Method, system, device and computer program product for wafer defect detection

Abstract

The present invention discloses a method, a system, a device and a computer program product for detecting wafer defects. Among them, the method for detecting wafer defects includes: obtaining a dark-field image of a wafer to be measured under dark-field illumination, and performing defocus detection and analysis using the dark-field image. In the technical solution of the present invention, by using a dark-field image to detect defects on the wafer surface, it is possible to detect some defects on the wafer surface with relatively small contrast that are difficult to detect in the detection of conventional bright-field images.

Description

Technical Field

[0001] The present invention relates to the technical field of optical detection equipment, and particularly to a method, system, device and computer program product for wafer defect detection. Background Art

[0002] During the wafer processing, it is necessary to perform multiple defect detections on the wafer to facilitate timely handling of defects and ensure the product yield.

[0003] In the existing wafer macro defect detection methods, obvious defects such as particles, dirt, and scratches can be detected. However, there is no good detection method for defects with relatively small bright-field imaging contrast such as defocus and shallow scratch color difference. Summary of the Invention

[0004] The purpose of the present invention is to provide a method, system, device and computer program product for wafer defect detection to solve the problem that it is difficult to detect defects with relatively small bright-field imaging contrast in the prior art.

[0005] To achieve the above object, the present invention proposes a method for detecting wafer defects in the first aspect, including:

[0006] Obtaining a dark-field image of a wafer to be measured under dark-field illumination;

[0007] Performing defocus detection and analysis using the dark-field image.

[0008] Preferably, it further includes obtaining a bright-field image of the wafer to be measured under bright-field illumination, and the bright-field image is used for positioning and other defect detection and analysis.

[0009] Preferably, obtaining a bright-field image of the wafer to be measured under bright-field illumination and obtaining a dark-field image under dark-field illumination include:

[0010] Loading the wafer to be measured at the best loading angle;

[0011] Respectively obtaining the bright-field image and the dark-field image of each wafer to be measured under the best loading angle.

[0012] Preferably, the method for obtaining the best loading angle includes:

[0013] Providing a first product wafer with defocus defects;

[0014] Taking the positioning notch directly below the image acquisition port as 0-degree loading, loading the first product wafer at different loading angles, and obtaining the dark-field image of the first product wafer at each loading angle until it rotates one week;

[0015] Calculating the gradient images of the dark-field images of all the first product wafers at the positions where the defocus defects are located;

[0016] Set the upper wafer angle corresponding to the image with the largest gray value among all gradient images as the optimal upper wafer angle.

[0017] Preferably, after respectively obtaining the bright-field image of the wafer to be measured under bright-field illumination and the dark-field image under dark-field illumination, it further includes: a step of correcting the image distortion of the bright-field image and the dark-field image.

[0018] Preferably, it further includes: calculating the offset using the bright-field image of the wafer to be measured to calibrate the horizontal error caused by the upper wafer.

[0019] Preferably, it further includes: correcting the position of the dark-field image of the wafer to be measured according to the offset calculated from the bright-field image of the wafer to be measured.

[0020] Preferably, it further includes: judging the defect position according to the gray value distribution of the dark-field image of the wafer to be measured.

[0021] On the other hand, the technical solution of the present invention also provides a method for detecting wafer defects. Preferably, it includes:

[0022] Provide n qualified second product wafers of the same batch, and respectively obtain the bright-field image and the dark-field image of each second product wafer at the optimal upper wafer angle, where n ∈ (3, 10);

[0023] Based on the n bright-field images and the dark-field images, establish a preset bright-field reference map template and a dark-field reference map template as the detection templates for defocus detection. The bright-field reference map template is used for positioning and other defect detection and analysis, and the dark-field reference map template is used for defocus detection and analysis.

[0024] Preferably, the method for obtaining the optimal upper wafer angle includes:

[0025] Provide a first product wafer with a defocus defect;

[0026] Take the positioning notch directly below the image acquisition port as 0-degree upper wafer, place the first product wafer at different upper wafer angles, and obtain the dark-field image of the first product wafer at each upper wafer angle until it rotates one week;

[0027] Calculate the gradient images of all the dark-field images of the first product wafers at the position where the defocus defect is located;

[0028] Set the upper wafer angle corresponding to the image with the largest gray value among all gradient images as the optimal upper wafer angle.

[0029] Preferably, the formation method of the preset bright-field reference map template includes:

[0030] Set the marked pattern on the bright-field image of the second product wafer as the alignment feature;

[0031] Calculate the horizontal difference in the loading positions of n second product wafers with respect to each other;

[0032] Move the bright-field images and dark-field images of n second product wafers to the same coordinate position to calibrate the horizontal error caused by wafer loading between different second product wafers.

[0033] Preferably, the step of calculating the horizontal difference in the loading positions of n wafers with respect to each other includes:

[0034] Taking the bright-field image and dark-field image of the first second product wafer as a reference, calculate the offsets of the bright-field images and dark-field images of the subsequent n - 1 second product wafers from the bright-field image and dark-field image of the first second product wafer, which are dx1, dy1,...., dxn - 1, dyn - 1 respectively;

[0035] Wherein, dx1, dx2,...., dxn - 1 are the offsets of the bright-field images of the subsequent n - 1 second product wafers from the bright-field image of the first product wafer, and dy1, dy2,...., dyn - 1 are the offsets of the dark-field images of the subsequent n - 1 second product wafers from the dark-field image of the first product wafer.

[0036] Preferably, it includes: The method for forming the preset bright-field reference map template further includes:

[0037] Calculate the mean value of the bright-field images of n second product wafers moved to the same coordinate position at each pixel point to generate a preset bright-field reference map template.

[0038] Preferably, the method for forming the preset dark-field reference map template includes:

[0039] Calculate the gray value distribution of the dark-field images of n second product wafers moved to the same coordinate position at the same pixel positions as the preset dark-field reference map template required for the detection process. The gray value distribution includes: the average gray value, the upper and lower gray limits, and the gray standard deviation at the same pixel positions.

