Lithography model construction method, equipment, medium and product

Abstract

The invention discloses a photoetching model construction method, equipment, a medium and a product, which are applied to the technical field of photoetching. According to the method, an alignment error is set to represent the offset existing when the mask pattern sample is aligned with the actual photoetching pattern. When the first photoetching model is trained, iteratively adjusting the alignment error and parameters of the first photoetching model until the cost function value meets an iteration stopping condition; and further obtaining a photoetching model capable of accurately simulating. According to the method, the alignment error is used as an unknown quantity to be solved, the value of the unknown quantity is calibrated in each round of iteration of the model, then a relatively accurate alignment error value is solved, and the simulation graph output by the model and the actual photoetching graph are aligned through the alignment error, so that the accuracy of the photoetching process is improved. Therefore, the influence caused by misalignment of the mask pattern sample and the actual photoetching pattern can be eliminated, so that the model training effect is ensured, and the simulation accuracy of the photoetching model is improved.

Description

Technical Field

[0001] The present application belongs to the field of photolithography technology, and in particular, relates to a photolithography model construction method, equipment, medium and product. Background Art

[0002] In order to determine the lithography model, it is necessary to use the actual exposure imaging data to calibrate the model parameters so that the model output results can accurately reflect the imaging process. During the parameter calibration process, the relative positions of the actual lithography pattern and the mask pattern in the scanning electron microscope (SEM) image need to be aligned, and the accuracy of the generated model depends on the alignment result. After the alignment is completed, the goal of parameter calibration is to reduce the deviation between the model simulation pattern and the actual lithography pattern in the SEM image. If there is an error in the alignment of the SEM image, it will have a great impact on the accuracy of the lithography model.

[0003] It can be seen that the current parameter calibration of the lithography model depends on the alignment accuracy of the mask pattern and the SEM image. However, due to the very large difference in shape between the mask pattern and the SEM image, it is currently difficult to achieve precise alignment between the two, which leads to low simulation accuracy of the lithography model. Summary of the invention

[0004] The embodiments of the present application provide a method, device, medium and product for constructing a lithography model, which can improve the accuracy of the lithography model.

[0005] On the one hand, an embodiment of the present application provides a method for constructing a lithography model, comprising:

[0006] Acquire a first lithography model to be trained, a mask pattern sample, an actual lithography pattern corresponding to the mask pattern sample, and an initial value of an alignment error;

[0007] Iteratively adjusting the value of the alignment error and the parameters of the first lithography model until the cost function value satisfies the iteration stop condition; the cost function value is determined based on the deviation value between the simulation pattern and the actual lithography pattern; the simulation pattern is obtained by simulating the first lithography model based on the mask pattern sample; the alignment error is used to align the simulation pattern with the actual lithography pattern;

[0008] When the cost function value satisfies the iteration stop condition, the current first lithography model is determined as the lithography model.

[0009] On the other hand, the first lithography model includes a plurality of sub-models;

[0010] The iterative adjustment of the value of the alignment error and the parameters of the first lithography model until the cost function value satisfies an iteration stop condition; when the cost function value satisfies the iteration stop condition, the current first lithography model is determined as the lithography model, including:

[0011] Adjusting a first parameter group of the first lithography model to obtain a second lithography model under the corresponding first parameter group; the first parameter group includes internal parameters of each sub-model;

[0012] Iteratively adjusting the value of the alignment error and the second parameter group of the second lithography model until the first cost function value satisfies the first iteration stop condition; the second parameter group includes the remaining parameters in the first lithography model except the first parameter group; the first cost function value is determined based on a first simulation pattern and the actual lithography pattern, and the first simulation pattern is obtained by simulating the adjusted second lithography model based on the mask pattern sample;

[0013] Determine a second cost function value based on the mask pattern sample, the actual lithography pattern, the final alignment error and the final second lithography model; if the second cost function value does not satisfy the second iteration stop condition, return to the step of adjusting the first parameter group of the first lithography model; until the second cost function value satisfies the second iteration stop condition;

[0014] When the second cost function value satisfies the second iteration stop condition, the second lithography model finally obtained is determined as the lithography model.

[0015] On the other hand, the iterative adjustment of the value of the alignment error and the second parameter group of the second lithography model until the first cost function value satisfies a first iteration stop condition includes:

[0016] Adjusting the value of the alignment error and the second parameter group of the second lithography model;

[0017] Inputting the mask pattern sample into the adjusted second lithography model to obtain the first simulation pattern;

[0018] Based on the adjusted alignment error, aligning the first simulation pattern with the actual lithography pattern;

[0019] Acquire a difference in a critical dimension between the first simulation pattern and the actual lithography pattern after alignment to obtain a first deviation;

[0020] Based on the first deviation, determining the first cost function value;

[0021] When the first cost function value does not satisfy the first iteration stop condition, return to the step of adjusting the value of the alignment error and the second parameter group of the second lithography model until the first cost function value finally satisfies the first iteration stop condition.

[0022] On the other hand, the first iteration stopping condition includes: a difference between the first cost function value obtained in the current iteration and the first cost function value obtained in the previous iteration is less than a first threshold.

[0023] On the other hand, adjusting the value of the alignment error and the second parameter group of the second lithography model includes:

[0024] According to a gradient-based optimization algorithm, the value of the alignment error and the second parameter group of the second lithography model are adjusted within a first value range.

