Wafer measurement method and device based on measurement mark points

Abstract

The embodiment of the invention discloses a wafer measurement method and device based on measurement mark points. The method comprises the following steps: acquiring wafer alignment coordinates of a plurality of measurement mark points of a target wafer; performing target capture on a plurality of first measurement mark points in the plurality of measurement mark points to obtain a coordinate offset of each first measurement mark point, and obtaining a first target capture coordinate of each first measurement mark point based on the wafer alignment coordinate and the coordinate offset of each first measurement mark point; inputting wafer alignment coordinates of a plurality of second measurement mark points except the plurality of first measurement mark points in the plurality of measurement mark points into a target capture coordinate prediction model to output a second target capture coordinate of each second measurement mark point; and measuring a plurality of first measurement mark points based on each first target capture coordinate, and measuring a plurality of second measurement mark points based on each second target capture coordinate, so as to realize wafer measurement of the target wafer, the marking and positioning time is short, and the measurement speed of the mark points is high.

Description

Technical Field

[0001] The present application relates to the field of semiconductor technology, and in particular to a wafer measurement method and device based on measurement mark points. Background Art

[0002] Semiconductor inspection technology plays an important role in the processes of semiconductor production, packaging and testing. Among them, the measurement of alignment error is an important part of the key process in the semiconductor production process. Alignment error is one of the most important errors that need to be controlled in the above key processes. It refers to the pattern alignment error between the previous layer and the current layer during the key process operation. There is a certain distribution pattern on the wafer and the exposure field. In order to ensure that the alignment error is within the threshold range, after each batch of wafers is exposed, it is necessary to perform alignment error measurements on many measurement mark points on the overlay measurement machine. After mathematical processing, the measurement values ​​​​are obtained to obtain the distribution model (overlay model terms) of the alignment error. The distribution model is fed back to the semiconductor production equipment for the alignment error control of the next batch of wafer exposure.

[0003] The measurement mark points for alignment error measurement are evenly distributed on the entire wafer. The optical measurement of alignment error requires that the wafer be accurately moved to the position of the measurement optical path before measurement, so that the camera can obtain a wafer measurement image containing information of both the previous layer and the current layer (i.e., a wafer scanning image containing alignment error measurement mark points), thereby obtaining the alignment error through the wafer measurement image. If the positioning of the measurement mark points is inaccurate, the wafer measurement image may be offset, thereby introducing measurement errors. Therefore, in order to improve the measurement accuracy of the alignment error, the positioning accuracy of the mark points is extremely high. Therefore, how to obtain accurate coordinates of the measurement mark points is an important part of the wafer measurement process.

[0004] In the process of improving the positioning accuracy of the measurement mark point, the measurement mark point is often accurately positioned by target capture. The existing technology achieves target capture by performing wafer image acquisition, wafer image transmission, algorithm calculation, feedback calculation results, motion stage control wafer movement and other steps for each measurement mark point in the wafer. It takes about 90ms to locate a measurement mark point by target capture, and the measurement time of a measurement mark point is generally 200-300ms. Therefore, target capture occupies a large extent of the measurement time of the measurement mark point, the positioning time of the measurement mark point is long, and the measurement speed of the measurement mark point is slow. Summary of the invention

[0005] The present application provides a wafer measurement method and device based on measurement mark points, which can shorten the positioning time of the measurement mark points and improve the measurement speed of the measurement mark points.

[0006] In a first aspect, the present application provides a wafer measurement method based on measurement mark points, including: obtaining wafer alignment coordinates of multiple measurement mark points of a target wafer; performing target capture on multiple first measurement mark points among the multiple measurement mark points to obtain the coordinate offset of each of the first measurement mark points, and obtaining the first target capture coordinates of each of the first measurement mark points based on the wafer alignment coordinates and the coordinate offset of each of the first measurement mark points; inputting the wafer alignment coordinates of multiple second measurement mark points among the multiple measurement mark points except the multiple first measurement mark points into a target capture coordinate prediction model, and outputting the second target capture coordinates of each of the second measurement mark points through the target capture coordinate prediction model; measuring the multiple first measurement mark points based on the first target capture coordinates of each of the first measurement mark points, and measuring the multiple second measurement mark points based on the second target capture coordinates of each of the second measurement mark points, so as to achieve wafer measurement of the target wafer. By adopting the present application, the first target capture coordinates can be obtained by performing target capture on the first measurement mark point, and the second target capture coordinates of the second measurement mark point can be quickly obtained through the target capture coordinate prediction model. Wafer measurement of the target wafer can be achieved without performing target capture on multiple second measurement mark points, which can shorten the positioning time of the measurement mark point and improve the measurement speed of the measurement mark point.

[0007] In a possible implementation of the first aspect, the first measurement mark point is a measurement mark point randomly selected from the plurality of measurement mark points, or a measurement mark point located on the target motion trajectory from the plurality of measurement mark points. The present application can determine the first measurement mark point in different ways, and has strong applicability.

[0008] In a possible implementation of the first aspect, the above-mentioned outputting the second target capture coordinates of each of the above-mentioned second measurement mark points through the above-mentioned target capture coordinate prediction model includes: obtaining the coordinate offset of each of the above-mentioned second measurement mark points based on the above-mentioned wafer alignment coordinates of each of the above-mentioned second measurement mark points through the above-mentioned target capture coordinate prediction model, and outputting the second target capture coordinates of each of the above-mentioned second measurement mark points based on the above-mentioned coordinate offsets of each of the above-mentioned second measurement mark points and the above-mentioned wafer alignment coordinates of each of the above-mentioned second measurement mark points through the above-mentioned target capture coordinate prediction model. With the present application, the wafer alignment coordinates can be compensated by obtaining the coordinate offset to obtain the second target capture coordinates with higher positioning accuracy. The target capture coordinate prediction model has a fast processing speed, which can further improve the positioning speed of the measurement mark points while ensuring the accuracy of the second target capture coordinates.

[0009] In a possible implementation of the first aspect, the target capture coordinate prediction model is trained by the following method: obtaining a first training sample set, the first training sample set including sample target capture coordinates and sample wafer alignment coordinates of multiple sample measurement mark points, wherein the multiple sample measurement mark points include the multiple first measurement mark points and / or the multiple measurement mark points included in the multiple sample wafers, the sample target capture coordinates and the sample wafer alignment coordinates include the first target capture coordinates and the wafer alignment coordinates of the multiple first measurement mark points, and / or the wafer alignment coordinates and target capture coordinates of the multiple measurement mark points included in the multiple sample wafers; obtaining multiple position error terms of each of the sample measurement mark points within and between fields of the target wafer based on the sample wafer alignment coordinates and the sample target capture coordinates of each of the sample measurement mark points, performing high-order polynomial fitting on the multiple position error terms to obtain the coordinate offset of each of the sample measurement mark points; establishing the target capture coordinate prediction model based on the sample wafer alignment coordinates and the coordinate offset of each of the sample measurement mark points. By adopting the present application, in addition to the target capture coordinates and wafer alignment coordinates of the sample wafer, the target capture coordinate prediction model can also be trained using the first target capture coordinates and wafer alignment coordinates corresponding to multiple first measurement mark points obtained during the measurement process, which can make the first training sample set of the target capture coordinate prediction model richer, and the target capture coordinate prediction model can be established by performing high-order polynomial fitting on multiple position error terms, which can improve the accuracy of the target capture coordinate prediction model, thereby improving the positioning accuracy of the measurement mark points.

[0010] In a possible implementation of the first aspect, the target capture coordinate prediction model is trained by the following method: obtaining a first training sample set, wherein the first training sample set includes sample target capture coordinates, sample wafer alignment coordinates and measurement quality evaluation indicators of multiple sample measurement mark points, wherein the multiple sample measurement mark points include the multiple first measurement mark points and / or the multiple measurement mark points included in the multiple sample wafers, and the sample target capture coordinates and the sample wafer alignment coordinates include the first target capture coordinates and the wafer alignment coordinates of the multiple first measurement mark points, and / or the wafer alignment coordinates and target capture coordinates of the multiple measurement mark points included in the multiple sample wafers; based on the sample target capture coordinates, the sample wafer alignment coordinates and the measurement quality evaluation indicators of each of the sample measurement mark points, the target capture coordinate prediction model is established using a target algorithm, wherein the target algorithm includes a gradient descent algorithm, a principal component analysis algorithm or a deep learning algorithm. Using the present application, the first training sample set may include the measurement quality evaluation indicator, so that the target capture coordinate prediction model can be established based on the measurement quality evaluation indicator through a variety of algorithms, which has strong applicability.

[0011] In a possible implementation of the first aspect, the first training sample set includes sample wafer alignment coordinates and sample target capture coordinates of multiple sample measurement mark points included in multiple sample wafers, and after obtaining the first target capture coordinates of each of the first measurement mark points based on the wafer alignment coordinates and the coordinate offset, the method further includes: adding the wafer alignment coordinates and the first target capture coordinates of each of the first measurement mark points to the first training sample set as correction training data of the target capture coordinate prediction model to perform correction training on the target capture coordinate prediction model. By adopting the present application, the target capture coordinate prediction model can be corrected and trained based on the first target capture coordinates after obtaining the first target capture coordinates, thereby further improving the accuracy of the target capture coordinate prediction model, thereby further improving the accuracy of the positioning of the mark point.

[0012] In a possible implementation of the first aspect, the above-mentioned acquisition of wafer alignment coordinates of multiple measurement mark points of the target wafer includes: acquiring wafer alignment coordinates of alignment mark points in the target wafer for wafer alignment; performing wafer alignment on the target wafer based on the wafer alignment coordinates of the alignment mark points to obtain the wafer alignment coordinates of the multiple measurement mark points. By adopting the present application, the wafer alignment coordinates of multiple measurement mark points can be acquired by wafer alignment, which can be applied to most scenarios for acquiring wafer alignment coordinates and has strong applicability.

[0013] In a possible implementation of the first aspect, the above-mentioned acquisition of wafer alignment coordinates of multiple measurement mark points of the target wafer includes: acquiring the wafer alignment coordinates of the alignment mark points used for wafer alignment in the above-mentioned target wafer, performing wafer alignment on the above-mentioned target wafer based on the wafer alignment coordinates of the above-mentioned alignment mark points to obtain the initial wafer alignment coordinates of the above-mentioned multiple measurement mark points; inputting the initial wafer alignment coordinates of each of the above-mentioned measurement mark points into a wafer alignment coordinate correction model, and outputting the wafer alignment coordinates of each of the above-mentioned measurement mark points through the above-mentioned wafer alignment coordinate correction model. By adopting the present application, the initial wafer alignment coordinates can be obtained through wafer alignment, and the initial wafer alignment coordinates can be corrected through the wafer alignment coordinate correction model to obtain the wafer alignment coordinates, which can further improve the accuracy of the wafer alignment coordinates.

[0014] In a possible implementation of the first aspect, the wafer alignment coordinate correction model is trained by the following method: obtaining a second training sample set, wherein the second training sample set includes sample target capture coordinates and sample wafer alignment coordinates of multiple sample measurement mark points included in multiple sample wafers, and one sample measurement mark point corresponds to one sample wafer alignment coordinate and one sample wafer alignment coordinate; calculating the coordinate difference between the sample wafer alignment coordinates and the sample target capture coordinates of each sample measurement mark point, and establishing the wafer alignment coordinate correction model based on the sample wafer alignment coordinates and the coordinate difference of each of the multiple sample measurement mark points. By adopting the present application, the wafer alignment coordinate correction model can be trained by the second training sample set, and the acquisition of the second training sample set is simple and has strong applicability.

