Wafer transmission mechanical finger surface evaluation method and system
By using multimodal data registration and fusion analysis, the problem of single-dimensional surface evaluation of wafer transfer mechanical fingers has been solved, enabling root cause diagnosis and lifetime prediction of surface defects, improving the depth and foresight of the evaluation, and supporting high-yield production in semiconductor manufacturing.
Patent Information
- Application Number
- CN202511653517.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-11-12
- Publication Date
- 2026-02-06
AI Technical Summary
Traditional methods for evaluating the surface of wafer transfer mechanical fingers have a single dimension, which cannot fully reveal the root cause of defects and makes it difficult to predict performance degradation, thus affecting the product yield and equipment stability of semiconductor manufacturing.
A multimodal dataset is collected, and a spatial coordinate system is established and registered using a three-dimensional geometric datum. By combining static and dynamic evaluations, the surface condition is analyzed in a fusion manner to achieve multidimensional evaluation.
It enables root cause diagnosis and lifetime prediction of surface defects, enhancing the depth and foresight of assessments and providing support for predictive maintenance and high-yield production in semiconductor manufacturing.
Smart Images
Figure CN121479196A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of equipment surface evaluation technology, and in particular to a method and system for evaluating the surface of a wafer transfer mechanical finger. Background Technology
[0002] Wafer transfer robotic fingers are critical components in semiconductor front-end equipment that directly contact the wafer. Their surface condition directly affects product yield and equipment stability. Microscopic particles, adsorbed chemical contaminants, or existing surface defects on the finger surface can directly transfer to the wafer surface, leading to serious problems such as short circuits and device failures, resulting in significant economic losses. Traditional evaluation methods mainly rely on single technical means, such as visual inspection with optical microscopy or roughness measurement. These methods suffer from limitations such as a single evaluation dimension, inability to reveal the root cause of defects, and a focus on static parameter measurement, making it difficult to predict performance degradation. As semiconductor process nodes continue to shrink, higher demands are placed on the surface condition evaluation of robotic fingers. Therefore, there is an urgent need for a comprehensive and intelligent surface evaluation method for wafer transfer robotic fingers. Summary of the Invention
[0003] This invention provides a method and system for evaluating the surface of a wafer-transfer mechanical finger, which can effectively solve the problems in the background art.
[0004] To achieve the above objectives, the technical solution adopted by the present invention is as follows: A method for evaluating the surface of a wafer-mounted mechanical finger, the method comprising: Image data and quantitative parameters of the surface of the wafer-transmitted robotic finger were collected to obtain a multimodal dataset; A spatial coordinate system is established based on the three-dimensional geometric reference of the wafer transfer mechanical finger. The wafer contact area is determined in the spatial coordinate system, and the multimodal dataset is registered on the spatial coordinate system. Correlation analysis is performed on the registered multimodal data to obtain the correlation analysis results. Based on the quantitative parameters and the correlation analysis results, a static diagnosis is performed on the surface of the wafer transfer mechanical finger to obtain the static diagnosis results. Under the simulation conditions of wafer transfer action, the particle behavior and friction behavior of the wafer transfer mechanical finger are evaluated to obtain dynamic evaluation results; The static diagnostic results and the dynamic evaluation results are fused and analyzed to generate the surface evaluation results of the wafer-transfer mechanical finger.
[0005] Furthermore, the acquisition of image data and quantitative parameters from the surface of the robotic finger via wafer transmission, resulting in a multimodal dataset, includes: The wafer transfer mechanical finger is divided into regions, and a sampling grid is established for each region; Two-dimensional and three-dimensional image data are collected within the sampling grid area, and quantitative parameters are collected at the sampling grid locations. The two-dimensional image data, the three-dimensional image data, and the quantitative parameters are associated and stored based on the sampling grid to construct a multimodal dataset.
[0006] Further, registering the multimodal dataset in the spatial coordinate system includes: Determine the initial positioning reference and align the two-dimensional image data to the spatial coordinate system; The three-dimensional image data is projected and its features are aligned with those of the two-dimensional image data. The acquisition location of the quantitative parameters is mapped to the spatial coordinate system and spatially bound to the two-dimensional image data; The registration error of the two-dimensional image data, the three-dimensional image data, and the quantitative parameters is verified.