[0040] Preferably, after respectively obtaining the bright-field image and dark-field image of the second product wafer at the optimal wafer loading angle, it further includes: correcting image distortion using a distortion template map.

[0041] Preferably, the method for forming the distortion template map includes:

[0042] Provide a calibration plate corresponding to the resolution of the lithography machine;

[0043] Calibrate the light source, lens, and camera, including: calibrating lens distortion, lens magnification, camera pixel size, and the center point of the image;

[0044] Calculate the distortion template map of the distortion mapping according to the lens distortion and the camera pixel size.

[0045] On the other hand, the technical solution of the present invention also provides a method for detecting wafer defects, including:

[0046] Obtain the bright-field image and the dark-field image of the wafer to be measured;

[0047] Perform position matching between the bright-field image of the wafer to be measured and a predetermined bright-field reference image, and calculate the offset;

[0048] Correct the position of the dark-field image of the wafer to be measured according to the offset;

[0049] Compare the corrected dark-field image of the wafer to be measured with the gray-scale average value and the gray-scale standard deviation value of a predetermined dark-field reference image group to determine the suspected defect position;

[0050] Among them, a predetermined bright-field reference image and a predetermined dark-field reference image group are formed according to the method described above.

[0051] Preferably, the obtaining of the bright-field image and the dark-field image of the wafer to be measured further includes: loading the wafer to be measured at the best loading angle.

[0052] Preferably, after obtaining the bright-field image and the dark-field image of the wafer to be measured, it further includes: correcting the image distortion of the bright-field image and the dark-field image.

[0053] Preferably, the step of comparing the corrected dark-field image of the wafer to be measured with the gray-scale average value and the gray-scale standard deviation value of a predetermined dark-field reference image group to determine the suspected defect position includes:

[0054] Determine the pixels outside the range of mean - k*std - g to mean + k*std + g as the suspected defect position, where mean is the gray-scale average value of the dark-field reference image group, std is the gray-scale standard deviation value of the dark-field reference image group, k is the set detection standard deviation coefficient, and g is the set detection gray-scale threshold.

[0055] Preferably, perform connected region analysis on the pixels of the suspected defect position, remove isolated pixel points to eliminate false detections caused by camera noise, and output defect data, where the defect data includes contrast, position, size, and / or area;

[0056] The output defect data is rejudged and classified according to the maximum defect area and size tolerable by production to eliminate unnecessary detection of small defects, so as to obtain the defect detection result.

[0057] The technical solution of the present invention also provides a wafer defect detection system, including:

[0058] An acquisition module for acquiring a bright-field image and a dark-field image of a wafer to be measured under dark-field illumination;

[0059] A detection module for performing defocus detection and analysis using the dark-field image.

[0060] The technical solution of the present invention also provides a wafer defect detection system, including:

[0061] An acquisition module for acquiring a bright-field image and a dark-field image of a wafer to be measured under dark-field illumination;

[0062] A position matching module for performing position matching between the bright-field image of the wafer to be measured and a predetermined bright-field reference image and calculating an offset;

[0063] A correction module for correcting the position of the dark-field image of the wafer to be measured according to the offset;

[0064] A detection and judgment module for comparing the corrected dark-field image of the wafer to be measured with a predetermined dark-field reference map template to judge the suspected defect position, and the predetermined dark-field reference map template is obtained by the method described in claim 19.

[0065] The technical solution of the present invention also provides a wafer defect detection device, including:

[0066] A processor,

[0067] A memory for storing a computer program or instruction, and the computer program or instruction, when executed by the processor, implements the steps of the method described in any one of the above to perform defect detection on an object.

[0068] The technical solution of the present invention also provides a computer program product, and the computer program product is stored on a computer-readable storage medium and includes a computer program or instruction, and the computer program or instruction, when executed by the processor, implements the steps of the method described in any one of the above.

[0069] Due to the application of the above solution, the present invention has the following advantages and effects compared with the prior art:

[0070] In the technical solution of the present invention, by using a dark-field image to perform defect detection on the wafer surface, it is possible to detect some defects on the wafer surface with relatively small contrast that are difficult to detect in the detection of conventional bright-field images.

[0071] Further, in the technical solution provided by the present invention, defect detection with a relatively small contrast of the wafer is performed by obtaining the bright-field image and the dark-field image of the wafer to be tested. Among them, the bright-field image of the wafer to be tested is position-matched with a predetermined bright-field reference image, the offset is calculated, and then the dark-field image of the wafer to be tested is adjusted and corrected with the offset to ensure the accuracy of the position of the dark-field image. Then, the corrected dark-field image of the wafer to be tested is compared with the gray-scale average value and the gray-scale standard deviation value of a predetermined dark-field reference image group, and the suspected defect position is judged by judging the gray-scale average value and the gray-scale standard deviation value. By comparing with the detection template composed of the predetermined bright-field reference image and the dark-field reference image group, the position of the image of the wafer to be tested can be quickly corrected, and the position of the defect can be preliminarily judged.

[0072] Further, by performing blob (connected domain) analysis on the preliminarily judged defect position, the false detection caused by isolated pixel points is removed.