[0015] In a possible implementation of the first aspect, after obtaining the first target capture coordinates of each of the first measurement mark points based on the wafer alignment coordinates and the coordinate offset of each of the first measurement mark points, the method further includes: obtaining the wafer alignment coordinate errors of each of the first measurement mark points based on the first target capture coordinates of each of the first measurement mark points and the wafer alignment coordinates of each of the first measurement mark points; adding the wafer alignment coordinates of each of the first measurement mark points and the wafer alignment coordinate errors to the second training sample set as the correction training data of the wafer alignment coordinate correction model to perform correction training on the wafer alignment coordinate correction model. By adopting the present application, the wafer alignment coordinate correction model can be corrected and trained by the first target capture coordinates obtained during the wafer measurement process, which can further improve the accuracy of the wafer alignment coordinate correction model, thereby improving the positioning accuracy of the measurement mark points.

[0016] In a possible implementation of the first aspect, the target wafer includes multiple fields, the multiple measurement mark points are multiple measurement mark points included in the target places in the multiple fields, and the distribution state of the measurement mark points in each of the multiple fields is the same; after the wafer measurement of the target wafer is realized, the method further includes: constructing an intra-field coordinate prediction model based on the wafer alignment coordinates and the first target capture coordinates of the multiple first measurement mark points included in the target place, and the wafer alignment coordinates and the second target capture coordinates of the multiple second measurement mark points; when the wafer alignment coordinates of the multiple measurement mark points included in any place other than the target field in the multiple fields are obtained, the wafer alignment coordinates of each measurement mark point included in any of the multiple fields are input into the intra-field coordinate prediction model, and the target capture coordinates of each measurement mark point in any of the fields are output through the intra-field coordinate prediction model to complete the measurement of each measurement mark point included in any of the fields. By adopting the present application, the target capture coordinates of each measurement mark point in any field can be output through the field coordinate prediction model, and the target capture coordinates of each measurement mark point can be obtained without performing a target capture operation, further shortening the positioning time of the measurement mark point and improving the measurement speed of the mark point.

[0017] In a second aspect, the present application provides a wafer measurement device based on measurement mark points, including: an acquisition module, which acquires wafer alignment coordinates of multiple measurement mark points of a target wafer; a target capture module, which performs target capture on multiple first measurement mark points among the above-mentioned multiple measurement mark points to obtain the coordinate offset of each of the above-mentioned first measurement mark points, and obtains the first target capture coordinates of each of the above-mentioned first measurement mark points based on the above-mentioned wafer alignment coordinates and the above-mentioned coordinate offset of each of the above-mentioned first measurement mark points; a prediction module, which inputs the above-mentioned wafer alignment coordinates of multiple second measurement mark points among the above-mentioned multiple measurement mark points except the above-mentioned multiple first measurement mark points into a target capture coordinate prediction model, and outputs the second target capture coordinates of each of the above-mentioned second measurement mark points through the above-mentioned target capture coordinate prediction model; a measurement module, which is used to measure the above-mentioned multiple first measurement mark points based on the above-mentioned first target capture coordinates of each of the above-mentioned first measurement mark points, and measure the above-mentioned multiple second measurement mark points based on the above-mentioned second target capture coordinates of each of the above-mentioned second measurement mark points, so as to realize wafer measurement of the above-mentioned target wafer.

[0018] In a third aspect, the present application provides a terminal device, comprising: an image acquisition device, a processor and a memory; the processor is connected to the memory, wherein the memory is used to store program code, and the image acquisition device and the processor are used to call the program code to execute the method provided in the first aspect and any possible implementation method of the first aspect.

[0019] In a fourth aspect, the present application provides a computer-readable storage medium, in which a computer program is stored. The computer program is suitable for being loaded by a processor and executing the method provided in the first aspect and any possible implementation method of the first aspect. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 It is a structural schematic diagram of a wafer measurement system based on measurement marking points provided by the present application;

[0021] Figure 2 It is a schematic diagram of the distribution of measurement mark points and alignment mark points of the target wafer provided by the present application;

[0022] Figure 3 It is a flow chart of a wafer measurement method based on measurement mark points provided by the present application;

[0023] Figure 4 is another flow chart of the wafer measurement method based on measurement mark points provided by the present application;

[0024] Figure 5 is another flow chart of the wafer measurement method based on measurement mark points provided by the present application;

[0025] Figure 6 is another flow chart of the wafer measurement method based on measurement mark points provided by the present application;

[0026] Figure 7 It is a structural schematic diagram of a wafer measurement device based on measurement marking points provided by the present application;

[0027] Figure 8 It is a structural diagram of the terminal device provided by this application. DETAILED DESCRIPTION

[0028] The technical solutions in the embodiments of the present application will be described below in conjunction with the drawings in the embodiments of the present application.

[0029] To facilitate understanding, some nouns are briefly explained as follows:

[0030] 1. Field: A partial area in the target wafer. A target wafer includes multiple fields. Each field has the same size and the distribution of the marking points in the field is the same. Usually a field includes one alignment marking point and multiple measurement marking points.

[0031] 2. Alignment mark point refers to the characteristic point on the target wafer used for wafer alignment, which is usually located at the geometric center of a field.

[0032] 3. Measurement mark points refer to characteristic points on the target wafer used to measure the wafer alignment error of the wafer, which are usually evenly distributed in each field.

[0033] 4. Target capture refers to the process of moving the measurement mark point in the field of view to the center position of the field of view when the wafer image acquisition device scans the target wafer to obtain the target wafer image, where the distance between the measurement mark point and the center position of the field of view of the target wafer image can be called the alignment error.

[0034] Embodiment 1

[0035] The wafer measurement method based on measurement mark points provided in this application can be applied to the alignment error measurement of wafer mark positioning, and can also be extended to application scenarios such as size measurement between marks, focus position measurement, etc., which require positioning and measurement of specific marks. For ease of description, this application takes the application scenario of alignment error measurement based on wafer mark positioning as an example, and takes the terminal device as the execution subject as an example. Figure 1 , Figure 1 Schematic diagram of the structure of the wafer measurement system based on the measurement mark point provided by the present application. Figure 1 As shown, the wafer measurement system based on measurement mark points may include a terminal device 1a, a motion stage 1b and a motion controller 1c.

[0036] In some feasible embodiments, the above-mentioned motion table 1b can be used to carry the target wafer and carry the target wafer to move based on the target motion trajectory. The above-mentioned motion table 1b can be an X / Z / Theta axis motion table, or it can be other motion tables that can carry the target wafer. This application does not limit this, and takes the X / Z / Theta axis motion table as an example for explanation. It can be understood that the above-mentioned motion table 1b can move on the X-axis, Z-axis and Theta axis, corresponding to lateral movement, vertical movement and rotational movement, respectively. The above-mentioned motion table 1b can be used to carry the target wafer and adjust the position of the target wafer by movement. Optionally, the above-mentioned motion table 1b can also include a Y-axis motion table, which can carry the target wafer to make the target wafer move longitudinally. The above-mentioned motion table 1b can move according to the target motion trajectory, and the above-mentioned target motion trajectory can be one of a spiral trajectory, a concentric circle trajectory, a square trajectory, a polygonal trajectory and a horizontal and vertical trajectory.

[0037] In some feasible implementations, the motion controller 1c may be a microcontroller, a central controller or other controller, which may be used to obtain or generate a target motion trajectory of the target wafer and control the motion stage 1b to move based on the target motion trajectory. Optionally, the motion controller 1c may also be integrated into a motion control module in the terminal device 1a, which is not limited in this application.

[0038] In some feasible implementations, the target wafer carried on the moving stage 1b includes multiple fields, each field includes multiple measurement mark points and at least one alignment mark point. For ease of description, this application takes each field including one alignment mark point as an example for explanation. Among them, the distribution states of the measurement mark points and the alignment mark points in each field are the same, and the size of each field is also the same. The alignment mark points are used to align the target wafer, and the measurement mark points are used to measure the alignment error of the target wafer. For ease of understanding the distribution of the measurement mark points and alignment mark points of the target wafer, please refer to Figure 2 , Figure 2 Schematic diagram of the distribution of measurement mark points and alignment mark points of the target wafer provided by the present application. Figure 2 As shown, the target wafer includes multiple fields of the same size such as field 1, field 2, field 3, field 4 and field 5 (not all of them are shown in the figure, only some of the fields are shown), each field includes an alignment mark point and multiple measurement mark points ( Figure 2 (Not all are shown in the figure, only some of the measurement mark points of field 1 are shown). Taking field 1 as an example, field 1 includes alignment mark points, measurement mark points 1, measurement mark points 2, measurement mark points 3 and measurement mark points 4. The alignment mark points may be located at the geometric center of the field, and the measurement mark points may be distributed as evenly as possible in each field.

[0039] In some feasible implementations, the terminal device 1a may include a memory, and the memory may store computer programs of models such as a target capture coordinate prediction model, a wafer alignment coordinate correction model, and an in-field coordinate prediction model. The terminal device 1a can be used to obtain the wafer alignment coordinates of multiple measurement mark points of the target wafer, wherein one measurement mark point is marked by a wafer alignment coordinate, and the multiple measurement mark points include multiple first measurement mark points and multiple second measurement mark points. It should be noted that the first measurement mark point can be determined according to the training requirements of the target capture coordinate prediction model, the wafer alignment coordinate correction model, and / or the in-field coordinate prediction model. For example, the target motion trajectory can be planned based on the training requirements, and the measurement mark point passed by the target motion trajectory is determined as the first measurement mark point.

[0040] In some feasible implementations, the terminal device 1a may include an image acquisition device, such as Figure 1The image acquisition device 1d shown in the figure may be a device such as a camera that can acquire a target wafer image of a target wafer. The image acquisition device 1d may be used to scan the target wafer to obtain a target wafer image of the target wafer, so as to perform operations such as target capture and wafer alignment of the target wafer. Optionally, the image acquisition device may be located outside the terminal device 1a. If the image acquisition device 1d is located outside the terminal device 1a, the target wafer image may be transmitted to the terminal device 1a after scanning the target wafer to obtain the target wafer image, so that the terminal device 1a performs operations such as target capture and wafer alignment based on the target wafer image. For ease of description, the present application takes the terminal device 1a including the image acquisition device 1d as an example for explanation, and no further description is given.

[0041] In some feasible embodiments, the terminal device 1a can perform target capture on the first measurement mark point based on the wafer alignment coordinates of the first measurement mark point. The terminal device 1a can determine the coordinate offset of the wafer alignment coordinates of the first measurement mark point through the target capture operation, and compensate the wafer alignment coordinates based on the coordinate offset, so as to obtain the first target capture coordinates of the first measurement mark point. For example, if the wafer alignment coordinates of the first measurement mark point are (1, 2), and the coordinate offset determined by target capture is (0, -0.5), then the first target capture coordinates may be (1, 1.5). It should be noted that the coordinate offset is used to indicate the difference between the wafer alignment coordinates of the first measurement mark point and the center of the field of view of the target wafer image. By compensating for the difference, the first measurement mark point can be moved to the center of the field of view of the target wafer image.

[0042] In some feasible implementations, a target capture coordinate prediction model may be stored in the terminal device 1a. The terminal device 1a may input the wafer alignment coordinates corresponding to the multiple second measurement mark points except the multiple first measurement mark points among the multiple measurement mark points into the target capture coordinate prediction model, thereby skipping the operation of directly performing target capture on the second measurement mark points, and predicting and outputting the second target capture coordinates corresponding to each of the above-mentioned second measurement mark points through the target capture coordinate prediction model, thereby reducing the time required for marking and positioning the second measurement mark points. For example, the wafer alignment coordinates of one of the multiple second measurement mark points may be (0, 1), then the terminal device 1a may input (0, 1) into the target capture coordinate prediction model, and the target capture coordinate prediction model may output the coordinates (1, 1) as the second target capture coordinates of the second measurement mark point with the wafer alignment coordinates of (0, 1).

