Incremental semiconductor image stitching and navigation method based on physical coordinate constraints

Abstract

The application provides a kind of incremental semiconductor image splicing and navigation method based on physical coordinate constraint, by utilizing the space-time sequence generated by automatic control flow, construct the visual navigation prior model with physical scale constraint, including the following operation logic: when the machine table moves according to the spiral search path, the machine table motion trajectory is synchronously associated and stored to form a map library;When the machine table fails to align, based on the map library, the machine table performs the operation logic of first-stage global coarse positioning and second-stage vector jump to find the alignment mark;When the machine table jumps to the specified coordinate in the map, but the environment changes, the map is automatically iteratively updated.The method of the application improves the efficiency and accuracy of machine table alignment, and realizes effective accumulation and utilization of information in the search process.

Description

Technical Field

[0001] This invention relates to the field of semiconductor manufacturing technology, and in particular to an incremental semiconductor image stitching and navigation method based on physical coordinate constraints. Background Technology

[0002] Global alignment (GA) of semiconductor equipment is a crucial step in semiconductor manufacturing processes such as photolithography, metrology, and etching. The process involves the equipment following a fixed path to locate and align alignment marks (GA marks). Existing equipment follows a fixed spiral path (e.g., Figure 1 When the machine moves along the 1-25 grid path to find the Mark, all previously searched images 1-24 and their coordinate information are discarded after the Mark is found. When the alignment fails again (GA fail), the machine needs to retrace the 1-25 path, wasting time and resources. Furthermore, the search process is blind and does not effectively utilize the search information. Summary of the Invention

[0003] The purpose of this invention is to provide an incremental semiconductor image stitching and navigation method based on physical coordinate constraints, which aims to solve the problems of resource waste and low efficiency in the machine alignment process in the prior art, and realize the transformation from blind search to efficient and accurate positioning based on memory navigation.

[0004] To achieve the above objectives, the technical solution of the present invention utilizes the spatiotemporal sequence generated by Automated Control Flow (RPA) to construct a visual navigation prior model with physical scale constraints, thereby improving the efficiency and accuracy of machine alignment and effectively accumulating and utilizing information during the search process.

[0005] This invention provides an incremental semiconductor image stitching and navigation method based on physical coordinate constraints, characterized by comprising: when the machine moves along a spiral search path, synchronously associating and storing the machine's movement trajectory to form a map library; when the machine fails to align, based on the map library, the machine performs a first-stage global coarse positioning and a second-stage vector jump operation logic to find the alignment mark; when the machine jumps to a specified coordinate in the map, but the environment changes, the map is automatically iteratively updated.

[0006] Furthermore, the incremental semiconductor image stitching and navigation method based on physical coordinate constraints provided by this invention may also have the following features: the synchronous association and storage of the machine tool's motion trajectory includes: during the machine tool's movement along a fixed spiral search path, when the machine tool receives a movement command and moves to the command position, an image corresponding to that position is acquired; a unique spatial coordinate index is established for each position in the spiral search path, and the spatial coordinate index is used to associate and store the coordinate information of that position with the image acquired at that position.

[0007] Furthermore, the incremental semiconductor image stitching and navigation method based on physical coordinate constraints provided by the present invention may also have the following features: wherein, the first stage of global coarse positioning includes: when the machine fails to align, acquiring an image of the current location, extracting features from the acquired image, matching features with each image stored in the map library, and if the match is successful, considering the coordinate position corresponding to the image as the possible location of the alignment mark.

[0008] Furthermore, the incremental semiconductor image stitching and navigation method based on physical coordinate constraints provided by the present invention may also have the following feature: wherein, the second-stage vector jump includes: when the feature matches a certain location in the map library, the coordinate difference between the current coordinates of the computer station and the coordinates of the matched location is used to drive the computer station to jump to the possible coordinates of the alignment mark with one click.

[0009] Furthermore, the incremental semiconductor image stitching and navigation method based on physical coordinate constraints provided by the present invention may also have the following features: environmental changes are determined by the following steps: when the machine jumps to a specified coordinate in the map, an image corresponding to that location is acquired; the acquired image is compared with the stored image corresponding to that location in the map library for similarity; when the similarity is lower than a set similarity threshold, it is determined that the environment at that location has changed.

[0010] Furthermore, the incremental semiconductor image stitching and navigation method based on physical coordinate constraints provided by this invention may also have the following feature: the automatic iterative map update includes: when the environment changes, automatically marking the area of ​​environmental change as Invalid, and re-triggering the machine to search according to a fixed spiral search path through the RPA control flow to realize the automatic iterative update of the map.