[0007] Furthermore, spatial binding with the two-dimensional image data includes: generating a continuous distribution map of the quantitative parameters using an interpolation method and overlaying it onto the two-dimensional image data.
[0008] Furthermore, the correlation analysis of the registered multimodal data includes: Within the contact area of the wafer, features are extracted from the multimodal data, and spatial colocation analysis is performed on the multimodal data based on the extracted features to obtain spatial distribution results; Correlation analysis is performed on the extracted features, the correlation coefficient is calculated, and the correlation analysis results are obtained.
[0009] Furthermore, static diagnostics of the surface of the wafer transfer mechanical finger includes: Collect historical datasets, which include historically acquired image data and quantitative parameters, as well as corresponding surface evaluation results; Construct a static diagnostic model, and train and optimize the static diagnostic model using the historical dataset; The quantitative parameters and the correlation analysis results are subjected to feature filtering. The filtered results are then input into the static diagnostic model for diagnosis, generating static diagnostic results.
[0010] Furthermore, particle behavior and friction behavior evaluations are performed on the wafer-level transport mechanical finger to obtain dynamic evaluation results, including: The particle release and adsorption behaviors of the wafer-transfer mechanical finger were tested and repeated to obtain the particle behavior results. The friction coefficient and wear characteristics of the wafer-level mechanical finger were tested and repeated to obtain the results of its friction behavior. The particle behavior results and the friction behavior results are integrated to form a dynamic evaluation result.
[0011] Furthermore, a method for evaluating the surface of a wafer-transfer mechanical finger, characterized in that the method includes: The data acquisition module collects image data and quantitative parameters from the surface of the wafer-transmitted robotic finger to obtain a multimodal dataset; The data analysis module establishes a spatial coordinate system based on the three-dimensional geometric reference of the wafer-transfer mechanical finger, determines the wafer contact area in the spatial coordinate system, registers the multimodal dataset on the spatial coordinate system, performs correlation analysis on the registered multimodal data, and obtains the correlation analysis results. The static diagnostic module performs static diagnostics on the surface of the wafer transfer mechanical finger based on the quantitative parameters and the correlation analysis results, and obtains static diagnostic results. The dynamic evaluation module evaluates the particle behavior and friction behavior of the wafer transfer mechanical finger under the simulation conditions of wafer transfer action, and obtains the dynamic evaluation results. The fusion analysis module integrates the static diagnostic results and the dynamic evaluation results to generate the surface evaluation results of the wafer-transfer mechanical finger.
[0012] Furthermore, the data acquisition module includes: The region division unit divides the wafer transfer mechanical finger into regions and establishes a sampling grid for each region; The data acquisition unit acquires two-dimensional image data and three-dimensional image data within the sampling grid area, and acquires quantitative parameters at the sampling grid locations; The associated storage unit stores the two-dimensional image data, the three-dimensional image data, and the quantitative parameters based on the sampling grid to construct a multimodal dataset.
[0013] Furthermore, the dynamic evaluation module includes: The particle behavior testing unit tests the particle release and adsorption behavior of the wafer-transfer mechanical finger and performs repeated tests to obtain particle behavior results. The friction behavior testing unit tests the friction coefficient and wear characteristics of the wafer-level mechanical finger and performs repeated tests to obtain the friction behavior results. The result integration unit integrates the particle behavior results and the friction behavior results to form a dynamic evaluation result. The technical solution of this invention achieves the following technical effects: It effectively solves the problem of a single evaluation dimension. Through multimodal data registration and correlation analysis, it realizes the root cause diagnosis of surface defects. By combining static and dynamic evaluation, it integrates static and dynamic data to achieve a comprehensive evaluation from phenomenon description to root cause diagnosis and lifetime prediction. It significantly improves the depth and foresight of the evaluation and provides strong support for predictive maintenance and high-yield production of semiconductor manufacturing equipment.
