A preparation method, system and photoresist for optimizing photoresist stability

Abstract

The present invention discloses a preparation method, system, and photoresist for optimizing photoresist stability, relating to the technical field of photoresists. The method comprises: using the identity characteristics of a target production line as a constraint, collecting photoresist preparation logs containing raw material ratios, reaction conditions, and stability indicators; constructing a stability evaluation model based on the logs, the model consisting of cascaded weak learners corresponding to multiple preparation steps; combining the evaluation model with a global optimization algorithm, iteratively searching for preparation parameters, outputting the model prediction results as the objective function, and using the preparation parameters as input; outputting parameters that meet the stability threshold as standard preparation parameters, and executing photoresist preparation accordingly. This achieves the technical effects of improving optimization efficiency, reducing optimization costs, and improving optimization accuracy.

Description

Technical Field

[0001] The present invention relates to the technical field of photoresists, and in particular to a preparation method and system for optimizing the stability of photoresists, and photoresists. Background Art

[0002] As integrated circuit manufacturing processes evolve toward smaller linewidths and higher integration densities, higher requirements are placed on the performance of photoresists, particularly their stability, which is directly related to pattern transfer accuracy, process window width, and production yield. Photoresist stability is influenced by numerous factors, including preparation parameters such as raw material ratio, reaction temperature, stirring speed, and reaction time, and these parameters exhibit complex nonlinear coupling relationships.

[0003] Currently, photoresist preparation processes rely heavily on experience and manual parameter adjustment for optimization. In actual production, engineers typically adjust preparation parameters through trial-and-error experiments and single-factor analysis to achieve optimal stability. However, these methods still have shortcomings in the following areas: Traditional methods often rely on local adjustments, making it difficult to achieve a global optimal search across multiple parameter spaces; some models fail to fully incorporate production line characteristics, resulting in poor transferability between different production lines and low prediction accuracy; and some methods fail to clearly translate model predictions into actionable preparation parameters, making them difficult to guide actual production. Summary of the Invention

[0004] The present invention provides a preparation method, system and photoresist for optimizing the stability of photoresist, so as to solve the technical problems of low optimization efficiency, high cost and insufficient optimization precision in the prior art, and achieve the technical effects of improving optimization efficiency, reducing optimization cost and improving optimization precision.

[0005] In a first aspect, the present invention provides a preparation method for optimizing the stability of a photoresist, wherein the method comprises:

[0006] With the production line identity characteristics of the target production line as a constraint, the preparation log of the photoresist is collected, wherein the preparation log includes raw material ratios, reaction condition parameters and corresponding stability index data.

[0007] Based on the preparation log, a stability evaluation model is constructed, wherein the stability evaluation model includes multiple groups of cascaded weak learners, and the multiple groups of weak learners correspond to multiple preparation links of the target production line.

[0008] The stability assessment model is combined with a global optimization algorithm to perform an iterative search on the preparation parameters, wherein the prediction result of the stability assessment model is used as the output of an objective function, and the input of the objective function is the preparation parameters.

[0009] The preparation parameters whose prediction results meet a preset stability threshold are output as standard preparation parameters, and photoresist preparation is performed according to the standard preparation parameters.

[0010] In one feasible implementation, the photoresist preparation log is collected based on the production line identity characteristics of the target production line, including:

[0011] Obtain the production line identification information of the target production line and determine its corresponding process configuration characteristics and production line label characteristics.

[0012] Based on the production line label features, historical records of the photoresist preparation process are extracted from the target production line to obtain a first data set including raw material ratios, reaction condition parameters and stability index data.

[0013] When the data volume of the first data set is lower than a preset sample threshold, homologous production lines with associated labels are identified based on the production line label features, and corresponding preparation logs are extracted from the homologous production lines to construct a second data set.

[0014] The first data set and the second data set are combined as a data basis for modeling to construct a preparation log that meets training requirements.

[0015] In a feasible implementation, a stability assessment model is constructed based on the production log, wherein the stability assessment model includes multiple groups of cascaded weak learners, and the multiple groups of weak learners correspond to multiple production links of the target production line, including:

[0016] The collected prepared log data are processed by filling missing values, eliminating outliers, normalizing values ​​and encoding categorical variables to obtain a modeling data set for training.

[0017] According to the key link information of the photoresist preparation process, the modeling data set is segmented according to the process stages to obtain multiple sub-data sets.

[0018] Based on the plurality of sub-data sets, multiple groups of weak learners are trained respectively, and the groups of weak learners are cascaded according to the preparation process sequence in combination with an ensemble learning method to form the stability evaluation model.

[0019] In a feasible implementation, multiple groups of weak learners are trained based on the multiple sub-datasets, including:

[0020] Co-variance coefficients between the plurality of process stages and the stability index are evaluated, and a first set of weight coefficients is defined based on the co-variance coefficients.

[0021] Based on the sequence of the multiple process stages in the photoresist preparation process flow, a second set of weight coefficients is defined, wherein the second weight coefficients decrease in a direction along the photoresist preparation process flow.

[0022] The first weight coefficient set is modified according to the second weight coefficient set, and the modified first weight coefficient set is normalized to obtain a magnitude weight coefficient set.

[0023] In a feasible implementation, multiple groups of weak learners are trained based on the multiple sub-data sets, and the groups of weak learners are cascaded according to the preparation process sequence in combination with an ensemble learning method to form the stability assessment model, including:

[0024] An evaluation computing power indicator of a target scenario is obtained, the evaluation computing power indicator is allocated according to the magnitude weight coefficient set, and group capacity constraints of multiple groups of weak learners are defined.