[0025] On the other hand, determining a second cost function value based on the mask pattern sample, the actual lithography pattern, the finally obtained alignment error and the finally obtained second lithography model includes:

[0026] Inputting the mask pattern sample into the second photolithography model finally obtained to obtain a second simulation pattern;

[0027] Based on the alignment error finally obtained, aligning the second simulation pattern with the actual lithography pattern;

[0028] Acquire a difference in a critical dimension between the aligned second simulation pattern and the actual lithography pattern to obtain a second deviation;

[0029] Based on the second deviation, the second cost function value is determined.

[0030] On the other hand, the second iteration stopping condition includes: a difference between the second cost function value obtained in the current iteration and the second cost function value obtained in the previous iteration is less than a second threshold.

[0031] On the other hand, adjusting the first parameter group of the first lithography model to obtain a second lithography model under the corresponding first parameter group includes:

[0032] According to a non-gradient nonlinear optimization algorithm, the first parameter group of the first lithography model is adjusted within a second value range to obtain the second lithography model.

[0033] On the other hand, an embodiment of the present application further provides a lithography model construction device, comprising: a processor and a memory storing computer program instructions;

[0034] When the processor executes the computer program instructions, the lithography model construction method as described above is implemented.

[0035] On the other hand, an embodiment of the present application further provides a computer-readable storage medium, on which computer program instructions are stored. When the computer program instructions are executed by a processor, the lithography model construction method as described above is implemented.

[0036] On the other hand, an embodiment of the present application further provides a computer program product. When instructions in the computer program product are executed by a processor of an electronic device, the electronic device executes any one of the above-described lithography model construction methods.

[0037] A method for constructing a lithography model provided in an embodiment of the present application sets an alignment error to characterize the offset existing when the mask pattern sample is aligned with the actual lithography pattern. When training the first lithography model, the alignment error and the parameters of the first lithography model are iteratively adjusted until the cost function value satisfies the iteration stop condition; thereby obtaining a lithography model that can be accurately simulated. Since the offset existing when the mask pattern sample is aligned with the actual lithography pattern will cause an offset between the simulation pattern output by the model and the actual lithography pattern when the model is trained, affecting the model training effect. Therefore, the present application treats the alignment error as an unknown quantity to be solved, and calibrates the value of this unknown quantity in each round of iteration of the model, thereby solving for a more accurate alignment error value, and then aligning the simulation pattern output by the model with the actual lithography pattern through the alignment error, thereby eliminating the influence caused by the misalignment of the mask pattern sample with the actual lithography pattern, thereby ensuring the effect of model training and improving the simulation accuracy of the lithography model. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] In order to more clearly illustrate the technical solution of the embodiments of the present application, the following is a brief introduction to the drawings required for use in the embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0039] Figure 1 A schematic diagram of a process of a first photolithography model construction method provided by an embodiment of the present application is shown;

[0040] Figure 2 A schematic flow chart of a second photolithography model construction method provided by an embodiment of the present application is shown;

[0041] Figure 3 A schematic diagram of an iterative process of an inner layer model provided by an embodiment of the present application is shown;

[0042] Figure 4A schematic diagram of a process for determining a second cost function value provided by an embodiment of the present application is shown;

[0043] Figure 5 A schematic diagram of the hardware structure of a lithography model building device provided in an embodiment of the present application is shown. DETAILED DESCRIPTION

[0044] The features and exemplary embodiments of various aspects of the present application will be described in detail below. In order to make the purpose, technical solutions and advantages of the present application clearer, the present application will be further described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are only intended to explain the present application, rather than to limit the present application. For those skilled in the art, the present application can be implemented without the need for some of these specific details. The following description of the embodiments is only to provide a better understanding of the present application by illustrating the examples of the present application.

[0045] It should be noted that, in this article, relational terms such as first and second, etc. are only used 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. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the statement "include..." does not exclude the presence of other identical elements in the process, method, article or device including the element.

[0046] Photolithography is a critical step in the chip manufacturing process. First, an optical system is used to form a light intensity distribution similar to the mask pattern in the photoresist, and then a series of physical and chemical reactions are performed to form a pattern on the silicon wafer that is similar to the design pattern. Figure 1 However, as the size of devices continues to shrink, gradually approaching the resolution limit of the optical system, and due to the diffraction-limited characteristics of the lithography system, as well as various system aberrations, errors and process deviations, the patterns formed in the silicon wafer seriously deviate from the designed patterns, ultimately causing the chip to fail to work properly.

[0047] In order to solve this problem, it is necessary to optimize and correct the mask pattern used for exposure so that the pattern finally obtained in the silicon wafer is consistent with the designed pattern. To optimize and correct the mask, it is necessary to establish an accurate lithography model of the aforementioned imaging process, and to perform iterative optimization under the guidance of the lithography model to obtain the final mask pattern.

[0048] The lithography model constructed in the present application can be applied to the above-mentioned scenarios. By simulating the lithography process through the lithography model, a simulation pattern corresponding to the mask pattern is obtained, thereby guiding the optical proximity effect correction process.

