Wafer overlay method and device, electronic equipment and storage medium

Abstract

The embodiment of the invention provides a wafer overlay method and device, electronic equipment and a storage medium. The wafer overlay method comprises the following steps: determining position information of a plurality of target measurement points of a target wafer by using a trained measurement point prediction model, and further determining an overlay compensation amount; performing overlay operation on the target wafer based on the overlay compensation amount; and based on the position information of the plurality of target measurement points and the difference between the first compensation residual error and the second compensation residual error, adjusting parameters of the measurement point prediction model for subsequent overlay operation. The target measurement point is intelligently and automatically determined, the determined target measurement point can more accurately reflect the overall overlay error distribution of the target wafer, measurement point redundancy is avoided, the measurement efficiency is improved, iterative optimization can be performed on the measurement point prediction model, data driving is realized, and the method dynamically adapts to a new process and a new design.

Description

Technical Field

[0001] This invention relates to the field of semiconductor technology. More specifically, it relates to a wafer overlay method, a wafer overlay apparatus, an electronic device, a storage medium, and a computer program product. Background Technology

[0002] Overlay refers to the alignment between different layers in multiple photolithography steps. Overlay error describes the alignment deviation between the current layer and the reference layer, and its magnitude directly determines device performance and production yield. To monitor and control overlay error, representative physical locations on the wafer surface need to be selected as measurement points. The selection of these measurement points needs to efficiently and comprehensively reflect the overlay error distribution across the entire wafer, thus providing a basis for overlay compensation. Too many measurement points will significantly increase measurement time, affecting production line efficiency; too few measurement points will fail to fully capture the error distribution characteristics on the wafer, resulting in insufficient overlay compensation and consequently reducing product yield.

[0003] In related technologies, the selection and updating of measurement points during the overlay process usually rely on manual experience, which not only leads to low accuracy in the selection of measurement points, but also makes it difficult to quickly adapt to constantly changing new processes and designs. Summary of the Invention

[0004] The present invention was proposed in view of the above-mentioned problems.

[0005] According to one aspect of the present invention, a wafer overlay method is provided. The wafer overlay method includes:

[0006] The location information of multiple target measurement points on the target wafer is determined by using a trained measurement point prediction model.

[0007] Based on the location information of multiple target measurement points, the overlay compensation amount is determined;

[0008] Overlay operation of the target wafer is performed based on overlay compensation amount;

[0009] Based on the location information of multiple target measurement points and the difference between the first compensation residual and the second compensation residual, the parameters of the measurement point prediction model are adjusted for subsequent overlay operations. The first compensation residual is calculated based on the current measurement point, and the second compensation residual is calculated based on the remaining measurement points obtained after deleting the measurement points to be measured from the current measurement point. The initial current measurement point consists of multiple target measurement points.

[0010] For example, the process of adjusting the parameters of the measurement point prediction model includes one or more deletion operations until a first preset requirement is met. The parameters of the measurement point prediction model are modified after each update of the current measurement point. The deletion operation includes: determining the first compensation residual of the current measurement point; determining the measurement point to be measured in the current measurement point; determining the second compensation residual of the remaining measurement points in the current measurement point; calculating the first residual difference index value based on the difference between the first compensation residual and the second compensation residual, wherein the larger the difference, the larger the first residual difference index value; if the first residual difference index value is less than or equal to the first difference threshold, then at least the measurement point to be measured is deleted in the current measurement point to update the current measurement point; if the first residual difference index value is greater than the first difference threshold, then the measurement point to be measured is determined to be non-deletable.

[0011] For example, multiple target measurement points are located in different exposure fields, and one or more deletion operations include: multiple deletion operations in the first elimination stage; in the first elimination stage, each time a deletion operation is performed, the determined measurement points to be measured are located in different exposure fields, and if the first residual difference index value is less than or equal to the first difference threshold, then all measurement points in the exposure field corresponding to the current measurement point to be measured are deleted in the current measurement point to update the current measurement point.

[0012] For example, one or more deletion operations further include: multiple groups of deletion operations in the second elimination stage; wherein, when the second preset requirement is met, the process transitions from the first elimination stage to the second elimination stage, each group of deletion operations includes at least one deletion operation executed consecutively, the quantitative measurement points determined in at least one deletion operation in each group of deletion operations are located in the same exposure field, and the quantitative measurement points determined in different groups of deletion operations are located in different exposure fields.

[0013] For example, one or more deletion operations may be executed synchronously, wherein the quantitative measurement points determined in the synchronously executed deletion operations are different.

[0014] For example, the measurement point prediction model is obtained by training with training data, which includes process parameter data of the training wafer. Using the trained measurement point prediction model, the location information of multiple target measurement points of the target wafer is determined, including: inputting the process parameter data of the target wafer into the trained measurement point prediction model to output the location information of multiple target measurement points of the target wafer.

[0015] For example, the training data also includes preset point selection requirements, which include one or more of the following: measurement point quantity requirements, measurement point distribution requirements, compensation effect requirements, and measurement capacity requirements. The method further includes: providing a first human-machine interface to a user; in response to a user's selection operation for at least one preset point selection requirement using the first human-machine interface, determining a target point selection requirement from the preset point selection requirements; inputting the process parameter data of the target wafer into the trained measurement point prediction model to output the location information of multiple target measurement points of the target wafer, including: inputting both the process parameter data and the target point selection requirements into the measurement point prediction model to obtain the location information of multiple target measurement points of the target wafer.

[0016] For example, the training data also includes the location information of historical measurement points on the training wafer and the overlay error measurement data corresponding to the historical measurement points.

[0017] For example, the training data also includes the measurement label type of historical measurement points, and the measurement point prediction model also outputs the measurement label type of each of the multiple target measurement points. The measurement label type of the target measurement point is used to determine the measurement method of the target measurement point. Based on the position information of the multiple target measurement points, the overlay compensation amount is determined, including: using the determined measurement method and based on the position information of the multiple target measurement points, the overlay compensation amount is determined.

[0018] According to another aspect of the present invention, a wafer overlay apparatus is also provided. The wafer overlay apparatus includes: a first determining module, a second determining module, an overlay module, and an adjustment module. The first determining module is used to determine the position information of multiple target measurement points on a target wafer using a trained measurement point prediction model; the second determining module is used to determine an overlay compensation amount based on the position information of the multiple target measurement points; the overlay module is used to perform an overlay operation on the target wafer based on the overlay compensation amount; the adjustment module is used to adjust the parameters of the measurement point prediction model based on the position information of the multiple target measurement points and the difference between the first compensation residual and the second compensation residual, for subsequent overlay operations, wherein the first compensation residual is calculated based on the current measurement point, and the second compensation residual is calculated based on the remaining measurement points obtained after deleting the measurement points to be measured from the current measurement point, and the initial current measurement point is the multiple target measurement points.

[0019] According to another aspect of the present invention, an electronic device is also provided, comprising: a processor and a memory, wherein the memory stores computer program instructions, which are executed by the processor to perform the wafer overlay method as described above.

[0020] According to another aspect of the present invention, a storage medium is also provided, on which program instructions are stored, which, when executed, are used to perform the wafer overlay method as described above.

