Analog method of blazed grating
By establishing predictive models and optimizing process parameters, the problem of low efficiency in the fabrication process of blazed gratings was solved, achieving high-precision and high-efficiency control of the grating structure and reducing costs and time.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- SHANGHAI NORTH OCEAN TECH CO LTD
- Filing Date
- 2024-12-18
- Publication Date
- 2026-06-19
AI Technical Summary
The existing technology for manufacturing blazed gratings is inefficient, requires multiple debugging and verification processes, consumes a lot of manpower and resources, and makes it difficult to achieve high-precision and efficient structural control.
By establishing a predictive model and using historical data sets for classification and regression analysis, the optimal predictive equation is determined, process parameters are optimized, and the grating structure can be obtained quickly, avoiding multiple process adjustments.
It improves the manufacturing precision and efficiency of blazed gratings, reduces process development time and labor costs, and optimizes process stability and reliability.
Smart Images
Figure CN122241949A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of semiconductor technology, and more particularly to a method for simulating a blazed grating. Background Technology
[0002] Blazed gratings play an important role in diffractive waveguide imaging. They can transfer light energy from the non-dispersive zero-order spectrum to other dispersive spectral orders, thereby concentrating the light energy to a desired diffraction order. This results in higher diffraction efficiency of the grating at specific wavelengths and orders, improving the efficiency of light utilization.
[0003] Blazed gratings, through their blaze angle design, can be adapted to a specific order of the spectrum within a particular wavelength band. This means that specific wavelengths of light can be selectively enhanced or suppressed as needed, which is extremely useful for applications requiring specific wavelengths of light. In AR glasses, diffractive waveguides combined with blazed gratings can achieve proportional image transfer and two-dimensional pupil expansion, enabling AR glasses to be thin, light, and have a wide field of view, making them suitable for a wider range of users and providing a good visual experience.
[0004] However, the transmission diffraction efficiency of blazed gratings is highly sensitive to changes in grating structural parameters, requiring precise control of the structural shape and blazing angle. The fabrication of blazed gratings involves multiple aspects, including high-precision process requirements, complex single-point technical breakthroughs, challenges in integration technology, crystal alignment and etching techniques, sensitive control of groove parameters, challenges in holographic ion beam etching technology, and the difficulty of fabricating convex gratings. Often, to achieve a better blazed structure, process parameters are continuously adjusted to obtain a stable structure, requiring significant manpower and resources, which is time-consuming and labor-intensive. Different processes are needed for different blazed structures, resulting in low efficiency. Therefore, how to improve the existing methods for fabricating blazed structures to increase process efficiency is an urgent problem for those skilled in the art. Summary of the Invention
[0005] This invention provides a simulation method for blazed gratings, which can optimize the structural and process parameters of blazed gratings, accurately describe the influence of different processes and parameters on the final etching results, and achieve high diffraction efficiency and selectivity for specific wavelengths. By deeply processing historical process data, a simulation prediction model is established to determine the influence of different process conditions on the final blazed structure, and this is used to determine the structure of the blazed grating under preset process conditions. This avoids multiple process debugging and verification. The simulation method is beneficial for quickly obtaining the grating structure under different process conditions.
[0006] A method for simulating a blazed grating includes the following steps:
[0007] (1) Obtain historical data sets of blazed grating etching under different etching process conditions using different equipment;
[0008] (2) Classify the acquired historical data sets, establish prediction models under different equipment and / or different etching process conditions, and determine the best prediction equation for each prediction model.
[0009] (3) Determine the process parameters of the corresponding blazed grating structure based on the simulated blazed grating structure parameters; and use these process parameters as new online data;
[0010] (4) Input the new online data from step (3) into the prediction model in step (2) and determine the accuracy of the prediction model under the optimal prediction equation;
[0011] When the accuracy rate is greater than 95%, proceed to step (5);
[0012] When the accuracy is less than 95%, the process parameters of the corresponding blazed grating structure are updated again, and the updated process parameters are used as updated online data and input into the prediction model in step (2). The accuracy of the prediction model under the optimal prediction equation is then determined. This process is repeated until the accuracy is greater than 95%, and then the process proceeds to step (5).