[0011] Furthermore, the incremental semiconductor image stitching and navigation method based on physical coordinate constraints provided by this invention may also have the following features: the method is implemented through an incremental semiconductor image stitching and navigation system based on physical coordinate constraints, which includes: a movement command generation module for generating machine movement commands; a shooting command generation module for generating machine shooting commands; an association module for associating coordinate information with acquired images; a map database module for storing images with associated coordinate information and constructing a map database; a feature extraction module for extracting features from images; a retrieval and matching module for matching extracted image features with images in the map database; a coordinate difference calculation module for calculating the difference between coordinates; a similarity judgment module for judging the similarity of images through image information entropy or consistency score, and determining environmental change when the similarity is lower than a threshold; a map iteration update module for automatically triggering the iteration update process when the environment changes; and a data transmission module for data transmission with the machine.

[0012] Compared with the prior art, the present invention has the following beneficial effects:

[0013] 1. Information Accumulation: This invention treats the trajectory traversed during the search as a "survey", and the search process is a modeling process. The coordinates of each grid in the spiral search path are correlated with the image in real time and serialized to construct a spatiotemporally correlated local search map, thereby achieving effective information accumulation.

[0014] 2. Highly efficient and accurate positioning: This invention adopts a two-stage jump logic that combines "coarse and fine" positioning. When the alignment fails again, it first performs global coarse positioning by quickly searching through feature fingerprints, and then realizes coordinate vector jump based on the matching results, instead of searching again, which greatly improves positioning efficiency and accuracy.

[0015] 3. Adaptability to complex situations: This invention proposes a non-rigid registration algorithm under coordinate constraints to solve the field distortion problem when the SEM machine moves over a large area; it judges environmental changes and realizes automatic iterative updates of the map, enabling the system to better adapt to complex and ever-changing working environments. Attached Figure Description

[0016] Figure 1 It is a schematic diagram of the machine moving along a spiral path of 1-25 grids.

[0017] Figure 2 This is a functional block diagram of the incremental semiconductor image stitching and navigation system based on physical coordinate constraints in an embodiment of the present invention. Detailed Implementation

[0018] To make the technical means, creative features, objectives and effects of the present invention easy to understand, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings.

[0019] Please see Figure 2 This embodiment provides an incremental semiconductor image stitching and navigation system based on physical coordinate constraints. The system operates through RPA (Robotic Process Automation) and includes the following functional modules: a movement command generation module, a shooting command generation module, an association module, a map database module, a feature extraction module, a retrieval and matching module, a similarity judgment module, a map iterative update module, and a data transmission module. The movement command generation module generates machine movement commands; the shooting command generation module generates machine shooting commands; the association module associates coordinate information with acquired images; the map database module stores images with associated coordinate information, constructing a map library; the feature extraction module extracts features from the images; the retrieval and matching module matches extracted image features with images in the map library; the coordinate difference calculation module calculates the difference between coordinates; the similarity judgment module judges the similarity of images based on their information entropy or consistency score, and determines environmental changes when the similarity is below a threshold; the map iterative update module automatically triggers an iterative update process when the environment changes; and the data transmission module transmits data with the machine.

[0020] This embodiment provides an incremental semiconductor image stitching and navigation method based on physical coordinate constraints, including:

[0021] 1. Track synchronization and association:

[0022] The movement command generation module generates movement commands for the machine to move to each position according to a fixed spiral search path. The movement commands are transmitted to the machine through the data transmission module to control the movement of the machine.

[0023] As the machine moves along a fixed spiral search path, each time it receives a movement command and reaches the designated position, the data transmission module acquires the coordinate information transmitted by the machine. At this time, the image capture command generation module generates an image capture command, which is transmitted to the machine via the data transmission module to control image acquisition. The machine receives the image capture command and captures one image at each position. For example, when the machine moves to a position such as… Figure 1 When the image is at position 13 (as shown), the image corresponding to position 13 is simultaneously captured.

[0024] The data transmission module acquires images captured by the machine, and the association module associates coordinate information with the acquired images: each grid cell is assigned a unique spatial coordinate index, which associates the coordinate information of that location with the image captured at that location. The map database module stores the images with associated coordinate information, constructing a map database.

[0025] Through the above-mentioned trajectory synchronization and association steps, even if the machine does not find the alignment mark during this search, these synchronously acquired images constitute a local "mini-map," providing basic data for subsequent positioning.

[0026] 2. Two-level jump combining "coarse-fine" approach:

[0027] First stage: Global coarse localization

[0028] When the machine fails to align, the data transmission module receives the GA fail signal transmitted from the machine, the shooting command generation module generates a shooting command for the machine, and the shooting command is transmitted to the machine through the data transmission module, and the machine takes an image of the current position.

[0029] The data transmission module acquires the currently captured image transmitted from the machine. The feature extraction module extracts features from the image using a predetermined image extraction algorithm (an existing algorithm can be used). The retrieval and matching module performs a fast feature matching by directly searching the image features extracted from the currently captured image against the image fingerprints of each grid image in the map database. If a match is successful, the coordinates corresponding to that image are considered to be the possible location of the alignment marker.

