Mask pattern determination method, device, medium, and product

Abstract

The disclosure provides a mask pattern determination method, device, medium, and product, applied in the field of semiconductor technology. In this method, the etch bias is determined by a first etch pattern and a first lithography pattern, while the first etch pattern is obtained by fusing a first sub-image and a second sub-image. The first sub-image is generated by a physical effect simulation unit based on the first lithography pattern, and the second sub-image is generated by a neural network unit based on the first lithography pattern. The neural network unit can effectively enhance the data fitting capability of the etch model, which can improve the accuracy of the etch model to obtain an accurate first etch bias. Meanwhile, the physical effect simulation unit for etching simulates actual physical effects, which can prevent overfitting of the etch model.

Description

[Technical Field]

[0001] This disclosure belongs to the field of semiconductor technology, and in particular relates to a method, apparatus, medium and product for determining mask patterns. [Previous Technology]

[0002] The most crucial step in semiconductor wafer manufacturing is transferring the wafer's design pattern onto a silicon wafer. Among the many process steps in wafer manufacturing, photolithography and etching are the processes directly related to pattern transfer. Both photolithography and etching processes introduce deviations, resulting in differences between the photolithographic pattern, the etched pattern, and the mask pattern. Therefore, when designing the mask pattern, it is necessary to compensate for photolithographic and etching deviations. Currently, optical proximity correction (OPC) is mainly used to compensate for photolithographic deviations. For etching deviations, the current approach is to establish an etching deviation table to store the etching deviations corresponding to different patterns. Then, based on the corresponding pattern style, the corresponding etching deviation is matched in the etching deviation table. For example, the corresponding etching deviation is determined based on the line width and spacing of the current polygon's edges; thus, etching deviation compensation is completed.

[0003] However, etching differs from photolithography in that its physical and chemical processes are more complex. For patterns with varying linewidths and spacings, etching deviations will change. For simple patterns, current methods based on etching deviation tables are applicable. However, in actual production, the line density of the pattern also affects the etching deviation. For complex patterns, current methods based on etching deviation tables are difficult to use effectively, leading to an inability to accurately determine the etching deviation, and consequently, the final designed mask pattern failing to meet actual requirements. [Summary of the Invention]

[0004] This disclosure provides a mask pattern determination method, device, medium and product, which can accurately obtain the first etching deviation of different photolithography patterns through etching model, and then determine the actual required mask pattern.

[0005] In one aspect, the present disclosure provides a method for determining a mask pattern, including:

[0006] Based on the first target etching pattern, determine the first mask pattern;

[0007] Obtain the first photolithographic pattern after photolithography of the first mask pattern;

[0008] The first lithographic pattern is input into at least one physical effect analog unit of the etching model to obtain at least one first sub-image, and the first lithographic pattern is input into the neural network unit of the etching model to obtain a second sub-image; the neural network unit is trained based on the lithographic pattern sample and the corresponding etching image label;

[0009] The fusion unit based on the etching model fuses the at least one first sub-image and the second sub-image to obtain the first etched pattern after etching the first lithographic pattern;

[0010] Based on the first etched pattern and the first photolithography pattern, determine the first etch deviation of the first photolithography pattern;

[0011] Based on the first etching deviation and the first target etching pattern, the first target mask pattern is determined.

[0012] On the other hand, determining the first target mask pattern based on the first etching deviation and the first target etching pattern includes:

[0013] Based on the first etching deviation, the first target etching pattern is compensated to obtain the first target photolithography pattern;

[0014] Based on the first target lithographic pattern, optical proximity effect correction is performed to obtain the second mask pattern;

[0015] If the first etching deviation does not meet the first preset condition, the first mask pattern is updated to the second mask pattern; and the step of obtaining the first photolithographic pattern after photolithography of the first mask pattern is returned until the first etching deviation meets the first preset condition.

[0016] When the first etching deviation meets the first preset condition, the current second mask pattern is determined as the first target mask pattern.

[0017] On the other hand, determining the first mask pattern based on the first target etching pattern includes:

[0018] Based on the first target etching pattern and the first correspondence, the corresponding second etching deviation is determined; the first correspondence is the correspondence between the etching pattern and the etching deviation;

[0019] Based on the second etching deviation, the first target etching pattern is compensated to obtain the corresponding first target lithography pattern;

[0020] The first mask pattern is obtained by performing optical proximity effect correction based on the first target lithographic pattern.

[0021] On the other hand, after the fusion unit based on the etching model fuses the at least one first sub-image and the second sub-image, the method further includes:

[0022] Obtain the first target etched pattern corresponding to the first mask pattern;

[0023] Based on the first target etching pattern and the first etching pattern, a training sample is constructed;

[0024] The model is trained based on the training samples to obtain the lithography compensation and lithography etching model.

[0025] On the other hand, after training the model based on the training samples to obtain the lithographic compensation and lithographic etching model, the method further includes:

[0026] Based on the etching pattern of the second target, determine the photolithography pattern of the second target;

[0027] Input the second target lithographic pattern into the lithographic compensation and lithographic etching model to obtain the second etched pattern, and input the second target lithographic pattern into the pre-trained lithographic compensation and lithographic model to obtain the second lithographic pattern;

[0028] Based on the second etching pattern and the second photolithography pattern, a third etching deviation is determined;

[0029] If the third etching deviation does not meet the second preset condition, the second target etching pattern is compensated based on the third etching deviation to obtain the corrected second target lithography pattern;

[0030] Update the second target lithographic pattern to the corrected second target lithographic pattern; and return to the steps of inputting the second target lithographic pattern into the lithographic compensation and lithographic etching model to obtain the second etched pattern, and inputting the second target lithographic pattern into the pre-trained lithographic compensation and lithographic model to obtain the second lithographic pattern, until the third etched deviation meets the second preset condition;

[0031] When the third etching deviation meets the second preset condition, the second target mask pattern is determined based on the current corrected second target lithography pattern.

[0032] On the other hand, determining the second target lithography pattern based on the second target etching pattern includes:

[0033] Based on the second target etching pattern and the first correspondence, the corresponding fourth etching deviation is determined; the first correspondence is the correspondence between the etching pattern and the etching deviation;

[0034] Based on the fourth etching deviation, the second target etching pattern is compensated to obtain the corresponding second target lithography pattern.

