomaly signal points and their surrounding background.
[0064] Correlation can be obtained by performing normalized cross-correlation on candidate detection region image patches and noise images. Correlation is used to evaluate the similarity between abnormal signal points and defect signals or noise signals.
[0065] Morphological features refer to the representation of a three-dimensional spatial structure, which can be approximated in a two-dimensional image using any of the following methods: profile line analysis, curvature estimation, height estimation, etc. Profile lines are analyzed by extracting a one-dimensional grayscale curve passing through the center of anomaly signal points and observing the steepness of its ascent / descent. Noise signals may exhibit spikes, while defect signals may show plateaus with abrupt edge changes. A larger curvature indicates more severe surface undulations, potentially indicating noise; a smaller curvature indicates a flatter surface, potentially indicating a defect.
[0066] Texture is used to describe local repetitive patterns in an image and is crucial for distinguishing noise signals from defect signals. It can be achieved by statistically analyzing the joint probability distribution of gray levels of two pixels spaced a certain distance apart to generate a gray-level co-occurrence matrix; or by encoding each pixel's neighborhood as an 8-bit binary number to generate a Local Binary Pattern (LBP) histogram; or by using Gabor kernels of different frequencies and orientations to convolve the image to extract highly directional texture responses.
[0067] S302. Use the features corresponding to the candidate detection region image blocks as the feature information corresponding to the abnormal signal points.
[0068] The feature information extracted from the candidate detection region image blocks corresponding to the abnormal signal points can be used as the feature information corresponding to the abnormal signal points.
[0069] Of course, in practical applications, the extracted feature information is not limited to one or more of the features listed above.
[0070] In some embodiments, historical data (historical patch image dataset) and the signal classification results corresponding to each patch image can be used to train a classification model, thereby performing feature extraction and signal classification processing on candidate detection region image blocks corresponding to abnormal signal points based on the trained classification model.
[0071] Figure 4 A schematic flowchart of the wafer defect detection method provided in the embodiments of this application. Figure 4Optionally, in step S104, based on the feature information corresponding to each abnormal signal point, defect signal points in the wafer to be inspected are identified, including: S401. Based on the characteristic information corresponding to each abnormal signal point and the standard characteristic information of noise signals and defect signals recorded in the database, each abnormal signal point is identified and classified. Noise signal points are initially screened out from each abnormal signal point to obtain each candidate defect signal point.
[0072] In some embodiments, a signal database can be pre-created, which records standard characteristic information of noise signals and standard characteristic information of defect signals. Of course, since noise signals and defect signals may be divided into many types, the database can record standard characteristic information corresponding to different noise signals and standard characteristic information of different defect signals respectively.
[0073] Therefore, based on the standard feature information recorded in the signal database, and according to the feature information corresponding to each abnormal signal point extracted above, each abnormal signal point can be preliminarily classified to filter out unwanted noise signal points and retain the required candidate defect signal points.
[0074] S402. Perform secondary screening on each candidate defect signal point, and based on the screening results, obtain the defect signal points in the wafer to be inspected.
[0075] In some embodiments, since some tiny noise signal points and defect signal points are still highly similar in terms of area, grayscale and other feature information, it is easy to misclassify when processing the classification boundary. Or, due to imaging noise, uneven illumination, etc., feature distortion will occur, which will also affect the result of feature-based classification.
[0076] Therefore, the remaining candidate defect signal points can be further screened to retain only the real defect signal points and remove as many noise signal points as possible that interfere with the defect signal points.
[0077] Figure 5 A schematic flowchart of the wafer defect detection method provided in the embodiments of this application. Figure 5 Optionally, in step S401, based on the feature information corresponding to each abnormal signal point and the standard feature information of noise signals and defect signals recorded in the database, each abnormal signal point is identified and classified, and noise signal points are initially screened out from each abnormal signal point to obtain each candidate defect signal point, including: S501. Perform feature matching processing on the feature information corresponding to each abnormal signal point with the standard feature information of the noise signal and the standard feature information of the defect signal, respectively. Based on the matching results, determine the abnormal signal point corresponding to the noise signal.
[0078] In some embodiments, feature matching can be performed between the feature information corresponding to the abnormal signal points and the standard feature information in the signal database to determine which abnormal signal points match the features of the noise signal and which abnormal signal points match the features of the defect signal, thereby identifying each abnormal signal point that matches the features of the noise signal as the point where the noise signal is located.
[0079] S502. Screen out the abnormal signal points corresponding to each noise signal from each abnormal signal point to obtain each candidate defect signal point.
[0080] Optionally, abnormal signal points identified as noise signals can be screened out from the set of abnormal signal points, and the remaining abnormal signal points can be used as candidate defect signal points.