[0021] According to another aspect of the present invention, a computer program product is also provided, comprising computer program instructions which, when executed, are used to perform the wafer overlay method as described above.

[0022] In the above technical solution, a trained measurement point prediction model is used to determine the position information of multiple target measurement points on the target wafer; based on the position information of multiple target measurement points, an overlay compensation amount is determined; an overlay operation is performed on the target wafer based on the overlay compensation amount; based on the position information of multiple target measurement points and the difference between the first compensation residual and the second compensation residual, the parameters of the measurement point prediction model are adjusted for subsequent overlay operations. Therefore, target measurement points can be determined intelligently and automatically. The determined target measurement points can more accurately reflect the overall overlay error distribution of the target wafer, avoid measurement point redundancy, improve measurement efficiency, and allow for iterative optimization of the measurement point prediction model to achieve data-driven, dynamic adaptation to new processes and designs.

[0023] The above description is merely an overview of the technical solution of the present invention. In order to better understand the technical means of the present invention and to implement it in accordance with the contents of the specification, and to make the above and other objects, features and advantages of the present invention more apparent and understandable, specific embodiments of the present invention are described below. Attached Figure Description

[0024] The above and other objects, features, and advantages of the present invention will become more apparent from the more detailed description of the embodiments of the invention in conjunction with the accompanying drawings. The drawings are provided to further illustrate the embodiments of the invention and form part of the specification. They are used together with the embodiments of the invention to explain the invention and do not constitute a limitation thereof.

[0025] Figure 1 A schematic flowchart of a wafer overlay method according to an embodiment of the present invention is shown;

[0026] Figure 2 A schematic flowchart of a deletion operation according to an embodiment of the present invention is shown;

[0027] Figure 3 A schematic block diagram of a wafer overlay apparatus according to an embodiment of the present invention is shown;

[0028] Figure 4 A schematic block diagram of an electronic device according to an embodiment of the present invention is shown. Detailed Implementation

[0029] To make the objectives, technical solutions, and advantages of the present invention more apparent, exemplary embodiments according to the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are merely a subset of embodiments of the present invention.

[0030] To at least address the aforementioned technical problems, this invention provides a wafer overlay method. This wafer overlay method utilizes a trained measurement point prediction model to determine the position information of multiple target measurement points, thereby determining the overlay compensation amount and performing the target wafer overlay operation. The parameters of the measurement point prediction model are then adjusted for subsequent overlay operations. Thus, target measurement points can be determined intelligently and automatically, and the measurement point prediction model can be iteratively optimized to achieve data-driven, dynamic adaptation to new processes and designs. This wafer overlay method can be applied to any electronic device, i.e., executed by any electronic device. Specifically, the wafer overlay method can be executed by the processor of any electronic device.

[0031] Figure 1 A schematic flowchart of a wafer overlay method according to an embodiment of the present invention is shown. Figure 1 As shown, the wafer overlay method includes steps S1100, S1200, S1300 and S1400.

[0032] In step S1100, the position information of multiple target measurement points on the target wafer is determined using a trained measurement point prediction model. The measurement point prediction model can be an algorithmic model built based on machine learning or statistical modeling. The measurement point prediction model can be used to determine the position information of multiple target measurement points on the target wafer. The measurement point prediction model can be implemented using various model architectures such as Convolutional Neural Networks (CNN) and Support Vector Machines (SVM). For example, the measurement point prediction model can be a multimodal intelligent model using a Transformer architecture. The trained measurement point prediction model can be obtained by training using training data. The training data may include wafer process parameter data, historical measurement point position information, and overlay error measurement data corresponding to historical measurement points. The trained measurement point prediction model can determine the position information of multiple target measurement points for the target wafer. In some embodiments, the trained measurement point prediction model can determine the position information of multiple target measurement points based on the process parameter data of the target wafer. In other embodiments, the trained measurement point prediction model can determine the location information of multiple target measurement points based on the physical characteristics of the target wafer (such as wafer warpage, film thickness uniformity, etc.).

[0033] The location information of multiple target measurement points can include the actual coordinates of each target measurement point on the target wafer. The location information of multiple target measurement points can also include the exposure field where each target measurement point is located, and the coordinates of that target measurement point within the exposure field.

[0034] In step S1200, the overlay compensation amount is determined based on the position information of multiple target measurement points. According to the position information of the multiple target measurement points, the specific position of each target measurement point in the target wafer can be determined. Measurement equipment can be used to measure each of the multiple target measurement points on the target wafer to obtain the overlay error measurement data for each target measurement point. Based on the overlay error measurement data for each target measurement point, the overlay compensation amount is calculated using an overlay error compensation model. The overlay compensation amount can include lateral displacement, longitudinal displacement, rotation angle, scaling, etc. The overlay error compensation model can include any existing or future overlay error compensation model. For example, the overlay error compensation model can be a high-order polynomial fitting model, which can fit the overlay error distribution based on the overlay error measurement data of each target measurement point, and then calculate the overlay compensation amount to compensate for the overlay error in the next overlay operation.

[0035] In step S1300, the overlay operation of the target wafer is performed based on the overlay compensation amount. During the next overlay operation, the overlay error can be compensated based on the overlay compensation amount, thereby reducing the overlay error. For example, when the photolithography equipment performs the next overlay operation on the target wafer, the relative position of the mask and the target wafer can be adjusted based on the overlay compensation amount to compensate for the overlay error and achieve high-precision overlay operation.

[0036] In step S1400, based on the location information of multiple target measurement points and the difference between the first compensation residual and the second compensation residual, the parameters of the measurement point prediction model are adjusted for subsequent overlay operations. The first compensation residual is calculated based on the current measurement point, and the second compensation residual is calculated based on the remaining measurement points obtained after deleting the measurement points to be measured from the current measurement point. Initially, the current measurement points are multiple target measurement points.

[0037] First, the first compensation residual can be calculated based on the current measurement point (initially, multiple target measurement points). The first compensation residual represents the difference between the predicted error and the actual error at the current measurement point. After deleting the measurement point to be measured from the current measurement point, the second compensation residual can be calculated based on the remaining measurement points. The second compensation residual represents the difference between the predicted error and the actual error at the remaining measurement points. If the difference between the first and second compensation residuals is less than or equal to a difference threshold, it can be determined that the measurement point to be measured has a small impact on the predicted error and can be deleted. The vector measurement point prediction model can then provide feedback that the measurement point to be measured can be deleted, and the output predicted measurement points may not include the measurement point to be measured, thus adjusting the parameters of the measurement point prediction model. If the difference between the first and second compensation residuals is greater than a difference threshold, it can be determined that the measurement point to be measured has a large impact on the predicted error and cannot be deleted. The vector measurement point prediction model can then provide feedback that the measurement point to be measured cannot be deleted, and the output predicted measurement points must include the measurement point to be measured, thus adjusting the parameters of the measurement point prediction model. For example, the weight parameters of the measurement point prediction model can be adjusted through backpropagation or reinforcement learning mechanisms, making it more inclined to retain measurement points that have a greater impact on prediction errors in future predictions. This adjusted measurement point prediction model can then be used for the overlay operation of the next wafer or the next batch of wafers. The first and second compensation residuals can be calculated in any way. For instance, based on the overlay error measurement data of each measurement point in the current measurement points, the overlay error prediction data can be calculated using the overlay error compensation model. Then, based on the overlay error measurement data and the overlay error prediction data of the current measurement point, the first compensation residual for the current measurement point can be determined. Similarly, the second compensation residual for the remaining measurement points can be determined based on the overlay error measurement data and the overlay error prediction data of the remaining measurement points.