[0014] The above description is only an overview of the technical solution of this application. In order to better understand the technical means of this application and to implement it in accordance with the contents of the specification, and to make the above and other objects, features and advantages of this application more obvious and understandable, the following are specific embodiments of this application. Attached Figure Description
[0015] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0016] Figure 1 This is a flowchart illustrating a method for evaluating the surface of a wafer-transfer mechanical finger. Figure 2 A flowchart illustrating the process of obtaining correlation analysis results; Figure 3 This is a schematic diagram of a wafer transfer mechanical finger surface evaluation system. Detailed Implementation
[0017] The technical solutions 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.
[0018] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.
[0019] Example 1 like Figure 1 As shown, a method for evaluating the surface of a wafer-level mechanical finger is provided, the method comprising: S1: Acquire image data and quantitative parameters of the surface of the wafer-transmitted mechanical finger to obtain a multimodal dataset; Specifically, to avoid the limitations of a single data source, this step collects image data and quantitative parameters of the surface of the wafer transfer robotic finger. The image data is used to visually reflect the defects, particles, and morphology information of the wafer transfer robotic finger surface. This step can acquire image data through industrial cameras, optical microscopes, etc. The quantitative parameters are used to numerically characterize the physical and chemical properties of the robotic finger surface, including surface roughness parameters, flatness parameters, electrical parameters, chemical cleanliness parameters, etc. The acquired image data and quantitative parameters form a multimodal dataset, which can comprehensively reflect the macroscopic defects, microscopic defects, particle distribution, and physicochemical properties of the wafer transfer robotic finger surface, providing a reliable data foundation for subsequent registration analysis, correlation analysis, and static diagnosis.
[0020] S2: Establish a spatial coordinate system based on the three-dimensional geometric reference of the wafer-transfer mechanical finger, determine the wafer contact area in the spatial coordinate system, register the multimodal dataset in the spatial coordinate system, perform correlation analysis on the registered multimodal data, and obtain the correlation analysis results; To address the spatial inconsistency between different data modalities, this embodiment performs registration on the acquired multimodal data. Image data is typically represented by pixel coordinates, while quantitative parameters are represented by instrument sampling points or region values. These differ in scale, resolution, and coordinate system; without registration, a correspondence cannot be established. Therefore, this embodiment uses feature point matching, geometric correction, pixel resampling, or a registration algorithm based on mutual information to align image data and quantitative parameters to the same spatial coordinate system, making data from different sources comparable and superimposed in terms of location. The spatial coordinate system is established with the three-dimensional geometric reference of the wafer transport robotic finger as a reference. This three-dimensional geometric reference is obtained from the CAD model, laser scanning model, or high-precision measurement data of the robotic finger. The establishment of the spatial coordinate system provides a unified spatial reference for data acquired by different detection devices. The wafer contact area refers to the area where the wafer directly or nearly contacts the surface of the robotic finger during transport. It can be determined based on the robotic finger structure, wafer diameter, adsorption pore distribution, and wafer positioning method. Determining the wafer contact area limits the evaluation area, allowing subsequent analysis to focus only on key areas that actually affect wafer cleanliness and handling reliability. After registration is completed, multimodal data can be correlated in the contact area of the wafer through methods such as spatial colocation analysis, threshold-triggered correlation, and correlation analysis. This reveals the correspondence and coupling law between defects, particle distribution and quantitative parameters in the image. It not only provides intuitive image evidence, but also establishes the connection between quantitative parameters and defect distribution, providing a more comprehensive and reliable data foundation for subsequent static diagnosis.
[0021] S3: Based on quantitative parameters and correlation analysis results, static diagnosis is performed on the surface of the wafer transfer mechanical finger to obtain static diagnosis results; Based on the above embodiments, in order to transform the collected and analyzed multimodal data into conclusions that can be used for surface quality assessment, this embodiment can use threshold determination methods, multi-index comprehensive scoring and other methods to diagnose quantitative parameters and correlation analysis results. It can simultaneously obtain numerical benchmark support and spatial positioning information, thereby achieving a more comprehensive and accurate static diagnosis, providing a reliable data foundation and risk warning information for subsequent dynamic assessment and comprehensive surface assessment.