[0030] Second-stage vector jump:

[0031] When a feature match is successful, the coordinate difference calculation module calculates the coordinate difference between the current coordinates of the computer platform and the matched position. Based on this coordinate difference, the movement command generation module drives the machine to jump to the possible coordinate position of the alignment mark with a single keystroke. The movement command is transmitted to the machine via the data transmission module to control its movement.

[0032] 3. Automatic map updates:

[0033] When the machine jumps to the specified coordinates on the map but the environment changes, the map is automatically updated iteratively. The specific process is as follows:

[0034] When the camera jumps to the specified coordinates on the map, the data transmission module receives the coordinate information transmitted by the camera, the shooting instruction generation module generates a shooting instruction for the camera, and the shooting instruction is transmitted to the camera through the data transmission module, and the camera takes an image of the current location.

[0035] The similarity judgment module compares the captured image with the images stored in the map. By calculating the information entropy or consistency score between the captured image and the corresponding image in the map, when the score is lower than a set threshold, it is determined that the environment at that location has changed (Wafer layer change, large contrast drift, etc.).

[0036] When an environmental change is detected, the map iteration update module automatically triggers the iteration update process: it automatically marks the area of ​​environmental change as Invalid, and re-triggers the logical process of the machine searching according to a fixed spiral search path, thereby realizing the automatic iteration update of the map.

[0037] Although the present invention has been described in detail with reference to the foregoing embodiments, it is obvious that the described embodiments are merely preferred embodiments of the present invention, and not all embodiments. For those skilled in the art, modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. An incremental semiconductor image stitching and navigation method based on physical coordinate constraints, characterized in that, include: When the machine moves along the spiral search path, the machine's movement trajectory is synchronously associated and stored to form a map library; When the machine fails to align, it uses the map library to perform a first-stage global coarse positioning and a second-stage vector jump operation to find the alignment mark. When the machine jumps to the specified coordinates on the map, but the environment changes, the map is automatically iterated and updated.

2. The incremental semiconductor image stitching and navigation method based on physical coordinate constraints as described in claim 1, characterized in that: in, Synchronous association and storage of machine tool motion trajectories includes: As the machine moves along a fixed spiral search path, it acquires an image corresponding to each position it receives and moves to. A unique spatial coordinate index is established for each position in the spiral search path. The spatial coordinate index is a storage that associates the coordinate information of the position with the image acquired at that position.

3. The incremental semiconductor image stitching and navigation method based on physical coordinate constraints as described in claim 1, characterized in that: in, The first stage of global coarse localization includes: when the machine fails to align, acquiring an image of the current location, extracting features from the acquired image, matching the features with images stored in the map library, and if the match is successful, considering the coordinate position corresponding to the image as the possible location of the alignment mark.

4. The incremental semiconductor image stitching and navigation method based on physical coordinate constraints as described in claim 3, characterized in that: in, The second stage of vector jump includes: when a feature matches a location in the map library, the difference between the current coordinates of the computer station and the coordinates of the matched location is used to drive the computer station to jump to the possible coordinates of the alignment mark with one click.

5. The incremental semiconductor image stitching and navigation method based on physical coordinate constraints as described in claim 1, characterized in that: in, The environmental changes are determined through the following steps: When the machine jumps to the specified coordinates on the map, it captures an image corresponding to that location. The collected image is compared with the stored image corresponding to that location in the map database; When the similarity is lower than the set similarity threshold, it is determined that the location environment has changed.

6. The incremental semiconductor image stitching and navigation method based on physical coordinate constraints as described in claim 1, characterized in that: in, The automatic iterative map update includes: when the environment changes, automatically marking the area of ​​environmental change as Invalid, and re-triggering the machine to search according to a fixed spiral search path through the RPA control process to achieve automatic iterative map update.

7. The incremental semiconductor image stitching and navigation method based on physical coordinate constraints as described in any one of claims 1 to 6, characterized in that, This method is implemented through an incremental semiconductor image stitching and navigation system based on physical coordinate constraints, which includes: The movement instruction generation module is used to generate machine movement instructions; The shooting instruction generation module is used to generate shooting instructions for the machine. The association module is used to associate coordinate information with the acquired image; The map database module is used to store images with associated coordinate information and build a map library; The feature extraction module is used to extract features from the image; The retrieval and matching module is used to match the extracted image features with the features of images in the map database; The coordinate difference calculation module is used to calculate the difference between coordinates; The similarity judgment module determines the similarity of images based on their information entropy or consistency score. When the similarity is below a threshold, it is determined to be an environmental change. The map iteration update module automatically triggers the iteration update process when the environment changes; The data transmission module is used for data transmission between the machine and the equipment.

Images