Model training method, device and equipment applied to wafer-level acceptable test

Abstract

The invention provides a model training method, device and equipment applied to a wafer-level acceptable test. The method comprises the following steps: acquiring historical measurement data of a preset point location on a wafer; the historical measurement data comprises historical online measurement values and wafer-level acceptable test data; according to the historical online measurement value, determining online measurement value image data in the historical time, and according to the historical wafer-level acceptable test data, determining image data of the wafer-level acceptable test data in the historical time; the image data of the online measurement values represent the online measurement values of all the position points, and the image data of the wafer-level acceptable test data represent the wafer-level acceptable test data of all the position points; inputting the image data of the online measurement value into the initial model to obtain an output image; outputting prediction data of the wafer-level acceptable test data of all position points on the output image representing the wafer; and if the output image meets the training completion condition, obtaining a prediction model of the wafer-level acceptable test data.

Description

Technical Field

[0001] The present application relates to data processing technology, and in particular to a model training method, device and equipment applied to wafer-level acceptability testing. Background Art

[0002] In semiconductor manufacturing, WAT (Wafer Acceptance Test) is a complete set of electrical characteristics testing processes designed to ensure that wafers have the highest quality and acceptable performance before leaving the manufacturing plant. It is a comprehensive and rigorous testing process that tests each device on the wafer to ensure that their quality and performance meet the expected level.

[0003] Therefore, WAT testing plays a vital role in the entire wafer manufacturing process. The current WAT testing requires a lot of manpower and time, and how to improve the efficiency of WAT testing has become an urgent problem to be solved. Summary of the invention

[0004] The present application provides a model training method, device and equipment for wafer-level acceptability testing to improve the efficiency of WAT testing.

[0005] In a first aspect, the present application provides a model training method for wafer-level acceptability testing, comprising:

[0006] Acquire historical measurement data corresponding to preset points on the wafer; wherein the historical measurement data includes online measurement values ​​and wafer-level acceptable test data obtained within a preset historical time, the online measurement values ​​are used to represent the size information of the wafer obtained in the wafer manufacturing process, and the wafer-level acceptable test data are used to represent the data obtained by performing a wafer-level acceptable test on the wafer; the preset points are used to represent preset position points on the wafer;

[0007] Determine the image data of the online measurement values ​​corresponding to the wafer within the historical time according to the online measurement values ​​in the historical measurement data, and determine the image data of the wafer-level acceptable test data corresponding to the wafer within the historical time according to the wafer-level acceptable test data in the historical measurement data; wherein the image data of the online measurement values ​​is used to represent the online measurement values ​​corresponding to all positions on the wafer, and the image data of the wafer-level acceptable test data is used to represent the wafer-level acceptable test data corresponding to all positions on the wafer;

[0008] Inputting the image data of the online measurement value corresponding to the wafer in the historical time into the pre-built initial model for processing to obtain an output image; wherein the output image is used to represent the predicted data of the wafer-level acceptable test data obtained by performing a wafer-level acceptable test on all positions on the wafer;

[0009] If it is determined that the output image meets the preset training completion condition based on the image data of the wafer-level acceptable test data corresponding to the wafer within the historical time, a prediction model of the wafer-level acceptable test data is obtained; wherein, the prediction model of the wafer-level acceptable test data is used to predict the wafer-level acceptable test data at any position point on the wafer.

[0010] In a second aspect, the present application provides a method for predicting wafer-level acceptable test data, comprising:

[0011] Obtain the current online measurement value of the wafer;

[0012] Input the current online measurement value of the wafer into the prediction model of wafer-level acceptable test data to obtain prediction data of wafer-level acceptable test data corresponding to the current online measurement value of the wafer; wherein the prediction model of wafer-level acceptable test data is a model obtained based on the method described in the first aspect.

[0013] In a third aspect, the present application provides a model training device for wafer-level acceptability testing, comprising:

[0014] A data acquisition module, used to acquire historical measurement data corresponding to preset points on the wafer; wherein the historical measurement data includes online measurement values ​​and wafer-level acceptable test data obtained within a preset historical time, wherein the online measurement values ​​are used to represent the size information of the wafer obtained in the wafer manufacturing process, and the wafer-level acceptable test data are used to represent the data obtained by performing a wafer-level acceptable test on the wafer; the preset points are used to represent preset position points on the wafer;

[0015] An image determination module is used to determine the image data of the online measurement values ​​corresponding to the wafer within the historical time according to the online measurement values ​​in the historical measurement data, and to determine the image data of the wafer-level acceptable test data corresponding to the wafer within the historical time according to the wafer-level acceptable test data in the historical measurement data; wherein the image data of the online measurement values ​​is used to represent the online measurement values ​​corresponding to all positions on the wafer, and the image data of the wafer-level acceptable test data is used to represent the wafer-level acceptable test data corresponding to all positions on the wafer;

[0016] A model training module, used for inputting the image data of the online measurement value corresponding to the wafer in the historical time into a pre-built initial model for processing to obtain an output image; wherein the output image is used to represent the predicted data of the wafer-level acceptable test data obtained by performing a wafer-level acceptable test on all positions on the wafer;

[0017] A training completion module is used to obtain a prediction model for wafer-level acceptable test data if it is determined that the output image meets a preset training completion condition based on the image data of the wafer-level acceptable test data corresponding to the wafer within the historical time; wherein the prediction model for wafer-level acceptable test data is used to predict the wafer-level acceptable test data for any position point on the wafer.

[0018] In a fourth aspect, the present application provides a device for predicting wafer-level acceptable test data, comprising:

[0019] A measurement value acquisition module, used to acquire the current online measurement value of the wafer; wherein the online measurement value is used to represent the size information of the wafer obtained in the wafer manufacturing process;

[0020] A data prediction module is used to input the current online measurement value of the wafer into a prediction model of wafer-level acceptable test data to obtain prediction data of wafer-level acceptable test data corresponding to the current online measurement value of the wafer; wherein the prediction model of wafer-level acceptable test data is a model obtained based on the device described in the third aspect.

[0021] In a fifth aspect, the present application provides an electronic device, comprising: a processor, and a memory communicatively connected to the processor;

[0022] The memory stores computer-executable instructions;

[0023] The processor executes the computer-executable instructions stored in the memory to implement the model training method applied to wafer-level acceptable test as described in the first aspect or the method for predicting wafer-level acceptable test data as described in the second aspect.

[0024] In a sixth aspect, the present application provides a computer-readable storage medium, which stores computer-executable instructions. When the computer-executable instructions are executed by a processor, they are used to implement the model training method for wafer-level acceptable testing as described in the first aspect or the method for predicting wafer-level acceptable test data as described in the second aspect.

[0025] In a seventh aspect, the present application provides a computer program product, comprising a computer program, which, when executed by a processor, implements the model training method for wafer-level acceptable testing described in the first aspect or the method for predicting wafer-level acceptable test data described in the second aspect.

[0026] The present application provides a model training method, device and equipment for wafer-level acceptable test, which obtains the online measurement values ​​and wafer-level acceptable test data corresponding to the preset points on the wafer in the historical time, and determines the online measurement values ​​and wafer-level acceptable test data of all points on the wafer in the historical time, thereby obtaining the image data of the online measurement values ​​and the image data of the wafer-level acceptable test data. The image data of the online measurement values ​​corresponding to the wafer in the historical time is input into the pre-built initial model to obtain the output image, and the output image can represent the predicted data of the wafer-level acceptable test data of all positions on the wafer. According to the image data of the wafer-level acceptable test data corresponding to the wafer in the historical time, it is determined whether the output image meets the preset training completion conditions. If so, it is determined to obtain the prediction model of the wafer-level acceptable test data, and the prediction model of the wafer-level acceptable test data can predict the wafer-level acceptable test data of any position on the wafer. By expanding the online measurement values ​​and wafer-level acceptable test data of preset points to the online measurement values ​​and wafer-level acceptable test data of all points, the amount of training data can be increased and the training accuracy of the model can be improved. The prediction model can be used to predict the wafer-level acceptable test data before the wafer is manufactured, avoiding abnormalities found in the test after the wafer has completed the process manufacturing, which causes the foundry to lose a lot of manufacturing costs and time costs, and improves the efficiency of wafer-level acceptable testing. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0028] Figure 1 A flow chart of a model training method for wafer-level acceptability testing provided in an embodiment of the present application;

[0029] Figure 2 A schematic diagram of preset points on a wafer provided in an embodiment of the present application;

[0030] Figure 3 A schematic diagram of image data of online measurement values ​​provided in an embodiment of the present application;

[0031] Figure 4 A flow chart of a model training method for wafer-level acceptability testing provided in an embodiment of the present application;

[0032] Figure 5 A schematic diagram of an initial image of historical online measurement values ​​provided in an embodiment of the present application;

[0033] Figure 6 A schematic flow chart of a method for predicting acceptable test data at the wafer level provided in an embodiment of the present application;

[0034] Figure 7 A structural block diagram of a model training device for wafer-level acceptability testing provided in an embodiment of the present application;

[0035] Figure 8 A structural block diagram of a model training device for wafer-level acceptability testing provided in an embodiment of the present application;

[0036] Fig. 9 A structural block diagram of a device for predicting acceptable test data at the wafer level provided in an embodiment of the present application;

[0037] Fig.10 A structural block diagram of an electronic device provided in an embodiment of the present application;

[0038] Fig.11 A structural block diagram of an electronic device provided in an embodiment of the present application.