[0073] Still further, by rejudging the defect data, unnecessary small defect detections are eliminated to ensure the product yield. Description of the Drawings

[0074] Attached Figure 1 is a flowchart of a method for detecting wafer defects provided in Embodiment 1 of this specification;

[0075] Attached Figure 2 is a flowchart of a method for obtaining the best wafer loading angle provided in Embodiment 1 of this specification;

[0076] Attached Figure 3 is a schematic diagram of a device for obtaining the best wafer loading angle provided in Embodiment 1 of this specification;

[0077] Attached Figure 4 is a flowchart of a method for detecting wafer defects provided in Embodiment 2 of this specification;

[0078] Attached Figure 5 is a flowchart of a method for generating a detection template for detecting wafer defects provided in Embodiment 3 of this specification;

[0079] Attached Figure 6 is a schematic diagram of a wafer defect detection device provided in Embodiment 4 of this specification;

[0080] Attached Figure 7 is a schematic diagram of a wafer defocus detection device provided in Embodiment 5 of this specification;

[0081] Attached Figure 8 is a flowchart of a method for detecting wafer defects provided in Embodiment 6 of this specification. Detailed Embodiments

[0082] To make the above objects, features, and advantages of the present application more apparent and understandable, the following provides a detailed description of the specific embodiments of the present application in conjunction with the accompanying drawings. Many specific details are set forth in the following description to facilitate a full understanding of the present application. However, the present application can be implemented in many other ways different from those described herein, and those skilled in the art can make similar improvements without departing from the connotation of the present application. Therefore, the present application is not limited by the specific embodiments disclosed below.

[0083] In the present invention, unless otherwise clearly defined and limited, terms such as "installation", "connection", "connection", "fixation", etc. shall be understood in a broad sense. For example, it can be a fixed connection, a detachable connection, or integrated; it can be a mechanical connection, a direct connection, or an indirect connection through an intermediate medium. It can be the communication inside two components or the interaction relationship between two components, unless otherwise clearly defined. For those of ordinary skill in the art, the specific meanings of the above terms in the present application can be understood according to specific circumstances.

[0084] In addition, terms such as "first", "second", etc. are only used for descriptive purposes and cannot be understood as indicating or implying relative importance or implicitly indicating the quantity of the indicated technical features. Thus, features defined with "first", "second" may explicitly or implicitly include at least one of such features. In the description of the present application, the meaning of "a plurality" is at least two, such as two, three, etc., unless otherwise clearly and specifically defined.

[0085] Due to factors such as partial warping on the surface of the silicon wafer, unevenness on the surface of the photoresist, and errors in light or the lens, the focal plane of the lithography machine is not completely consistent with the surface of the silicon wafer, resulting in defocus defects in the lithography exposure process.

[0086] Generally, wafer inspection mainly uses bright-field images. Under bright-field illumination conditions, there is no obvious difference between the positions with defocus defects and the defect-free positions on the wafer, which brings certain difficulties to accurate inspection.

[0087] The inventors of the present invention have found that at a specific angle under dark-field illumination conditions, the defocus defects will exhibit relatively large contrast.

[0088] Based on this, the present invention provides a method for detecting wafer defects (defocus). This method adopts an image analysis and detection method that combines bright-field illumination (coaxial light incidence) and dark-field illumination (small-angle oblique incidence), and automatically tests the wafer loading angle suitable for imaging different types of wafer defocus defects.

[0089] Furthermore, the technical solution of the present invention can also be applied to the detection of other defects with relatively small contrast on the wafer surface under common bright-field illumination conditions in wafer inspection.

[0090] The technical concept of the present invention will be elaborated in detail by way of specific embodiments as follows.

[0091] Among them, the following situations involving several wafers are interpreted as follows:

[0092] Wafer to be measured: wafers of the same batch that need to be defocused detected, and it is necessary to determine whether there are defocus defects on its surface through detection.

[0093] First product wafer: a non - qualified product wafer that has been determined to have defocus defects on its surface through detection and has the same specifications as the wafer to be measured. In the embodiments provided by the present invention, it is mainly used to obtain the best wafer - loading angle.

[0094] Second product wafer: a qualified product wafer that has been determined to have no defocus defects on its surface through detection and has the same specifications as the wafer to be measured. In the embodiments provided by the present invention, it is mainly used to obtain the preset bright - field reference - image template and dark - field reference - image template for constructing automatic detection.

[0095] Embodiment 1

[0096] Specifically, Figure 1 The flowchart of a method for detecting wafer defects provided in this embodiment includes:

[0097] Execute step S100: Obtain the dark - field image of the wafer to be measured under dark - field illumination;

[0098] Execute step S200: Perform defocus detection and analysis using the dark - field image.

[0099] Specifically, in this embodiment, step S100 may further include: obtaining the bright - field image of the wafer to be measured under bright - field illumination; wherein, the bright - field image is used for positioning and other defect detection and analysis.

[0100] Furthermore, obtaining the bright - field image of the wafer to be measured under bright - field illumination and obtaining the dark - field image under dark - field illumination include:

[0101] Execute step S110: Load the wafer to be measured at the best wafer - loading angle;

[0102] Figure 2 The flowchart of the method for obtaining the best wafer - loading angle is shown as follows, Figure 3 The schematic diagram of the device for obtaining the best wafer - loading angle is shown.

[0103] Through further, in combination with reference to Figures 2 to 3 As shown, the method for obtaining the best wafer - loading angle includes:

[0104] Perform step A11: Provide a first product wafer 900 with defocus defects;

[0105] In this step, the specific positions of the defocus defects on the wafer surface of the first product wafer 900 have been obtained through other tests or observations.

[0106] Perform step A12: Use the alignment notch directly below the image acquisition port as 0-degree wafer loading, load the first product wafer at different wafer loading angles, and obtain the dark-field images of the first product wafer at each wafer loading angle until one full rotation;

[0107] Specifically, refer to Figure 3 As described, the first product wafer 900 is carried below the image acquisition port 444. The entire circumference is evenly divided into 8 positions corresponding to the same circular angles, which are C1 to C8 in sequence.

[0108] When the alignment notch of the first product wafer 900 is at the C1 position, it is exactly directly below the image acquisition port 444, and this is 0-degree wafer loading at this time. Then, the first product wafer 900 is respectively set at the positions of C2 to C8 for wafer loading.