[0025] Weak learners are added to the multiple groups of weak learners with the group capacity constraint as the filling upper limit.

[0026] According to the feature dimensions and data sample size contained in each sub-dataset, an adaptive sampling strategy is used to enhance or crop the data, and cross-validation and early stopping mechanisms are used to train each group of weak learners until the stability evaluation index converges on the validation set.

[0027] The multiple groups of trained weak learners are integrated, and the weights of the weak learners are dynamically fine-tuned according to the training error distribution to form the stability evaluation model.

[0028] In a feasible implementation, the stability assessment model is combined with a global optimization algorithm to perform an iterative search on the preparation parameters, wherein the prediction result of the stability assessment model is used as the output of an objective function, and the input of the objective function is the preparation parameters, including:

[0029] The preparation parameters are acquired in real time and input into the stability evaluation model.

[0030] The first group of weak learners of the stability evaluation model takes the preparation parameters as input, evaluates and outputs a first stability index.

[0031] The first stability index and the preparation parameters are input into a second group of weak learners to obtain a second stability index.

[0032] The first stability index, the second stability index and the preparation parameters are input into a third group of weak learners to obtain a third stability index, and so on until N groups of weak learners of the stability evaluation model are traversed to obtain an Nth stability index as the prediction result.

[0033] In a feasible implementation, the process of preparing the photoresist according to the standard preparation parameters may further include:

[0034] The photoresist preparation experiment was carried out according to the standard preparation parameters.

[0035] If the test results of the photoresist preparation of the first consecutive preset test batches meet the stability requirement, the photoresist preparation is performed based on the standard preparation parameters.

[0036] If the test results of the photoresist preparation of the second consecutive preset test batch do not meet the stability requirement, the stability evaluation model and the optimization algorithm are dynamically modified based on the experimental results.

[0037] In a second aspect, the present invention further provides a preparation system for optimizing the stability of a photoresist, wherein the system comprises:

[0038] The log collection module is used to collect the preparation log of the photoresist based on the production line identity characteristics of the target production line as a constraint, wherein the preparation log includes raw material ratios, reaction condition parameters and corresponding stability index data.

[0039] An evaluation model construction module is used to construct a stability evaluation model based on the preparation log, wherein the stability evaluation model includes multiple groups of cascaded weak learners, and the multiple groups of weak learners correspond to multiple preparation links of the target production line.

[0040] An optimization search module is used to combine the stability assessment model with a global optimization algorithm to iteratively search the preparation parameters, wherein the prediction result of the stability assessment model is used as the output of an objective function, and the input of the objective function is the preparation parameters.

[0041] The parameter output module is used to output the preparation parameters whose prediction results meet the preset stability threshold as standard preparation parameters, and prepare the photoresist according to the standard preparation parameters.

[0042] In a third aspect, the present invention further provides a photoresist, which is obtained by implementing any one of the above-mentioned preparation methods for optimizing the stability of the photoresist.

[0043] The present invention discloses a preparation method, system and photoresist for optimizing the stability of photoresist, comprising: collecting a preparation log of the photoresist with the production line identity characteristics of a target production line as a constraint, the preparation log covering raw material ratios, reaction condition parameters and corresponding stability index data; constructing a stability evaluation model based on the preparation log, the model consisting of multiple groups of cascaded weak learners, and each weak learner corresponds to multiple preparation links of the target production line; by combining the stability evaluation model with a global optimization algorithm, iteratively searching the preparation parameters, using the prediction results of the stability evaluation model as the output of the objective function, and the preparation parameters as the input; finally, determining the preparation parameters whose prediction results meet a preset stability threshold as standard preparation parameters, and carrying out photoresist preparation accordingly. The present invention solves the technical problems of low optimization efficiency, high cost and insufficient optimization accuracy, and achieves the technical effects of improving optimization efficiency, reducing optimization cost and improving optimization accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 The present invention is a schematic flow chart of a preparation method for optimizing the stability of a photoresist.

[0045] Figure 2 This is a schematic structural diagram of a preparation system for optimizing photoresist stability according to the present invention.

[0046] Description of the accompanying drawings: log collection module 11, evaluation model construction module 12, optimization search module 13, parameter output module 14. DETAILED DESCRIPTION

[0047] The above technical solution will be described in detail below in conjunction with the accompanying drawings and specific implementation methods of the specification to better understand the above technical solution. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments of the present invention. It should be understood that the present invention is not limited to the example embodiments used only to explain the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention. In addition, it should be noted that, for the convenience of description, only the parts related to the present invention, rather than all, are shown in the drawings.

[0048] Example 1, as Figure 1 The present invention is a schematic flow diagram of a preparation method for optimizing the stability of a photoresist, wherein the method comprises:

[0049] With the production line identity characteristics of the target production line as a constraint, the preparation log of the photoresist is collected, wherein the preparation log includes raw material ratios, reaction condition parameters and corresponding stability index data.

[0050] Specifically, the preparation log refers to a historical data set that records all relevant information in the photoresist preparation process, including at least the raw material ratio (the proportion of each component), reaction condition parameters (such as temperature, pressure, stirring intensity, etc.), and corresponding stability index data (such as the storage stability of the photoresist).