[0038] In the above technical solution, a trained measurement point prediction model is used to determine the position information of multiple target measurement points on the target wafer; based on the position information of multiple target measurement points, an overlay compensation amount is determined; an overlay operation is performed on the target wafer based on the overlay compensation amount; based on the position information of multiple target measurement points and the difference between the first compensation residual and the second compensation residual, the parameters of the measurement point prediction model are adjusted for subsequent overlay operations. Therefore, target measurement points can be determined intelligently and automatically. The determined target measurement points can more accurately reflect the overall overlay error distribution of the target wafer, avoid measurement point redundancy, improve measurement efficiency, and allow for iterative optimization of the measurement point prediction model to achieve data-driven, dynamic adaptation to new processes and designs.

[0039] For example, the process of adjusting the parameters of the measurement point prediction model includes one or more deletion operations until a first preset requirement is met. The parameters of the measurement point prediction model are modified after each update of the current measurement point. In the multiple deletion operations, at least some deletion operations may delete at least one measurement point to be measured from the current measurement point to update the current measurement point. After each update of the current measurement point, feedback can be provided to the measurement point prediction model to modify its parameters. For instance, based on the updated current measurement point, the weight parameters of the measurement point prediction model can be adjusted through backpropagation or reinforcement learning mechanisms.

[0040] Figure 2 A schematic flowchart of a deletion operation according to an embodiment of the present invention is shown. Figure 2 As shown, the deletion operation includes steps S2100, S2200, S2300, S2400, S2500 and S2600.

[0041] In step S2100, the first compensation residual of the current measurement point is determined. Multiple target measurement points output by the measurement point prediction model can be used as the current measurement point when the deletion operation is first performed. The compensation residual represents a local deviation determined based on the measurement point that cannot be covered by the global model. Its sources include local wafer stress, uneven photoresist coating, and local fluctuations in the etching process.

[0042] In a specific example, actual overlay measurements can be performed on each measurement point in the current measurement point to determine its respective overlay error measurement data. An overlay error compensation model can be used to fit the measured data of each measurement point to determine the predicted overlay error data for each measurement point in the current measurement point. The aforementioned overlay error compensation model can be any existing or future overlay error compensation model. Optionally, an overlay error compensation model suitable for the wafer's process parameters can be selected. Then, based on the overlay error measurement data and predicted overlay error data of the current measurement point, the first compensation residual for the current measurement point can be determined. For example, the average difference between the measured overlay error and the predicted overlay error in the lateral direction of all measurement points in the current measurement point can be calculated as the first difference, and the average difference between the measured overlay error and the predicted overlay error in the longitudinal direction of all measurement points in the current measurement point can be calculated as the second difference. The first compensation residual can be calculated based on the first and second differences.

[0043] It is understandable that any existing or future research and development method can be used to determine the first compensation residual of the current measurement point.

[0044] In step S2200, the measurement point to be measured is determined in the current measurement point.

[0045] The measurement points to be quantified can include one or more measurement points from the current measurement points. Optionally, the user can set the number of measurement points to be quantified. For example, measurement points to be quantified can be randomly determined from the current measurement points. It can be understood that measurement points determined as non-deletable are those that have undergone deletion operations but cannot be removed. Therefore, in a newly executed deletion operation, measurement points to be quantified can be determined from measurement points other than those determined as non-deletable in the current measurement points. Optionally, the measurement point prediction model can also be used to determine measurement points to be quantified from the current measurement points.

[0046] In step S2300, the second compensation residual of the remaining measurement points in the current measurement point is determined. The remaining measurement points include other measurement points besides the one to be measured in the current measurement point. For example, based on the overlay error measurement data of each measurement point in the remaining measurement points, the predicted overlay error data of each measurement point in the remaining measurement points can be determined using an overlay error compensation model. Then, based on the overlay error measurement data and the predicted overlay error data of each measurement point in the remaining measurement points, the second compensation residual of the remaining measurement points can be determined. The same or different overlay error compensation model as in step S2100 above can be used to determine the predicted overlay error data of each measurement point in the remaining measurement points. The same or different calculation method as in step S2100 above can be used to determine the second compensation residual of the remaining measurement points. For example, the average of the differences between the measured overlay error and the predicted overlay error in the lateral direction of all measurement points in the remaining measurement points can be calculated as the third difference. The average of the differences between the measured overprinting error and the predicted overprinting error in the longitudinal direction of all remaining measurement points can be calculated as the fourth difference. The second compensation residual can then be calculated based on the third and fourth differences. Any existing or future developed method can be used to determine the second compensation residual for the remaining measurement points.

[0047] In step S2400, based on the difference between the first compensation residual and the second compensation residual, the first residual difference index value is calculated, wherein the larger the difference, the larger the first residual difference index value.

[0048] In some embodiments, the difference between the first and second compensated residuals can be assessed by the difference between them. A larger difference indicates a greater difference, and thus a higher first residual difference index value. For example, the absolute value of the difference between the first and second compensated residuals can be calculated as the first residual difference index value. In other embodiments, the difference between the first and second compensated residuals can be assessed by the ratio of the first and second compensated residuals. The ratio of the larger of the first and second compensated residuals to the smaller of the second compensated residuals can be used as the first residual difference index value; a larger ratio indicates a higher first residual difference index value.

[0049] Each deletion operation can be performed by selecting either step S2500 or step S2600 based on the relationship between the first residual difference index value and the first difference threshold.

[0050] In step S2500, if the first residual difference index value is less than or equal to the first difference threshold, at least one measurement point to be measured is deleted from the current measurement point to update the current measurement point; step S2500 can be executed if the first residual difference index value is less than or equal to the first difference threshold. In other words, if the first residual difference index value is less than or equal to the first difference threshold, at least one measurement point to be measured can be deleted from the current measurement point to update the current measurement point.

[0051] In some embodiments, only the measurement point to be measured can be deleted from the current measurement point. In other embodiments, the measurement point to be measured and its associated measurement points can be deleted from the current measurement point. For example, the measurement point to be measured and all measurement points in its corresponding exposure field can be deleted from the current measurement point. It is understood that if the first residual difference index value is less than or equal to the first difference threshold, it indicates that a good compensation effect can still be achieved after the measurement point to be measured is deleted. The measurement point to be measured can be considered a redundant measurement point, and therefore it can be deleted. The measurement point prediction model can be used to provide feedback that the measurement point to be measured can be deleted, so as to modify the parameters of the measurement point prediction model. If the measurement point to be measured is redundant, the measurement points associated with it, such as measurement points whose distance to it is less than the distance threshold, are also likely to be redundant. Optionally, other measurement points associated with it can be deleted at the same time as the measurement point to be measured.