[0049] The purpose of this application is to calibrate the parameters of the lithography model, so as to construct a lithography model that can be accurately simulated. In the traditional solution, it is necessary to first obtain a certain number of mask patterns and SEM images of these mask patterns after lithography, use the mask patterns as samples, and use the actual lithography patterns in the SEM images as labels to iteratively adjust the parameters of the lithography model, so that the final lithography model can output simulation patterns that are consistent with the actual lithography patterns.

[0050] During the parameter calibration process, the relative positions of the actual lithography pattern and the mask pattern in the SEM image need to be aligned first, and the accuracy of the subsequently generated lithography model depends on the alignment result. However, in actual applications, due to the large difference in shape between the mask pattern and the SEM image, it is difficult to achieve accurate alignment between the actual lithography pattern and the mask pattern, which leads to low simulation accuracy of the lithography model.

[0051] It can be seen that it is too difficult to accurately align the actual lithography pattern in the SEM image with the mask pattern, and alignment errors are inevitable in the alignment process. Therefore, the present application thinks that since it is impossible to eliminate the alignment error during alignment, the alignment error can be used as a hidden variable to be substituted into the iterative training process of the model, and the accurate alignment error value can be obtained by continuously calibrating the alignment error value, and then the simulation pattern output by the model and the actual lithography pattern are aligned based on the determined alignment error, and then the model training is performed based on the above data to eliminate the influence of the alignment error on the model training.

[0052] Based on this, the embodiment of the present application provides a method, device, medium and product for constructing a photolithography model. The following first introduces the method for constructing a photolithography model provided by the embodiment of the present application. Figure 1 FIG. 1 is a flow chart of a first lithography model construction method provided by an embodiment of the present application. Figure 1 As shown, the method includes the following steps: S101 to S103.

[0053] S101: Obtain a first lithography model to be trained, a mask pattern sample, an actual lithography pattern corresponding to the mask pattern sample, and an initial value of an alignment error.

[0054] The photolithography process includes many different types of reactions such as optical imaging, physics, and chemistry. Different reactions need to be simulated based on different sub-models, so the first photolithography model includes multiple sub-models. In practical applications, the first photolithography model including multiple sub-models is first established, and the initial values ​​of the model parameters in the first photolithography model can be set.

[0055] Like the traditional lithography model training scheme, the embodiment of the present application also needs to first obtain a certain number of mask patterns and SEM images of these mask patterns after lithography, use the mask patterns as model training samples, and use the actual lithography patterns in the SEM images as labels.

[0056] The mask pattern samples correspond to the actual lithography patterns one by one. During the training process, the mask pattern samples are usually input into the lithography model, and then the difference between the simulation pattern output by the model and the actual lithography pattern is obtained. The model parameters are adjusted based on the difference so that the final model can output results consistent with the actual lithography.

[0057] The alignment error is used to characterize the offset that exists when the mask pattern sample is aligned with the actual lithography pattern. In order to ensure the training effect of the model, it is necessary to ensure that the two patterns are aligned when obtaining the first deviation between the simulation pattern and the actual lithography pattern. However, since the actual lithography pattern is extracted from the SEM image, the mask pattern sample and the actual lithography pattern need to be aligned first, but there is a certain offset (i.e., alignment error) during alignment. In order to eliminate this alignment error, this solution will calculate the value of the alignment error, and then use the alignment error to align the simulation pattern with the actual lithography pattern.

[0058] S102: Iteratively adjust the value of the alignment error and the parameters of the first lithography model until the cost function value meets the iteration stop condition.

[0059] In the embodiment of the present application, the cost function value is determined based on the deviation value between the simulation pattern and the actual lithography pattern; the simulation pattern is obtained by simulating the first lithography model based on the mask pattern sample. When there is a large deviation between the simulation pattern and the actual lithography pattern, it means that the simulation accuracy of the lithography model is not enough, and its parameters need to be adjusted. Finally, the lithography model can achieve a relatively accurate simulation, that is, the deviation between the simulation pattern and the actual lithography pattern is small.

[0060] Alignment error is generated when the mask pattern sample is aligned with the actual lithography pattern in the SEM image. Due to the limitation of alignment technology, the existence of alignment error cannot be avoided in actual application, and the value of alignment error cannot be known, and the influence of alignment error on model training cannot be eliminated, resulting in low model accuracy. The purpose of this implementation method is to obtain a more accurate alignment error value through iterative model training, thereby eliminating its influence on model accuracy.

[0061] In this embodiment, the alignment error is used to align the simulation pattern with the actual lithography pattern. In the case where the simulation pattern is not aligned with the actual lithography pattern, the model training result is bound to be affected. In this application, an alignment error parameter is designed in the model, and the simulation pattern is aligned with the actual lithography pattern based on the alignment error, thereby eliminating the impact caused by the misalignment between the mask pattern sample and the actual lithography pattern.

[0062] However, it is currently impossible to directly obtain the exact value of the alignment error. Therefore, the present application calibrates the value of the alignment error during each iteration of the model to obtain a more accurate alignment error value. Based on this alignment error value, the simulated pattern and the actual lithography pattern are aligned, which can better eliminate the impact of the misalignment between the mask pattern sample and the actual lithography pattern.

[0063] S103: When the cost function value satisfies the iteration stop condition, the current first lithography model is determined as the lithography model.