[0051] Specifically, by collecting and preparing logs based on the identity characteristics of the target production line, we can ensure the accuracy and consistency of the data and avoid model deviations caused by inconsistent data sources.

[0052] In some embodiments, collecting photoresist preparation logs based on the production line identity characteristics of the target production line includes:

[0053] Obtain the production line identification information of the target production line and determine its corresponding process configuration characteristics and production line label characteristics; based on the production line label characteristics, extract the historical records of the photoresist preparation process from the target production line to obtain a first data set including raw material ratios, reaction condition parameters and stability index data; when the data volume of the first data set is lower than a preset sample threshold, identify the homologous production line with the associated label based on the production line label characteristics, and extract the corresponding preparation log from the homologous production line to construct a second data set; merge the first data set and the second data set as the data basis for modeling to construct a preparation log that meets the training requirements.

[0054] Specifically, the production line identity features are the unique identifier of the target production line, including process configuration features (such as equipment type and process parameter range) and production line label features (such as the process type and equipment model of the production line). The sample threshold is a pre-set minimum number of samples to ensure that the data used for modeling is statistically representative and to avoid insufficient samples. Homologous production lines refer to other production lines with similar process configurations or label features as the target production line. They are used to supplement data when the target production line's own data is insufficient.

[0055] Specifically, first, through database query or data interface, the unique identifier of the target production line (such as ID number, area number) is obtained, and its corresponding process configuration characteristics (such as equipment type "photolithography machine type A") and production line label characteristics are determined. Then, based on the production line label characteristics, relevant data on the photoresist preparation process is extracted from the historical database of the target production line to form the first data set. For example, all photoresist preparation records from the past year are extracted, including raw material ratios (such as solvent A: reactant B: additive C = 45:40:15); reaction condition parameters (such as temperature 85°C, stirring 300 rpm, reaction time 6 hours); stability indicators (such as sedimentation rate after 30 days, percentage change in particle size, and photosensitivity decay rate).

[0056] Furthermore, if the number of records in the first data set is less than a set threshold (for example, less than 50 records), the homologous production line identification mechanism is activated, and the label matching algorithm is used to find production lines with other process label matching degrees higher than the set value (such as 90%). The corresponding photoresist preparation logs are extracted from these production lines, and the second data set is constructed. The first and second data sets are cleaned and standardized and then merged to finally form a preparation log that meets the modeling requirements.

[0057] The aforementioned method steps, by supplementing data from homologous production lines, can improve data integrity and diversity, thereby enhancing the model's generalization and prediction accuracy. Furthermore, this data collection method effectively incorporates production line characteristics, providing high-quality data support for subsequent model construction and parameter optimization, ultimately improving the stability and efficiency of photoresist preparation.

[0058] Based on the preparation log, a stability evaluation model is constructed, wherein the stability evaluation model includes multiple groups of cascaded weak learners, and the multiple groups of weak learners correspond to multiple preparation links of the target production line.

[0059] Specifically, the stability evaluation model is an analytical model based on machine learning, which is used to predict and evaluate the stability performance of photoresist under different preparation parameters. The stability evaluation model includes multiple cascaded groups of weak learners, and each group of weak learners focuses on evaluating a specific link in the photoresist preparation process; by cascading multiple groups of weak learners (such as decision trees, neural networks, etc.) in sequence, it is possible to consider the impact of upstream links on downstream links, which helps to improve the evaluation accuracy of the model.

[0060] In some embodiments, a stability assessment model is constructed based on the production log, wherein the stability assessment model includes multiple groups of cascaded weak learners, and the multiple groups of weak learners correspond to multiple production links of the target production line, including:

[0061] The collected preparation log data is processed by filling missing values, eliminating outliers, normalizing values, and encoding categorical variables to obtain a modeling data set for training; based on key link information of the photoresist preparation process, the modeling data set is segmented according to the process stage to obtain multiple sub-data sets; based on the multiple sub-data sets, multiple groups of weak learners are trained respectively, and combined with an ensemble learning method, each group of weak learners is cascaded according to the preparation process sequence to form the stability assessment model.

[0062] Specifically, the collected preparation log data is first processed for missing value filling, outlier removal, numerical normalization, and categorical variable encoding to ensure data quality. For example, temperature data is converted from Celsius to Kelvin, and equipment types are encoded as numerical labels. Then, based on the typical photoresist preparation process (e.g., raw material premixing → temperature-controlled polymerization → post-processing → inspection), the modeling dataset is split into several sub-datasets, each corresponding to a specific process stage. For example, the data is divided into the raw material mixing stage, the reaction stage, the stirring stage, etc.

[0063] Specifically, each sub-dataset is used as a type of training data, and multiple groups of weak learners are trained separately. Each group of weak learners focuses on assessing the stability of a specific preparation step. For example, the first group of weak learners assesses the stability of the raw material mixing stage, while the second group assesses the stability of the reaction stage. Preferably, the multiple weak learners included in each group are based on multiple machine learning models with different structures and parameters.

[0064] Furthermore, each group of weak learners is cascaded sequentially based on the process flow, with the output of the previous group serving as partial input to the next group's model. This allows for information transfer and forms a complete stability assessment model. For example, the groups of weak learners corresponding to steps such as raw material mixing, reaction, and stirring are sequentially connected, ultimately outputting an overall stability assessment result for the photoresist. Each group of weak learners outputs its final assessment result through an ensemble learning approach.

