Method and apparatus for determining wafer yield loss

Abstract

The application provides a wafer yield loss determination method and device, and relates to the technical field of semiconductors. The method comprises the following steps: obtaining defect distribution data and electrical test yield data corresponding to a wafer; processing the defect distribution data to obtain a first gray image; processing the electrical test yield data to obtain a second gray image; determining the similarity between the first gray image and the second gray image; and determining the yield loss of the wafer based on the similarity. Thus, the similarity between the first gray image corresponding to the defect distribution data and the second gray image corresponding to the electrical test yield data can be determined, and the yield loss of the wafer can be determined based on the similarity. Since the similarity is determined from the perspective of the images, the similarity and the determined yield loss are more accurate and reliable, and the wafer yield can be better tracked and traced, and the wafer yield can be improved.

Description

Technical Field

[0001] This application relates to the field of semiconductor technology, and in particular to a method and apparatus for determining wafer yield loss. Background Technology

[0002] In the process of integrated circuit manufacturing, there are many processes involved, and each process may result in wafer contamination, damage to the wafer surface, etc., which may lead to poor wafer yield.

[0003] In related technologies, based on the number of defects provided by various inspection stations, manual screening is first performed to identify the types and corresponding quantities of defects. Then, based on the type of defect, the number of defects is multiplied by a killratio (KR) to obtain the loss rate corresponding to that type of defect. Finally, the loss rates for each type of defect are summed and converted into the final yield loss. Determining wafer yield loss in this way requires significant manpower and is not very accurate. Therefore, improving the accuracy of wafer yield loss determination is crucial. Summary of the Invention

[0004] This application provides a method and apparatus for determining wafer yield loss, in order to solve the technical problem that the accuracy of wafer yield loss determined in the prior art is low.

[0005] According to a first aspect of this application, a method for determining wafer yield loss is provided. The method includes: acquiring defect distribution data and electrical test yield data corresponding to the wafer; processing the defect distribution data to obtain a first grayscale image; processing the electrical test yield data to obtain a second grayscale image; determining the similarity between the first grayscale image and the second grayscale image; and determining the yield loss corresponding to the wafer based on the similarity.

[0006] In some embodiments, processing the defect distribution data to obtain a first grayscale image includes: clustering the defect distribution data according to coordinates to obtain a first clustering result corresponding to the defect distribution data; determining a second clustering result corresponding to the defect distribution data based on the defect coordinates of the defect distribution data, the range of the defect within the wafer, and the background image of the defect; merging the first clustering result and the second clustering result to obtain a merged defect distribution map; and performing grayscale processing on the merged defect distribution map to obtain a first grayscale image.

[0007] In some embodiments, determining the second clustering result corresponding to the defect distribution data based on the defect coordinates, the extent of the defect within the wafer, and the background image of the defect includes: encoding the defect coordinates, the extent of the defect within the wafer, and the background image of the defect according to a mapping channel to form a color image; performing grayscale processing on the color image to obtain a third grayscale image; performing binary processing on the third grayscale image to obtain a binary image; and determining the second clustering result corresponding to the defect distribution data based on the binary image.

[0008] In some embodiments, processing the electrical test yield data to obtain a second grayscale image includes: processing the electrical test results corresponding to each grain in the electrical test yield data based on a preset mapping table to obtain a second grayscale image.

[0009] In some implementations, determining the similarity between the first grayscale image and the second grayscale image includes: inputting the first grayscale image into a first classification model to obtain a first classification result; inputting the second grayscale image into a second classification model to obtain a second classification result; and determining the similarity between the first grayscale image and the second grayscale image when the first classification result and the second classification result are the same.

[0010] In some embodiments, determining the similarity between the first grayscale image and the second grayscale image when the first classification result and the second classification result are the same includes: determining similarity parameters between the first grayscale image and the second grayscale image, wherein the similarity parameters include at least one of the following: signal-to-noise ratio, structural similarity, and mean square error; and when the similarity parameters include multiple items, fusing the similarity parameters to determine the similarity between the first grayscale image and the second grayscale image.

[0011] In some implementations, determining the yield loss corresponding to the wafer based on the similarity includes: determining that the wafer passes inspection if the similarity is less than a threshold; determining fatal defects in the wafer from the defect distribution map corresponding to the defect distribution data based on the electrical test yield map corresponding to the electrical test yield data if the similarity is greater than or equal to the threshold; and determining the yield loss corresponding to the wafer based on the fatal defects in the wafer.

[0012] In some embodiments, after inputting the second grayscale image into the second classification model to obtain the second classification result, the method further includes: if the first classification result is different from the second classification result, determining the yield loss corresponding to the wafer based on the defect distribution data and electrical test yield data corresponding to each grain in the wafer.

[0013] According to a second aspect of this application, an apparatus for determining wafer yield loss is provided, comprising: an acquisition module for acquiring defect distribution data and electrical test yield data corresponding to the wafer; a first processing module for processing the defect distribution data to obtain a first grayscale image; a second processing module for processing the electrical test yield data to obtain a second grayscale image; a first determination module for determining the similarity between the first grayscale image and the second grayscale image; and a second determination module for determining the yield loss corresponding to the wafer based on the similarity.