[0035] On the other hand, before inputting the first lithographic pattern into at least one physical effect analog unit of the etching model to obtain at least one first sub-image, and inputting the first lithographic pattern into a neural network unit of the etching model to obtain a second sub-image, the method further includes:

[0036] Obtain the lithographic pattern sample and the corresponding etching image label;

[0037] The lithographic pattern sample is input into the at least one physical effect analog unit to obtain at least one third sub-image, and the lithographic pattern sample is input into the initial neural network unit to obtain a fourth sub-image;

[0038] The at least one third sub-image and the fourth sub-image are fused based on the fusion unit to obtain the second etched image of the lithographic pattern sample after etching;

[0039] Obtain the size difference between the first graphic key dimension of the etched image label and the second graphic key dimension of the second etched image;

[0040] Based on the size difference, the initial neural network unit is trained to obtain the neural network unit;

[0041] The etching model is determined based on the neural network unit, the physical effect analogy unit, and the fusion unit.

[0042] In another aspect, the present disclosure provides a mask pattern determination device, including: a processor and a memory storing computer program instructions;

[0043] When the processor executes the computer program instructions, it implements the mask pattern determination method as described above.

[0044] In another aspect, the present disclosure provides a computer-readable storage medium storing computer program instructions, which, when executed by a processor, implement the mask pattern determination method as described above.

[0045] In another aspect, the present disclosure provides a computer program product, wherein when the instructions in the computer program product are executed by the processor of an electronic device, the electronic device performs the mask pattern determination method as described above.

[0046] This disclosure embodiment provides a mask pattern determination method. In this scheme, the etching deviation used for etching deviation compensation is determined by a first etching pattern and a first lithographic pattern. The first etching pattern is obtained by fusing a first sub-image and a second sub-image. The first sub-image is generated by a physical effect analogy unit based on the first lithographic pattern, and the second sub-image is generated by a neural network unit based on the first lithographic pattern. The neural network unit effectively improves the etching model's ability to fit data, increases the accuracy of the etching model, and obtains an accurate first etching deviation, thereby ensuring the final determination of the actual required mask pattern. Simultaneously, the etching physical effect analogy unit analogizes actual physical effects, preventing overfitting of the etching model.

Implementation Method

[0047] The features and exemplary embodiments of various aspects of this disclosure will now be described in detail. To make the objectives, technical solutions, and advantages of this disclosure clearer, the disclosure will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are intended only to explain this disclosure and not to limit it. For those skilled in the art, this disclosure can be implemented without some of these specific details. The following description of the embodiments is merely to provide a better understanding of this disclosure by illustrating examples.

[0048] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0049] In the current mainstream integrated circuit manufacturing process, photolithography technology involves chemically changing photoresist under exposure to light of a specific wavelength, and then transferring the pattern designed on the mask into a photoresist morphology on the silicon wafer through development. The etching process, on the other hand, is based on selectively removing unwanted materials with the help of the photoresist morphology, thereby ultimately creating the required micro-patterns on the silicon wafer.

[0050] As chip technology nodes continue to evolve, the size of the design pattern is now much smaller than the wavelength of photolithography (193nm), which leads to optical proximity effect and thus photolithography deviation. To compensate for this deviation, OPC is required.

[0051] In the manufacturing process of integrated circuits, after photolithography, an etching process is required. Etching can cause etching deviations, that is, the linewidth of the photoresist before and after etching is inconsistent. For example, when a wafer contains both sparse and dense patterns to be etched, the etching rate of the dense region is lower than that of the sparse region. This is called the micro-loading effect. Another example is in the etching of high aspect ratio structures, such as deep holes or deep trenches. The etching rate of smaller holes or trenches is lower than that of larger holes or trenches. This is called the aperture effect.

[0052] It is evident that different etching rates correspond to different areas. This not only leads to changes in the overall pattern size after etching, but more importantly, patterns of the same size after photolithography will also have different sizes after etching. When designing mask patterns, etching deviations must be considered; otherwise, even if the size of the photolithographic pattern obtained after photolithography exposure meets expectations, it will still deviate from the target size after etching.

[0053] Regarding etching deviation, the current approach mainly involves establishing an etching deviation table (Bias-Table) to store the etching deviations corresponding to different graphics. Then, based on the corresponding graphic style, the corresponding etching deviation is matched in the etching deviation table. Currently, the Bias-Table mainly determines the corresponding etching deviation based on the line width and spacing of the current polygon. For simple graphics, this can barely meet the requirements, but when the graphics become more complex, the accuracy of the traditional method decreases.

[0054] For two edges with identical line width and spacing, if the density of the surrounding patterns is different, the final etching deviation will also be different. Matching etching deviations using the existing bias-table method will inevitably lead to errors. As semiconductor technology nodes continue to evolve, the precision requirements for pattern transfer (lithography and etching) are becoming increasingly higher; and traditional methods cannot meet the actual needs.

[0055] The accuracy of the Bias-Table-based approach is limited. To address this issue, this disclosure proposes an etching model to analogize the etching process. After simulating the etching image, the etching deviation can be determined based on the patterns in the image. First, a physical effect analogy unit, Term, can be established to build a partial model based on the actual physical effects generated during the etching process. Since the accuracy of etching simulation using Term alone is insufficient, this disclosure further enhances the accuracy of the etching model by incorporating a neural network unit, i.e., introducing Artificial Intelligence (AI). AI assists in etching modeling, resulting in a higher accuracy of the etching model and enabling the accurate determination of the etching deviation.

[0056] Based on this, the present disclosure provides a mask pattern determination method, device, medium, and product. The mask pattern determination method provided by the present disclosure embodiment will be described below first. Figure 1 shows a flowchart of a mask pattern determination method provided by an embodiment of the present disclosure. As shown in Figure 1, the method may include the following steps: S101 to S106.

[0057] S101: Determine the first mask pattern based on the first target etching pattern.

[0058] This disclosure is often used in mask pattern design scenarios. Therefore, before obtaining the first photolithographic pattern after the first mask pattern has been photolithographically processed, the first target etched pattern can be obtained, and the first mask pattern can be determined based on the first target etched pattern.

[0059] In the design scenario of mask pattern, the first target etched pattern is the actual desired wafer pattern. As an optional implementation method, etching deviation compensation and OPC operations can be performed on the first target etched pattern to obtain the first mask pattern. At this time, the etching deviation can be determined by the traditional method. However, since the etching deviation determined by the traditional method is not accurate, after obtaining the first mask pattern, this implementation method re-obtains the first etching deviation based on this method. Then, the first target mask pattern can be re-determined based on the first etching deviation, and the first target mask pattern can be used as the final designed mask pattern.

[0060] S102: Obtain the first photolithographic pattern after the first mask pattern has been photolithographically processed.

[0061] The first mask pattern is a pattern on a mask, which may be a pattern in the design process or a pattern that actually exists in the mask. After the mask containing the first mask pattern is processed by photolithography or photolithography simulation, a first photolithographic image is obtained, and the pattern in the first photolithographic image is the first photolithographic pattern.