[0081] Optionally, in step S402, a secondary screening is performed on each candidate defect signal point, and the defect signal points in the wafer to be detected are obtained based on the screening results. This includes: performing a secondary screening on each candidate defect signal point according to a preset signal screening threshold and noise equivalent flux to obtain the defect signal points in the wafer to be detected.
[0082] Optionally, secondary filtering can be implemented by combining threshold control and noise equivalent flux (NEF) filtering. Threshold control can be based on preset signal filtering thresholds, which include, but are not limited to, grayscale difference thresholds and area thresholds.
[0083] For example, when the grayscale difference of a candidate defect signal point is greater than 50, the candidate defect signal point is considered to be a noise signal and can be marked; or when the area of a candidate defect signal point is less than 0.1 μm² or greater than 2.0 μm², it is considered to be a noise signal and the candidate defect signal point is marked in the same way.
[0084] Since some noise signals and defect signals may have very similar intensity and size, for example, areas of 0.18–0.3 μm² and grayscale differences of 60, threshold alone cannot effectively distinguish them.
[0085] Therefore, the marked candidate defect signal points can be filtered again based on spatial distribution characteristics using the NEF filtering method.
[0086] NEF filtering typically employs two implementation methods: The first is neighborhood exclusion filtering, where multiple similar signals exist within a certain range around a given signal, indicating a periodic structure (such as a Grain array) and thus excluding it; isolated signals are more likely to be genuine defect signals. The second is nearest-nearest-expected-feature matching, which matches the current signal with the closest expected defect pattern in a standard template library, discarding signals with scores below a certain confidence level. Therefore, NEF filtering can re-evaluate marked candidate defect signal points to identify noise signals that are confusing the defect signal.
[0087] It is worth noting that the above-mentioned secondary filtering involves performing threshold control and NEF filtering sequentially; in some embodiments, one of the two methods can be selected for execution, or both methods can be executed simultaneously. In the simultaneous execution mode, a portion of candidate defect signal points can be filtered out first through threshold control, and then a portion of candidate defect signal points can be filtered out through NEF filtering.
[0088] The candidate defect signal points that remain are then used as the actual defect signal points.
[0089] Optionally, this method further includes: optimizing the standard feature information recorded in the database based on the defect signal point identification results.
[0090] In some embodiments, the classification results of the automated process can be compared again by manual review, so that the misjudged or missed judgment results can be fed back to the database to update the standard feature information in the database, thereby realizing the optimized feedback of the database and improving the accuracy of subsequent feature-based classification.
[0091] Figure 6 This is a schematic diagram illustrating a classification result provided in an embodiment of this application. The diagram uses the Grain signal as a noise signal and the Bridge signal as a defect signal for example. Figure 6 (a) shows a schematic diagram of the classification results after classification using feature information. Through feature classification, the classified Grain signals can be filtered out, and only the Bridge signals are retained. Figure 6 (b) shows the true defect signal retained after threshold control and NEF filtering for the retained Bridge signal.
[0092] In summary, the wafer defect detection method provided in this embodiment includes: scanning the surface of the wafer to be inspected to acquire an original image of the wafer; performing anomaly detection on the original wafer image to determine at least one anomalous signal point, and determining candidate detection region image blocks corresponding to each anomalous signal point based on each anomalous signal point; performing image feature extraction on the candidate detection region image blocks corresponding to each anomalous signal point to obtain feature information corresponding to each anomalous signal point; and identifying defect signal points in the wafer to be inspected based on the feature information corresponding to each anomalous signal point. This solution determines candidate detection region image blocks corresponding to anomalous signal points through preliminary detection, thereby performing feature extraction on anomalous signal points based on the candidate detection region image blocks, and performing defect signal and noise signal classification based on the extracted feature information. This can improve the accuracy of defect signal detection, ensuring the online detection rate while avoiding the loss of effective defect signal points.
[0093] Secondly, based on feature information-based classification, this scheme further employs threshold control combined with NEF filtering to filter candidate defect signal points again. Through multi-level filtering, it effectively filters out possible noise signal points and retains as many valid defect signal points as possible.
[0094] This method can reduce the probability of missed or false detections and greatly improve the accuracy of defect signal point detection results.
[0095] The apparatus, equipment, and storage medium used to implement the wafer defect detection method provided in this application are described below. The specific implementation process and technical effects are as described above and will not be repeated below.
[0096] Figure 7 This is a schematic diagram of a wafer defect detection device provided in an embodiment of this application. The function of this wafer defect detection device corresponds to the steps performed by the method described above. This device can be deployed in the aforementioned equipment. Figure 7 As shown, the device includes: an acquisition module 100, a determination module 200, a processing module 300, and an identification module 400; The acquisition module 100 is used to scan the surface of the wafer to be inspected and acquire the original image of the wafer; The determination module 200 is used to perform anomaly detection on the original wafer image, determine at least one abnormal signal point, and determine the candidate detection region image block corresponding to each abnormal signal point based on each abnormal signal point. The processing module 300 is used to extract image features from the candidate detection region image blocks corresponding to each abnormal signal point to obtain the feature information corresponding to each abnormal signal point. The identification module 400 is used to identify defect signal points in the wafer to be inspected based on the feature information corresponding to each abnormal signal point.
