Photoresist model optimization method, program product, device and storage medium

Abstract

The invention relates to the technical field of photoetching, in particular to a photoresist model optimization method, a program product, a device and a storage medium. An initial photoresist model is provided, and a plurality of to-be-optimized initial parameters of the initial photoresist model are obtained; providing a genetic algorithm model to carry out iterative solution on all the to-be-optimized initial parameters to obtain optimized parameters; a fitness function set is provided, and the fitness function set comprises a minimum error function and further comprises at least one of an allowable error loss function, a grid loss function and a coefficient loss function; in the iteration process, all the optimization parameters are evaluated through the fitness function set, the fitness value of each optimization parameter is obtained, and when the next round of iteration of the optimization parameters is executed, the fitness values serve as the optimization weights of the corresponding optimization parameters; and when a preset condition is satisfied, stopping iteration and outputting a target parameter finally obtained by the genetic algorithm model. According to the method provided by the invention, the technical problems of an over-fitting phenomenon and poor consistency existing in the model can be solved.

Description

[Technical field]

[0001] The present invention relates to the field of photolithography technology, and in particular to a photoresist model optimization method, program product, device and storage medium. [Background technology]

[0002] The photolithography process can be described by optical and chemical models with the help of mathematical formulas. When light is irradiated on the mask, it diffracts. The diffraction orders are collected by the projection lens and converged on the surface of the photoresist. This imaging process is an optical process. The image projected on the photoresist stimulates a photochemical reaction. After baking, the photoresist becomes partially soluble in the developer. This is a chemical process. The model that describes this physical and chemical process is the photoresist model. The mathematical formula that describes a certain optical or chemical model is a term. Each term has its own parameters and weights. The photoresist model is formed by superimposing all the selected terms according to the weights.

[0003] The existing technical solution mainly uses the gradient descent algorithm to solve the model parameters. Gradient descent is an optimization algorithm based on solving the gradient of the function. The goal is to minimize the loss function by adjusting the model parameters through continuous iteration. In each iteration, the parameters are updated by calculating the gradient direction of the current point in the hope of finding the global optimal solution.

[0004] However, the main disadvantage of the gradient descent method is that it needs to satisfy the prerequisite that the function is differentiable, and for complex non-convex functions, it is easy to fall into the local optimal solution. In addition, general genetic algorithms also have overfitting phenomena and perform poorly in terms of consistency. [Summary of the invention]

[0005] In order to solve the technical problems of overfitting and poor consistency in a photoresist model optimization model, the present invention provides a photoresist model optimization method, program product, device and storage medium.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions: a photoresist model optimization method, providing an initial photoresist model, obtaining multiple initial parameters to be optimized of the initial photoresist model; providing a genetic algorithm model to iteratively solve all the initial parameters to be optimized to obtain the optimization parameters; providing a fitness function group, the fitness function group includes a minimization error function, and also includes at least one of an allowable error loss function, a grid loss function and a coefficient loss function; in the iterative process, using the fitness function group to evaluate all the optimization parameters to obtain the fitness value of each optimization parameter, and when executing the next round of iteration of the optimization parameters, using the fitness value as the optimization weight of the corresponding optimization parameter; when the preset conditions are met, stopping the iteration and outputting the target parameters finally obtained by the genetic algorithm model; replacing the initial parameters to be optimized with the target parameters to optimize the photoresist model.

[0007] Preferably, the fitness value includes an allowable error loss value, which is calculated by calculating the deviation between the line width of the pattern in the prediction result of the photoresist model and the line width on the corresponding actual photoresist pattern based on the allowable error loss function.

[0008] Preferably, the fitness value includes a grid loss value, and the grid loss value is calculated as follows: a measurement point is determined in the prediction result obtained by calculating the photoresist model and an offset point is obtained after dithering the measurement point. The grid loss function obtains the line width of the measurement point and its corresponding offset point, and calculates the degree of deviation between the predicted line width of the offset point and the predicted line width of the measurement point to obtain the fitness value.

[0009] Preferably, the fitness value includes a coefficient loss value, and the coefficient loss value is calculated as follows: the coefficient loss function obtains the fitness value by calculating the average square sum of the coefficients of various initial parameters to be optimized.

[0010] Preferably, the initial photoresist model includes at least one model formula item, and the initial parameters to be optimized include parameters in the model formula item and weight parameters of each model formula item, and the weight parameters are used to adjust the relative importance of different model items.