[0043] In some feasible implementations, the terminal device 1a can also measure the multiple first measurement mark points based on the first target capture coordinates corresponding to each first measurement mark point, and measure the multiple second measurement mark points based on the second target capture coordinates corresponding to each second measurement mark point, thereby realizing wafer measurement of the target wafer. For example, the terminal device 1a can control the motion stage 1b to carry the target wafer to move to the first target capture coordinates or the second target capture coordinates through the motion controller 1c or the motion control module in the terminal device 1a to perform wafer measurement on the target wafer, thereby avoiding the poor detection accuracy of the wafer alignment error caused by the positioning error of the first measurement mark point or the second measurement mark point. The first measurement mark point can be a measurement mark point randomly selected from the multiple measurement mark points. Optionally, the target motion trajectory can be planned based on the training requirements of the target capture coordinate prediction model, and the measurement mark point located on the target motion trajectory can be determined as the first measurement mark point, or the first measurement mark point can be determined according to other rules, which is not limited by the present application.

[0044] In some feasible implementations, the terminal device 1a can predict the coordinate offset of each second measurement mark point based on the wafer alignment coordinates of each second measurement mark point through a target capture coordinate prediction model, and then compensate the wafer alignment coordinates of each second measurement mark point based on the coordinate offset of each second measurement mark point through the target capture coordinate prediction model to output the second target capture coordinates of each of the above second measurement mark points. For example, if the target capture coordinate prediction model detects that the coordinate offset of a second measurement mark point is (1, 0), and the wafer alignment coordinates of the above second measurement mark point are (0, 1), then the second target capture coordinates of the above second measurement mark point can be obtained as (1, 1) through the target capture coordinate prediction model.

[0045] In some feasible implementations, the terminal device 1a may obtain a first training sample set of the target capture coordinate prediction model, and obtain the target capture coordinate prediction model based on the training sample data included in the first training sample set. The first training sample set may include sample target capture coordinates and sample wafer alignment coordinates of multiple sample measurement mark points. Among them, the multiple sample measurement mark points include multiple first measurement mark points and / or multiple measurement mark points included in multiple sample wafers, and the sample target capture coordinates and sample wafer alignment coordinates include the first target capture coordinates and wafer alignment coordinates of multiple first measurement mark points, and / or the wafer alignment coordinates and target capture coordinates of multiple measurement mark points included in multiple sample wafers. In other words, if the terminal device 1a does not obtain the first target capture coordinates of the first measurement mark point, the target capture coordinate prediction model may be trained by obtaining the wafer alignment coordinates and target capture coordinates of multiple measurement mark points included in multiple sample wafers as the sample wafer alignment coordinates and sample target capture coordinates. Among them, the sample wafer may be a wafer that has completed wafer measurement before wafer measurement of the target wafer. The target capture coordinate prediction model is trained using the first training sample set. The data of the first training sample set is convenient to obtain and the data of the first training sample set is rich, so that the prediction accuracy of the target capture coordinate prediction model obtained by training the first training sample set can be enhanced, and the adaptability is strong. If the terminal device 1a has obtained multiple first target capture coordinates obtained by target capture of multiple first measurement mark points, the first target capture coordinates and wafer alignment coordinates corresponding to the multiple first measurement mark points can be used as sample target capture coordinates and sample wafer alignment coordinates in the first training sample set to train the target capture coordinate prediction model. Optionally, the first training sample set can also include the first target capture coordinates and wafer alignment coordinates of the measurement mark points of the target wafer and the target capture coordinates and wafer alignment coordinates of the measurement mark points of the sample wafer. The first target capture coordinates and wafer alignment coordinates obtained during the wafer measurement of the target wafer are added to the first training sample set to train the target capture coordinate prediction model, which can make the sample data of the first training sample set richer and enhance the prediction accuracy of the target capture coordinate prediction model obtained by training the first training sample set.

[0046] In some feasible implementations, the target capture coordinate prediction model may be a high-order fitting model, and the terminal device 1a may obtain multiple position error terms of each sample measurement mark point within and between fields of the target wafer based on the sample wafer alignment coordinates and sample target capture coordinates of each sample measurement mark point in the first training sample set through operations such as scaling, translation, and distortion. Among them, multiple position error terms of the same sample measurement mark point may be obtained by performing a variety of different operations on the sample wafer alignment coordinates and sample target capture coordinates of the same sample measurement mark point, and then a high-order polynomial fitting may be performed on the multiple position error terms obtained based on the same sample measurement mark point, thereby obtaining the coordinate offset of the sample measurement mark point. The terminal device 1a may perform a high-order polynomial fitting on the multiple position error terms obtained based on each sample measurement mark point, thereby obtaining the coordinate offset of each sample measurement mark point (the above-mentioned coordinate offset may also be referred to as a deviation value), and then establish a target capture coordinate prediction model based on the sample wafer alignment coordinates and coordinate offset of each sample measurement mark point.

[0047] Optionally, in some feasible implementations, the target capture coordinate prediction model may also be an artificial intelligence algorithm model, and the first training sample set may also include a measurement quality evaluation index. The sample target capture coordinate of a sample measurement mark point corresponds to a measurement quality evaluation index, and the measurement quality evaluation index is used to evaluate the credibility of the sample target capture coordinate. When the credibility reaches a threshold, it indicates that the sample target capture coordinate can be regarded as the actual target capture coordinate. The threshold may be a pre-set threshold such as 80%, 90%, etc., which should be determined according to the specific form of the product, and this application does not limit it. It should be noted that the first target capture coordinate obtained by the terminal device 1a through target capture is a measurement quality evaluation index corresponding to the measurement quality evaluation index that reaches the threshold. The terminal device 1a can establish a target capture coordinate prediction model based on the sample target capture coordinates of each sample measurement mark point, the sample wafer alignment coordinates and the measurement quality evaluation index using a target algorithm, wherein the target algorithm includes but is not limited to a gradient descent algorithm, a principal component analysis algorithm or a deep learning algorithm, and this application does not limit it.

[0048] Optionally, in some feasible implementations, if the training sample data used by the terminal device 1a when training the target capture coordinate prediction model is the wafer alignment coordinates and target capture coordinates of the multiple sample measurement mark points corresponding to the above-mentioned sample wafer, then after the terminal device 1a performs target capture on multiple first measurement mark points and generates the first target capture coordinates, the wafer alignment coordinates and the first target capture coordinates corresponding to each first measurement mark point are added to the above-mentioned first training sample set, so as to input the target capture coordinate prediction model as the corrected training data of the target capture coordinate prediction model, and then correct the target capture coordinate prediction model. For example, the terminal device 1a can obtain the predicted target capture coordinates of the first measurement mark point by inputting the wafer alignment coordinates of the first measurement mark point into the target capture coordinate prediction model, and then can obtain the corrected training data of the target capture coordinate prediction model (i.e., the corrected training data for the first measurement mark point) by comparing the above-mentioned predicted target capture coordinates with the first target capture coordinates of the first measurement mark point, so as to obtain the corrected training data of the multiple measurement mark points and correct the target capture coordinate prediction model based on the corrected training data.

[0049] In some feasible implementations, the terminal device 1a can obtain the wafer alignment coordinates of the above-mentioned multiple measurement mark points through a wafer alignment operation. The terminal device 1a can obtain the wafer alignment coordinates of the alignment mark points in the target wafer for wafer alignment, and then perform wafer alignment on the above-mentioned target wafer based on the wafer alignment coordinates of the alignment mark points to obtain the wafer alignment coordinates of the above-mentioned multiple measurement mark points. For example, the reference mark points in the reference layer of the target wafer are aligned with the alignment mark points so that the reference layer is aligned with the target wafer, thereby obtaining the wafer alignment coordinates of the multiple measurement mark points after alignment.

[0050] Optionally, in some feasible implementations, the terminal device 1a may also correct the wafer alignment coordinates obtained by the above-mentioned wafer alignment through the wafer alignment coordinate correction model, so as to obtain the corrected wafer alignment coordinates of the above-mentioned multiple measurement mark points. For ease of understanding, if the terminal device 1a uses the wafer alignment coordinate correction model for correction, the uncorrected wafer alignment coordinates obtained by the above-mentioned wafer alignment can be determined as the initial wafer alignment coordinates. The terminal device 1a can obtain the initial wafer alignment coordinates of multiple measurement mark points and input the multiple initial wafer alignment coordinates into the wafer alignment coordinate correction model, so that the wafer alignment coordinates obtained after correction for each initial wafer alignment coordinate can be output through the wafer alignment coordinate correction model. For example, if the wafer alignment coordinates of a measurement mark point among multiple measurement mark points are (1, 2), the terminal device 1a can input (1, 2) into the wafer alignment coordinate correction model. The wafer alignment coordinate correction model can correct the initial wafer alignment coordinates and output coordinates (2, 2) as the wafer alignment coordinates of the measurement mark point whose initial wafer alignment coordinates are (1, 2), which is used to obtain the target capture coordinates of the wafer.

[0051] In some feasible implementations, the terminal device 1a may train the above-mentioned wafer alignment coordinate correction model. The terminal device 1a may obtain a second training sample set of the wafer alignment coordinate correction model, and the above-mentioned second training sample set may include sample target capture coordinates and sample wafer alignment coordinates of multiple sample measurement mark points included in multiple sample wafers, wherein one sample measurement mark point corresponds to one sample target capture coordinate and one sample wafer alignment coordinate. The terminal device 1a may calculate the coordinate difference between the sample target capture coordinates and the sample wafer alignment coordinates of each sample measurement mark point, thereby establishing a wafer alignment coordinate correction model based on the above-mentioned coordinate differences, so that the wafer alignment coordinate correction model obtains the ability to output wafer alignment coordinates corresponding to any initial wafer alignment coordinate based on any input initial wafer alignment coordinate.

[0052] In some feasible implementations, the terminal device 1a may obtain the wafer alignment coordinate error of each first measurement mark point based on the first target capture coordinates and wafer alignment coordinates corresponding to each first measurement mark point after obtaining the first target capture coordinates of each first measurement mark point. For example, if the above-mentioned first target capture coordinates are (1, 1.5) and the above-mentioned wafer alignment coordinates are (1, 2), then the above-mentioned wafer alignment coordinate error may be (0, -0.5). The terminal device 1a may input the wafer alignment coordinates and wafer alignment coordinate errors of each first measurement mark point into the wafer alignment coordinate correction model, and perform correction training on the wafer alignment coordinate correction model as correction training data of the wafer alignment coordinate correction model. For example, the terminal device 1a may input the wafer alignment coordinate error (0, -0.5) and the wafer alignment coordinates (1, 2) as correction training data into the wafer alignment coordinate correction model, thereby performing correction training based on the correction training data.