[0052] In step S2600, if the first residual difference index value is greater than the first difference threshold, then the measurement point to be quantified is determined to be non-deletable. Step S2600 is executed when the first residual difference index value is greater than the first difference threshold. When the first residual difference index value is greater than the first difference threshold, it can be determined that the measurement point to be quantified is non-deletable. For example, the measurement point to be quantified can be marked as non-deletable in the current measurement points. It can be understood that if the first residual difference index value is greater than the first difference threshold, it indicates that deleting the measurement point to be quantified would not achieve a good compensation effect, and the measurement point to be quantified can be considered a necessary measurement point; therefore, the measurement point to be quantified can be determined as non-deletable.

[0053] Optionally, the vector measurement point prediction model can provide feedback that the measurement point to be measured cannot be deleted, and the output predicted measurement points must include the measurement point to be measured.

[0054] In the above technical solution, the process of adjusting the parameters of the measurement point prediction model includes one or more deletion operations until the first preset requirement is met. The parameters of the measurement point prediction model are modified after each update of the current measurement point. The deletion operation includes: determining the first compensation residual of the current measurement point; determining the measurement point to be measured in the current measurement point; determining the second compensation residual of the remaining measurement points in the current measurement point; calculating the first residual difference index value based on the difference between the first and second compensation residuals; if the first residual difference index value is less than or equal to the first difference threshold, then at least the measurement point to be measured is deleted from the current measurement point to update the current measurement point; if the first residual difference index value is greater than the first difference threshold, then the measurement point to be measured is determined to be non-deletable. Therefore, by filtering the target measurement points using the first residual difference index value, it can be ensured that the target measurement points output by the adjusted measurement point prediction model can both reduce redundancy of target measurement points and ensure comprehensive coverage of target measurement points, thereby improving the performance of the measurement point prediction model.

[0055] For example, the first preset requirement may include: all measurement points in the current measurement point cannot be deleted.

[0056] After performing one or more deletion operations, if all measurement points in the current measurement point are marked as non-deletable, then the first preset requirement is met, and further deletion operations can be discontinued. The measurement points in the initial measurement point set, excluding those deleted at the end, are defined as the final measurement points. For example, if the current measurement point set includes three measurement points when the first deletion operation is performed, and measurement point 1 is deleted, then the current measurement point set includes measurement points 2 and 3. If the second deletion operation determines that measurement point 2 is non-deletable, then the current measurement point set still includes measurement points 2 and 3, and measurement point 2 is marked as non-deletable. If the third deletion operation determines that measurement point 3 is non-deletable, then the current measurement point set still includes measurement points 2 and 3, and both measurement points 2 and 3 are marked as non-deletable. At this point, all measurement points in the current measurement point set are non-deletable, meeting the first preset requirement.

[0057] In the above technical solution, the first preset requirement includes: all measurement points in the current measurement point cannot be deleted. This ensures comprehensive measurement point coverage, better supporting the overlay error compensation model, and allows the deletion operation to terminate promptly once all measurement points are deemed undeletable, guaranteeing its effectiveness.

[0058] For example, the first preset requirement may include: the current number of measurement points equals a measurement point number threshold. After each deletion operation, the current number of measurement points can be determined. If the determined number equals the measurement point number threshold, then the first preset requirement is met. For example, the measurement point number threshold is 3. If, during the first deletion operation, the current number of measurement points includes 5 points, and the first deletion operation deletes measurement point 1, the second deletion operation determines that measurement point 2 cannot be deleted, and the third deletion operation deletes measurement point 4, then the current number of measurement points is 3, which equals the measurement point number threshold and meets the first preset requirement. Users can set the measurement point number threshold as needed. For example, the measurement point number threshold can be set according to the requirements of the overlay error compensation model.

[0059] In the above technical solution, the first preset requirement includes: the current number of measurement points equals the measurement point number threshold. This avoids deleting too many measurement points, ensures comprehensive measurement point coverage, and better supports the overlay error compensation model.

[0060] For example, it can be determined at the same time whether all measurement points in the current measurement point are not deletable, and whether the number of current measurement points is equal to the measurement point number threshold. If "all measurement points in the current measurement point are not deletable" or "the number of current measurement points is equal to the measurement point number threshold", it can be determined that the first preset requirement is met.

[0061] For example, multiple target measurement points are located in different exposure fields. The target wafer may include multiple exposure fields, and the target measurement points may be located in the exposure fields. It is understood that the exposure fields in which the multiple target measurement points are located may be part or all of the exposure fields of the target wafer.

[0062] The aforementioned one or more deletion operations include: multiple deletion operations in the first elimination stage; in the first elimination stage, each time a deletion operation is performed, the determined measurement point to be measured is located in a different exposure field, and if the first residual difference index value is less than or equal to the first difference threshold, then all measurement points in the exposure field corresponding to the current measurement point to be measured are deleted in the current measurement point to update the current measurement point.

[0063] The first elimination phase may include performing at least one deletion operation. During the deletion operation in the first elimination phase, the measurement point to be measured, determined in step S2200, is located in a different exposure field each time the deletion operation is performed. In some embodiments, the current measurement point at the time of the first deletion operation is located in 10 exposure fields. One measurement point can be determined in each of these 10 exposure fields, resulting in 10 measurement points. When performing the deletion operation corresponding to the first elimination phase, one of these 10 measurement points can be determined as the measurement point to be measured. Since these 10 measurement points are located in different exposure fields, it can be ensured that the measurement point to be measured determined during the deletion operation corresponding to the first elimination phase is located in different exposure fields. In other embodiments, the exposure fields where the current measurement point is located can be determined. When determining the measurement point to be measured during the deletion operation in the first elimination phase, any one or more measurement points in different exposure fields can be determined as the measurement point to be measured in this deletion operation, based on the exposure fields corresponding to the measurement points to be measured determined in previous deletion operations in the first elimination phase. For example, if the current measurement point is located in three exposure fields, namely exposure field 1, exposure field 2 and exposure field 3, during the first elimination stage, when performing the first deletion operation, the measurement point to be measured can be determined in any of the three exposure fields. For example, the measurement point to be measured can be determined in exposure field 1. When performing the second deletion operation, the measurement point to be measured can be determined in any of the two exposure fields, exposure field 2 and exposure field 3.

[0064] During the first elimination phase, if the determined first residual difference index value is less than or equal to the first difference threshold, all measurement points in the exposure field corresponding to the measurement point to be measured are deleted from the current measurement point to update the current measurement point. Alternatively, if the first residual difference index value is less than or equal to the first difference threshold during the first elimination phase, all measurement points in the exposure field where the measurement point to be measured is located, as determined by the deletion operation, can be identified as redundant measurement points. In this case, all measurement points in the exposure field corresponding to the measurement point to be measured can be deleted from the current measurement point to update the current measurement point. This deletion operation is performed once or multiple times until the first preset requirement is met.