[0013] (5) The new online data is used as input, and the blazed grating is etched based on the new online data as process parameters.
[0014] Furthermore, in step (1), the historical data set includes the straight tooth mask height, period, aspect ratio, grating tilt angle, etching time, and blazed grating height of the blazed grating structure under different equipment and / or different process conditions.
[0015] Furthermore, in step (2), the acquired historical data group is classified in detail. This classification is based on etching data from different devices under the same process conditions, or etching data from the same device under different process conditions, or multiple etching data from the same device under the same process conditions, or etching data from different devices under different process conditions. The etching data under this classification is used as the historical data group for classification.
[0016] Furthermore, in step (2), the optimal prediction equation is determined based on the predictive power of the prediction model and the reliability during regression analysis.
[0017] Furthermore, the predictive capability of the prediction model is determined based on the blazed grating structure simulated by the prediction model and the parameters of the blazed grating structure to be simulated.
[0018] In some other embodiments, step (2) further includes iteratively optimizing the prediction model to update the prediction model under different etching process conditions.
[0019] In some embodiments, in step (3), the process parameters of the corresponding blazed grating structure are determined based on etching process experience using the simulated blazed grating structure parameters.
[0020] Further, in step (4), specifically, the process parameters determined in step (3) are used as new online data and input into the prediction model in step (2). Based on the prediction models corresponding to at least different equipment and / or different etching process conditions, the blazed grating structure obtained by each prediction model under the optimal prediction equation is compared with the simulated blazed grating structure parameters. When the error accuracy between the output result of any prediction model and the simulated blazed grating structure parameters is greater than 95%, then proceed to step (5). When the error accuracy between the output result of any prediction model and the designed structure parameters is less than 95%, the process parameters of the corresponding blazed grating structure are updated again, and the updated process parameters are used as updated online data and input into the prediction model in step (2). The accuracy of the prediction model under the optimal prediction equation is then judged. This process is repeated until the accuracy is greater than 95%, and then proceed to step (5).
[0021] Furthermore, the present invention also includes, in step (4), storing the prediction model corresponding to the updated process parameters in the corresponding prediction model database, so as to continuously update the prediction model database in real time.
[0022] Furthermore, in step (4), when calculating the accuracy, the blazed grating structure obtained based on the new online data as process parameters is compared with the simulated blazed grating structure parameters in step (3); when the difference between the two blazed structure parameters meets a certain threshold, it is determined that the process parameters meet the simulation requirements of the blazed grating; otherwise, step (3) is repeated until the difference between the two blazed structure parameters meets the threshold.
[0023] In the method provided by this invention, in step (1), regression analysis is performed on multiple different discrete historical data sets; and in step (2), a prediction model is established based on the historical data sets, and the best prediction equation for different prediction models is determined through regression analysis; of course, in step (4), the established prediction model can also be continuously optimized and iterated, and different prediction model databases can be updated to achieve the expected model prediction ability and credibility, and the best prediction equation is finally determined by analyzing, iterating, and re-regressing the prediction results.
[0024] This invention analyzes limited and discrete data and constructs a prediction model with strong predictive capabilities based on a suitable regression model.
[0025] On the one hand, the method of the present invention can establish the mapping relationship between parameters such as straight tooth mask height, period, duty cycle, tooth width, slot width, grating tilt angle, and etching time and blazing height and blazing angle, so that the previously single and discrete process results can be used to establish a systematic process prediction model database, and the prediction model database can be updated based on different process data, thereby reducing the barrier of semiconductor etching process technology.
[0026] On the other hand, the predictive equations obtained based on regression analysis can predict new design schemes in advance, obtain directional process results, and predict the structures etched by different etching processes in advance, providing a clear direction for actual process verification. This greatly reduces the time and labor costs of process development, reduces the difficulty of fabricating blazed gratings, and improves production efficiency, thereby saving enterprise production costs. It also helps to improve the fabrication accuracy of blazed gratings, reduce process errors caused by production equipment and process stability, and optimize process parameters to improve process stability and reliability. Attached Figure Description
[0027] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0028] Figure 1 This is a flowchart illustrating a method for simulating a blazed grating provided by the present invention. Detailed Implementation
[0029] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.
