Wafer surface pollution defect detection method and device

Abstract

The invention relates to the technical field of wafer detection, in particular to a wafer surface pollution defect detection method and device, and the method comprises the steps: carrying out the dark field shooting of a target wafer, obtaining a surface dark field scattering image of the wafer, traversing each pixel in the dark field image, forming an SNR estimation graph through the neighborhood setting and SNR estimation calculation, and carrying out the detection of the surface pollution defect of the wafer. Global threshold segmentation is carried out on the SNR estimation image through a double-threshold principle to obtain a low-threshold binary image and a high-threshold binary image, a connected region in the low-threshold binary image is traversed, if at least one pixel in the connected region is marked as a defect in the high-threshold binary image, all pixels in the connected region are marked as real defects, and if at least one pixel in the connected region is marked as a defect in the high-threshold binary image, all pixels in the connected region are marked as real defects. Otherwise, marking the defect as a pseudo defect. According to the invention, under the conditions of large dynamic range of background noise, small number of defect pixels, weak defect signals and the like, defects and defect contours can still be detected stably.

Description

Technical Field

[0001] The present invention belongs to the technical field of wafer detection, and particularly relates to a method and device for detecting contamination defects on the wafer surface. Background Art

[0002] The detection of contamination defects on the wafer surface is a crucial step in the semiconductor manufacturing process. The main purpose is to detect the contamination situation on the wafer surface to ensure the quality of chip manufacturing and the performance of the final product. Contamination defects include particle contamination, scratches, cracks, etc. These defects will have a significant impact on chip manufacturing and the performance of the final product.

[0003] Dark field detection is a method for detecting defects on the wafer surface. Its working principle is to rely on a beam of laser to irradiate the wafer surface from the upper side, and then use a detector to collect the scattered light and convert it into an image signal. Since the reflected light on the wafer surface is avoided, only the scattered light at the defect can reach the detector. Therefore, the defect will appear brighter, while other defect-free areas will appear darker. Due to the high sensitivity of dark field detection to scattered light, it is very effective for detecting tiny surface defects.

[0004] One major challenge in the current dark field inspection process is that the cleanliness standards and yield requirements for the wafer surface vary, resulting in a wide range of defect and dirt sizes that need to be detected. This poses high requirements for the ultimate sensitivity of the detection system and algorithm. In addition, the diversity of wafer manufacturing processes further increases the detection difficulty. Although dark field imaging technology can obtain the wafer scattering intensity picture, the background signal (Haze) in the picture varies greatly. This large dynamic range often masks the subtle defect and dirt signals on the wafer surface, which will cause great interference to accurate detection.

[0005] In an industrial production environment, the detection algorithm not only needs to have high stability, but also needs to take into account efficiency and yield standards. This undoubtedly poses more stringent requirements for the design and optimization of the algorithm. Therefore, there is an urgent need for a method for detecting contamination defects on the wafer surface that can effectively solve the problems such as complex background signals and weak defect signals in dark field detection. Summary of the Invention

[0006] In view of this, the present invention aims to provide a method for detecting contamination defects on the wafer surface. By using the principle of neighborhood SNR (signal-to-noise ratio) estimation, the background signal intensity around the current signal is accurately estimated, the background signal and the defect signal are effectively segmented and distinguished, and then a set of high and low thresholds are used for double-threshold segmentation to improve the detection stability and enhance the detection ability for subtle defects and dirt on the wafer surface.

[0007] To achieve the above object, the technical solution of the present invention is realized as follows: On the one hand, the present invention provides a method for detecting contamination defects on the surface of a wafer, including: for the SNR estimation map of the wafer dark-field image, at least two thresholds are set to segment it, pixels with gray values greater than the threshold are marked as defects, and the remaining pixels are marked as normal, forming multiple binary maps corresponding to different thresholds; Combine any two thresholds; In each threshold combination, traverse the connected regions in the binary map with the lower threshold. If there is at least one pixel in the connected region marked as a defect in the binary map with the higher threshold, then all pixels in the connected region are marked as real defects, otherwise they are marked as pseudo-defects.