[0109] Specifically, at the eight wafer loading positions of C1 to C8, image acquisition is respectively performed on the first product wafer 900. The first product wafer images corresponding to the wafer loading at 8 positions are respectively obtained.

[0110] Among them, preferably, the acquired images are dark-field images obtained under dark-field illumination at the positions corresponding to each wafer loading angle. In other embodiments, the acquired images may also include obtaining bright-field images under bright-field illumination and obtaining dark-field images under dark-field illumination.

[0111] Then, perform step A13: Calculate the gradient images of all the dark-field images of the first product wafer at the positions where the defocus defects are located;

[0112] Then, perform step A14: Calculate the gray values in all the gradient images, and use the image with the largest gray value and its corresponding wafer loading angle as the best wafer loading angle with the largest defocus imaging contrast.

[0113] Through the above steps, the best wafer loading angle can be obtained. Loading at the best wafer loading angle can obtain the wafer image with the most obvious defocus defects.

[0114] After obtaining the best wafer loading angle, load the wafer to be tested at the best wafer loading angle;

[0115] Then, execute step S120: Respectively obtain the bright-field image and the dark-field image of each wafer to be tested under the best wafer loading angle.

[0116] In this embodiment, a bright-field image of the wafer to be measured under bright-field illumination and a dark-field image of the wafer to be measured under dark-field illumination are respectively obtained.

[0117] Then, it further includes: performing distortion correction on the bright-field image and the dark-field image. Specifically, in this embodiment, a distortion template image is used to correct image distortion.

[0118] Preferably, the method for forming the distortion template image includes:

[0119] Providing a calibration board corresponding to the resolution of the lithography machine;

[0120] Calibrating the light source, the lens, and the camera, including: calibrating lens distortion, lens magnification, camera pixel size, and the center point of the image;

[0121] Calculating the distortion template image of the distortion mapping according to the distortion and the camera pixel size.

[0122] In this embodiment, the error caused by the distortion introduced in the image of the wafer collected by the image acquisition port can be eliminated through the distortion template image. And the pixel size of the wafer image can be corrected to a standard value to eliminate the differences between machines.

[0123] After correcting the image distortion, it includes: calculating the offset using the bright-field image of the wafer to be measured to calibrate the horizontal error caused by wafer loading.

[0124] Preferably, it further includes: correcting the position of the dark-field image of the wafer to be measured according to the offset calculated from the bright-field image of the wafer to be measured.

[0125] Preferably, it further includes: calculating the gray value distribution at the same pixel position of the dark-field images of different wafers to be measured to determine the position where the defocus defect is located. Among them, the dark-field image used is the corrected dark-field image of the wafer to be measured.

[0126] In this embodiment, the offset between different wafers is calculated through the significant marking features in the bright-field image of the wafer, and then used to correct the position of the dark-field image of the wafer to ensure accurate positioning of the dark-field image. Then, through the gray value distribution, the accurate position and characteristics of the defect are determined. In this way, the accurate judgment of the defect position is ensured.

[0127] Embodiment Two

[0128] Figure 4 As shown, it is a flowchart of a method for detecting wafer defects provided in this embodiment, which includes:

[0129] Step S310: Obtaining a bright-field image and a dark-field image of the wafer to be measured;

[0130] Step S320: Perform position matching between the bright-field image of the wafer to be measured and a predetermined bright-field reference image, and calculate the offset;

[0131] Step S330: Correct the position of the dark-field image of the wafer to be measured according to the offset;

[0132] Step S340: Compare the corrected dark-field image of the wafer to be measured with the gray-scale average value and gray-scale standard deviation value of a predetermined dark-field reference image group to determine the suspected defect position.

[0133] Specifically, execute Step S310: Obtain the bright-field image and dark-field image of the wafer to be measured;

[0134] Preferably, the obtaining of the bright-field image and dark-field image of the wafer to be measured further includes: loading the wafer to be measured at the optimal loading angle.

[0135] Preferably, the method for obtaining the optimal loading angle includes:

[0136] Provide a product wafer with defocus defects;

[0137] Take the positioning notch directly below the image acquisition port as the 0-degree loading, rotate the product wafer at regular intervals of a certain angle, load the wafer at different loading angles, and obtain the dark-field image of each loading angle until one full rotation;

[0138] Calculate the gradient images of the product wafers at the defocus positions of all images, then calculate the image number with the largest gray-scale value in all gradient images and the corresponding loading angle of its original image, and take this angle as the optimal loading angle with the largest defocus imaging contrast.

[0139] Preferably, after respectively obtaining the bright-field image and dark-field image of the wafer to be measured under bright-field illumination and dark-field illumination, it further includes: correcting image distortion using a distortion template image.

[0140] Preferably, the method for forming the distortion template image includes:

[0141] Provide a calibration plate corresponding to the resolution of the lithography machine;

[0142] Calibrate the light source, lens, and camera, including: calibration of lens distortion, lens magnification, camera pixel size, and image center point;

[0143] Calculate the distortion template image of the distortion mapping according to the distortion and camera pixel size.

[0144] Next, execute Step S320: Perform position matching between the bright-field image of the wafer to be measured and a predetermined bright-field reference image, and calculate the offset;

[0145] In this embodiment, the selected feature markers in the predetermined bright-field reference image can be compared with the feature markers of the wafer to be measured for position matching, so as to calculate the offset between the two.

[0146] Next, step S330 is executed: correcting the position of the dark-field image of the wafer to be measured according to the offset.

[0147] In this embodiment, the offset between different wafers is calculated through the prominent marker features in the bright-field image of the wafer, and then used to correct the position of the dark-field image of the wafer to ensure accurate positioning of the dark-field image.

[0148] Step S340: Comparing the corrected dark-field image of the wafer to be measured with the gray-scale average value and gray-scale standard deviation value of the predetermined dark-field reference image group to determine the suspected defect position.

