Alignment method of AOI (Automatic Optic Inspection) machine table before and after wafer cutting

Abstract

The invention relates to the technical field of semiconductor manufacturing, and provides an AOI machine alignment method before and after wafer cutting in order to solve the problem of alignment offset caused by dislocation of a wafer on a film. The obtained jigsaw image is subjected to binarization processing, and then the alignment mark is grabbed, so that the alignment mark exists in the jigsaw image, the alignment success is improved, and the error detection risk is reduced.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of semiconductor manufacturing. BACKGROUND

[0002] In the production of different batches of wafers, wafer film sticking is a key preparation process before the semiconductor packaging process. After completing the previous process (manufacturing chip circuits on the wafer), and after thinning and grinding, a special polymer film, usually a blue film or a UV film, is precisely attached to the back of the wafer, providing mechanical support for the wafer, facilitating subsequent wafer cutting and picking. Due to the ductility of the film and the slight difference in the position of the wafer, the X and Y directions of the wafers in different batches are offset on the film, and the position of the wafer on the AOI machine workbench is deviated, resulting in a failure of position alignment.

[0003] Currently, a one-template alignment scheme is used: an AOI machine is used to identify a preset alignment mark, which is located on the scribe line of the wafer and serves as the coordinate system for the entire wafer. Each corner or specific position of the wafer has a unique design pattern (such as an L-shaped mark or a cross line). By capturing the mark on the wafer, the captured image is compared with the preset standard image, and it must be completely consistent to pass. This method can result in incorrect photos of some offset wafers, leading to alarms and low production efficiency. SUMMARY

[0004] The purpose of the present application is to solve the problem of wafer alignment offset caused by film sticking on the film. The present application provides an alignment method for an AOI machine before and after wafer cutting, comprising the following steps: Finding the alignment mark retained on the current wafer in the FOV field of view of the AOI machine, calculating the offset and rotation of the entire wafer, and using the offset to preliminarily correct the expected position of all wafers; After global compensation, control the camera of the machine to move to the position of the alignment mark relative to the FOV field of view; In a search area at the current position, use the mark template image to search and match, find the alignment mark in the FOV field of view of the AOI machine and create an RCP mark setting position; Take a picture at a certain distance outside the setting position, and form a gray-scale image after splicing the nine pictures taken; After binarization processing of the spliced image, capture the alignment mark.

[0005] Technical effects: This invention proposes a multi-angle, multi-image stitching alignment method. This novel alignment method improves alignment success rate, thereby reducing the risk of false tests. The method is considered successful as long as alignment marks appear in the stitched images. This improves production efficiency and solves the problem of alignment misalignment caused by dislocations on the wafer thin film. Through core logic ranging from coarse positioning to feature alignment mark identification, to outward area imaging, and then to local fine matching, the method ensures the accuracy and traceability of wafer inspection in semiconductor manufacturing. Attached Figure Description

[0006] Figure 1 This is a schematic diagram of the overall process of this embodiment.

[0007] Figure 2 This is a schematic diagram of the binarization process for a jigsaw puzzle image. Detailed Implementation

[0008] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention.

[0009] like Figure 1 As shown in the figure, this embodiment proposes an alignment method for AOI equipment before and after wafer dicing, including the following steps: Provide a wafer to be aligned, locate the alignment mark retained on the current wafer in the FOV field of view of the AOI machine, and calculate the overall offset and rotation of the wafer. The offset is used to initially correct the expected position of all wafers. After global compensation, the camera on the control unit is moved to the position of the alignment mark relative to the field of view (FOV) after the specified position is reached. Within a search area at the current location, use the marker template image for search matching, find the alignment mark in the FOV field of view of the AOI machine, and create an RCP mark to set the position; Take photos at a certain distance around the designated location, and stitch the nine photos together to form a grayscale image; preferably, the outward distance is 3mm or 5mm.

[0010] After binarizing the obtained jigsaw puzzle image, the alignment marks are extracted to ensure that the alignment marks definitely exist in the jigsaw puzzle image. For example... Figure 2 As shown, Figure 2 The leftmost image is the original FOV (Field of View) image. Alignment markers typically have a noticeable black-and-white difference from their surroundings. After binarization, the alignment marker boundaries are all 0s, and the underlying background is all 1s. 0s are defined as feature points. This clearly separates the black lines of L-shaped or crosshair markers from the white background (or oxide layer background), facilitating accurate calculation of the marker's center point coordinates, angle, and contour by the algorithm. This avoids lower recognition scores due to color differences caused by previous processes, thereby improving alignment success and reducing the risk of false positives.

[0011] What is not specifically mentioned in the present application is well known to those skilled in the art, and the above only describes certain exemplary embodiments of the present application in a manner of illustration, and it is needless to say that those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of the present application. Therefore, the above drawings and descriptions are illustrative in nature, and should not be understood as limiting the scope of protection of the claims of the present application.

Claims

1. A method for aligning an AOI (Automated Optical Inspection) machine before and after wafer dicing, characterized in that, Includes the following steps: Locate the alignment marks remaining on the current wafer within the FOV of the AOI machine, and calculate the overall offset and rotation of the wafer. The offset is used to initially correct the expected position of all wafers. After global compensation, the camera on the control unit is moved to the position of the alignment mark relative to the field of view (FOV) after the specified position is reached. Within a search area at the current location, use the marker template image for search matching, find the alignment mark in the FOV field of view of the AOI machine, and create an RCP mark to set the position; Take photos at a certain distance around the designated location, and stitch the nine photos together to form a grayscale image; The obtained jigsaw puzzle image is binarized before the alignment mark is captured.

2. The alignment method of an AOI machine before and after wafer dicing according to claim 1, characterized in that, Take a photo at a distance of 3mm or 5mm beyond the set position.

3. The alignment method of an AOI machine before and after wafer dicing according to claim 1, characterized in that, After binarizing the jigsaw puzzle image, the alignment mark boundaries are all 0, and the underlying background is all 1. 0 is defined as a feature point, thereby accurately calculating the center point coordinates, angle, and contour of the alignment mark.

Images