[0022] S4: Under the simulation conditions of wafer transfer action, the particle behavior and friction behavior of the wafer transfer mechanical finger are evaluated to obtain dynamic evaluation results; Specifically, particle behavior and friction behavior are both dynamic degradation characteristics that occur in wafer-to-wafer transport robotic fingers during wafer transport, which cannot be fully revealed by static detection methods. This step dynamically evaluates particle behavior and friction behavior in an experimental or simulation environment. Particle behavior evaluation characterizes the particle release of the robotic finger during operation. This can be achieved by monitoring and statistically analyzing particle behavior using a particle monitoring device in a clean environment, or by judging the strength of particle release behavior through surface comparison tests. Friction behavior evaluation characterizes the friction and wear characteristics of the robotic finger when it moves relative to the wafer contact surface. This can be measured using a tribological testing machine or a simulated transport device, or by predicting friction behavior using tribological modeling methods. Incorporating both particle behavior and friction behavior into the dynamic evaluation not only supplements the deficiencies of static diagnosis but also enables the prediction of the reliability and service life of the robotic finger.
[0023] S5: Combine and analyze static diagnostic results and dynamic evaluation results to generate surface evaluation results for the wafer-transfer mechanical finger.
[0024] Specifically, static diagnostic results reflect the basic attributes of the robotic finger surface, such as defect distribution, roughness, smoothness, electrical properties, and cleanliness. Dynamic evaluation results reflect the particle release behavior and frictional wear characteristics of the robotic finger during actual operation. This step involves cross-validation and comprehensive calculation to form a fusion analysis result, which in turn generates surface evaluation results, including surface condition diagnosis, conformity assessment, defect risk level, and lifespan prediction. This fusion analysis avoids the limitations of single results, enabling a comprehensive evaluation of the robotic finger's surface condition and performance. It allows for complementary verification; defects discovered in static diagnosis can be verified in dynamic evaluation to determine if they truly lead to particle release or frictional failure. High particle release rates or frictional anomalies observed in dynamic evaluation can be traced back to defect areas discovered in static diagnosis, improving the reliability of the analysis. Simultaneously, it supports risk grading and lifespan prediction, making the evaluation results more aligned with the needs of actual engineering applications.
[0025] This invention effectively solves the problem of a single evaluation dimension. Through multimodal data registration and correlation analysis, it enables root cause diagnosis of surface defects. By combining static and dynamic evaluation, it integrates static and dynamic data to achieve a comprehensive evaluation from phenomenon description to root cause diagnosis and lifetime prediction, significantly improving the depth and foresight of the evaluation and providing strong support for predictive maintenance and high-yield production of semiconductor manufacturing equipment.
[0026] As a preferred embodiment, image data and quantitative parameters of the surface of the wafer-transmitted mechanical finger are acquired to obtain a multimodal dataset including: S11: Divide the wafer transfer mechanical finger into regions and establish a sampling grid for each region; Based on the structural characteristics of the mechanical finger, its surface can be divided into functional areas such as a support area, an adsorption hole area, an edge area, and a transition area. A regular sampling grid can be further established in each functional area. The sampling grid can be a rectangular grid or a square grid. The density and size of the grid can be set according to the importance of the area and the detection requirements. For example, a higher density grid can be used in the edge area of the adsorption hole. Through area division and grid establishment, a unified spatial reference system can be provided for subsequent image acquisition and quantitative parameter acquisition, thereby ensuring the alignment and traceability of data from different modalities.
[0027] S12: Collect two-dimensional and three-dimensional image data within the sampling grid area, and collect quantitative parameters at the sampling grid location; To ensure a comprehensive assessment of the surface condition of the mechanical finger being transferred, the acquired image data preferably includes two-dimensional and three-dimensional image data. Two-dimensional image data can efficiently acquire macroscopic morphological information of large-area surfaces and can be obtained through industrial cameras, optical microscopes, and scanning electron microscopes to cover macroscopic defects, microscopic scratches, and particle distribution. Three-dimensional image data can be acquired through atomic force microscopy or laser confocal microscopy to reflect the surface height distribution and roughness characteristics. At specific locations on the sampling grid, quantitative parameter data, including surface roughness, flatness, electrical parameters, and chemical cleanliness, are further acquired. These image data and quantitative parameter data correspond to the sampling grid, thereby ensuring the spatial consistency of data from different modalities.
[0028] S13: Associate and store two-dimensional image data, three-dimensional image data, and quantitative parameters based on a sampling grid to construct a multimodal dataset.