[0011] Preferably, providing a genetic algorithm model to iteratively solve all initial parameters to be optimized to obtain optimized parameters includes: setting genetic algorithm parameters; obtaining several initial parameters to be optimized as an initial population; based on the genetic algorithm parameters, iteratively calculating the initial population with three operators of selection, crossover and mutation to obtain the next generation population, and the next generation population after the iteration is the optimized parameters.

[0012] Preferably, based on the genetic algorithm parameters, the initial population is iteratively operated with the three operators of selection, crossover and mutation to obtain the next generation population. The next generation population after the iteration is the optimization parameter including: when executing the selection operator operation: selecting a number of initial parameters to be optimized from the population, and the initial parameters to be optimized with high fitness have a higher probability of being selected; when executing the crossover operator operation: selecting two initial parameters to be optimized from the population, and through the exchange combination of two chromosomes, the excellent characteristics of the parent string are inherited to the child string to generate new initial parameters to be optimized; when executing the mutation operator operation: changing certain gene values ​​in the initial parameters to be optimized to generate new initial parameters to be optimized.

[0013] In order to solve the above technical problems, the present invention provides another technical solution as follows: a computer program product, including a computer program, which implements a photoresist model optimization method when executed by a processor.

[0014] In order to solve the above technical problems, the present invention provides another technical solution as follows: a computer-readable storage medium includes a storage device, a processor and a computer program stored on the storage device, and the processor executes the computer program to implement the steps of the photoresist model optimization method.

[0015] In order to solve the above technical problems, the present invention provides another technical solution as follows: a computer-readable storage medium, which implements the steps of the photoresist model parameter solving method when the computer program is executed by a processor.

[0016] Compared with the prior art, the photoresist model optimization method, device, computer device and computer-readable storage medium provided by the present invention have the following advantages:

[0017] 1. The present invention provides a photoresist model optimization method, which more accurately calculates the fitness of initial parameters by introducing a fitness function group, and obtains better target parameters according to the fitness. Through the comprehensive use of a genetic algorithm and a specific fitness function group, not only can the global optimal solution or parameters close to the optimal solution be found, but also the model can be ensured to have consistent performance under different data disturbances, thus avoiding overfitting and improving the generalization ability of the model. Specifically, the genetic algorithm can find the global optimal solution in complex, high-dimensional nonlinear problems, and does not require the target function to be differentiable. After the fitness function group is introduced, when the initial parameter fitness calculated based on the fitness function group is used in the genetic algorithm, by using the fitness value as the optimization weight in the genetic algorithm, the overfitting phenomenon can be better avoided, and the consistency performance is better, thereby improving the accuracy of the target parameters finally outputted. The optimized parameters can accelerate the optimization process, improve the calculation efficiency, and shorten the optimization time.

[0018] 2. The present invention provides a photoresist model optimization method, which allows the error loss function to quantify the deviation between the predicted result and the actual result. The loss error obtained by calculating the line width in the predicted result and the line width loss on the actual photoresist pattern helps to evaluate the accuracy of the model prediction and gives a lower loss value within the allowable error range to reduce overfitting.

[0019] 3. The present invention provides a photoresist model optimization method, in which the grid loss function evaluates the stability of the model under data disturbance and improves the consistency of the model by calculating the average error of the offset points.

[0020] 4. The present invention provides a photoresist model optimization method, in which the coefficient loss function is used to describe the quality of the coefficient combination of each item in the model, and quantifies the importance of each coefficient of the model. By limiting the size of the coefficient in the model, the impact on the mask image can be minimized to avoid the occurrence of overfitting.

[0021] 5. The present invention provides a photoresist model optimization method, which optimizes various parameters in the initial photoresist model to improve the accuracy of the photoresist model and better describe the photoresist development process. The weight parameters can reflect the relative importance of each model formula item. By adjusting the relative importance of different model items, it helps to balance the contribution between model items and improve model performance.

[0022] 6. The present invention provides a photoresist model optimization method, which sets genetic algorithm parameters to control the operation of the genetic algorithm. According to the characteristics and requirements of this scheme, the parameters are reasonably set to obtain the best optimization effect; the random generation of the initial population can provide a diverse starting point for the genetic algorithm; through operations such as selection, crossover and mutation, the fitness of the population is gradually improved, which can accelerate the optimization process and improve the search efficiency.

[0023] 7. The present invention provides a photoresist model optimization method, which uses a genetic algorithm to select individuals for reproduction according to fitness, thereby increasing the probability of selecting initial parameters with high fitness, helping to retain parameters with high fitness, accelerating the optimization process, and improving search efficiency; and increasing the number and diversity of initial parameters through crossover and mutation, and gradually improving the fitness of the selected individuals during the cycle, making it easier to select the individual with the maximum fitness for output.

