Defect classification device

Abstract

The present invention accurately classifies patterns of defects occurring in an object to be inspected.　Specifically, a defect classification device 1 includes a classification unit 10 for classifying patterns of defect X occurring in an object W to be inspected that has been imaged so as to be divided into a plurality of visual fields F. The plurality of visual fields include a first visual field F1 and a second visual field F2 adjacent to each other. A first end part F1p on the second visual field side of the first visual field and a second end part F2p on the first visual field side of the second visual field overlap each other. The first visual field includes a first defect X1. The second visual field includes a second defect X2. The classification unit cuts out a region in which the first defect appears in the first visual field as a first defect image C1, and cuts out a region in which the second defect appears in the second visual field as a second defect image C2. When the first defect image includes the first end part and the second defect image includes the second end part, the classification unit classifies the patterns of the defects in a state in which the first defect image and the second defect image are aligned such that the first defect and the second defect overlap.

Description

Defect Classification Device

[0001] The present disclosure relates to a defect classification device.

[0002] For example, in the manufacturing process of semiconductor devices, numerous circuit patterns are formed on the surface of a semiconductor wafer (substrate) by performing processes such as exposure, development, and etching on the surface of the semiconductor wafer. Then, the semiconductor wafer on which numerous circuit patterns have been formed is diced into multiple chip components. The diced chip components are packaged and then shipped as individual electronic components or incorporated into electrical products.

[0003] Before the semiconductor wafer is divided into individual chip components, defects on the surface of the semiconductor wafer are detected based on an inspection image captured by a camera connected to an optical microscope or a scanning electron microscope, for example.

[0004] Then, for semiconductor wafers on which defects have been detected, defect patterns are classified based on feature quantities such as the shape, size, color, etc. Defect patterns include, for example, foreign matter and scratches.

[0005] Conventionally, defect patterns have been classified by visual inspection by an operator. However, in visual inspection, the classification results of defect patterns can vary depending on the subjective judgment of the operator, and therefore it has not been possible to classify defect patterns with high accuracy.

[0006] Therefore, as disclosed in Japanese Patent Laid-Open No. 2003-144999, an automatic defect classification (ADC) device is known that automatically classifies defect patterns occurring on the surface of a semiconductor wafer.

[0007] Japanese Patent Application Laid-Open No. 2004-47939

[0008] However, even if the automatic defect classification device according to Patent Document 1 is used to classify defect patterns without relying on the subjectivity of the operator, there is still a problem with the accuracy of defect pattern classification.

[0009] The semiconductor wafer to be inspected is usually larger than the field of view (imaging range) of the camera. Therefore, the semiconductor wafer is divided into multiple fields of view and images are captured by the camera, thereby obtaining inspection images for each of the multiple fields of view. In this case, defects that occur on the semiconductor wafer may span multiple fields of view.

[0010] For this reason, when defect patterns are classified based on only one field of view, the defect patterns may not be classified accurately. For example, if only a small part of a defect is captured at the edge of the one field of view, the original feature quantities (shape, size, color, etc.) of the partially captured defect cannot be fully grasped, and therefore the defect pattern cannot be accurately grasped.

[0011] The above problem also applies to objects to be inspected other than semiconductor wafers.

[0012] The present disclosure has been made in view of the above points, and an object thereof is to accurately classify patterns of defects occurring in an object to be inspected.

[0013] A defect classification device according to the present disclosure includes a classification unit that classifies a pattern of defects that have occurred in an object to be inspected, the image of which has been divided into a plurality of fields of view, the plurality of fields of view including a first field of view and a second field of view that are adjacent to each other, a first end portion of the first field of view that is the end portion of the second field of view that is the end portion of the second field of view that is the end portion of the first ... second field of view that is the end portion of the first field of view that is the end portion of the first field of view that is the end portion of the first field of view that is the end portion of the first field of view that

[0014] According to this embodiment, the first defect image extracted from the first field of view and the second defect image extracted from the second field of view are combined so that the first defects and the second defects overlap each other, and the defect patterns are classified in a combined state of the first defect image and the second defect image.