[0029] During data storage, the spatial coordinates of the aforementioned sampling grid are used as an index to bind and store the two-dimensional images, three-dimensional topographic data, and quantitative parameters corresponding to the same grid location, thereby forming a multimodal dataset with spatial correlation. The multimodal dataset not only contains intuitive image information but also contains precise numerical parameters, which can provide complete data support for subsequent spatial registration, correlation analysis, and static diagnosis, significantly improving the accuracy and reliability of surface assessment.
[0030] Based on the above embodiments, such as Figure 2 As shown, registering the multimodal dataset in a spatial coordinate system includes: S21: Determine the initial positioning reference and align the two-dimensional image data to the spatial coordinate system; Before performing multimodal data fusion, based on the established spatial coordinate system with the three-dimensional geometric reference of the wafer-transfer robotic finger as the reference, structural feature points such as the center of the adsorption hole, edge points, or positioning holes on the surface of the robotic finger are selected as calibration points. On this basis, the coordinates of the corresponding calibration points are identified in the two-dimensional image data, and the acquired two-dimensional image data is aligned to the spatial coordinate system. Through translation, rotation, and scaling operations based on the geometric reference, the initial alignment of the two-dimensional image in the spatial coordinate system can be achieved, thereby providing a global reference for subsequent multimodal registration and ensuring that the two-dimensional image data has spatial consistency in a unified coordinate system.
[0031] S22: Project the 3D image data and align its features with the 2D image data; Specifically, for three-dimensional image data, the three-dimensional image can be projected into a two-dimensional form such as a surface contour map or a surface gradient distribution map. By extracting common features such as surface edges, scratches, adsorption pore locations, or particle distribution in the three-dimensional projection view and the two-dimensional image, feature matching or mutual information optimization methods are used to align the features of the three-dimensional image and the two-dimensional image, so that the surface morphology information can correspond to the macroscopic image information, ensuring the accurate positioning of the three-dimensional image in a unified coordinate system.
[0032] S23: Map the acquisition location of quantitative parameters to a spatial coordinate system and spatially bind it with two-dimensional image data; S24: Verify the registration error of two-dimensional image data, three-dimensional image data, and quantitative parameters.
[0033] Quantitative parameters are collected at sampling grid locations. This step establishes a correspondence between the sampling grid and the spatial coordinate system, mapping the collection point of each quantitative parameter to a specific coordinate location in the spatial coordinate system. These parameters are then spatially bound to the registered 2D image, displaying or overlaying the parameter's numerical information at the corresponding grid location in the 2D image. This achieves the combination of quantitative numerical data and 2D image data, ensuring spatial correspondence between image and numerical information and supporting subsequent multimodal comprehensive analysis. After completing the registration of the 2D image, 3D image, and quantitative parameters, the registration accuracy can be verified. By detecting the reference point residual, feature overlap, or mutual information index, the registration error can be quantitatively evaluated, improving the reliability of the registration results and ensuring the credibility of multimodal data fusion analysis.
[0034] Furthermore, spatial binding with two-dimensional image data includes generating a continuous distribution map of quantitative parameters using interpolation methods and overlaying it onto the two-dimensional image data.
[0035] Specifically, quantitative parameters are discrete values collected at the sampling grid locations of the robotic finger. These parameters are collected only at a limited number of points and cannot directly reflect the continuous distribution of the entire surface. The sampling point data can be processed by interpolation methods, including bilinear interpolation, cubic spline interpolation, kriging interpolation, or surface fitting algorithms based on machine learning, so as to transform the discrete quantitative parameter values into a continuous distribution map covering the entire sampling area. The continuous distribution map is then superimposed on the registered two-dimensional image through methods such as direct overlay, transparency fusion, or color mapping to form a comprehensive effect of image and parameters.
[0036] As a preferred embodiment, the correlation analysis of the registered multimodal data includes: S25: Within the contact area of the wafer, feature extraction is performed on the multimodal data, and spatial colocation analysis is conducted on the multimodal data based on the extracted features to obtain the spatial distribution results; S26: Perform correlation analysis on the extracted features, calculate the correlation coefficient, and obtain the correlation analysis results.