[0065] By constructing multiple cascaded weak learners, the aforementioned method fully considers the order and extent of the impact of different preparation steps on photoresist stability, ensuring that the most impactful / front-end steps are evaluated first, thereby improving the model's prediction accuracy. Furthermore, by decomposing a complex problem into multiple subproblems, the complexity of individual learners is reduced, improving the model's training efficiency and generalization capabilities.

[0066] In some implementations, the steps of training multiple groups of weak learners based on the multiple sub-datasets include:

[0067] Evaluate the co-variation coefficients between multiple process stages and stability indicators, and define a first weight coefficient set based on the co-variation coefficients; define a second weight coefficient set based on the sequence of multiple process stages in the photoresist preparation process flow, wherein the second weight coefficients decrease in the direction of the photoresist preparation process flow; correct the first weight coefficient set based on the second weight coefficient set, and normalize the corrected first weight coefficient set to obtain a magnitude weight coefficient set.

[0068] Specifically, the coefficient of variation is a measure of the degree of coordinated change between two variables, and is used to measure the statistical correlation between the process stage characteristics and the finished product stability indicators. The larger the coefficient of variation, the greater the impact of the corresponding process stage on the stability of the photoresist.

[0069] Specifically, the first set of weight coefficients is calculated by the coefficient of covariance between each process stage and the target stability index, which is used to quantitatively reflect the weight of the direct impact of different stages on the stability index; the second set of weight coefficients is an attenuated weight set defined according to the sequential relationship of each stage in the photoresist preparation process flow, which represents the structural importance of the process stage in the overall process, among which the process stage at the upstream in the process flow has a higher weight.

[0070] Specifically, the magnitude weight coefficient set is the final weight set obtained by fusing the above two weight factors and normalizing them, which is used to adjust the contribution ratio of each weak learner in the integrated model.

[0071] Specifically, the process-stability correlation is evaluated to form the first set of weight coefficients. First, a statistical analysis is performed on the key process parameters (such as temperature, rotation speed, time, concentration) and target stability indicators (such as sedimentation rate, particle size change, and light transmittance) of each process stage in the modeling data set; for example, the covariance coefficient between each parameter and the target indicator is calculated using methods such as Pearson covariance or Spearman; then, the covariance coefficient between the parameters and the target indicator of each stage is weighted averaged to obtain the first set of weight coefficients representing the contribution of each stage.

[0072] Specifically, a second weight coefficient is assigned to each stage according to the process sequence of the photoresist preparation process. This second weight coefficient is a decreasing weight, for example, linearly decreasing from 1.0 at the front end of the process to 0.4 at the end, thereby giving higher attention to the process stages at the front. The first weight coefficient set is then modified with the second weight coefficient set to comprehensively consider the process sequence and direct impact. The modified weight coefficients are then normalized to ensure that the sum of the weights is 1, resulting in the final set of magnitude weight coefficients used to adjust the contribution of the weak learner. The modified weights can better emphasize the influence of the early process stages while retaining sensitivity to stability indicators.

[0073] The above-mentioned method steps can more accurately reflect the real impact of each process stage on the stability of the photoresist. Among them, the first set of weight coefficients is used to ensure the sensitivity of the model to key process parameters, and the second set of weight coefficients introduces the sequential nature of the process flow, so that the model can dynamically adjust the degree of attention to different stages. The weak learner training guided by the obtained magnitude weight coefficient set helps to ensure that the model is more scientific and efficient when evaluating the stability of the photoresist, thereby improving the accuracy and efficiency of the preparation parameter optimization.

[0074] In some implementations, multiple groups of weak learners are trained based on the multiple sub-datasets, and the groups of weak learners are cascaded according to the preparation process sequence in combination with an ensemble learning method to form the stability assessment model, including:

[0075] Obtain an evaluation computing power indicator of the target scenario, allocate the evaluation computing power indicator according to the magnitude weight coefficient set, and define group capacity constraints for multiple groups of weak learners; use the group capacity constraint as the filling upper limit to add weak learners to the multiple groups of weak learners; based on the feature dimensions and data sample size contained in each sub-data set, adopt an adaptive sampling strategy to enhance or crop the data, and use cross-validation and early stopping mechanisms to train each group of weak learners until the stability evaluation indicator convergence requirements are met on the validation set; integrate the multiple groups of weak learners that have been trained, and dynamically fine-tune the weights of the weak learners according to the training error distribution to form the stability evaluation model.

[0076] Specifically, the evaluation computing power metric is a scalar quantity, defined as the computing resource capacity allocated to the training and inference of the stability assessment model, determined by factors such as the number of CPU cores, GPU memory, and model execution frame rate or response time. Based on this evaluation computing power metric and a set of magnitude weight coefficients, a maximum limit on the number of weak learners in each group (i.e., a group capacity constraint) can be defined to ensure resource feasibility of the training process in the target scenario.

[0077] Specifically, the adaptive sampling strategy is a process of data augmentation (such as repetition, noise addition, and interpolation) or cropping (such as undersampling and cluster compression) of training samples based on the number of samples and feature dimensions of each sub-dataset, which is used to improve the generalization ability of the model; cross-validation is used to evaluate the generalization error of the model, and the early stopping mechanism is used to terminate training when the accuracy of the validation set reaches the convergence condition to prevent overfitting.