[0015] Even if a defect that has occurred in the object to be inspected straddles both the first field of view and the second field of view, the original feature amount of the defect can be grasped.

[0016] As described above, the patterns of defects occurring in the object to be inspected can be classified with high accuracy.

[0017] In one embodiment, the classification unit extracts an area in the first defect image that includes the first end as a matching image, and sets an area in the second defect image that includes the second end as a search range, and the classification unit extracts multiple target images that correspond to the matching images from the search range, and the classification unit aligns the first defect image and the second defect image so as to overlay the matching image on the target image that is closest to the matching image among the multiple target images.

[0018] According to this configuration, the first defect and the second defect can be more accurately superimposed by aligning the first defect image and the second defect image so that the matching image is superimposed on the target image that is closest to the matching image among the multiple target images T. This is advantageous in accurately classifying defect patterns that have occurred on the object under inspection.

[0019] In one embodiment, the classification unit determines the features of the matching image and each of the plurality of target images, compares the features of the matching image with the features of each of the plurality of target images, and overlays the matching image on the target image among the plurality of target images that has features closest to the features of the matching image.

[0020] According to this configuration, by comparing the feature amounts of the matching image with the feature amounts of each of the multiple target images, it is possible to more accurately determine the target image that is closest to the matching image from among the multiple target images.

[0021] In one embodiment, the classification unit aligns the first defect image and the second defect image in an adjacent direction in which the first field of view and the second field of view are adjacent, and in an intersecting direction intersecting the adjacent direction.

[0022] According to this configuration, the first defect image and the second defect image can be suitably aligned in at least two directions, that is, the adjacent direction and the intersecting direction.

[0023] According to the present disclosure, it is possible to accurately classify patterns of defects that occur in an object to be inspected.

[0024] Fig. 1 shows a schematic diagram of a defect classification method using a defect classification device, Fig. 2 shows a combination of a first defect image and a second defect image, and Fig. 3 shows a superposition of a first defect and a second defect.

[0025]

[0023] An embodiment of the present disclosure will be described in detail below with reference to the accompanying drawings. The following description of the preferred embodiment is merely exemplary in nature and is not intended to limit the present disclosure, its application, or its uses.

[0026] (Defect Classification) Fig. 1 schematically shows a defect classification method using a defect classification device 1. The defect classification device 1 is an automatic defect classification device (ADC). The defect classification device 1 includes a classification unit 10. The classification unit 10 is built into the main body of the defect classification device 1. The classification unit 10 is, for example, a computer. The classification unit 10 includes, for example, a processor mounted on a board and a memory device that stores software for operating the processor.

[0027] The classification unit 10 of the defect classification device 1 classifies patterns of defects X that occur on an object to be inspected W. In this example, the object to be inspected W is the surface of a semiconductor wafer (substrate) in a semiconductor manufacturing process.

[0028] A defect detection device (not shown) detects defects X before the defect classification device 1 classifies the patterns of defects X. In detecting defects X, an inspection image J is obtained by capturing an image of an object W to be inspected using a camera (not shown) connected to an optical microscope or a scanning electron microscope, and the inspection image J is compared with a reference image K.

[0029] The reference image K is also called a fitting image or a statistically acceptable image. The reference image K is obtained, for example, by averaging images of the inspection object W that has been determined to be acceptable. The reference image K is set in advance. When detecting the defect X, the inspection image J is superimposed on the reference image K.

[0030] The classification unit 10 of the defect classification device 1 classifies the pattern of defect X only for the inspection object W in which defect X has been detected by the defect detection device. Here, examples of defect X patterns include foreign matter, scratches, poor edge rinsing, through-hole abnormalities, coating unevenness, peeling, scratches, bonding pad abnormalities, color unevenness, pattern abnormalities, stains, discoloration, deformation, etc.

