A sorter F champion algorithm

Abstract

The present application relates to wafer sorting technical field, and disclose a kind of sorting machine F champion algorithm, the algorithm includes the following steps: 1) loading F champion die map document, F champion die is evenly distributed on wafer, each F champion die has map row coordinate, first by the way of loading file to obtain the map row coordinate of all F champion die;2) visual positioning F champion die initial position;3) F champion die area division and management.This sorting machine F champion algorithm, to use the F champion die that appearance is inconsistent with ordinary die, cooperate direct input F champion map row coordinate, visual positioning accurate initial position and corresponding area division and management of F champion algorithm, to when film shrinkage is more serious, even if chip offset is more than half chip spacing, will be due to F champion appearance special, effectively avoid visual identification to find wrong champion die, reduce scanning champion time, realize the improvement of production efficiency.

Description

Technical Field

[0001] This invention relates to the field of wafer sorting technology, specifically to a sorting machine F-scorer algorithm. Background Technology

[0002] In the normal wafer sorting process, the ejector pins puncture the blue film, causing the film to shrink and the position of the chips on it to change. To prevent excessive chip position changes from causing missorting, the chips on the wafer need to be divided into small areas. One chip is selected from each area as a marker. Every short interval (usually 5 minutes), the positions of all markers are visually scanned to scan and inspect the wafer. However, because the position of the chips in the area will change and update with the position of the markers, sorting cannot be performed during the scanning of the markers, resulting in slow production capacity.

[0003] Furthermore, in the traditional method of using ordinary chips as benchmarks, the benchmark chips are indistinguishable from ordinary chips in appearance. Therefore, when the film shrinkage is severe and the chip position changes by more than half a chip spacing, the visual system may mistakenly identify the chips around the benchmark as the benchmark, causing all chips in the entire area to be misaligned and resulting in chaotic sorting accidents. To avoid this situation, the time interval for rescanning the benchmark must be shortened. The benchmark needs to be rescanned before the chip position shifts by half a chip spacing, which will further reduce the production efficiency. Therefore, a new algorithm that can solve this problem is needed. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides an F-marker algorithm for sorting machines, which has the advantages of using F-marker dies as markers, eliminating the need for rescanning markers during the production process, reducing marker scanning time to improve production capacity, and solving the problem of low production efficiency caused by the need for frequent rescanning of markers during the traditional method of using ordinary dies as markers.

[0005] To achieve the goal of increasing production capacity by using F-type benchmark dies as benchmarks and eliminating the need for rescanning benchmarks during the production process, thereby reducing benchmark scanning time, this invention provides the following technical solution: an F-type benchmark algorithm for a sorting machine, which includes the following steps: 1) Load the F-marker die map document. The F-marker dies are uniformly distributed on the wafer. Each F-marker die has map row and column coordinates. First, obtain the map row and column coordinates of all F-marker dies by loading the file. 2) Visually locate the initial position of the F-marker grain. After the sorting process merges the map, based on the map row and column coordinates of the F-marker grain, find the nearest merged grain within the nine-square grid range adjacent to the F-marker grain. Based on the position of this grain and the row and column interval between it and the F-marker grain, calculate the theoretical position of the F-marker grain. Then, take a picture at the theoretical position to obtain the actual axis position of the F-marker grain. 3) F-marker die region division and management: Based on the map row and column coordinates of the F-marker die, the entire wafer region is evenly divided into rectangular regions centered on the F-marker die for management.

[0006] Preferably, the specific coordinate values ​​of the F-marker crystal in step 1 are stored in a plain text file, where each line of data in the file represents the row and column coordinates of an F-marker crystal.

[0007] Preferably, in step 2, the theoretical position is photographed and imaged using a template matching method. Visual matching is performed based on the F-marker grain image template to check whether the F-marker grain can be matched. If it cannot be matched, the data row and column coordinates of the F-marker grain are recorded first.

[0008] Preferably, if visual matching can directly identify a match, the center pixel coordinates of the matched F-marker crystal are converted into the actual axis position based on the camera and axis calibration results.

[0009] Preferably, in step 2, the same method should be used to determine whether each F-marker grain has an initial actual axis position in order to ensure the accuracy of the initial actual axis position of each F-marker grain.

[0010] Preferably, after the region is divided in step 3, when the position of all grains in the divided region changes or shifts, they will uniformly follow the F-marker grain management. That is, whatever the F-marker grain shift value is, the shift value of all grains in that region will also be the same.

[0011] Preferably, step 3 further includes searching the surrounding areas above, below, left, and right of the missing F-marker grains in the wafer edge and the missing F-marker grains that were not visually matched in step 2. The area containing the F-marker grain is found in the surrounding area, and the F-marker grain closest to the center of the missing area is found in the surrounding area. Then, the missing F-marker area is merged into the area where the F-marker grain is located.

[0012] Preferably, step 3 further includes merging the missing marker region into the region where the F marker grain is located, so that the positional change of all grains in the missing region will follow the positional change of the F marker grain in the adjacent region, so that the region where there is no F marker grain can also obtain the position of the film shrinkage change, accurately complete the sorting and picking, and avoid the loss of grain yield.