[0008] Preferably, perform dark-field imaging on the target wafer to obtain the dark-field image of the wafer surface.

[0009] Preferably, the method for obtaining the SNR estimation map is as follows: Traverse each pixel on the dark-field image, and select an n×n neighborhood containing the pixel, and calculate the SNR estimation value of each pixel.

[0010] Preferably, the method for obtaining the SNR estimation map is as follows: Traverse each pixel on the dark-field image, and select an n×m neighborhood containing the pixel, and calculate the SNR estimation value of each pixel.

[0011] Preferably, the calculation formula for the SNR estimation value is: ; where, represents the SNR estimation value of the pixel, represents the gray value of the pixel in the dark-field image, represents the mean value of the background noise of the pixel neighborhood; represents the amplitude of the background noise of the pixel neighborhood, represents the maximum value of the background noise of the pixel neighborhood, represents the minimum value of the background noise of the pixel neighborhood.

[0012] Preferably, the mean value of the background noise of the pixel neighborhood takes the value of: the median gray value of all pixels in the pixel neighborhood, or the average gray value of all pixels in the pixel neighborhood, or is calculated by the quadratic std (standard deviation) mean estimation algorithm .

[0013] Preferably, the steps of calculating using the quadratic std mean estimation algorithm are as follows: Calculate the average gray value and standard deviation of all pixels in the pixel neighborhood; Pixels less than (average - Theta × standard deviation) and greater than (average + Theta × standard deviation) are removed, where Theta is a set constant; Calculate the average gray value of the remaining pixels and use this average gray value as the background noise mean of the pixel neighborhood. .

[0014] Preferably, the background noise amplitude of the pixel neighborhood takes the value of: (gray median of the pixel neighborhood - gray minimum of the pixel neighborhood) × 2, or is calculated by the 6std algorithm. .

[0015] Preferably, the step of calculating using the 6std algorithm is as follows: Calculate the gray standard deviation of all pixels in the pixel neighborhood; = 6 × gray standard deviation.

[0016] Preferably, by increasing the size of the neighborhood window, the accuracy of the SNR estimate value is improved.

[0017] Preferably, the way to expand or shrink the neighborhood window size is: The aspect ratio of the neighborhood window is fixed for proportional expansion or contraction; Or, the aspect ratio of the neighborhood window is not fixed for non-proportional expansion or contraction.

[0018] Preferably, when setting two thresholds to segment the SNR estimate map, the pixels marked as real defects in the low-threshold binary map are the defects on the wafer surface.

[0019] Preferably, when setting more than two thresholds to segment the SNR estimate map, each threshold combination marks and obtains a binary map with real defects and pseudo-defects marked; perform a pixel OR operation on all threshold combinations to obtain the final binary map, and the pixels marked as real defects in the final binary map are the defects on the wafer surface.

[0020] Preferably, in each threshold combination, the high threshold takes a value greater than or equal to 3, and the low threshold takes a value less than or equal to 2.

[0021] A wafer surface contamination defect detection device includes a wafer surface contamination defect detection module, and the wafer surface contamination defect detection module uses the wafer surface contamination defect detection method to detect contamination defects.

[0022] An electronic device includes: at least one processor; and a memory communicatively connected to the at least one processor; Among them, the memory stores instructions executable by at least one processor. When the instructions are executed by at least one processor, the at least one processor is caused to execute a method for detecting contamination defects on the wafer surface.

[0023] A computer-readable storage medium stores computer instructions. When the computer instructions are executed by a processor, the processor is caused to execute a method for detecting contamination defects on the wafer surface.

