Method for analyzing and optimizing optical parameters of focusing sensor and technological parameters of silicon wafer film layer

Abstract

The invention provides an analysis and optimization method for optical parameters of a focusing sensor and technological parameters of a silicon wafer film layer. The method comprises the following steps: constructing an error agent model which takes optical parameters of a focusing sensor and technological parameters of a silicon wafer film layer as inputs and takes a height technological correlation error as an output; constructing a training data set to train the error agent model until the model converges; and according to a preset error target and a parameter optimization range, performing global optimization on the optical parameters of the focusing sensor and the technological parameters of the silicon wafer film layer by using the error agent model to obtain an optimal parameter set conforming to the preset error target. According to the method, the error agent model is established, a focusing sensor and a physical model of a silicon wafer multi-film structure are approximated with high precision in principle, the evaluation precision of the error agent model on the real situation is improved through training optimization, the optimal parameter design is realized in combination with parameter optimization, and the method does not depend on analytical calculation of the physical model; and the parameter analysis and optimization process has higher realizability and higher calculation speed.

Description

Technical Field

[0001] The present disclosure relates to the field of semiconductor manufacturing technology, and in particular to a method, device, storage medium and electronic equipment for analyzing and optimizing optical parameters of a focus sensor and process parameters of a silicon wafer film layer. Background Art

[0002] Focus sensors used in semiconductor manufacturing and testing equipment are mainly used for measuring the surface height and topography of silicon wafers. The focus sensor can obtain the surface height value of the silicon wafer online, calculate the surface topography of the silicon wafer, and convert it into the vertical adjustment amount of the silicon wafer stage through further data processing. The silicon wafer stage makes real-time vertical adjustments to ensure that the silicon wafer is always within the focal depth range of the exposure or detection optical system and near the optimal focal plane, thereby achieving the most ideal exposure or detection effect.

[0003] The incident light beam of the focus sensor is reflected on the surface of the silicon wafer. The focus sensor receives the reflected light beam and performs photoelectric conversion and online calculation to obtain the surface height value of the silicon wafer. Since the silicon wafer has a multi-layer structure during the photolithography exposure process, the incident light beam will be reflected and refracted multiple times in each layer. The reflected light beam will produce an overall phase shift, which directly leads to the measurement error of the silicon wafer surface height, that is, the height process correlation error. The influencing factors of the height process correlation error mainly include two aspects: the optical parameters of the focus sensor and the process parameters of the silicon wafer film layer. The traditional method of calculating the height process correlation error is to calculate the characteristic matrix of the incident light and the reflected light at the junction of each film layer based on the multi-layer electromagnetic wave propagation theory, and finally obtain the overall reflectivity and average phase shift of the multiple film layer structure on the silicon wafer surface, and then obtain the height process correlation error through complex partial derivatives and multiple integral operations. Because the optical model of the height process correlation error is too complex, it is difficult to perform subsequent sensitivity analysis and parameter optimization design through analytical methods. Summary of the invention

[0004] The purpose of the embodiments of the present disclosure is to provide a method for analyzing and optimizing the optical parameters of a focus sensor and the process parameters of a silicon wafer film layer, so as to solve the above-mentioned problems existing in the prior art.

[0005] The embodiments of the present disclosure adopt the following technical scheme: a method for analyzing and optimizing the optical parameters of a focus sensor and the process parameters of a silicon wafer film layer, comprising: constructing an error proxy model with the optical parameters of a focus sensor and the process parameters of a silicon wafer film layer as input and with a high degree of process correlation error as output; constructing a training data set to train the error proxy model until the model converges; and according to preset error targets and parameter optimization ranges, using the error proxy model to globally optimize the optical parameters of the focus sensor and the process parameters of the silicon wafer film layer to obtain an optimal parameter set that meets the preset error targets.