[0062] This disclosure is commonly used in mask pattern design scenarios. In the mask pattern design process, it is necessary to first obtain the desired wafer pattern (i.e., the pattern on the wafer after photolithography and etching processes), and then perform etching deviation compensation and OPC on the wafer pattern to obtain the mask pattern; the above process is a common mask pattern design process. As some feasible implementation methods, the first mask pattern mentioned in this disclosure can be the above-mentioned mask pattern.

[0063] S103: Input the first lithographic pattern into at least one physical effect analog unit of the etching model to obtain at least one first sub-image, and input the first lithographic pattern into the neural network unit of the etching model to obtain a second sub-image.

[0064] The etching model provided in this embodiment is used to process an image to simulate the etching process, thereby obtaining an etched image. The first lithographic pattern obtained above mainly represents the pattern style and cannot be directly input into the etching model. Therefore, to analogize the etching process, a corresponding image needs to be generated. In practical applications, the processor can directly generate a corresponding optical image based on the pattern style of the first mask pattern, and then perform etching simulation based on the optical image of the first lithographic pattern to generate a corresponding sub-image.

[0065] The etching model in this embodiment consists of neural network units and at least one physical effect analog unit.

[0066] The physical effect analogy unit is used to analogize some physical effects in the actual etching process, such as micro-loading effect and aperture effect. By inputting the mask pattern into different physical effect analogy units, the first sub-image after corresponding processing can be obtained. For example, Gaussian convolution terms can be used to simulate the diffusion process.

[0067] The neural network units are trained based on lithographic pattern samples and corresponding etching image labels. Adding neural network units to the etching model can greatly improve the model's ability to fit data. This solves the problem that existing modeling methods cannot quickly obtain good model calibration results based on measurement results due to the complexity of the etching process, thus improving the speed and accuracy of model calibration.

[0068] Furthermore, neural network units may suffer from overfitting. In this implementation, by using physical effect analogy units to represent actual physical effects, overfitting is prevented. Moreover, training the neural network units can be based on a large amount of scanning electron microscope (SEM) image data, further reducing the problem of overfitting in the network model.

[0069] S104: The fusion unit based on the etching model fuses at least one first sub-image and a second sub-image to obtain the first etched pattern after etching the first lithographic pattern.

[0070] After obtaining the first sub-image output by the physical effect analog unit and the second sub-image output by the neural network unit, the images need to be fused to obtain the first etched image after the first lithographic pattern is etched. The specific fusion idea of ​​the etching model can be referred to the following formula: OutputImage=Termi(InputImage)+Neurm_NET(InputImage) (1)

[0071] Where OutputImage is the first etched image, InputImage is the optical image of the input etched model, Termi(InputImage) is the first sub-image, n is the number of the first sub-images, and Neurom_NET(InputImage) is the second sub-image.

[0072] Figure 2 shows a schematic diagram of a model simulation process provided in an embodiment of this disclosure. As shown in Figure 2, a first lithographic pattern 201 is first acquired, and then an optical image 202 of the first lithographic pattern 201 is generated. The optical image 202 is input into each physical effect analogy unit to obtain each first sub-image 203. The optical image 202 is then input into a neural network unit to obtain a second sub-image 204. Finally, a fusion unit based on the etching model fuses each first sub-image 203 and the second sub-image 204 to obtain a first etched image 205 after the first lithographic pattern has been etched.

[0073] For the first etched image obtained by fusing the first sub-image and the second sub-image, the first etched pattern in the first etched image is extracted. As an optional implementation, the first etched image can be binarized based on a preset threshold to obtain the first etched pattern (i.e., etched contour).

[0074] S105: Based on the first etched pattern and the first photolithography pattern, determine the first etch deviation of the first photolithography pattern.

[0075] After extracting the first etched pattern, the first etched pattern can be compared with the initial first lithographic pattern to find the differences between the two and determine the difference, i.e., the etching deviation. In practical applications, the pattern is usually broken, and then the first etching deviation corresponding to each edge is determined when determining the first etching deviation.

[0076] S106: Determine the first target mask pattern based on the first etching deviation and the first target etching pattern.

[0077] In specific implementation, there is no limitation on how to determine the first target mask pattern based on the first etching deviation and the first target etching pattern; the corresponding scheme can be selected according to the actual situation. As an optional implementation method, the first target etching pattern can be compensated based on the first etching deviation to obtain the target lithography pattern; then, optical proximity effect correction can be performed based on the target lithography pattern to obtain the first target mask pattern.

[0078] In addition, after obtaining the first etching deviation, the first etching deviation and the etching deviation obtained based on the traditional method can be compared. If the difference between the two is small, it indicates that the etching deviation obtained based on the traditional method is accurate enough. At this time, the first mask pattern can be directly used as the first target mask pattern.

[0079] Furthermore, as mentioned above, the graphic is usually broken into multiple edges, each with a corresponding first etching deviation. Therefore, as an optional implementation, the above process can be performed sequentially for each edge individually.

[0080] Since the etching deviation determined in the traditional solution is not accurate, in this implementation, after the first mask pattern is determined, the first etching deviation corresponding to the first mask pattern is determined based on the etching model proposed above. Since the first etching deviation obtained based on the etching model is more accurate, the first target mask pattern that is more in line with the actual needs can be determined.

[0081] Furthermore, in practical applications, because the OPC script, lithography compensation, and lithography model are responsible for generating all the mask data for the corresponding process layer of this technology node, any change to one affects the whole system and cannot be easily modified. Therefore, during actual process development and yield improvement of specific wafer products, the etching script is frequently adjusted. Therefore, in practice, when the etching script changes, the above-mentioned etching model can be dynamically adjusted based on the collected silicon wafer data.

[0082] This disclosure embodiment provides a mask pattern determination method, wherein the etching deviation is determined by a first etching pattern and a first lithographic pattern, and the first etching image corresponding to the first etching pattern is obtained by fusing a first sub-image and a second sub-image. The first sub-image is generated by a physical effect analogy unit based on the first lithographic pattern, and the second sub-image is generated by a neural network unit based on the first lithographic pattern. The neural network unit can effectively improve the etching model's ability to fit data, thereby improving the accuracy of the etching model to obtain an accurate first etching deviation. Simultaneously, the etching physical effect analogy unit analogizes actual physical effects, preventing overfitting of the etching model.