[0024] Compared with the prior art, the present invention can achieve the following beneficial effects: The present invention adopts the principle of neighborhood SNR estimation and optimally designs an SNR estimation formula that can accurately estimate the background signal intensity of each pixel neighborhood, which can effectively segment and distinguish the background signal and the defect signal. And through the double-threshold calibration method, the recognition ability of extremely weak defect intensity signals in the dark-field image is improved. Even under the conditions of small defect size and low image signal-to-noise ratio, it can accurately and stably detect the fine defects and dirt on the wafer surface. In addition, the method of the present invention has good robustness. In a complex environment with a large dynamic range of the background signal, it can still stably detect defect signals with a small pixel size and a weak signal-to-noise ratio, and stably detect the defect contour, effectively overcoming the influence of background noise on the detection result.

[0025] The method of the present invention has a certain advantage in defect detection efficiency and meets the comprehensive requirements of industrial production for the stability, efficiency, and yield standards of the detection algorithm.

[0026] In addition, the parameter values in the SNR estimation formula optimally proposed by the present invention can be adjusted to different degrees according to the actual application situation. For example in terms of the value of, the gray average value or the quadratic std average value estimation result can be used to replace the gray median value, making it more suitable for the situation where the dynamic range of the original image background is small and the requirement for the algorithm efficiency is high; the value of can also adopt the output result of the 6std algorithm, making it more suitable for the situation where the dynamic range of the original image background is small and the requirement for the algorithm efficiency is high; and the setting of the neighborhood window size can be scaled or unscaled according to factors such as calculation efficiency and defect contour type, etc., to improve the detection accuracy while not reducing the detection efficiency as much as possible.

[0027] Based on the double-threshold segmentation design, according to the defect contour type and light intensity distribution, different segmented multi-threshold logics can be changed to further improve the defect detection ability in complex scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] The drawings constituting a part of the present invention are used to provide a further understanding of the present invention. The schematic embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the drawings: Figure 1 is a flowchart of a method for detecting contamination defects on the surface of a wafer provided according to an embodiment of the present invention; Figure 2 is a schematic diagram of the principle of SNR estimation calculation provided according to an embodiment of the present invention; Figure 3 is a schematic diagram of a strip-shaped defect provided according to an embodiment of the present invention; Figure 4 is a schematic diagram of a defect on a dark-field image of a wafer provided according to an embodiment of the present invention; Figure 5 is an SNR estimation diagram of a dark-field image of a wafer provided according to an embodiment of the present invention; Figure 6 is a mask diagram provided according to an embodiment of the present invention; Figure 7 is a maker diagram provided according to an embodiment of the present invention. Detailed implementation manners

[0029] In order to make the objectives, technical solutions and advantages of the present invention clearer, the present invention will be further described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and do not constitute a limitation to the present invention. Similar elements in different embodiments are labeled with related similar element numbers. In the following embodiments, many detailed descriptions are provided to enable a better understanding of the present invention. However, those skilled in the art can easily recognize that some of the features can be omitted in different situations, or can be replaced by other elements, materials, or methods. In some cases, some operations related to the present invention are not shown or described in the specification, which is to avoid the core part of the present invention being overwhelmed by excessive descriptions. For those skilled in the art, it is not necessary to describe these related operations in detail, and they can fully understand the related operations based on the descriptions in the specification and the general technical knowledge in the art.

[0030] It should be noted that, without conflict, the embodiments and features in the embodiments of the present invention can be combined with each other to form various implementation manners. At the same time, the steps or actions in the method descriptions can also be reordered or adjusted in a manner obvious to those skilled in the art. Therefore, the various sequences in the specification and drawings are only for clearly describing a certain embodiment and do not mean that they are the necessary sequences, unless it is stated that a certain sequence must be followed.

[0031] In the description of the present invention, it should be understood that the orientation or positional relationship indicated by the terms "center", "longitudinal", "transverse", "length", "width", "thickness", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", "clockwise", "counterclockwise", etc. is based on the orientation or positional relationship shown in the drawings. It is only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be construed as a limitation to the present invention. In addition, the terms "first", "second", etc. are only used for descriptive purposes and cannot be understood as indicating or implying relative importance or implicitly specifying the quantity of the indicated technical features. Thus, the features defined with "first", "second", etc. may explicitly or implicitly include one or more of such features. In the description of the present invention, unless otherwise specified, the meaning of "a plurality of" is two or more.