[0039] The above drawings have shown clear embodiments of the present application, which will be described in more detail later. These drawings and text descriptions are not intended to limit the scope of the present application in any way, but to illustrate the concept of the present application to those skilled in the art by referring to specific embodiments. DETAILED DESCRIPTION

[0040] In order to make the objectives, technical solutions and advantages of the present application more clear, the embodiments of the present application will be further described in detail below with reference to the accompanying drawings.

[0041] It should be clear that the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in the field without creative work are within the scope of protection of the present application.

[0042] When the following description refers to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present application. Instead, they are only examples of devices and methods consistent with some aspects of the present application as detailed in the attached claims.

[0043] In the description of the present application, it should be understood that the terms "first", "second", "third", etc. are only used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence, nor can they be understood as indicating or implying relative importance. For those of ordinary skill in the art, the specific meanings of the above terms in the present application can be understood according to the specific circumstances. In addition, in the description of the present application, unless otherwise specified, "multiple" refers to two or more. "And / or" describes the association relationship of associated objects, indicating that three relationships may exist. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. The character " / " generally indicates that the objects associated before and after are in an "or" relationship.

[0044] It should be noted that due to space limitations, this application specification does not list all optional implementation methods. After reading this application specification, those skilled in the art should be able to think that as long as the technical features do not contradict each other, any combination of technical features can constitute an optional implementation method. Each embodiment is described in detail below.

[0045] In semiconductor manufacturing, WAT is a complete set of electrical characteristics test processes designed to ensure that wafers have the highest quality and acceptable performance before leaving the manufacturing plant. WAT testing is one of the last few checkpoints in wafer manufacturing, used to verify whether the electrical parameters of semiconductor devices meet the specification requirements. It is a comprehensive and rigorous testing process that tests each device on the wafer to ensure that their quality and performance meet the expected level.

[0046] WAT test data can monitor whether the wafer manufacturing process is abnormal, determine whether the product meets the electrical specifications of the process technology platform, and correlate WAT data with yield, check the most correlated parameters, identify problematic process steps, and then test and analyze specific WAT test parameters to improve the process or develop the next generation of process technology platforms.

[0047] Traditional WAT testing plays a vital role in the entire wafer manufacturing process, but it still has some shortcomings and challenges. WAT testing usually requires a lot of time and resources to test and analyze multiple devices on each wafer one by one. The test of key measurement parameters is between BEOL (Back End of Line) and OQA (Outgoing Quality Assure). At this time, the wafer has completed the process manufacturing. If the WAT test is abnormal, it will cause the foundry to lose a lot of manufacturing costs and time costs, and the test efficiency is low.

[0048] The present application provides a model training method, device and equipment for wafer-level acceptability testing, which aims to solve the above technical problems in the prior art.

[0049] The technical solution of the present application and how the technical solution of the present application solves the above-mentioned technical problems are described in detail below with specific embodiments. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be repeated in some embodiments. The embodiments of the present application will be described below in conjunction with the accompanying drawings.

[0050] Figure 1 1 is a flow chart of a model training method for wafer-level acceptability testing provided in accordance with an embodiment of the present application. The method can be executed by a model training device for wafer-level acceptability testing. Figure 1 As shown, the method comprises the following steps:

[0051] S101. Obtain historical measurement data corresponding to preset points on the wafer; wherein the historical measurement data includes online measurement values ​​and wafer-level acceptable test data obtained within a preset historical time, the online measurement values ​​are used to represent the size information of the wafer obtained in the wafer manufacturing process, and the wafer-level acceptable test data are used to represent the data obtained by performing a wafer-level acceptable test on the wafer; the preset points are used to represent preset position points on the wafer.

[0052] Exemplarily, a wafer is a circular silicon chip, and a plurality of position points can be preset on the plane of the wafer as preset points. Figure 2 Schematic diagram of preset points on the wafer. Figure 2 There are five preset points, which are located at the top, bottom, left, right and center of the wafer respectively. The measurement data of the preset points within the preset historical time is obtained as the historical measurement data. For example, the measurement data of the preset points one month ago can be obtained. The measurement data may include Inline data and WAT data, that is, the historical measurement data may include Inline data and WAT data obtained within the preset historical time, and Inline data may refer to online measurement values.

[0053] Online measurement values ​​can be used to represent the size information of the wafer obtained in the wafer production process. The wafer production process may include multiple process passes, and each process pass can obtain Inline data at different positions. For example, Inline data can include data categories such as THK (thickness), OVL (deviation), CD (critical dimension) and OCD (optical critical dimension). WAT data can be used to represent the data obtained from wafer-level acceptability testing of wafers. WAT data can be the electrical parameters of the wafer.

[0054] Obtaining historical measurement data corresponding to preset points on a wafer may be to obtain the Inline data and WAT data at the preset points of a wafer that has undergone WAT. The historical measurement data of the preset points under multiple process passes may be obtained. For example, if there are three process passes, for a wafer that has been manufactured, the Inline data and WAT data of each preset point of the wafer under the three process passes may be obtained respectively.

[0055] S102. Determine the image data of the online measurement values ​​corresponding to the wafer within the historical time based on the online measurement values ​​in the historical measurement data, and determine the image data of the wafer-level acceptable test data corresponding to the wafer within the historical time based on the wafer-level acceptable test data in the historical measurement data; wherein the image data of the online measurement values ​​is used to represent the online measurement values ​​corresponding to all position points on the wafer, and the image data of the wafer-level acceptable test data is used to represent the wafer-level acceptable test data corresponding to all position points on the wafer.

[0056] Exemplarily, the historical measurement data obtained is the historical measurement data of some points on the wafer, rather than the historical measurement data of the entire wafer. The historical measurement data of all positions on the wafer can be determined based on the historical measurement data of some points. This avoids storing the historical measurement data of all positions on the wafer, effectively saving storage space.

[0057] The Inline data of the preset points in the historical measurement data can be determined, and the Inline data corresponding to all the positions in the wafer in the historical time can be calculated based on the Inline data of the preset points. For example, the average value of the Inline data of the preset points can be calculated, and the average value can be determined as the Inline data of the positions other than the preset points. Based on the Inline data of all the positions in the wafer, the image data of the Inline data is obtained. The image data of the Inline data can be used to represent the Inline data corresponding to all the positions on the wafer. For example, color can be used on the image data to represent the size of the Inline data. Figure 3 Schematic diagram of image data of online measurement values. Figure 3 A square in the image is a position point on the wafer, and all the squares form a complete wafer, that is, a circle. The pixel value of each square represents the size of the Inline data at the position point. Different image data can be generated for Inline data of different data categories. For example, image data of THK, image data of OVL, image data of CD, and image data of OCD can be generated.

[0058] The WAT data of the preset points in the historical measurement data can be determined, and the WAT data corresponding to all the positions in the wafer in the historical time can be calculated based on the determined WAT data of the preset points. For example, the average value of the WAT data of the preset points can be calculated, and the average value is determined as the WAT data of the positions other than the preset points. Based on the WAT data of all the positions in the wafer, the image data of the WAT data is obtained. The image data of the WAT data can be used to represent the WAT data corresponding to all the positions on the wafer. For example, color can be used on the image data to represent the size of the WAT data.

[0059] S103. Input the image data of the online measurement values ​​corresponding to the wafer in the historical time into the pre-built initial model for processing to obtain an output image; wherein the output image is used to represent the predicted data of the wafer-level acceptable test data obtained by performing wafer-level acceptable test on all positions on the wafer.

[0060] Exemplarily, an initial model for predicting WAT data is pre-built. In this embodiment, an initial neural network model can be constructed for training. The image data of the Inline data corresponding to the Wafer determined in S102 in the historical time is input as input data into the pre-built initial model. If image data of different data categories are generated, the image data of different data categories can be used as input data and input into the model. The initial model is provided with network layers such as convolutional layers, pooling layers, and fully connected layers, and the input data can be processed to obtain the output data corresponding to the input data, that is, to obtain the output image corresponding to the image data of the Inline data. The trained model is used to predict the WAT data of the wafer. During training, the input data represents the Inline data of all positions on the wafer, and the output image can be used to represent the predicted data of the WAT data obtained by performing WAT on all positions on the wafer.