[0078] Specifically, first, the computing power evaluation indicators for the target production line deployment scenario are read or calculated, such as the computing power under FP16, and the overall computing power indicators are proportionally distributed to each group of weak learners based on the magnitude weight coefficient set generated above. Next, within each group of weak learners, weak learners are gradually added, with the group capacity constraint as the upper limit, until the group capacity upper limit is reached. Then, for each sub-dataset, the sample distribution is dynamically adjusted based on its sample size and feature dimension. At the same time, when there are too few samples, enhancement strategies such as SMOTE synthesis and local Gaussian noise perturbation are adopted. When there are too many samples, K-means clustering representative sampling is performed. Then, cross-validation and early stopping mechanisms are used to train each group of weak learners to ensure that the model meets the convergence requirements of the stability evaluation indicators on the validation set.

[0079] Furthermore, multiple sets of trained weak learners are integrated, and the weights of the weak learners are dynamically fine-tuned based on the training error distribution to further optimize model performance and obtain a complete stability assessment model. For example, if a weak learner has a large error on a specific dataset, its weight is appropriately reduced to reduce its impact on the overall model.

[0080] The above process enables efficient utilization of limited computing resources, ensuring sufficient training support for weak learners in key process stages. Group capacity constraints and adaptive sampling strategies improve data utilization efficiency, while cross-validation and early stopping prevent overfitting. Dynamic fine-tuning further optimizes model performance, enabling the stability assessment model to more accurately predict photoresist stability. This provides reliable evaluation support for subsequent parameter optimization, helping to improve the stability and efficiency of photoresist preparation.

[0081] The stability assessment model is combined with a global optimization algorithm to perform an iterative search on the preparation parameters, wherein the prediction result of the stability assessment model is used as the output of an objective function, and the input of the objective function is the preparation parameters.

[0082] Specifically, the stability evaluation model is used as the objective function or evaluation function of the global optimization algorithm to perform iterative global optimization of the preparation parameters, wherein the input of the objective function is the preparation parameters and the output is the prediction result of the stability evaluation model.

[0083] For example, a set of preparation parameters is first initialized as an initial population based on the parameter range of the photoresist preparation process and the existing preparation parameters acquired in real time. Each parameter combination in the initial population is then input into the stability evaluation model to obtain a predicted stability index. For example, the storage stability of the photoresist under a certain parameter combination is 92% (scalar). Next, based on the predicted results (the predicted stability index), the best performing parameter combinations are selected as new populations. New parameter combinations are generated through crossover operations. For example, a parameter combination with high stability is crossed with a parameter combination with medium stability to generate new parameter combinations. These new parameter combinations are then mutated, introducing randomness to explore more possibilities. Furthermore, the evaluation, selection, crossover, and mutation processes are repeated to gradually optimize the parameter combinations until a preset number of iterations is reached or the stability index converges.

[0084] The above steps, combined with a stability assessment model and a global optimization algorithm, enable efficient search for optimal preparation parameters in a multi-parameter space. The stability assessment model provides accurate prediction support, guiding the optimization algorithm to quickly converge to a region of highly stable parameters. This not only improves optimization efficiency but also helps reduce trial-and-error costs.

[0085] In some embodiments, the stability assessment model is combined with a global optimization algorithm to perform an iterative search on the preparation parameters, wherein the prediction result of the stability assessment model is used as the output of an objective function whose input is the preparation parameters, including:

[0086] Acquire preparation parameters in real time and input the preparation parameters into the stability assessment model; a first group of weak learners of the stability assessment model uses the preparation parameters as input, evaluates and outputs a first stability index; the first stability index and the preparation parameters are input into a second group of weak learners to obtain a second stability index; the first stability index, the second stability index and the preparation parameters are input into a third group of weak learners to obtain a third stability index, and so on, until N groups of weak learners of the stability assessment model are traversed, and the Nth stability index is obtained as the prediction result.

[0087] Optionally, global optimization algorithms include particle swarm optimization (PSO), genetic algorithm (GA) or Bayesian optimization, which are used to search for the preparation parameter combination that optimizes the objective function in a complex multivariable space; the stability index output by the stability assessment model is used as a performance evaluation function of the optimization algorithm to drive the parameter search direction.

[0088] Specifically, first, obtain the preparation parameter X to be evaluated in the current round, and input the parameter X into the first group of weak learners of the stability evaluation model to obtain the first stability index y1. For example, the first group of weak learners focuses on the raw material mixing stage, and the output stability index is the mixing uniformity score; then, {X, y1} is used as new input data and input into the second group of weak learners to output the second stability index y2. For example, the second group of weak learners combines the mixing uniformity score and the reaction temperature to output the reaction stability score; then, {X, y1, y2} is input into the third group of weak learners to obtain the third stability index y3, and so on, until all N groups of weak learners are traversed, and finally the Nth stability index is obtained as the prediction result of the model to evaluate the pros and cons of the current preparation parameter combination.

[0089] This approach, through the gradual assessment of stability through the cascade of multiple weak learners, fully leverages information from each process stage. Each weak learner is specific to a single manufacturing step, reducing model complexity for that step and improving training efficiency. The cascaded structure enables the model to dynamically integrate stability information from different stages, ultimately resulting in more reliable predictions.

[0090] The preparation parameters whose prediction results meet a preset stability threshold are output as standard preparation parameters, and photoresist preparation is performed according to the standard preparation parameters.