[0014] In some embodiments, the first processing module includes: a first processing unit, configured to cluster the defect distribution data according to coordinates to obtain a first clustering result corresponding to the defect distribution data; a first determining unit, configured to determine a second clustering result corresponding to the defect distribution data based on the defect coordinates of the defect distribution data, the range of the defect within the wafer, and the background image of the defect; a merging unit, configured to merge the first clustering result and the second clustering result to obtain a merged defect distribution map; and a second processing unit, configured to perform grayscale processing on the merged defect distribution map to obtain a first grayscale image.

[0015] In some embodiments, the first determining unit is specifically configured to: encode the defect coordinates of the defect distribution data, the range of the defect within the wafer, and the background image of the defect according to a mapping channel to form a color image; perform grayscale processing on the color image to obtain a third grayscale image; perform binary processing on the third grayscale image to obtain a binary image; and determine the second clustering result corresponding to the defect distribution data based on the binary image.

[0016] In some embodiments, the second processing module is specifically used to: process the electrical test results corresponding to each grain in the electrical test yield data based on a preset mapping table to obtain a second grayscale image.

[0017] In some embodiments, the first determining module includes: a first input unit for inputting the first grayscale image into a first classification model to obtain a first classification result; a second input unit for inputting the second grayscale image into a second classification model to obtain a second classification result; and a second determining unit for determining the similarity between the first grayscale image and the second grayscale image when the first classification result and the second classification result are the same.

[0018] In some embodiments, the second determining unit is specifically configured to: determine similarity parameters between the first grayscale image and the second grayscale image, wherein the similarity parameters include at least one of the following: signal-to-noise ratio, structural similarity, and mean square error; and, when the similarity parameters include multiple items, perform fusion processing on the similarity parameters to determine the similarity between the first grayscale image and the second grayscale image.

[0019] In some embodiments, the second determining module is specifically used to: determine that the wafer passes inspection if the similarity is less than a threshold; determine fatal defects in the wafer from the defect distribution map corresponding to the defect distribution data based on the electrical test yield map corresponding to the electrical test yield data if the similarity is greater than or equal to the threshold; and determine the yield loss corresponding to the wafer based on the fatal defects in the wafer.

[0020] In some embodiments, the second determining unit is further configured to determine the yield loss corresponding to the wafer based on the defect distribution data and electrical test yield data corresponding to each grain in the wafer when the first classification result is different from the second classification result.

[0021] According to a third aspect of this application, an electronic device is provided, comprising: a processor and a memory storing computer program instructions; the processor, when executing the computer program instructions, implements any of the above-described methods for determining wafer yield loss.

[0022] According to a fourth aspect of this application, a computer-readable storage medium is provided, characterized in that the computer-readable storage medium stores computer program instructions, which, when executed by a processor, implement any of the above-described methods for determining wafer yield loss.

[0023] In summary, the wafer yield loss determination method and apparatus provided in this application have at least the following beneficial effects: First, defect distribution data and electrical test yield data corresponding to the wafer can be acquired. Then, the defect distribution data can be processed to obtain a first grayscale image, and the electrical test yield data can be processed to obtain a second grayscale image. Next, the similarity between the first and second grayscale images is determined, and the wafer yield loss is determined based on this similarity. Thus, the similarity between the first grayscale image corresponding to the defect distribution data and the second grayscale image corresponding to the electrical test yield data can be determined, and the wafer yield loss can be determined based on this similarity. Since the similarity is determined from an image perspective, it is more comprehensive and reliable. The yield loss determined using this similarity is also more accurate and reliable. Furthermore, based on this yield loss, better traceability can be achieved, saving labor costs and improving efficiency, thus providing conditions for subsequently improving the product quality and wafer yield of wafer manufacturing. Attached Figure Description

[0024] To more clearly illustrate the specific embodiments of this application or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0025] Figure 1 A flowchart illustrating a method for determining wafer yield loss provided for embodiments of this application;

[0026] Figure 2 A schematic diagram of a second grayscale image provided for an embodiment of this application;

[0027] Figure 3 A flowchart illustrating a method for determining wafer yield loss provided for embodiments of this application;

[0028] Figure 4 A schematic diagram of a third grayscale image provided for an embodiment of this application;

[0029] Figure 5 A structural diagram of a device for determining wafer yield loss provided for an embodiment of this application;

[0030] Figure 6 This is a structural diagram of an electronic device provided as an embodiment of the present application. Detailed Implementation

[0031] To make the above and other features and advantages of this application clearer, the application is further described below with reference to the accompanying drawings. It should be understood that the specific embodiments given herein are for the purpose of explanation to those skilled in the art, and are exemplary only, not restrictive.

[0032] In the following description, numerous specific details are set forth to provide a thorough understanding of this application. However, it will be apparent to those skilled in the art that the specific details are not required to practice this application. In other instances, well-known steps or operations have not been described in detail to avoid obscuring this application.

[0033] The method for determining wafer yield loss provided in this application embodiment can be executed by the wafer yield loss determination device provided in this application embodiment, which can be configured in an electronic device.

[0034] refer to Figure 1 This application provides a method for determining wafer yield loss, the method comprising:

[0035] Step 101: Obtain the defect distribution data and electrical test yield data corresponding to the wafer.

[0036] The defect distribution data can be used to characterize the defects that may exist on the wafer. It can be obtained by a defect scanner or by any desirable means, and this application does not limit it.

[0037] The electrical test yield data can be used to characterize the electrical properties of each die in the wafer. For example, by performing electrical tests on each die in the wafer, the electrical test results of each die are the electrical test yield data of the wafer. The electrical test yield data of the wafer can be determined in any acceptable way, and this application does not limit this.