[0024] 8. A computer program product provided by the present invention has the same beneficial effects as the above-mentioned photoresist model optimization method, which will not be described in detail here.

[0025] 9. The computer device provided by the present invention has the same beneficial effects as the above-mentioned photoresist model optimization method, which will not be described in detail here.

[0026] 10. The computer-readable storage medium provided by the present invention has the same beneficial effects as the above-mentioned photoresist model optimization method, which will not be described in detail here.

Brief Description of the Drawings

[0027] Figure 1 It is a step flow chart of a photoresist model optimization method provided by the first embodiment of the present invention.

[0028] Figure 2 It is a flow chart of step S1 in a photoresist model optimization method provided by the first embodiment of the present invention.

[0029] Figure 3 It is a flowchart of step S2 in a photoresist model optimization method provided by the first embodiment of the present invention.

[0030] Figure 4 It is a flow chart of step S23 in a photoresist model optimization method provided by the first embodiment of the present invention.

[0031] Figure 5 It is a flow chart of step S4 in a photoresist model optimization method provided by the first embodiment of the present invention.

[0032] Figure 6 It is a schematic diagram of a computer program product provided according to the second embodiment of the present invention.

[0033] Figure 7 FIG. 4 is a schematic diagram of a computer device provided according to a third embodiment of the present invention.

[0034] Figure 8 It is a schematic diagram of a computer-readable storage medium provided according to the fourth embodiment of the present invention.

[0035] Description of the accompanying drawings:

[0036] 1. Computer program product; 2. Computer device; 3. Computer readable storage medium;

[0037] 11. Computer program; 21. Storage; 22. Processor; 23. Computer program; 31. Computer program. [Specific implementation method]

[0038] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and implementation examples. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0039] It should be noted that the photoresist model is a model used to describe the overall process of photoresist from exposure to development. Its basic principle involves optical and chemical processes: light is diffracted when it is irradiated on the mask, and the diffraction orders are collected by the projection lens and converged on the surface of the photoresist. This is an optical process; and the image projected on the photoresist will stimulate a photochemical reaction, and after baking, the photoresist will be partially soluble in the developer. This is a chemical process. The photoresist model is used to describe this physical and chemical process.

[0040] See also Figure 1 The first embodiment of the present invention provides a photoresist model optimization method, comprising the following steps:

[0041] S1: providing an initial photoresist model and obtaining a plurality of initial parameters to be optimized of the initial photoresist model;

[0042] S2: Provide a genetic algorithm model to iteratively solve all initial parameters to be optimized to obtain optimized parameters;

[0043] S3: providing a fitness function group, the fitness function group including a minimization error function and at least one of an allowable error loss function, a grid loss function and a coefficient loss function; in the iteration process, using the fitness function group to evaluate all optimization parameters to obtain the fitness value of each optimization parameter, and when executing the next round of iteration of the optimization parameters, using the fitness value as the optimization weight of the corresponding optimization parameter;

[0044] S4: When the preset conditions are met, the iteration is stopped and the target parameters finally obtained by the genetic algorithm model are output;

[0045] S5: replacing the initial parameters to be optimized with the target parameters to optimize the photoresist model.

[0046] Specifically, by introducing a fitness function group, the fitness of the initial parameters can be calculated more accurately, and better target parameters can be obtained based on the fitness. Through the comprehensive use of genetic algorithms and specific fitness function groups, not only can the global optimal solution or parameters close to the optimal solution be found, but also the model can be ensured to have consistent performance under different data disturbances, avoiding overfitting and improving the generalization ability of the model. Optimizing parameters can accelerate the optimization process, improve computing efficiency and shorten optimization time.

[0047] It should be noted that the initial photoresist model includes at least one model formula item, and the initial parameters to be optimized include parameters in the model formula item and weight parameters of each model formula item. The photoresist model is obtained by superimposing each model formula item according to its weight parameter.

[0048] It should be noted that the fitness function group includes a minimization error function and at least one of an allowable error loss function, a grid loss function and a coefficient loss function, and the loss function type and coefficient of the fitness function group are set according to the training data, that is, a linear combination of the minimization error function and at least one of the allowable error loss function, the grid loss function and the coefficient loss function together constitutes the fitness function, so as to improve the fitness of the output optimization parameters through more accurate fitness evaluation and processing of optimization parameters, so that the solved photoresist model can effectively describe the development pattern, which can significantly improve the accuracy and efficiency of microelectronics design work and reduce the cost of repeated tape-outs.