[0037] Information such as the number of surface particles, defect area, grayscale contrast, and texture features are extracted from two-dimensional image data. Information such as surface height distribution, roughness, and morphological undulation parameters are extracted from three-dimensional image data. Combined with a continuous distribution map generated from quantitative parameters, the differences in feature distribution in different regions are compared under a unified spatial coordinate system. The differences between defective and normal regions in various feature values are identified, thus obtaining spatial distribution results such as regional difference maps and heat maps of particle and roughness regions. At the same time, for the image features, three-dimensional features, and quantitative parameters of the same sampling grid or region, Pearson correlation coefficient, Spearman rank correlation coefficient, or mutual information metric can be used to calculate the correlation between different modal data, quantify the relationship between each modal data, and obtain correlation analysis results such as correlation matrix and abnormal regions, providing input features for subsequent diagnosis.
[0038] Furthermore, static diagnostics of the surface of the wafer-transfer mechanical finger includes: S31: Collect historical datasets, which include historically acquired image data, quantitative parameters, and corresponding surface evaluation results; S32: Construct a static diagnostic model and train and optimize it using historical datasets; S33: Perform feature filtering on quantitative parameters and correlation analysis results, input the filtered results into the static diagnostic model for diagnosis, and generate static diagnostic results.
[0039] Specifically, historical multimodal data and corresponding surface assessment results of the surface of the wafer-transfer robotic finger are collected. Spatial distribution characteristics, correlation indicators, and abnormal area markers are obtained after correlation analysis of the multimodal data. Surface assessment results can be surface assessment levels, surface defect judgments, cleanliness level judgments, or surface quality scores obtained based on manual judgment or standard testing methods, and are used as labels for model training. A static diagnostic model is constructed using the collected historical dataset. The static diagnostic model can adopt a classification model or a regression model. Through training and cross-validation on the historical dataset, the model parameters are optimized to accurately reflect the patterns of surface assessment. In the actual diagnostic process, new robotic finger surface data are processed, quantitative parameters and correlation analysis results are extracted, and feature selection is performed. The selected features are input into the static diagnostic model, and the model outputs static diagnostic results.
[0040] Based on the above embodiments, particle behavior and friction behavior evaluations were performed on the wafer transfer mechanical finger, and the dynamic evaluation results included: S41: Test the particle release and adsorption behavior of the wafer-transfer mechanical finger, and repeat the test to obtain the particle behavior results; Specifically, the particle behavior of the robotic finger was evaluated by placing it in a standard cleanroom environment and driving it to perform multiple simulated transmission actions. The concentration of airborne particles was detected using an online particle counter to obtain the number of particles released by the robotic finger during the action. At the same time, a laser scattering detection system was used to detect the particle distribution on the test wafer surface to evaluate the adsorption characteristics of particles by the robotic finger. The above operations were repeated multiple times under the same conditions to reduce random errors and improve data reliability. Finally, the particle behavior results were obtained, including particle release rate, particle adsorption rate, and particle distribution characteristics.
[0041] S42: Test the friction coefficient and wear characteristics of the wafer-level mechanical finger, and repeat the test to obtain the results of the friction behavior; This embodiment uses a tribological testing machine to simulate the actual friction state of a mechanical finger with a wafer or corresponding contact substrate. The friction coefficient of the mechanical finger material surface is measured, and wear volume and wear track morphology data are collected under certain loads and cycles. At the same time, an infrared thermal imager is used to monitor the temperature rise during the friction process. The above tests are also carried out under repeated experimental conditions to ensure the accuracy and repeatability of the results. Finally, the friction behavior results are obtained, including friction coefficient, wear rate, wear track morphology, and tribothermal characteristics.
[0042] S43: Integrate particle behavior results and friction behavior results to form dynamic evaluation results.