[0032] In the description of the present invention, it should be noted that, unless otherwise clearly specified and limited, the terms "mounted", "connected" and "coupled" shall be understood in a broad sense. For example, it may be a fixed connection, a detachable connection or an integral connection; it may be a mechanical connection or an electrical connection; it may be directly connected or indirectly connected through an intermediate medium, and it may be the communication inside two elements. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood through specific situations.

[0033] The present invention will be described in detail below with reference to the drawings and in conjunction with embodiments.

[0034] Please refer to Figure 1 , in an embodiment of the present invention, a method for detecting contamination defects on the surface of a wafer is provided. Based on the double-threshold neighborhood SNR estimation, the defect signal and the background signal of the wafer dark-field scattering image are segmented, effectively solving the problems of complex background signals and weak defect signals in dark-field detection, and meeting the high-precision and high-efficiency requirements for detecting contamination defects on the surface of the wafer in semiconductor manufacturing. The specific detection process is as follows: First, perform dark-field imaging on the target wafer to be measured to obtain its surface dark-field scattering image, providing raw data for subsequent detection of contamination defects. Dark-field detection is a prior art, and the specific detection process is not within the protection scope of the present invention. The main principle of dark-field imaging is: irradiate the wafer with a laser from the upper side, causing scattered light at the defect location, while almost no scattered light is generated in the defect-free area, thereby forming an image on the detector with a brighter defect and a darker background, highlighting the contamination defects on the surface of the wafer and laying a foundation for subsequent signal processing and defect recognition.

[0035] By traversing each pixel in the dark-field image and performing double-threshold neighborhood SNR estimation on each pixel, the distinction between defect signals and background noise is achieved. When calculating the SNR estimation value of a pixel, a neighborhood needs to be selected first. The neighborhood refers to, when calculating the SNR estimation value of each pixel, taking the current pixel as the center and selecting a neighborhood with a window size of n×n pixels or n×m pixels. The gray values of the pixels within the neighborhood window will be used to calculate the SNR estimation value of the current pixel. The window size of the neighborhood is determined according to the actual detection requirements and defect characteristics. A larger window can cover more background information and improve the accuracy of defect detection, but it will increase the computational amount; while a smaller window has a faster calculation speed, but may be more affected by local noise. Therefore, the neighborhood can be adjusted according to experience and actual needs. The ways to expand or shrink the neighborhood window size include two types: one is to keep the aspect ratio of the neighborhood window fixed and expand or shrink it proportionally; the other is to not keep the aspect ratio of the neighborhood window fixed and expand or shrink it non-proportionally.

[0036] As a preferred embodiment, the n×n neighborhood corresponds to a square neighborhood, and the n×m neighborhood corresponds to a rectangular neighborhood. According to the shape of the defect, the specific window shape and size of the neighborhood can be selected. The rectangular neighborhood is suitable for narrow defects or long-strip defects such as slip or scratch. For example, presetting a rectangular neighborhood with a longer horizontal side than the vertical side will improve the SNR accuracy of detecting vertical defects. Although expanding the neighborhood size will result in more pixels for estimation and make the SNR estimation more accurate, after the neighborhood size is expanded, the computational amount of SNR estimation will increase significantly by multiples, and thus the time-consuming overhead will be very large. Therefore, selecting a specific-direction size expansion according to the defect shape to improve the SNR estimation accuracy has a better effect and can balance the detection ability and detection time consumption.

