A defect detection method

Abstract

The present invention provides a defect detection method, including: performing electron microscopy scanning on abnormal positions on the surface of a wafer to obtain a scanned image, and classifying the corresponding abnormal positions into true defects and interference defects according to the scanned image; performing line processing on the first scanned image and the second scanned image of the interference defects respectively to obtain corresponding first line-processed pictures and second line-processed pictures; comparing the first line-processed picture with the second line-processed picture, and classifying the interference defects into false defects and non-feature defects according to the comparison result; and giving an early warning for the non-feature defects. The defect detection method provided by the present invention generates a first line-processed picture and a second line-processed picture of the abnormal position through image post-processing, and by comparing the first line-processed picture and the second line-processed picture, it is convenient and intuitive to perform secondary identification on non-feature defects, improves the recognition accuracy of non-feature defects, avoids batch anomalies, and improves the product yield.

Description

Technical Field

[0001] The present invention belongs to the technical field of defect detection, and particularly relates to a defect detection method. Background Art

[0002] There are yield problems in wafers themselves, and various defects may also exist on the wafer surface. In order to prevent wafers with defects from flowing into the subsequent packaging process, it is necessary to use optical detection equipment to identify, classify, and mark the defects on the wafer surface to assist in wafer sorting. Traditionally, defect classification is carried out by yield engineers using a scanning electron microscope. The general process is as follows: the scanner defines the abnormal position, takes a photo of the abnormal position with a scanning electron microscope (SEM) to generate an SEM scanning image, and the engineer identifies the scanning image. According to the identification result, the abnormal position is classified into true defects and nuisance defects, and based on this, it is decided whether to notify the corresponding machine to conduct defect investigation. In the detection process of the prior art, some non-characteristic defects, or defects with unclear characteristics, do not show obvious characteristic defect structures in the scanning image, and are often misclassified as nuisance defects by yield engineers, and thus are not timely warned, resulting in batch abnormalities in wafers and loss of product yield. The characteristic defect structures include spots, patches, pits, protrusions, scratches, deformations, color differences, defects, etc. When the scanning image shows these characteristic defect structures, the yield engineer will classify the abnormal position corresponding to the scanning image as a true defect. When the scanning image does not show these characteristic defect structures, the yield engineer will classify the abnormal position corresponding to the scanning image as a nuisance defect. In addition, when the detection sensitivity is low or the characteristic size of the defect structure is small (for example, less than 100 nm), the number of non-characteristic defects will increase, which will further reduce the detection accuracy. Due to the above characteristics of non-characteristic defects, there is still a lack of an efficient and accurate detection method for them in the existing detection technology.

[0003] Therefore, it is necessary to propose a defect detection method that can efficiently and accurately identify and detect non-characteristic defects, and accordingly improve the processes and machines that cause defects, avoid batch abnormalities, and improve product yield. Summary of the Invention

[0004] To solve the problem that non-characteristic defects are difficult to detect, resulting in batch abnormalities in wafers and reduction of product yield, the present invention provides a defect detection method.

[0005] The defect detection method provided by the present invention includes:

[0006] Performing electron microscope scanning on the abnormal positions on the wafer surface to obtain scanning images of the abnormal positions, and each scanning image of the abnormal position includes at least a first scanning image and a second scanning image;

[0007] Identify the scanned images of abnormal positions, and classify the abnormal positions into true defects and interference defects according to the first classification criterion;

[0008] Perform line processing on the first scanned image and the second scanned image of the interference defect respectively to obtain the corresponding first line-processed picture and second line-processed picture;

[0009] Compare the first line-processed picture with the second line-drawn picture, and classify the interference defect into false defect and non-characteristic defect according to the comparison result;

[0010] Give an early warning for non-characteristic defects.

[0011] Among them, the first classification criterion includes: if a certain part of the first scanned image or the second scanned image of the abnormal position shows a characteristic defect structure, the abnormal position is classified as a true defect, otherwise the abnormal position is classified as an interference defect; the characteristic defect structure includes at least one of unexpected spots, patches, pits, protrusions, scratches, deformations, color differences, and defects.

[0012] Among them, classifying the interference defect into false defect and non-characteristic defect according to the comparison result includes: superimposing and canceling the first line-processed picture and the second line-processed picture to obtain a line-processed defect picture corresponding to the interference defect, and classifying the interference defect into false defect and non-characteristic defect according to the second classification criterion.