[0049] It should be noted that the coefficient of the minimization error function is non-zero by default. During the iteration process, the value of a loss function calculated by the fitness function group may be zero. The function value is zero, that is, the fitness value is zero, indicating that the parameters of the current model perform well on this loss and the loss is low.

[0050] It can be understood that the fitness value is obtained by calculating and adding each loss function in the fitness function group. The fitness value includes the minimization loss value, the allowable error loss value, the grid loss value and the coefficient loss value, and is obtained by adding the linear combination of the minimization loss value, the allowable error loss value, the grid loss value and the coefficient loss value.

[0051] Specifically, the calculation method of the allowable error loss value is as follows: the allowable error loss function obtains the fitness value by calculating the deviation between the line width of the graphic in the predicted result of the photoresist model and the line width on the actual photoresist graphic, and quantifies the deviation between the predicted result and the actual result. The allowable error loss function represents the acceptable degree of loss error, which helps to evaluate the accuracy of the model prediction and gives a lower loss value within the allowable error range to reduce overfitting.

[0052] Furthermore, the prediction result is an output pattern of the initial photoresist model, and the line width of the pattern in the prediction result is obtained by calculating the pattern in the photoresist model.

[0053] Furthermore, in this embodiment, the allowable error loss function is:

[0054]

[0055] Where cost rs represents the allowable error loss function, ModelCD i To predict the line width of the graphics in the result, WaferCD i is the line width of the graphics in the actual photoresist model, n represents the number of line widths of the graphics in the predicted results / line widths on the actual photoresist graphics, Weight i Represents the weight of the line width of the graphic in the prediction result; when the error between the line width of the graphic in the prediction result and the line width of the graphic in the actual photoresist model is within the allowable error, the ReLU function transforms the function value into a number close to 0, and when the error exceeds the allowable error, the ReLU function transforms the function value into a calculated value, and the weighted average of all samples in the training set is the allowable error loss.

[0056] Specifically, the grid loss value is calculated as follows: the measurement point is determined in the prediction result obtained by calculating the photoresist model and the offset point is obtained after dithering the measurement point. The grid loss function obtains the line width of the measurement point and its corresponding offset point, and calculates the degree of deviation between the predicted line width after the offset and the predicted line width of the measurement point to obtain the fitness value, so as to evaluate the stability of the photoresist model under data disturbance, and improve the consistency of the model by calculating the average error of the offset point.

[0057] Furthermore, a measurement point is selected in the graphic to measure the line width of the point, and after jittering the measurement point to different degrees, the point where the position has changed slightly, that is, the line width at the offset point, is measured again. By calculating the error or difference between the predicted value and the actual value, the consistency between the line width value in the graphic predicted by the quantification model and the line width value of the offset point obtained by actual measurement is determined.

[0058] Furthermore, the grid loss value is obtained by calculating the average error after the offset of multiple measurement points to obtain a more accurate loss value, find the parameter combination that optimizes the algorithm performance, and improve the consistency of the model.

[0059] Furthermore, in this embodiment, the grid loss function is

[0060]

[0061] Where cost grid represents the grid loss function, n represents the number of line widths of the pattern in the predicted result / line widths on the actual photoresist pattern, r represents the number of times the measurement point is offset, WaferCD i is the line width of the ith measurement point actually measured, ModelCD i,j The line width of the i-th measurement point after the j-th offset. After multiple offsets of the measurement points, the average error of the offset points is calculated by the line width of the measurement points in the prediction results and the line width of the measurement points after the offset to improve the consistency of the model.

[0062] It should be noted that poor consistency performance may lead to overfitting, performance fluctuations, and difficulty in adjusting parameters and optimizing models, affecting the accuracy of the results provided by the model. When the measurement point jitters slightly, the grid loss function can capture the error caused by this jitter and balance the prediction accuracy and stability of the model, reducing this error during the optimization process, thereby improving the consistency of the model.

[0063] Specifically, the coefficient loss value is calculated as follows: The coefficient loss function calculates the average square sum of the coefficients of the initial parameters to be optimized to obtain the fitness value. The coefficient loss function is used to describe the quality of the coefficient combination of each item in the model, quantify the importance of each coefficient of the model, and minimize the impact on the mask image by limiting the size of the coefficients in the model, so as to minimize the impact of the mask operation on the original image and avoid overfitting.

[0064] It should be noted that an overfitting model performs well on the training set, but performs poorly on unseen test sets or data in actual applications, making the model less stable and affecting the accuracy of the model.