[0064] When the cost function value satisfies the iteration stop condition, the characterization model can be simulated relatively accurately at this time, and therefore, the current first lithography model is determined as the final lithography model.

[0065] A method for constructing a lithography model provided in an embodiment of the present application sets an alignment error to characterize the offset existing when the mask pattern sample is aligned with the actual lithography pattern. When training the first lithography model, the alignment error and the parameters of the first lithography model are iteratively adjusted until the cost function value satisfies the iteration stop condition; thereby obtaining a lithography model that can be accurately simulated. Since the offset existing when the mask pattern sample is aligned with the actual lithography pattern will cause an offset between the simulation pattern output by the model and the actual lithography pattern when the model is trained, affecting the model training effect. Therefore, the present application treats the alignment error as an unknown quantity to be solved, and calibrates the value of this unknown quantity in each round of iteration of the model, thereby solving for a more accurate alignment error value, and then aligning the simulation pattern output by the model with the actual lithography pattern through the alignment error, thereby eliminating the influence caused by the misalignment of the mask pattern sample with the actual lithography pattern, thereby ensuring the effect of model training and improving the simulation accuracy of the lithography model.

[0066] The above embodiment does not specify how to iteratively adjust the alignment error and the parameters of the first lithography model. Because the first lithography model has many parameters, if the adjustment method is unreasonable, it will lead to low efficiency of model training. Therefore, the embodiment of the present application provides a specific implementation method. Figure 2 FIG. 1 is a flow chart of a second method for constructing a photolithography model provided by an embodiment of the present application. Figure 2As shown, S102 may include the following steps: S1021 to S1024; S103 may specifically include S1031.

[0067] S1021: Adjust the first parameter group of the first lithography model to obtain a second lithography model.

[0068] In the embodiment of the present application, the lithography model construction process is divided into two layers, including a first parameter group of the outer layer and a second parameter group of the inner layer. The first parameter group may include the internal parameters of each sub-model; the second parameter group may include the remaining parameters except the first parameter group, such as the weight coefficient and model threshold of each sub-model. The weight coefficient and model threshold can be better adjusted only after the internal parameters of each sub-model are determined.

[0069] As mentioned above, the initial first lithography model is first obtained, and during parameter calibration, the first parameter group of the first lithography model is adjusted to obtain the second lithography model. That is, in each round of iteration of the outer layer, the first parameter group needs to be adjusted once.

[0070] It should be noted that each parameter in the first parameter group generally needs to be within a reasonable value range. Therefore, a search space for each parameter may be given in advance, and a search is performed within the search space in each round of iteration.

[0071] S1022: Iteratively adjust the value of the alignment error and the second parameter group of the second lithography model until the first cost function value satisfies the first iteration stop condition.

[0072] The second parameter group includes weight coefficients and model thresholds of each sub-model; the first cost function value is determined based on the first simulation figure and the actual lithography figure, the adjusted alignment error is used to align the first simulation figure with the actual lithography figure, and the first simulation figure is obtained by simulating the adjusted second lithography model based on the mask figure sample.

[0073] After the parameters of the first parameter group of the outer layer are fixed, the inner layer is iteratively adjusted. In addition to adjusting the parameters of the second parameter group, the inner layer also needs to adjust the alignment error between the actual lithography pattern and the mask pattern sample.

[0074] The specific content of the first iteration stop condition is not limited and can be set according to actual needs. As an optional implementation, a maximum number of iterations can be set, and the first iteration stop condition includes that the number of iterations reaches the maximum number of iterations. It is also possible to judge whether the model has converged by the cost function value. For example, the deviation value between the simulation figure output by the model and the actual lithography figure can be calculated, and the deviation value is input into the cost function to obtain a cost function value. The difference between the cost function values ​​of two adjacent iterations is used to judge whether the model has converged. It should be noted that at this time, the convergence of the model only represents that the second parameter group of the inner layer and the alignment error have reached a local optimal solution.

[0075] As mentioned above, when calculating the deviation value between the simulation pattern output by the model and the actual lithography pattern, the alignment error between the mask pattern sample and the actual lithography pattern will also cause the alignment error between the simulation pattern and the actual lithography pattern, resulting in inaccurate calculation results, and thus ultimately obtaining a lithography model with low accuracy. Therefore, when calculating the deviation value, an alignment error value can be set to characterize the alignment error between the mask pattern sample and the actual lithography pattern, and the simulation pattern and the actual lithography pattern are aligned based on the set alignment error, thereby eliminating this alignment error.

[0076] This implementation uses the alignment error as a hidden variable in the inner iteration. Each round of iterative adjustment can calibrate the alignment error value, and a more accurate alignment error can improve the model accuracy, that is, a parameter value that better meets the requirements. After improving the model accuracy, the alignment error can be adjusted again in the next round of iteration to obtain a more accurate alignment error value.

[0077] S1023: Determine a second cost function value based on the mask pattern sample, the actual lithography pattern, the final alignment error, and the final second lithography model.

[0078] S1024: Determine whether the second cost function value satisfies the second iteration stop condition; if not, return to S1021; if satisfied, execute S1031.

[0079] S1031: Determine the finally obtained second lithography model as the lithography model.