[0061] In this embodiment, the image data of the online measurement values ​​corresponding to the wafer in the historical time is input into the pre-built initial model for processing to obtain an output image, including: inputting the image data of the online measurement values ​​corresponding to the wafer in the historical time into the pre-built initial model; wherein the pre-built initial model is a U-shaped network structure; according to the deconvolution network layer preset in the initial model, the image data of the online measurement values ​​corresponding to the wafer in the historical time is size-expanded to obtain the image data after the size-expanded processing; according to the convolution network layer preset in the initial model, the image data after the size-expanded processing is feature extracted to obtain the output image.

[0062] Specifically, a suitable convolutional neural network model is designed as an initial model for training to predict WAT data. The network structure may adopt a UNet (U-type network) structure, and a deconvolution network layer may be added to the network structure to amplify the image size of the input data.

[0063] The image data of the Inline data corresponding to the Wafer in the historical time is input into the pre-built initial model. According to the preset deconvolution network layer in the initial model, the image data of the Inline data corresponding to the Wafer in the historical time is scaled up to obtain the image data after the scaled up processing. For example, the image size is scaled up by the deconvolution component, and the image of the input data is scaled up by 4 times from 9×11, that is, to 36×44, to avoid the image being too small after multiple convolutions. After passing through the network layers such as the convolution layer, the image data after the scaled up processing is subjected to feature extraction and calculation, and finally the output image of the WAT is output.

[0064] The beneficial effect of this setting is that by designing a neural network model, WAT data can be automatically predicted before wafer manufacturing is completed, improving the efficiency of WAT testing. The deconvolution layer in the model can provide the neural network with more fine-grained high-dimensional features, improving the accuracy of WAT parameter prediction.

[0065] S104. If it is determined that the output image meets the preset training completion condition based on the image data of the wafer-level acceptable test data corresponding to the wafer in the historical time, a prediction model of the wafer-level acceptable test data is obtained; wherein the prediction model of the wafer-level acceptable test data is used to predict the wafer-level acceptable test data of any position point on the wafer.

[0066] Exemplarily, the training completion condition is pre-set, and the initial model is iteratively trained multiple times. After each output image is obtained, it is necessary to determine whether the output image meets the preset training completion condition. If so, it is determined that the model training is completed and a prediction model of wafer-level acceptable test data is obtained; if not, it is determined that the model training is not completed and training needs to be continued. The prediction model of wafer-level acceptable test data after training can be used to predict wafer-level acceptable test data at any position on the wafer.

[0067] A loss function can be designed in advance, which can be used to calculate the difference between the output image and the image data of the wafer-level acceptable test data corresponding to the wafer in the historical time, and determine whether the calculated difference meets the preset training completion condition. If not, the parameters of the network model are updated iteratively through back propagation and the optimizer, and finally the training of the prediction model is completed.

[0068] The embodiment of the present application provides a model training method for wafer-level acceptable test, which obtains the online measurement values ​​and wafer-level acceptable test data corresponding to the preset points on the wafer in the historical time, determines the online measurement values ​​and wafer-level acceptable test data of all points on the wafer in the historical time, and thus obtains the image data of the online measurement values ​​and the image data of the wafer-level acceptable test data. The image data of the online measurement values ​​corresponding to the wafer in the historical time is input into the pre-built initial model, and the output image is output, and the output image can represent the predicted data of the wafer-level acceptable test data of all positions on the wafer. According to the image data of the wafer-level acceptable test data corresponding to the wafer in the historical time, it is determined whether the output image meets the preset training completion conditions. If so, it is determined to obtain the prediction model of the wafer-level acceptable test data, and the prediction model of the wafer-level acceptable test data can predict the wafer-level acceptable test data of any position on the wafer. By expanding the online measurement values ​​and wafer-level acceptable test data of preset points to the online measurement values ​​and wafer-level acceptable test data of all points, the amount of training data can be increased and the training accuracy of the model can be improved. The prediction model can be used to predict the wafer-level acceptable test data before the wafer is manufactured, avoiding the loss of a large amount of manufacturing cost and time cost for the foundry due to abnormalities found in the test after the wafer has completed the process manufacturing, and improving the efficiency of wafer-level acceptable testing.

[0069] Figure 4 A flow chart of a model training method for wafer-level acceptability testing provided in an embodiment of the present application is provided, and this embodiment is an optional embodiment based on the above-mentioned embodiment.

[0070] In this embodiment, obtaining historical measurement data corresponding to preset points on the wafer can be refined as follows: obtaining online measurement values ​​corresponding to a first preset point on the wafer obtained within a preset historical time, which are historical online measurement values, and obtaining wafer-level acceptable test data corresponding to a second preset point on the wafer obtained within a preset historical time, which are historical wafer-level acceptable test data; wherein the first point includes at least one position point, and the second point includes at least one position point; preprocessing the historical online measurement values ​​and the historical wafer-level acceptable test data, determining target points in the first point and the second point, and obtaining historical online measurement values ​​and historical wafer-level acceptable test data corresponding to the target points.

[0071] like Figure 4 As shown, the method comprises the following steps:

[0072] S401. Obtain an online measurement value corresponding to a first point position preset on the wafer obtained within a preset historical time, which is a historical online measurement value, and obtain wafer-level acceptable test data corresponding to a second point position preset on the wafer obtained within a preset historical time, which is a historical wafer-level acceptable test data; wherein the first point position includes at least one position point, and the second point position includes at least one position point.

[0073] Exemplarily, multiple positions may be preset as the first position. The first position is a position for obtaining Inline data, that is, the online measurement value corresponding to the preset first position on the wafer is obtained from the Inline data obtained in the historical time as the historical online measurement value.

[0074] Multiple position points can also be preset as the second point. The position point of the second point can be completely consistent with the position point of the first point, or there may be inconsistencies. For example, the first point has 13 position points, and the second point has 21 position points, among which 9 position points of the first point and the second point are located in the same position. The second point is a position point for obtaining WAT data, that is, from the WAT data obtained in the historical time, the WAT data corresponding to the preset second point on the wafer is obtained as the historical wafer-level acceptable test data. During the manufacturing process of the wafer, the Inline data of each first point and the WAT data of the second point can be stored in real time, which is convenient for the acquisition of historical measurement data. And there is no need to store the Inline data and WAT data of each position point on the wafer, which effectively saves storage space.

[0075] In this embodiment, the process of manufacturing wafers includes at least two process passes; after obtaining the online measurement value corresponding to the preset first point on the wafer obtained within the preset historical time as the historical online measurement value, it also includes: determining the historical online measurement value corresponding to each process pass obtained within the preset historical time; determining the correlation between the historical online measurement values ​​corresponding to two process passes; if the correlation meets the preset elimination condition, retaining the historical online measurement value corresponding to one process pass from the historical online measurement values ​​corresponding to the two process passes.

[0076] Specifically, the wafer manufacturing process may include multiple process passes, and the finished wafer can only be obtained after multiple process passes. In each process pass, the inline data of the wafer may change. Therefore, when obtaining historical online measurement values, historical online measurement values ​​under multiple process passes can be obtained. That is, the online measurement values ​​of each first point under different process passes within the preset historical time are obtained.

[0077] From the historical online measurement values ​​obtained, determine the historical online measurement values ​​corresponding to each process pass. For example, if there are three process passes, the historical online measurement values ​​of each first point under these three process passes can be obtained. Determine the correlation between the historical online measurement values ​​corresponding to two process passes. The correlation between the historical online measurement values ​​under each two process passes can be determined, or one process pass can be determined as the main link and the other process passes as the secondary links, and the historical online measurement values ​​of the main link can be compared with the historical online measurement values ​​of each secondary link to determine the correlation between the main link and each secondary link. For example, the three process passes can be compared in pairs to determine the correlation between the historical online measurement values ​​of the first process pass and the second process pass, the correlation between the historical online measurement values ​​of the first process pass and the third process pass, and the correlation between the historical online measurement values ​​of the second process pass and the third process pass. For another example, take the first process pass as the main link, determine the correlation between the historical online measurement values ​​of the first process pass and the second process pass, and determine the correlation between the historical online measurement values ​​of the first process pass and the third process pass.