[0097] Optionally, the determining module 200 is specifically used to perform differential imaging processing on the original wafer image to determine at least one abnormal signal point and determine the position information of each abnormal signal point; Based on the location information of each abnormal signal point, a local image region of a preset size is extracted from the original wafer image with each abnormal signal point as the center, and a candidate detection region image block corresponding to each abnormal signal point is obtained.
[0098] Optionally, the processing module 300 is specifically used to extract one or more of the following features corresponding to the candidate detection region image blocks corresponding to the abnormal signal points: morphological features, intensity features, contrast, correlation and texture features. The features corresponding to the candidate detection region image blocks are used as the feature information corresponding to the abnormal signal points.
[0099] Optionally, the identification module 400 is specifically used to identify and classify each abnormal signal point based on the feature information corresponding to each abnormal signal point and the standard feature information of noise signals and defect signals recorded in the database, and to preliminarily screen out noise signal points from each abnormal signal point to obtain each candidate defect signal point. Each candidate defect signal point is screened a second time, and based on the screening results, the defect signal points in the wafer to be inspected are obtained.
[0100] Optionally, the identification module 400 is specifically used to perform feature matching processing on the feature information corresponding to each abnormal signal point with the standard feature information of the noise signal and the standard feature information of the defect signal, and determine the abnormal signal point corresponding to the noise signal based on the matching result. The abnormal signal points corresponding to each noise signal are screened out from each abnormal signal point to obtain each candidate defect signal point.
[0101] Optionally, the identification module 400 is specifically used to perform secondary screening of each candidate defect signal point according to a preset signal screening threshold and noise equivalent flux to obtain the defect signal points in the wafer to be detected.
[0102] Optionally, the processing module 300 is also used to optimize the standard feature information recorded in the database based on the defect signal point identification results.
[0103] The above-described device is used to execute the method provided in the foregoing embodiments, and its implementation principle and technical effect are similar, so they will not be described again here.
[0104] These modules can be one or more integrated circuits configured to implement the above methods, such as one or more Application Specific Integrated Circuits (ASICs), one or more digital signal processors (DSPs), or one or more Field Programmable Gate Arrays (FPGAs). Alternatively, when a module is implemented using processing element scheduler code, the processing element can be a general-purpose processor, such as a Central Processing Unit (CPU) or other processor capable of calling program code. Furthermore, these modules can be integrated together as a system-on-a-chip (SOC).
[0105] The modules described above can be connected or communicate with each other via wired or wireless connections. Wired connections can include metal cables, optical fibers, hybrid cables, or any combination thereof. Wireless connections can include connections via LAN, WAN, Bluetooth, ZigBee, or NFC, or any combination thereof. Two or more modules can be combined into a single module, and any module can be divided into two or more units. Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems and devices described above can be referred to the corresponding processes in the method embodiments, and will not be repeated here.
[0106] Figure 8 This is a schematic diagram of an electronic device provided in an embodiment of this application. The electronic device can be the aforementioned machine tool, and it has image acquisition and data processing functions.
[0107] The device includes: a processor 801 and a storage medium 802.
[0108] Storage medium 802 is used to store programs, and processor 801 calls the programs stored in storage medium 802 to execute the above method embodiments. The specific implementation and technical effects are similar, and will not be described in detail here.
[0109] The storage medium 802 stores program code, which, when executed by the processor 801, causes the processor 801 to perform various steps in the wafer defect detection method according to various exemplary embodiments of this application as described in the "Exemplary Methods" section above.
[0110] The processor 801 can be a general-purpose processor, such as a central processing unit (CPU), digital signal processor (DSP), application-specific integrated circuit (ASIC), field-programmable gate array (FPGA), or other programmable logic device, discrete gate or transistor logic device, or discrete hardware component, capable of implementing or executing the methods, steps, and logic block diagrams disclosed in the embodiments of this application. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the methods disclosed in the embodiments of this application can be directly manifested as being executed by a hardware processor, or executed by a combination of hardware and software modules within the processor.
[0111] Storage medium 802, as a non-volatile computer-readable storage medium, can be used to store non-volatile software programs, non-volatile computer-executable programs, and modules. The storage medium can include at least one type of storage medium, such as flash memory, hard disk, multimedia card, card-type storage medium, random access memory (RAM), static random access memory (SRAM), programmable read-only memory (PROM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), magnetic storage medium, magnetic disk, optical disk, etc. The storage medium is any other medium capable of carrying or storing desired program code in the form of instructions or data structures that can be accessed by a computer, but is not limited thereto. In the embodiments of this application, storage medium 802 can also be a circuit or any other device capable of implementing storage functions for storing program instructions and / or data.