[0006] The disclosed embodiment also provides an analysis and optimization device for the optical parameters of a focus sensor and the process parameters of a silicon wafer film layer, including: a proxy model construction module, used to construct an error proxy model with the optical parameters of a focus sensor and the process parameters of a silicon wafer film layer as input and with a high degree of process correlation error as output; a training module, used to construct a training data set to train the error proxy model until the model converges; a parameter optimization module, used to use the error proxy model to perform global optimization on the optical parameters of the focus sensor and the process parameters of the silicon wafer film layer according to preset error targets and parameter optimization ranges, so as to obtain an optimal parameter set that meets the preset error targets.

[0007] The embodiment of the present disclosure further provides a storage medium storing a computer program, which, when executed by a processor, implements the steps of the method for analyzing and optimizing the optical parameters of the focus sensor and the process parameters of the silicon wafer film layer as described in the first embodiment of the present disclosure.

[0008] An embodiment of the present disclosure also provides an electronic device, comprising at least a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program on the memory, the processor implements the steps of the method for analyzing and optimizing the optical parameters of the focus sensor and the process parameters of the silicon wafer film layer as described in the first embodiment of the present disclosure.

[0009] The beneficial effects of the embodiments of the present disclosure are: establishing an error proxy model for highly process-related errors based on machine learning, which in principle can approximate the physical model of the focus sensor and the multi-film layer structure of the silicon wafer with high precision, and optimizing the training of the error proxy model through a training data set formed by training with a large amount of simulation or actual test data, which can improve the accuracy of the error proxy model in assessing the actual situation, and at the same time combine parameter optimization to achieve the optimal parameter design without relying on the analytical calculation of the physical model, so that the parameter analysis optimization process has higher feasibility and faster calculation speed. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] In order to more clearly illustrate one or more embodiments of this specification or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this specification. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0011] Figure 1 It is a flow chart of the method for analyzing and optimizing the optical parameters of the focus sensor and the process parameters of the silicon wafer film layer in the first embodiment of the present disclosure;

[0012] Figure 2 This is a schematic diagram of a high degree of process correlation error in the first embodiment of the present disclosure;

[0013] Figure 3 A schematic diagram of the optical design and parameters of the focus sensor in the first embodiment of the present disclosure;

[0014] Figure 4 This is a schematic diagram of an error proxy model in the first embodiment of the present disclosure;

[0015] Figure 5 It is a schematic diagram showing the variation of the high process correlation error with the photoresist thickness in the first embodiment of the present disclosure;

[0016] Figure 6 Schematic diagram of the variation of the high process correlation error with the thickness of the anti-reflection layer in the first embodiment of the present disclosure;

[0017] Figure 7 It is a schematic diagram showing the variation of the high process correlation error with the hard mask thickness in the first embodiment of the present disclosure;

[0018] Figure 8 It is a schematic diagram showing the variation of the height process correlation error with the incident angle in the first embodiment of the present disclosure;

[0019] Fig. 9 It is a schematic diagram of the variation of the height process correlation error with the photoresist thickness when different anti-reflection layer thicknesses and hard mask thicknesses are used in the first embodiment of the present disclosure;

[0020] Fig.10 It is a schematic diagram of the variation of the height process correlation error with the photoresist thickness when different focus sensor optical parameters are used in the first embodiment of the present disclosure;

[0021] Fig.11 The global sensitivity analysis result based on the variance method in the first embodiment of the present disclosure;

[0022] Fig.12 A schematic diagram of the Pareto frontier for multi-objective optimization of the optical parameters of the focus sensor and the process parameters of the silicon wafer film layer using a genetic algorithm in the first embodiment of the present disclosure;

[0023] Fig.13 Schematic diagram of the structure of the device for analyzing and optimizing the optical parameters of the focus sensor and the process parameters of the silicon wafer film layer in the second embodiment of the present disclosure. DETAILED DESCRIPTION

[0024] In order to enable those skilled in the art to better understand the technical solutions in one or more embodiments of this specification, the following will be combined with the drawings in one or more embodiments of this specification to clearly and completely describe the technical solutions in one or more embodiments of this specification. Obviously, the described embodiments are only part of the embodiments of this specification, not all of the embodiments. Based on one or more embodiments of this specification, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of this document.