[0038] Step 102: Process the defect distribution data to obtain the first grayscale image.

[0039] In this process, the defect distribution data can be first converted into a color image, and then the color image can be processed into grayscale to obtain a first grayscale image.

[0040] For example, defect distribution data can be encoded to obtain a color image, which can then be processed to obtain a first grayscale image. For instance, the defect distribution data can be encoded according to its coordinates, shape, and contour to obtain a color image, which can then be converted to grayscale to obtain a first grayscale image, and so on. Alternatively, the coordinates of the defect distribution data, the extent of the defect distribution data within the wafer, and the background image of the defect can be mapped to the red (red, R), green (green, G), and blue (blue, B) channels, respectively, i.e., a color image can be formed through encoding, which can then be converted to grayscale to obtain a first grayscale image, and so on. This application does not limit the scope of this method.

[0041] Alternatively, the defect distribution data can be clustered in different ways to obtain corresponding clustering results, and then the clustering results can be processed to obtain the first grayscale image.

[0042] Optionally, the defect distribution data can be clustered according to coordinates to obtain the first clustering result. Then, based on the defect coordinates, the extent of the defect within the wafer, and the background image of the defect, the second clustering result can be determined. The first and second clustering results are then merged to obtain a merged defect distribution map. Finally, the merged defect distribution map is processed into grayscale to obtain the first grayscale image.

[0043] For example, a distance threshold can be pre-set, and any two defects whose distance is less than the threshold can be grouped into one category. This allows all defect data in the defect distribution data to be clustered according to their coordinates, yielding the corresponding first clustering result. Alternatively, after clustering any two defects whose distance is less than the threshold, if another defect's distance to the clustered defect is still less than the threshold, further clustering can be performed until all defects in the defect distribution data are traversed. This application does not limit this approach.

[0044] Subsequently, the coordinates of the defect distribution data, the range of the defect within the wafer, and the background image of the defect can be mapped to the red channel, green channel, and blue channel, respectively, thus forming a color image through encoding. Then, the color image is grayscaled and binarized to obtain the second clustering result. The first clustering result and the second clustering result are then merged and processed to obtain the first grayscale image, etc. This application does not limit this.

[0045] Step 103: Process the electrical test yield data to obtain a second grayscale image.

[0046] The second grayscale image can be obtained based on the electrical test results corresponding to each grain in the electrical test yield data.

[0047] Optionally, based on a preset mapping table, the electrical test results corresponding to each grain in the electrical test yield data can be processed to obtain a second grayscale image.

[0048] The preset mapping table can be used to characterize the mapping relationship between the electrical test results and grayscale values ​​of the grains. For example, the mapping table can be: electrical test result [0] corresponds to grayscale value 0, [1] corresponds to grayscale value [1, 15], [2] corresponds to grayscale value [16, 30], [3] corresponds to grayscale value [30, 45], ...,

[15] corresponds to grayscale value [210, 225]. Alternatively, the mapping table can also be: electrical test result passing corresponds to grayscale value [0], electrical test result failing corresponds to grayscale value [1], etc. This application does not limit this.

[0049] For example, if there are 1-15 grains, and electrical tests are performed on each grain, the resulting electrical test yields are: 1, 2, 2, 3, 5, 0; 6, 7, 7, 9, 10, 10, 11, 12, 14, 15. A preset mapping table can characterize the mapping relationship between the electrical test results [0, 15] and the grayscale values ​​[0, 255]. For example, the preset mapping table could be: electrical test result [0] corresponds to grayscale value 0, [1] corresponds to any grayscale value [1, 15], [2] corresponds to grayscale values ​​[16, 30], [3] corresponds to any grayscale value [30, 45], ...,

[15] corresponds to any grayscale value [210, 225]. Then, based on the electrical test results of each grain and the above mapping relationship, the grayscale value corresponding to each grain can be obtained, thus obtaining a second grayscale image. For example, Figure 2 This is a second grayscale image converted from electrical test yield data.

[0050] It should be noted that the above examples are merely illustrative and should not be construed as limiting the methods for obtaining the second grayscale image in the embodiments of this application.

[0051] It should be noted that step 102 can be executed first, followed by step 103; or step 103 can be executed first, followed by step 102; or steps 102 and 103 can be executed in parallel, etc. This application does not limit this.

[0052] Step 104: Determine the similarity between the first grayscale image and the second grayscale image.

[0053] There are several ways to determine the similarity between the first grayscale image and the second grayscale image. For example, the first grayscale image and the second grayscale image can be subtracted to obtain the fourth grayscale image. Then, the similarity between the first grayscale image and the second grayscale image can be determined by calculating the signal-to-noise ratio, structural similarity, etc. of the fourth grayscale image. This application does not limit the method.

[0054] It is understandable that, since the defect distribution data and the electrical test yield data are obtained based on different dimensions, in this embodiment of the application, the defect distribution data is processed to obtain a first grayscale image, and the electrical test yield data is processed to obtain a second grayscale image. Then, the similarity between the first grayscale image and the second grayscale image can be determined. Since the image information is fully considered in the process of determining the similarity, the determined similarity can be more accurate, comprehensive and complete, providing conditions for the subsequent determination of wafer yield loss.

[0055] Step 105: Determine the yield loss corresponding to the wafer based on similarity.