[0031] The classification unit 10 of the defect classification device 1 classifies the pattern of the defect X in the inspection image J of the object W to be inspected based on a predetermined feature amount of the defect X.

[0032] The feature quantities of the defect X include, for example, area, brightness, color, luminance, shape, clarity, length, width, width / length, angle, transparency, etc. The feature quantities of the defect X are registered in advance in the classification unit 10 of the defect classification device 1.

[0033] The classification unit 10 of the defect classification device 1 may classify the pattern of defect X for an inspection image J of an inspection object W based on classification conditions generated by AI (Artificial Intelligence) learning. Examples of AI learning include machine learning and deep learning. Also, examples of machine learning include "supervised learning," "unsupervised learning," and "reinforcement learning."

[0034] (Division of Field of View) Here, the object to be inspected W is usually larger than the field of view (imaging range) F of the camera. Therefore, the object to be inspected W is imaged so as to be divided into a plurality of fields of view F. As a result, an inspection image J is obtained for each of the plurality of fields of view F.

[0035] Each field of view F (each inspection image J) is adjacent to one another. In this example, each field of view F is rectangular. Each field of view F is arranged in a rectangular grid pattern. In other words, each field of view F is adjacent to one another vertically and horizontally. Each field of view F is further divided into a plurality of pixels.

[0036] Hereinafter, the field of view F may be referred to as FOV (field of view).

[0037] In FIG. 1, the plurality of fields of view F are exemplified by an upper left FOV 11, a lower left FOV 12, an upper right FOV 21, and a lower right FOV 22.

[0038] The FOV 11 and the FOV 12 are adjacent to each other in the vertical direction. A lower end 11c of the FOV 11, which is the end of the FOV 11 on the FOV 12 side, and an upper end 12a of the FOV 12, which is the end of the FOV 11 side, overlap each other.

[0039] The FOV 11 and the FOV 21 are adjacent to each other in the left-right direction. A right end 11b of the FOV 11, which is the end of the FOV 21 on the FOV 21 side, and a left end 21d of the FOV 21, which is the end of the FOV 11 on the FOV 11 side, overlap each other.

[0040] The FOV 21 and the FOV 22 are adjacent to each other in the vertical direction. A lower end 21c of the FOV 21, which is the end of the FOV 21 on the FOV 22 side, and an upper end 22a of the FOV 22, which is the end of the FOV 21 side, overlap each other.

[0041] The FOV 12 and the FOV 22 are adjacent to each other in the left-right direction. A right end 12b of the FOV 12, which is the end of the FOV 22, and a left end 22d of the FOV 22, which is the end of the FOV 12, overlap each other.

[0042] The "edge" mentioned above includes the edge of the field of view F and the area slightly inward from the edge.

[0043] The camera obtains four inspection images J11, J12, J21, and J22 corresponding to the four fields of view F (FOV11, FOV12, FOV21, and FOV22).

[0044] The reference image K is similar to the inspection image J. That is, four reference images K11, K12, K21, and K22 are prepared in advance so as to correspond to the four fields of view F (FOV11, FOV12, FOV21, and FOV22).

[0045] Here, a defect X occurring in the object to be inspected W may span multiple fields of view F (inspection images J). In the example of Fig. 1, the defect X occurring in the object to be inspected W spans the FOV11 (inspection image J11) and FOV21 (inspection image J21) in the left-right direction.

[0046] 1, defect X extends from the middle of the FOV 11 (inspection image J11) to the right end 11b, and then reaches the left end 21d of the FOV 21 (inspection image J21). Defect X has a jagged shape like the teeth of a saw.

[0047] Here, the defect detection device detects defect X in FOV11 (inspection image J11) and FOV21 (inspection image J21) of the inspection object W. On the other hand, the defect detection device does not detect defect X in FOV12 (inspection image J12) and FOV22 (inspection image J22) of the inspection object W. Note that defect X does not exist in any of the four reference images K11, K12, K21, and K22.