[0053] In some feasible implementations, the above-mentioned multiple measurement mark points may be multiple measurement mark points included in the target places in multiple fields in the target wafer. After performing wafer measurement on the target wafer, the terminal device 1a may construct an intra-field coordinate prediction model based on the wafer alignment coordinates and first target capture coordinates of multiple first measurement mark points included in the target places in multiple fields, as well as the wafer alignment coordinates and second target capture coordinates of multiple second measurement mark points, and then predict the target capture coordinates of multiple measurement mark points in any field outside the target field through the intra-field coordinate prediction model. For example, the terminal device 1a may use the wafer alignment coordinates and first target capture coordinates of multiple first measurement mark points in the above-mentioned target field, as well as the wafer alignment coordinates and second target capture coordinates of multiple second measurement mark points as construction samples of the intra-field coordinate prediction model, and iteratively train the intra-field coordinate prediction model based on the construction samples, thereby constructing an intra-field coordinate prediction model. The above-mentioned intra-field coordinate prediction model has the ability to output the target capture coordinates corresponding to the wafer alignment coordinates based on the wafer alignment coordinates of the measurement mark points included in any field outside the target field. When the terminal device 1a obtains the wafer alignment coordinates of multiple measurement mark points included in any one of the multiple fields except the above-mentioned target field, the wafer alignment coordinates of each measurement mark point included in any one of the fields can be input into the above-mentioned in-field coordinate prediction model, and the target capture coordinates of each measurement mark point in any one of the fields are output through the in-field coordinate prediction model to complete the measurement of each measurement mark point included in any one of the fields. For example, if the wafer alignment coordinates of one of the multiple measurement mark points included in any field are (3, 3), the terminal device 1a can input (3, 3) into the in-field coordinate prediction model, and the in-field coordinate prediction model can output the coordinates (4, 4) as the target capture coordinates of the measurement mark point with the wafer alignment coordinates (3, 3).

[0054] It should be noted that the above-mentioned target capture coordinate prediction model, wafer alignment coordinate correction model and in-field coordinate prediction model can be high-order fitting models, artificial intelligence algorithm models, neural network models and other models. This application does not limit the types of the above-mentioned models.

[0055] By adopting the present application, the wafer measurement system based on the measurement mark point can obtain the first target capture coordinates by performing target capture on the first measurement mark point, and quickly and accurately obtain the second target capture coordinates of the second measurement mark point through the target capture coordinate prediction model, thereby realizing wafer measurement of the target wafer, shortening the positioning time of the measurement mark point, and improving the measurement speed of the measurement mark point. The operation is simple and the adaptability is strong. In addition, the wafer measurement system can also obtain the wafer alignment coordinates of each measurement mark point through the wafer alignment coordinate correction model, and obtain the target capture coordinates of the measurement mark points of multiple fields through the in-field coordinate prediction model, which can improve the efficiency of obtaining the wafer alignment coordinates and / or target capture coordinates of the measurement mark point, thereby improving the efficiency of wafer measurement. The operation is simple and the applicability is strong.

[0056] Embodiment 2

[0057] See also Figure 3 , Figure 3 It is a flow chart of a wafer measurement method based on measurement mark points provided in the present application. Figure 3 The wafer metrology method based on metrology marks shown is suitable for Figure 1 For the convenience of description, the wafer measurement system based on measurement mark points shown in the present application is described with the terminal device as the execution subject. Figure 3 As shown, the wafer measurement method based on measurement mark points may include the following steps:

[0058] Step S101, obtaining wafer alignment coordinates of a plurality of measurement mark points of a target wafer.

[0059] In some feasible implementations, the terminal device can obtain the wafer alignment coordinates of multiple measurement mark points in the target wafer. Optionally, the terminal device can directly obtain the wafer alignment coordinates of multiple measurement mark points by wafer alignment. The terminal device can obtain the wafer alignment coordinates of the alignment mark points in the target wafer for wafer alignment, for example, the terminal device can obtain the wafer alignment coordinates of the alignment mark points as (5, 5). The terminal device can perform wafer alignment on the target wafer based on the wafer alignment coordinates of the alignment mark points, for example, moving the reference mark points in the reference layer to the wafer alignment coordinates (5, 5) so that the reference mark points are aligned with the alignment mark points. Further, on the basis of the alignment of the reference mark points with the alignment mark points, the terminal device can detect the wafer alignment coordinates of multiple measurement mark points by performing image recognition on the target wafer image of the target wafer, etc., wherein the above-mentioned target wafer image can be obtained by scanning the target wafer by an image acquisition device in the terminal device; the above-mentioned image acquisition device can also be located outside the terminal device, and the terminal device can obtain the target wafer image from the image acquisition device. Further optionally, the terminal device can correct the above-mentioned wafer alignment coordinates by a wafer alignment coordinate correction model. The terminal device can obtain the initial wafer alignment coordinates of multiple measurement mark points in the target wafer through wafer alignment, and then input the multiple initial wafer alignment coordinates into the wafer alignment coordinate correction model, so as to output each wafer alignment coordinate through the wafer alignment coordinate correction model. For example, if the initial wafer alignment coordinates of one of the multiple measurement mark points are (1, 2), the terminal device can input (1, 2) into the wafer alignment coordinate correction model, and the wafer alignment coordinate correction model can output the coordinates (2, 2) as the wafer alignment coordinates of the measurement mark point with the initial wafer alignment coordinates (1, 2).

[0060] Step S102, target capture is performed on multiple first measurement mark points among the above-mentioned multiple measurement mark points to obtain the coordinate offset of each of the above-mentioned first measurement mark points, and the first target capture coordinates of each of the above-mentioned first measurement mark points are obtained based on the above-mentioned wafer alignment coordinates and the above-mentioned coordinate offset of each of the above-mentioned first measurement mark points.

[0061] In some feasible implementations, the terminal device may perform target capture based on the wafer alignment coordinates of multiple first measurement mark points to obtain the coordinate offset of each first measurement mark point, thereby correcting the wafer alignment coordinates based on the coordinate offset of each first measurement mark point to obtain each first target capture coordinate. For example, if the wafer alignment coordinates of the above-mentioned first measurement mark point are (2, 2), and the coordinate offset of the first measurement mark point determined by target capture is (0, -0.5), then the first target capture coordinates of the first measurement mark point can be obtained as (2, 1.5). Among them, the above-mentioned multiple first measurement mark points can be multiple measurement mark points randomly selected from the multiple measurement mark points included in the target wafer, or can be multiple measurement mark points included in the target motion trajectory, or can be determined according to other methods, which should be determined according to the actual product form and are not limited here. It can be understood that since the target motion trajectory can be planned according to the training requirements of the target capture coordinate prediction model, determining the above-mentioned multiple first measurement mark points based on the planned target motion trajectory can better meet the training requirements of the target capture coordinate prediction model, and can avoid the additional target wafer movement time caused by the unplanned motion path. Therefore, the positioning time of the measurement mark points can be further shortened, and the measurement speed of the measurement mark points can be improved.

[0062] In some feasible implementations, the terminal device can obtain the target capture coordinate prediction model through training. The terminal device can obtain the first training sample set of the target capture coordinate prediction model, and the above-mentioned first training sample set may include sample target capture coordinates and sample wafer alignment coordinates of multiple sample measurement mark points. Among them, the multiple sample measurement mark points include multiple first measurement mark points and / or multiple measurement mark points included in multiple sample wafers, and the sample target capture coordinates and sample wafer alignment coordinates include the first target capture coordinates and wafer alignment coordinates of multiple first measurement mark points, and / or the wafer alignment coordinates and target capture coordinates of multiple measurement mark points included in multiple sample wafers. In other words, the above-mentioned first training sample set may include the first target capture coordinates and wafer alignment coordinates corresponding to multiple first measurement mark points, or may include the wafer alignment coordinates and target capture coordinates of multiple measurement mark points included in multiple sample wafers, or may include the wafer alignment coordinates and target capture coordinates of multiple measurement mark points included in multiple sample wafers and the first target capture coordinates and wafer alignment coordinates corresponding to multiple first measurement mark points. The sample wafer may be a wafer that has completed wafer measurement before wafer measurement of the target wafer. The implementation method of using the sample wafer to train the target capture coordinate prediction model can utilize the convenience of data acquisition of the first training sample set and the data richness of the first training sample set to enhance the prediction accuracy of the target capture coordinate prediction model obtained by training the first training sample set. The first training sample set has rich acquisition sources and strong adaptability.

[0063] In some feasible implementations, the target capture coordinate prediction model may be a high-order fitting model, and the terminal device may obtain multiple position error terms of each sample measurement mark point within and between fields of the target wafer based on the sample wafer alignment coordinates and sample target capture coordinates of each sample measurement mark point in the first training sample set through operations such as scaling, translation, and distortion. Among them, multiple position error terms of the same sample measurement mark point may be obtained by performing a variety of different operations on the sample wafer alignment coordinates and sample target capture coordinates of the same sample measurement mark point, and then a high-order polynomial fitting may be performed on the multiple position error terms obtained based on the same sample measurement mark point, thereby obtaining the coordinate offset of the sample measurement mark point. The terminal device may perform a high-order polynomial fitting on the multiple position error terms obtained based on each sample measurement mark point, thereby obtaining the coordinate offset of each sample measurement mark point (the above-mentioned coordinate offset may also be referred to as a deviation value), and then establish a target capture coordinate prediction model based on the sample wafer alignment coordinates and coordinate offset of each sample measurement mark point.

[0064] Optionally, in some feasible implementations, the target capture coordinate prediction model may also be an artificial intelligence algorithm model, and the first training sample set may also include a measurement quality evaluation index. The sample target capture coordinate of a sample measurement mark point corresponds to a measurement quality evaluation index, and the measurement quality evaluation index is used to evaluate the credibility of the sample target capture coordinate. When the credibility reaches a threshold, it indicates that the sample target capture coordinate can be regarded as the actual target capture coordinate. The threshold may be a pre-set threshold such as 80%, 90%, etc., which should be determined according to the specific form of the product, and this application does not limit it. It should be noted that the first target capture coordinate obtained by the terminal device through target capture is a measurement quality evaluation index corresponding to the measurement quality evaluation index that reaches the threshold. The terminal device can establish a target capture coordinate prediction model based on the sample target capture coordinates, sample wafer alignment coordinates and measurement quality evaluation indicators of each sample measurement mark point using a target algorithm, wherein the target algorithm includes but is not limited to a gradient descent algorithm, a principal component analysis algorithm or a deep learning algorithm, and this application does not limit it. It is understandable that if the terminal device has not obtained the first target capture coordinates and wafer alignment coordinates corresponding to the first measurement mark point, the first training sample set includes the wafer alignment coordinates and target capture coordinates of multiple measurement mark points in multiple sample wafers. If the terminal device has obtained the first target capture coordinates and wafer alignment coordinates corresponding to multiple first measurement mark points of the target wafer, the historical data of the first target capture coordinates and wafer alignment coordinates of the above-mentioned multiple first measurement mark points can be added to the first training sample set. By adding the multiple first target capture coordinates obtained during the wafer measurement of the target wafer to the first training sample set, the diversity of the data acquisition paths of the first training sample set can be improved, the richness of the sample data of the first training sample set can be improved, the operation is simple, and the applicability is strong.

[0065] In some feasible implementations, if the training sample data used by the terminal device to train the target capture coordinate prediction model are the sample wafer alignment coordinates and sample target capture coordinates of multiple sample measurement mark points corresponding to the sample wafer, then after the terminal device obtains the first target capture coordinates of each first measurement mark point, the wafer alignment coordinates and the first target capture coordinates corresponding to each first measurement mark point can also be added to the first training sample set as the correction training data of the target capture coordinate prediction model to perform correction training on the target capture coordinate prediction model. In other words, the terminal device can correct the target capture coordinate prediction model after performing target capture on multiple first measurement mark points, so the second target capture coordinate output by the target capture coordinate prediction model will gradually approach the actual value in the process of performing target capture on multiple first measurement mark points, which can further improve the prediction accuracy of the target capture coordinate prediction model.