[0083] As mentioned in the above embodiments, the first target mask pattern can be determined based on the first etching deviation and the first target etching pattern, and a specific determination method is provided. However, in practical applications, since the first lithographic pattern of the input etching model is obtained based on traditional methods when the first etching deviation is determined for the first time, the accuracy is low, which may lead to the accuracy of the obtained first etching deviation not meeting the requirements. Therefore, another feasible implementation method is provided here. Figure 3 shows a schematic diagram of a first process for designing the first target mask pattern according to an embodiment of this disclosure. As shown in Figure 3, S106 may include the following steps:

[0084] S1061: Based on the first etching deviation, the first target etching pattern is compensated to obtain the first target photolithography pattern.

[0085] S1062: Based on the first target lithographic pattern, perform optical proximity effect correction to obtain the second mask pattern.

[0086] S1063: Determine whether the first etching deviation does not meet the first preset condition; if the first etching deviation does not meet the first preset condition, execute S1064; if the first etching deviation meets the first preset condition, execute S1065.

[0087] S1064: Update the first mask pattern to the second mask pattern; and return to S102.

[0088] S1065: The final obtained second mask pattern is determined as the first target mask pattern.

[0089] This implementation further improves the accuracy of the determined first etching deviation, thereby ensuring that the first target mask pattern better meets the actual requirements. In this implementation, multiple iterative calculations are performed to update the first mask pattern to a more accurate second mask pattern. Based on the second mask pattern, a more accurate first etching deviation can be obtained. This loop continuously improves the accuracy of the second mask pattern.

[0090] As an optional implementation, the first preset condition may be that the difference between the two corresponding first etching deviations in two adjacent rounds of iterative calculations is less than a threshold. The threshold value can be set according to requirements. When the difference between the first etching deviations corresponding to two adjacent rounds of iterative calculations is not large, it indicates that the first etching deviation obtained at this time is accurate enough, so the iterative calculation can be stopped, and the final obtained second mask pattern can be determined as the first target mask pattern.

[0091] A specific implementation method for designing a first target mask pattern is provided here. Figure 4 shows a second flowchart of designing a first target mask pattern according to an embodiment of this disclosure. As shown in Figure 4, a first target etched pattern 401 is first obtained. The etching deviation can be determined based on a traditional scheme and compensated for, resulting in a first target lithographic pattern 402. After OPC, a mask pattern is obtained. A lithographic simulation is performed to obtain a first lithographic pattern. Then, the etching model 403 is input for etching simulation to obtain the first etched pattern. Based on the first lithographic pattern and the first etched pattern, a first etching deviation is obtained.

[0092] Further, after obtaining the first etching deviation, the first target etching pattern 401 is compensated based on the first etching deviation to obtain a new first target lithographic pattern 402. This process is repeated until the first etching deviation meets the first preset condition, and the final mask pattern is determined as the first target mask pattern.

[0093] In the above embodiments, since the first lithographic pattern of the input etching model is obtained based on traditional methods when the first etching deviation is determined for the first time, the accuracy is low, which may lead to the accuracy of the obtained first etching deviation not meeting the requirements. Therefore, in this implementation, the value of the first etching deviation is continuously calibrated through multiple repeated calculations, so that the accuracy of the final first target mask pattern is higher.

[0094] As mentioned above, when determining the etching deviation using the etching model provided in this disclosure, it is necessary to obtain the first mask pattern in advance. However, during the process from the first target etching pattern to the first mask pattern, it is necessary to obtain the etching deviation to compensate for the first target etching pattern, which cannot be achieved based on the etching model. Therefore, in this implementation, the method of establishing a first correspondence is used to determine the initial etching deviation. As an optional implementation method, determining the first mask pattern based on the first target etching pattern may include:

[0095] Based on the first target etching pattern and the first correspondence, determine the corresponding second etching deviation;

[0096] Based on the second etching deviation, the etching pattern of the first target is compensated to obtain the corresponding photolithography pattern of the first target;

[0097] Based on the first target lithographic pattern, optical proximity effect correction is performed to obtain the first mask pattern.

[0098] In this embodiment, the first correspondence is the correspondence between the etched pattern and the etch deviation. As an optional implementation, the first correspondence can be implemented based on a table, which stores the correspondence between different graphic styles (including line width, spacing, etc.) and etch deviations. When it is necessary to determine the etch deviation, the graphic style corresponding to the first target etched pattern is first matched from the table, and then the corresponding second etch deviation is found.

[0099] In addition, in practical applications, the first correspondence can be corrected by the etching deviation obtained from the etching model, thereby improving accuracy.

[0100] In this implementation, the second etching deviation is first determined by the first correspondence. Although the etching deviation obtained by this method has low accuracy, it can meet the basic requirements of accuracy. Moreover, the solution is simple and efficient.

[0101] In practical applications, the process from the target etched pattern to the etched pattern requires etching deviation compensation, OPC, and simulation, which is time-consuming. Therefore, an optional implementation method is provided here. After the fusion unit based on the etching model fuses at least one first sub-image and a second sub-image, the method may further include:

[0102] Obtain the first target etching pattern corresponding to the first mask pattern;

[0103] Construct training samples based on the first target etching pattern and the first etching pattern;

[0104] The model is trained based on the training samples to obtain the lithography compensation and lithography etching model.

[0105] In this implementation, the lithography compensation and lithography etching model are used to complete processes such as OPC, lithography simulation, and etching simulation. The corresponding training data can be obtained through the etching model mentioned above; that is, based on the etching model, the first etched pattern corresponding to the first mask pattern is obtained, and since the first mask pattern corresponds to the first target etched pattern, training samples can be constructed based on the correspondence between the first target etched pattern and the first etched pattern, thereby training the model to obtain the lithography compensation and lithography etching model.

[0106] In this implementation, the lithography compensation and lithography etching models are directly trained to complete the process from lithographic pattern to etched pattern. However, training the lithography compensation and lithography etching models requires a large amount of first etched pattern label data, while in practical applications, the efficiency of obtaining label data through scanning electron microscopy is extremely low and cannot meet the requirements. Therefore, this implementation uses the etching model mentioned above to obtain a large amount of training sample data, and then trains the lithography compensation and lithography etching models.

[0107] As mentioned in the above embodiments, in order to obtain the final target mask pattern, the value of the first etching deviation can be calibrated by repeated calculations. However, each repeated calculation requires etching deviation compensation, OPC, and simulation processes, which are time-consuming and affect production efficiency. Therefore, a specific implementation method is provided here. Figure 5 shows a schematic diagram of the first process for designing the second target mask pattern according to an embodiment of this disclosure. As shown in Figure 5, after training the model based on the training samples to obtain the lithography compensation and lithography etching model, this method may further include the following steps:

[0108] S501: Determine the second target lithography pattern based on the second target etching pattern.

[0109] S502: Input the second target lithographic pattern into the lithographic compensation and lithographic etching model to obtain the second etched pattern, and input the second target lithographic pattern into the pre-trained lithographic compensation and lithographic model to obtain the second lithographic pattern.