[0080] Similarly, the specific content of the second iteration stop condition is not limited and can be set according to actual needs. As an optional implementation, a maximum number of iterations can be set, and the second iteration stop condition includes the number of iterations reaching the maximum number of iterations. It is also possible to determine whether the model has converged by using a cost function. For example, the deviation value between the simulation graphics output by the model and the actual lithography graphics can be calculated, and the deviation value can be input into the cost function to obtain a cost function value. The difference between the cost function values ​​of two adjacent iterations is used to determine whether the model has converged.

[0081] In the embodiment of the present application, the value of the first parameter group is first fixed, and then the second parameter group and the alignment error are iteratively adjusted to obtain the optimal solution of the second parameter group and the alignment error under the current value of the first parameter group. It is determined whether the lithography model converges under the current first parameter group and the second parameter group. If it does not converge, the value of the first parameter group is adjusted, and the second parameter group and the alignment error are iteratively adjusted again to obtain the optimal solution until the model converges. It can be seen that this solution divides the process of adjusting the model parameters into two layers, the inner and outer layers. First, the outer first parameter group is adjusted to the local optimum, and then the inner second parameter group is adjusted, thereby improving the training efficiency of the model.

[0082] As mentioned above, in this application, model training is divided into an inner layer and an outer layer. For the iteration of the inner layer, an embodiment of this application provides a specific implementation method. Figure 3 FIG. 1 shows a schematic diagram of an iterative process of an inner layer model provided by an embodiment of the present application. Figure 3 As shown, S1023 may include: S10231 to S10236.

[0083] S10231: Adjust the value of the alignment error and the second parameter group of the second lithography model.

[0084] The second parameter group and the alignment error both need to be within a reasonable range of values, so the search space for each parameter can be given in advance, and in each round of iteration, the search space is searched. This implementation does not limit how to adjust the second parameter group and the alignment error. As an optional implementation, the second parameter group and the alignment error can be adjusted according to a gradient-based optimization algorithm to ensure the accuracy of the solution.

[0085] S10232: Input the mask pattern sample into the adjusted second lithography model to obtain a first simulation pattern.

[0086] In this implementation, the lithography model is used to simulate the lithography process. Therefore, after the mask pattern sample is input into the adjusted second lithography model, the second lithography model will simulate the lithography process and output the corresponding first simulation pattern.

[0087] S10233: Based on the adjusted alignment error, align the first simulation pattern with the actual lithography pattern.

[0088] In the embodiment of the present application, it is not limited how to determine the first deviation, because whether the first simulation pattern is aligned with the actual lithography pattern is based on the alignment between the mask pattern sample and the actual lithography pattern. The first simulation pattern can be aligned with the actual lithography pattern based on the alignment deviation between the mask pattern sample and the actual lithography pattern.

[0089] S10234: Obtain a difference in a critical dimension between the aligned first simulation pattern and the actual lithography pattern to obtain a first deviation.

[0090] This implementation provides a specific solution for iteratively adjusting the second parameter group and the alignment error. During the model iteration process, the simulation pattern is aligned with the actual lithography pattern through the alignment error adjusted in each round. Because multiple rounds of iteration can gradually improve the accuracy of the alignment error, the first deviation between the first simulation pattern and the actual lithography pattern can be determined when the two are aligned. It is ensured that the second lithography model finally obtained reaches the local optimum, thereby improving the accuracy of the final lithography model.

[0091] S10235: Determine a first cost function value based on the first deviation.

[0092] Furthermore, after aligning the first simulation pattern with the actual lithography pattern, the difference between the key dimensions of the two can be obtained, thereby obtaining a first deviation, and then inputting the first deviation into a preset cost function to obtain a first cost function value.

[0093] S10236: Determine whether the first cost function value satisfies the first iteration stop condition; if not, return to S1031; if satisfied, proceed to S1024.

[0094] The embodiment of the present application does not limit the specific content of the first iteration stopping condition. As an optional implementation, the first iteration stopping condition may include: the difference between the first cost function value obtained in this round of iteration and the first cost function value obtained in the previous round of iteration is less than a first threshold.

[0095] The size of the first threshold is not limited, and the specific size can be set according to the actual situation. When the difference in the cost function value of two adjacent iterations is less than the first threshold, the inner optimization of the representation model has been completed, and the optimal solution of the second parameter group and the alignment error under the current first parameter group can be obtained.

[0096] In this implementation, a specific example of a first iteration stop condition is provided. By adopting a suitable iteration stop condition, the accuracy of the obtained solution can be ensured while the solution is obtained quickly. In addition, the corresponding cost function value is determined by the first deviation, and the cost function value can accurately determine whether the inner iteration of the model has converged, so that the iteration can be stopped in time. This solution improves the efficiency of model training.

[0097] This implementation provides a specific solution for iteratively adjusting the second parameter group and the alignment error, and ensures that the second lithography model finally obtained reaches a local optimum through multiple iterations.

[0098] The above does not limit how to adjust the first parameter group of the first lithography model. As a feasible implementation method, the first parameter group of the first lithography model can be adjusted according to a gradient-free nonlinear optimization algorithm to obtain a second lithography model, thereby quickly obtaining a more accurate parameter solution.

[0099] In addition, the above provides a specific embodiment of the first iteration stop condition, and the specific content of the second iteration stop condition is also not limited. As an optional implementation, the second iteration stop condition may include: the difference between the second cost function value obtained in the current iteration and the second cost function value obtained in the previous iteration is less than the second threshold value. The size of the second threshold value can also be determined according to actual conditions.