[0025] Focus sensors used in semiconductor manufacturing and testing equipment are mainly used for measuring the surface height and topography of silicon wafers. The focus sensor can obtain the surface height value of the silicon wafer online, calculate the surface topography of the silicon wafer, and convert it into the vertical adjustment amount of the silicon wafer stage through further data processing. The silicon wafer stage makes real-time vertical adjustments to ensure that the silicon wafer is always within the focal depth range of the exposure or detection optical system and near the optimal focal plane, thereby achieving the most ideal exposure or detection effect.

[0026] The incident light beam of the focus sensor is reflected on the surface of the silicon wafer. The focus sensor receives the reflected light beam and performs photoelectric conversion and online calculation to obtain the surface height value of the silicon wafer. Since the silicon wafer has a multi-layer structure during the photolithography exposure process, the incident light beam will be reflected and refracted multiple times in each layer. The reflected light beam will produce an overall phase shift, which directly leads to the measurement error of the silicon wafer surface height, that is, the height process correlation error. The influencing factors of the height process correlation error mainly include two aspects: the optical parameters of the focus sensor and the process parameters of the silicon wafer film layer. The traditional method of calculating the height process correlation error is to calculate the characteristic matrix of the incident light and the reflected light at the junction of each film layer based on the multi-layer electromagnetic wave propagation theory, and finally obtain the overall reflectivity and average phase shift of the multiple film layer structure on the silicon wafer surface, and then obtain the height process correlation error through complex partial derivatives and multiple integral operations. Because the optical model of the height process correlation error is too complex, it is difficult to perform subsequent sensitivity analysis and parameter optimization design through analytical methods.

[0027] In order to solve the above problems, the first embodiment of the present disclosure provides a method for analyzing and optimizing the optical parameters of a focus sensor and the process parameters of a silicon wafer film layer, which is used to guide the design of the focus sensor optical system and the design of the silicon wafer multi-film layer parameters during the semiconductor manufacturing process. Figure 1 A flow chart of the analysis and optimization method of this embodiment is shown. Figure 1 As shown, the method mainly includes steps S10 to S30:

[0028] S10, constructing an error proxy model with the focus sensor optical parameters and the silicon wafer film layer process parameters as input and the highly process-related error as output.

[0029] The height process correlation error is due to the multi-layer structure of the silicon wafer, which causes the incident light beam and the reflected light beam of the focus sensor to form multiple reflections and refractions between different film layers, resulting in an overall phase shift of the reflected light beam, which ultimately causes an offset between the measured height of the silicon wafer surface and the actual height, such as Figure 2 As shown in the figure. The conventional evaluation model for highly process-dependent errors is a complex optical model of the optical parameters of the focus sensor and the process parameters of the silicon wafer film layer. Due to the complexity of the physical principles, it needs to be modeled using an approximate method, which reduces the accuracy of the evaluation model. In addition, the optical calculation formula for highly process-dependent errors is complex and cannot explicitly include the optical parameters of the focus sensor and the process parameters of the silicon wafer film layer, making it difficult to perform sensitivity analysis and optimal design of the design parameters using analytical methods.

[0030] In response to the above problems, this embodiment establishes an error proxy model for highly process-correlation errors. The error proxy model takes the optical parameters of the focus sensor and the process parameters of the silicon wafer film layer as input, and takes the highly process-correlation errors as output. It approximates the optical model of the focus sensor and the multi-film layer structure of the silicon wafer with high precision based on machine learning, avoids analyzing the highly process-correlation errors through complex methods of optical models, solves the problems of waste of computing power and time consumption, and optimizes the calculation time of the highly process-correlation errors.