[0030] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.
[0031] This invention provides a simulation method for blazed gratings to solve or optimize existing blazed grating fabrication processes, determine the structure of blazed gratings under different preset process conditions, and determine the optimal etching process route, avoiding multiple process debugging and verification.
[0032] More specifically, a method for simulating a blazed grating includes the following steps:
[0033] (1) Obtain historical data sets of blazed grating etching under different etching process conditions using different equipment;
[0034] (2) Classify the acquired historical data sets, establish prediction models under different equipment and / or different etching process conditions, and determine the best prediction equation for each prediction model.
[0035] (3) Determine the process parameters of the corresponding blazed grating structure based on the simulated blazed grating structure parameters; and use these process parameters as new online data;
[0036] (4) Input the new online data into the prediction model in step (2) and determine the accuracy of the prediction model under the optimal prediction equation;
[0037] When the accuracy rate is greater than 95%, proceed to step (5);
[0038] When the accuracy is less than 95%, the process parameters of the corresponding blazed grating structure are updated again, and the updated process parameters are used as updated online data and input into the prediction model in step (2). The accuracy of the prediction model under the optimal prediction equation is then determined. This process is repeated until the accuracy is greater than 95%, and then the process proceeds to step (5).
[0039] (5) The new online data is used as input, and the blazed grating is etched based on the new online data as process parameters.
[0040] More specifically, in step (1), historical data sets of blazed grating etching for different equipment under different process conditions are obtained. In this step, based on different etching process equipment, there are different etching process routes or process parameters. In order to ensure the reliability of the blazed structure simulation, it is first necessary to determine the etching data of blazed gratings for different equipment under different process conditions, including the straight tooth mask height, period, aspect ratio, grating tilt angle, etching time, etc. of the blazed structure. The etching data of the blazed grating is then combined with different process equipment to form a historical data set under that condition. Several different historical data sets are then established by combining different process conditions.
[0041] In step (1), the present invention first collects a certain amount of previous discrete process data as a reference standard, and uses it as the basis for the simulation of the present invention. It makes full use of discrete data under different process conditions to characterize different process errors and etching results, deeply integrates the etching structure of the blazed grating with the discrete data, and establishes the correspondence between the process data and the blazed grating structure.
[0042] In step (2), the acquired historical data sets are classified, prediction models under different etching process conditions are established, and the optimal prediction equations for different prediction models are determined. In this step, in detail, the historical data sets formed by discrete data under different process conditions are classified, and more specifically, the acquired historical data sets are classified. This classification is based on etching data from different equipment under the same process conditions, or etching data from the same equipment under different process conditions, or multiple etching data from the same equipment under the same process conditions, or etching data from different equipment under different process conditions, and the etching data under this classification is used as the historical data set for classification.
[0043] Based on different historical data groups after classification, prediction models are established under different process conditions, including linear models and / or quadratic models, i.e., multiple mathematical models with different parameters. Regression analysis is then performed on each prediction model, and the optimal prediction equation is obtained based on the different regression analysis results. In this step, the regression analysis results are iteratively optimized and regressed. Based on the predictive ability of the prediction model and the confidence value of the regression analysis, the optimal prediction equation in the prediction model is determined. Among the optimal prediction equations, the prediction equation with some grating structure parameters can be selected based on the regression analysis results and confidence value.
[0044] Of course, the predictive capability of the prediction model is determined based on the blazed grating structure simulated by the model and the parameters of the simulated blazed grating structure. Specifically, the predictive capability of the model is judged by comparing the blazed grating structure formed in the prediction model based on the etching process parameters with the parameters of the simulated blazed grating structure. The simulated blazed grating structure parameters are the blazed grating structure to be etched and formed by the designer according to different blazed grating requirements, or the standard blazed grating structure parameters.