[0065] In the above technical solution, during the first elimination stage, each time a deletion operation is performed, the determined measurement points to be quantified are located in different exposure fields. Furthermore, if the first residual difference index value is less than or equal to the first difference threshold, all measurement points in the exposure field corresponding to the measurement point to be quantified are deleted from the current measurement point to update the current measurement point. This speeds up the execution efficiency of the deletion operation, reduces the number of deletion operations, and allows for rapid adjustment of the measurement point prediction model.

[0066] For example, the above-mentioned one or more deletion operations further include: multiple groups of deletion operations in the second elimination stage, wherein when the second preset requirement is met, the process is transferred from the first elimination stage to the second elimination stage, each group of deletion operations includes at least one deletion operation executed consecutively, the quantitative measurement points determined in at least one deletion operation in each group of deletion operations are located in the same exposure field, and the quantitative measurement points determined in different groups of deletion operations are located in different exposure fields.

[0067] When the second preset requirement is met, the first elimination stage can end and the second elimination stage can begin. The second preset requirement may include that the current number of measurement points is less than or equal to a measurement point quantity threshold. It may also include that the number of times the deletion operation is executed in the first elimination stage equals the first threshold. Furthermore, it may include that the number of times the action "delete all measurement points in the exposure field corresponding to the current measurement point" is executed equals the second threshold. Users can set any form of second preset requirement as needed.

[0068] The second removal phase can include multiple groups of deletion operations. Each group of deletion operations can be executed at least once. These groups can be executed in parallel or sequentially. For example, parallel execution means that multiple groups of deletion operations can be executed simultaneously; sequential execution means that the next group can only begin execution after the previous group has finished. Users can configure the execution of multiple groups of operations sequentially or in parallel as needed.

[0069] For example, at the end of the first elimination stage, the number of multiple deletion operations in the second elimination stage can be determined based on the number of exposure fields where the current measurement points are located at the end of the first elimination stage. Thus, multiple deletion operations can correspond one-to-one with the exposure fields where the current measurement points are located at the end of the first elimination stage. When at least one deletion operation of each group is executed, the measurement points to be measured determined in step S2200 of different groups are located in different exposure fields. At least one deletion operation of each group can determine the measurement points to be measured within the exposure field corresponding to that group of deletion operations; that is, when at least one deletion operation of that group executes step S2200, all determined measurement points to be measured are located within the exposure field corresponding to that group of deletion operations. Therefore, measurement points within the exposure field can be filtered and deleted. For example, at the end of the first elimination stage, the current measurement point is located in three exposure fields, namely exposure field 1, exposure field 2 and exposure field 3. The second elimination stage may include three sets of deletion operations. The first set of deletion operations can be performed on the measurement points in exposure field 1. When performing the deletion operation corresponding to the first set of deletion operations, the measurement point to be measured can be determined only among the measurement points in exposure field 1. Similarly, the second set of deletion operations can be performed on the measurement points in exposure field 2, and the third set of deletion operations can be performed on the measurement points in exposure field 3.

[0070] In the above technical solution, the multiple deletion operations in the second elimination stage transition from the first elimination stage to the second elimination stage when the second preset requirements are met. Each group of deletion operations includes at least one consecutive deletion operation. The measurement points to be measured in at least one deletion operation within each group are located within the same exposure field, and the measurement points to be measured in different groups of deletion operations are located within different exposure fields. Therefore, measurement points within each exposure field can be screened separately, allowing for the precise elimination of redundant measurement points and the retention of necessary ones. This enables fine-tuning of the measurement point prediction model. Furthermore, multiple groups of deletion operations can be executed serially, saving computational resources, or in parallel, accelerating processing speed.

[0071] For example, one or more deletion operations may be executed synchronously, wherein the quantitative measurement points determined in the synchronously executed deletion operations are different.

[0072] When there are a large number of measurement points on a wafer, multiple deletion operations can be performed simultaneously, each identifying a different measurement point to be measured. After one round of simultaneous deletion operations is completed, the next round of simultaneous deletion operations can identify different measurement points based on the points identified in the previous round, and these identified measurement points can be different from those identified in each previous round. The number of deletion operations in each round can be the same or different. For example, in the initial execution, if there are 36 measurement points, 18 deletion operations can be performed simultaneously. Each deletion operation selects one measurement point as the measurement point to be measured. After performing 18 deletion operations simultaneously, another 18 deletion operations can be performed simultaneously. By performing two consecutive rounds of simultaneous deletion operations, it is possible to determine which measurement points can be deleted and which cannot, quickly determining the final measurement points.

[0073] In the above technical solution, one or more deletion operations may involve simultaneous deletion operations, each identifying different measurement points. This significantly improves processing speed, enhances the ability to select measurement points, and allows for rapid adjustment of the measurement point prediction model.

[0074] For example, the measurement point prediction model is obtained by training with training data, which includes process parameter data of the training wafer. The process parameter data may include some or all of the following process parameter items: lithography machine model, lithography process node, wafer diameter, exposure energy, development time, film thickness, polishing pressure, exposure field size, and center exposure field position. The lithography machine model can represent the brand and specific model of the lithography machine; the lithography process node can represent the type or process node of the wafer; the wafer diameter can represent the physical size of the wafer; the exposure energy can represent the exposure energy of the lithography operation; the development time can represent the development time; the film thickness can represent the thickness of the layer where the measurement mark corresponding to the measurement point is located; the polishing pressure can represent the polishing pressure during wafer thinning; the exposure field size can represent the size of one exposure field on the wafer; and the center exposure field position can represent the offset coordinates of the center exposure field on the wafer surface relative to the physical center of the wafer.

[0075] Measuring point prediction models can be trained using process parameter data from training wafers. The training data can include process parameter data from multiple training wafers. The process parameter data from the training wafers can include some or all of the process parameter items mentioned above. An initial measuring point prediction model can be trained using this training data to obtain a trained measuring point prediction model. Fluctuations in process parameters can change the number and distribution of measuring points, thus affecting the accuracy of overlay error measurement. For example, the lithography process node can determine the density (number) and accuracy level of measuring points. The smaller the process node, the smaller the critical dimensions of components such as transistors on the wafer, the smaller the tolerance margin for overlay errors, and the higher the requirements for the density and coverage of measuring points. Specifically, for 90nm and above: overlay error requirements are more lenient (3σ≤5nm), error distribution is uniform, the number of measuring points is small, and the density is low. A distribution ratio of "one measuring point for every 20 to 30 chips" is sufficient, covering only a few locations at the wafer center and edges. For 28nm and below (advanced processes): Strict overlay error requirements (3σ≤1nm). Due to stress and equipment distortion, wafer edges exhibit significantly greater errors than the center, necessitating a substantial increase in the number and density of measurement points. A distribution of one measurement point for every 5 to 10 chips is recommended, with particular emphasis on sensitive areas such as edges and corners. For 3nm wafers, the number of measurement points can be 5 to 8 times that of 90nm wafers, with additional points added at the edge of each exposure field to capture local deviations between fields. Furthermore, the lithography machine model determines the lithography technology used on the wafer, significantly influencing the location of measurement points. Different lithography technologies have different error sources, requiring targeted placement of measurement points in error-sensitive locations. For example, in DUV lithography, errors primarily arise from lens distortion and wafer thermal expansion. Measurement points can be evenly distributed at the wafer center, edges, and corners, covering typical areas of lens distortion (e.g., edge distortion is greater than center distortion). DUV multiple exposure: Requires multiple overlay alignments, leading to interlayer error accumulation. Measurement points can be placed on critical layers of each pattern, such as contact hole layers and gate layers, coinciding with measurement points on previous layers to facilitate tracking accumulated errors. EUV lithography: Errors mainly arise from weak marker signals and wafer stress distortion. Measurement points can be placed in high-reflectivity areas of the dicing traces to match EUV-specific markers, avoiding marker obstruction caused by thin film deposition, and densifying stress-concentrated wafer edges. For thick-film processes with significant film thickness, such as SiC power devices: Etching and polishing processes can damage the overlay marks. Measurement points can be placed in areas with high process tolerance, such as redundant areas at chip corners, and the number of measurement points should be increased to prevent some points from failing due to process damage, thus reserving spare points. Furthermore, wafer diameter, exposure field size, and the position of the center exposure field all directly affect the location of measurement points.In summary, the wafer's process parameters determine the areas on the wafer prone to errors and the magnitude of those errors. Therefore, the number and location of measurement points can be determined based on these parameters. The more complex the error, the more measurement points are needed. Areas more prone to errors require more measurement points. Optionally, the training data also includes labeled data on the error distribution of the training wafer. The trained measurement point prediction model can learn the error distribution of different wafers and then determine measurement points based on this distribution, thereby reducing overlay errors on the wafer.