[0091] Specifically, the standard preparation parameters are a set of preparation process parameters whose corresponding stability prediction results meet the preset threshold requirements, which are determined jointly by the stability evaluation model and the global optimization algorithm and are used for photoresist preparation; the preset stability threshold refers to the minimum acceptable stability index set for the target performance index (such as solubility, stability, particle size distribution, etc.) during the photoresist preparation process, and is used to screen qualified parameter combinations.

[0092] Specifically, when the final stability index predicted by the model is not lower than the stability threshold, it means that the current parameter combination can meet the actual production requirements for photoresist quality. Then this set of parameters will be used as standard preparation parameters, and the above standard preparation parameters will be sent to the photoresist production system to automatically control the reactor, feeding device, temperature control system and other execution equipment to carry out actual photoresist preparation.

[0093] Preferably, from multiple parameter combinations that meet the stability requirements, the parameter group with the lowest model prediction error or the best secondary index performance (such as minimum energy consumption, shortest reaction time, etc.) is selected as the final standard preparation parameter.

[0094] In some embodiments, the process of preparing the photoresist according to the standard preparation parameters may further include:

[0095] A photoresist preparation test is performed according to the standard preparation parameters; if the photoresist preparation test results of the first consecutive preset test batch meet the stability requirements, photoresist preparation is performed based on the standard preparation parameters; if the photoresist preparation test results of the second consecutive preset test batch do not meet the stability requirements, the stability evaluation model and optimization algorithm are dynamically corrected based on the experimental results.

[0096] Specifically, the first preset test batch is a threshold for the number of consecutive batches set to evaluate the recommendation effect of the model. If the stability indicators of the first preset test batches meet the requirements continuously, the standard preparation parameters can be considered to be executable; the second preset test batch is a threshold for the number of consecutive failed batches set to determine the failure of the model prediction. If it is not met continuously, it can be considered that the performance of the standard preparation parameters in actual production is inconsistent with the results predicted by the model, and the model needs to be corrected and learned.

[0097] Specifically, based on the standard fabrication parameters output by the model, small-scale fabrication experiments (i.e., photoresist fabrication tests) are conducted under a controlled environment. Corresponding performance indicators, such as thermal stability, dispersibility, and shelf life, are collected to verify the feasibility and stability of the recommended parameter combination under actual process conditions. If the test results of the first predetermined batch (e.g., three batches) meet the set stability requirement threshold (e.g., deviation ≤5%), the standard fabrication parameters are deemed viable and the batch fabrication process is initiated.

[0098] Specifically, if the test results of the second consecutive preset batch (such as 2 batches) do not meet the stability requirements, it means that the current model prediction error is large or there is a systematic deviation. The test data of the failed batch will be fed back to the model training module, and the weights and structure of the weak learner will be updated based on the actual performance, or the objective function constraints will be readjusted to improve the accuracy of the next round of prediction and optimization.

[0099] Through the aforementioned test verification steps, the validity and reliability of standard preparation parameters can be confirmed before formal production, reducing the risk of mass production. This ensures that only fully verified parameters are used in actual production, thereby improving the stability and consistency of the photoresist. In addition, a dynamic correction mechanism enables the model and algorithm to be continuously optimized based on actual test results, enhancing the adaptability and robustness of the overall optimization.

[0100] In summary, the preparation method for optimizing the stability of photoresist provided by the present invention has the following technical effects:

[0101] By using the production line identity characteristics of the target production line as a constraint, the preparation log of the photoresist is collected, and the preparation log covers the raw material ratio, reaction condition parameters and corresponding stability index data; based on the preparation log, a stability assessment model is constructed, and the model consists of multiple groups of cascaded weak learners, and each weak learner corresponds to multiple preparation links of the target production line; by combining the stability assessment model with the global optimization algorithm, the preparation parameters are iteratively searched, and the prediction results of the stability assessment model are used as the output of the objective function, and the preparation parameters are used as the input; finally, the preparation parameters whose prediction results meet the preset stability threshold are determined as standard preparation parameters, and photoresist preparation is carried out accordingly, thereby achieving the technical effect of improving optimization efficiency, reducing optimization cost and improving optimization accuracy.

[0102] Example 2, as Figure 2 Schematic diagram of a system for preparing photoresist for optimizing stability according to the present invention. Figure 1 The schematic flow chart of a preparation method for optimizing the stability of a photoresist of the present invention can be shown as follows: Figure 2 The structure shown is implemented.

[0103] Based on the same concept as the preparation method for optimizing the stability of the photoresist in the embodiment, the present invention also provides a preparation system for optimizing the stability of the photoresist, comprising:

[0104] The log collection module 11 is used to collect the photoresist preparation log based on the production line identity characteristics of the target production line, wherein the preparation log includes raw material ratios, reaction condition parameters and corresponding stability index data.

[0105] The evaluation model construction module 12 is used to construct a stability evaluation model based on the preparation log, wherein the stability evaluation model includes multiple groups of cascaded weak learners, and the multiple groups of weak learners correspond to multiple preparation links of the target production line.

[0106] The optimization search module 13 is used to combine the stability evaluation model with the global optimization algorithm to iteratively search the preparation parameters, wherein the prediction result of the stability evaluation model is used as the output of the objective function, and the input of the objective function is the preparation parameters.

[0107] The parameter output module 14 is configured to output the preparation parameters whose prediction results satisfy a preset stability threshold as standard preparation parameters, and perform photoresist preparation according to the standard preparation parameters.