[0149] The corrected dark-field image to be measured is compared with the gray-scale average value and gray-scale standard deviation value of the dark-field reference image group. It is considered that the pixel positions between mean - k*std - g and mean + k*std + g are not defect positions, and the pixels outside this range are regarded as suspected defect positions. Where mean is the gray-scale average value of the dark-field reference image group, std is the gray-scale standard deviation value of the dark-field reference image group, k is the set detection standard deviation coefficient, and g is the detection gray-scale threshold.

[0150] Furthermore, in this embodiment, it also includes performing blob (connected region) analysis on the pixel positions of the suspected defects to remove isolated pixel points to eliminate false detections caused by camera noise.

[0151] Blob in computer vision refers to a connected region in an image. Blob analysis is to perform connected component extraction and marking on the binary image after foreground / background separation. Each marked blob represents a foreground object, and then some related features of the blob can be calculated, such as geometric features like area, centroid, circumscribed rectangle, etc., and color and texture features of the blob can also be calculated. These features can all be used as the basis for tracking.

[0152] Furthermore, in this embodiment, it also includes outputting defect data, and the defect data includes the contrast, position, size, area, etc. of the defocus defect.

[0153] Furthermore, in this embodiment, it also includes rechecking and classifying the output defect data according to the maximum defect area and size tolerable in production to eliminate unnecessary detection of small defects and ensure the product yield.

[0154] Specifically, in the above steps, the formation methods of the predetermined bright-field reference image and the dark-field reference image group include:

[0155] Step B21: Provide n qualified product wafers of the same batch, and obtain the bright-field image and dark-field image of each product wafer at the optimal wafer-loading angle, where n ∈ (3, 10);

[0156] Step B22: Set a marking pattern on the bright-field image of the product wafer as an alignment feature;

[0157] Step B23: Calculate the horizontal difference in wafer-loading positions between the n product wafers (dx1, dy1,...., dxn-1, dyn-1), eliminate the wafer-loading error, and generate a reference bright-field image;

[0158] Step B24: Move the bright-field images and dark-field images of the n product wafers to the same coordinate position to calibrate the horizontal error caused by wafer loading.

[0159] Among them, calculating the horizontal difference in wafer-loading positions between the n wafers (dx1, dy1,...., dxn-1, dyn-1) includes: using the bright-field image and dark-field image of the first product wafer as a reference, and calculating the offsets of the bright-field images and dark-field images of the subsequent n-1 product wafers from the first wafer.

[0160] Specifically, after correcting the image positions in the previous step, a predetermined bright-field reference image and a set of dark-field reference images can be calculated. Specifically as follows:

[0161] The method for forming the predetermined bright-field reference image includes:

[0162] Step B25-1: Calculate the mean value of each pixel point of the bright-field images of the n product wafers to generate a predetermined bright-field reference image.

[0163] Specifically, the method for forming the predetermined set of dark-field reference images includes:

[0164] Step B25-2: Calculate the gray value distribution of the dark-field images of different product wafers at the same pixel position, and calculate the average gray value, upper and lower gray limits, and gray standard deviation at the same pixel position as the predetermined set of dark-field reference images required for the detection process.

[0165] In this embodiment, the bright-field image and the dark-field image of the wafer to be measured are obtained for defect detection of the wafer. Among them, the bright-field image of the wafer to be measured is position-matched with a predetermined bright-field reference image, and the offset is calculated. Then, the offset is used to adjust and correct the dark-field image of the wafer to be measured to ensure the accuracy of the position of the dark-field image. Then, the corrected dark-field image of the wafer to be measured is compared with the gray-scale average value and the gray-scale standard deviation value of a predetermined dark-field reference image group. Through the judgment of the gray-scale average value and the gray-scale standard deviation value, the preliminarily judged defect data is obtained. The defect data includes: suspected defect positions, geometric features of the defects (area, length, width, etc.), and brightness features (brightness, contrast, etc.).

[0166] In this embodiment, by comparing with a detection template composed of a predetermined bright-field reference image and a dark-field reference image group, the position of the image of the wafer to be measured can be quickly corrected, and the preliminarily judged defect data can be obtained.

[0167] Further, in this embodiment, by performing blob analysis on the positions of the preliminarily judged defects, false detections caused by isolated pixel points are removed.

[0168] Still further, by re-judging the defect data, unnecessary small defect detections are eliminated to ensure the product yield.

[0169] Embodiment III

[0170] Reference Figure 5 As shown, this embodiment provides a method for generating a detection template for detecting wafer defects, including:

[0171] Step S410: Provide n qualified second product wafers of the same batch, and respectively obtain the bright-field image and the dark-field image of each second product wafer at the best wafer loading angle, where n ∈ (3, 10);

[0172] Among them, the method for obtaining the best wafer loading angle is similar to the method provided in the above embodiment and will not be elaborated here.

[0173] Preferably, after respectively obtaining the bright-field image and the dark-field image of the product wafer under bright-field illumination and dark-field illumination, it further includes: using a distortion template image to correct image distortion.

[0174] Preferably, the method for forming the distortion template image includes:

[0175] Provide a calibration plate corresponding to the resolution of the lithography machine;

[0176] Calibrate the light source, lens, and camera, including: calibration of lens distortion, lens magnification, camera pixel size, and image center point;

[0177] Calculate the distortion template image of the distortion mapping according to the distortion and the camera pixel size.

[0178] The n bright-field images and n dark-field images after the above-mentioned distortion correction are used as the basic image data for subsequent processing.

[0179] Step S420: Based on the n bright-field images and the dark-field image, establish a preset bright-field reference map template and a dark-field reference map template as the detection templates for defocus detection. The bright-field reference map template is used for positioning and other defect detection and analysis, and the dark-field reference map template is used for defocus detection and analysis.