[0056] It is understandable that a lower yield loss indicates a higher wafer yield, and a higher yield loss indicates a lower wafer yield. The similarity between the first grayscale image and the second grayscale image can be positively correlated with the wafer yield loss, that is, the higher the similarity, the higher the yield loss of the corresponding wafer, etc. This application does not limit this.

[0057] Optionally, if the similarity is less than a threshold, the wafer can be determined to pass the inspection; if the similarity is greater than or equal to the threshold, the wafer can be determined to fail the inspection, resulting in a yield loss.

[0058] The threshold can be a pre-set value, such as 60%, 95%, etc., and this application does not limit it.

[0059] For example, with a threshold of 60%, if the similarity between the current first grayscale image and the second grayscale image is 80%, then the wafer can be determined to have passed the inspection, and a "pass" or "qualified" message can be issued. Alternatively, with a threshold of 95%, if the similarity between the current first grayscale image and the second grayscale image is 96%, then the wafer can be determined to have failed the inspection, and a "fail" or "unqualified" message can be issued. This application does not limit the scope of the application.

[0060] Optionally, yield loss tracking prompts can be provided based on wafer yield loss to show users the locations of dies with large yield losses or those prone to yield loss. This allows users to trace the source based on the tracking prompts, which can significantly reduce the workload of manually finding fatal defects, save labor costs, and thus improve the product quality and yield of wafer manufacturing, while also saving efficiency.

[0061] Therefore, in this embodiment, the defect distribution data can be processed first to obtain a first grayscale image, and the electrical test yield data can be processed to obtain a second grayscale image. Then, based on the first and second grayscale images, the similarity between them is determined from an image perspective. Subsequently, based on this similarity, the yield loss corresponding to the wafer is determined. Since the similarity determined from an image perspective is more comprehensive and complete, the yield loss of the wafer determined based on this comprehensive and complete similarity is also more accurate and reliable. Furthermore, based on this wafer yield loss, it is easier to trace and track the source, thus providing conditions for subsequently improving the product quality of wafer manufacturing.

[0062] In this embodiment, defect distribution data and electrical test yield data corresponding to the wafer can be acquired first. Then, the defect distribution data can be processed to obtain a first grayscale image, and the electrical test yield data can be processed to obtain a second grayscale image. Next, the similarity between the first and second grayscale images is determined, and based on this similarity, the yield loss corresponding to the wafer is determined. Thus, the similarity between the first grayscale image corresponding to the defect distribution data and the second grayscale image corresponding to the electrical test yield data can be determined, and the wafer yield loss can be determined based on this similarity. Since the similarity is determined from an image perspective, it is more comprehensive and reliable. The yield loss determined using this similarity is also more accurate and reliable. Furthermore, based on this yield loss, better traceability can be achieved, saving labor costs and improving efficiency, providing conditions for subsequently improving the product quality and wafer yield in wafer manufacturing.

[0063] like Figure 3 As shown, the method for determining the wafer yield loss may include the following steps:

[0064] Step 301: Obtain the defect distribution data and electrical test yield data corresponding to the wafer.

[0065] The defect distribution data is clustered according to coordinates to obtain the first clustering result corresponding to the defect distribution data.

[0066] In this context, a wafer can typically contain multiple grains, and the reference coordinate system used for each grain may be the same or different. In this case, the coordinates of all defects in the wafer can be transformed into unified absolute coordinates according to the grain coordinates, the coordinates within the grain, and the coordinates of the wafer center. For ease of explanation, this can be abbreviated as "defect coordinates, etc." This application does not limit this.

[0067] It is understandable that after standardizing the coordinates of all defects in the wafer, the defect coordinates can be stored in an ordered container, the distance between each defect and the defects following it can be calculated, and then defects that meet the distance threshold can be clustered into one class. If the distance between a certain defect and the clustered defects is less than the distance threshold, then that certain defect can be merged with the clustered defects. Thus, through the above clustering process, the first clustering result corresponding to the defect distribution data can be obtained, and so on. This application does not limit this.

[0068] Step 303: Based on the defect coordinates, the extent of the defect within the wafer, and the background image of the defect in the defect distribution data, determine the second clustering result corresponding to the defect distribution data.

[0069] It is understandable that, in addition to processing defects in a wafer by considering the distance between two defects, information such as the shape, location, or contour of the defects can also be taken into account to cluster the defects in the wafer from an image perspective, etc. This application does not limit this.

[0070] Optionally, the defect coordinates, the extent of the defect within the wafer, and the background image of the defect in the defect distribution data can be encoded according to the mapping channel to form a color image. Then, the color image can be processed into grayscale to obtain a third grayscale image. The third grayscale image can then be processed into binary image to obtain a binary image. Based on the binary image, the second clustering result corresponding to the defect distribution data can be determined.

[0071] The mapping channels can be R channels, G channels, and B channels. Then, the defect coordinates in the defect distribution data can be mapped to the R channel of the image, the range of the defect within the wafer can be mapped to the G channel of the image, and the background image of the defect can be mapped to the B channel of the image. Based on the RGB channels in the image, a color image is formed through encoding. Then, the color image can be processed into grayscale to obtain a third grayscale image.

[0072] For example, in such Figure 4In the schematic diagram shown, part (a) is the image formed by mapping the defect coordinates to the R channel, part (b) is the image formed by mapping the defect coordinates to the G channel, part (c) is the image formed by mapping the defect coordinates to the B channel, and part (d) is the third grayscale image formed by clustering the images shown in parts (a), (b), and (c) to obtain a color image. There are various grayscale processing methods, and this application does not limit them.