[0110] In this implementation, the lithography compensation and lithography etching model are used to complete processes such as OPC, lithography simulation, and etching simulation. The lithography compensation and lithography model are used to implement steps such as OPC and lithography simulation.

[0111] Currently, when determining the second etching pattern from the second target lithography pattern, it is necessary to first perform OPC to obtain the corresponding mask pattern, and then perform lithography and etching simulation to obtain the second etching pattern. However, this implementation method can directly obtain the second etching pattern based on the second target lithography pattern through lithography compensation and lithography-etching model.

[0112] Accordingly, currently, when determining the second lithographic pattern based on the second target lithographic pattern, it is necessary to first perform OPC to obtain the corresponding mask pattern, and then perform lithographic simulation to obtain the second lithographic pattern. However, this implementation method can directly obtain the second lithographic pattern based on the second target lithographic pattern through lithographic compensation and lithographic model.

[0113] S503: Determine the third etching deviation based on the second etching pattern and the second photolithography pattern.

[0114] S504: Based on the third etching deviation, the second target etching pattern is compensated to obtain the corrected second target lithography pattern.

[0115] S505: Determine whether the third etching deviation meets the second preset condition; if the third etching deviation does not meet the second preset condition, execute S506; if the third etching deviation meets the second preset condition, execute S507.

[0116] S506: Update the second target lithography pattern to the corrected second target lithography pattern; and return to S501.

[0117] S507: Based on the final corrected second target lithography pattern, determine the second target mask pattern.

[0118] Corresponding to the first target etched pattern mentioned above, in the design scenario of the mask pattern, the second target etched pattern in this implementation is the actual desired wafer pattern, and the second target lithography pattern can be obtained by etch deviation compensation of the first target etched pattern.

[0119] As mentioned above, the graphic is usually broken, and each edge corresponds to a third etching deviation. Therefore, as an optional implementation, the above process can be performed sequentially for each edge individually.

[0120] As an optional implementation, the second preset condition can be that the difference between the two corresponding third etching deviations in two adjacent rounds of iterative calculations is less than a threshold. The threshold value can be set according to requirements. When the difference between the third etching deviations corresponding to two adjacent rounds of iterative calculations is not large, it indicates that the obtained third etching deviation is accurate enough, so the iterative calculation can be stopped.

[0121] When performing etching deviation compensation for the first time, the etching deviation can be determined by the traditional method. However, since the etching deviation determined by the traditional method is not accurate, this implementation method updates the third etching deviation by repeatedly calculating the lithography compensation and lithography etching model, and finally obtains a sufficiently accurate third etching deviation, thereby obtaining an accurate second target lithography pattern. Then, OPC is performed to obtain the actual required second target mask pattern, and finally the second target mask pattern is used as the final designed mask pattern.

[0122] A specific implementation method for designing a second target mask pattern is provided here. Figure 6 shows a schematic diagram of a second process for designing a second target mask pattern according to an embodiment of this disclosure. As shown in Figure 6, the second target etched pattern 601 is first obtained. The etching deviation can be determined based on the traditional scheme and the second target etched pattern 601 can be compensated to obtain the second target lithographic pattern 602. The second target lithographic pattern 602 is then input into the lithographic compensation and lithographic etching model 603 and the lithographic compensation and lithographic model 604 respectively to obtain the second lithographic pattern and the second etched pattern, and then the third etching deviation is obtained.

[0123] Further, after obtaining the third etching deviation, the second target etching pattern 601 is compensated based on the third etching deviation to obtain a new second target lithography pattern 602, and the above process is repeated until the third etching deviation meets the second preset condition. Finally, based on the finally obtained second target lithography pattern 602, OPC is performed to obtain the actual required second target mask pattern.

[0124] In this implementation, the process from the second target lithographic pattern to the second etched pattern and the second lithographic pattern is directly completed by using the lithographic compensation and lithographic etching model established in the above embodiments and the pre-trained lithographic compensation and lithographic model. This saves a lot of time in repeated calculations, and can quickly determine the second target mask pattern that meets the requirements, thereby improving production efficiency.

[0125] As mentioned above, when performing etching deviation compensation for the first time, the etching deviation can be determined using a traditional method. As an optional implementation, determining the second target lithography pattern based on the second target etching pattern can include:

[0126] Based on the second target etching pattern and the first correspondence, determine the corresponding fourth etching deviation;

[0127] Based on the fourth etching deviation, the second target etching pattern is compensated to obtain the corresponding second target lithography pattern.

[0128] In this embodiment, the first correspondence is the correspondence between the etched pattern and the etch deviation. As an optional implementation, the first correspondence can be implemented based on a table, which stores the correspondence between different graphic styles (including line width, spacing, etc.) and etch deviations. When it is necessary to determine the etch deviation, the graphic style corresponding to the second target etched pattern is first matched from the table, and then the corresponding fourth etch deviation is found.

[0129] In this implementation, the fourth etching deviation is first determined by the first correspondence. Although the etching deviation obtained by this method has low accuracy, it can meet the basic requirements of accuracy. Moreover, the solution is simple and efficient.

[0130] As mentioned in the above embodiments, the etching model provided by this disclosure includes neural network units and at least one physical effect analog unit. The neural network units need to be trained to ensure that the etching model outputs accurate results. Therefore, an optional implementation method is provided here. Figure 7 shows a schematic diagram of the training process of an etching model provided by an embodiment of this disclosure. As shown in Figure 7, before using the etching model, the method further includes the following steps:

[0131] S701: Obtain the lithography pattern sample and the corresponding etching image label.

[0132] S702: Input the lithographic pattern sample into at least one physical effect analog unit to obtain at least one third sub-image, and input the lithographic pattern sample into the initial neural network unit to obtain a fourth sub-image.

[0133] S703: Based on the fusion unit, at least one third sub-image and a fourth sub-image are fused to obtain a second etched image after the photolithographic pattern sample is etched.

[0134] S704: Obtain the size difference between the first graphic key dimension of the etched image label and the second graphic key dimension of the second etched image.

[0135] S705: Based on the size difference, train the initial neural network unit to obtain the neural network unit.

[0136] S706: Determine the etching model based on neural network units, physical effect analog units, and fusion units.

[0137] In this implementation, the lithographic pattern sample corresponds to the etched image label, and the lithographic pattern sample becomes the etched image label after etching. Therefore, by training the neural network unit with the lithographic pattern sample and the corresponding etched image label, the etching model can output accurate results.