[0108] In some embodiments, the log collection module 11 includes:

[0109] The production line identification information acquisition unit is used to obtain the production line identification information of the target production line and determine its corresponding process configuration characteristics and production line label characteristics.

[0110] The first data set acquisition unit is used to extract historical records of the photoresist preparation process from the target production line based on the production line label characteristics, and acquire a first data set including raw material ratios, reaction condition parameters and stability index data.

[0111] The second data set construction unit is used to identify the homologous production line with the associated label based on the production line label feature when the data volume of the first data set is lower than the preset sample threshold, and extract the corresponding preparation log from the homologous production line to construct the second data set.

[0112] The preparation log integration unit is used to merge the first data set and the second data set as the data basis for modeling, and construct a preparation log that meets the training requirements.

[0113] In some embodiments, the assessment model building module 12 includes:

[0114] The data preprocessing unit is used to perform missing value filling, outlier removal, numerical normalization and categorical variable encoding processing on the collected prepared log data to obtain a modeling data set for training.

[0115] The sub-dataset segmentation unit is used to segment the modeling dataset according to the process stages according to the key link information of the photoresist preparation process to obtain multiple sub-datasets.

[0116] Multiple weak learner training units are used to train multiple groups of weak learners based on multiple sub-data sets, and combine the ensemble learning method to cascade each group of weak learners in the order of the preparation process to form the stability evaluation model.

[0117] In some implementations, the multiple groups of weak learner training units in the evaluation model building module 12 include:

[0118] The co-variation coefficient evaluation unit is used to evaluate the co-variation coefficients between multiple process stages and stability indicators, and define a first weight coefficient set according to the co-variation coefficients.

[0119] The second weight coefficient set definition unit is used to define a second weight coefficient set based on the sequence of multiple process stages in the photoresist preparation process flow, wherein the second weight coefficient decreases along the direction of the photoresist preparation process flow.

[0120] The magnitude weight coefficient set acquisition unit is configured to modify the first weight coefficient set according to the second weight coefficient set, and normalize the modified first weight coefficient set to obtain a magnitude weight coefficient set.

[0121] In some implementations, the multiple groups of weak learner training units in the evaluation model building module 12 further include:

[0122] An evaluation computing power indicator allocation unit is used to obtain an evaluation computing power indicator of a target scenario, allocate the evaluation computing power indicator according to the magnitude weight coefficient set, and define group capacity constraints for multiple groups of weak learners.

[0123] The weak learner filling unit is used to add weak learners to multiple groups of weak learners with the group capacity constraint as the filling upper limit.

[0124] The weak learner training unit is used to enhance or crop the data using an adaptive sampling strategy based on the feature dimensions and data sample size contained in each sub-dataset, and to train each group of weak learners using cross-validation and early stopping mechanisms until the stability evaluation index converges on the validation set.

[0125] The stability evaluation model integration unit is used to integrate multiple groups of weak learners that have been trained and dynamically fine-tune the weights of the weak learners according to the training error distribution to form the stability evaluation model.

[0126] In some embodiments, the optimized search module 13 includes:

[0127] The preparation parameter real-time acquisition unit is used to acquire the preparation parameters in real time and input the preparation parameters into the stability evaluation model.

[0128] A stability index evaluation unit is used for a first group of weak learners of the stability evaluation model to evaluate and output a first stability index using the preparation parameters as input.

[0129] The multi-level stability index transmission unit is configured to input the first stability index and the preparation parameters into a second group of weak learners to obtain a second stability index. The first stability index, the second stability index, and the preparation parameters are input into a third group of weak learners to obtain a third stability index. The multi-level stability index transmission unit is configured to input the first stability index, the second stability index, and the preparation parameters into a third group of weak learners to obtain a third stability index. The multi-level stability index transmission unit is configured to input the first stability index ...

[0130] In some embodiments, the parameter output module 14 includes:

[0131] The photoresist preparation test unit is used to perform a photoresist preparation test according to the standard preparation parameters.

[0132] The stability requirement satisfaction determination unit is configured to perform photoresist preparation based on the standard preparation parameters if the test results of the photoresist preparation of the first consecutive preset test batches meet the stability requirement.

[0133] The model dynamic correction unit is used to dynamically correct the stability evaluation model and the optimization algorithm based on the experimental results if the test results of the photoresist preparation of the second consecutive preset test batch do not meet the stability requirement.

[0134] Example 3: Based on the same concept as a preparation method for optimizing the stability of a photoresist in the above-mentioned embodiment, the present application also provides a photoresist, which can be prepared by any of the methods described in the above-mentioned embodiments.

[0135] It should be understood that the embodiments mentioned in this specification focus on their differences from other embodiments. The specific embodiments in the aforementioned embodiment one are also applicable to a preparation system for optimizing photoresist stability described in embodiment two. For the sake of brevity of the specification, no further elaboration will be given here.

[0136] It should be understood that the embodiments disclosed in the present invention and the above description can enable those skilled in the art to use the present invention to implement the present invention. At the same time, the present invention is not limited to the embodiments mentioned above. It should be understood that those skilled in the art can still modify the technical solutions described in the above embodiments or replace some of the technical features therein with equivalents; and such modifications or replacements do not deviate from the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention and are all included in the scope of protection of the present invention.