[0013] Compared with existing technologies, this invention provides a sorting machine F-spot algorithm, which has the following beneficial effects: The sorting machine's F-marker algorithm uses F-marker dies with appearances different from ordinary dies. Combined with direct input of F-marker map row and column coordinates, accurate initial position for visual positioning, and corresponding area division and management, the F-marker algorithm effectively avoids visual identification of the wrong marker die even when the film shrinkage is severe and the chip is offset by more than half a chip spacing. Therefore, the special appearance of the F-marker die can effectively prevent visual identification from misidentifying the marker die. Thus, it is not necessary to rescan the marker die, reducing the scanning time and improving production efficiency. Attached Figure Description

[0014] Figure 1 This is a schematic diagram of the F-scorer algorithm for a sorting machine proposed in this invention. Detailed Implementation

[0015] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0016] Please see Figure 1 A sorting machine F-spot algorithm, which includes the following steps: 1) Load the F-marker die map document. The F-marker dies are uniformly distributed on the wafer. Each F-marker die has map row and column coordinates. First, obtain the map row and column coordinates of all F-marker dies by loading the file. The specific coordinate values ​​of the F-marker dies in step 1 will be saved in a plain text file, such as a .csv file. Each line of data in the file represents the row and column coordinates of an F-marker die. 2) Visually locate the initial position of the F-marker crystal. After the sorting process merges the map images, based on the map row and column coordinates of the F-marker crystal, find the nearest merged crystal within the adjacent nine-square grid range of the F-marker crystal. Based on the position of this crystal and the row and column intervals between it and the F-marker crystal, calculate the theoretical position of the F-marker crystal. Then, take a picture at this theoretical position. In step 2, the theoretical position is photographed and captured by template matching. Visual matching is performed based on the F-marker crystal image template to check if the F-marker crystal can be matched. If it cannot be matched, record the data row and column coordinates of the F-marker crystal. If the visual matching can directly identify the match, convert the center pixel coordinates of the matched F-marker crystal into the actual axis position according to the camera and axis calibration results to obtain the actual axis position of the F-marker crystal. In step 2, the same method should be used to determine whether there is an initial actual axis position for each F-marker crystal to ensure the accuracy of the initial actual axis position of each F-marker crystal. 3) F-marker die region division and management: Based on the map row and column coordinates of the F-marker die, the entire wafer area is uniformly divided into rectangular regions centered on the F-marker die for management. After the region division in step 3, any positional shifts of all dies within the divided region will uniformly follow the F-marker die management; that is, the offset value of the F-marker die will be the offset value of all dies within that region. Step 3 also includes regions on the wafer edge without F-marker dies, and missing marker regions where F-marker dies were not visually matched in step 2, by adjusting their positions in all directions. The search is conducted in the surrounding area to find the region where the F-marker crystal exists. The F-marker crystal closest to the center of the missing region is found in the surrounding area, and then its missing region is merged into the region where the F-marker crystal is located. Step 3 also includes the fact that after the missing region is merged into the region where the F-marker crystal is located, the positional change of all crystals in the missing region will follow the positional change of the F-marker crystal in the surrounding area. This allows the position of the film shrinkage change to be obtained even in areas where there is no F-marker crystal, so as to accurately complete the sorting and picking and avoid the loss of crystal yield.

[0017] In summary, the F-marker algorithm of this sorting machine uses F-marker dies with appearances different from ordinary dies. Combined with the direct input of F-marker map row and column coordinates, accurate initial position for visual positioning, and corresponding area division and management, the F-marker algorithm effectively avoids visual identification of the wrong marker die even when the film shrinkage is severe and the chip is offset by more than half a chip spacing. Therefore, the special appearance of the F-marker die can effectively prevent visual identification from misidentifying the marker die. Thus, it is not necessary to rescan the marker die, reducing the scanning time and improving production efficiency.

[0018] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one" does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0019] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A sorting machine F-spot algorithm, characterized in that: The algorithm includes the following steps: 1) Load the F-marker die map document. The F-marker dies are uniformly distributed on the wafer, and each F-marker die has map row and column coordinates. The specific coordinate values ​​of the F-marker dies are stored in a plain text file, where each line of data in the file represents the row and column coordinates of an F-marker die. First, obtain the map row and column coordinates of all F-marker dies by loading the file. 2) Visually locate the initial position of the F-marker grain. After the sorting process merges the map, based on the map row and column coordinates of the F-marker grain, find the nearest merged grain within the nine-square grid range adjacent to the F-marker grain. Based on the position of this grain and the row and column interval between it and the F-marker grain, calculate the theoretical position of the F-marker grain. Then, take a picture at the theoretical position to obtain the actual axis position of the F-marker grain. The theoretical position image acquisition is achieved through template matching. Visual matching is performed based on the F-marker grain image template to check if the F-marker grain can be matched. If no match is found, the data row and column coordinates of the F-marker grain are recorded. If the visual matching can directly identify the match, the center pixel coordinates of the matched F-marker grain are converted into the actual axis position according to the camera and axis calibration results. The same method needs to be used to determine whether there is an initial actual axis position for each F-marker grain in order to ensure the accuracy of the initial actual axis position of each F-marker grain. 3) F-marker die region division and management: Based on the map row and column coordinates of the F-marker die, the entire wafer region is evenly divided into rectangular regions centered on the F-marker die for management.

2. The sorting machine F-spot algorithm according to claim 1, characterized in that: After the region is divided in step 3, when the position of all the grains in the divided region changes or shifts, they will uniformly follow the F-marker grain management. That is, whatever the F-marker grain offset value is, the offset value of all the grains in that region will also be the same.

3. The sorting machine F-spot algorithm according to claim 2, characterized in that: Step 3 also includes searching the surrounding areas around the wafer edge where there are no F-marker dies and the missing marker areas where no F-marker dies were visually matched in Step 2. The area containing the F-marker die is found in the surrounding area, and the F-marker die closest to the center of the missing area is found in the surrounding area. Then the missing marker area is merged into the area where the F-marker die is located.

4. The sorting machine F-spot algorithm according to claim 3, characterized in that: Step 3 also includes merging the missing marker region into the region where the F marker grain is located. The positional changes of all grains in the missing region will follow the positional changes of the F marker grains in the adjacent region. This allows the region without F marker grains to obtain the position of the film shrinkage change, accurately complete the sorting and picking, and avoid grain yield loss.

Images