[0048] In classifying the pattern of defect X by the defect classification device 1, only the field of view F (inspection image J) in which defect X is detected by the defect detection device is extracted and compared with the reference image K. In the example of FIG. 1 , FOV11 (inspection image J11) and FOV21 (inspection image J21) are first extracted as the field of view F (inspection image J) in which defect X is detected. Then, FOV11 (inspection image J11) is compared with the reference image K11, and FOV21 (inspection image J21) is compared with the reference image K21.

[0049] If an attempt is made to classify the pattern of defect X based on only one field of view F, the pattern of defect X may not be accurately classified. For example, the FOV21 (inspection image J21) only shows a small portion of defect X near its left end 21d. Therefore, even if the FOV21 (inspection image J21) is compared with the reference image K22, the original feature quantities (shape, size, color, etc.) of defect X cannot be fully grasped, and the pattern of defect X cannot be accurately grasped.

[0050] On the other hand, the defect X is visible in most of the FOV11 (inspection image J11) from its middle to the right end 11b. Therefore, when the FOV11 (inspection image J11) is compared with the reference image K11, it may be possible to grasp the original feature amount of the defect X and to grasp the accurate pattern of the defect X. For example, it may be possible to classify the pattern of the defect X as a "flaw" based on the jagged shape of the defect X visible in the FOV11 (inspection image J11).

[0051] However, if defect X is only slightly visible in FOV11 (inspection image J11) as in FOV21 (inspection image J21), the original feature amount of defect X cannot be grasped, and the accurate pattern of defect X cannot be grasped.

[0052] In this way, when a defect X occurring in an object to be inspected W spans a plurality of fields of view F (inspection images J), it is difficult to classify the pattern of the defect X with high accuracy.

[0053] In the defect classification device 1 according to this embodiment, by implementing the following measures, it is possible to accurately classify the pattern of defect X even when the defect X occurring in the inspection object W spans multiple fields of view F (inspection images J).

[0054] (Cutting out a defect image) Hereinafter, the FOV11 will be referred to as the first field of view F1. The FOV21 will be referred to as the second field of view F2. As shown in Fig. 1 , the multiple fields of view F include the FOV11 as the first field of view F1 and the FOV21 as the second field of view F2. The FOV11 (first field of view F1) and the FOV21 (second field of view F2) are adjacent to each other in the left-right direction.

[0055] The right end 11b, which serves as the first end F1p, which is the end of the FOV11 (first field of view F1) on the FOV21 (second field of view F2) side, and the left end 21d, which serves as the second end F2p, which is the end of the FOV21 (second field of view F2) on the FOV11 (first field of view F1) side, overlap each other.

[0056] The right end 11b (first end F1p) of the FOV 11 (first field of view F1) includes the right end of the FOV 11 (first field of view F1) and a region slightly to the left of the right end. The left end 21d (second end F2p) of the FOV 21 (second field of view F2) includes the left end of the FOV 21 (second field of view F2) and a region slightly to the right of the left end.

[0057] The right end of the FOV11 (first field of view F1) on the FOV21 (second field of view F2) side is located closer to the right than the left end of the FOV21 (second field of view F2), i.e., closer to the FOV21 (second field of view F2). The left end of the FOV21 (second field of view F2) on the FOV11 (first field of view F1) side is located closer to the left than the right end of the FOV11 (first field of view F1), i.e., closer to the FOV11 (first field of view F1).

[0058] As shown in FIG. 1 , the FOV11 (first field of view F1) includes a first defect X1 as the defect X. The FOV21 (second field of view F2) includes a second defect X2 as the defect X. The first defect X1 is a part of the defect X. The second defect X2 is a part of the defect X.

[0059] 1, the classification unit 10 of the defect classification device 1 cuts out an area in the FOV 11 (first field of view F1) where the first defect X1 appears as a first defect image C1. The classification unit 10 of the defect classification device 1 cuts out an area in the FOV 21 (second field of view F2) where the second defect X2 appears as a second defect image C2.