Claims

1. A preparation method for optimizing the stability of a photoresist, characterized in that: The method comprises: Using the production line identity characteristics of the target production line as a constraint, collect photoresist preparation logs, where the preparation logs include raw material ratios, reaction condition parameters, and corresponding stability index data; Based on the production log, a stability assessment model is constructed, wherein the stability assessment model includes multiple groups of cascaded weak learners, and the multiple groups of weak learners correspond to multiple production links of the target production line, including: Filling missing values, removing outliers, normalizing values, and encoding categorical variables are performed on the collected prepared log data to obtain a modeling data set for training; According to the key link information of the photoresist preparation process, the modeling data set is segmented according to the process stage to obtain multiple sub-data sets; Obtaining an evaluation computing power indicator for the target scenario, allocating the evaluation computing power indicator based on a pre-built set of magnitude weight coefficients, and defining group capacity constraints for multiple groups of weak learners; adding weak learners to the plurality of groups of weak learners with the group capacity constraint as a fill upper limit; Based on the feature dimensions and data sample size contained in each sub-dataset, an adaptive sampling strategy is used to enhance or crop the data, and each group of weak learners is trained using cross-validation and early stopping mechanisms until the stability evaluation index converges on the validation set. Integrating multiple groups of trained weak learners and dynamically fine-tuning the weights of the weak learners according to the training error distribution to form the stability evaluation model; Combining the stability assessment model with a global optimization algorithm, an iterative search is performed on the preparation parameters, wherein the prediction result of the stability assessment model is used as the output of an objective function, and the input of the objective function is the preparation parameters; The preparation parameters whose prediction results meet a preset stability threshold are output as standard preparation parameters, and photoresist preparation is performed according to the standard preparation parameters.

2. A preparation method for optimizing the stability of a photoresist according to claim 1, characterized in that: Based on the production line identity characteristics of the target production line, collect photoresist preparation logs, including: Obtain the production line identification information of the target production line and determine its corresponding process configuration characteristics and production line label characteristics; Based on the production line label features, extracting historical records of the photoresist preparation process from the target production line to obtain a first data set including raw material ratios, reaction condition parameters, and stability index data; When the data volume of the first data set is lower than a preset sample threshold, identifying homologous production lines with associated labels based on the production line label features, and extracting corresponding preparation logs from the homologous production lines to construct a second data set; The first data set and the second data set are combined as a data basis for modeling to construct a preparation log that meets training requirements.

3. A preparation method for optimizing the stability of a photoresist according to claim 2, characterized in that: Combine the pre-built magnitude weight coefficient set to allocate the evaluation computing power indicator and define the group capacity constraints of multiple groups of weak learners, including: evaluating co-variance coefficients between a plurality of process stages and stability indicators, and defining a first set of weight coefficients based on the co-variance coefficients; Based on the order of the multiple process stages in the photoresist preparation process flow, a second set of weight coefficients is defined, wherein the second weight coefficients decrease in a direction along the photoresist preparation process flow; The first weight coefficient set is modified according to the second weight coefficient set, and the modified first weight coefficient set is normalized to obtain a magnitude weight coefficient set.

4. A preparation method for optimizing the stability of a photoresist according to claim 1, characterized in that: Combining the stability assessment model with a global optimization algorithm, an iterative search is performed on the preparation parameters, wherein the prediction result of the stability assessment model is used as the output of an objective function whose input is the preparation parameters, including: Acquiring preparation parameters in real time and inputting the preparation parameters into the stability assessment model; The first group of weak learners of the stability evaluation model takes the preparation parameters as input, evaluates and outputs a first stability index; Inputting the first stability index and the preparation parameters into a second group of weak learners to obtain a second stability index; The first stability index, the second stability index and the preparation parameters are input into a third group of weak learners to obtain a third stability index, and so on until N groups of weak learners of the stability evaluation model are traversed to obtain an Nth stability index as the prediction result.

5. A preparation method for optimizing the stability of a photoresist according to claim 1, characterized in that: The photoresist is prepared according to the standard preparation parameters, and before that, the method further includes: Performing a photoresist preparation test according to the standard preparation parameters; If the test results of the photoresist preparation of the first consecutive preset test batches meet the stability requirement, then the photoresist preparation is performed based on the standard preparation parameters; If the test results of the photoresist preparation of the second consecutive preset test batch do not meet the stability requirement, the stability evaluation model and the optimization algorithm are dynamically modified based on the test results.

6. A preparation system for optimizing the stability of photoresist, characterized in that: A preparation method for optimizing the stability of a photoresist according to any one of claims 1 to 5, comprising: A log collection module, configured to collect photoresist preparation logs based on the production line identity characteristics of the target production line, wherein the preparation logs include raw material ratios, reaction condition parameters, and corresponding stability index data; An evaluation model construction module, configured to construct a stability evaluation model based on the production log, wherein the stability evaluation model includes a plurality of cascaded weak learners corresponding to a plurality of production links of the target production line; an optimization search module, configured to perform an iterative search on the preparation parameters by combining the stability assessment model with a global optimization algorithm, wherein the prediction result of the stability assessment model is used as the output of an objective function whose input is the preparation parameters; The parameter output module is used to output the preparation parameters whose prediction results meet the preset stability threshold as standard preparation parameters, and prepare the photoresist according to the standard preparation parameters.

7. A photoresist, characterized in that The photoresist is obtained by a preparation method for optimizing the stability of the photoresist according to any one of claims 1 to 5.

Images