[0078] A calculation formula for correlation can be preset, and the correlation can represent the similarity between historical online measurement values. For example, the average value of the historical online measurement value of the first point under each process pass can be determined, and the difference between the average values ​​corresponding to the two process passes can be determined. The smaller the difference, the greater the correlation between the historical online measurement values ​​of the two process passes. The online measurement values ​​can include multiple data categories, for example, CD, OCD, THK, and OVL. When determining the correlation, the correlation calculation can be performed separately for the online measurement values ​​of different data categories. For example, the correlation is determined for the CD data under the first process pass and the second process pass, and the correlation is determined for the OCD data under the first process pass and the second process pass.

[0079] A rejection condition is set in advance, and the rejection condition can be used to determine whether to reject the historical online measurement values ​​under the process pass. Determine whether the correlation between the historical online measurement values ​​corresponding to the two process passes meets the preset rejection condition. For example, the preset correlation threshold is 0.9, and determine whether the calculated correlation is greater than 0.9. If so, the historical online measurement value corresponding to one of the two process passes is retained, and the historical online measurement value corresponding to the other process pass is eliminated. If not, the historical online measurement values ​​under the two process passes are retained. When eliminating, a process pass can be randomly selected from the two process passes for data elimination. It is also possible to preset rejection rules. For example, the process pass with a later process flow can be selected from the two process passes for data elimination, or the amount of data under the two process passes can be determined, and the process pass with a smaller amount of data can be eliminated.

[0080] The beneficial effect of this setting is that it traverses the data in different process passes, eliminates data, reduces the amount of data for subsequent calculations, achieves data dimensionality reduction, and improves the efficiency of model training.

[0081] In this embodiment, after obtaining the online measurement value corresponding to the preset first point on the wafer obtained within the preset historical time, it also includes: if the number of first points is not a preset point number threshold, the online measurement value corresponding to the preset first point on the wafer is deleted.

[0082] Specifically, the number of first points and the number of second points may be preset by the foundry of the wafer, and the number of first points set by different foundries may be different. A threshold value for the number of first points can be preset to determine whether the number of first points of the acquired wafer is the preset threshold value for the number of points. If so, the subsequent data processing continues; if not, the historical measurement data of the wafer can be deleted, that is, the data of the wafer is not used for training. For example, the preset threshold value for the number of points is 13, and the Inline data whose total number of first points is not 13 is removed.

[0083] The beneficial effect of this setting is that historical measurement data can be processed uniformly to avoid the different data volumes of different wafers affecting training efficiency and accuracy.

[0084] In the present embodiment, the wafer-level acceptable test data corresponding to the preset second point on the wafer obtained within the preset historical time is obtained as the historical wafer-level acceptable test data, including: obtaining all the wafer-level acceptable test data corresponding to the preset second point on the wafer obtained within the preset historical time; selecting the wafer-level acceptable test data of the preset data category from all the wafer-level acceptable test data as the historical wafer-level acceptable test data.

[0085] Specifically, in addition to screening the historical online measurement values ​​of the first point, the historical wafer-level acceptable test data of the second point can also be screened. There can also be different data categories in the wafer-level acceptable test data, and one or more data categories are pre-set as the data categories to be retained. Obtain all the wafer-level acceptable test data corresponding to the second point on the wafer obtained within a preset historical time, and select the wafer-level acceptable test data of the preset data category from all the wafer-level acceptable test data as the retained historical wafer-level acceptable test data. That is, eliminate information irrelevant to WAT prediction from all WAT data.

[0086] The beneficial effect of this setting is that it retains important WAT information, avoids the impact of information redundancy on model training, and improves the efficiency and accuracy of model training.

[0087] S402, pre-processing the historical online measurement values ​​and the historical wafer-level acceptable test data, determining the target point among the first point and the second point, and obtaining the historical online measurement values ​​and the historical wafer-level acceptable test data corresponding to the target point.

[0088] Exemplarily, the number and position of the first point and the second point may be different. Therefore, after obtaining the historical online measurement value of the first point and the historical wafer-level acceptable test data of the second point, the historical online measurement value and the historical wafer-level acceptable test data can be preprocessed. The preprocessing can be to align the historical online measurement value and the historical wafer-level acceptable test data. One or more target points are determined from the first point and the second point. The target point in the first point is consistent with the target point in the second point, that is, the point in the first point that repeats with the second point can be determined as the target point. Determine the historical online measurement value corresponding to the target point and the historical wafer-level acceptable test data corresponding to the target point.

[0089] In this embodiment, historical online measurement values ​​and historical wafer-level acceptable test data are preprocessed to determine a target point in the first point and the second point, including: adding the first point to a preset spatial coordinate system to obtain an initial image of the historical online measurement values; wherein, in the initial image of the historical online measurement values, the coordinate point of the first point is used to indicate the position of the first point in the wafer, and the pixel value of the first point is used to indicate the historical online measurement value of the first point; adding the second point to the preset spatial coordinate system to obtain an initial image of the historical wafer-level acceptable test data; wherein, in the initial image of the historical wafer-level acceptable test data, the coordinate point of the second point is used to indicate the position of the second point in the wafer, and the pixel value of the second point is used to indicate the historical wafer-level acceptable test data of the second point; overlapping the initial image of the historical online measurement values ​​with the initial image of the historical wafer-level acceptable test data, and determining the points that overlap in the initial image of the historical online measurement values ​​and the initial image of the historical wafer-level acceptable test data as the target points.

[0090] Specifically, according to the historical online measurement value of the first point, data in the form of an image of the historical online measurement value is obtained as an initial image. For example, the first point can be added to the preset spatial coordinates. The initial image can indicate the position of the first point in the wafer and the size of the historical online measurement value at the first point. That is, the coordinate point of the first point can be used to indicate the position of the first point in the wafer, and the size of the historical online measurement value can be indicated by the size of the pixel value. For different data categories in the historical online measurement values, an initial image corresponding to each data category can be generated. That is, for the CD data of a certain process pass, an initial image can be generated; for the OCD data of a certain process pass, an initial image can also be generated. Figure 5 This is a schematic diagram of the initial image of historical online measurement values. Figure 5 There are 21 first points in the image, which are distributed on the wafer. The pixel value of each square can represent the historical online measurement value of the location.

[0091] According to the historical wafer-level acceptable test data of the second point, data in the form of an image of the historical wafer-level acceptable test data is obtained as an initial image. For example, the second point can be added to the preset spatial coordinates. The initial image can indicate the position of the second point in the wafer and the size of the historical wafer-level acceptable test data of the second point. That is, the coordinate point of the second point can be used to indicate the position of the second point in the wafer, and the size of the historical wafer-level acceptable test data can be indicated by the pixel size. For different data categories in the historical wafer-level acceptable test data, an initial image corresponding to each data category can be generated.

[0092] After obtaining the initial image of the historical online measurement value and the initial image of the historical wafer-level acceptable test data, the initial image of the historical online measurement value and the initial image of the historical wafer-level acceptable test data can be overlapped and aligned. The overlapping points in the initial image of the historical online measurement value and the initial image of the historical wafer-level acceptable test data are determined as the target points.

[0093] The beneficial effect of this setting is that, based on image processing technology, historical measurement data can be visualized in the form of images, making data analysis and problem identification more intuitive and efficient, and improving the efficiency of data processing. And through the alignment of the initial image, data redundancy can be reduced, further improving the efficiency of model training.

[0094] S403. Determine the image data of the online measurement values ​​corresponding to the wafer within the historical time based on the online measurement values ​​in the historical measurement data, and determine the image data of the wafer-level acceptable test data corresponding to the wafer within the historical time based on the wafer-level acceptable test data in the historical measurement data; wherein the image data of the online measurement values ​​is used to represent the online measurement values ​​corresponding to all position points on the wafer, and the image data of the wafer-level acceptable test data is used to represent the wafer-level acceptable test data corresponding to all position points on the wafer.

[0095] Exemplarily, the historical online measurement values ​​and the historical wafer-level acceptable test data are only the Inline data and WAT data at the target point in the wafer, and there are still many positions on the entire wafer where the Inline data and WAT data are unknown. After obtaining the historical online measurement values ​​and the historical wafer-level acceptable test data, the historical online measurement values ​​and the historical wafer-level acceptable test data can be supplemented. According to the historical online measurement values ​​and the historical wafer-level acceptable test data of the target point, the historical online measurement values ​​and the historical wafer-level acceptable test data of all positions on the wafer are determined. The historical online measurement values ​​of all positions on the wafer are determined as the image data of the historical online measurement values, and the historical wafer-level acceptable test data of all positions on the wafer are determined as the image data of the historical wafer-level acceptable test data. The image data of the historical online measurement values ​​can be determined on the basis of the initial image of the historical online measurement values, and the image data of the historical wafer-level acceptable test data can be determined on the basis of the initial image of the historical wafer-level acceptable test data.