[0065] Furthermore, in this embodiment, the coefficient loss function is

[0066]

[0067] Where cost coeffs represents the coefficient loss function, coeff is the weight parameter of the model formula term, m represents the number of weight parameters, and the coefficient loss function is used to ensure that the impact on the mask image is minimized as much as possible, so that the impact of the mask operation on the original image is minimized, while maintaining the necessary processing effect and avoiding overfitting.

[0068] Specifically, the minimization error function is used to measure the difference between the line width in the model prediction graph and the line width actually measured, so as to adjust the parameters of the model so that the prediction results of the model are closer to the actual values, thereby improving the prediction accuracy of the model.

[0069] Furthermore, the minimized error function is

[0070]

[0071] Where cost represents the minimization error function, ModelCD i ModelCD' is the line width of the measured point in the prediction result. i,j The line width after the measurement point is offset.

[0072] It should be noted that the genetic algorithm is an optimization algorithm that simulates the biological evolution process. Its goal is to generate individuals with higher fitness through continuous iteration by simulating natural selection and genetic mechanisms. In the genetic algorithm, each individual has a set of genes. In each iteration, new gene combinations are generated through operations such as crossover, mutation, and selection in the hope of finding the global optimal solution. The advantage of the genetic algorithm is that it can find the global optimal solution in complex, high-dimensional nonlinear problems, and does not require the objective function to be differentiable.

[0073] In this embodiment, an individual in the genetic algorithm includes a parameter in a model formula item and a weight parameter of the model formula item.

[0074] Please note that Figure 2 , step S1 specifically includes the following steps:

[0075] S11: Select a combination of model formula items to establish a photoresist model;

[0076] S12: extracting parameters and weight parameters in each model formula item as initial parameters to be optimized;

[0077] S13: Use binary coding to map feasible solutions from the solution space to the search space of the genetic algorithm.

[0078] It should be noted that binary coding is one of the most commonly used coding methods in genetic algorithms, and mapping is achieved by representing the solution to the problem as a binary string.

[0079] In this embodiment, the value range of the model formula items is taken as a feasible solution, and the feasible solution is mapped to the search space of the genetic algorithm.

[0080] Furthermore, the search space of a genetic algorithm refers to the set or range of all possible solutions at the virtual level when the algorithm is solving a problem.

[0081] Specifically, the feasible decoding operation in step S13 simplifies the implementation of genetic operations, such as selection, crossover, and mutation, and enables the algorithm to run efficiently on a computer.

[0082] Please note that Figure 3 , step S2 specifically includes the following steps:

[0083] S21: Setting genetic algorithm parameters;

[0084] S22: obtaining a number of initial parameters to be optimized as an initial population;

[0085] S23: Based on the genetic algorithm parameters, the initial population is iterated using the three operators of selection, crossover and mutation to obtain the next generation population. The next generation population after the iteration is the optimization parameter.

[0086] It should be noted that the genetic algorithm parameters include the maximum evolutionary generation, population size, crossover probability and mutation probability. The maximum evolutionary generation refers to the maximum number of iterations of the genetic algorithm, that is, the maximum number of generations of population evolution before the algorithm stops; the population size refers to the number of individuals in each generation; the crossover probability refers to the probability of two parent individuals performing a crossover operation; the mutation probability refers to the probability of an individual gene mutating. Genetic algorithm parameters play a vital role in the algorithm and jointly determine the performance and results of the algorithm.

[0087] Optionally, in step S22, several initial parameters to be optimized can be obtained by random generation as the initial population; or the training data in the historical model can be used as the initial population. When the size of the training data in the historical model is larger than the initial population size, part of the training data in the historical model is selected as the initial population size. When the size of the training data in the historical model is smaller, the missing parameters are supplemented by operations such as crossover and mutation.

[0088] It should be noted that step S23 specifically includes the following steps:

[0089] S231: When executing the selection operator operation: a number of initial parameters to be optimized are selected from the population, and the initial parameters to be optimized with high fitness are more likely to be selected;

[0090] S232: When executing the crossover operator operation: two initial parameters to be optimized are selected from the population, and the excellent features of the parent string are inherited to the child string through the exchange combination of the two chromosomes, so as to generate new initial parameters to be optimized;

[0091] S233: When executing the mutation operator operation: changing certain gene values ​​in the initial parameters to be optimized to generate new initial parameters to be optimized.