[0045] The grating structure parameters in the optimal prediction equation include linear equations and / or quadratic equations, etc., determined based on parameters such as straight-tooth mask height, and / or period, and / or aspect ratio, and / or tooth width, and / or slot width, and / or grating tilt angle, and / or etching time, and / or blazed grating height. In one embodiment, the parameters may be determined based on the confidence value of linear regression, as in linear regression. Of course, linear regression is merely one way to fit based on all etching data; other embodiments include multiple curve equation fitting, etc., with different fitting methods performed according to different process conditions, all of which are within the scope of protection of this invention. No limitations are imposed here.
[0046] As an example, Table 1 below shows the process parameters for different blazing structures.
[0047] Table 1 Process parameters for different blazed grating structures
[0048]
[0049] Based on the actual etching data in Table 1 above, combined with the mapping relationship between scorch angle and scorch height, as a historical data set, different prediction equations were fitted, and the best prediction equation was determined based on the confidence level of the different prediction equations.
[0050] In detail, based on the different blazed grating structures designed in the above serial numbers 1 to 7, experimental etching information of the blazed structure was obtained through known etching processes. Combined with the process parameters corresponding to the different grating structures, blazed prediction information was established, the mapping relationship between different parameters was determined, and the numerical values corresponding to the blazed angle and blazed height were determined through the determined optimal prediction equation, as shown in Table 2 below.
[0051] Table 2 shows the error magnitudes when performing blaze simulations using both the first-order and second-order models, and provides a detailed analysis based on these errors.
[0052] Table 2 Comparison of Parameters Between Etching and Simulation of Blazed Grating Structure
[0053]
[0054] The scintillation etching information refers to the scintillation structures prepared using different etching processes as shown in Table 1. Based on this scintillation etching information and the established prediction model for the corresponding etching process, and based on the optimal prediction equation corresponding to the prediction model, the corresponding scintillation prediction information is calculated. This includes selecting a primary model and a secondary model, and using the optimal prediction equation determined by the primary and secondary models to judge the scintillation prediction information, thus obtaining the corresponding scintillation angle and scintillation height information. As shown in Table 2, the angle difference of the scintillation angle and the height difference of the scintillation height simulated by different prediction models are different. The accuracy of the prediction model is judged based on the corresponding error magnitude. Therefore, this is important for the present invention and is a key factor in improving the model's predictive ability.
[0055] Furthermore, in step (3), the process parameters of the corresponding blazed grating structure are determined based on the simulated blazed grating structure parameters; and the process parameters are used as new online data; more specifically, the process parameters of the corresponding blazed grating structure are determined based on etching process experience using the simulated blazed grating structure parameters; that is, the etching process parameters corresponding to the simulated blazed structure parameters are determined by the process experience formed by those skilled in the art during the etching process, and the etching process parameters are as described above.
[0056] Furthermore, the present invention also includes step (4) inputting new online data into the prediction model in step (2) and judging the accuracy of the prediction model under the optimal prediction equation; when the accuracy is greater than 95%, proceed to step (5); when the accuracy is less than 95%, update the process parameters of the corresponding blazed grating structure, and use the updated process parameters as updated online data, input them into the prediction model in step (2), and judge the accuracy of the prediction model under the optimal prediction equation; repeat this process until the accuracy is greater than 95%, and then proceed to step (5); more specifically, based on the newly given blazed grating parameters in the design, determine various process parameters, use them as new online data, input them into the prediction model determined in step (2), and based on the prediction models corresponding to at least different equipment and / or different etching process conditions, compare and judge the blazed grating structure obtained by each prediction model with the designed blazed grating structure parameters.
[0057] The prediction model, at least under different equipment and / or different etching process conditions, includes two aspects. First, based on the new online data, when input into the prediction model determined in step (2), the prediction model is determined based on the grouping corresponding to the new online data, and then input into the prediction model to simulate the corresponding blazed grating process data. This is, of course, only one embodiment.
[0058] On the other hand, in another embodiment, it also includes inputting new online data into the prediction model determined in step (2), and traversing each prediction model under different groups, and comparing and judging the blazed grating structure obtained by each prediction model with the designed blazed grating structure parameters.