[0096] In this embodiment, based on the online measurement values ​​in the historical measurement data, the image data of the online measurement values ​​corresponding to the wafer in the historical time are determined, including: obtaining the target point image of the historical online measurement values ​​based on the overlapping results of the initial image of the historical online measurement values ​​and the initial image of the historical wafer-level acceptable test data; wherein the target point image of the historical online measurement values ​​is used to represent the part of the initial image of the historical online measurement values ​​that overlaps with the initial image of the historical wafer-level acceptable test data; according to the preset spline interpolation algorithm, the target point image of the historical online measurement values ​​is supplemented and processed to obtain the image data of the online measurement values ​​corresponding to the wafer in the historical time.

[0097] Specifically, the initial image of the historical online measurement value is overlapped with the initial image of the historical wafer-level acceptable test data to obtain an overlap result, which may be an image of the overlapped portion of the initial image of the historical online measurement value and the initial image of the historical wafer-level acceptable test data. The preset points in the image of the overlapped portion are the target points, that is, according to the overlap result, the portion of the initial image of the historical online measurement value that overlaps with the initial image of the historical wafer-level acceptable test data is determined as the target point image of the historical online measurement value.

[0098] The spline interpolation algorithm is pre-set. If there are blank positions in the target point image of the historical online measurement value, the historical online measurement value of the blank position can be calculated according to the spline interpolation algorithm, thereby supplementing the target point image of the historical online measurement value. For example, Figure 5 For the image before filling, Figure 3 is the padded image, and after the padded image data of 9×11 can be obtained. In this embodiment, the specific calculation process of the spline interpolation algorithm is not specifically limited, for example, a cubic interpolation algorithm can be used.

[0099] The beneficial effect of this setting is that by filling in blank location points, the amount of data of historical online measurement values ​​is increased, thereby providing a large amount of data sets for training and improving the accuracy of model training.

[0100] In this embodiment, based on the wafer-level acceptable test data in the historical measurement data, the image data of the wafer-level acceptable test data corresponding to the wafer in the historical time is determined, including: according to the overlapping results of the initial image of the historical online measurement value and the initial image of the historical wafer-level acceptable test data, the target point image of the historical wafer-level acceptable test data is obtained; wherein, the target point image of the historical wafer-level acceptable test data is used to represent the part of the initial image of the historical wafer-level acceptable test data that overlaps with the initial image of the historical online measurement value; according to the preset spline interpolation algorithm, the target point image of the historical wafer-level acceptable test data is supplemented and processed to obtain the image data of the wafer-level acceptable test data corresponding to the wafer in the historical time.

[0101] Specifically, the initial image of the historical online measurement value is overlapped with the initial image of the historical wafer-level acceptable test data to obtain an overlap result, which may be an image of the overlapping portion of the initial image of the historical online measurement value and the initial image of the historical wafer-level acceptable test data. The points in the image of the overlapping portion are target points, that is, according to the overlap result, the portion of the initial image of the historical wafer-level acceptable test data that overlaps with the initial image of the historical online measurement value is determined as the target point image of the historical wafer-level acceptable test data. The target point image of the historical online measurement value has the same position and number of target points as the target point image of the historical wafer-level acceptable test data, but the meaning represented by the pixel value of the target point is different. The pixel value of the target point in the target point image of the historical online measurement value represents the historical online measurement value of the point, and the pixel value of the target point in the target point image of the historical wafer-level acceptable test data represents the historical wafer-level acceptable test data of the point.

[0102] The spline interpolation algorithm is preset, and there are blank positions in the target point image of the historical wafer-level acceptable test data. According to the spline interpolation algorithm, the historical wafer-level acceptable test data of the blank positions can be calculated, thereby supplementing the target point image of the historical wafer-level acceptable test data. For example, Figure 5 For the image before filling, Figure 3 is the padded image, and after the padded image data of 9×11 can be obtained. In this embodiment, the specific calculation process of the spline interpolation algorithm is not specifically limited, for example, a cubic interpolation algorithm can be used.

[0103] The beneficial effect of this setting is that by filling in blank position points, the amount of acceptable test data at the historical wafer level is increased, thereby providing a large amount of data sets for training and improving the accuracy of model training.

[0104] S404. Input the image data of the online measurement values ​​corresponding to the wafer in the historical time into the pre-built initial model for processing to obtain an output image; wherein the output image is used to represent the predicted data of the wafer-level acceptable test data obtained by performing wafer-level acceptable test on all positions on the wafer.

[0105] Exemplarily, this step may refer to the above-mentioned step S103 and will not be described in detail.

[0106] S405. If it is determined that the output image meets the preset training completion condition based on the image data of the wafer-level acceptable test data corresponding to the wafer in the historical time, a prediction model of the wafer-level acceptable test data is obtained; wherein the prediction model of the wafer-level acceptable test data is used to predict the wafer-level acceptable test data of any position point on the wafer.

[0107] Exemplarily, this step may refer to the above-mentioned step S104 and will not be described in detail.

[0108] The embodiment of the present application provides a model training method for wafer-level acceptable test, which obtains the online measurement values ​​and wafer-level acceptable test data corresponding to the preset points on the wafer in the historical time, determines the online measurement values ​​and wafer-level acceptable test data of all points on the wafer in the historical time, and thus obtains the image data of the online measurement values ​​and the image data of the wafer-level acceptable test data. The image data of the online measurement values ​​corresponding to the wafer in the historical time is input into the pre-built initial model, and the output image is output, and the output image can represent the predicted data of the wafer-level acceptable test data of all positions on the wafer. According to the image data of the wafer-level acceptable test data corresponding to the wafer in the historical time, it is determined whether the output image meets the preset training completion conditions. If so, it is determined to obtain the prediction model of the wafer-level acceptable test data, and the prediction model of the wafer-level acceptable test data can predict the wafer-level acceptable test data of any position on the wafer. By expanding the online measurement values ​​and wafer-level acceptable test data of preset points to the online measurement values ​​and wafer-level acceptable test data of all points, the amount of training data can be increased and the training accuracy of the model can be improved. The prediction model can be used to predict the wafer-level acceptable test data before the wafer is manufactured, avoiding the loss of a large amount of manufacturing cost and time cost for the foundry due to abnormalities found in the test after the wafer has completed the process manufacturing, and improving the efficiency of wafer-level acceptable testing.

[0109] Figure 6 FIG. 1 is a flow chart of a method for predicting acceptable test data at the wafer level according to an embodiment of the present application. The method can be performed by a device for predicting acceptable test data at the wafer level. Figure 6 As shown, the method comprises the following steps:

[0110] S601, obtaining the current online measurement value of the wafer.

[0111] For example, after the prediction model of wafer-level acceptable test data is trained, the WAT data can be predicted during the wafer manufacturing process, so that the electrical parameters of the wafer can be analyzed in advance based on the predicted data.

[0112] The current Inline data of the wafer can be obtained in real time or periodically during the wafer manufacturing process. The current Inline data can be the Inline data of the process pass that has been currently experienced. For example, there are three process passes in the wafer manufacturing process, and the second process pass has been completed. The current Inline data obtained can be the Inline data of the first process pass and the Inline data of the second process pass.

[0113] The acquired Inline data may be the Inline data of a preset point in the wafer, and the preset point may be a preset position point of part of the wafer, or may be a preset position point of all the wafer. The position and number of the preset points may be set according to actual needs.

[0114] The obtained online measurement values ​​can be in the form of images, and the obtained online measurement values ​​of preset points can be added to the preset spatial coordinate system to obtain an image. The coordinate point of the preset point in the spatial coordinate system is represented as the position of the preset point in the wafer, and the pixel value at the preset point is represented as the online measurement value at the preset point. The online measurement value can include multiple data categories, and for each process pass, an image can be generated for each data category. For example, if there are three process passes and four data categories, 12 images of online measurement values ​​can be obtained.

[0115] S602 , inputting the current online measurement value of the wafer into a prediction model of wafer-level acceptable test data, and obtaining prediction data of wafer-level acceptable test data corresponding to the current online measurement value of the wafer.

[0116] Exemplarily, the current online measurement value is used as input data, for example, the image corresponding to the online measurement value can be used as input data. The current online measurement value of the wafer is input into the prediction model of the trained wafer-level acceptable test data, and the input data is processed through the network layers such as the deconvolution layer and the convolution layer in the model to obtain the output data of the model. The output data can be in the form of an image, and the coordinate position of the preset site can be represented in the image of the output data, and the pixel value at the preset site is represented as the predicted value of the WAT data at the site. In the prediction model of wafer-level acceptable test data, a calculation formula for WAT data can be set, for example, WAT data can be calculated based on the data of the manufacturing process such as Inline data and a preset recipe (formula) as prediction data.