[0092] Specifically, individuals are selected for reproduction based on fitness, which increases the probability of selecting initial parameters with high fitness, helps to retain parameters with high fitness, accelerates the optimization process, and improves search efficiency; the fitness of the initial parameters to be optimized is improved by executing the selection operator, and the number and diversity of the parameters to be optimized are increased by executing the crossover and mutation operators. During the cycle, the fitness of the parameters to be optimized is gradually improved, and the next generation population after the iteration is the optimized parameter, which is convenient for finally selecting the parameter with the maximum fitness as output.

[0093] Specifically, in step S3, the fitness function group is used to evaluate all optimization parameters to obtain the fitness value of each optimization parameter. In the process of executing the next round of iterations of the optimization parameters, when executing the selection operator operation, the fitness value is used as the optimization weight of the corresponding optimization parameter. The initial parameters to be optimized with high fitness have a higher probability of being selected. After multiple iterations, the fitness of the optimization parameters is gradually improved, so that after the iteration is completed, the individual with the largest fitness is selected as the optimal solution.

[0094] Specifically, the fitness value is obtained by adding the linear combination of the loss values ​​calculated by each loss function in the fitness function group.

[0095] It should be noted that the smaller the fitness value, the higher the fitness.

[0096] Specifically, step S4 includes the following steps:

[0097] S41: Determine whether the number of iterations meets the maximum evolutionary generation;

[0098] S42a: If the number of iterations does not satisfy the maximum evolutionary generation, continue to iterate;

[0099] S42b: If the maximum evolutionary generation is satisfied, the optimization parameter with the minimum fitness value is selected from all populations obtained in the iterative process as the best solution;

[0100] S43b: Decode the best solution to obtain target parameters, and use the target parameters to replace the initial parameters to be optimized to optimize the photoresist model.

[0101] Specifically, in step S13, the feasible solution is encoded and mapped from the solution space to the search space of the genetic algorithm, and in step S43b, the solution is decoded accordingly, and the target parameters are obtained after decoding.

[0102] In summary, when optimizing the parameters of the photoresist model, the genetic algorithm is used for iteration to solve the optimization parameters in the hope of finding the global optimal solution in the complex nonlinear problem; in the solution process, the fitness of the parameters to be optimized is calculated through the fitness function group, and when performing the selection operator operation, the fitness is used as an evaluation, and the parameters with high fitness have a higher probability of being selected. In the iterative process, the fitness of the parameters to be optimized is gradually improved, so that the optimization parameters with the maximum fitness can be finally selected as the best solution output; the best solution is decoded to obtain the target parameters, so as to achieve the optimization of the photoresist model.

[0103] The second embodiment of the present invention provides a computer program product 1, including a computer program 11, which, when executed by a processor, implements the photoresist model optimization method provided in the first embodiment.

[0104] The third embodiment of the present invention provides a computer device 2, including a storage 21, a processor 22 and a computer program 23 stored in the storage 21, and the processor 22 executes the computer program 23 to implement the steps of the photoresist model optimization method provided in the first embodiment.

[0105] The fourth embodiment of the present invention provides a computer-readable storage medium 3 on which a computer program 31 is stored. When the computer program 31 is executed by a processor, the steps of the method for solving the photoresist model parameters provided in the first embodiment are implemented.

[0106] It is understandable that, according to the embodiments disclosed in the present invention, the process described above with reference to the flowchart can be implemented as a computer software program. For example, the embodiments disclosed in the present invention include a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program includes a program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from the network through the communication part, and / or installed from a removable medium. When the computer program is executed by the central processing unit (CPU), the above functions defined in the method of the present application are executed. It should be noted that the computer-readable medium described in the present application can be a computer-readable signal medium or a computer-readable storage medium or any combination of the above two. Computer-readable storage media, for example, include but are not limited to electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices or devices, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to, an electrical connection with one or more conductors, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present application, a computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, an apparatus or a device. In the present application, a computer-readable signal medium may include a data signal propagated in a baseband or as part of a carrier wave, which carries a computer-readable program code. Such propagated data signals may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. A computer-readable signal medium may also be any computer-readable medium other than a computer-readable storage medium, which may send, propagate, or transmit a program for use by or in combination with an instruction execution system, an apparatus or a device. The program code contained on the computer-readable medium may be transmitted using any appropriate medium, including but not limited to: wireless, wire, optical cable, RF, etc., or any suitable combination of the above.

[0107] Computer program code for performing the operations of the present application may be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, C++, and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as an independent software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer via any type of network including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0108] In the embodiments provided by the present invention, it should be understood that "B corresponding to A" means that B is associated with A, and B can be determined according to A. However, it should also be understood that determining B according to A does not mean determining B only according to A, and B can also be determined according to A and / or other information.