[0138] In this embodiment, the first key dimension of the graphic is the key dimension corresponding to the graphic in the etched image label, and the second key dimension of the graphic is the key dimension corresponding to the graphic in the second etched image, which can be calculated based on the corresponding graphic. The difference between the calculated first key dimension and the second key dimension of the graphic is obtained, and then the model convergence is judged based on the evaluation function. If the model does not converge, the parameters of the initial neural network unit can be adjusted repeatedly to finally obtain the neural network unit that meets the requirements. In addition, in practical applications, the combination of physical effect analog units can be adjusted through the above training method to optimize the entire etching model.

[0139] In the process of judging whether the model converges based on the evaluation function, the evaluation can also be carried out on each side of the graph. As an optional implementation method, evaluation points can be set on the side that needs to be evaluated, and then the positions of the evaluation points of the two graphs at the same position are compared. The evaluation function value is determined based on the distance difference, and then the accuracy of the output result of the model is judged.

[0140] The above embodiments also mention that, in practice, etching scripts frequently change. When the etching script changes, the etching model can be dynamically adjusted based on the collected silicon wafer data. Since this disclosure can train the model using actual data, a new etching model can be quickly obtained, thereby improving production efficiency.

[0141] Furthermore, automated continuous modeling can be achieved by continuously collecting SEM images from the actual production process and automatically updating the etching model after process changes. For example, wafer images after etching in actual production can be obtained through SEM, and then the graphic contour of each SEM image can be extracted as etching image labels, and the corresponding lithography pattern can be used as lithography pattern samples for training the model.

[0142] This embodiment provides a training method for an etching model. An initial neural network unit is trained using photolithographic pattern samples and corresponding etching pattern labels. When there are differences between the simulated image and the label image, the neural network unit is adjusted so that the etching model can output results that match reality.

[0143] Based on the mask pattern determination method provided in the above embodiments, this disclosure also provides a mask pattern determination device. FIG8 shows a schematic diagram of the structure of the mask pattern determination device provided in this disclosure embodiment. As shown in FIG8, the device includes the following modules:

[0144] Determine module 801, used to determine the first mask pattern based on the first target etching pattern;

[0145] Acquisition module 802 is used to acquire the first photolithographic pattern after the first mask pattern has undergone photolithography;

[0146] Input module 803 is used to input the first lithographic pattern into at least one physical effect analog unit of the etching model to obtain at least one first sub-image, and to input the first lithographic pattern into the neural network unit of the etching model to obtain a second sub-image; the neural network unit is trained based on the lithographic pattern sample and the corresponding etching image label;

[0147] The fusion module 804 is used to fuse the at least one first sub-image and the second sub-image based on the fusion unit of the etching model to obtain the first etched pattern after etching the first lithographic pattern;

[0148] The module 801 is further configured to determine a first etching deviation of the first lithography pattern based on the first etched pattern and the first lithography pattern;

[0149] The module 801 is further configured to determine a first target mask pattern based on the first etching deviation and the first target etching pattern.

[0150] The apparatus provided in this disclosure embodiment is the same as the method in the above embodiment, so the two have the same embodiments and beneficial effects, which will not be repeated here.

[0151] In some embodiments, module 801 is specifically used for:

[0152] Based on the first etching deviation, the first target etching pattern is compensated to obtain the first target lithography pattern;

[0153] Based on the first target lithographic pattern, optical proximity effect correction is performed to obtain the second mask pattern;

[0154] Update the first mask pattern to the second mask pattern; and return to the step of obtaining the first photolithographic pattern after photolithography of the first mask pattern, until the first etching deviation meets the first preset condition;

[0155] The final obtained second mask pattern is determined as the first target mask pattern.

[0156] In this implementation, the value of the first etching deviation is continuously calibrated through repeated calculations, so that the accuracy of the final first target mask pattern is higher.

[0157] In some embodiments, module 801 is specifically used for:

[0158] Based on the first target etching pattern and the first correspondence, determine the corresponding second etching deviation; the first correspondence is the correspondence between the etching pattern and the etching deviation;

[0159] Based on the second etching deviation, the etching pattern of the first target is compensated to obtain the corresponding photolithography pattern of the first target;

[0160] Based on the first target lithographic pattern, optical proximity effect correction is performed to obtain the first mask pattern.

[0161] In this implementation, the second etching deviation is first determined by the first correspondence relationship. The solution is simple and efficient.

[0162] In some embodiments, the acquisition module 802 is further configured to acquire the first target etched pattern corresponding to the first mask pattern after the fusion unit based on the etched model fuses at least one first sub-image and the second sub-image;

[0163] The mask pattern determination device further includes: an etching module, used to construct training samples based on the first target etching pattern and the first etching pattern;

[0164] Training module, used to train the model based on the training samples to obtain the lithography compensation and lithography etching model.

[0165] This implementation method obtains a large amount of training sample data through the etching model mentioned above, and then trains a more accurate lithography compensation and lithography etching model.

[0166] In some embodiments, the module 801 is further configured to determine the second target lithography pattern based on the second target etch pattern after model training based on training samples to obtain a lithography compensation and lithography etching model.

[0167] The input module 803 is also used to input the second target lithography pattern into the lithography compensation and lithography etching model to obtain the second etched pattern, and to input the second target lithography pattern into the pre-trained lithography compensation and lithography model to obtain the second lithography pattern.

[0168] The module 801 is also used to determine the third etching deviation based on the second etching pattern and the second photolithography pattern;

[0169] The mask pattern determination device further includes: a compensation module, used to compensate the second target etching pattern based on the third etching deviation, so as to obtain the corrected second target photolithography pattern;

[0170] Update module, used to update the second target lithography pattern to the corrected second target lithography pattern; and return to the steps of inputting the second target lithography pattern into the lithography compensation and lithography etching model to obtain the second etched pattern, and inputting the second target lithography pattern into the pre-trained lithography compensation and lithography model to obtain the second lithography pattern, until the third etch deviation meets the second preset condition.

[0171] Based on the final corrected second target lithography pattern, determine the second target mask pattern.

[0172] This implementation saves a lot of time in repeated calculations and can quickly determine the second target mask pattern that meets the requirements, thereby improving production efficiency.

[0173] In some embodiments, module 801 is specifically used for:

[0174] Based on the second target etching pattern and the first correspondence, determine the corresponding fourth etching deviation; the first correspondence is the correspondence between the etching pattern and the etching deviation;

[0175] Based on the fourth etching deviation, the second target etching pattern is compensated to obtain the corresponding second target lithography pattern.

[0176] The method of determining etching deviation in this implementation is simple and efficient.

[0177] In some embodiments, the acquisition module 802 is further configured to acquire a lithographic pattern sample and a corresponding etching image label before inputting the first lithographic pattern into at least one physical effect analog unit of the etching model to obtain at least one first sub-image, and before inputting the first lithographic pattern into the neural network unit of the etching model to obtain a second sub-image.