[0013] Among them, the second classification criterion includes: when the line-processed defect picture shows lines, the interference defect corresponding to the line-processed defect picture is classified as a non-characteristic defect; when the line-processed defect picture does not show any lines, the interference defect corresponding to the line-processed defect picture is classified as a false defect.

[0014] Among them, the electron microscope scanning includes scanning electron microscope scanning.

[0015] Among them, the first scanned image is: the scanned image obtained by scanning the abnormal position with an initial electron beam at a first incident angle; the second scanned image is: the scanned image obtained by scanning the abnormal position with an initial electron beam at a second incident angle different from the first incident angle, and the other scanning conditions and parameters for obtaining the first scanned image and the second scanned image are the same, and the incident angle is defined as the angle between the axis of the electron beam and the normal of the wafer surface.

[0016] Among them, the first incident angle is 0° to 15°, and the second incident angle is 45° to 60°.

[0017] Among them, the method for obtaining the abnormal position includes: scanning the wafer surface with a defect scanner to obtain the signal value of each position, and if the signal value exceeds a predetermined threshold, it is an abnormal value, and the position corresponding to the abnormal value is the abnormal position.

[0018] Among them, the surface of the wafer has a trench structure, and the abnormal position is located on or inside the surface of the trench structure.

[0019] Among them, the backgrounds of the first and second line images are white, and the lines are black.

[0020] Compared with the prior art, the present invention has the following advantages:

[0021] The non-feature defect detection method provided by the present invention generates corresponding first and second line images by performing image post-processing on the first and second scanned images of interference defects, and compares the first line image with the second line image, so as to visually and accurately perform secondary identification of non-feature defects, improve the recognition accuracy of non-feature defects, avoid wafer batch abnormalities caused by misclassification of non-feature defects, and improve the product yield. Description of the Drawings

[0022] Figure 1 is a flowchart of the defect detection method provided in an embodiment;

[0023] Figure 2 is an arrow diagram of the defect detection method provided in an embodiment;

[0024] Figures 3A to 3E are respectively the first scanned image, the second scanned image, the first line image, the second line image, and the line defect image of the first abnormal position in an embodiment;

[0025] Figures 4A to 4D are respectively the first scanned image, the second scanned image, the first line image, and the second line image of the second abnormal position in an embodiment.

[0026] The description of the reference numerals is as follows:

[0027] 1 - trench structure; 10 - faint imprint; 11 - abnormal line; 12 - line structure; 2 - normal structure. Detailed Embodiments

[0028] To make the objectives, advantages, and features of the present invention clearer, the following further details the non-feature defect detection method provided by the present invention with reference to the accompanying drawings. It should be noted that the drawings are in very simplified forms and use non-precise scales, only for conveniently and clearly assisting in explaining the objectives of the embodiments of the present invention.

[0029] Please refer to Figure 1 , Figure 1 which is a flowchart of the defect detection method provided in this embodiment. The defect detection method includes the following steps:

[0030] Step S1: Electron microscope scanning is performed on the abnormal positions on the wafer surface to obtain scanning images of the abnormal positions. The scanning image of each abnormal position includes at least a first scanning image and a second scanning image;

[0031] Step S2: Identify the scanning images of the abnormal positions, and classify the abnormal positions as true defects and interference defects according to the first classification criterion;

[0032] Step S3: Perform line processing on the first scanning image and the second scanning image of the interference defect respectively to obtain corresponding first line-processed pictures and second line-processed pictures;

[0033] Step S4: Compare the first line-processed picture with the second line-processed picture, and classify the interference defect as a false defect and a non-characteristic defect according to the comparison result;

[0034] Step S5: Give an early warning for the non-characteristic defect.

[0035] Hereinafter, in conjunction with Figure 2 , Figures 3A to 3E and Figures 4A to 4D each of the above steps will be described in detail, where Figure 2 is the arrow diagram of the defect detection method provided in an embodiment; Figure 3A ~Figure 3E are respectively the first scanning image, the second scanning image, the first line-processed picture, the second line-processed picture and the line-processed defect picture of the first abnormal position in an embodiment; Figures 4A to 4D are respectively the first scanning image, the second scanning image, the first line-processed picture and the second line-processed picture of the second abnormal position in an embodiment.

[0036] Execute Step S1, perform electron microscope scanning on the abnormal positions on the wafer surface to obtain the scanning images of each abnormal position. The scanning image of each abnormal position includes at least a first scanning image and a second scanning image.