[0031] In this embodiment, the focus sensor optical module generates an incident light beam, which passes through the surface of the silicon wafer to be tested to form a reflected light beam. The focus sensor receives the reflected light beam and then calculates the height value of the silicon wafer surface according to the intensity of the reflected light beam. The principle diagram is as follows: Figure 3 As shown. The optical parameters of the focus sensor mainly include at least one of the following parameters: numerical aperture NA, incident light angle of incidence, incident light center wavelength, incident light wavelength bandwidth, incident light spectral distribution, incident light polarization coefficient; wherein, numerical aperture NA is the product of the refractive index n of the incident light in the transmission medium and the sine of the aperture angle θ. Since the optical path of the focus sensor detection light is in the air, NA = sin(θ); the incident light angle is the angle between the incident light and the normal of the silicon wafer surface; the focus sensor uses a wide spectrum light source, and the incident light center wavelength is the middle value of the spectrum wavelength; the incident light bandwidth makes the incident light effective in the above spectrum; the incident light spectral distribution is the distribution function of the incident light intensity with the wavelength; the incident light has two polarization states, s light and p light, and the incident light polarization coefficient is the intensity ratio of these two polarization states in the total light intensity, and the range is [0, 1]. Some or all of the optical parameters described above can be used as input parameters of the error proxy model. At the same time, the optical parameter information used in this embodiment may include but is not limited to the above parameters.

[0032] Different process treatments are required during the processing of silicon wafers. The focus sensor is mainly used in the exposure stage or detection stage of silicon wafers. The silicon wafer to be tested generally has multiple process film layers. The silicon wafer film layer process parameters mainly include at least one of the following parameters: the film layer structure of the silicon wafer, the material type of each film layer, the refractive index n and extinction coefficient k of each film layer, and the film layer thickness; wherein the film layer structure parameters describe how many film layers the surface of the silicon wafer to be tested is composed of and the function of each film layer. Typical film layers include: photoresist film layer, top or bottom anti-reflective film layer, hard mask film layer, substrate layer; the film layer material type is the material model of each film layer. The same type of film layer of different models may have significantly different optical properties. Some or all of the process parameters described above can be used as input parameters of the error proxy model. At the same time, the process parameter information used in this embodiment may include but is not limited to the above parameters.

[0033] The height process correlation error is the height process correlation error evaluation value obtained by the proxy model calculation when the optical parameters of the focus sensor and the process parameters of the silicon wafer film layer are given as input. The error proxy model of this embodiment is established using a machine learning method, such as Figure 4 As shown, the machine learning method includes at least one of the following methods: neural network, decision tree, random forest, support vector regression; the neural network model constructed when the neural network method is adopted can adopt any one of the following models: multi-layer perceptron (MLP), residual neural network (RNN), convolutional neural network (CNN), deep neural network (DNN), and this embodiment does not make specific restrictions.

[0034] S20, constructing a training data set to train the error proxy model until the model converges.

[0035] After the error proxy model is constructed, it can be trained through the training data set to make it reflect the real physical model. Specifically, the training data set can be established through optical model simulation calculation, or established using experimental data, or a combination of the two. The training data set constructed in this embodiment should at least include the optical parameters of the focus sensor and the process parameters of the silicon wafer film layer used in the simulation or experiment, and also include the simulation error or experimental error corresponding to the highly process-related error under different optical parameter and process parameter settings. The training data set is then used to train the error proxy model until the model converges. The specific convergence conditions can be set according to conditions such as the amount of data or accuracy requirements, and this embodiment does not impose specific restrictions.

[0036] In actual operation, the training data set can also be divided into training samples and test samples. The training samples are used for model training, and the test samples are used to evaluate the trained model and fine-tune the model based on the evaluation results.