[0100] In practical applications, when judging whether the model has converged, it is necessary to determine the value of the second cost function. Here, a specific determination scheme is provided. Figure 4 FIG. 2 shows a schematic diagram of a process for determining a second cost function value provided by an embodiment of the present application. Figure 4 As shown, specifically, S1024 includes:

[0101] S10241: Input the mask pattern sample into the final second lithography model to obtain a second simulation pattern.

[0102] The second lithography model finally obtained, specifically the second lithography model obtained after the inner layer iteration is completed, that is, the optimal solution of the second parameter group and the alignment error is determined. At this time, it is necessary to perform the convergence judgment of the outer layer iteration, so the mask pattern sample is input into the second lithography model finally obtained to obtain the second simulation pattern.

[0103] S10242: Based on the final alignment error, align the second simulation pattern with the actual lithography pattern.

[0104] S10243: Obtain a difference in a critical dimension between the aligned second simulation pattern and the actual lithography pattern to obtain a second deviation.

[0105] Similarly, since the alignment between the second simulation pattern and the actual lithography pattern is based on the alignment between the mask pattern sample and the actual lithography pattern, the second simulation pattern and the actual lithography pattern can be aligned based on the alignment deviation between the mask pattern sample and the actual lithography pattern to obtain the second deviation.

[0106] S10244: Determine a second cost function value based on the second deviation.

[0107] This implementation method determines the corresponding cost function value through the second deviation. The cost function value can accurately determine whether the outer iteration of the model converges, thereby stopping the iteration in time and improving the efficiency of model training.

[0108] In this implementation, the corresponding cost function value is determined by the second deviation. The corresponding cost function value can accurately determine whether the entire model has converged, so that the iteration can be stopped in time. This solution also improves the efficiency of model training.

[0109] The following is a specific example of the method for constructing a lithography model provided by the present application. The lithography model can be expressed in the following form:

[0110]

[0111] Where AI(x,y) is the spatial image formed by the incident light at a certain position in the photoresist after passing through the mask and the optical system, and F 0 (…) is the spatial image calculation model, mask is the mask pattern used, is the parameter group to be calibrated in the model (usually including the defocus distance of the optical system, the imaging position in the photoresist, etc.).

[0112] RI (x, y) is the developed image, and the outline of the image obtained by photolithography is formed by connecting the points with a value of 0 on the image. i (…)i=1,2,3,…,m is the semi-empirical model expression, which is called the i-th model component, which takes the basic image as input (without loss of generality, the spatial image is used as input here, and it can also be a pixelated image of the mask, etc., are the parameters to be calibrated for the model component (the first parameter group). i is the weight coefficient of the corresponding model component, t is the threshold, c i and t are the second parameter group.

[0113] The semi-empirical model expression can be in many forms. For example, a semi-empirical model expression can be a Gaussian convolution on the input image to simulate the diffusion process, as described by the following formula:

[0114]

[0115] in is the convolution operation, is the Gaussian convolution kernel, g k The parameters to be calibrated.

[0116] For the first parameter group, a non-linear optimization algorithm engine based on gradient-free optimization (such as genetic algorithm, simulated annealing algorithm, non-linear optimization (Nonlinear Optimization) library, etc.) can be used for optimization. First, the search space of each parameter in the first parameter group is given, and the non-linear algorithm engine generates candidate seed parameters, that is, determines the values ​​of each parameter in the first parameter group in this round of iteration.

[0117] At this time, the basic model component images under the selected parameter seeds can be calculated, and then the optimization search of the second parameter group can be performed based on these images to obtain the minimum cost function when the first parameter group is selected as the candidate seed, that is, the optimal solution of the second parameter group under this round of iteration. The optimization result is then fed back to the gradient-free nonlinear optimization algorithm engine and the convergence is judged. If the optimization converges, the obtained model parameters are output; if not, the first parameter group seed is continued to be generated and the optimization search is continued.

[0118] For the optimization search of the second parameter group, under a certain first parameter group, each model component image F has been calculated 1 (x,y),F 2 (x,y),…,F m (x,y), the developed image can be expressed as:

[0119] RI(x,y)=c 1 ×F 1 (x,y)+c 2 ×F 2 (x,y)+…+c m ×F m (x,y)-t (4)

[0120] For the first contour calibration point, its coordinates are (x l ,y l ), whose normal direction vector is (nx l ,ny l ), the contour point comes from the dth SEM image, and the alignment error between the actual lithography pattern of the SEM image and the mask is (P d ,Q d ), P d and Q d are the alignment errors in the x and y directions, respectively, and their initial values ​​are both set to 0.

[0121] Iteratively optimize the alignment error of each SEM image and the second parameter group of the model, and the alignment error value of the current SEM image is The second deviation of the lth calibration point is:

[0122]

[0123] in:

[0124]

[0125] in Can be calculated directly from the model component images.

[0126] Substitute the above calculation results into the cost function:

[0127] Λ(c 1 , c 2 , ..., c m ,t,P 1 , Q 1 , ..., P n , Q n )=∑ l |EPE l | 2 +others (11)

[0128] The above cost function is a basic one about the model parameter c 1 ,c 2 ,…,c m ,t and the quadratic expression of the alignment error of each SEM image, the optimal solution can be easily obtained using the gradient-based optimization algorithm.