[0178] The input module 803 is also used to input the lithographic pattern sample into at least one physical effect analog unit to obtain at least one third sub-image, and to input the lithographic pattern sample into the initial neural network unit to obtain a fourth sub-image;

[0179] The fusion module 804 is also used to fuse at least one third sub-image and a fourth sub-image based on the fusion unit to obtain a second etched image after the photolithography pattern sample is etched;

[0180] The module 802 is also used to obtain the size difference between the first graphic key dimension of the etched image label and the second graphic key dimension of the second etched image;

[0181] The training module is also used to train initial neural network units based on size differences to obtain neural network units;

[0182] The module 801 is also used to determine the etching model based on the neural network unit, the physical effect analog unit, and the fusion unit.

[0183] This implementation uses lithographic pattern samples and corresponding etching pattern labels to train the initial neural network units, enabling the etching model to output accurate results that match reality.

[0184] FIG9 shows a schematic diagram of the hardware structure of the mask pattern determination device provided in the present disclosure embodiment. As shown in FIG9, the mask pattern determination device may include a processor 901 and a memory 902 storing computer program instructions.

[0185] Specifically, the processor 901 may include a central processing unit (CPU), or an application specific integrated circuit (ASIC), or one or more integrated circuits that can be configured to implement the embodiments disclosed herein.

[0186] Memory 902 may include a large storage area for data or instructions. For example, and not limitingly, memory 902 may include a hard disk drive (HDD), floppy disk drive, flash memory, optical disc, magneto-optical disc, magnetic tape, or Universal Serial Bus (USB) drive, or a combination of two or more of these. Where appropriate, memory 902 may include removable or non-removable (or fixed) media. Where appropriate, memory 902 may be internal or external to an integrated gateway disaster recovery device. In a particular embodiment, memory 902 is non-volatile solid-state memory.

[0187] Memory 902 may include read-only memory (ROM), random access memory (RAM), magnetic storage medium device, optical storage medium device, flash memory device, electrical, optical or other physical / tangible memory storage device. Therefore, generally, memory includes one or more tangible (non-transitory) computer-readable storage media (e.g., memory devices) encoded with software including computer-executable instructions, and when the software is executed (e.g., by one or more processors), it is operable to perform the operations described with reference to the method according to one aspect of this disclosure.

[0188] The processor 901 reads and executes computer program instructions stored in the memory 902 to implement any of the mask pattern determination methods in the above embodiments.

[0189] In one example, the mask pattern determination device may also include a communication interface 903 and a bus 904. The processor 901, memory 902, and communication interface 903 are connected via bus 904 and communicate with each other.

[0190] Communication interface 903 is mainly used to realize communication between modules, devices, units and / or equipment in the embodiments disclosed herein.

[0191] Bus 904 includes hardware, software, or both, that couples together components of a mask pattern defining device. For example, and not as a limitation, a bus may include an Accelerate Graphical Port (AGP) or other graphics buses, an Enhanced Industry Standard Architecture (EISA) bus, a Front Side Bus (FSB), Hyper Transport (HT) interconnect, an Industry Standard Architecture (ISA) bus, an Infinite Bandwidth Interconnect, a Low Pin Count (LPC) bus, a memory bus, a Micro Channel Architecture (MCA) bus, a Peripheral Component Interconnect (PCI) bus, a PCI-Express (PCI-X) bus, a Serial Advanced Technology Attachment (SATA) bus, a Video Electronics Standards Association Local Bus (VLB) bus, or other suitable buses, or a combination of two or more of these. Where appropriate, bus 904 may include one or more buses. Although specific bus configurations are described and illustrated in this disclosure, this disclosure considers any suitable bus or interconnection.

[0192] In addition, in conjunction with the mask pattern determination method in the above embodiments, this disclosure embodiment can provide a computer storage medium for implementation. The computer storage medium stores computer program instructions; when the computer program instructions are executed by a processor, they implement any one of the mask pattern determination methods in the above embodiments.

[0193] This disclosure embodiment also provides a computer program product, including a computer program, which, when executed, implements any of the mask pattern determination methods in the above embodiments.

[0194] It should be clarified that this disclosure is not limited to the specific configurations and processes described above and shown in the figures. For the sake of brevity, detailed descriptions of known methods are omitted here. In the above embodiments, several specific steps are described and shown as examples. However, the method process of this disclosure is not limited to the specific steps described and shown. Those skilled in the art can make various changes, modifications, and additions, or change the order of steps, after understanding the spirit of this disclosure.

[0195] The functional blocks shown in the above structural diagram can be implemented as hardware, software, firmware, or a combination thereof. When implemented in hardware, they can be, for example, electronic circuits, ASICs, appropriate firmware, plug-ins, function cards, etc. When implemented in software, the elements disclosed herein are programs or code segments used to perform the required tasks. The programs or code segments can be stored in a machine-readable medium or transmitted over a transmission medium or communication link via data signals carried in a carrier wave. "Machine-readable medium" can include any medium capable of storing or transmitting information. Examples of machine-readable media include electronic circuits, semiconductor memory devices, ROMs, flash memory, erasable ROMs (EROMs), floppy disks, compact disc read-only memory (CD-ROMs), optical disks, hard disks, fiber optic media, radio frequency (RF) links, etc. The code segments can be downloaded via computer networks such as the Internet, intranets, etc.

[0196] It should also be noted that the exemplary embodiments mentioned in this disclosure describe some methods or systems based on a series of steps or apparatus. However, this disclosure is not limited to the order of the above steps. That is, the steps can be performed sequentially as mentioned in the embodiments, or in a different order than that in the embodiments, or several steps can be performed simultaneously.

[0197] Aspects of the present disclosure are described above with reference to a flow diagram and / or block diagram of one mask graphic determination method, equipment, medium and product in accordance with embodiments of the present disclosure. It should be understood that each box of the flowchart and / or block diagram and a combination of each box in the flowchart and / or block diagram may be implemented by computer program instructions. Such computer program instructions may be provided to a processor of a general-purpose computer, a specialized computer, or other programmable data processing device to produce a machine such that these instructions executed via a processor of a computer or other programmable data processing device enable the implementation of the function / action specified in one or more boxes of the flow diagram and / or block diagram. Such processors may be, but are not limited to, general-purpose processors, specialized processors, special application processors, or field programmable logic circuits. It is also understood that each box of the block diagram and / or flow diagram and a combination of the boxes of the block diagram and / or flow diagram may also be implemented by a specialized hardware performing a specified function or action, or may be implemented by a combination of specialized hardware and computer instructions.