[0109] It should be understood that the "one embodiment" or "an embodiment" mentioned throughout the specification means that the specific features, structures or characteristics related to the embodiment are included in at least one embodiment of the present invention. Therefore, "in one embodiment" or "in an embodiment" appearing throughout the specification does not necessarily refer to the same embodiment. In addition, these specific features, structures or characteristics can be combined in one or more embodiments in any suitable manner. Those skilled in the art should also be aware that the embodiments described in the specification are all optional embodiments, and the actions and modules involved are not necessarily required by the present invention.

[0110] In various embodiments of the present invention, it should be understood that the size of the serial numbers of the above-mentioned processes does not mean the necessary order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0111] The flow chart and block diagram in the accompanying drawings of the present invention illustrate the possible architecture, function and operation of the system, method and computer program product according to various embodiments of the present application. In this regard, each box in the flow chart or block diagram can represent a module, a program segment or a part of a code, and the module, the program segment or a part of the code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a sequence different from that marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, which is determined based on the functions involved. It should be particularly noted that each box in the block diagram and / or flow chart, and the combination of the boxes in the block diagram and / or flow chart can be implemented by a dedicated hardware-based system that performs a specified function or operation, or can be implemented by a combination of dedicated hardware and computer instructions.

[0112] Compared with the prior art, the photoresist model optimization method, program product, device and storage medium provided by the present invention have the following beneficial effects:

[0113] 1. The present invention provides a photoresist model optimization method, which more accurately calculates the fitness of initial parameters by introducing a fitness function group, and obtains better target parameters according to the fitness. Through the comprehensive use of a genetic algorithm and a specific fitness function group, not only can the global optimal solution or parameters close to the optimal solution be found, but also the model can be ensured to have consistent performance under different data disturbances, thus avoiding overfitting and improving the generalization ability of the model. Specifically, the genetic algorithm can find the global optimal solution in complex, high-dimensional nonlinear problems, and does not require the target function to be differentiable. After the fitness function group is introduced, when the initial parameter fitness calculated based on the fitness function group is used in the genetic algorithm, by using the fitness value as the optimization weight in the genetic algorithm, the overfitting phenomenon can be better avoided, and the consistency performance is better, thereby improving the accuracy of the target parameters finally outputted. The optimized parameters can accelerate the optimization process, improve the calculation efficiency, and shorten the optimization time.

[0114] 2. The present invention provides a photoresist model optimization method, which allows the error loss function to quantify the deviation between the predicted result and the actual result. The loss error obtained by calculating the line width in the predicted result and the line width loss on the actual photoresist pattern helps to evaluate the accuracy of the model prediction and gives a lower loss value within the allowable error range to reduce overfitting.

[0115] 3. The present invention provides a photoresist model optimization method, in which the grid loss function evaluates the stability of the model under data disturbance and improves the consistency of the model by calculating the average error of the offset points.

[0116] 4. The present invention provides a photoresist model optimization method, in which the coefficient loss function is used to describe the quality of the coefficient combination of each item in the model, and quantifies the importance of each coefficient of the model. By limiting the size of the coefficient in the model, the impact on the mask image can be minimized to avoid the occurrence of overfitting.

[0117] 5. The present invention provides a photoresist model optimization method, which optimizes various parameters in the initial photoresist model to improve the accuracy of the photoresist model and better describe the photoresist development process. The weight parameters can reflect the relative importance of each model formula item. By adjusting the relative importance of different model items, it helps to balance the contribution between model items and improve model performance.

[0118] 6. The present invention provides a photoresist model optimization method, which sets genetic algorithm parameters to control the operation of the genetic algorithm. According to the characteristics and requirements of this scheme, the parameters are reasonably set to obtain the best optimization effect; the random generation of the initial population can provide a diverse starting point for the genetic algorithm; through operations such as selection, crossover and mutation, the fitness of the population is gradually improved, which can accelerate the optimization process and improve the search efficiency.

[0119] 7. The present invention provides a photoresist model optimization method, which uses a genetic algorithm to select individuals for reproduction according to fitness, thereby increasing the probability of selecting initial parameters with high fitness, helping to retain parameters with high fitness, accelerating the optimization process, and improving search efficiency; and increasing the number and diversity of initial parameters through crossover and mutation, and gradually improving the fitness of the selected individuals during the cycle, making it easier to select the individual with the maximum fitness for output.

[0120] 8. A computer program product provided by the present invention has the same beneficial effects as the above-mentioned photoresist model optimization method, which will not be described in detail here.