[0037] Take Figure 3 the situation shown as an example for illustration. Assume that the black line in the figure is a vertical defect (narrow defect or long-strip defect such as slip or scratch), and the red and blue frames are two optional rectangular neighborhoods respectively. The cross-sectional area of the red and blue frames is an optional square neighborhood. Since the blue and red frames are larger than the square frame and contain more pixels, the SNR estimation is less affected by extreme pixels and the estimation value will be more stable. Therefore, when dealing with defects with a certain shape tendency, the rectangular neighborhood can be preferred. And since the red frame will contain more defect pixels than the blue frame, the maximum background gray value of its estimation will increase, which will in turn cause the denominator of the following SNR estimation value calculation formula (1) to increase, resulting in a smaller SNR estimation value and a greater impact on the SNR estimation. Therefore, the defect SNR signal estimated by selecting the blue-frame neighborhood will be better than the defect SNR signal obtained by selecting the red-frame neighborhood. Therefore, a specific-direction expansion can be selected to improve the defect detection ability.

[0038] After the neighborhood is selected, the SNR estimation value of each pixel is calculated in combination with the following calculation formula (1) of the SNR estimation value: (1); Among them, represents the SNR estimation value of the current pixel, and its value is mainly used to reflect the intensity of the current pixel signal relative to the background noise, that is, the magnitude of the current pixel gray value relative to the neighborhood gray value, represents the signal intensity of the current pixel in the dark field image, that is, the gray value of the current pixel in the dark field image, represents the average background noise of the pixel neighborhood; represents the background noise amplitude of the pixel neighborhood, represents the maximum background noise of the pixel neighborhood, represents the minimum background noise of the pixel neighborhood, and Generally, they cannot be directly obtained, so the background noise amplitude cannot be obtained by taking the difference between the specific values of and Therefore, it is necessary to estimate as a whole through algorithm design, rather than actually obtaining and calculating and specific values.

[0039] As a preferred embodiment, the average background noise of the pixel neighborhood generally takes the median gray value of all pixels in the pixel neighborhood. The median has certain noise resistance and representativeness for the background noise. Therefore, when the background dynamic range of the dark field image is large and the detection accuracy requirements for pollution defects are high, takes the median gray value of all pixels in the neighborhood. In the case where the dynamic range of the dark field image is small and the defect detection efficiency is mainly concerned, can also take the average gray value of all pixels in the pixel neighborhood, or calculate through the quadratic std mean estimation algorithm. Among them, the specific process of calculating using the quadratic std mean estimation algorithm is as follows: First, calculate the average gray value and standard deviation of all pixels in the pixel neighborhood; then remove all pixels in the neighborhood with gray values less than (average - Theta × standard deviation), and all pixels in the neighborhood with gray values greater than (average + Theta × standard deviation), where Theta is a constant set by humans and obtained based on experience. Generally, the value range of Theta is between 1 and 3. After removing the abnormal pixel points with large deviations, the remaining pixels in the neighborhood are averaged, and this average value is assigned to .

[0040] As a preferred embodiment, the background noise amplitude of the pixel neighborhood Generally, it is taken as twice the difference between the median gray value and the minimum gray value of the neighborhood, that is: =(Median gray value of pixel neighborhood - Minimum gray value of pixel neighborhood)×2.

[0041] In the case where the dynamic range of the dark field image is small and the defect detection efficiency is mainly concerned, The value of can also be replaced by the calculation result of the 6std noise estimation algorithm. The specific process of calculating using the 6std noise international algorithm is as follows: The specific process is: Calculate the gray standard deviation of all pixels in the pixel neighborhood, and copy 6 times the gray standard deviation to , that is: =6×Gray standard deviation.

[0042] Please refer to Figure 4 , which is the dark field image obtained by dark field shooting. The boxed area is the selected field according to the defect shape, and the pixels for punctuation are the actual defect pixels (when actually detecting by the method, it is initially impossible to determine which specific pixels are the actual defect pixels). By combining the signal intensity of the pixel with the mean and amplitude of the neighborhood background noise through the selected neighborhood and the calculation formula (1) of the SNR estimate value, the signal-to-noise ratio of the current pixel is quantitatively evaluated, so as to highlight the potential defect signals that are not easily recognized, and calculate to form an SNR estimate map of the wafer dark field image as shown in Figure 5 shown.

