Wafer bump surface defect self-correlation detection method and device and storage medium

Abstract

The present application relates to wafer detection technical field, and relates to a kind of wafer bump surface defect autocorrelation detection method, device and storage medium.The method has the following steps: S1, obtaining wafer image to be detected;S2, for any pixel size uniform background image, obtain corresponding image reduction set;S3, in the order of zooming factor reduction, the reduced image in image reduction set is detected in sequence, and when detecting the defect in current reduced image, end detection and determine that there is defect in corresponding uniform background image, when all reduced images in image reduction set do not exist defect, determine that there is no defect in corresponding uniform background image;S4, obtain all uniform background images with defect and as the surface defect detection result set of wafer image to be detected.The device and storage medium can implement the above method.The present application can preferably improve the robustness of defect detection.

Description

Technical Field

[0001] This invention relates to the field of wafer inspection technology, and more specifically, to an autocorrelation detection method, apparatus, and storage medium for wafer bump surface defects. Background Technology

[0002] Currently, when detecting surface defects on wafer bumps using visual inspection, a region of interest (ROI) is typically defined based on a bump map and a gold template. This ROI serves as the detection template. The acquired image of the wafer to be inspected is then directly matched with the detection template, and the surface defect detection of the wafer bumps is achieved based on the matching result. (See also...) Figure 1 This is a gold template that marks the region of interest.

[0003] When acquiring images of wafers for inspection, illumination is achieved using light sources such as bright and dark fields. The background of the acquired wafer images varies under different lighting conditions. This leads to detection errors in existing direct image matching methods. Specifically: when matching a wafer image with a light background to a gold template with a dark background, defect features in the wafer image may disappear, resulting in missed detections; conversely, when matching a wafer image with a dark background to a gold template with a light background, over-detection may occur, misclassifying good wafers as defective. All of these factors contribute to a decrease in the accuracy of defect detection.

[0004] Seen in Figure 2 During wafer inspection, due to the limited field of view of the lens, the entire wafer image is actually pieced together. This means that even under the same ambient lighting, there will be background differences between images of the wafer being inspected acquired in a single session. This further complicates the accuracy control of conventional inspection methods. Figure 2 In the diagram, 'a' is a wafer image, 'b' is a die image, and 'c' is a bump image. Summary of the Invention

[0005] This invention provides an autocorrelation detection method for surface defects of wafer bumps, which can overcome some or more defects of the prior art.

[0006] According to the present invention, an autocorrelation detection method for surface defects of wafer bumps comprises the following steps:

[0007] S1. Obtain the image A of the wafer to be inspected, A = {I} k |k∈N +}; where I k Let N be the k-th uniform background image in the wafer image A to be inspected. + It is a positive integer;

[0008] S2. For any uniform background image I with pixel size [P, Q] k Obtain the corresponding image reduction set B. Among them, I kn For a uniform background image I k A scaled-down image of size n, where P is a uniform background image I. k The row size, Q is the uniform background image I. k The column size, n is a natural number, and m is the maximum scaling factor;

[0009] S3. Shrink the reduced images I in image set B in descending order of scaling factor. kn Perform defect detection and reduce the size of the image I. kn When a defect is detected, the detection process ends and the corresponding uniform background image I is determined. k There is a defect in the image set B, specifically in the case of all resized images I. kn When there are no defects, the corresponding uniform background image I is determined. k There are no defects in it;

[0010] S4. Obtain all uniform background images containing defects. k It is used as a set of surface defect detection results for the wafer image A to be inspected.

[0011] This method can improve detection speed by performing stepwise defect detection on a scaled-down uniform background image, and improve detection accuracy by performing autocorrelation detection on individual wafer images with uniform backgrounds, thereby improving robustness.

[0012] Preferably, the value of m is set to the uniform background image I after being reduced m times. k The pixel size is no less than 10. Therefore, it can better ensure a uniform background image I at the maximum reduction factor. k It can still exhibit defective characteristics relatively well.

[0013] Preferably, in step S3, when processing any reduced image I... kn When performing defect detection, the following steps should be followed:

[0014] S31. Obtain each pixel I knl The grayscale value (i, j);

[0015] S32, For any pixel I knl Obtain the I of any given pixel. knl The surrounding r pixels and any pixel I knl The grayscale value deviation;

[0016] S33. Obtain the number s of pixels whose grayscale value deviation exceeds the set deviation threshold among the r pixels, and calculate the proportion of pixels exceeding the limit T, T = s / r;

[0017] S34. When the proportion T of pixels exceeding the limit exceeds the set proportion threshold, determine the value of any pixel I. knl For defective pixels, simultaneously determine any reduced image I kn It has defects.