[0180] Specifically, the specific method of step S420 includes:

[0181] Step C31: Set a marking pattern on the bright-field image of the second product wafer as an alignment feature;

[0182] Step C32: Calculate the horizontal difference amounts (dx1, dy1,...., dxn-1, dyn-1) of the wafer loading positions among the n second product wafers, eliminate the wafer loading error, and generate a bright-field reference image;

[0183] Step C33: Move the bright-field images and dark-field images of the n second product wafers to the same coordinate position to calibrate the horizontal error caused by wafer loading.

[0184] Among them, the calculation of the horizontal difference amounts (dx1, dy1,...., dxn-1, dyn-1) of the wafer loading positions among the n second product wafers includes: taking the bright-field image and dark-field image of the first second product wafer as references, and calculating the offsets of the bright-field images and dark-field images of the subsequent n-1 wafers from the first second product wafer.

[0185] Specifically, after the correction of the image position in the previous step and the calibration of the horizontal error caused by wafer loading, a predetermined bright-field reference map and a dark-field reference image group can be calculated. Specifically as follows:

[0186] The formation method of the predetermined bright-field reference map includes:

[0187] Step C34-1: Calculate the mean value of each pixel point in the images of the n second product wafers to generate a predetermined bright-field reference image.

[0188] The formation method of the predetermined dark-field reference image group includes:

[0189] Step C34-2: Calculate the gray value distribution of the dark-field images of different second product wafers at the same pixel position, and calculate the average gray value, the upper and lower limits of the gray value, and the gray standard deviation at the same pixel position as the predetermined dark-field reference image group required for the detection process.

[0190] In this embodiment, a method for forming a reference template for wafer defocus detection is provided. Based on the images of n defect-free second product wafers, a bright-field reference image template and a group of dark-field reference images are established, which can provide a fast, intelligent, and automatic detection process.

[0191] Furthermore, in this embodiment, after obtaining the bright-field image and the dark-field image of the second product wafer, image distortion correction is also performed, which can eliminate the error caused by the distortion introduced in the image of the wafer collected by the image acquisition port. And the pixel size of the wafer image can be corrected to within one pixel to eliminate the differences between machines. Ensure the accuracy of the detection template.

[0192] Furthermore, in this embodiment, the offset between different wafers is calculated through the prominent marking features in the bright-field image of the wafer, and then used to correct the position of the dark-field image of the wafer to ensure accurate positioning of the dark-field image. Further ensure the accuracy of the detection template.

[0193] Embodiment 4

[0194] Reference Figure 6 As shown, this embodiment provides a schematic diagram of a wafer defect detection device, including:

[0195] An acquisition module 10, configured to acquire a bright-field image and a dark-field image of a wafer to be measured under dark-field illumination;

[0196] A detection module 20, configured to perform defocus detection and analysis using the dark-field image.

[0197] Embodiment 5

[0198] Reference Figure 7 As shown, this embodiment provides a schematic diagram of a wafer defect detection device, including:

[0199] An acquisition module 11, configured to acquire a bright-field image and a dark-field image of a wafer to be measured under dark-field illumination;

[0200] A position matching module 21, configured to perform position matching between the bright-field image of the wafer to be measured and a predetermined bright-field reference image, and calculate the offset;

[0201] A correction module 31, configured to correct the position of the dark-field image of the wafer to be measured according to the offset;

[0202] A detection and judgment module 41, configured to compare the corrected dark-field image of the wafer to be measured with a predetermined dark-field reference map template 41a to judge the suspected defect position, and the predetermined dark-field reference map template is obtained by the method described in Embodiment 3.

[0203] Embodiment 6

[0204] Reference Figure 8, the method for detecting wafer defects (defocus) provided in this embodiment includes a calibration step F1, a modeling step F2, and a detection step F3. Specifically, the execution process of the method for detecting wafer defects is described as follows:

[0205] 1) Calibration step F1

[0206] Prepare a calibration plate corresponding to the resolution of the lithography machine, and calibrate the light source, lens, and camera, including lens distortion, lens magnification, camera pixel size, and calibration of the image center point. And calculate the map image of the distortion mapping according to the distortion and pixel size, which can map the distorted image into a non-distorted image and correct the pixel size to a standard value to eliminate the differences between machines.

[0207] 2) Modeling step F2

[0208] Reference Figure 2 , test the product wafers with defocus defects, and use the alignment notch (notch / flat notch) directly below the image acquisition device as the 0-degree wafer loading. Rotate the wafer at regular intervals and load the wafer at different wafer loading angles to obtain the dark-field images at each wafer loading angle until one full rotation is completed. Calculate the gradient images of the defocus positions of all images, and then calculate the image number with the largest gray value in all gradient images and the corresponding wafer loading angle of its original image. Take this angle as the wafer loading angle theta with the largest defocus imaging contrast.

[0209] Prepare n product wafers of the same batch and load all product wafers at the theta angle, and respectively obtain the bright-field images and dark-field images of each product wafer at the theta angle, where n ∈ (3, 10).

[0210] Use the distortion map to correct the image distortion and machine errors. Use the pattern on the bright-field image of the product wafer as the alignment feature (a unique pattern within a certain range), and calculate the offset of the horizontal position of the n wafer loadings (dx1, dy1,...., dxn-1, dyn-1). Move the bright-field images and dark-field images of the n wafers to the same coordinate position to calibrate the horizontal error caused by wafer loading. Calculate the average value of each pixel point of the n images to generate a bright-field reference image for alignment.

[0211] After the correction of the image position in the previous step, calculate the gray value distribution of the dark-field images of different wafers at the same pixel position, and calculate the average gray value, upper and lower gray limits, and gray standard deviation at the same pixel position as the dark-field reference image group of the detection process.

[0212] 3) Detection step F3

[0213] Load the wafer to be measured at the optimal loading angle theta to obtain the bright-field image and dark-field image of the wafer to be measured. Use the distortion map to correct the image distortion, match the position of the bright-field image of the wafer to be measured with the previous bright-field reference image, and calculate the offset. Correct the position of the dark-field image of the wafer to be measured.