[0076] Step S1100 uses a trained measurement point prediction model to determine the location information of multiple target measurement points on the target wafer, including step S1110. In step S1110, the process parameter data of the target wafer is input into the trained measurement point prediction model to output the location information of multiple target measurement points on the target wafer.

[0077] A human-machine interface (HMI) can be provided, offering one or more process parameter data setting controls. Users can input process parameter data by triggering these controls, allowing the system to determine the process parameter data for the target wafer. For example, a user can use the process parameter data setting controls to determine the lithography machine needed to process the target wafer, and the system can automatically match the remaining process parameter data required by that machine for processing the target wafer. Optionally, the user can adjust the automatically matched process parameter data. In other embodiments, the process parameter data for the target wafer can be automatically determined. For example, the process parameter data can be automatically determined based on the type of the target wafer and processing requirements. Another example is acquiring pre-stored process parameter data for the target wafer sent from other electronic devices. The process parameter data for the target wafer can be input into a trained measurement point prediction model, which can output the location information of multiple target measurement points on the target wafer.

[0078] In the above technical solution, the measurement point prediction model is trained using training data, including wafer process parameter data. The process parameter data of the target wafer is input into the trained measurement point prediction model to output the location information of multiple target measurement points on the target wafer. Therefore, by determining target measurement points based on process parameter data, data-driven, intelligent, and automated target measurement point determination is possible. Furthermore, the determined target measurement points are adaptable to the process parameters of the target wafer, can capture local errors caused by specific process parameters, and can improve the effectiveness of the target measurement points.

[0079] For example, the training data also includes preset point selection requirements, which may include one or more of the following: measurement point quantity requirements, measurement point distribution requirements, compensation effect requirements, and measurement capacity requirements. When training the measurement point prediction model, information about the preset point selection requirements can be incorporated to ensure that the predicted measurement points output by the model meet these requirements. The measurement point prediction model can be trained based on one or more of the following: measurement point quantity requirements, measurement point distribution requirements, compensation effect requirements, and measurement capacity requirements. The measurement point quantity requirement can be used to limit the maximum or minimum number of predicted measurement points output. The measurement point distribution requirement can be used to limit the spatial distribution of the predicted measurement points output, such as limiting the dispersion of the predicted measurement points. The compensation effect requirement can be used to limit the effectiveness of the predicted measurement points for the overlay error compensation model; for example, it can limit the situation where the predicted measurement points can effectively characterize the error distribution to support the overlay error compensation model in accurately compensating for errors. Measurement capacity requirements can be used to limit the maximum or minimum capacity corresponding to the predicted measurement points. For example, the total measurement time for the predicted measurement points can be limited, thereby limiting the capacity corresponding to the predicted measurement points.

[0080] The wafer overlay method further includes steps S1500 and S1600. In step S1500, a first human-computer interaction interface is provided to the user. The first human-computer interaction interface can be an interface for input / output devices that establish a connection and exchange information with the computer system. Input / output devices may include a monitor, mouse, keyboard, trackball, touchpad, touch screen, etc.

[0081] In step S1600, in response to a user's selection operation using the first human-computer interaction interface for at least one preset selection point requirement, a target selection point requirement is determined from the preset selection point requirements. For example, a human-computer interaction interface can be displayed on the screen, and the interface can contain multiple controls. Each control corresponds one-to-one with a preset selection point requirement supported by the measurement point prediction model. The user can use an input device to click one or more controls, and the preset selection point requirement corresponding to the clicked control is the target selection point requirement. For example, if the measurement point prediction model supports three preset selection point requirements, the user can select some or all of these three preset selection point requirements as the target selection point requirement through a selection operation.

[0082] Step S1110 inputs the process parameter data of the target wafer into the trained measurement point prediction model to output the location information of multiple target measurement points on the target wafer, including: inputting both the process parameter data and the target point selection requirements into the measurement point prediction model to obtain the location information of multiple target measurement points on the target wafer.

[0083] The process parameter data of the target wafer can serve as the primary reference for a trained measurement point prediction model to determine target measurement points. Target point selection requirements can serve as auxiliary constraints for the same model, meaning that the target measurement points determined by the trained model based on the process parameter data must meet these requirements. For example, the trained measurement point prediction model can fuse process parameter data and target point selection requirements through methods such as embedding encoding or attention mechanisms to determine the location information of multiple target measurement points.

[0084] In the above technical solution, the training data also includes preset point selection requirements. These preset point selection requirements include one or more of the following: measurement point quantity requirements, measurement point distribution requirements, compensation effect requirements, and measurement capacity requirements. This provides the user with a first human-machine interface. In response to the user's selection operation using the first human-machine interface for at least one preset point selection requirement, the target point selection requirements are determined from the preset requirements. Both process parameter data and the target point selection requirements are input into the measurement point prediction model to obtain the location information of multiple target measurement points on the target wafer. Therefore, the determined target measurement points can meet different preset requirements, allowing the user to flexibly determine the target point selection requirements according to their needs, thus improving the flexibility and applicability of determining target measurement points.

[0085] For example, the training data also includes the location information of historical measurement points on the training wafer and the overlay error measurement data corresponding to the historical measurement points.

[0086] The location information of historical measurement points can include the positions of the measurement points within the training wafer during the actual measurement process; these measurement points are also called historical measurement points. The overlay error measurement data corresponding to the historical measurement points includes the overlay error data obtained after actual measurements were performed on the historical measurement points. Training data can consist of process parameter data for each wafer in multiple training wafers, the location information of historical measurement points, and the overlay error measurement data corresponding to the historical measurement points. For example, the position and priority of each measurement point in the historical measurement points can be marked to obtain training data. The measurement point prediction model can use an attention mechanism to learn the contribution of measurement points in different regions of the training wafer to the overall error distribution, ultimately outputting predicted measurement points. The model loss value can be calculated using the location information of the output predicted measurement points and the overlay error measurement data to ensure that the predicted measurement point positions are accurate and effectively represent the error distribution.