[0018] Through the above, it is possible to better utilize any scaled-down image I kn The pixel deviation value between any pixel and its surrounding pixels enables autocorrelation defect detection, thus better avoiding the inability to guarantee detection accuracy due to background differences between the wafer image to be inspected and the inspection template.

[0019] Preferably, in step S31, any pixel I is used knl Using the center point as the reference point, select any pixel I. knl The pixels within a range of h rows and h columns are considered as any given pixel I. knl The system has r surrounding pixels. Therefore, it can effectively obtain the grayscale value difference between any pixel and its surrounding pixels.

[0020] Preferably, h = m - n + 1, where m is the maximum scaling factor and n is the current scaled-down image I. kn The magnification factor is [value missing]. Therefore, it can better consider the overall size of the reduced image and achieve the discovery of the largest defect features.

[0021] As a preferred option, the deviation threshold is set to 25.

[0022] As a preferred option, the ratio threshold is set to 0.2.

[0023] Furthermore, the present invention also provides an autocorrelation detection device for wafer bump surface defects, including a memory and a processor. The memory stores a computer program, and when the processor executes the computer program, it implements the steps of any of the autocorrelation detection methods described above.

[0024] Furthermore, the present invention provides a computer-readable medium having a computer program stored thereon, wherein the computer program, when executed, implements the steps of any of the autocorrelation detection methods described herein. Attached Figure Description

[0025] Figure 1 This is a schematic diagram of a standard gold template;

[0026] Figure 2 This is a schematic diagram of a standard wafer under inspection.

[0027] Figure 3 This is a flowchart illustrating an autocorrelation detection method for surface defects on wafer bumps, as described in Example 1. Detailed Implementation

[0028] To further understand the content of this invention, the invention will be described in detail with reference to the embodiments. It should be understood that the embodiments are merely illustrative and not limiting of the invention.

[0029] Example 1

[0030] Seen in Figure 3 This embodiment provides an autocorrelation detection method for surface defects on wafer bumps, which includes the following steps:

[0031] S1. Obtain the image A of the wafer to be inspected, A = {I} k |k∈N +}; where I k Let N be the k-th uniform background image in the wafer image A to be inspected. + It is a positive integer;

[0032] S2. For any uniform background image I with pixel size [P, Q] k Obtain the corresponding image reduction set B. Among them, I kn For a uniform background image I k A scaled-down image of size n, where P is a uniform background image I. k The row size, Q is the uniform background image I. k The column size, n is a natural number, and m is the maximum scaling factor;

[0033] S3. Shrink the reduced images I in image set B in descending order of scaling factor. kn Perform defect detection and reduce the size of the image I. kn When a defect is detected, the detection process ends and the corresponding uniform background image I is determined. k There is a defect in the image set B, specifically in the case of all resized images I. kn When there are no defects, the corresponding uniform background image I is determined. k There are no defects in it;

[0034] S4. Obtain all uniform background images containing defects. k It is used as a set of surface defect detection results for the wafer image A to be inspected.

[0035] The method in this embodiment can improve the detection speed by performing stepwise defect detection on a scaled-down uniform background image, and improve the detection accuracy by performing autocorrelation detection on the wafer image under inspection with a uniform background, thereby improving robustness.

[0036] Among them, the wafer image A to be inspected can be obtained based on the detection template, that is, it can be obtained by filtering and other processing on the region in the acquired image that corresponds to the position of the detection template.

[0037] In this context, a uniform background image refers to an image acquired in a single snapshot. It can be understood that a complete wafer image is composed of multiple snapshot images stitched together; therefore, snapshot images can be considered to have essentially the same background.

[0038] In this embodiment, the value of m is set to the uniform background image I after being reduced m times. k The pixel size is no less than 10 (meaning that the pixel size of both rows and columns is no less than 10). Therefore, it can better ensure a uniform background image I at the maximum reduction factor. k It can still exhibit defective characteristics relatively well.

[0039] In step S3 of this embodiment, when processing any scaled-down image I... kn When performing defect detection, the following steps should be followed:

[0040] S31. Obtain each pixel I knl The grayscale value (i, j);

[0041] S32, For any pixel I knl Obtain the I of any given pixel. knl The surrounding r pixels and any pixel I knl The grayscale value deviation;

[0042] S33. Obtain the number s of pixels whose grayscale value deviation exceeds the set deviation threshold among the r pixels, and calculate the proportion of pixels exceeding the limit T, T = s / r;

[0043] S34. When the proportion T of pixels exceeding the limit exceeds the set proportion threshold, determine the value of any pixel I. knl For defective pixels, simultaneously determine any reduced image I kn It has defects.