[0214] Compare the corrected dark-field image of the wafer to be measured with the gray-scale average value and gray-scale standard deviation value of the dark-field reference image group. It is considered that the pixel positions between mean - k * std - g and mean + k * std + g are not defect positions, and the pixels outside this range are regarded as suspected defect positions. Among them, mean is the gray-scale average value of the dark-field reference image group, std is the gray-scale standard deviation value of the dark-field reference image group, k is the detection standard deviation coefficient, and g is the detection gray-scale threshold.

[0215] Perform blob analysis on the pixel positions of suspected defects to remove isolated pixels to eliminate false detections caused by camera noise.

[0216] Output the defect results, including contrast, position, size, area, etc., and rejudge and classify the output defects according to the maximum defect area and size tolerable in production to eliminate unnecessary detection of small defects and ensure the product yield.

[0217] Example Seven

[0218] An embodiment of the present invention also provides a wafer defect detection device, including:

[0219] A processor,

[0220] A memory for storing computer programs or instructions, which when executed by the processor implement the steps of the method as described above to perform defect detection on an object.

[0221] In a specific implementation, the processor may include a Central Processing Unit (CPU), a Graphics Processing Unit (GPU), a Field Programmable Gate Array (FPGA), etc. The memory may include a Random Access Memory (RAM), a Read-Only Memory (ROM), a Non-Volatile Memory (NVM), etc.

[0222] Example Eight

[0223] Embodiments of the present invention also provide a computer program product. The computer program product is stored on a computer-readable storage medium and includes a computer program or instructions. When the computer program or instructions are executed by a processor, the steps of the method described in any of the above embodiments are implemented.

[0224] In specific implementations, computer instructions may include any suitable type of code implemented by using any suitable high-level, low-level, object-oriented, visual, compiled, and / or interpreted programming language. For example, source code, compiled code, interpreted code, executable code, static code, dynamic code, encrypted code, etc.

[0225] Among them, the computer-readable storage medium may include any suitable type of memory unit, memory device, memory item, memory medium, storage device, storage item, storage medium, and / or storage unit. For example, memory, removable or non-removable media, erasable or non-erasable media, writable or rewritable media, digital or analog media, hard disks, floppy disks, optical discs, etc. And, computer instructions may include any suitable type of code implemented by using any suitable high-level, low-level, object-oriented, visual, compiled, and / or interpreted programming language. For example, source code, compiled code, interpreted code, executable code, static code, dynamic code, encrypted code, etc.

[0226] In this application, the mention of "embodiment" or "implementation manner" means that the specific features, structures, or characteristics described in connection with the embodiment may be included in at least one embodiment of this application. The phrase appearing in various positions in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment mutually exclusive with other embodiments. Those skilled in the art explicitly and implicitly understand that the embodiments described in this application can be combined with other embodiments. In addition, it should also be understood that the features, structures, or characteristics described in each embodiment of this application can be combined arbitrarily without contradiction to form another embodiment that does not deviate from the spirit and scope of the technical solution of this application. Finally, it should be noted that the above implementation manners are only used to illustrate the technical solution of this application and not to limit it. Although the technical solution of this application has been described in detail with reference to the above preferred implementation manners, those of ordinary skill in the art should understand that modifications or equivalent replacements of the technical solution of this application should not deviate from the spirit and scope of the technical solution of this application.

Claims

1. A method for detecting wafer defects, characterized in that, Including: Loading the wafer to be measured at the optimal loading angle; Respectively obtaining the bright-field image and the dark-field image of each wafer to be measured under the optimal loading angle; Performing defocus detection and analysis using the dark-field image, and using the bright-field image for positioning and other defect detection and analysis; The method for obtaining the optimal loading angle includes: Providing a first product wafer with defocus defects; Taking the position where the alignment notch is directly below the image acquisition port as 0-degree loading, loading the first product wafer at different loading angles, and obtaining the dark-field images of the first product wafer at each loading angle until a full rotation; Calculating the gradient images of the dark-field images of all the first product wafers at the position of the defocus defect; Setting the loading angle corresponding to the image with the largest gray value among all the gradient images as the optimal loading angle.

2. The method according to claim 1, wherein After respectively obtaining the bright-field image of the wafer to be measured under bright-field illumination and the dark-field image under dark-field illumination, it further includes the step of correcting the image distortion of the bright-field image and the dark-field image.

3. The method according to claim 1, wherein It also includes: Calculating the offset using the bright-field image of the wafer to be measured to calibrate the horizontal error caused by loading.

4. The method according to claim 3, wherein It also includes: Correcting the position of the dark-field image of the wafer to be measured according to the offset calculated from the bright-field image of the wafer to be measured.

5. The method according to claim 1 or 4, characterized in that It also includes: Judging the defect position according to the gray value distribution of the dark-field image of the wafer to be measured.

6. A method for detecting wafer defects, characterized in that, Including: Providing n second product wafers of the same batch and qualified, respectively obtaining the bright-field image and the dark-field image of each second product wafer under the optimal loading angle, where n ∈ (3, 10); Setting the marked pattern on the bright-field image of the second product wafer as the alignment feature; Calculating the horizontal difference amount of the loading positions among the n second product wafers based on the alignment feature, and thus moving both the bright-field image and the dark-field image to the same coordinate position based on their respective horizontal difference amounts to calibrate the horizontal error caused by loading between different second product wafers; Based on the calibrated bright-field image and dark-field image, establishing a bright-field reference map template and a dark-field reference map template as the detection templates for defocus detection, where the dark-field reference map template is used for defocus detection and analysis, and the bright-field reference map template is used for positioning and other defect detection and analysis; The method for obtaining the optimal loading angle includes: Providing a first product wafer with defocus defects; Taking the position where the alignment notch is directly below the image acquisition port as 0-degree loading, loading the first product wafer at different loading angles, and obtaining the dark-field images of the first product wafer at each loading angle until a full rotation; Calculating the gradient images of the dark-field images of all the first product wafers at the position of the defocus defect; Setting the loading angle corresponding to the image with the largest gray value among all the gradient images as the optimal loading angle.