[0043] For the SNR estimate map of the wafer dark field image, the embodiment of the present invention uses high and low double thresholds for segmentation, and then realizes defect calibration. Specifically, first, according to the actual detection requirements and empirical data, a suitable high threshold and a suitable low threshold are selected to segment the SNR estimate map respectively. Among them, the high threshold T1 is taken as: T1≥3, and the low threshold T2 is taken as: T2≤2. The high threshold is used to extract more obvious and stronger defect signals, and the low threshold is used to capture relatively weak defect signals while minimizing the interference of background noise as much as possible. Denote the binary map obtained by high threshold segmentation as shown in Figure 6 as the maker map (defect marking map), and denote the binary map obtained by low threshold segmentation as shown in Figure 7 as the mask map (defect contour mask map).

[0044] The specific global threshold segmentation process is: Segment the SNR estimation graph using a high threshold, mark the pixel points with gray values greater than the high threshold as true (i.e., defects), and the remaining pixel points as false (i.e., normal, non-defective pixels) to generate a maker binary graph. This binary graph mainly contains relatively significant defective pixels, which can be used as a reference mark for subsequent defect recognition.

[0045] Segment the SNR estimation graph using a low threshold, mark the pixel points with gray values greater than the low threshold as true (i.e., defect contour mask), and the remaining pixel points as false (i.e., normal) to generate a mask binary graph. This binary graph covers more potential defect areas, including some weak defect signals and possible pseudo-defect areas, providing more comprehensive information for subsequent connected region analysis. The combined segmentation and cross-validation of high and low thresholds effectively improve the detection ability of micro-defects.

[0046] Traverse all connected regions in the mask binary graph, and according to the marking situation in the maker graph, determine whether the connected region is a real defect or a pseudo-defect, and finally generate a detected defect binary graph. The specific identification and calibration process of defect authenticity is as follows: First, in the mask binary graph, find and extract all connected regions. Here, a connected region refers to a region composed of adjacent pixels whose initial calibration results by the low threshold are all true, and each connected region may correspond to a potential defect.

[0047] For each extracted connected region, check whether at least one pixel in the region is marked as true in the maker graph. That is, the pixel is true after calibration by both the high threshold and the low threshold. If there is an overlap between the connected region and the defect mark in the maker graph, it indicates that the connected region is likely to contain real defects. If there is no pixel in the connected region that is true in the maker graph, the connected region is likely to be a pseudo-defect caused by background noise or other interference factors.

[0048] According to the above judgment basis, mark the connected regions containing true pixels in the maker graph as real defects, and mark the connected regions where all pixels are false in the maker graph in the remaining regions as pseudo-defects. Then, reset all pixels marked as pseudo-defects in the mask graph to false, delete the pseudo-defects, and keep the pixels of real defects marked as true. Finally, a detected defect binary graph is obtained. This binary graph clearly shows the position and range contour of the contamination defects on the wafer surface.

[0049] In the above embodiments, the SNR estimation graph is segmented only with two thresholds. However, in complex cases, according to the defect contour type and light intensity distribution, it can be changed to a multi-threshold segmentation SNR estimation graph with different segments. The multi-threshold segmentation still follows the dual-threshold segmentation logic, that is, the multi-thresholds are combined in pairs in series to form multiple dual-threshold combinations. For each dual-threshold combination, the dual-threshold identification processing method in the above embodiments is used for processing. At this time, finally, defect binary graphs corresponding to multiple different threshold combinations will be detected. Then, the multiple defect binary graphs are fused to further improve the defect detection ability. In order to improve the detection ability of fine defects and contamination as much as possible, the fusion method of multiple defect binary graphs generally uses pixel OR operation on all defect binary graphs, that is, taking the union of the pixels marked as true on all defect binary graphs.