[0066] In some feasible implementations, the terminal device may also determine the measurement mark points required for target capture based on the training requirements of the above-mentioned target capture coordinate prediction model, thereby determining multiple first measurement mark points from multiple measurement mark points for target capture. The terminal device may plan the target motion trajectory of the target wafer according to the training requirements of the target capture coordinate prediction model. Since the training requirements of the target capture coordinate prediction model are taken into account when generating the target motion trajectory, the efficiency of measuring multiple measurement mark points can be improved by planning the target motion trajectory, further shortening the time for wafer measurement of the target wafer. Based on the above-mentioned target motion trajectory, the terminal device may determine the multiple measurement mark points passed by the target motion trajectory as multiple first measurement mark points to perform target capture on the first measurement mark points.

[0067] Step S103, inputting the wafer alignment coordinates of the plurality of second measurement mark points among the plurality of measurement mark points except the plurality of first measurement mark points into the target capture coordinate prediction model, and outputting the second target capture coordinates of each of the second measurement mark points through the target capture coordinate prediction model.

[0068] In some feasible implementations, the terminal device may obtain the second target capture coordinates corresponding to each second measurement mark point other than the first measurement mark point among the multiple measurement mark points through a target capture coordinate prediction model. For example, if the wafer alignment coordinates of a second measurement mark point among the multiple second measurement mark points are (0, 1), the terminal device may input (0, 1) into the above target capture coordinate prediction model, and the target capture coordinate prediction model may predict and output the coordinates (1, 1) as the second target capture coordinates of the second measurement mark point with the wafer alignment coordinates of (0, 1). The above target capture coordinate prediction model may skip the process of directly performing target capture on the second measurement mark point, and quickly output the target capture coordinates of the second measurement mark point, which may further shorten the positioning time of the measurement mark point and improve the measurement speed of the measurement mark point.

[0069] In some feasible embodiments, the terminal device can obtain the coordinate offset of each second measurement mark point based on the wafer alignment coordinates of each second measurement mark point through the above-mentioned target capture prediction model, and thus compensate the wafer alignment coordinates of each second measurement mark point based on the coordinate offset of each second measurement mark point through the target capture coordinate prediction model to output the second target capture coordinates of each second measurement mark point.

[0070] Step S104, measuring the above-mentioned multiple first measurement mark points based on the above-mentioned first target capture coordinates corresponding to each of the above-mentioned first measurement mark points, and measuring the above-mentioned multiple second measurement mark points based on the above-mentioned second target capture coordinates corresponding to each of the above-mentioned second measurement mark points, so as to achieve wafer measurement of the above-mentioned target wafer.

[0071] In some feasible implementations, the terminal device may measure a plurality of first measurement mark points based on a plurality of first target capture coordinates, and may measure a plurality of second measurement mark points based on a plurality of second target capture coordinates. The terminal device may trigger the target wafer to move to the first target capture coordinates or the second target capture coordinates, thereby measuring the target wafer. When all first measurement mark points and all second measurement mark points on the target wafer are measured, wafer measurement of the target wafer may be achieved.

[0072] By adopting the present application, the first target capture coordinates can be obtained by performing target capture on the first measurement mark point, and the second target capture coordinates of the second measurement mark point can be quickly and accurately obtained through the target capture coordinate prediction model, thereby realizing wafer measurement of the target wafer, shortening the positioning time of the measurement mark point, and improving the measurement speed of the measurement mark point. The operation is simple and the adaptability is strong. In addition, the present application can quickly locate multiple measurement mark points through the target capture coordinate prediction model and the wafer alignment coordinate correction model, and the above multiple models can be applied to subsequent wafer measurements without re-establishing the model, thereby further shortening the wafer measurement time, and the above multiple models can be corrected during the wafer measurement process, so the prediction accuracy of the above multiple models is high.

[0073] To understand the wafer measurement method based on measurement mark points, see Figure 4 , Figure 4 FIG. 1 is another flow chart of the wafer measurement method based on the measurement mark point provided by the present application. Figure 4 As shown, the wafer measurement method based on the measurement mark point may include the following steps:

[0074] Step S201, obtaining wafer alignment coordinates of a plurality of measurement mark points of a target wafer.

[0075] In some feasible implementations, the terminal device may obtain wafer alignment coordinates of multiple measurement mark points. For ease of understanding, the implementation method of the above step S201 may be referred to the description of the above step S101 and will not be repeated here.

[0076] Step S202 , triggering the target wafer to move so as to sequentially move each measurement mark point to the corresponding wafer alignment coordinate.

[0077] In some feasible implementations, the terminal device triggers the motion stage to move to carry the target wafer to move, so as to move each measurement mark point in turn to the wafer alignment coordinates corresponding to each measurement mark point, and align the target wafer. For example, if the target motion trajectory indicates that the wafer alignment coordinates of the measurement mark point to be measured are (4, 4), the terminal device can trigger the motion stage to carry the target wafer to move through the motion controller to move the measurement mark point to the wafer alignment coordinates (4, 4), so as to perform wafer alignment based on the measurement mark point with the wafer alignment coordinates (4, 4).

[0078] Step S203 , determining whether the measurement mark point to which the target wafer currently moves is the first measurement mark point.

[0079] In some feasible implementations, the terminal device can identify the measurement mark point indicated by the wafer alignment coordinate corresponding to the wafer alignment coordinate to which the target wafer currently moves, and determine whether the above-mentioned wafer alignment coordinate point is the first measurement mark point. If the terminal device identifies the measurement mark point as the first measurement mark point, step S204 is executed; if the terminal device identifies the measurement mark point as the second measurement mark point, step S205 is executed.

[0080] Step S204: performing target capture on the first measurement mark point to obtain a first target capture coordinate corresponding to the first measurement mark point.

[0081] In some feasible implementations, if the terminal device identifies the measurement mark point as the first measurement mark point, the target capture may be performed on the first measurement mark point to generate the first target capture coordinates. Optionally, the terminal device may also add the first target capture coordinates and wafer alignment coordinates corresponding to the first measurement mark point to the first training sample set of the target capture coordinate prediction model, thereby training the target capture coordinate prediction model based on the updated first training sample set. For ease of understanding, the implementation method of the above step S204 can be referred to the description of the above step S102, and will not be repeated.

[0082] Step S205 , inputting the wafer alignment coordinates of each second measurement mark point into a target capture coordinate prediction model, and outputting the second target capture coordinates corresponding to each second measurement mark point through the target capture coordinate prediction model.

[0083] In some feasible implementations, if the terminal device identifies the measurement mark point as the second measurement mark point, that is, when the terminal device identifies that the measurement mark point reached by the current movement is not the first measurement mark point, the wafer alignment coordinates of the above-mentioned second measurement mark point can be input into the target capture coordinate prediction model, and the second target capture coordinates corresponding to the second measurement mark point are output through the target capture coordinate prediction model. It can be understood that the terminal device can identify each measurement mark point in turn to determine each second measurement mark point, thereby inputting the wafer alignment coordinates of each second measurement mark point into the target capture coordinate prediction model so that the target capture coordinate prediction model outputs the second target capture coordinates of each second measurement mark point. For ease of understanding, the implementation method of the above-mentioned step S206 can be referred to the description of the above-mentioned step S103, and will not be repeated.

[0084] In some feasible implementations, before the terminal device calls the target capture coordinate prediction model to predict the second target capture coordinates, steps S2051 and S2052 may be executed to train the target capture coordinate prediction model.

[0085] Step S2051, obtaining a first training sample set of the target capture coordinate prediction model.

[0086] In some feasible implementations, the terminal device may obtain a first training sample set of the target capture coordinate prediction model, wherein the first set includes a plurality of sample target capture coordinates and a plurality of sample wafer alignment coordinates. The sample target capture coordinates may include the first target capture coordinates of the first measurement mark point of the target wafer and / or the summed target capture coordinates of the measurement mark points of the sample wafer; the sample wafer alignment coordinates may include the wafer alignment coordinates of the first measurement mark point of the target wafer and / or the summed wafer alignment coordinates of the measurement mark points of the sample wafer.

[0087] Optionally, if the target capture coordinate prediction model is an artificial intelligence algorithm model, the first training sample set also includes a measurement quality evaluation index. The implementation of step S2051 can be found in the description of the first training sample set in step S102, which will not be described in detail.

[0088] Step S2052: training a target capture coordinate prediction model based on the first training sample set.

[0089] In some feasible implementations, the terminal device may train the target capture coordinate prediction model based on the first training sample set. For example, if the target capture coordinate prediction model is a high-order polynomial model, the terminal device may perform high-order polynomial fitting training on the target capture coordinate prediction model based on the first training sample set. Optionally, if the target capture coordinate prediction model is an artificial intelligence algorithm model, the terminal device may train the target capture coordinate prediction model based on the first training sample set through a gradient descent algorithm, a principal component analysis algorithm, or a deep learning algorithm. Through the above-mentioned training of the target capture coordinate prediction model based on the first training sample set, the target capture coordinate prediction model can obtain the ability to output the target capture coordinates of the measurement mark point corresponding to any wafer alignment coordinate based on any input wafer alignment coordinate.

[0090] In some feasible implementations, the terminal device may plan the target motion trajectory according to the training requirements of the target capture coordinate prediction model, so that a plurality of first measurement mark points may be determined based on the target motion trajectory. Optionally, the terminal device may not plan the target motion trajectory, which is not limited in this application. The implementation method of the above step S2052 can be referred to the description of the target capture coordinate prediction model in the above step S102, which will not be repeated here.

[0091] Step S206, measuring each first measurement mark point based on the first target capture coordinates corresponding to each first measurement mark point, and measuring each second measurement mark point based on the second target capture coordinates corresponding to each second measurement mark point.

[0092] In some feasible implementations, if the terminal device obtains the first target capture coordinates of each first measurement mark point, the target wafer movement can be triggered to move each first measurement mark point to the above-mentioned first target capture coordinate position to measure each first measurement mark point; if the terminal device obtains the second target capture coordinates of each second measurement mark point, the target wafer movement can be triggered to move each second measurement mark point to the above-mentioned second target capture coordinate position to measure the second measurement mark point. For ease of understanding, the implementation method of the above step S206 can be referred to the description of the above step S104, which will not be repeated.

[0093] Step S207, determining whether the measurement of all measurement mark points is completed.

[0094] In some feasible implementations, the terminal device can determine whether all the measurement mark points indicated by the target motion trajectory have been measured based on the target motion trajectory. If so, step S208 is executed to end the process; if not, step S202 is executed to measure the next measurement mark point indicated by the target motion trajectory. For example, if the wafer alignment coordinates of the next measurement mark point are (5, 5), the terminal device can trigger the target wafer to move to the position of coordinates (5, 5).

[0095] Step S208, completing wafer measurement of the target wafer.

[0096] When the terminal device completes the measurement of all measurement mark points, the terminal device can output information such as a signal that the wafer measurement of the target wafer has been completed, so as to complete the wafer measurement of the target wafer.

[0097] By adopting the present application, the second target capture coordinates of the second measurement mark point can be quickly output through the target capture coordinate prediction model, and the sample data of the target capture coordinate prediction model is rich and the sample data is easy to obtain; in addition, the terminal device can correct and train the target capture coordinate prediction model according to the target capture result of the first measurement mark point (i.e., the first target capture coordinates), which can further improve the prediction accuracy of the target capture coordinate prediction model. Therefore, by adopting the present application, the positioning time of the measurement mark point can be shortened and the measurement speed of the measurement mark point can be improved.