[0117] The embodiment of the present application provides a method for predicting acceptable test data at the wafer level. By obtaining the current online measurement value of the wafer, the WAT data can be predicted before the wafer manufacturing is completed. The online measurement value can be used as input data and input into the prediction model of the wafer-level acceptable test data to automatically obtain the predicted data of the WAT data, thereby improving the efficiency of the WAT test. It saves a lot of time and resources, analyzes the electrical parameters of the wafer before the wafer manufacturing is completed, predicts and identifies potential problems, and provides strong support for quality control and optimization in the semiconductor manufacturing process.

[0118] Figure 7 This is a structural block diagram of a model training device for wafer-level acceptability testing provided by an embodiment of the present application. For ease of explanation, only the parts related to the embodiment of the present disclosure are shown. Figure 7 The device includes: a data acquisition module 701, an image determination module 702, a model training module 703 and a training completion module 704.

[0119] The data acquisition module 701 is used to acquire historical measurement data corresponding to preset points on the wafer; wherein the historical measurement data includes online measurement values ​​and wafer-level acceptable test data obtained within a preset historical time, wherein the online measurement values ​​are used to represent the size information of the wafer obtained in the wafer manufacturing process, and the wafer-level acceptable test data are used to represent the data obtained by performing a wafer-level acceptable test on the wafer; the preset points are used to represent preset position points on the wafer;

[0120] The image determination module 702 is used to determine the image data of the online measurement values ​​corresponding to the wafer within the historical time according to the online measurement values ​​in the historical measurement data, and to determine the image data of the wafer-level acceptable test data corresponding to the wafer within the historical time according to the wafer-level acceptable test data in the historical measurement data; wherein the image data of the online measurement values ​​is used to represent the online measurement values ​​corresponding to all positions on the wafer, and the image data of the wafer-level acceptable test data is used to represent the wafer-level acceptable test data corresponding to all positions on the wafer;

[0121] The model training module 703 is used to input the image data of the online measurement value corresponding to the wafer in the historical time into the pre-built initial model for processing to obtain an output image; wherein the output image is used to represent the predicted data of the wafer-level acceptable test data obtained by performing a wafer-level acceptable test on all positions on the wafer;

[0122] The training completion module 704 is used to obtain a prediction model of wafer-level acceptable test data if it is determined that the output image meets the preset training completion condition based on the image data of the wafer-level acceptable test data corresponding to the wafer within the historical time; wherein the prediction model of wafer-level acceptable test data is used to predict the wafer-level acceptable test data at any position on the wafer.

[0123] Figure 8 A structural block diagram of a model training device for wafer-level acceptability testing provided in an embodiment of the present application, Figure 7 Based on the embodiment shown, Figure 8 As shown, the data acquisition module 701 includes an acquisition unit 7011 and a preprocessing unit 7012 .

[0124] The acquisition unit 7011 is used to acquire the online measurement value corresponding to the first point position preset on the wafer obtained within a preset historical time, which is the historical online measurement value, and to acquire the wafer-level acceptable test data corresponding to the second point position preset on the wafer obtained within a preset historical time, which is the historical wafer-level acceptable test data; wherein the first point position includes at least one position point, and the second point position includes at least one position point;

[0125] The preprocessing unit 7012 is used to preprocess the historical online measurement values ​​and the historical wafer-level acceptable test data, determine the target point among the first point and the second point, and obtain the historical online measurement values ​​and historical wafer-level acceptable test data corresponding to the target point.

[0126] In one example, the preprocessing unit 7012 is specifically configured to:

[0127] Adding the first point to a preset spatial coordinate system to obtain an initial image of the historical online measurement value; wherein, in the initial image of the historical online measurement value, the coordinate point of the first point is used to represent the position of the first point in the wafer, and the pixel value of the first point is used to represent the historical online measurement value of the first point;

[0128] Adding the second point to a preset spatial coordinate system to obtain an initial image of the historical wafer-level acceptable test data; wherein, in the initial image of the historical wafer-level acceptable test data, the coordinate point of the second point is used to represent the position of the second point in the wafer, and the pixel value of the second point is used to represent the historical wafer-level acceptable test data of the second point;

[0129] The initial image of the historical online measurement values ​​is overlapped with the initial image of the historical wafer-level acceptable test data, and the overlapping points in the initial image of the historical online measurement values ​​and the initial image of the historical wafer-level acceptable test data are determined as the target points.

[0130] In one example, the process of manufacturing the wafer includes at least two process passes;

[0131] The device also includes:

[0132] A data elimination module is used to determine the historical online measurement value corresponding to each process pass obtained within the preset historical time after obtaining the online measurement value corresponding to the preset first point on the wafer obtained within the preset historical time as the historical online measurement value;

[0133] Determine the correlation between the historical online measurement values ​​corresponding to the two process passes;

[0134] If the correlation satisfies a preset elimination condition, the historical online measurement value corresponding to one process pass is retained from the historical online measurement values ​​corresponding to the two process passes.

[0135] In one example, the device further includes:

[0136] The data deletion module is used to delete the online measurement value corresponding to the first point preset on the wafer after obtaining the online measurement value corresponding to the first point preset on the wafer obtained within a preset historical time, if the number of the first point is not a preset point number threshold.

[0137] In one example, the acquisition unit 7011 includes:

[0138] A WAT data acquisition subunit is used to acquire all wafer-level acceptable test data corresponding to a preset second point on the wafer obtained within a preset historical time;

[0139] The WAT data screening subunit is used to select wafer-level acceptable test data of a preset data category from all the wafer-level acceptable test data as the historical wafer-level acceptable test data.

[0140] In one example, the image determination module 702 includes:

[0141] A first obtaining unit is used to obtain a target point image of the historical online measurement value according to an overlap result of the initial image of the historical online measurement value and the initial image of the historical wafer-level acceptable test data; wherein the target point image of the historical online measurement value is used to represent a portion of the initial image of the historical online measurement value that overlaps with the initial image of the historical wafer-level acceptable test data;

[0142] The first supplementing unit is used to supplement the target point image of the historical online measurement value according to a preset spline interpolation algorithm to obtain the image data of the online measurement value corresponding to the wafer within the historical time.

[0143] In one example, the image determination module 702 includes:

[0144] A second obtaining unit is used to obtain a target point image of the historical wafer-level acceptable test data according to an overlap result of the initial image of the historical online measurement value and the initial image of the historical wafer-level acceptable test data; wherein the target point image of the historical wafer-level acceptable test data is used to represent a portion of the initial image of the historical wafer-level acceptable test data that overlaps with the initial image of the historical online measurement value;

[0145] The second supplementing unit is used to supplement the target point image of the historical wafer-level acceptable test data according to a preset spline interpolation algorithm to obtain the image data of the wafer-level acceptable test data corresponding to the wafer within the historical time.

[0146] In one example, the model training module 703 is specifically used to:

[0147] Inputting the image data of the online measurement value corresponding to the wafer in the historical time into a pre-built initial model; wherein the pre-built initial model is a U-shaped network structure;

[0148] According to the deconvolution network layer preset in the initial model, size-enlargement processing is performed on the image data of the online measurement value corresponding to the wafer within the historical time to obtain the image data after size-enlargement processing;

[0149] According to the convolutional network layer preset in the initial model, feature extraction is performed on the image data after the size expansion process to obtain the output image.

[0150] Fig. 9 This is a structural block diagram of a device for predicting acceptable test data at the wafer level provided in an embodiment of the present application. For ease of explanation, only the part related to the embodiment of the present disclosure is shown. Fig. 9 The device includes: a measurement value acquisition module 901 and a data prediction module 902.

[0151] The measurement value acquisition module 901 is used to acquire the current online measurement value of the wafer; wherein the online measurement value is used to represent the size information of the wafer obtained in the wafer manufacturing process;

[0152] The data prediction module 902 is used to input the current online measurement value of the wafer into the prediction model of wafer-level acceptable test data to obtain prediction data of wafer-level acceptable test data corresponding to the current online measurement value of the wafer; wherein the prediction model of wafer-level acceptable test data is a model obtained based on the device described in any embodiment of the present application.

[0153] Fig.10 A structural block diagram of an electronic device provided in an embodiment of the present application, such as Fig.10 As shown, the electronic device includes: a memory 1001 and a processor 1002; the memory 1001 is a memory for storing instructions executable by the processor 1002.

[0154] The processor 1002 is configured to execute the method provided in the above embodiment.

[0155] The electronic device further includes a receiver 1003 and a transmitter 1004. The receiver 1003 is used to receive instructions and data sent by other devices, and the transmitter 1004 is used to send instructions and data to external devices.