[0060] The first defect image C1 is smaller than the FOV11 (first field of view F1). The second defect image C2 is smaller than the FOV21 (second field of view F2). The size and position of the first defect image C1 are set so that the first defect X1 is easily visible. The size and position of the second defect image C2 are set so that the second defect X2 is easily visible. The size and position of the first defect image C1 and the size and position of the second defect image C2 may be the same as or different from each other.

[0061] (Combining Defect Images) The combining of the first defect image C1 and the second defect image C2 will be described in detail. Fig. 2 shows the combining of the first defect image C1 and the second defect image C2. When the first defect image C1 includes the right end 11b (first end F1p) of the FOV 11 (first field of view F1) and the second defect image C2 includes the left end 21d (second end F2p) of the FOV 21 (second field of view F2), the classification unit 10 of the defect classification device 1 classifies the pattern of defect X in a state in which the first defect image C1 and the second defect image C2 are aligned so that the first defect X1 and the second defect X2 overlap.

[0062] In this example, the first defect image C1 includes the right end 11b of the FOV 11. The second defect image C2 includes the left end 21d of the FOV 21. Therefore, the first defect image C1 and the second defect image C2 are aligned so that the first defect X1 and the second defect X2 overlap. Then, in the aligned state of the first defect image C1 and the second defect image C2 (in the state where the first defect X1 and the second defect X2 overlap), the pattern of the defect X is classified.

[0063] Here, the left-right direction in which the FOV11 (first field of view F1) and the FOV21 (second field of view F2) are adjacent to each other is referred to as the adjacent direction A. The adjacent direction A is also the direction in which the FOV11 and the FOV21 overlap. The up-down direction that intersects (more specifically, is perpendicular to) the adjacent direction A, which is the left-right direction, is referred to as the intersecting direction B. The intersecting direction B is also a direction parallel to the right side of the FOV11 and the left side of the FOV21.

[0064] The classification unit 10 of the defect classification device 1 aligns the first defect image C1 and the second defect image C2 in the adjacent direction (left-right direction) A and the intersecting direction (up-down direction) B. That is, the first defect image C1 and the second defect image C2 are aligned while being moved in the adjacent direction A and the intersecting direction B so that the first defect X1 and the second defect X2 overlap each other.

[0065] Conversely, when the first defect image C1 does not include the right end 11b (first end F1p) of the FOV 11 (first field of view F1) or the second defect image C2 does not include the left end 21d (second end F2p) of the FOV 21 (second field of view F2), the classification unit 10 of the defect classification device 1 does not align the first defect image C1 with the second defect image C2 (do not overlap the first defect X1 with the second defect X2). In this case, the classification unit 10 of the defect classification device 1 classifies the pattern of the defect X based on the first defect image C1 alone and / or the second defect image C2 alone.

[0066] (Overlay of First Defect and Second Defect) The overlay of the first defect X1 and the second defect X2 will be described in detail. Fig. 3 shows the overlay of the first defect X1 and the second defect X2. The classification unit 10 of the defect classification device 1 cuts out a part of an area including the right end 11b (first end F1p) of the FOV 11 (first field of view F1) in the first defect image C1 as a matching image M.

[0067] Next, the classification unit 10 of the defect classification device 1 sets, as a search range N, an area including the left end 21d (second end F2p) of the FOV 21 (second field of view F2) in the second defect image C2.

[0068] The matching image M is smaller than the first defect image C1. The search range N is smaller than the second defect image C2. The matching image M is smaller than the search range N. The matching image M may be cut out after the search range N is set in the reverse order to the above.

[0069] The classification unit 10 of the defect classification device 1 extracts a plurality of target images T (T1, T2, ... Tn) from a search range N set in the second defect image C2. Each target image T extracted from the search range N of the second defect image C2 corresponds to a matching image M extracted from the first defect image C1. The size of each target image T is equal to the size of the matching image M.