[0129] Determine whether further rounds of optimization are needed (usually a maximum number of optimization iterations can be set. When the maximum number is not reached, or when the cost function of this round continues to decrease compared to the cost function of the previous round, further optimization is performed). If so, update the SEM image alignment error value and the value of the second parameter group and perform the next round of iterative optimization.

[0130] Figure 5 FIG. 1 is a schematic diagram showing the hardware structure of the lithography model building device provided in an embodiment of the present application. Figure 5 As shown, the lithography model building device may include a processor 501 and a memory 502 storing computer program instructions.

[0131] Specifically, the processor 501 may include a central processing unit (CPU), or an application specific integrated circuit (ASIC), or may be configured to implement one or more integrated circuits of the embodiments of the present application.

[0132] The memory 502 may include a large capacity memory for data or instructions. By way of example and not limitation, the memory 502 may include a hard disk drive (HDD), a floppy disk drive, a flash memory, an optical disk, a magneto-optical disk, a magnetic tape, or a universal serial bus (USB) drive or a combination of two or more of these. In appropriate cases, the memory 502 may include a removable or non-removable (or fixed) medium. In appropriate cases, the memory 502 may be inside or outside the integrated gateway disaster recovery device. In a specific embodiment, the memory 502 is a non-volatile solid-state memory.

[0133] The memory 502 may include a read-only memory (ROM), a random access memory (RAM), a magnetic disk storage medium device, an optical storage medium device, a flash memory device, an electrical, optical or other physical / tangible memory storage device. Therefore, generally, the 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 the present disclosure.

[0134] The processor 501 implements any one of the photolithography model building methods in the above embodiments by reading and executing computer program instructions stored in the memory 502 .

[0135] In one example, the lithography model building device may further include a communication interface 503 and a bus 504. The processor 501, the memory 502, and the communication interface 503 are connected via the bus 504 and communicate with each other.

[0136] The communication interface 503 is mainly used to implement communication between various modules, devices, units and / or equipment in the embodiments of the present application.

[0137] The bus 504 includes hardware, software, or both, coupling the components of the lithography model building device to each other. By way of example and not limitation, the bus may include an Accelerate Graphical Port (AGP) or other graphics bus, an Enhanced Industry Standard Architecture (EISA) bus, a Front Side Bus (FSB), a Hyper Transport (HT) interconnect, an Industry Standard Architecture (ISA) bus, an InfiniBand 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 the above. Where appropriate, the bus 504 may include one or more buses. Although embodiments of the present application describe and illustrate a particular bus, the present application contemplates any suitable bus or interconnect.

[0138] In addition, in combination with the lithography model construction method in the above embodiment, the embodiment of the present application 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, any one of the lithography model construction methods in the above embodiment is implemented.

[0139] An embodiment of the present application also provides a computer program product, including a computer program, which implements any one of the lithography model construction methods in the above embodiments when the computer program is processed and executed.

[0140] It should be clear that the present application is not limited to the specific configuration and processing described above and shown in the figures. For the sake of simplicity, a detailed description of the known method is omitted here. In the above embodiments, several specific steps are described and shown as examples. However, the method process of the present application is not limited to the specific steps described and shown, and those skilled in the art can make various changes, modifications and additions, or change the order between the steps after understanding the spirit of the present application.

[0141] The functional blocks shown in the block diagram above can be implemented as hardware, software, firmware or a combination thereof. When implemented in hardware, it can be, for example, an electronic circuit, an ASIC, appropriate firmware, a plug-in, a function card, etc. When implemented in software, the elements of the present application are programs or code segments used to perform the required tasks. The program or code segment can be stored in a machine-readable medium, or transmitted on a transmission medium or a communication link by a data signal 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, ROM, flash memory, erasable ROM (Erasable ROM, EROM), floppy disks, compact disc read-only memory (Compact Disc Read-Only Memory, CD-ROM), optical discs, hard disks, optical fiber media, radio frequency (Radio Frequency, RF) links, etc. Code segments can be downloaded via computer networks such as the Internet, intranets, etc.

[0142] It should also be noted that the exemplary embodiments mentioned in this application describe some methods or systems based on a series of steps or devices. However, this application is not limited to the order of the above steps, that is, the steps can be performed in the order mentioned in the embodiment, or in a different order from the embodiment, or several steps can be performed simultaneously.

[0143] The above reference is made to the flowchart and / or block diagram of a lithography model construction method, device, medium and product according to an embodiment of the present disclosure. Various aspects of the present disclosure are described. It should be understood that each box in the flowchart and / or block diagram and the combination of each box in the flowchart and / or block diagram can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device to produce a machine so that these instructions executed by the processor of the computer or other programmable data processing device enable the implementation of the functions / actions specified in one or more boxes of the flowchart and / or block diagram. Such a processor can be, but is not limited to, a general-purpose processor, a special-purpose processor, a special application processor or a field programmable logic circuit. It can also be understood that each box in the block diagram and / or flowchart and the combination of boxes in the block diagram and / or flowchart can also be implemented by dedicated hardware that performs a specified function or action, or can be implemented by a combination of dedicated hardware and computer instructions.