[0121] 9. The computer device provided by the present invention has the same beneficial effects as the above-mentioned photoresist model optimization method, which will not be described in detail here.

[0122] 10. The computer-readable storage medium provided by the present invention has the same beneficial effects as the above-mentioned photoresist model optimization method, which will not be described in detail here.

[0123] The above is a detailed introduction to a photoresist model optimization method, program product, device and storage medium disclosed in an embodiment of the present invention. Specific examples are used in this article to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only used to help understand the method of the present invention and its core idea; at the same time, for a person skilled in the art, according to the idea of ​​the present invention, there will be changes in the specific implementation method and application scope. In summary, the content of this specification should not be understood as a limitation on the present invention. Any modifications, equivalent substitutions and improvements made within the principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A photoresist model optimization method, characterized in that: include: Providing an initial photoresist model, and obtaining a plurality of initial parameters to be optimized of the initial photoresist model; Providing a genetic algorithm model to iteratively solve all the initial parameters to be optimized to obtain optimized parameters; Providing a fitness function group, the fitness function group including a minimization error function and at least one of an allowable error loss function, a grid loss function and a coefficient loss function; in an iterative process, using the fitness function group to evaluate all the optimization parameters to obtain a fitness value of each of the optimization parameters, and when executing the next round of iteration of the optimization parameters, using the fitness value as the optimization weight corresponding to the optimization parameter; When the preset conditions are met, the iteration is stopped and the target parameters finally obtained by the genetic algorithm model are output; The target parameters are used to replace the initial parameters to be optimized to optimize the photoresist model.

2. A photoresist model optimization method as claimed in claim 1, characterized in that: The fitness value includes an allowable error loss value, and the allowable error loss value is calculated by calculating the deviation between the line width of the pattern in the prediction result of the photoresist model and the line width on the corresponding actual photoresist pattern based on the allowable error loss function to obtain the fitness value.

3. A photoresist model optimization method as claimed in claim 1, characterized in that: The fitness value includes a grid loss value, and the grid loss value is calculated as follows: a measurement point is determined in a prediction result obtained by calculating a photoresist model and an offset point is obtained after the measurement point is jittered. The grid loss function obtains the measurement point and the line width corresponding to the offset point, and calculates the degree of deviation between the predicted line width of the offset point and the predicted line width of the measurement point to obtain the fitness value.

4. A photoresist model optimization method as claimed in claim 1, characterized in that: The fitness value includes a coefficient loss value, and the coefficient loss value is calculated as follows: the coefficient loss function obtains the fitness value by calculating the average square sum of the coefficients of each of the initial parameters to be optimized.

5. A photoresist model optimization method as claimed in claim 1, characterized in that: The initial photoresist model includes at least one model formula item, and the initial parameters to be optimized include parameters in the model formula item and weight parameters of each of the model formula items, and the weight parameters are used to adjust the relative importance of different model formula items.

6. A photoresist model optimization method as claimed in claim 1, characterized in that: Providing a genetic algorithm model to iteratively solve all the initial parameters to be optimized to obtain the optimized parameters includes: Set genetic algorithm parameters; Obtaining a number of the initial parameters to be optimized as an initial population; Based on the genetic algorithm parameters, the initial population is iteratively operated using three operators, namely, selection, crossover and mutation, to obtain a next generation population, and the next generation population after the iteration is the optimization parameter.

7. A photoresist model optimization method as claimed in claim 6, characterized in that: Based on the genetic algorithm parameters, the initial population is iterated by using the three operators of selection, crossover and mutation to obtain the next generation population. The next generation population after the iteration is the optimization parameter including: When executing the selection operator operation: selecting a number of the initial parameters to be optimized from the population, the initial parameters to be optimized with high fitness have a higher probability of being selected; When performing the crossover operator operation: two initial parameters to be optimized are selected from the population, and the excellent features of the parent string are inherited to the child string through the exchange combination of the two chromosomes, so as to generate new initial parameters to be optimized; When executing the mutation operator operation: some gene values ​​in the initial parameters to be optimized are changed to generate new initial parameters to be optimized.

8. A computer program product, characterized in that: It comprises a computer program, and when the computer program is executed by a processor, it implements the photoresist model optimization method as described in any one of claims 1 to 7.

9. A computer device, characterized in that: The invention comprises a storage, a processor and a computer program stored in the storage, wherein the processor executes the computer program to implement the steps of the photoresist model optimization method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method for solving photoresist model parameters as claimed in any one of claims 1 to 7 are implemented.

Images