[0070] The classification unit 10 of the defect classification device 1 obtains a feature amount Ma of the matching image M. The classification unit 10 of the defect classification device 1 obtains a feature amount Ta (Ta1, Ta2, ... Tan) of each of a plurality of target images T (T1, T2, ... Tn).

[0071] The classification unit 10 of the defect classification device 1 compares the feature amount Ma of the matching image M with the feature amount Ta (Ta1, Ta2, . . . Tan) of each of a plurality of target images T (T1, T2, . . . Tn).

[0072] The classification unit 10 of the defect classification device 1 determines the target image T that is closest to the matching image M among the multiple target images T as the closest target image TC.

[0073] The classification unit 10 of the defect classification device 1 aligns the first defect image C1 and the second defect image C2 so as to overlay the matching image M on the closest target image TC (the target image T among the multiple target images T that is closest to the matching image M).

[0074] Specifically, the classification unit 10 of the defect classification device 1 determines, among the multiple target images T, the target image T having the feature amount Ta closest to the feature amount Ma of the matching image M as the closest target image TC.

[0075] The classification unit 10 of the defect classification device 1 superimposes the matching image M on the most recent target image TC (the target image T having the feature value Ta closest to the feature value Ma of the matching image M among the plurality of target images T).

[0076] When comparing the feature amount Ma of the matching image M with the feature amount Ta of each target image T, for example, the SIFT (Scale-Invariant Feature Transform) method is used. Feature points are detected from the matching image M. Feature vectors are obtained in the surrounding areas of the feature points of the matching image M. Feature points are detected from each target image T. Feature vectors are obtained in the surrounding areas of the feature points of each target image T.

[0077] The Euclidean distance is calculated between the feature vector of the matching image M and the feature vector of each target image T. The smaller the Euclidean distance, the higher the similarity between the feature points of the matching image M and the feature points of the target image T. Based on this, the identity of the feature points of the matching image M and the feature points of the target image T is determined.

[0078] In this example, a part of the area including the right end 11b of the FOV 11 in the first defect image C1 is cut out as a matching image M. An area including the left end 21d of the FOV 21 in the second defect image C2 is set as a search range N. From the search range N, multiple target images T (T1, T2, ... Tn) are cut out.

[0079] The feature points of the matching image M are obtained to obtain a feature amount Ma. The feature points of each target image T (T1, T2, ... Tn) are obtained to obtain a feature amount Ta (Ta1, Ta2, ... Tan).

[0080] The feature amount Ma of the feature points of the matching image M is matched with the feature amount Ta (Ta1, Ta2, ... Tan) of the feature points of each target image T (T1, T2, ... Tn). The target image T having the largest overlapping area between the feature amount Ma of the feature points of the matching image M and the feature amount Ta (Ta1, Ta2, ... Tan) of the feature points of each target image T (T1, T2, ... Tn) is determined and designated as the nearest target image TC. The overlap value between the first defect image C1 and the second defect image C2 is determined from the deviation amount between the matching image M and the nearest target image TC, and the first defect image C1 and the second defect image C2 are combined.

[0081] In this way, the classification unit 10 of the defect classification device 1 classifies the pattern of defect X in a state in which the first defect image C1 and the second defect image C2 are aligned (a state in which the first defect X1 and the second defect X2 are overlapped). As a result, the pattern of defect X that has occurred in the inspection object W and straddles the right end 11b of the FOV 11 and the left end 21d of the FOV 21 is correctly classified as a "flaw."

[0082] (Effects) The first defect image C1 extracted from the FOV11 (first field of view F1) and the second defect image C2 extracted from the FOV21 (second field of view F2) are combined so that the first defect X1 and the second defect X2 overlap. In the combined state of the first defect image C1 and the second defect image C2, the pattern of the defect X is classified.

[0083] Even if a defect X that has occurred in the object to be inspected W spans both FOV11 (first field of view F1) and FOV21 (second field of view F2), the original feature amount of the defect X can be grasped.