[0144] The above contents are only specific implementation methods of the present application. Those skilled in the art can clearly understand that for the convenience and simplicity of description, the specific working processes of the systems, modules and units described above can refer to the corresponding processes in the aforementioned method embodiments, and will not be repeated here. It should be understood that the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of various equivalent modifications or replacements within the technical scope disclosed in the present application, and these modifications or replacements should be included in the protection scope of the present application.

Claims

1. A method for constructing a lithography model, characterized in that: include: Acquire a first lithography model to be trained, a mask pattern sample, an actual lithography pattern corresponding to the mask pattern sample, and an initial value of an alignment error; Iteratively adjusting the value of the alignment error and the parameters of the first lithography model until the cost function value satisfies the iteration stop condition; the cost function value is determined based on the deviation value between the simulation pattern and the actual lithography pattern; the simulation pattern is obtained by simulating the first lithography model based on the mask pattern sample; The alignment error is used to align the simulation pattern with the actual lithography pattern; When the cost function value satisfies the iteration stop condition, the current first lithography model is determined as the lithography model.

2. The method for constructing a lithography model according to claim 1, characterized in that: The first lithography model includes a plurality of sub-models; The iterative adjustment of the value of the alignment error and the parameters of the first lithography model until the cost function value satisfies an iteration stop condition; when the cost function value satisfies the iteration stop condition, the current first lithography model is determined as the lithography model, including: Adjusting a first parameter group of the first lithography model to obtain a second lithography model under the corresponding first parameter group; the first parameter group includes internal parameters of each sub-model; Iteratively adjusting the value of the alignment error and the second parameter group of the second lithography model until the first cost function value satisfies the first iteration stop condition; the second parameter group includes the remaining parameters in the first lithography model except the first parameter group; the first cost function value is determined based on a first simulation pattern and the actual lithography pattern, and the first simulation pattern is obtained by simulating the adjusted second lithography model based on the mask pattern sample; Determine a second cost function value based on the mask pattern sample, the actual lithography pattern, the final alignment error and the final second lithography model; if the second cost function value does not satisfy the second iteration stop condition, return to the step of adjusting the first parameter group of the first lithography model; until the second cost function value satisfies the second iteration stop condition; When the second cost function value satisfies the second iteration stop condition, the second lithography model finally obtained is determined as the lithography model.

3. The method for constructing a lithography model according to claim 2, characterized in that: The iteratively adjusting the value of the alignment error and the second parameter group of the second lithography model until the first cost function value satisfies a first iteration stop condition includes: Adjusting the value of the alignment error and the second parameter group of the second lithography model; Inputting the mask pattern sample into the adjusted second lithography model to obtain the first simulation pattern; Based on the adjusted alignment error, aligning the first simulation pattern with the actual lithography pattern; Acquire a difference in a critical dimension between the first simulation pattern and the actual lithography pattern after alignment to obtain a first deviation; Based on the first deviation, determining the first cost function value; When the first cost function value does not satisfy the first iteration stop condition, return to the step of adjusting the value of the alignment error and the second parameter group of the second lithography model until the first cost function value finally satisfies the first iteration stop condition.

4. The method for constructing a lithography model according to claim 3, characterized in that: The first iteration stopping condition includes: a difference between the first cost function value obtained in the current iteration and the first cost function value obtained in the previous iteration is less than a first threshold.

5. The method for constructing a lithography model according to claim 3 or 4, characterized in that: Adjusting the value of the alignment error and the second parameter group of the second lithography model includes: According to a gradient-based optimization algorithm, the value of the alignment error and the second parameter group of the second lithography model are adjusted within a first value range.

6. The method for constructing a lithography model according to claim 2, characterized in that: Determining a second cost function value based on the mask pattern sample, the actual lithography pattern, the finally obtained alignment error, and the finally obtained second lithography model includes: Inputting the mask pattern sample into the second photolithography model finally obtained to obtain a second simulation pattern; Based on the alignment error finally obtained, aligning the second simulation pattern with the actual lithography pattern; Acquire a difference in a critical dimension between the aligned second simulation pattern and the actual lithography pattern to obtain a second deviation; Based on the second deviation, the second cost function value is determined.

7. The method for constructing a lithography model according to claim 6, characterized in that: The second iteration stopping condition includes: a difference between the second cost function value obtained in the current iteration and the second cost function value obtained in the previous iteration is less than a second threshold.

8. The method for constructing a lithography model according to claim 6 or 7, characterized in that: Adjusting the first parameter group of the first lithography model to obtain a second lithography model under the corresponding first parameter group includes: According to a non-gradient nonlinear optimization algorithm, the first parameter group of the first lithography model is adjusted within a second value range to obtain the second lithography model.

9. A photolithography model building device, characterized in that: include: a processor and a memory storing computer program instructions; When the processor executes the computer program instructions, the lithography model construction method according to any one of claims 1 to 8 is implemented.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer program instructions, and when the computer program instructions are executed by a processor, the lithography model construction method according to any one of claims 1 to 8 is implemented.

11. A computer program product, characterized in that When the instructions in the computer program product are executed by a processor of an electronic device, the electronic device executes the lithography model construction method according to any one of claims 1 to 8.

Images