[0044] Through the above, it is possible to better utilize any scaled-down image I kn The pixel deviation value between any pixel and its surrounding pixels enables autocorrelation defect detection, thus better avoiding the inability to guarantee detection accuracy due to background differences between the wafer image to be inspected and the inspection template.

[0045] In step S31 of this embodiment, any pixel I knl Using the center point as the reference point, select any pixel I. knlThe pixels within a range of h rows and h columns are considered as any given pixel I. knl The system has r surrounding pixels. Therefore, it can effectively obtain the grayscale value difference between any pixel and its surrounding pixels.

[0046] Where h = m - n + 1, m is the maximum scaling factor, and n is the current scaled-down image I. kn The magnification factor is [value missing]. Therefore, it can better consider the overall size of the reduced image and achieve the discovery of the largest defect features.

[0047] The deviation threshold can be set to, for example, 25.

[0048] The set ratio threshold can be, for example, 0.2.

[0049] In addition, this embodiment also provides an autocorrelation detection device for wafer bump surface defects, including a memory and a processor. The memory stores a computer program, and when the processor executes the computer program, it implements the steps of the above-described autocorrelation detection method.

[0050] Furthermore, this embodiment also provides a computer-readable medium having a computer program stored thereon, characterized in that the computer program, when executed, implements the steps of the above-described autocorrelation detection method.

[0051] It is readily understood that those skilled in the art can combine, split, or reorganize the embodiments provided in this application to obtain other embodiments, all of which do not exceed the protection scope of this application.

[0052] The present invention and its embodiments have been described above illustratively. This description is not restrictive, and the embodiments shown are only part of the embodiments of the present invention. The actual structure is not limited thereto. Therefore, if those skilled in the art are inspired by this description and design similar structures and embodiments without departing from the spirit of the present invention, they should all fall within the protection scope of the present invention.

Claims

1. An autocorrelation detection method for surface defects on wafer bumps, comprising the following steps: S1. Obtain image A of the wafer to be inspected. ;in, This is the k-th uniform background image in the wafer image A to be inspected. It is a positive integer; S2. For any uniform background image with pixel size [P, Q] Obtain the corresponding image reduction set B. ;in, Uniform background image A scaled-down image of size n, where P is a uniform background image. The row size, Q is the uniform background image. The column size, n is a natural number, and m is the maximum scaling factor and not less than 1; S3. Shrink the images in set B in descending order of scaling factor. Perform defect detection and reduce the image size as needed. When a defect is detected, the detection process ends and the corresponding uniform background image is determined. There is a defect in the image set B, which contains all the scaled-down images. When there are no defects, the corresponding uniform background image is determined. There are no defects in it; S4. Obtain all uniform background images containing defects. It is used as a set of surface defect detection results for the wafer image A to be inspected.

2. The autocorrelation detection method for wafer bump surface defects according to claim 1, characterized in that: The value of m is set to the uniform background image after being reduced m times. The pixel size is no less than 10.

3. The autocorrelation detection method for wafer bump surface defects according to claim 1, characterized in that: In step S3, for any reduced image When performing defect detection, the following steps should be followed: S31. Obtain each pixel. The grayscale values ​​(i, j); S32, For any pixel point Obtain each pixel point The surrounding r pixels and any pixel The grayscale value deviation; S33. Obtain the number s of the r pixels whose grayscale value deviation exceeds the set deviation threshold, and calculate the proportion of pixels exceeding the threshold T, T=s / r; S34. When the proportion T of pixels exceeding the limit exceeds the set proportion threshold, determine the value of any pixel. For defective pixels, determine the size of any reduced image. It has defects.

4. The autocorrelation detection method for wafer bump surface defects according to claim 3, characterized in that: In step S31, using any one of the pixels... Using the center point as the reference point, select any pixel point. The pixels within a range of h rows and h columns are considered as any given pixel. The surrounding r pixels.

5. The autocorrelation detection method for wafer bump surface defects according to claim 4, characterized in that: h = m - n + 1, where m is the maximum scaling factor and n is the currently scaled-down image. The scaling factor.

6. The autocorrelation detection method for wafer bump surface defects according to claim 3, characterized in that: Set the deviation threshold to 25.

7. The autocorrelation detection method for wafer bump surface defects according to claim 3, characterized in that: Set the ratio threshold to 0.

2.

8. An autocorrelation detection device for surface defects of wafer bumps, comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, it implements the steps of the autocorrelation detection method according to any one of claims 1 to 7.

9. A computer-readable medium having a computer program stored thereon, characterized in that, When the computer program is executed, it implements the steps of the autocorrelation detection method according to any one of claims 1 to 7.

Images