[0198] The above content is only the specific embodiment of this disclosure. Those with usual knowledge in the field can clearly understand that for the convenience and conciseness of the description, the specific working processes of the systems, modules and units described above can be referred to the corresponding processes in the embodiments of the aforementioned methods. It is to be understood that the scope of protection of this disclosure is not limited to this, and that any usual knowledgeable person familiar with the technical area may readily think of various equivalent modifications or substitutions within the scope of the technology disclosed in this disclosure, which shall be covered within the scope of protection of this disclosure. [Brief explanation of the diagram]

[0199] To more clearly illustrate the technical solutions of the embodiments disclosed herein, the accompanying drawings used in the embodiments of this disclosure will be briefly introduced below. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. Figure 1 shows a flowchart of a mask pattern determination method provided in an embodiment of this disclosure; Figure 2 shows a schematic diagram of a model simulation process provided in an embodiment of this disclosure; Figure 3 shows a first flowchart of a first target mask pattern design provided in an embodiment of this disclosure; Figure 4 shows a second flowchart of a first target mask pattern design provided in an embodiment of this disclosure; Figure 5 shows a first flowchart of a second target mask pattern design provided in an embodiment of this disclosure; Figure 6 shows a second flowchart of a second target mask pattern design provided in an embodiment of this disclosure; Figure 7 shows a flowchart of an etching model training process provided in an embodiment of this disclosure; Figure 8 shows a structural schematic diagram of a mask pattern determination device provided in an embodiment of this disclosure; Figure 9 shows a hardware structural schematic diagram of a mask pattern determination device provided in an embodiment of this disclosure.

Claims

1. A method for determining a mask pattern, comprising: Based on the first target etching pattern, determine the first mask pattern; The process involves: acquiring a first lithographic pattern after photolithography of the first mask pattern; inputting the first lithographic pattern into at least one physical effect analog unit of an etching model to obtain at least one first sub-image; inputting the first lithographic pattern into a neural network unit of the etching model to obtain a second sub-image; training the neural network unit based on lithographic pattern samples and corresponding etching image labels; fusing the at least one first sub-image and the second sub-image using a fusion unit of the etching model to obtain a first etched pattern after etching the first lithographic pattern; determining a first etching deviation of the first lithographic pattern based on the first etched pattern and the first lithographic pattern; and determining a first target mask pattern based on the first etching deviation and the first target etched pattern.

2. The mask pattern determination method as described in claim 1, wherein, The method of determining a first target mask pattern based on the first etching deviation and the first target etching pattern includes: compensating the first target etching pattern based on the first etching deviation to obtain a first target lithographic pattern; performing optical proximity correction based on the first target lithographic pattern to obtain a second mask pattern; updating the first mask pattern to the second mask pattern if the first etching deviation does not meet a first preset condition; and returning to the step of obtaining the first lithographic pattern after lithography of the first mask pattern until the first etching deviation meets the first preset condition; and determining the current second mask pattern as the first target mask pattern if the first etching deviation meets the first preset condition.

3. The mask pattern determination method as described in claim 2, wherein, The method of determining the first mask pattern based on the first target etched pattern includes: determining a corresponding second etch deviation based on the first target etched pattern and a first correspondence relationship; the first correspondence relationship being the correspondence between the etched pattern and the etch deviation; compensating the first target etched pattern based on the second etch deviation to obtain the corresponding first target lithographic pattern; and performing optical proximity effect correction based on the first target lithographic pattern to obtain the first mask pattern.

4. The mask pattern determination method as described in claim 1, wherein, After the fusion unit based on the etching model fuses the at least one first sub-image and the second sub-image, the method further includes: obtaining a first target etching pattern corresponding to the first mask pattern; constructing training samples based on the first target etching pattern and the first etching pattern; and training the model based on the training samples to obtain a lithographic compensation and lithographic etching model.

5. The mask pattern determination method as described in claim 4, wherein, After training the model based on the training samples to obtain the lithography compensation and lithography etching model, the method further includes: determining a second target lithography pattern based on the second target etching pattern; inputting the second target lithography pattern into the lithography compensation and lithography etching model to obtain a second lithography pattern, and inputting the second target lithography pattern into a pre-trained lithography compensation and lithography model to obtain a second lithography pattern; determining a third etching deviation based on the second lithography pattern and the second lithography pattern; if the third etching deviation does not meet the second preset condition, compensating the second target lithography pattern based on the third etching deviation to obtain a corrected second target lithography pattern; updating the second target lithography pattern to the corrected second target lithography pattern; and returning to the steps of inputting the second target lithography pattern into the lithography compensation and lithography etching model to obtain a second lithography pattern, and inputting the second target lithography pattern into the pre-trained lithography compensation and lithography model to obtain a second lithography pattern, until the third etching deviation meets the second preset condition; If the third etching deviation meets the second preset condition, the second target mask pattern is determined based on the current corrected second target lithography pattern.

6. The mask pattern determination method as described in claim 5, wherein, The method of determining the second target lithography pattern based on the second target etching pattern includes: determining the corresponding fourth etching deviation based on the second target etching pattern and the first correspondence relationship; the first correspondence relationship being the correspondence between the etching pattern and the etching deviation; and compensating the second target etching pattern based on the fourth etching deviation to obtain the corresponding second target lithography pattern.

7. The mask pattern determination method as described in any one of claims 1 to 6, wherein, Before inputting the first lithographic pattern into at least one physical effect analog unit of the etching model to obtain at least one first sub-image, and inputting the first lithographic pattern into the neural network unit of the etching model to obtain a second sub-image, the method further includes: acquiring the lithographic pattern sample and the corresponding etching image label; inputting the lithographic pattern sample into the at least one physical effect analog unit to obtain at least one third sub-image, and inputting the lithographic pattern sample into an initial neural network unit to obtain a fourth sub-image; fusing the at least one third sub-image and the fourth sub-image based on the fusion unit to obtain a second etched image of the lithographic pattern sample after etching; acquiring the size difference between the first key dimension of the etching image label and the second key dimension of the second etched image; training the initial neural network unit based on the size difference to obtain the neural network unit; and determining the etching model based on the neural network unit, the physical effect analog unit, and the fusion unit.

8. A mask pattern determining device, comprising: Processor and memory storing computer program instructions; When the processor executes the computer program instructions, it implements the mask pattern determination method as described in any one of requests 1 to 7.

9. A computer-readable storage medium storing computer program instructions that, when executed by a processor, implement the mask pattern determination method as described in any one of claims 1 to 7.

10. A computer program product, wherein instructions in the computer program product, when executed by a processor of an electronic device, cause the electronic device to perform the mask pattern determination method as described in any one of claims 1 to 7.

Images