[0073] Optionally, the color image can be processed into grayscale to obtain a third grayscale image. This third grayscale image can then be downsampled, and morphological opening and closing operations can be performed on the downsampled image to merge the discrete defects. Next, by setting a grayscale threshold, the processed third grayscale image can be binarized to form a binary image. Then, based on the contours in the binary image, a contour image of a certain size can be obtained through downsampling, which is the second clustering result, etc. This application does not limit the scope of this method.

[0074] Step 304: Merge the first clustering result and the second clustering result to obtain the merged defect distribution map.

[0075] When merging the first clustering result and the second clustering result, there are multiple merging methods, such as merging based on proximity of positions, approximation of contour moments, slope of contour, intercept, etc. This application does not limit the specific methods used.

[0076] For example, if the outline of defect 1 in the first clustering result is identical in shape and position to the outline of defect 2 in the second clustering result, then they can be considered as the same defect, and either one can be retained. Alternatively, if the defects in the first clustering result and the defects in the second clustering result are located close to each other, then the defects in the first clustering result and the defects in the second clustering result can be superimposed according to their positions to obtain a merged defect distribution map, etc. This application does not impose any limitations on this.

[0077] Therefore, in this embodiment, not only can clustering be performed based on defect coordinates to obtain a first clustering result, but the defects can also be processed by combining defect coordinates, the range of defects within the wafer, and the background image of the defects in the defect distribution data to obtain a second clustering result. By comprehensively considering the detailed information of the defects and the macroscopic information such as the shape and position of the defects in the image unit, the clustering result can be made more accurate and reliable, providing conditions for improving the accuracy of wafer yield loss in the future.

[0078] Step 305: Perform grayscale processing on the merged defect distribution map to obtain the first grayscale image.

[0079] Step 306: Process the electrical test yield data to obtain a second grayscale image.

[0080] Step 307: Input the first grayscale image into the first classification model to obtain the first classification result.

[0081] The defects can be of various types, such as center, donut, edge-loc, edge-ring, loc, near-full, random, and scratch. Correspondingly, the first classification result may be one or more of these defect types; this application does not limit this.

[0082] The first classification model can be used to process the first grayscale image in the input to obtain the corresponding first classification result, that is, to obtain the defect type corresponding to the first grayscale image, etc. This application does not limit this.

[0083] Optionally, an initial classification model can be trained first to obtain a corresponding first classification model. The initial classification model can be any model capable of classification, and grayscale images corresponding to the defect distribution data can be used as the training dataset. The initial classification model is continuously trained to obtain a trained first classification model. This first classification model can then be used to process the input first grayscale image to obtain the first classification result corresponding to the first grayscale image, etc. This application does not impose any limitations on this.

[0084] Step 308: Input the second grayscale image into the second classification model to obtain the second classification result.

[0085] The second classification model can be used to process the second grayscale image in the input to obtain the corresponding second classification result, that is, to obtain the defect type corresponding to the second grayscale image, etc. This application does not limit this.

[0086] Optionally, an initial classification model can be trained first to obtain a corresponding second classification model. The initial classification model can be any model capable of classification; for example, grayscale images corresponding to electrical test yield data can be used as the training dataset. The initial classification model is continuously trained to obtain a trained second classification model. This second classification model can then be used to process the input second grayscale image to obtain the second classification result corresponding to the second grayscale image, etc. This application does not impose any limitations on this.

[0087] Step 309: If the first classification result and the second classification result are the same, determine the similarity between the first grayscale image and the second grayscale image.

[0088] The first classification result and the second classification result may be the same or different. For example, if the first classification result contains only one defect type and the second classification result also contains only one defect type, and they are identical, then the first classification result and the second classification result can be considered the same. Alternatively, if the first classification result contains multiple defect types and the second classification result also contains multiple defect types, and they have some different defect types, then the first classification result and the second classification result can be considered the same, and so on. This application does not limit this. Furthermore, any desirable method can be used to determine the similarity between the first grayscale image and the second grayscale image, and this application does not limit this.

[0089] Optionally, the similarity parameters between the first grayscale image and the second grayscale image can be determined first, and then the similarity between the first grayscale image and the second grayscale image can be determined based on the similarity parameters.

[0090] The similarity parameters may include at least one of the following: signal-to-noise ratio (SNR), structural similarity (SSIM), mean square error (MSE), etc., which are not limited in this application.

[0091] Furthermore, if the similarity parameter contains only one of the above-mentioned parameters, then the similarity is simply the numerical value of the similarity parameter. If the similarity parameter contains multiple parameters, they can be fused to determine the similarity between the first grayscale image and the second grayscale image.

[0092] For example, if we subtract the first grayscale image from the second grayscale image based on their similarity points to obtain the fourth grayscale image, we can further calculate the similarity parameters corresponding to this fourth grayscale image, such as determining the corresponding SNR, SSIM, and MSE. Then, based on the weights corresponding to SNR, SSIM, and MSE, we can fuse these similarity parameters to obtain the corresponding similarity score.

[0093] The weights corresponding to each similarity parameter can be pre-set or adjusted according to actual conditions, etc., and this application does not limit this.

[0094] It should be noted that the above examples are merely illustrative and should not be construed as limiting the methods for determining the similarity between the first grayscale image and the second grayscale image in the embodiments of this application.