7. The method according to claim 6, wherein The step of calculating the horizontal difference amount of the loading positions among n wafers includes: Taking the bright-field image and the dark-field image of the first second product wafer as references, calculating the offsets of the bright-field images and the dark-field images of the subsequent n - 1 second product wafers from the bright-field image and the dark-field image of the first second product wafer, which are dx1, dy1,...., dxn-1, dyn-1 respectively; Wherein, dx1, dx2,...., dxn-1 are the offsets of the bright-field images of the subsequent n-1 second product wafers from the first product wafer, and dy1, dy2,...., dyn-1 are the offsets of the dark-field images of the subsequent n-1 second product wafers from the first product wafer.

8. The method according to claim 6, wherein Including: The method for forming the bright-field reference map template further includes: Calculating the mean value of the bright-field images of n second product wafers moved to the same coordinate position at each pixel point to generate a bright-field reference map template.

9. The method according to claim 6, characterized in that, The method for forming a preset dark-field reference map template includes: Calculating the gray value distribution of the dark-field images of n second product wafers moved to the same coordinate position at the same pixel position as the predetermined dark-field reference map template required for the detection process. The gray value distribution includes: the average gray value, the upper and lower gray limits, and the gray standard deviation at the same pixel position.

10. The method according to claim 6, wherein After respectively obtaining the bright-field image and the dark-field image of the second product wafer at the optimal wafer loading angle, it further includes: correcting the image distortion using a distortion template map.

11. The method according to claim 10, wherein The method for forming the distortion template map includes: Providing a calibration board corresponding to the resolution of the lithography machine; Calibrating the light source, lens, and camera, including: calibrating lens distortion, lens magnification, camera pixel size, and image center point; Calculating the distortion template map of the distortion mapping according to the lens distortion and the camera pixel size.

12. A method for detecting wafer defects, characterized in that, Including: Obtaining the bright-field image and the dark-field image of the wafer to be measured; Performing position matching between the bright-field image of the wafer to be measured and the bright-field reference map template, and calculating the offset; Correcting the position of the dark-field image of the wafer to be measured according to the offset; Comparing the gray average value and the gray standard deviation value of the corrected dark-field image of the wafer to be measured with those of the dark-field reference map template to determine the suspected defect positions, including: determining the pixels outside the range from mean - k*std - g to mean + k*std + g as the suspected defect positions, where mean is the gray average value of the dark-field reference image group, std is the gray standard deviation value of the dark-field reference image group, k is the set detection standard deviation coefficient, and g is the set detection gray threshold; Wherein, the bright-field reference map template and the dark-field reference map template are formed according to the method described in claim 6.

13. The method according to claim 12, characterized in that, The obtaining of the bright-field image and the dark-field image of the wafer to be measured further includes: loading the wafer to be measured at the optimal wafer loading angle.

14. The method according to claim 12, wherein After obtaining the bright-field image and the dark-field image of the wafer to be measured, it further includes: the step of correcting the image distortion of the bright-field image and the dark-field image.

15. The method according to claim 12, wherein The method further includes: Performing connected region analysis on the pixels at the suspected defect positions, removing isolated pixel points to eliminate false detections caused by camera noise, and outputting defect data, where the defect data includes contrast, position, size, and / or area; And performing rejudgment and classification on the output defect data according to the maximum defect area and size tolerable in production to eliminate unnecessary small defect detections to obtain the defect detection result.

16. A wafer defect detection system, characterized in that, Including: An obtaining module for providing a first product wafer with defocus defects; Taking the positioning port directly below the image acquisition port as the 0-degree wafer loading, loading the first product wafer at different wafer loading angles, and acquiring the dark-field images of the first product wafer at each wafer loading angle until one full rotation; calculating the gradient images of all the dark-field images of the first product wafer at the positions of the defocus defects; setting the wafer loading angle corresponding to the image with the largest gray value among all the gradient images as the optimal wafer loading angle, and loading the wafers to be tested at the optimal wafer loading angle; respectively acquiring the bright-field images and dark-field images of each wafer to be tested under the optimal wafer loading angle; A detection module that uses the dark-field image for defocus detection and analysis, and the bright-field image is used for positioning and other defect detection and analysis.

17. A wafer defect detection system, characterized in that, Comprising: An acquisition module for acquiring the bright-field image and dark-field image of the wafer to be tested under dark-field illumination; A position matching module for performing position matching between the bright-field image of the wafer to be tested and the bright-field reference map template, and calculating the offset; A correction module for correcting the position of the dark-field image of the wafer to be tested according to the offset; A detection and judgment module for comparing the corrected dark-field image of the wafer to be tested with the dark-field reference map template to judge the suspected defect positions, including: determining the pixels outside the range from mean - k*std - g to mean + k*std + g as the suspected defect positions, where mean is the gray average value of the dark-field reference map template, std is the gray standard deviation value of the dark-field reference map template, k is the set detection standard deviation coefficient, g is the set detection gray threshold, and the dark-field reference map template is obtained by the method described in claim 6.

18. A wafer defect detection device, characterized in that, Comprising: A processor, A memory that stores computer programs or instructions, and the computer programs or instructions, when executed by the processor, implement the steps of the method according to any one of claims 1 to 15 to perform defect detection on an object.

19. A computer program product, characterized in that, The computer program product is stored on a computer-readable storage medium and includes computer programs or instructions, and the computer programs or instructions, when executed by the processor, implement the steps of the method according to any one of claims 1 to 15.

Images