[0112] Optionally, this application also provides a program product, such as a computer-readable storage medium, including a program that, when executed by a processor, performs the above-described method embodiments.
[0113] In the several embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.
[0114] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.
[0115] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or in a combination of hardware and software functional units.
[0116] The integrated units implemented as software functional units described above can be stored in a computer-readable storage medium. These software functional units, stored in a storage medium, include several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) or processor to execute some steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.
Claims
1. A method for detecting wafer defects, characterized in that, include: Scan the surface of the wafer to be inspected to obtain the original image of the wafer; Anomaly detection is performed on the original wafer image to identify at least one abnormal signal point, and candidate detection region image blocks corresponding to each abnormal signal point are determined based on each abnormal signal point. For each candidate detection region image block corresponding to an abnormal signal point, image features are extracted to obtain the feature information corresponding to each abnormal signal point. Based on the feature information corresponding to each abnormal signal point, the defect signal points in the wafer to be detected are identified.
2. The method according to claim 1, characterized in that, The step of performing anomaly detection on the original wafer image, identifying at least one anomalous signal point, and determining candidate detection region image blocks corresponding to each anomalous signal point, includes: Differential imaging processing is performed on the original wafer image to determine at least one abnormal signal point and the location information of each abnormal signal point; Based on the location information of each abnormal signal point, a local image region of a preset size is cropped from the original wafer image with each abnormal signal point as the center, to obtain the candidate detection region image block corresponding to each abnormal signal point.
3. The method according to claim 1, characterized in that, The step involves extracting image features from the candidate detection region image blocks corresponding to each anomalous signal point to obtain feature information corresponding to each anomalous signal point, including: Based on the candidate detection region image blocks corresponding to the abnormal signal points, extract one or more of the following features corresponding to the candidate detection region image blocks: morphological features, intensity features, contrast, correlation, and texture features. The features corresponding to the candidate detection region image blocks are used as the feature information corresponding to the abnormal signal points.
4. The method according to claim 1, characterized in that, The step of identifying defect signal points in the wafer to be inspected based on the feature information corresponding to each abnormal signal point includes: Based on the characteristic information corresponding to each abnormal signal point and the standard characteristic information of noise signals and defect signals recorded in the database, each abnormal signal point is identified and classified. Noise signal points are initially screened out from each abnormal signal point to obtain each candidate defect signal point. Each candidate defect signal point is further screened, and based on the screening results, the defect signal points in the wafer to be detected are obtained.
5. The method according to claim 4, characterized in that, The process involves identifying and classifying each abnormal signal point based on its corresponding feature information, as well as the standard feature information of noise signals and defect signals recorded in the database. Noise signal points are initially screened out from these abnormal signal points to obtain candidate defect signal points, including: The feature information corresponding to each abnormal signal point is matched with the standard feature information of the noise signal and the standard feature information of the defect signal, respectively. Based on the matching results, the abnormal signal point corresponding to the noise signal is determined. From the abnormal signal points, the abnormal signal points corresponding to each noise signal are screened out to obtain each candidate defect signal point.
6. The method according to claim 4, characterized in that, The process of performing secondary screening on each candidate defect signal point, and obtaining the defect signal points in the wafer to be inspected based on the screening results, includes: Based on the preset signal filtering threshold and noise equivalent flux, each candidate defect signal point is screened a second time to obtain the defect signal points in the wafer to be detected.
7. The method according to claim 4, characterized in that, Also includes: Based on the defect signal point identification results, the standard feature information recorded in the database is optimized.
8. A wafer defect detection device, characterized in that, include: The module includes an acquisition module, a determination module, a processing module, and an identification module. The acquisition module is used to scan the surface of the wafer to be inspected and acquire the original image of the wafer; The determining module is used to perform anomaly detection on the original wafer image, determine at least one abnormal signal point, and determine the candidate detection region image block corresponding to each abnormal signal point based on each abnormal signal point. The processing module is used to extract image features from the candidate detection region image blocks corresponding to each abnormal signal point to obtain the feature information corresponding to each abnormal signal point. The identification module is used to identify defect signal points in the wafer to be inspected based on the feature information corresponding to each abnormal signal point.
9. An electronic device, characterized in that, include: The device includes a processor, a storage medium, and a bus, wherein the storage medium stores program instructions executable by the processor, and when the electronic device is running, the processor communicates with the storage medium via the bus, and the processor executes the program instructions to implement the wafer defect detection method as described in any one of claims 1 to 7.
10. A computer-readable storage medium, characterized in that, The storage medium stores a computer program, which is executed by a processor to implement the wafer defect detection method as described in any one of claims 1 to 7.