[0095] Step 310: Determine the yield loss corresponding to the wafer based on similarity.

[0096] Optionally, if the similarity is less than a threshold, the wafer can be determined to pass inspection. If the similarity is greater than or equal to the threshold, fatal defects in the wafer can be identified from the defect distribution map corresponding to the defect distribution data based on the electrical test yield map corresponding to the electrical test yield data, and then the yield loss corresponding to the wafer can be determined based on the fatal defects in the wafer.

[0097] For example, if the similarity between the first grayscale image and the second grayscale image is 96% when the threshold is 93%, then it can be determined that the wafer has failed the inspection. Then, based on the electrical test yield map corresponding to the electrical test yield data, the first position of yield loss in the wafer can be determined. Next, from the defect distribution map corresponding to the defect distribution data, the second position corresponding to the first position can be determined. Based on whether a defect exists at the second position and the defect type, the fatal defect in the wafer can be determined. Then, based on the fatal defect in the wafer, the corresponding yield loss of the wafer can be determined, and so on. This application does not limit this process.

[0098] Alternatively, any desirable method can be used to determine whether there are fatal defects in the wafer.

[0099] For example, if a defect has a review image, we can further determine whether there is a pattern in the image containing the defect. If so, the defect can be identified as a fatal flaw. Alternatively, we can determine whether a defect is fatal based on its size and location in the corresponding Design Data Sheet (GDS) file. For instance, if the GDS file indicates that the defect is located at the center, it can be considered a fatal flaw. Or, if a defect does not have a corresponding review image, we can determine whether it is a fatal flaw based on whether it forms a cluster. For example, during the initial clustering process, if a defect is clustered and merged to form a cluster, it can be considered a fatal flaw.

[0100] It should be noted that the above examples are merely illustrative and should not be construed as limiting the methods for determining fatal defects in the embodiments of this application.

[0101] It is understandable that the more critical defects there are, the higher the yield loss of the wafer; the fewer critical defects there are, the lower the yield loss of the wafer. Therefore, the yield loss of the wafer can be determined based on the number of critical defects, etc., but this application does not limit this.

[0102] Optionally, after determining the yield loss corresponding to the wafer, a yield loss tracking prompt can be provided. This prompt can include the wafer yield loss, the type of critical defect, the location of the critical defect, the shape of the critical defect, and so on. Based on this prompt, users can better trace and address critical defects, significantly reducing the workload of manually searching for them, saving labor costs, and ultimately improving the product quality and yield of wafer manufacturing, while also increasing efficiency.

[0103] Step 311: If the first classification result is different from the second classification result, determine the yield loss of the wafer based on the defect distribution data and electrical test yield data corresponding to each grain in the wafer.

[0104] Understandably, when the first classification result and the second classification result are different, the first yield loss can be determined based on the defect distribution data corresponding to each die in the wafer. Then, based on the electrical yield data corresponding to each die in the wafer, the second yield loss corresponding to the wafer can be determined. Finally, the first yield loss and the second yield loss are fused together to obtain the yield loss corresponding to the wafer.

[0105] Alternatively, the defect distribution map corresponding to the defect distribution data and the electrical test yield map corresponding to the electrical test yield data can be superimposed. If, in the same grain, the defect in the defect distribution map is in the same position as the failed grain in the electrical test yield map, it is recorded as a repetition. Then, the repetition rate can be determined based on the ratio of the number of defective grains in the electrical test yield map to the number of defective grains in the defect distribution map. Finally, the yield loss corresponding to the wafer can be determined based on the repetition rate.

[0106] It should be noted that the above examples are merely illustrative and should not be construed as limiting the methods for determining the yield loss of wafers in the embodiments of this application.

[0107] In this embodiment, defect distribution data and electrical test yield data corresponding to the wafer can be obtained first. Then, the defect distribution data can be clustered according to coordinates to obtain a first clustering result. Based on the defect coordinates, the range of the defect within the wafer, and the background image of the defect in the defect distribution data, a second clustering result can be determined. The first and second clustering results can then be merged to obtain a merged defect distribution map. The merged defect distribution map is then processed into grayscale to obtain a first grayscale image. The electrical test yield data is processed to obtain a second grayscale image. The first grayscale image can then be input into a first classification model to obtain a first classification result. The second grayscale image can then be input into a second classification model to obtain a second classification result. If the first and second classification results are the same, the similarity between the first and second grayscale images is determined. Based on the similarity, the yield loss corresponding to the wafer is determined. If the first and second classification results are different, the yield loss corresponding to the wafer is determined based on the defect distribution data and electrical test yield data corresponding to each die in the wafer. Therefore, the similarity between the first grayscale image corresponding to the defect distribution data and the second grayscale image corresponding to the electrical test yield data can be determined. Based on this similarity, the wafer yield loss can be determined. Since the determination of similarity fully considers image information, the similarity is more comprehensive and reliable. The yield loss determined by using this similarity is also more accurate and reliable. Furthermore, based on the yield loss, better traceability can be achieved, providing conditions for improving the product quality and wafer yield in subsequent wafer manufacturing.

[0108] According to this application, a device for determining wafer yield loss is provided, such as... Figure 5 As shown, the device includes an acquisition module 510, a first processing module 520, a second processing module 530, a first determination module 540, and a second determination module 550.