[0084] As described above, the patterns of defects X occurring in the object W to be inspected can be classified with high accuracy.

[0085] By combining the first defect image C1 extracted from FOV11 and the second defect image C2 extracted from FOV21, rather than directly combining FOV11 and FOV21, the amount of image storage capacity in the classification unit 10 of the defect classification device 1 can be saved.

[0086] The first defect image C1 and the second defect image C2 are aligned so that the matching image M overlaps with the target image T (closest target image TC) that is closest to the matching image M among the multiple target images T. This allows the first defect X1 and the second defect X2 to be more accurately aligned. This is advantageous in accurately classifying the pattern of the defect X that has occurred in the inspection object W.

[0087] By comparing the feature value Ma of the matching image M with the feature value Ta of each of the multiple target images T, it is possible to more accurately determine the target image T (closest target image TC) that is closest to the matching image M from among the multiple target images T.

[0088] The first defect image C1 and the second defect image C2 can be suitably aligned in at least two directions, the adjacent direction A and the cross direction B.

[0089] (Other Embodiments) Although the present disclosure has been described above with reference to preferred embodiments, such descriptions are not limiting, and it goes without saying that various modifications, substitutions, and combinations are possible.

[0090] In the above embodiment, the defect X straddles the FOV 11 as the first field of view F1 and the FOV 21 as the second field of view F2 in the left-right direction, but this is not limited to this. For example, the defect X may straddle the FOV 11 as the first field of view F1 and the FOV 12 as the second field of view F2 in the up-down direction. In this case, the up-down direction is the adjacent direction A, and the left-right direction is the intersecting direction B. Furthermore, the lower end 11c of the FOV 11 is the first end F1p, and the upper end 12a of the FOV 12 is the second end F2p.

[0091] Alternatively, the defect X may span horizontally between FOV12 as the first field of view F1 and FOV22 as the second field of view F2, or vertically between FOV21 as the first field of view F1 and FOV22 as the second field of view F2.

[0092] In the above-described example case, the first field of view F1 and the second field of view F2 may be applied in reverse.

[0093] The names "first..." and "second..." are merely given for the convenience of explanation in the specification.

[0094] In the above embodiment, the upper left FOV 11, the lower left FOV 12, the upper right FOV 21, and the lower right FOV 22 are exemplified as the multiple fields of view F, but many other fields of view may also be included.

[0095] The field of view F is not limited to a rectangle, but may be a polygon other than a rectangle, or may have a shape other than a polygon.

[0096] The defect classification device 1 may be provided with a judgment unit for determining whether the defect X obtained when the first defect image C1 and the second defect image C2 are aligned (when the first defect X1 and the second defect X2 are overlapping) appears natural.

[0097] When comparing the feature amount Ma of the matching image M with the feature amount Ta of each target image T, other methods (for example, a method using a machine learning model) may be applied instead of the SIFT method.

[0098] The classification unit 10 may not be built into the main body of the defect classification device 1, but may be provided in a server or the like outside the main body of the defect classification device 1.

[0099] The object to be inspected W is not limited to a semiconductor wafer, and various objects can be considered. The object to be inspected W may be, for example, glass, metal, resin, etc. The object to be inspected W does not have to be a plate material, and may be, for example, an electric wire, etc.

[0100] The present disclosure is applicable to defect classification devices and is therefore extremely useful and has high industrial applicability.