[0043] Example 2 Based on the same inventive concept as the method described in the foregoing embodiments, this invention also provides a wafer-transfer mechanical finger surface evaluation system, such as... Figure 3 As shown, the system includes: The data acquisition module collects image data and quantitative parameters from the surface of the wafer-transmitted robotic finger to obtain a multimodal dataset; The data analysis module establishes a spatial coordinate system based on the three-dimensional geometric reference of the wafer-transfer mechanical finger, determines the wafer contact area in the spatial coordinate system, registers the multimodal dataset in the spatial coordinate system, performs correlation analysis on the registered multimodal data, and obtains the correlation analysis results. The static diagnostic module performs static diagnostics on the surface of the wafer transfer mechanical finger based on quantitative parameters and correlation analysis results, and obtains static diagnostic results. The dynamic evaluation module evaluates the particle behavior and friction behavior of the wafer transfer mechanical finger under the simulation conditions of wafer transfer action, and obtains the dynamic evaluation results. The fusion analysis module integrates and analyzes static diagnostic results and dynamic evaluation results to generate surface evaluation results for the wafer-transfer mechanical finger.
[0044] The adjustment system described above in this invention can effectively realize the wafer transfer mechanical finger surface evaluation method, and the technical effects it can achieve are as described in the above embodiments, and will not be repeated here.
[0045] Furthermore, the data acquisition module includes: The region division unit divides the wafer transfer mechanical finger into regions and establishes a sampling grid for each region; The data acquisition unit collects two-dimensional and three-dimensional image data within the sampling grid area, and collects quantitative parameters at the sampling grid locations; The associated storage unit stores two-dimensional image data, three-dimensional image data, and quantitative parameters based on a sampling grid to construct a multimodal dataset.
[0046] Furthermore, the dynamic evaluation module includes: The particle behavior testing unit tests the particle release and adsorption behavior of the wafer-transfer mechanical finger, and performs repeated tests to obtain particle behavior results. The friction behavior testing unit tests the friction coefficient and wear characteristics of the wafer-level mechanical finger and performs repeated tests to obtain the friction behavior results. The results integration unit integrates particle behavior results and friction behavior results to form dynamic evaluation results.
[0047] Similarly, the above-mentioned optimization schemes for the system can also achieve the optimization effects corresponding to the methods in Embodiment 1, which will not be repeated here.
[0048] Although this application has been described in conjunction with specific features and embodiments, it is obvious that various modifications and combinations can be made thereto without departing from the spirit and scope of this application. Accordingly, this specification and accompanying drawings are merely exemplary illustrations of the application as defined herein, and are to be considered as covering any and all modifications, variations, combinations, or equivalents within the scope of this application. Clearly, those skilled in the art can make various alterations and modifications to this application without departing from its scope. Thus, if such modifications and modifications fall within the scope of this application and its equivalents, this application intends to include such modifications and modifications.
Claims
1. A method for evaluating the surface of a wafer-level transfer mechanical finger, characterized in that, The method includes: Image data and quantitative parameters of the surface of the wafer-transmitted robotic finger were collected to obtain a multimodal dataset; A spatial coordinate system is established based on the three-dimensional geometric reference of the wafer transfer mechanical finger. The wafer contact area is determined in the spatial coordinate system, and the multimodal dataset is registered on the spatial coordinate system. Correlation analysis is performed on the registered multimodal data to obtain the correlation analysis results. Based on the quantitative parameters and the correlation analysis results, a static diagnosis is performed on the surface of the wafer transfer mechanical finger to obtain the static diagnosis results. Under the simulation conditions of wafer transfer action, the particle behavior and friction behavior of the wafer transfer mechanical finger are evaluated to obtain dynamic evaluation results; The static diagnostic results and the dynamic evaluation results are fused and analyzed to generate the surface evaluation results of the wafer-transfer mechanical finger.
2. The method for evaluating the surface of a wafer-level transfer mechanical finger according to claim 1, characterized in that, The acquisition of image data and quantitative parameters from the surface of the robotic finger via wafer transmission yields a multimodal dataset, including: The wafer transfer mechanical finger is divided into regions, and a sampling grid is established for each region; Two-dimensional and three-dimensional image data are collected within the sampling grid area, and quantitative parameters are collected at the sampling grid locations. The two-dimensional image data, the three-dimensional image data, and the quantitative parameters are associated and stored based on the sampling grid to construct a multimodal dataset.