[0156] Fig.11 It is a block diagram of an electronic device according to an exemplary embodiment, which device may be a mobile phone, a computer, a digital broadcast terminal, a messaging device, a game console, a tablet device, a medical device, a fitness device, a personal digital assistant, etc.

[0157] Device 1100 may include one or more of the following components: a processing component 1102 , a memory 1104 , a power component 1106 , a multimedia component 1108 , an audio component 1110 , an input / output (I / O) interface 1112 , a sensor component 1114 , and a communication component 1116 .

[0158] The processing component 1102 generally controls the overall operation of the device 1100, such as operations associated with display, phone calls, data communications, camera operations, and recording operations. The processing component 1102 may include one or more processors 1120 to execute instructions to complete all or part of the steps of the above-mentioned method. In addition, the processing component 1102 may include one or more modules to facilitate the interaction between the processing component 1102 and other components. For example, the processing component 1102 may include a multimedia module to facilitate the interaction between the multimedia component 1108 and the processing component 1102.

[0159] The memory 1104 is configured to store various types of data to support operations on the device 1100. Examples of such data include instructions for any application or method operating on the device 1100, contact data, phone book data, messages, pictures, videos, etc. The memory 1104 can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk or optical disk.

[0160] The power supply component 1106 provides power to the various components of the device 1100. The power supply component 1106 can include a power management system, one or more power supplies, and other components associated with generating, managing, and distributing power for the device 1100.

[0161] The multimedia component 1108 includes a screen that provides an output interface between the device 1100 and the user. In some embodiments, the screen may include a liquid crystal display (LCD) and a touch panel (TP). If the screen includes a touch panel, the screen may be implemented as a touch screen to receive input signals from the user. The touch panel includes one or more touch sensors to sense touch, slide, and gestures on the touch panel. The touch sensor may not only sense the boundaries of the touch or slide action, but also detect the duration and pressure associated with the touch or slide operation. In some embodiments, the multimedia component 1108 includes a front camera and / or a rear camera. When the device 1100 is in an operating mode, such as a shooting mode or a video mode, the front camera and / or the rear camera may receive external multimedia data. Each front camera and rear camera may be a fixed optical lens system or have a focal length and optical zoom capability.

[0162] The audio component 1110 is configured to output and / or input audio signals. For example, the audio component 1110 includes a microphone (MIC), and when the device 1100 is in an operating mode, such as a call mode, a recording mode, and a speech recognition mode, the microphone is configured to receive an external audio signal. The received audio signal can be further stored in the memory 1104 or sent via the communication component 1116. In some embodiments, the audio component 1110 also includes a speaker for outputting audio signals.

[0163] I / O interface 1112 provides an interface between processing component 1102 and peripheral interface modules, which may be keyboards, click wheels, buttons, etc. These buttons may include but are not limited to: a home button, a volume button, a start button, and a lock button.

[0164] The sensor assembly 1114 includes one or more sensors for providing various aspects of status assessment for the device 1100. For example, the sensor assembly 1114 can detect the open / closed state of the device 1100, the relative positioning of components, such as the display and keypad of the device 1100, and the sensor assembly 1114 can also detect the position change of the device 1100 or a component of the device 1100, the presence or absence of user contact with the device 1100, the orientation or acceleration / deceleration of the device 1100, and the temperature change of the device 1100. The sensor assembly 1114 may include a proximity sensor configured to detect the presence of a nearby object without any physical contact. The sensor assembly 1114 may also include an optical sensor, such as a CMOS or CCD image sensor, for use in imaging applications. In some embodiments, the sensor assembly 1114 may also include an accelerometer, a gyroscope sensor, a magnetic sensor, a pressure sensor, or a temperature sensor.

[0165] The communication component 1116 is configured to facilitate wired or wireless communication between the device 1100 and other devices. The device 1100 can access a wireless network based on a communication standard, such as WiFi, 2G or 3G, or a combination thereof. In an exemplary embodiment, the communication component 1116 receives a broadcast signal or broadcast-related information from an external broadcast management system via a broadcast channel. In an exemplary embodiment, the communication component 1116 also includes a near field communication (NFC) module to facilitate short-range communication. For example, the NFC module can be implemented based on radio frequency identification (RFID) technology, infrared data association (IrDA) technology, ultra-wideband (UWB) technology, Bluetooth (BT) technology and other technologies.

[0166] In an exemplary embodiment, device 1100 may be implemented by one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors, or other electronic components to perform the above methods.

[0167] In an exemplary embodiment, a non-transitory computer-readable storage medium including instructions is also provided, such as a memory 1104 including instructions, and the instructions can be executed by the processor 1120 of the device 1100 to perform the above method. For example, the non-transitory computer-readable storage medium can be a ROM, a random access memory (RAM), a CD-ROM, a tape, a floppy disk, etc.

[0168] A non-temporary computer-readable storage medium, when the instructions in the storage medium are executed by a processor of a terminal device, enables the terminal device to execute the model training method for wafer-level acceptable testing and the prediction method for wafer-level acceptable test data of the above-mentioned terminal device.

[0169] The present application also discloses a computer program product, including a computer program, which implements the method described in this embodiment when executed by a processor.

[0170] Various embodiments of the systems and techniques described above in the present application can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include: being implemented in one or more computer programs, which can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor, which can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0171] The program code for implementing the method of the present application can be written in any combination of one or more programming languages. These program codes can be provided to a processor or controller of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that the program code, when executed by the processor or controller, enables the functions / operations specified in the flow chart and / or block diagram to be implemented. The program code can be executed entirely on the machine, partially on the machine, partially on the machine as a stand-alone software package and partially on a remote machine, or entirely on a remote machine or electronic device.

[0172] In the context of the present application, a machine-readable medium may be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, device, or equipment. A machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, an electronic, magnetic, electromagnetic, infrared, or semiconductor system, device, or device, or any suitable combination of the foregoing. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), a portable compact disk read-only memory (CD-ROM), a magnetic storage device, or any suitable combination of the foregoing.

[0173] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user can provide input to the computer. Other types of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0174] The systems and techniques described herein can be implemented in a computing system that includes back-end components (e.g., as a data electronic device), or a computing system that includes middleware components (e.g., an application electronic device), or a computing system that includes front-end components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such back-end components, middleware components, or front-end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), and the Internet.

[0175] The computer system may include a client and an electronic device. The client and the electronic device are generally far away from each other and usually interact through a communication network. The relationship between the client and the electronic device is generated by computer programs running on corresponding computers and having a client-electronic device relationship with each other. The electronic device may be a cloud electronic device, also known as a cloud computing electronic device or a cloud host, which is a host product in a cloud computing service system to solve the defects of difficult management and weak business scalability in traditional physical hosts and VPS services ("Virtual Private Server", or "VPS" for short). The electronic device may also be an electronic device of a distributed system, or an electronic device combined with a blockchain. It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps recorded in this application may be executed in parallel, sequentially, or in different orders, as long as the desired results of the technical solution disclosed in this application can be achieved, and this document is not limited here.

[0176] Those skilled in the art will readily appreciate other embodiments of the present application after considering the specification and practicing the invention disclosed herein. The present application is intended to cover any modification, use or adaptation of the present application, which follows the general principles of the present application and includes common knowledge or customary techniques in the art that are not disclosed in the present application. The specification and examples are intended to be exemplary only, and the true scope and spirit of the present application are indicated by the following claims.

[0177] It should be understood that the present application is not limited to the precise structures that have been described above and shown in the drawings, and that various modifications and changes may be made without departing from the scope thereof. The scope of the present application is limited only by the appended claims.

Claims

1. A model training method for wafer-level acceptability testing, It is characterized in that include: Acquire historical measurement data corresponding to preset points on the wafer; wherein the historical measurement data includes online measurement values ​​and wafer-level acceptable test data obtained within a preset historical time, the online measurement values ​​are used to represent the size information of the wafer obtained in the wafer manufacturing process, and the wafer-level acceptable test data are used to represent the data obtained by performing a wafer-level acceptable test on the wafer; the preset points are used to represent preset position points on the wafer; Determine the image data of the online measurement values ​​corresponding to the wafer within the historical time according to the online measurement values ​​in the historical measurement data, and determine the image data of the wafer-level acceptable test data corresponding to the wafer within the historical time according to the wafer-level acceptable test data in the historical measurement data; wherein the image data of the online measurement values ​​is used to represent the online measurement values ​​corresponding to all position points on the wafer, and the image data of the wafer-level acceptable test data is used to represent the wafer-level acceptable test data corresponding to all position points on the wafer; Inputting the image data of the online measurement value corresponding to the wafer in the historical time into the pre-built initial model for processing to obtain an output image; wherein the output image is used to represent the predicted data of the wafer-level acceptable test data obtained by performing a wafer-level acceptable test on all positions on the wafer; If it is determined that the output image meets the preset training completion condition based on the image data of the wafer-level acceptable test data corresponding to the wafer within the historical time, a prediction model of the wafer-level acceptable test data is obtained; wherein, the prediction model of the wafer-level acceptable test data is used to predict the wafer-level acceptable test data at any position point on the wafer.