[0059] Both of the above embodiments are methods of prediction.
[0060] When the error accuracy between the output of any prediction model and the designed structural parameters is greater than 95%, proceed to step (5). This indicates that the selected prediction model meets the design requirements. Then, the scintillation structure is simulated based on the best prediction equation under the prediction model, and the next etching process is carried out based on the process parameters under the condition.
[0061] When the error accuracy between the output of any prediction model and the designed structural parameters is less than 95%, it indicates that none of the prediction models meet the requirements. Since there is no model among all the prediction models that can accurately simulate the blazed structure, the process parameters of the corresponding blazed grating structure are updated, and the updated process parameters are used as updated online data and input into the prediction model in step (2). The accuracy of the prediction model under the optimal prediction equation is then determined. This process is repeated until the accuracy is greater than 95%, and then proceeds to step (5). In this step, the process parameters of the corresponding blazed grating structure are determined based on etching process experience, and the updated process parameters are used as updated online data and input into the prediction model in step (2). The accuracy of the prediction model under the optimal prediction equation is then determined and repeated until the accuracy of the model is greater than 95%. The prediction model corresponding to the updated process parameters is stored in the corresponding prediction model database to continuously update the prediction model database in real time.
[0062] Of course, in another embodiment, if after traversing all the prediction models, the accuracy of the prediction model under the optimal prediction equation cannot reach 95%, it means that there is no corresponding prediction model based on the historical data set. The blazed grating structure cannot be simulated using this method for structural prediction. In this case, etching can only be performed based on empirical process parameters, and the etching process parameters are used as the data set to establish a corresponding model to update the prediction model database.
[0063] Furthermore, the method includes storing the updated process parameters and their corresponding prediction models in a corresponding prediction model database for continuous real-time updates. In this step, the updated process parameters are used as input to build a new prediction model; different prediction models are iteratively updated, and the latest optimal prediction equation under that model is determined. The structure of the blazed grating is then simulated using the prediction equation determined by the prediction model under those conditions. In this way, the prediction model database is continuously iteratively optimized and updated, using new online data as the basis for updates and using this data as historical data sets to build more prediction models, thus achieving database establishment across a wider data range. This continuous updating approach to building the database for different blazed structures under different process routes significantly improves the prediction capability and accuracy of the invention.
[0064] As previously shown, when the error accuracy of the two is greater than 95%, the process proceeds to step (5). In step (5), new online data is used as input, and the blazed grating is etched based on the new online data as process parameters. Specifically, in this step, the new online data is input into the optimal prediction equation of the determined different prediction models to obtain the predicted blazed grating; simultaneously, the process parameters corresponding to the prediction model are determined, and etching is performed based on the etching process parameters to obtain the corresponding blazed grating.
[0065] In step (5), the preset blazed grating is formed by etching based on the relevant etching process parameters.
[0066] Furthermore, step (5) includes comparing the scintillation grating formed by etching with the scintillation grating structure to be etched as designed (i.e., the standard scintillation grating structure) to verify the accuracy of the prediction model. This is to achieve feedback verification of the prediction results and to verify the differences between the two.
[0067] Of course, in other cases, if the blazed grating formed by etching differs greatly from the blazed grating structure to be formed by etching given in the design, the process parameters will be continuously adjusted based on step (3) to meet the preset blazed structure requirements.
[0068] Furthermore, in step (4), when judging accuracy, the blazed grating structure obtained based on the new online data obtained in the aforementioned steps as process parameters is compared with the simulated blazed grating structure parameters in step (3). When the difference between the two blazed structure parameters meets a certain threshold, it is determined that the process parameter meets the simulation requirements of the blazed grating; otherwise, step (3) is repeated until the difference between the two blazed structure parameters meets the threshold. Of course, based on different process performance requirements, the difference between the two blazed structure parameters is limited, such as a blazed height less than or equal to 3nm, or less than or equal to 5nm, or less than or equal to 10nm, etc.; a blazed angle difference less than or equal to 0.1°, or less than or equal to 0.2°, or less than or equal to 0.5°, less than or equal to 0.8°, etc. The present invention does not limit the size of this threshold in detail, but determines it based on different performance requirements, process requirements, and design parameter requirements.