[0050] Based on the above wafer surface contamination defect detection method, an embodiment of the present invention also constructs a wafer surface contamination defect detection device. This device at least includes a wafer surface contamination defect detection module, which uses the above wafer surface contamination defect detection method to perform a full-range scan and analysis of the wafer surface, and accurately identifies various tiny contamination particles and defects. In addition, this device also has high-efficient data processing capabilities, and can quickly classify and record the detected contamination defect information, and output the detection results.

[0051] An electronic device is designed to provide computing support for the wafer surface contamination defect detection method. This electronic device is at least equipped with a high-performance processor, which is used for calculation and data processing, and can execute the above wafer surface contamination defect detection method and data analysis tasks. At the same time, this electronic device also includes a memory closely connected to the processor, and an instruction set executable by the processor is stored in the memory. When the processor runs these instructions, it can accurately execute each step in the wafer surface contamination defect detection method, from data acquisition, image processing to defect identification and classification, ensuring the efficiency and accuracy of the entire detection process.

[0052] A computer-readable storage medium stores computer instructions specifically used to mobilize the wafer surface contamination defect detection. When these computer instructions are loaded and executed by the processor, the processor drives the entire detection system to work according to the preset process and algorithm according to the wafer surface contamination defect detection method, realizing the automatic and intelligent detection of the wafer surface contamination defects.

[0053] In summary, the above description is only the preferred embodiments of this specification, and is not used to limit the protection scope of this specification. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of this specification shall be included in the protection scope of this specification.

[0054] The systems, devices, modules or units described in one or more of the above embodiments may specifically be implemented by a computer chip or an entity, or by a product with certain functions. A typical implementation device is a computer. Specifically, the computer may be, for example, a personal computer, a laptop computer, a cellular phone, a camera phone, a smart phone, a personal digital assistant, a media player, a navigation device, an email device, a game console, a tablet computer, a wearable device, or any combination of these devices.

[0055] It should also be noted that the term "comprising", "including" or any other variant thereof is intended to cover a non-exclusive inclusion, such that a process, method, commodity or device comprising a series of elements not only includes those elements but also other elements not expressly listed, or elements inherent to such process, method, commodity or device. Without further limitation, an element defined by the statement "comprising a..." does not exclude the presence of additional identical elements in the process, method, commodity or device comprising the element.

[0056] Each embodiment in this specification is described in a progressive manner, and the same or similar parts among the embodiments may be referred to each other. Each embodiment focuses on the differences from other embodiments. In particular, for system embodiments, since they are basically similar to method embodiments, the description is relatively simple, and the relevant parts may be referred to the description of the method embodiments.

[0057] The above describes specific embodiments of this specification. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims may be performed in a different order than in the embodiments and still achieve the desired result. Additionally, the processes depicted in the figures do not necessarily require the specific order or sequential order shown to achieve the desired result. In certain embodiments, multitasking and parallel processing are also possible or may be advantageous.

Claims

1. A method for detecting contamination defects on the surface of a wafer, characterized in that, Including: For the SNR estimation map of the wafer dark-field image, at least two thresholds are set to segment it. Pixels with gray values greater than the threshold are marked as defects, and the remaining pixels are marked as normal, forming multiple binary maps corresponding to different thresholds. Combine any two thresholds. In each threshold combination, traverse the connected regions in the low-threshold binary map. If there is at least one pixel in the connected region marked as a defect in the high-threshold binary map, then mark all pixels in the connected region as real defects; otherwise, mark them as pseudo-defects.

2. The method for detecting wafer surface contamination defects according to claim 1, wherein Perform dark-field imaging on the target wafer to obtain the dark-field image of the wafer surface.

3. The method for detecting wafer surface contamination defects according to claim 1, wherein The method for obtaining the SNR estimation map is as follows: Traverse each pixel on the dark-field image, and select an n×n neighborhood containing the pixel, and calculate the SNR estimation value of each pixel.