[0087] In the above technical solution, the training data also includes the location information of historical measurement points on the training wafer and the corresponding overlay error measurement data. Therefore, the measurement point prediction model also learns the location information of historical measurement points and the corresponding overlay error measurement data. The output measurement points can be better allocated to areas with historically large errors, avoiding wasting measurement resources in low-error areas. Furthermore, it can better identify common error patterns across batches and devices, and better adapt to new processes and designs.

[0088] For example, the training data also includes measurement label types for historical measurement points, and the measurement point prediction model also outputs the measurement label types for multiple target measurement points. The measurement label types of the target measurement points are used to determine the measurement method for those target measurement points.

[0089] The measurement marker type determines the measurement method used when measuring the corresponding measurement point. For example, measurement marker types can include box markers, periodic grating markers, and cross grating markers. When training the measurement point prediction model, the training data also includes the measurement marker types of historical measurement points. Therefore, the measurement point prediction model can output multiple target measurement points, as well as the measurement marker type corresponding to each target measurement point. The measurement method for each target measurement point can be determined based on its measurement marker type.

[0090] Step S1200, determining the overlay compensation amount based on the position information of multiple target measurement points, includes: using the determined measurement method and based on the position information of multiple target measurement points, determining the overlay compensation amount.

[0091] The measurement method for each target measurement point can be determined based on its measurement mark type. Furthermore, based on its location information, its position within the target wafer can be determined. This allows for the application of the corresponding measurement method to measure the target measurement point and obtain overlay error measurement data. For example, the measurement point prediction model outputs three target measurement points based on the process parameters of the target wafer. Target measurement point 1 and target measurement point 2 correspond to measurement mark type 1, and target measurement point 3 corresponds to measurement mark type 2. Based on the location information and measurement mark type of measurement point 1, the corresponding measurement method can be applied to measurement point 1. Similarly, based on the location information and measurement mark type of measurement points 2 and 3, the corresponding measurement methods can be applied to measurement points 2 and 3 respectively, yielding overlay error measurement data for each measurement point.

[0092] Based on the overlay error measurement data for each target measurement point, the overlay compensation amount is calculated using an overlay error compensation model. The overlay compensation amount can include lateral displacement, longitudinal displacement, rotation angle, scaling, etc. The overlay error compensation model can be any existing or future overlay error compensation model. For example, the overlay error compensation model can be a high-order polynomial fitting model, which can fit the overlay error distribution based on the overlay error measurement data for each target measurement point, and then calculate the overlay compensation amount to compensate for the overlay error in the next overlay operation.

[0093] In the above technical solution, the training data also includes the measurement mark types of historical measurement points. The measurement point prediction model also outputs the measurement mark types of each of the multiple target measurement points. The measurement mark types of the target measurement points are used to determine the measurement method for that target measurement point. Using the determined measurement method and based on the position information of multiple target measurement points, the overlay compensation amount is determined. Therefore, the method for measuring target measurement points can be determined intelligently and automatically, avoiding manual operation and improving efficiency.

[0094] By way of example, according to another aspect of the present invention, a wafer overlay apparatus is also provided. Figure 3 A schematic block diagram of a wafer overlay apparatus 300 according to an embodiment of the present invention is shown. The wafer overlay apparatus 300 includes a first determining module 310, a second determining module 320, an overlay module 330, and an adjustment module 340.

[0095] The first determining module 310 is used to determine the position information of multiple target measurement points on the target wafer using a trained measurement point prediction model. The second determining module 320 is used to determine the overlay compensation amount based on the position information of the multiple target measurement points. The overlay module 330 is used to perform an overlay operation on the target wafer based on the overlay compensation amount. The adjusting module 340 is used to adjust the parameters of the measurement point prediction model based on the position information of the multiple target measurement points and the difference between the first compensation residual and the second compensation residual for subsequent overlay operations. The first compensation residual is calculated based on the current measurement point, and the second compensation residual is calculated based on the remaining measurement points obtained after deleting the measurement points to be measured from the current measurement point. The initial current measurement point is the multiple target measurement points.

[0096] By way of example, according to another aspect of the present invention, an electronic device is also provided. Figure 4 A schematic block diagram of an electronic device 400 according to an embodiment of the present invention is shown. The electronic device 400 includes a processor 410 and a memory 420. The memory 420 stores computer program instructions, which, when executed by the processor 410, are used to perform the wafer overlay method as described above.

[0097] By way of example, according to another aspect of the present invention, a storage medium is also provided, on which program instructions are stored, which, when executed, are used to perform the wafer overlay method as described above. The storage medium may, for example, include an erasable programmable read-only memory (EPROM), a portable read-only memory (CD-ROM), a USB memory, or any combination of the above storage media. The storage medium may be any combination of one or more computer-readable storage media.

[0098] By way of example, according to another aspect of the present invention, a computer program product is also provided, including computer program instructions, which, when executed, are used to perform the wafer overlay method as described above.

[0099] Those skilled in the art can understand the specific implementation schemes and beneficial effects of the above-described wafer overlay method apparatus, electronic equipment, storage medium, and computer program product by reading the relevant descriptions of the wafer overlay method. For the sake of brevity, they will not be described in detail here.

[0100] Although exemplary embodiments have been described herein with reference to the accompanying drawings, it should be understood that the above exemplary embodiments are merely illustrative and are not intended to limit the scope of this application. Various changes and modifications can be made therein by those skilled in the art without departing from the scope and spirit of this application. All such changes and modifications are intended to be included within the scope of this application as claimed in the appended claims.

[0101] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can 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.

[0102] In the several embodiments provided in this application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another device, or some features may be ignored or not executed.

[0103] Numerous specific details are set forth in the specification provided herein. However, it will be understood that embodiments of this application may be practiced without these specific details. In some instances, well-known methods, structures, and techniques have not been shown in detail so as not to obscure the understanding of this specification.

[0104] Similarly, it should be understood that, in order to streamline this application and aid in understanding one or more of the various inventive aspects, features of this application may sometimes be grouped together in a single embodiment, figure, or description thereof in the description of exemplary embodiments of this application. However, this approach should not be construed as reflecting an intention that the claimed application requires more features than are expressly recited in each claim. Rather, as reflected in the corresponding claims, its inventive point lies in solving the corresponding technical problem with features fewer than all features of a single disclosed embodiment. Therefore, the claims following the detailed description are hereby expressly incorporated into that detailed description, wherein each claim itself is a separate embodiment of this application.

[0105] Those skilled in the art will understand that, apart from the mutual exclusion of features, all features disclosed in this specification (including the accompanying claims, abstract, and drawings) and all processes or units of any method or apparatus so disclosed can be combined in any combination. Unless otherwise expressly stated, each feature disclosed in this specification (including the accompanying claims, abstract, and drawings) may be replaced by an alternative feature that serves the same, equivalent, or similar purpose.