[0069] In some other embodiments, step (2) further includes iterative optimization of the prediction model established under different process conditions. By training the prediction model, the latest prediction model under different process conditions is updated in real time to adapt to the changes brought about by different processes or different equipment, so as to improve the applicability of the method of the present invention.
[0070] In detail, based on the blaze prediction information in Table 2, different prediction models are used to predict the blaze angle and height of different blaze structures. Due to the large prediction error of the quadratic model, this invention updates and optimizes the blaze prediction for cases where the quadratic model has a large prediction error, as shown in Table 3 below:
[0071] Table 3 compares the scintillation parameters before and after the prediction model optimization.
[0072]
[0073] In this embodiment, as shown in Table 3, the prediction model is iteratively optimized. For example, the prediction model is set to be a quadratic model for optimization. Comparing the parameters before and after optimization, the optimized blaze angle and blaze height show significantly reduced errors in height and angle, indicating a significant improvement in prediction accuracy. Therefore, in this step, different designs can be used to optimize the prediction model to further improve prediction accuracy.
[0074] In some other embodiments, in step (2), the optimal prediction equation for different prediction models is determined. For example, based on the process parameters in Table 1, combined with the prediction model, the optimal prediction equation corresponding to the prediction model is obtained. For instance, assuming the optimal prediction equation of the fitted linear model, through regression fitting analysis, the relationship between the prediction equation and the straight tooth mask height A, tooth width B, slot width C, grating tilt angle D, and etching time E is defined as follows: Blaze angle = 61.21 - 0.0304A - 0.01748C + 0.01439B + 21.18D - 0.287E, Blaze height = -165.6 + 0.0618A - 0.135B + 91.3C - 7.04D - 0.985E.
[0075] On the other hand, the best prediction equation for limiting the brilliance height is a quadratic model, then brilliance height = 801 - 2.19B + 0.3839A + 0.3372C + 152.4D - 9.77E - 0.00958B*B + 0.0547B*E.
[0076] The above-mentioned fitting prediction equations for blaze angle and blaze height are merely one example equation determined based on the reliability of the fitting equation under certain circumstances. They are not the only optimal prediction equations for the first-order or second-order models defined in this invention. This equation is only for illustrative purposes.
[0077] Furthermore, Tables 2 and 3 show prediction information including primary or secondary models. However, in the process of establishing the prediction model, the optimal prediction equation for the blaze angle can be determined based on the optimal equation of the primary model, and the optimal prediction equation for the blaze height can be determined based on the optimal prediction equation of the optimized secondary model. This also falls within the scope of protection of this invention.
[0078] In the method of the present invention, in step (1), regression analysis is performed on multiple different discrete historical data sets; and in step (2), a prediction model is established based on the historical data sets, and the best prediction equation for different prediction models is determined through regression analysis; of course, the established prediction model can also be continuously optimized and iterated, and different prediction model databases can be updated to achieve the expected model prediction ability and credibility, and the best prediction equation is finally determined by analyzing, iterating, and re-regressing the prediction results.
[0079] This invention analyzes limited and discrete data and constructs a predictive model with strong predictive capabilities based on a suitable regression model. On one hand, the method of this invention establishes a mapping relationship between parameters such as straight-tooth mask height, period, duty cycle, tooth width, slot width, grating tilt angle, and etching time, and blazing height and blazing angle. This allows previously singular and discrete process results to be transformed into a systematic process prediction model database, which is updated based on different process data, thus lowering the barriers to semiconductor etching technology. On the other hand, based on the predictive equations obtained from regression analysis, new design schemes can be predicted in advance, obtaining directional process results and predicting the structures etched by different etching processes. This provides a clear direction for actual process verification, significantly reducing process development time and labor costs, lowering the difficulty of fabricating blazed gratings, and improving production efficiency to save enterprise production costs. It also helps improve the fabrication accuracy of blazed gratings, reducing process errors caused by production equipment and process stability, and optimizing process parameters to improve process stability and reliability.