[0109] The acquisition module 510 is used to acquire defect distribution data and electrical test yield data corresponding to the wafer; the first processing module 520 is used to process the defect distribution data to obtain a first grayscale image; the second processing module 530 is used to process the electrical test yield data to obtain a second grayscale image; the first determining module 540 is used to determine the similarity between the first grayscale image and the second grayscale image; and the second determining module 550 is used to determine the yield loss corresponding to the wafer based on the similarity.

[0110] In some embodiments, the first processing module 510 includes: a first processing unit for clustering the defect distribution data according to coordinates to obtain a first clustering result corresponding to the defect distribution data; a first determining unit for determining a second clustering result corresponding to the defect distribution data based on the defect coordinates of the defect distribution data, the range of the defect within the wafer, and the background image of the defect; a merging unit for merging the first clustering result and the second clustering result to obtain a merged defect distribution map; and a second processing unit for performing grayscale processing on the merged defect distribution map to obtain a first grayscale image.

[0111] In some embodiments, the first determining unit is specifically used to: encode the defect coordinates of the defect distribution data, the range of the defect within the wafer, and the background image of the defect according to a mapping channel to form a color image; perform grayscale processing on the color image to obtain a third grayscale image; perform binary processing on the third grayscale image to obtain a binary image; and determine the second clustering result corresponding to the defect distribution data based on the binary image.

[0112] In some embodiments, the second processing module 530 is specifically used to: process the electrical test results corresponding to each grain in the electrical test yield data based on a preset mapping table to obtain a second grayscale image.

[0113] In some embodiments, the first determining module 540 includes: a first input unit for inputting the first grayscale image into a first classification model to obtain a first classification result; a second input unit for inputting the second grayscale image into a second classification model to obtain a second classification result; and a second determining unit for determining the similarity between the first grayscale image and the second grayscale image when the first classification result and the second classification result are the same.

[0114] In some embodiments, the second determining unit is specifically used to: determine similarity parameters between the first grayscale image and the second grayscale image, wherein the similarity parameters include at least one of the following: signal-to-noise ratio, structural similarity, and mean square error; and when the similarity parameters include multiple items, perform fusion processing on the similarity parameters to determine the similarity between the first grayscale image and the second grayscale image.

[0115] In some embodiments, the second determining module 550 is specifically used to: determine that the wafer passes inspection when the similarity is less than a threshold; determine fatal defects in the wafer from the defect distribution map corresponding to the defect distribution data based on the electrical test yield map corresponding to the electrical test yield data when the similarity is greater than or equal to the threshold; and determine the yield loss corresponding to the wafer based on the fatal defects in the wafer.

[0116] In some embodiments, the second determining unit is further configured to determine the yield loss corresponding to the wafer based on the defect distribution data and electrical test yield data corresponding to each grain in the wafer when the first classification result is different from the second classification result.

[0117] The wafer yield loss determination apparatus provided in this application can first acquire defect distribution data and electrical test yield data corresponding to the wafer. Then, the defect distribution data can be processed to obtain a first grayscale image, and the electrical test yield data can be processed to obtain a second grayscale image. Next, the similarity between the first and second grayscale images is determined, and the wafer yield loss is determined based on this similarity. Thus, the similarity between the first grayscale image corresponding to the defect distribution data and the second grayscale image corresponding to the electrical test yield data can be determined, and the wafer yield loss is then determined based on this similarity. Since the similarity is determined from an image perspective, it is more comprehensive and reliable, and the yield loss determined using this similarity is more accurate and reliable. Furthermore, based on this yield loss, better traceability can be achieved, saving labor costs and improving efficiency, thus providing conditions for subsequently improving product quality and wafer yield in wafer manufacturing.

[0118] It should be understood that the specific features, operations, and details described herein with respect to the methods of this application can also be similarly applied to the apparatus and system of this application, or vice versa. Furthermore, each step of the methods of this application described above can be performed by a corresponding component or unit of the apparatus or system of this application.

[0119] It should be understood that the various modules / units of the device of this application can be implemented wholly or partially through software, hardware, firmware, or a combination thereof. Each module / unit can be embedded in the processor of the electronic device in hardware or firmware form or independent of the processor, or it can be stored in the memory of the electronic device in software form for the processor to call to execute the operation of each module / unit. Each module / unit can be implemented as an independent component or module, or two or more modules / units can be implemented as a single component or module.

[0120] like Figure 6As shown, this application provides an electronic device 600, which includes a processor 601 and a memory 602 storing computer program instructions. The processor 601 executes the computer program instructions to implement the steps of the aforementioned method for determining wafer yield loss. This electronic device 600 can be broadly categorized as a server, terminal, or any other electronic device with the necessary computing and / or processing capabilities.

[0121] In one embodiment, the electronic device 600 may include a processor, memory, network interface, communication interface, etc., connected via a system bus. The processor of the electronic device 600 can be used to provide necessary computing, processing, and / or control capabilities. The memory of the electronic device 600 may include non-volatile storage media and internal memory. The non-volatile storage media may store an operating system, computer programs, etc. The internal memory can provide an environment for the operation of the operating system and computer programs in the non-volatile storage media. The network interface and communication interface of the electronic device 600 can be used to connect and communicate with external devices via a network. When the computer program is executed by the processor, it performs the steps of the method of this application.

[0122] This application provides a computer-readable storage medium storing computer program instructions, which, when executed by a processor, implement the aforementioned method for determining wafer yield loss.