[0101] W: Inspected object X: Defect X1: First defect X2: Second defect J: Inspection image K: Reference image F: Field of view FOV11 F1: First field of view FOV21 F2: Second field of view 11b: Right end F1p: First end 21d: Left end F2p: Second end C1: First defect image C2: Second defect image A: Adjacent direction B: Intersecting direction M: Matching image N: Search range T: Target image TC: Closest target image Ma: Feature amount Ta: Feature amount 1: Defect classification device 10: Classification unit

Claims

1. [Incorporation by reference (Rules 20.6) 05.12.2024] A defect classification device comprising: a classification unit that classifies a pattern of defects that have occurred in an inspected object, the image of which has been divided into a plurality of fields of view, the plurality of fields of view including a first field of view and a second field of view that are adjacent to each other, a first end of the first field of view that is an end of the second field of view on the side of the first field of view and a second end of the second field of view that is an end of the first field of view on the side of the first field of view overlap each other, the first field of view includes a first defect as the defect, and the second field of view includes a second defect as the defect, the classification unit extracts an area in the first field of view in which the first defect is captured as a first defect image, and extracts an area in the second field of view in which the second defect is captured as a second defect image, and when the first defect image includes the first end and the second defect image includes the second end, the classification unit classifies the pattern of the defects in a state in which the first defect image and the second defect image are aligned such that the first defect and the second defect overlap each other.

2. [Incorporation by reference (Rules 20.6) 05.12.2024] The defect classification device of claim 1, wherein the classification unit extracts a region including the first end in the first defect image as a matching image and sets a region including the second end in the second defect image as a search range, the classification unit extracts a plurality of target images corresponding to the matching images from the search range, and the classification unit aligns the first defect image and the second defect image so as to overlay the matching image on the target image that is closest to the matching image among the plurality of target images.

3. [Incorporation by reference (Rule 20.6) 05.12.2024] The defect classification device of claim 2, wherein the classification unit determines features of the matching image and each of the plurality of target images, the classification unit compares the features of the matching image with the features of each of the plurality of target images, and the classification unit overlays the matching image on the target image among the plurality of target images that has features closest to the features of the matching image.

4. [Incorporation by reference (Rules 20.6) 05.12.2024] A defect classification device as described in any one of claims 1 to 3, wherein the classification unit aligns the first defect image and the second defect image in an adjacent direction in which the first field of view and the second field of view are adjacent, and in a cross direction that crosses the adjacent direction.

1. [Incorrect submission (Rule 20.5-2)] A substrate holding device comprising: a stage for holding a substrate, the stage including a mounting surface on which the substrate is placed, and a plurality of suction grooves formed by recesses provided in the mounting surface and arranged concentrically, the suction grooves suction the substrate by negative pressure, the mounting surface including a boundary portion crossing the suction grooves, and the suction grooves being divided by the boundary portion.

2. [Incorrect submission (Rule 20.5-2)] The substrate holding device of claim 1, wherein the substrate is positioned on the stage by aligning an edge of the substrate with the suction groove.

3. [Incorrect submission (Rule 20.5-2)] A substrate holding device as described in claim 2, further comprising a determination unit that determines, based on the dimensions of the substrate, the suction groove to which the edge of the substrate should be aligned, from among a plurality of concentrically arranged suction grooves.

4. [Incorrect submission (Rule 20.5-2)] A substrate holding device as described in any one of claims 1 to 3, comprising: a negative pressure pump that sucks in gas from the suction groove; a plurality of valves that open and close the gas passage between the negative pressure pump and the suction groove; and a judgment unit, wherein the valves correspond to both of the portions of the suction groove divided by the boundary portion for each of the plurality of concentrically arranged suction grooves, and the judgment unit judges which of the plurality of valves should be opened based on the dimensions of the substrate.

5. [Incorrect submission (Rule 20.5-2)] The stage includes an undivided suction groove that is not divided at the boundary, at a position that is not concentric with the plurality of concentrically arranged suction grooves, and the undivided suction groove adsorbs the substrate by negative pressure. A substrate holding device as described in any one of claims 1 to 3.

6. [Incorrect submission (Rule 20.5-2)] A substrate holding device as described in any one of claims 1 to 3, wherein the stage includes an adjacent groove formed by a recess provided in the placement surface and adjacent to the suction groove.

7. [Incorrect submission (Rule 20.5-2)] A substrate holding device according to any one of claims 1 to 3, wherein the suction groove is rectangular in shape.

Images