4. The method for detecting wafer surface contamination defects according to claim 1, wherein The method for obtaining the SNR estimation map is as follows: Traverse each pixel on the dark-field image, and select an n×m neighborhood containing the pixel, and calculate the SNR estimation value of each pixel.

5. The method for detecting wafer surface contamination defects according to claim 3 or 4, characterized in that, The calculation formula for the SNR estimation value is: ； Among them, represents the SNR estimated value of the pixel, represents the gray value of the pixel in the dark field image, represents the mean background noise of the pixel neighborhood; represents the amplitude of the background noise of the pixel neighborhood, represents the maximum value of the background noise of the pixel neighborhood, represents the minimum value of the background noise of the pixel neighborhood.

6. The method for detecting contamination defects on the surface of a wafer according to claim 5, characterized in that, Background noise mean of pixel neighborhood The value is: the median gray value of all pixels in the pixel neighborhood, or the average gray value of all pixels in the pixel neighborhood, or calculated by the quadratic std mean estimation algorithm .

7. The method for detecting wafer surface contamination defects according to claim 6, characterized in that, Calculate using the quadratic std mean estimation algorithm The steps are as follows: Calculate the gray average value and standard deviation of all pixels in the pixel neighborhood. Exclude pixels less than (average value - Theta × standard deviation) and pixels greater than (average value + Theta × standard deviation), where Theta is a set constant. Calculate the gray - scale average value of the remaining pixels and use this gray - scale average value as the background noise mean of the pixel neighborhood .

8. The method for detecting wafer surface contamination defects according to claim 6, wherein Background noise amplitude of the pixel neighborhood The value is: (median gray value of the pixel neighborhood - minimum gray value of the pixel neighborhood) × 2, or calculated by the 6std algorithm .

9. The method for detecting contamination defects on the surface of a wafer according to claim 8, wherein, Calculate using the 6std algorithm The steps are as follows: Calculate the gray standard deviation of all pixels in the pixel neighborhood. = 6 × standard deviation of gray level.

10. The method for detecting contamination defects on the surface of a wafer according to claim 3 or 4, characterized in that, By increasing the size of the neighborhood window, improve the accuracy of the SNR estimation value.

11. The method for detecting wafer surface contamination defects according to claim 10, wherein The expansion or contraction method of the neighborhood window size is as follows: The aspect ratio of the neighborhood window is fixed for equal-proportion expansion or contraction; Or, the aspect ratio of the neighborhood window is not fixed for non-equal-proportion expansion or contraction.

12. The method for detecting wafer surface contamination defects according to claim 1, wherein, When setting two thresholds to segment the SNR estimation map, the pixels marked as real defects in the low-threshold binary map are the defects on the wafer surface.

13. The method for detecting wafer surface contamination defects according to claim 1, characterized in that When setting more than two thresholds to segment the SNR estimation map, each threshold combination marks and obtains a binary map with real defects and pseudo-defects marked; perform pixel OR operation on all threshold combinations to obtain the final binary map, and the pixels marked as real defects in the final binary map are the defects on the wafer surface.

14. The method for detecting wafer surface contamination defects according to claim 1, wherein In each threshold combination, the high threshold value is greater than or equal to 3, and the low threshold value is less than or equal to 2.

15. A wafer surface contamination defect detection device, characterized in that, Including a wafer surface contamination defect detection module, and the wafer surface contamination defect detection module uses the wafer surface contamination defect detection method described in any one of claims 1 to 14 to detect contamination defects.

16. An electronic device, characterized in that, Including: At least one processor; And A memory communicatively connected to the at least one processor; Wherein, the memory stores instructions executable by the at least one processor, and when the instructions are executed by the at least one processor, the at least one processor executes the wafer surface contamination defect detection method described in any one of claims 1 to 14 above.

17. A computer-readable storage medium storing computer instructions, characterized in that, When the computer instructions are executed by the processor, the processor executes the wafer surface contamination defect detection method described in any one of claims 1 to 14 above.

Images