[0098] Embodiment 3

[0099] In some feasible implementations, the terminal device can obtain the wafer alignment coordinates of each measurement mark point in the target wafer through the wafer alignment coordinate correction model. To facilitate understanding of the application of the wafer alignment coordinate correction model, please refer to Figure 5 , Figure 5 FIG. 1 is another flow chart of the wafer measurement method based on the measurement mark point provided by the present application. Figure 5 As shown, the wafer measurement method based on the measurement mark point may include the following steps:

[0100] Step S301, obtaining a second training sample set of a wafer alignment coordinate correction model.

[0101] In some feasible implementations, the terminal device may obtain a second training sample set of the wafer alignment coordinate correction model, wherein the second training sample set includes sample target capture coordinates and sample wafer alignment coordinates of multiple sample measurement mark points included in multiple sample wafers, and one sample measurement mark point corresponds to one sample target capture coordinate and one sample wafer alignment coordinate. Here, the implementation method of obtaining the second training sample set claimed in step S301 can be referred to the description of the above step S101, and will not be repeated here.

[0102] Step S302: training a wafer alignment coordinate correction model based on a second training sample set.

[0103] In some feasible embodiments, the terminal device may train a wafer alignment coordinate correction model based on multiple sample target capture coordinates and sample wafer alignment coordinates of multiple sample measurement mark points, for example, by calculating the coordinate difference between the sample wafer alignment coordinates and the sample target capture coordinates of each sample measurement mark point, thereby establishing a wafer alignment coordinate correction model based on the sample wafer alignment coordinates and the coordinate difference of each multiple sample measurement mark point, so that the wafer alignment coordinate correction model acquires the ability to output wafer alignment coordinates corresponding to any initial wafer alignment coordinate inputted.

[0104] Step S303 , obtaining wafer alignment coordinates of each measurement mark point of the target wafer through a wafer alignment coordinate correction model.

[0105] In some feasible implementations, a trained wafer alignment coordinate correction model may be stored in the terminal device. The terminal device may perform wafer alignment on the target wafer to obtain the initial wafer alignment coordinates of each measurement mark point of the target wafer, and input the above-mentioned initial wafer alignment coordinates into the wafer alignment coordinate correction model, thereby correcting the above-mentioned initial wafer alignment coordinates through the wafer alignment coordinate correction model to output the wafer alignment coordinates corresponding to the above-mentioned measurement mark points. Here, the implementation method of obtaining the wafer alignment coordinates of each measurement mark point of the target wafer requested for protection in step S303 can be referred to the description of the above-mentioned step S101, and will not be repeated.

[0106] Step S304 , triggering the target wafer to move each measurement mark point to the corresponding wafer alignment coordinate in sequence.

[0107] In some feasible implementations, after the terminal device obtains the wafer alignment coordinates of each measurement mark point, it can trigger the motion stage to carry the target wafer to move, so as to move each measurement mark point to the wafer alignment coordinates corresponding to each measurement mark point in sequence (for example, (2, 2)). For ease of understanding, the implementation method of the above step S304 can be referred to the description of the above step S102 and step S204, and will not be repeated here.

[0108] Step S305 , determining whether the measurement mark point corresponding to the position to which the target wafer currently moves is the first measurement mark point.

[0109] In some feasible implementations, the terminal device can identify whether the position to which the target wafer currently moves is the wafer alignment coordinate of the first measurement mark point. For example, if the target wafer moves according to the target motion trajectory obtained by the training requirement planning based on the target capture coordinate prediction model, it can identify whether the current position to which the target wafer currently moves is the motion identifier of the measurement mark point on the target motion trajectory. If so, the measurement mark point corresponding to the current position to which the target wafer currently moves can be determined as the first measurement mark point. If not, it can be determined that the measurement mark point corresponding to the current position to which the target wafer currently moves is not the first measurement mark point. If the terminal device determines that the measurement mark point corresponding to the current position to which the target wafer currently moves is the first measurement mark point, step S306 is executed, and if not, step S304 is executed.

[0110] Step S306, performing target capture on the first measurement mark point to obtain first target capture coordinates.

[0111] In some feasible implementations, the terminal device may perform target capture on the first measurement mark point based on the target wafer image of the target wafer, and obtain the first target capture coordinates of the first measurement mark point by image recognition, etc. For ease of understanding, the implementation of the above step S306 can be referred to the description of the above step S102, and will not be repeated here.

[0112] Step S307 , performing correction training on the wafer alignment coordinate correction model based on the first target capture coordinates and the wafer alignment coordinates.

[0113] In some feasible implementations, the terminal device may calculate the wafer alignment coordinate error of the first measurement mark point based on the first target capture coordinate and the wafer alignment coordinate corresponding to the first measurement mark point. For example, if the first target capture coordinate is (2, 1.5) and the wafer alignment coordinate is (2, 2), the wafer alignment coordinate error is (0, -0.5). The terminal device may also add the wafer alignment coordinate and the wafer alignment coordinate error of the first measurement mark point to the second training sample set as the correction training data of the wafer alignment coordinate correction model to perform correction training on the above-mentioned wafer alignment coordinate correction model.

[0114] Step S308, determining whether the measurement of all first measurement mark points is completed.

[0115] In some feasible implementations, the terminal device may determine whether the measurement of all first measurement mark points is completed based on the target motion trajectory or the wafer alignment coordinates of the first measurement mark point. For example, if the target wafer has moved to the end point of the target motion trajectory, it indicates that the measurement of all first measurement mark points is completed. Alternatively, if the terminal device detects that the initial wafer alignment coordinates input to the wafer alignment coordinate correction model are the initial wafer alignment coordinates of the last first measurement mark point, it indicates that the measurement of all first measurement mark points is completed. Similarly, if the target wafer does not move to the end point of the target motion trajectory, it indicates that the measurement of all first measurement mark points is not completed. Alternatively, if the terminal device detects that the initial wafer alignment coordinates input to the wafer alignment coordinate correction model are not the wafer alignment coordinates of the last first measurement mark point, it indicates that the measurement of all first measurement mark points is not completed. When the terminal device determines that the measurement of all first measurement mark points is completed, step S309 can be executed; when the terminal device determines that the measurement of all first measurement mark points is not completed, step S304 is executed.

[0116] Step S309, storing the wafer alignment coordinate correction model.

[0117] When the measurement of all the first measurement mark points is completed, the terminal device may store the wafer alignment coordinate correction model so that the wafer alignment coordinate correction model can be applied to subsequent wafer measurements.

[0118] By adopting the present application, the wafer alignment coordinates corresponding to each measurement mark point can be quickly obtained through the wafer alignment coordinate correction model, which can further shorten the positioning time of the measurement mark point and improve the measurement speed of the measurement mark point. In addition, the first target capture coordinates obtained by target capture during the wafer measurement process can be compared with the wafer alignment coordinates predicted by the wafer alignment coordinate correction model, and the wafer alignment coordinate correction model can be corrected by the difference between the two, so as to improve the prediction accuracy of the wafer alignment coordinate correction model.

[0119] Embodiment 4

[0120] In some feasible embodiments, the target wafer includes multiple fields, and the multiple measurement mark points described in the above-mentioned second and third embodiments may be multiple measurement mark points included in the target places in the multiple fields included in the target wafer, and the distribution state of the measurement mark points in each of the multiple fields is the same. The terminal device may construct an intra-field coordinate prediction model after realizing the wafer measurement of the target field included in the above-mentioned target wafer. The above-mentioned intra-field coordinate prediction model may also be constructed by the wafer data that has been measured, and may be stored in the terminal device after the construction is completed, so as to be directly called in the wafer measurement of the target wafer, so as to complete the measurement of each measurement mark point included in any place in the target wafer except the above-mentioned target field. For ease of understanding, refer to Figure 6 , Figure 6FIG. 1 is another flow chart of the wafer measurement method based on the measurement mark point provided by the present application. Figure 6 As shown, the wafer measurement method based on the measurement mark point may include the following steps:

[0121] Step S401, constructing an on-site coordinate prediction model.

[0122] In some feasible implementations, the terminal device may construct an intra-field coordinate prediction model based on the wafer alignment coordinates and first target capture coordinates of multiple first measurement mark points included in the target site that has been measured, as well as the wafer alignment coordinates and second target capture coordinates corresponding to multiple second measurement mark points. For example, the wafer alignment coordinates and first target capture coordinates corresponding to multiple first measurement mark points included in the target site, as well as the wafer alignment coordinates and second target capture coordinates corresponding to multiple second measurement mark points may be used as training samples of the intra-field coordinate prediction model, and the intra-field coordinate prediction model is iteratively trained based on the above training samples, so that the intra-field coordinate prediction model has the ability to output the target capture coordinates of any measurement mark point included in any site outside the target site according to the wafer alignment coordinates of the above measurement mark point.

[0123] Step S402: outputting the target capture coordinates of each measurement mark point in any field in sequence through the field coordinate prediction model.

[0124] In some feasible implementations, when the terminal device obtains the wafer alignment coordinates of multiple measurement mark points included in any one of the multiple fields except the target field, the wafer alignment coordinates of each measurement mark point included in any one of the fields may be input into the field coordinate prediction model, and the target capture coordinates of each measurement mark point in any one of the fields may be output through the field coordinate prediction model to complete the measurement of each measurement mark point included in any one of the fields. For example, if the multiple wafer alignment coordinates of multiple measurement mark points in a field are (7, 7), (8, 8), (9, 9), respectively, the terminal device may input (7, 7), (8, 8), (9, 9) into the field coordinate prediction model, and the field coordinate prediction model may output (7.5, 7), (8.5, 8), (9.5, 9) as the target capture coordinates of the measurement mark points with the above-mentioned wafer alignment coordinates of (7, 7), (8, 8), (9, 9), respectively. It can be understood that after the terminal device obtains the target capture coordinates of multiple measurement mark points included in any place in the target wafer except the target field, the measurement of the next field can be performed, for example, the wafer alignment coordinates of the multiple measurement mark points included in the next field are input into the in-field coordinate prediction model to output the target capture coordinates of each measurement mark point in any field in turn.

[0125] Step S403 , correcting and training the in-field coordinate prediction model based on the wafer alignment coordinates and target capture coordinates of each measurement mark point in any field.

[0126] In some feasible implementations, after the terminal device obtains the target capture coordinates of each measurement mark point in any of the above fields through the in-field coordinate prediction model, the target capture coordinates and wafer alignment coordinates of each measurement mark point in any of the above fields can be input into the in-field coordinate prediction model to correct and train the in-field coordinate prediction model. For example, the terminal device can determine the loss value of the loss function of the in-field coordinate prediction model through multiple target capture coordinates, and update the in-field coordinate prediction model by compensating for the above loss value to improve the accuracy of the target capture coordinates output by the in-field coordinate prediction model. In other words, after the in-field coordinate prediction model outputs multiple target capture coordinates of any field, the in-field coordinate prediction model can correct the in-field coordinate prediction model based on the output multiple target capture coordinates, thereby enriching the training data of the in-field coordinate prediction model and improving the accuracy of the prediction of the in-field coordinate prediction model.

[0127] Step S404, measuring each measurement mark point based on the target capture coordinates of each measurement mark point in any field.

[0128] In some feasible embodiments, the terminal device can trigger the motion table carrying the target wafer to move based on the target capture coordinates of each measurement mark point in any field, so as to move each measurement mark point in turn to the target capture coordinates corresponding to each measurement mark point, so as to measure each measurement mark point.

[0129] Step S405, determining whether the measurement of all fields is completed.

[0130] In some feasible implementations, the terminal device can determine whether the measurement of all fields is completed by determining whether the wafer alignment coordinates of the measurement mark point are the wafer alignment coordinates corresponding to the last measurement mark point. For example, the terminal device can determine whether the measurement of all fields is completed based on the wafer alignment coordinates of the last measurement mark point indicated by the target motion trajectory. If the wafer alignment coordinates of the last measurement mark point are (9, 9), then when the terminal device detects that the wafer alignment coordinates of the current measurement mark point are (9, 9), it can be determined that the measurement of all fields is completed and step S406 is executed; otherwise, it is determined that the measurement of all fields is not completed and step S402 is executed.