[0080] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.
Claims
1. A method of simulating a blazed grating, characterized by, Includes the following steps: (1) Obtain historical data sets of blazed grating etching under different etching process conditions using different equipment; (2) Classify the acquired historical data sets, establish prediction models under different equipment and / or different etching process conditions, and determine the best prediction equation for each prediction model. (3) Determine the process parameters of the corresponding blazed grating structure based on the simulated blazed grating structure parameters; and use these process parameters as new online data; (4) Input the new online data from step (3) into the prediction model in step (2) and determine the accuracy of the prediction model under the optimal prediction equation; When the accuracy rate is greater than 95%, proceed to step (5); When the accuracy is less than 95%, the process parameters of the corresponding simulated blazed grating structure are updated again, and the updated process parameters are used as updated online data and input into the prediction model in step (2), and the accuracy of the prediction model under the optimal prediction equation is judged; this is repeated until the accuracy is greater than 95%, and then step (5) is entered. (5) The new online data is used as input, and the blazed grating is etched based on the new online data as process parameters.
2. The method of simulating a blazed grating according to claim 1, wherein, In step (1), the historical data set includes the straight tooth mask height, period, aspect ratio, grating tilt angle, etching time, and blazed grating height of the blazed grating structure under different equipment and / or different process conditions.
3. The method of simulating a blazed grating according to claim 1 or 2, wherein, In step (2), the acquired historical data group is classified in detail. This classification is based on etching data of different equipment under the same process conditions, or etching data of the same equipment under different process conditions, or multiple etching data of the same equipment under the same process conditions, or etching data of different equipment under different process conditions. The etching data under this classification is used as the historical data group for classification.
4. The method for simulating a blazed grating according to claim 3, characterized in that, In step (2), the optimal prediction equation is determined based on the predictive power of the prediction model and the reliability of the regression analysis.
5. The method for simulating a blazed grating according to claim 4, characterized in that, The predictive capability of the prediction model is determined based on the blazed grating structure simulated by the prediction model and the parameters of the blazed grating structure to be simulated.
6. The method for simulating a blazed grating according to claim 1, characterized in that, In step (2), the prediction model is iteratively optimized to update the prediction model under different etching process conditions.
7. The method for simulating a blazed grating according to claim 6, characterized in that, In step (3), the process parameters of the corresponding blazed grating structure are determined based on etching process experience using the simulated blazed grating structure parameters.
8. The method for simulating a blazed grating according to claim 6, characterized in that, In step (4), specifically, the process parameters determined in step (3) are used as new online data and input into the prediction model in step (2). Based on the prediction models corresponding to at least different equipment and / or different etching process conditions, the blazed grating structure obtained by each prediction model under the optimal prediction equation is compared with the simulated blazed grating structure parameters. When the error accuracy between the output result of any prediction model and the simulated blazed grating structure parameters is greater than 95%, the process proceeds to step (5). When the error accuracy between the output result of any prediction model and the designed structure parameters is less than 95%, the process parameters of the corresponding simulated blazed grating structure are updated, and the updated process parameters are used as updated online data and input into the prediction model in step (2). The accuracy of the prediction model under the optimal prediction equation is then determined. This process is repeated until the accuracy is greater than 95%, and then the process proceeds to step (5).
9. The method for simulating a blazed grating according to claim 7, characterized in that, In step (4), the method further includes storing the prediction model corresponding to the updated process parameters in the corresponding prediction model database so as to continuously update the prediction model database in real time.
10. The method for simulating a blazed grating according to claim 8, characterized in that, In step (4), when calculating the accuracy, the blazed grating structure obtained based on the new online data as process parameters is compared with the simulated blazed grating structure parameters in step (3). When the difference between the two blazed structure parameters meets a certain threshold, it is determined that the process parameters meet the simulation requirements of the blazed grating. Otherwise, step (3) is repeated until the difference between the two blazed structure parameters meets the threshold.