[0123] Those skilled in the art will understand that the method steps of this application can be performed by a computer program instructing related hardware, such as electronic device 600 or a processor. The computer program can be stored in a non-transitory computer-readable storage medium, and its execution causes the steps of this application to be performed. Depending on the context, any reference herein to memory, storage, or other media may include non-volatile or volatile memory. Examples of non-volatile memory include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), flash memory, magnetic tape, floppy disk, magneto-optical data storage device, optical data storage device, hard disk, solid-state drive, etc. Examples of volatile memory include random access memory (RAM), external cache memory, etc.

[0124] The technical features described above can be combined arbitrarily. Although not all possible combinations of these technical features are described, any combination of these technical features should be considered to be covered by this specification, provided that such combination does not contain contradictions.

[0125] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of this application.

Claims

1. A method for determining wafer yield loss, characterized in that, include: Obtain defect distribution data and electrical test yield data for the corresponding wafer; The defect distribution data and the electrical test yield data are obtained based on different dimensions; The defect distribution data is clustered to obtain a first grayscale image; The electrical test yield data is processed to map a second grayscale image. Determine the similarity between the first grayscale image and the second grayscale image; Based on the similarity, the yield loss corresponding to the wafer is determined; The step of clustering the defect distribution data to map it into a first grayscale image includes: The defect distribution data is clustered according to coordinates to obtain the first clustering result corresponding to the defect distribution data; Based on the defect coordinates of the defect distribution data, the extent of the defect within the wafer, and the background image of the defect, a second clustering result corresponding to the defect distribution data is determined. The first clustering result and the second clustering result are merged to obtain a merged defect distribution map; The merged defect distribution map is processed into grayscale to obtain a first grayscale image.

2. The method for determining wafer yield loss as described in claim 1, characterized in that, The determination of the second clustering result corresponding to the defect distribution data based on the defect coordinates, the extent of the defect within the wafer, and the background image of the defect includes: The defect coordinates, the extent of the defect within the wafer, and the background image of the defect in the defect distribution data are encoded according to the mapping channel to form a color image; The color image is processed into grayscale to obtain a third grayscale image; The third grayscale image is subjected to binary processing to obtain a binary image; Based on the binary image, the second clustering result corresponding to the defect distribution data is determined.

3. The method for determining wafer yield loss as described in claim 1, characterized in that, The step of processing the electrical test yield data to map it into a second grayscale image includes: Based on a preset mapping table, the electrical test results corresponding to each grain in the electrical test yield data are processed to map and obtain a second grayscale image.

4. The method for determining wafer yield loss as described in claim 1, characterized in that, Determining the similarity between the first grayscale image and the second grayscale image includes: The first grayscale image is input into the first classification model to obtain the first classification result; The second grayscale image is input into the second classification model to obtain the second classification result; If the first classification result is the same as the second classification result, the similarity between the first grayscale image and the second grayscale image is determined.

5. The method for determining wafer yield loss as described in claim 4, characterized in that, Determining the similarity between the first grayscale image and the second grayscale image when the first classification result and the second classification result are the same includes: Determine similarity parameters between the first grayscale image and the second grayscale image, wherein the similarity parameters include at least one of the following: signal-to-noise ratio, structural similarity, and mean square error; When the similarity parameters include multiple parameters, the similarity parameters are fused to determine the similarity between the first grayscale image and the second grayscale image.

6. The method for determining wafer yield loss as described in claim 4, characterized in that, The determination of the yield loss corresponding to the wafer based on the similarity includes: If the similarity is less than a threshold, the wafer is determined to pass the inspection. If the similarity is greater than or equal to the threshold, based on the electrical test yield map corresponding to the electrical test yield data, the fatal defects in the wafer are determined from the defect distribution map corresponding to the defect distribution data; Based on the fatal defects in the wafer, determine the yield loss corresponding to the wafer.

7. The method for determining wafer yield loss as described in claim 4, characterized in that, After inputting the second grayscale image into the second classification model to obtain the second classification result, the method further includes: If the first classification result differs from the second classification result, the yield loss corresponding to the wafer is determined based on the defect distribution data and electrical test yield data corresponding to each grain in the wafer.

8. A device for determining wafer yield loss, characterized in that, The device includes: The acquisition module is used to acquire defect distribution data and electrical test yield data corresponding to the wafer; the defect distribution data and the electrical test yield data are obtained based on different dimensions; The first processing module is used to cluster the defect distribution data to map it into a first grayscale image; The second processing module is used to process the electrical test yield data to map it into a second grayscale image. The first determining module is used to determine the similarity between the first grayscale image and the second grayscale image; The second determining module is used to determine the yield loss corresponding to the wafer based on the similarity. The first processing module includes: a first processing unit for clustering the defect distribution data according to coordinates to obtain a first clustering result corresponding to the defect distribution data; a first determining unit for determining a second clustering result corresponding to the defect distribution data based on the defect coordinates of the defect distribution data, the range of the defect within the wafer, and the background image of the defect; a merging unit for merging the first clustering result and the second clustering result to obtain a merged defect distribution map; and a second processing unit for performing grayscale processing on the merged defect distribution map to obtain a first grayscale image.

9. An electronic device, characterized in that, The electronic device includes: a processor and a memory storing computer program instructions; When the processor executes the computer program instructions, it implements the method for determining wafer yield loss as described in any one of claims 1-7.

Images