[0131] Step S406, storing the in-field coordinate prediction model.

[0132] In some feasible implementations, if the terminal device determines that the measurement of all fields is completed, the intra-field coordinate prediction model obtained after the measurement of all fields is completed can be stored so that the intra-field coordinate prediction model can be applied to subsequent wafer measurement of other target wafers.

[0133] By adopting the present application, the target capture coordinates of multiple measurement mark points in any field other than the target field in the target wafer can be quickly output through the in-field coordinate prediction model. There is no need to perform target capture and other operations on the multiple measurement mark points in any of the above fields. Each measurement mark point can be measured based on the target capture coordinates output by the in-field coordinate prediction model. Therefore, the positioning time of the measurement mark point is short and the measurement speed of the measurement mark point is fast.

[0134] Embodiment 5

[0135] See also Figure 7 , Figure 7 Schematic diagram of the structure of the wafer measurement device based on the measurement mark point provided by the present application. Figure 7 The wafer measurement device 10 based on measurement mark points shown can be a computer device, which has a computer program (including program code) for implementing the above-mentioned wafer image scanning method. The wafer measurement device 10 based on measurement mark points can also be a computer program, for example, the computer program is an application software. The device can be used to execute the corresponding steps in the wafer measurement method based on measurement mark points provided in the present application, such as Figure 1 The terminal equipment shown. Figure 7 As shown, the wafer measurement device 10 based on measurement mark points may include: a first acquisition module 101, a target capture module 102, a first prediction module 103 and a measurement module 104:

[0136] A first acquisition module 101 is used to acquire wafer alignment coordinates of a plurality of measurement mark points of a target wafer;

[0137] A target capture module 102 is used to perform target capture on a plurality of first measurement mark points among the plurality of measurement mark points to obtain a coordinate offset of each of the first measurement mark points, and obtain a first target capture coordinate of each of the first measurement mark points based on the wafer alignment coordinates of each of the first measurement mark points and the coordinate offset;

[0138] A first prediction module 103 is used to input the wafer alignment coordinates of the plurality of second measurement mark points other than the plurality of first measurement mark points among the plurality of measurement mark points into a target capture coordinate prediction model, and output the second target capture coordinates of each of the second measurement mark points through the target capture coordinate prediction model;

[0139] The measurement module 104 is used to measure the above-mentioned multiple first measurement mark points based on the above-mentioned first target capture coordinates of each of the above-mentioned first measurement mark points, and to measure the above-mentioned multiple second measurement mark points based on the above-mentioned second target capture coordinates of each of the above-mentioned second measurement mark points, so as to realize wafer measurement of the above-mentioned target wafer.

[0140] In some feasible implementations, the first measurement mark point in the above-mentioned wafer measurement device based on measurement mark points is a measurement mark point randomly selected from the above-mentioned multiple measurement mark points, or a measurement mark point located on the target motion trajectory from the above-mentioned multiple measurement mark points.

[0141] In some feasible implementations, the first prediction module 103 further includes an output unit 1031:

[0142] The output unit 1031 is used to obtain the coordinate offset of each of the above-mentioned second measurement mark points based on the above-mentioned wafer alignment coordinates of each of the above-mentioned second measurement mark points through the above-mentioned target capture coordinate prediction model, and output the second target capture coordinates of each of the above-mentioned second measurement mark points based on the above-mentioned coordinate offset of each of the above-mentioned second measurement mark points and the above-mentioned wafer alignment coordinates of each of the above-mentioned second measurement mark points through the above-mentioned target capture coordinate prediction model.

[0143] In some feasible implementations, the wafer measurement device based on measurement mark points further includes a second acquisition module 105, a fitting module 106 and a first establishment module 107:

[0144] The second acquisition module 105 is used to acquire a first training sample set, wherein the first training sample set includes sample target capture coordinates and sample wafer alignment coordinates of a plurality of sample measurement mark points, wherein the plurality of sample measurement mark points include the plurality of first measurement mark points and / or the plurality of sample wafers, and the sample target capture coordinates and the sample wafer alignment coordinates include the first target capture coordinates and the wafer alignment coordinates of the plurality of first measurement mark points, and / or the wafer alignment coordinates and target capture coordinates of the plurality of measurement mark points included in the plurality of sample wafers;

[0145] A fitting module 106 is used to obtain a plurality of position error terms of each of the sample measurement mark points within and between fields of the target wafer based on the sample wafer alignment coordinates and the sample target capture coordinates of each of the sample measurement mark points, and perform high-order polynomial fitting on the plurality of position error terms to obtain a coordinate offset of each of the sample measurement mark points;

[0146] The first establishing module 107 is used to establish the target capture coordinate prediction model based on the sample wafer alignment coordinates and the coordinate offset of each of the sample measurement mark points.

[0147] In some feasible implementations, the wafer measurement device based on measurement mark points further includes a third acquisition module 108:

[0148] The third acquisition module 108 is used to acquire a first training sample set, wherein the first training sample set includes sample target capture coordinates, sample wafer alignment coordinates and measurement quality evaluation indicators of multiple sample measurement mark points, wherein the multiple sample measurement mark points include the multiple first measurement mark points and / or the multiple measurement mark points included in the multiple sample wafers, the sample target capture coordinates and the sample wafer alignment coordinates include the first target capture coordinates and the wafer alignment coordinates of the multiple first measurement mark points, and / or the wafer alignment coordinates and target capture coordinates of the multiple measurement mark points included in the multiple sample wafers.

[0149] In some feasible implementations, the wafer measurement device based on measurement mark points further includes a second establishment module 109:

[0150] The second establishment module 109 is used to establish the target capture coordinate prediction model using a target algorithm based on the sample target capture coordinates of each of the sample measurement mark points, the sample wafer alignment coordinates and the measurement quality evaluation index, wherein the target algorithm includes a gradient descent algorithm, a principal component analysis algorithm or a deep learning algorithm.

[0151] In some feasible implementations, the wafer measurement device based on measurement mark points further includes a first correction module 110:

[0152] The first correction module 110 is used to add the wafer alignment coordinates and the first target capture coordinates of each of the first measurement mark points to the first training sample set as correction training data of the target capture coordinate prediction model to perform correction training on the target capture coordinate prediction model.

[0153] In some feasible implementations, the first acquisition module 101 includes a first acquisition unit 1011 and an alignment unit 1012:

[0154] A first acquisition unit 1011 is used to acquire the wafer alignment coordinates of the alignment mark points used for wafer alignment in the target wafer;

[0155] The alignment unit 1012 is used to perform wafer alignment on the target wafer based on the wafer alignment coordinates of the alignment mark points to obtain the wafer alignment coordinates of the multiple measurement mark points.

[0156] In some feasible implementations, the first acquisition module 101 further includes a second acquisition unit 1013 and a prediction unit 1014:

[0157] A second acquisition unit 1013 is used to acquire wafer alignment coordinates of alignment mark points in the target wafer for wafer alignment, and perform wafer alignment on the target wafer based on the wafer alignment coordinates of the alignment mark points to obtain initial wafer alignment coordinates of the plurality of measurement mark points;

[0158] The prediction unit 1014 is used to input the initial wafer alignment coordinates of each of the above-mentioned measurement mark points into the wafer alignment coordinate correction model, and output the wafer alignment coordinates of each of the above-mentioned measurement mark points through the wafer alignment coordinate correction model.

[0159] In some feasible implementations, the first acquisition module 101 further includes a third acquisition unit 1015 and a training unit 1016:

[0160] The third acquisition unit 1015 is used to acquire a second training sample set, wherein the second training sample set includes sample target capture coordinates and sample wafer alignment coordinates of multiple sample measurement mark points included in multiple sample wafers, and one of the sample measurement mark points corresponds to one of the sample wafer alignment coordinates and one of the sample wafer alignment coordinates;

[0161] The training unit 1016 is used to calculate the coordinate difference between the sample wafer alignment coordinates of each of the sample measurement mark points and the sample target capture coordinates, and establish the wafer alignment coordinate correction model based on the sample wafer alignment coordinates of each of the multiple sample measurement mark points and the coordinate difference.

[0162] In some feasible implementations, the wafer measurement device based on measurement mark points further includes a third acquisition module 111 and a second correction module 112:

[0163] A third acquisition module 111 is used to obtain the wafer alignment coordinate error of each of the first measurement mark points based on the first target capture coordinates of each of the first measurement mark points and the wafer alignment coordinates of each of the first measurement mark points;

[0164] The second correction module 112 is used to add the wafer alignment coordinates and the wafer alignment coordinate errors of each of the first measurement mark points to the second training sample set as correction training data of the wafer alignment coordinate correction model to perform correction training on the wafer alignment coordinate correction model.

[0165] In some feasible implementations, the wafer measurement device based on measurement mark points further includes a construction module 113 and a second prediction module 114:

[0166] A construction module 113, configured to construct an in-field coordinate prediction model based on the wafer alignment coordinates and the first target capture coordinates of the plurality of first measurement mark points included in the target field, and the wafer alignment coordinates and the second target capture coordinates of the plurality of second measurement mark points;

[0167] The second prediction module 114 is used to input the wafer alignment coordinates of each measurement mark point included in any one of the above-mentioned multiple fields except the above-mentioned target field into the above-mentioned in-field coordinate prediction model when the wafer alignment coordinates of multiple measurement mark points included in any one of the above-mentioned multiple fields except the above-mentioned target field are acquired, and output the target capture coordinates of the above-mentioned each measurement mark point in any one of the above-mentioned fields through the above-mentioned in-field coordinate prediction model to complete the measurement of the above-mentioned each measurement mark point included in any one of the above-mentioned fields.

[0168] In some feasible implementations, the implementation of each module and / or unit included in the above wafer measurement device can refer to the above Figures 3 to 6 The implementation methods provided in each step of the wafer measurement method based on measurement mark points are not described in detail.

[0169] In the present application, a wafer measurement device based on measurement mark points can obtain a first target capture coordinate by performing target capture on a first measurement mark point, and quickly and accurately obtain a second target capture coordinate of a second measurement mark point through a target capture coordinate prediction model, thereby realizing wafer measurement of a target wafer, shortening the positioning time of the measurement mark points, and improving the measurement speed of the measurement mark points. The operation is simple and the adaptability is strong. In addition, a wafer measurement device based on measurement mark points can quickly locate multiple measurement mark points through a target capture coordinate prediction model and a wafer alignment coordinate correction model, and the above multiple models can be applied to subsequent wafer measurements without re-establishing the model, thereby further shortening the wafer measurement time, and the above multiple models can be corrected during the wafer measurement process, so the prediction accuracy of the above multiple models is high.