[0037] In this embodiment, the method for obtaining the abnormal position is as follows: Use a defect scanner to scan the surface of the wafer to obtain the signal value at the corresponding position. When the signal value exceeds a predetermined threshold, it is an abnormal value, and the position corresponding to the abnormal value is the abnormal position, where the predetermined threshold can be set according to actual process requirements. The electron microscope scan is performed using a scanning electron microscope (SEM). The first scan image is an SEM scan image obtained by scanning the initial electron beam of the SEM at a first incident angle under certain conditions. The second scan image is an SEM scan image obtained by scanning the initial electron beam of the SEM at a second incident angle under the same conditions. The incident angle is defined as the angle between the axis of the electron beam and the normal of the wafer surface. In this description, the wafer surface is an ideal plane obtained by ignoring local unevenness. Preferably, the first incident angle is 0° to 15°. At this time, the electron beam is nearly perpendicular to the wafer surface, and the corresponding scan image can collect the morphological information of the flat area on the wafer surface. In this embodiment, the first incident angle is 0°. The second incident angle is 45° to 60°. At this time, there is a large angle between the electron beam and the normal of the wafer surface, and the corresponding scan image can collect more morphological information of the uneven area on the wafer surface. In this embodiment, the second incident angle is 45°. Selecting appropriate first and second incident angles can ensure that the obtained first and second scan images can comprehensively display the surface morphological characteristics of the corresponding abnormal position. When the abnormal position is on the surface or inside the trench structure on the wafer surface, the corresponding first scan image contains more morphological information of the trench bottom, and the corresponding second scan image contains more morphological information of the trench sidewall, and the background gray levels of the first scan image and the second scan image are different, so there is also a large difference in image contrast. In this embodiment, the scan images corresponding to two abnormal positions are used for comparison to better illustrate the advantages and features of the defect detection method provided by the present invention. As Figure 3A and Figure 3B shown, after scanning the first abnormal position, the corresponding first scan image and second scan image are obtained. The first abnormal position is on the surface or inside the trench structure 1; as Figure 4A and 4B shown, the first scan image and the second scan image obtained after scanning the second abnormal position are respectively shown. The second abnormal position is in a normal structure 2, so the scan images of the second abnormal position all show a normal structure without characteristic defect structures.

[0038] Execute step S2, identify the scanned image of the abnormal position, and classify the abnormal position into a true defect and an interference defect according to the first classification criterion. The first classification criterion includes: if a certain part of the first scanned image or the second scanned image of the abnormal position shows a characteristic defect structure, then classify the abnormal position as a true defect; otherwise, classify the abnormal position as an interference defect. The characteristic defect structure includes at least one of unexpected spots, patches, pits, protrusions, scratches, deformations, color differences, and defects. In this embodiment, the identification and classification of the abnormal position are performed by a yield engineer. The specific process is as follows: The yield engineer observes and compares the first scanned image and the second scanned image corresponding to a certain abnormal position. If the first scanned image or the second scanned image shows a characteristic defect structure, then classify the corresponding abnormal position as a true defect. These characteristic defect structures generally correspond to some real defects, such as particles, chemical residue, glue overflow, scratches, chip defects, etc. If neither the first scanned image nor the second scanned image shows a characteristic defect structure, then classify the corresponding abnormal position as an interference defect.

[0039] Figure 3B shows a normal structure, but in Figure 3A a local area shows an indistinct imprint 10. The structural features shown by this indistinct imprint 10 are not obvious mainly because the image contrast is too low, that is, the brightness contrast between the indistinct imprint 10 and the surrounding structure is too low, so the structural features of the indistinct imprint 10 are masked. After the engineer's initial identification, the first abnormal position is generally classified as an interference defect. In the prior art, when classifying this abnormal position as an interference defect, this abnormal position is also determined as a false defect at the same time, and there is no subsequent procedure to detect the non-characteristic defect represented by the indistinct imprint 10, resulting in misclassification and causing batch abnormalities. To prevent non-characteristic defects such as those shown in FIG. 3A from being misclassified as false defects, in this embodiment, the interference defect will be secondarily identified and classified by means of image post-processing. In Figure 4A and 4B no characteristic defect structure is shown. Therefore, after the engineer's identification, the corresponding abnormal position will be classified as an interference defect.