3. The method for evaluating the surface of a wafer-level transfer mechanical finger according to claim 1, characterized in that, Registering the multimodal dataset in the spatial coordinate system includes: Determine the initial positioning reference and align the two-dimensional image data to the spatial coordinate system; The three-dimensional image data is projected and its features are aligned with those of the two-dimensional image data. The acquisition location of the quantitative parameters is mapped to the spatial coordinate system and spatially bound to the two-dimensional image data; The registration error of the two-dimensional image data, the three-dimensional image data, and the quantitative parameters is verified.
4. The method for evaluating the surface of a wafer-level transfer mechanical finger according to claim 3, characterized in that, Spatial binding with the two-dimensional image data includes: generating a continuous distribution map of the quantitative parameters using an interpolation method and overlaying it onto the two-dimensional image data.
5. The method for evaluating the surface of a wafer-level transfer mechanical finger according to claim 1, characterized in that, The correlation analysis of the registered multimodal data includes: Within the contact area of the wafer, features are extracted from the multimodal data, and spatial colocation analysis is performed on the multimodal data based on the extracted features to obtain spatial distribution results; Correlation analysis is performed on the extracted features, the correlation coefficient is calculated, and the correlation analysis results are obtained.
6. The method for evaluating the surface of a wafer-level transfer mechanical finger according to claim 1, characterized in that, Static diagnostics of the surface of the wafer transfer mechanical finger include: Collect historical datasets, which include historically acquired image data and quantitative parameters, as well as corresponding surface evaluation results; Construct a static diagnostic model, and train and optimize the static diagnostic model using the historical dataset; The quantitative parameters and the correlation analysis results are subjected to feature filtering. The filtered results are then input into the static diagnostic model for diagnosis, generating static diagnostic results.
7. The method for evaluating the surface of a wafer-level transfer mechanical finger according to claim 1, characterized in that, The particle behavior and friction behavior of the wafer transfer mechanical finger were evaluated, and the dynamic evaluation results include: The particle release and adsorption behaviors of the wafer-transfer mechanical finger were tested and repeated to obtain the particle behavior results. The friction coefficient and wear characteristics of the wafer-level mechanical finger were tested and repeated to obtain the results of its friction behavior. The particle behavior results and the friction behavior results are integrated to form a dynamic evaluation result.
8. A method for evaluating the surface of a wafer-mounted mechanical finger, characterized in that, The method includes: The data acquisition module collects image data and quantitative parameters from the surface of the wafer-transmitted robotic finger to obtain a multimodal dataset; The data analysis module establishes a spatial coordinate system based on the three-dimensional geometric reference of the wafer-transfer mechanical finger, determines the wafer contact area in the spatial coordinate system, registers the multimodal dataset on the spatial coordinate system, performs correlation analysis on the registered multimodal data, and obtains the correlation analysis results. The static diagnostic module performs static diagnostics on the surface of the wafer transfer mechanical finger based on the quantitative parameters and the correlation analysis results, and obtains static diagnostic results. The dynamic evaluation module evaluates the particle behavior and friction behavior of the wafer transfer mechanical finger under the simulation conditions of wafer transfer action, and obtains the dynamic evaluation results. The fusion analysis module integrates the static diagnostic results and the dynamic evaluation results to generate the surface evaluation results of the wafer-transfer mechanical finger.
9. The method for evaluating the surface of a wafer-level transfer mechanical finger according to claim 8, characterized in that, The data acquisition module includes: The region division unit divides the wafer transfer mechanical finger into regions and establishes a sampling grid for each region; The data acquisition unit acquires two-dimensional image data and three-dimensional image data within the sampling grid area, and acquires quantitative parameters at the sampling grid locations; The associated storage unit stores the two-dimensional image data, the three-dimensional image data, and the quantitative parameters based on the sampling grid to construct a multimodal dataset.
10. The method for evaluating the surface of a wafer-level transfer mechanical finger according to claim 8, characterized in that, The dynamic evaluation module includes: The particle behavior testing unit tests the particle release and adsorption behavior of the wafer-transfer mechanical finger and performs repeated tests to obtain particle behavior results. The friction behavior testing unit tests the friction coefficient and wear characteristics of the wafer-level mechanical finger and performs repeated tests to obtain the friction behavior results. The results integration unit integrates the particle behavior results and the friction behavior results to form a dynamic evaluation result.