[0106] Furthermore, those skilled in the art will understand that although some embodiments described herein include certain features but not others included in other embodiments, combinations of features from different embodiments are intended to be within the scope of this application and form different embodiments. For example, in the claims, any one of the claimed embodiments can be used in any combination.

[0107] The various component embodiments of this application can be implemented in hardware, or as software modules running on one or more processors, or a combination thereof. Those skilled in the art will understand that microprocessors or digital signal processors (DSPs) can be used in practice to implement some or all of the functions of some modules in the wafer overlay apparatus according to embodiments of this application. This application can also be implemented as an apparatus program (e.g., a computer program and computer program product) for performing part or all of the methods described herein. Such an implementation of this application can be stored on a computer-readable medium, or can be in the form of one or more signals. Such signals can be downloaded from an Internet website, provided on a carrier signal, or provided in any other form.

[0108] It should be noted that the above embodiments are illustrative of this application and not restrictive, and that those skilled in the art can devise alternative embodiments without departing from the scope of the appended claims. In the claims, any reference signs placed between parentheses should not be construed as limiting the claims. The word "comprising" does not exclude the presence of elements or steps not listed in the claims. The word "a" or "an" preceding an element does not exclude the presence of a plurality of such elements. This application can be implemented by means of hardware comprising several different elements and by means of a suitably programmed computer. In the unit claims enumerating several means, several of these means may be embodied by the same item of hardware. The use of the words first, second, and third, etc., does not indicate any order. These words can be interpreted as names.

[0109] The above description is merely a specific embodiment or illustration of the embodiments of this application. The scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. The scope of protection of this application shall be determined by the scope of the claims.

Claims

1. A wafer overlay method, characterized in that, include: The location information of multiple target measurement points on the target wafer is determined by using a trained measurement point prediction model. Based on the position information of the multiple target measurement points, the overlay compensation amount is determined; The overlay operation of the target wafer is performed based on the overlay compensation amount; Based on the location information of the multiple target measurement points and the difference between the first compensation residual and the second compensation residual, the parameters of the measurement point prediction model are adjusted for subsequent overlay operations. The first compensation residual is calculated based on the current measurement point, and the second compensation residual is calculated based on the remaining measurement points obtained after deleting the measurement points to be measured from the current measurement point. The initial current measurement point is the multiple target measurement points.

2. The wafer overlay method according to claim 1, characterized in that, The process of adjusting the parameters of the measurement point prediction model includes one or more deletion operations until the first preset requirement is met. The parameters of the measurement point prediction model are modified after each update of the current measurement point. The deletion operation includes: Determine the first compensation residual for the current measurement point; Determine the measurement points to be measured from the current measurement points; Determine the second compensation residual for the remaining measurement points in the current measurement point; Based on the difference between the first compensation residual and the second compensation residual, a first residual difference index value is calculated, wherein the larger the difference, the larger the first residual difference index value. If the value of the first residual difference index is less than or equal to the first difference threshold, then at least the measurement point to be measured is deleted from the current measurement point to update the current measurement point; If the value of the first residual difference index is greater than the first difference threshold, then the measurement point to be quantified is determined to be non-deletable.

3. The wafer overlay method according to claim 2, characterized in that, The multiple target measurement points are located in different exposure fields. The one or more deletion operations include: multiple deletion operations in the first removal phase; In the first elimination stage, each time a deletion operation is performed, the determined measurement points to be measured are located in different exposure fields, and if the first residual difference index value is less than or equal to the first difference threshold, then all measurement points in the exposure field corresponding to the current measurement point to be measured are deleted in the current measurement point to update the current measurement point.

4. The wafer overlay method according to claim 3, characterized in that, The one or more deletion operations also include: multiple sets of deletion operations in the second elimination stage; When the second preset requirement is met, the process transitions from the first elimination stage to the second elimination stage. Each group of deletion operations includes at least one deletion operation executed consecutively. The quantitative measurement points determined in at least one deletion operation in each group of deletion operations are located in the same exposure field, and the quantitative measurement points determined in different groups of deletion operations are located in different exposure fields.

5. The wafer overlay method according to claim 2, characterized in that, Among the one or more deletion operations, there are simultaneous deletion operations, wherein the quantitative measurement points determined in the simultaneous deletion operations are different.

6. The wafer overlay method according to claim 1, characterized in that, The measurement point prediction model is obtained by training with training data, which includes: process parameter data of the training wafer. The process of determining the location information of multiple target measurement points on the target wafer using a trained measurement point prediction model includes: The process parameter data of the target wafer is input into the trained measurement point prediction model to output the location information of multiple target measurement points of the target wafer.

7. The wafer overlay method according to claim 6, characterized in that, The training data also includes preset point selection requirements, which may include one or more of the following: requirements for the number of measurement points, requirements for the distribution of measurement points, requirements for compensation effect, and requirements for measurement capacity. The method further includes: Provide users with the primary human-computer interaction interface; In response to a user's selection operation using the first human-computer interaction interface for at least one preset selection point requirement, a target selection point requirement is determined from the preset selection point requirements; The step of inputting the process parameter data of the target wafer into the trained measurement point prediction model to output the location information of multiple target measurement points on the target wafer includes: The process parameter data and the target selection requirements are both input into the measurement point prediction model to obtain the location information of multiple target measurement points on the target wafer.

8. The wafer overlay method according to claim 6, characterized in that, The training data also includes the location information of historical measurement points on the training wafer and the overlay error measurement data corresponding to the historical measurement points.

9. The wafer overlay method according to claim 8, characterized in that, The training data also includes the measurement label type of the historical measurement points. The measurement point prediction model also outputs the measurement label type of each of the multiple target measurement points. The measurement label type of the target measurement point is used to determine the measurement method of the target measurement point. The determination of the overlay compensation amount based on the position information of the multiple target measurement points includes: The overlay compensation amount is determined using the established measurement method and based on the position information of the multiple target measurement points.

10. A wafer overlay apparatus, characterized in that, include: The first determination module is used to determine the location information of multiple target measurement points on the target wafer using a trained measurement point prediction model. The second determining module is used to determine the overlay compensation amount based on the position information of the multiple target measurement points; An overlay module is used to perform an overlay operation on the target wafer based on the overlay compensation amount; An adjustment module is used to adjust the parameters of the measurement point prediction model based on the location information of the multiple target measurement points and the difference between the first compensation residual and the second compensation residual, for subsequent overlay operations. The first compensation residual is calculated based on the current measurement point, and the second compensation residual is calculated based on the remaining measurement points obtained after deleting the measurement points to be measured from the current measurement point. The initial current measurement point is the multiple target measurement points.

11. An electronic device, comprising: Processor and memory, characterized in that, The memory stores computer program instructions, which, when executed by the processor, are used to perform the wafer overlay method as described in any one of claims 1 to 9.

12. A storage medium on which program instructions are stored, characterized in that, The program instructions, when executed, are used to perform the wafer overlay method as described in any one of claims 1 to 9.

13. A computer program product comprising computer program instructions, characterized in that, The computer program instructions, when executed, are used to perform the wafer overlay method as described in any one of claims 1 to 9.

Images