[0040] Please refer to Figure 2, after classifying the abnormal positions, the wafers or chips with true defects will normally enter the yield engineer handle process. This online processing process includes counting, analyzing the number of defects, and notifying the corresponding machine tools to conduct defect investigation. The processing method in this step is the same as that in the prior art. In the prior art, the interference defects are not further detected, and the wafers or chips with the interference defects are regarded as normal wafers or chips for normal release processing; while in this embodiment, the interference defects will be further detected in order to detect the non-characteristic defects from the interference defects.

[0041] Execute step S3, perform line drawing processing on the first scan image and the second scan image of the interference defect respectively to obtain the corresponding first line drawing picture and second line drawing picture. The line drawing processing belongs to a common method of image post-processing, which can simplify the image blocks in the image into lines and obtain a picture with a white background and black lines. Since the background grayscales of the first scan image (for example Figure 3A ) and the second scan image (for example Figure 3B ) are different, it is difficult to perform image comparison processing; after line drawing processing, the generated first line drawing picture and second line drawing picture are both with a white background and black lines, which can facilitate subsequent comparison processing (such as superposition and cancellation processing) of the line drawing pictures. The line drawing processing is preferably performed in the data processing unit equipped with the existing detection machine tool.

[0042] Please refer to Figure 3C , Figure 3D , Figure 4C and Figure 4D , where Figure 3C and Figure 3D are respectively Figure 3A and Figure 3B the line drawing pictures obtained after performing the line drawing processing, that is, Figure 3C and Figure 3D are respectively the first line drawing picture and the second line drawing picture of the first abnormal position; where Figure 4C and Figure 4D are respectively Figure 4A and Figure 4B the line drawing pictures obtained after performing the line drawing processing, that is, Figure 4C and Figure 4D are respectively the first line drawing picture and the second line drawing picture of the second abnormal position. Figure 3C , Figure 3D , Figure 4C and Figure 4D generally show the line frameworks corresponding to the abnormal positions. Figure 3A The image block showing the unclear imprint 10 in Figure 3Cis displayed as abnormal line 11, while Figure 3D , Figure 4C and Figure 4D do not display any abnormal line 11.

[0043] Execute step S4, compare the first line-drawn picture with the second line-drawn picture, and classify the interference defects into false defects and non-feature defects according to the comparison result.

[0044] The specific method for comparing the first line-drawn picture with the second line-drawn picture is as follows: superimpose and cancel the first line-drawn picture and the second line-drawn picture to obtain a line-drawn defect picture corresponding to the interference defect, and classify the interference defect into false defects and non-feature defects according to the second classification criterion. Since the first line-drawn picture and the second line-drawn picture correspond to the line-drawn pictures of the same position structure, they contain substantially the same information and similar line-drawn pictures. By performing the picture superimposing and canceling process commonly used in the post-processing of the first line-drawn picture and the second line-drawn picture, a line-drawn defect picture can be obtained. According to this line-drawn defect picture and the second classification criterion, the interference defect can be classified into false defects and non-feature defects. Preferably, the second classification criterion is: when the line-drawn defect picture shows lines, the corresponding interference defect is classified as a non-feature defect; when the line-drawn defect picture does not show any lines, the corresponding interference defect is classified as a false defect.

[0045] As Figure 3E shown, it is the line-drawn defect picture of the first abnormal position. Figure 3E In addition to showing a white background, it also shows several lines. Therefore, according to the above second classification criterion, the first abnormal position can be classified as a non-feature defect, and Figure 3E the lines in it form a certain line structure 12, and this line structure 12 corresponds to the two-dimensional structure of this non-feature defect. In contrast, the line-drawn defect picture of the second abnormal position only shows a white background and does not show any lines. Therefore, the second abnormal position can be classified as a false defect according to the above second classification criterion. For wafers or chips in which all abnormal positions are classified as false defects, the yield engineer performs normal shipping processing. For wafers or chips with non-feature defects, corresponding warning processing needs to be performed.

[0046] Execute step S5 to give a warning for the non-feature defect. Through the above warning, remind the engineer to perform a secondary confirmation on the non-feature defect. Specifically, the engineer can analyze by combining the above line-drawn defect picture and the SEM image corresponding to the non-feature defect, determine the structure and type of the non-feature defect, evaluate the impact of the non-feature defect, and decide whether to perform subsequent defect troubleshooting work.