[0170] Embodiment 6

[0171] See also Figure 8 , Figure 8 is a schematic diagram of the structure of the terminal device provided by this application. Figure 8As shown, the terminal device 100 may include: a processor 1001, a network interface 1004, a memory 1005, a controller 1006 and an image acquisition device 1007. In addition, the terminal device 100 may also include: a user interface 1003, and at least one communication bus 1002. The communication bus 1002 is used to realize the connection and communication between these components. The user interface 1003 may include a display screen (Display) and a keyboard (Keyboard), and the optional user interface 1003 may also include a standard wired interface and a wireless interface. The network interface 1004 may optionally include a standard wired interface and a wireless interface (such as a WI-FI interface). The memory 1005 includes a random access memory (RAM) and a non-volatile memory (NVM), such as an erasable programmable read-only memory (EPROM). The memory 1005 may also be at least one storage device located away from the aforementioned processor 1001. The controller 1006 can be used to receive data sent by the processor and generate control instructions to control devices such as a sports platform, and can execute the above Figures 1 to 6 The functions implemented by the controller involved in the illustrated embodiment. The image acquisition device 1007 may be a linear array camera or other device that can be used to scan the target wafer to obtain the target wafer image. Figure 8 As shown, the memory 1005 as a computer-readable storage medium may include an operating system, a network communication module, a user interface module, and a device control application. Figure 1 The processor 1001 described in this embodiment can execute the above Figure 1 The functions performed by the processor provided in the illustrated embodiment.

[0172] In such Figure 8 In the terminal device 100 shown, the network interface 1004 can provide a network communication function; the user interface 1003 is mainly used to provide an input interface for the user; and the processor 1001 and the image acquisition device 1007 can be used to call the device control application stored in the memory 1005 to implement the aforementioned Figures 3 to 6 The wafer measurement method based on measurement mark points in the illustrated embodiment.

[0173] It should be understood that the terminal device 100 described in the embodiment of the present application can execute the above Figures 3 to 6The wafer measurement method based on the measurement mark point of the embodiment shown is not described in detail. In addition, the description of the beneficial effects of the same method is not described in detail either.

[0174] In addition, it should be pointed out here that: the embodiment of the present application also provides a computer-readable storage medium, and the computer-readable storage medium stores a computer program executed by the wafer measurement method based on the measurement mark point mentioned above, and the computer program includes program instructions. When the processor executes the program instructions, the above-mentioned Figures 3 to 6 The above-mentioned wafer measurement method based on measurement mark points provided in the illustrated embodiment is therefore not described in detail. In addition, the description of the beneficial effects of using the same method is not described in detail. For technical details not disclosed in the computer-readable storage medium embodiment involved in this application, please refer to the description of the method embodiment of this application.

[0175] The above-mentioned computer-readable storage medium may be an internal storage unit of the terminal device provided in any of the aforementioned embodiments, such as a hard disk or memory of the terminal device. The computer-readable storage medium may also be an external storage device of the terminal device, such as a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, etc. equipped on the terminal device. Further, the computer-readable storage medium may also include both an internal storage unit of the terminal device and an external storage device. The computer-readable storage medium is used to store the computer program and other programs and data required by the terminal device. The computer-readable storage medium may also be used to temporarily store data that has been output or is to be output.

[0176] In addition, it should be noted that: the embodiment of the present application also provides a computer program product or a computer program, the computer program product or the computer program includes computer instructions, and the computer instructions are stored in a computer-readable storage medium. The processor of the terminal device loads and executes the computer instructions, so that the terminal device can execute the above Figures 3 to 6 The method provided by the illustrated embodiment.

[0177] Those skilled in the art will appreciate that the systems and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented with electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described in terms of function in the above description. Professional and technical personnel may use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

Claims

1. A wafer measurement method based on measurement mark points, characterized in that: include: Obtaining wafer alignment coordinates of a plurality of measurement mark points of a target wafer; Performing target capture on a plurality of first measurement mark points among the plurality of measurement mark points to obtain a coordinate offset of each of the first measurement mark points, and obtaining a first target capture coordinate of each of the first measurement mark points based on the wafer alignment coordinates and the coordinate offset of each of the first measurement mark points; Inputting the wafer alignment coordinates of a plurality of second measurement mark points among the plurality of measurement mark points except the plurality of first measurement mark points into a target capture coordinate prediction model, and outputting second target capture coordinates of each of the second measurement mark points through the target capture coordinate prediction model; The multiple first measurement mark points are measured based on the first target capture coordinates of each of the first measurement mark points, and the multiple second measurement mark points are measured based on the second target capture coordinates of each of the second measurement mark points to achieve wafer measurement of the target wafer.

2. The method according to claim 1, characterized in that: The first measurement mark point is a measurement mark point randomly selected from the multiple measurement mark points, or a measurement mark point located on the target motion trajectory from the multiple measurement mark points.

3. The method according to claim 1, characterized in that The outputting the second target capture coordinates of each second measurement mark point through the target capture coordinate prediction model includes: The coordinate offset of each second measurement mark point is obtained based on the wafer alignment coordinates of each second measurement mark point through the target capture coordinate prediction model, and the second target capture coordinates of each second measurement mark point are output based on the coordinate offset of each second measurement mark point and the wafer alignment coordinates of each second measurement mark point through the target capture coordinate prediction model.

4. The method according to any one of claims 1 to 3, characterized in that: The target capture coordinate prediction model is trained by the following method: Acquire a first training sample set, wherein the first training sample set includes sample target capture coordinates and sample wafer alignment coordinates of a plurality of sample measurement mark points, wherein the plurality of sample measurement mark points include the plurality of first measurement mark points and / or the plurality of measurement mark points included in the plurality of sample wafers, and the sample target capture coordinates and the sample wafer alignment coordinates include the first target capture coordinates and the wafer alignment coordinates of the plurality of first measurement mark points, and / or the wafer alignment coordinates and target capture coordinates of the plurality of measurement mark points included in the plurality of sample wafers; Based on the sample wafer alignment coordinates and the sample target capture coordinates of each of the sample measurement mark points, a plurality of position error terms of each of the sample measurement mark points within and between fields of the target wafer are obtained, and a high-order polynomial fitting is performed on the plurality of position error terms to obtain a coordinate offset of each of the sample measurement mark points; The target capture coordinate prediction model is established based on the sample wafer alignment coordinates and the coordinate offsets of each of the sample measurement mark points.

5. The method according to any one of claims 1 to 3, characterized in that: The target capture coordinate prediction model is trained by the following method: Acquire a first training sample set, wherein the first training sample set includes sample target capture coordinates, sample wafer alignment coordinates and measurement quality evaluation indicators of multiple sample measurement mark points, wherein the multiple sample measurement mark points include the multiple first measurement mark points and / or the multiple measurement mark points included in the multiple sample wafers, and the sample target capture coordinates and the sample wafer alignment coordinates include the first target capture coordinates and the wafer alignment coordinates of the multiple first measurement mark points, and / or the wafer alignment coordinates and target capture coordinates of the multiple measurement mark points included in the multiple sample wafers; Based on the sample target capture coordinates of each of the sample measurement mark points, the sample wafer alignment coordinates and the measurement quality evaluation index, a target capture coordinate prediction model is established using a target algorithm, wherein the target algorithm includes a gradient descent algorithm, a principal component analysis algorithm or a deep learning algorithm.

6. The method according to claim 4 or 5, characterized in that: The first training sample set includes sample wafer alignment coordinates and sample target capture coordinates of multiple sample measurement mark points included in multiple sample wafers. After obtaining the first target capture coordinates of each first measurement mark point based on the wafer alignment coordinates and the coordinate offset of each first measurement mark point, the method further includes: The wafer alignment coordinates and the first target capture coordinates of each of the first measurement mark points are added to the first training sample set as correction training data of the target capture coordinate prediction model to perform correction training on the target capture coordinate prediction model.

7. The method according to any one of claims 1 to 6, characterized in that: The step of obtaining wafer alignment coordinates of a plurality of measurement mark points of a target wafer includes: Acquiring wafer alignment coordinates of alignment mark points in the target wafer for wafer alignment; Wafer alignment is performed on the target wafer based on the wafer alignment coordinates of the alignment mark points to obtain the wafer alignment coordinates of the plurality of measurement mark points.

8. The method according to any one of claims 1 to 6, characterized in that: The step of obtaining wafer alignment coordinates of a plurality of measurement mark points of a target wafer includes: Acquire wafer alignment coordinates of alignment mark points in the target wafer for wafer alignment, and perform wafer alignment on the target wafer based on the wafer alignment coordinates of the alignment mark points to obtain initial wafer alignment coordinates of the plurality of measurement mark points; The initial wafer alignment coordinates of each of the measurement mark points are input into a wafer alignment coordinate correction model, and the wafer alignment coordinates of each of the measurement mark points are output through the wafer alignment coordinate correction model.

9. The method according to claim 8, characterized in that The wafer alignment coordinate correction model is trained by the following method: Acquire a second training sample set, wherein the second training sample set includes sample target capture coordinates and sample wafer alignment coordinates of a plurality of sample measurement mark points included in a plurality of sample wafers, wherein one of the sample measurement mark points corresponds to one of the sample wafer alignment coordinates and one of the sample wafer alignment coordinates; The coordinate difference between the sample wafer alignment coordinates of each of the sample measurement mark points and the sample target capture coordinates is calculated, and the wafer alignment coordinate correction model is established based on the sample wafer alignment coordinates of each of the plurality of sample measurement mark points and the coordinate difference.

10. The method according to claim 9, characterized in that After obtaining the first target capture coordinates of each of the first measurement mark points based on the wafer alignment coordinates and the coordinate offset of each of the first measurement mark points, the method further includes: Obtaining wafer alignment coordinate errors of each first measurement mark point based on the first target capture coordinates of each first measurement mark point and the wafer alignment coordinates of each first measurement mark point; The wafer alignment coordinates and the wafer alignment coordinate errors of each of the first measurement mark points are added to the second training sample set as correction training data of the wafer alignment coordinate correction model to perform correction training on the wafer alignment coordinate correction model.

11. The method according to any one of claims 1 to 6, characterized in that: The target wafer includes a plurality of fields, the plurality of measurement mark points are a plurality of measurement mark points included in the target field in the plurality of fields, and the distribution state of the measurement mark points in each field in the plurality of fields is the same; After the wafer measurement of the target wafer is achieved, the method further includes: Constructing an in-field coordinate prediction model based on the wafer alignment coordinates and the first target capture coordinates of the plurality of first measurement mark points included in the target field, and the wafer alignment coordinates and the second target capture coordinates of the plurality of second measurement mark points; When the wafer alignment coordinates of multiple measurement mark points included in any one of the multiple fields except the target field are obtained, the wafer alignment coordinates of each measurement mark point included in any one of the multiple fields are input into the in-field coordinate prediction model, and the target capture coordinates of each measurement mark point in any one of the fields are output through the in-field coordinate prediction model to complete the measurement of each measurement mark point included in any one of the fields.

12. A wafer measurement device based on measurement mark points, characterized in that: include: An acquisition module, for acquiring wafer alignment coordinates of a plurality of measurement mark points of a target wafer; a target capture module, performing target capture on a plurality of first measurement mark points among the plurality of measurement mark points to obtain a coordinate offset of each of the first measurement mark points, and obtaining a first target capture coordinate of each of the first measurement mark points based on the wafer alignment coordinates and the coordinate offset of each of the first measurement mark points; A prediction module, inputting the wafer alignment coordinates of a plurality of second measurement mark points among the plurality of measurement mark points except the plurality of first measurement mark points into a target capture coordinate prediction model, and outputting second target capture coordinates of each of the second measurement mark points through the target capture coordinate prediction model; A measurement module is used to measure the multiple first measurement mark points based on the first target capture coordinates of each of the first measurement mark points, and to measure the multiple second measurement mark points based on the second target capture coordinates of each of the second measurement mark points, so as to achieve wafer measurement of the target wafer.

13. A terminal device, characterized in that: include: image acquisition device, processor and memory; The processor is connected to the memory, wherein the memory is used to store program codes, and the image acquisition device and the processor are used to call the program codes to execute any one of the methods described in claims 1-11.

14. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and the computer program is suitable for being loaded by a processor and executing the method according to any one of claims 1 to 11.

Images