2. The method according to claim 1, It is characterized in that Obtain historical measurement data corresponding to preset points on the wafer, including: Acquire an online measurement value corresponding to a first point on the wafer obtained within a preset historical time, which is a historical online measurement value, and acquire wafer-level acceptable test data corresponding to a second point on the wafer obtained within a preset historical time, which is a historical wafer-level acceptable test data; wherein the first point includes at least one position point, and the second point includes at least one position point; The historical online measurement values ​​and the historical wafer-level acceptable test data are preprocessed to determine a target point among the first point and the second point, and to obtain the historical online measurement values ​​and the historical wafer-level acceptable test data corresponding to the target point.

3. The method according to claim 2, It is characterized in that Preprocessing the historical online measurement values ​​and the historical wafer-level acceptable test data to determine a target point in the first point and the second point includes: Adding the first point to a preset spatial coordinate system to obtain an initial image of the historical online measurement value; wherein, in the initial image of the historical online measurement value, the coordinate point of the first point is used to represent the position of the first point in the wafer, and the pixel value of the first point is used to represent the historical online measurement value of the first point; Adding the second point to a preset spatial coordinate system to obtain an initial image of the historical wafer-level acceptable test data; wherein, in the initial image of the historical wafer-level acceptable test data, the coordinate point of the second point is used to represent the position of the second point in the wafer, and the pixel value of the second point is used to represent the historical wafer-level acceptable test data of the second point; The initial image of the historical online measurement values ​​is overlapped with the initial image of the historical wafer-level acceptable test data, and the overlapping points in the initial image of the historical online measurement values ​​and the initial image of the historical wafer-level acceptable test data are determined as the target points.

4. The method according to claim 2, It is characterized in that The process of manufacturing the wafer includes at least two process passes; After obtaining the online measurement value corresponding to the preset first point on the wafer obtained within the preset historical time as the historical online measurement value, the method further includes: Determine the historical online measurement value corresponding to each process pass obtained within the preset historical time; Determine the correlation between the historical online measurement values ​​corresponding to the two process passes; If the correlation satisfies a preset elimination condition, the historical online measurement value corresponding to one process pass is retained from the historical online measurement values ​​corresponding to the two process passes.

5. The method according to claim 2, It is characterized in that After obtaining the online measurement value corresponding to the preset first point on the wafer obtained within the preset historical time, the method further includes: If the number of the first points is not a preset point number threshold, the online measurement value corresponding to the preset first point on the wafer is deleted.

6. The method according to claim 2, It is characterized in that The wafer-level acceptable test data corresponding to the preset second point on the wafer obtained within a preset historical time is obtained as the historical wafer-level acceptable test data, including: Acquire all wafer-level acceptable test data corresponding to a preset second point on the wafer obtained within a preset historical time; Wafer-level acceptable test data of a preset data category is selected from all the wafer-level acceptable test data as the historical wafer-level acceptable test data.

7. The method according to claim 3, It is characterized in that Determining image data of the online measurement values ​​corresponding to the wafer within the historical time according to the online measurement values ​​in the historical measurement data includes: According to the overlapping result of the initial image of the historical online measurement value and the initial image of the historical wafer-level acceptable test data, a target point image of the historical online measurement value is obtained; wherein the target point image of the historical online measurement value is used to represent the portion of the initial image of the historical online measurement value that overlaps with the initial image of the historical wafer-level acceptable test data; According to a preset spline interpolation algorithm, the target point image of the historical online measurement value is supplemented and processed to obtain the image data of the online measurement value corresponding to the wafer within the historical time.

8. The method according to claim 3, It is characterized in that Determining image data of the wafer-level acceptable test data corresponding to the wafer within the historical time according to the wafer-level acceptable test data in the historical measurement data includes: According to the overlapping result of the initial image of the historical online measurement value and the initial image of the historical wafer-level acceptable test data, a target point image of the historical wafer-level acceptable test data is obtained; wherein the target point image of the historical wafer-level acceptable test data is used to represent the portion of the initial image of the historical wafer-level acceptable test data that overlaps with the initial image of the historical online measurement value; According to a preset spline interpolation algorithm, the target point image of the historical wafer-level acceptable test data is supplementarily processed to obtain image data of the wafer-level acceptable test data corresponding to the wafer within the historical time.

9. The method according to any one of claims 1 to 8, It is characterized in that The image data of the online measurement value corresponding to the wafer in the historical time is input into the pre-built initial model for processing to obtain an output image, including: Inputting the image data of the online measurement value corresponding to the wafer in the historical time into a pre-built initial model; wherein the pre-built initial model is a U-shaped network structure; According to the deconvolution network layer preset in the initial model, size-enlargement processing is performed on the image data of the online measurement value corresponding to the wafer within the historical time to obtain the image data after size-enlargement processing; According to the convolutional network layer preset in the initial model, feature extraction is performed on the image data after the size expansion process to obtain the output image.

10. A method for predicting acceptable test data at wafer level, It is characterized in that include: Obtain the current online measurement value of the wafer; Input the current online measurement value of the wafer into the prediction model of wafer-level acceptable test data to obtain prediction data of wafer-level acceptable test data corresponding to the current online measurement value of the wafer; wherein the prediction model of wafer-level acceptable test data is a model obtained based on the method described in any one of claims 1-9.

11. A model training device for wafer-level acceptability testing, It is characterized in that include: A data acquisition module, used to acquire historical measurement data corresponding to preset points on the wafer; wherein the historical measurement data includes online measurement values ​​and wafer-level acceptable test data obtained within a preset historical time, wherein the online measurement values ​​are used to represent the size information of the wafer obtained in the wafer manufacturing process, and the wafer-level acceptable test data are used to represent the data obtained by performing a wafer-level acceptable test on the wafer; the preset points are used to represent preset position points on the wafer; An image determination module is used to determine the image data of the online measurement values ​​corresponding to the wafer within the historical time according to the online measurement values ​​in the historical measurement data, and to determine the image data of the wafer-level acceptable test data corresponding to the wafer within the historical time according to the wafer-level acceptable test data in the historical measurement data; wherein the image data of the online measurement values ​​is used to represent the online measurement values ​​corresponding to all positions on the wafer, and the image data of the wafer-level acceptable test data is used to represent the wafer-level acceptable test data corresponding to all positions on the wafer; A model training module, used for inputting the image data of the online measurement value corresponding to the wafer in the historical time into a pre-built initial model for processing to obtain an output image; wherein the output image is used to represent the predicted data of the wafer-level acceptable test data obtained by performing a wafer-level acceptable test on all positions on the wafer; A training completion module is used to obtain a prediction model for wafer-level acceptable test data if it is determined that the output image meets a preset training completion condition based on the image data of the wafer-level acceptable test data corresponding to the wafer within the historical time; wherein the prediction model for wafer-level acceptable test data is used to predict the wafer-level acceptable test data for any position point on the wafer.

12. A device for predicting acceptable test data at wafer level, It is characterized in that include: A measurement value acquisition module, used to acquire the current online measurement value of the wafer; wherein the online measurement value is used to represent the size information of the wafer obtained in the wafer manufacturing process; A data prediction module is used to input the current online measurement value of the wafer into a prediction model of wafer-level acceptable test data to obtain prediction data of wafer-level acceptable test data corresponding to the current online measurement value of the wafer; wherein the prediction model of wafer-level acceptable test data is a model obtained based on the device described in claim 11.

13. An electronic device, It is characterized in that include: A processor, and a memory communicatively connected to the processor; The memory stores computer-executable instructions; The processor executes the computer-executable instructions stored in the memory to implement the model training method for wafer-level acceptable test as described in any one of claims 1-9 or the method for predicting wafer-level acceptable test data as described in claim 10.

14. A computer-readable storage medium, It is characterized in that The computer-readable storage medium stores computer-executable instructions, which, when executed by a processor, are used to implement the model training method for wafer-level acceptable testing as described in any one of claims 1 to 9 or the method for predicting wafer-level acceptable test data as described in claim 10.

15. A computer program product, It is characterized in that It comprises a computer program which, when executed by a processor, implements the model training method for wafer-level acceptable test described in any one of claims 1 to 9 or the method for predicting wafer-level acceptable test data described in claim 10.

Images