[0047] In summary, for a defect detection method provided by the present invention, after the abnormal positions are firstly identified and classified, the scanned images of interfering defects are subjected to image post-processing to generate corresponding first and second line images. By comparing the first line image and the second line image, the interfering defects can be secondly identified and classified, and non-characteristic defects can be detected therefrom. Therefore, the detection method provided by the present invention can conveniently and intuitively secondly identify non-characteristic defects, improve the recognition accuracy of the non-characteristic defects, avoid the batch abnormality of wafers caused by the misclassification of the non-characteristic defects, and improve the product yield.

[0048] In addition, it can be understood that although the present invention has been disclosed above with preferred embodiments, the above embodiments are not intended to limit the present invention. For any person skilled in the art, without departing from the scope of the technical solution of the present invention, many possible changes and modifications can be made to the technical solution of the present invention by using the technical content disclosed above, or it can be modified into an equivalent embodiment with equivalent changes. Therefore, any simple modification, equivalent change and modification made to the above embodiments according to the technical essence of the present invention without departing from the content of the technical solution of the present invention still fall within the scope of protection of the technical solution of the present invention. It should also be understood that the present invention is not limited to the specific methods, compounds, materials, manufacturing techniques, uses and applications described herein, and they can be changed. It should also be understood that the terms described herein are only used to describe specific embodiments, rather than to limit the scope of the present invention. It must be noted that the singular forms "a", "an" and "the" used herein and in the appended claims include plural references unless the context clearly dictates otherwise. Thus, for example, a reference to "a step" means a reference to one or more steps and may include sub-steps. All conjunctions used should be understood in their broadest sense. Thus, the word "or" should be understood to have the defined logical "or" rather than the logical "exclusive or" definition unless the context clearly dictates otherwise. The structures described herein will be understood to also refer to functional equivalents of such structures. Language that may be construed as approximate should be so understood unless the context clearly dictates otherwise.

Claims

1. A defect detection method, characterized in that, Including the following steps: Performing electron microscope scanning on abnormal positions on the wafer surface to obtain scanning images of the abnormal positions, and the scanning images of each abnormal position include at least a first scanning image and a second scanning image; Identifying the scanning images of the abnormal positions and classifying the abnormal positions into true defects and interference defects according to a first classification criterion; Performing lineation processing on the first scanning image and the second scanning image of the interference defects respectively to obtain corresponding first lineated pictures and second lineated pictures; Comparing the first lineated picture with the second lineated picture and classifying the interference defects into false defects and non-feature defects according to the comparison result; Giving an early warning for the non-feature defects; The electron microscope scanning includes scanning electron microscope scanning; 2. The defect detection method according to claim 1, wherein The first scanning image is: a scanning image obtained by scanning an abnormal position with an initial electron beam at a first incident angle; the second scanning image is: a scanning image obtained by scanning an abnormal position with the initial electron beam at a second incident angle different from the first incident angle, and other scanning conditions and parameters for obtaining the first scanning image and the second scanning image are the same, and the incident angle is defined as the angle between the axis of the electron beam and the normal line of the wafer surface.

3. The defect detection method according to claim 1, wherein The first classification criterion includes: if a certain part of the first scanning image or the second scanning image of the abnormal position shows a characteristic defect structure, the abnormal position is classified as a true defect, otherwise the abnormal position is classified as an interference defect; the characteristic defect structure includes at least one of unexpected spots, patches, pits, protrusions, scratches, deformations, color differences, and defects.

4. The defect detection method according to claim 3, wherein Classifying the interference defects into false defects and non-feature defects according to the comparison result includes: superimposing and canceling the first lineated picture and the second lineated picture to obtain a lineated defect picture corresponding to the interference defect, and classifying the interference defect into false defects and non-feature defects according to a second classification criterion.

5. The defect detection method according to claim 1, wherein The second classification criterion includes: when the lineated defect picture shows lines, the interference defect corresponding to the lineated defect picture is classified as a non-feature defect; when the lineated defect picture does not show any lines, the interference defect corresponding to the lineated defect picture is classified as a false defect.

6. The defect detection method according to claim 1, characterized in that The first incident angle is 0° to 15°, and the second incident angle is 45° to 60°.

7. The defect detection method according to claim 1, wherein The method for obtaining the abnormal positions includes: scanning the wafer surface with a defect scanner to obtain the signal value of each position, and if the signal value exceeds a predetermined threshold, it is an abnormal value, and the position corresponding to the abnormal value is the abnormal position.

8. The defect detection method according to claim 1, characterized in that, The wafer surface has a trench structure, and the abnormal positions are located on the surface or inside of the trench structure. The backgrounds of the first lineated picture and the second lineated picture are white, and the lines are black.

Images