[0037] S30, according to the preset error target and parameter optimization range, the error proxy model is used to perform global optimization on the optical parameters of the focus sensor and the process parameters of the silicon wafer film layer to obtain an optimal parameter set that meets the preset error target.

[0038] Parameter optimization refers to the global optimization of the optical parameters of the focus sensor and the process parameters of the silicon wafer film layer under given optimization objectives and constraints, so as to obtain the optimal parameter set that minimizes the optimization objectives. In this embodiment, the above-mentioned parameter optimization function can be realized by calling the parameter optimization model, and its input is the parameter optimization range of the optical parameters of the focus sensor and the process parameters of the silicon wafer film layer, and the preset optimization objectives of the high process correlation error. At the same time, in the parameter optimization process, it is also necessary to call the error proxy model to cooperate in the error calculation and verification functions, and finally output the optimized parameter set. In some embodiments, the preset optimization target is a set of functions about the high process correlation error, which are used to describe the size or change speed of the high process correlation error value, and the optimal parameter optimization set can give at least one solution set that meets the above-mentioned optimization objectives and constraints. The operator can select one of the solution sets from multiple solution sets as the analysis and optimization results of each parameter based on the actual needs.

[0039] In some embodiments, the parameter optimization model executes a parameter optimization algorithm to implement the above optimization process, and the parameter optimization algorithm includes at least one of the following methods: genetic algorithm, evolutionary algorithm, particle swarm algorithm.

[0040] In some embodiments, after the error proxy model is trained, a global sensitivity analysis step of parameters may also be included. Specifically, the global sensitivity analysis may be performed on all parameters to determine the degree of influence of each parameter on the error of high process correlation, and the degree of influence may be described by the sensitivity index. After the sensitivity index of all parameters is determined, at least one parameter with a sensitivity index greater than a preset threshold is selected as a key parameter. When performing parameter optimization in the subsequent process, only the key parameter optimization range of the key parameter is set, and the error proxy model is used to perform global optimization on the key parameters, so that the error calculated by the obtained optimal key parameter set meets the preset error target. As the key parameters that have a more obvious influence on the error of high process correlation, only these parameters are optimized and the parameters with small influence are ignored, thereby reducing the complexity of the model and algorithm and improving the calculation speed.

[0041] Furthermore, after determining the key parameters, the training data set can be further screened according to the key parameters. For example, only the sample data corresponding to the key parameters are retained in the training data set, and the data of other parameters with less impact on the error can be directly deleted, and the error proxy model is optimized using the screened training data set to improve the fitting effect of the key parameters, reduce the amount of data for the training model, and improve the optimization speed and model accuracy. In actual use, the steps of global sensitivity analysis of the above parameters can be implemented by calling the global sensitivity analysis model, and the global sensitivity analysis model can be constructed based on the variance method or the density function method.

[0042] Combine the following Figures 5 to 12 A possible implementation process of the analysis and optimization method of this embodiment is described.

[0043] The high process correlation error analysis and optimization process includes three parts: the error proxy model of high process correlation error, the global sensitivity analysis model and the parameter optimization model. The function of the error proxy model is to calculate the high process correlation error evaluation value through machine learning methods. The input is the focus sensor optical parameters and silicon wafer film layer process parameters, and the output is the high process correlation error; the function of the global sensitivity analysis model is to quantitatively analyze the influence weight of the focus sensor optical parameters and silicon wafer film layer process parameters on the high process correlation error. The input is the focus sensor optical parameters, silicon wafer film layer process parameters and the high process correlation error proxy model. The output is the sensitivity index of the above parameters, that is, the influence weight on the high process correlation error. The model uses the error proxy model for the calculation of the high process correlation error; the function of the parameter optimization model is to perform global optimization of the focus sensor optical parameters and silicon wafer film layer process parameters under given optimization objectives and constraints, and obtain the optimal parameter set that minimizes the optimization objective. The input is the optimization range of the focus sensor optical parameters and silicon wafer film layer process parameters and the preset error optimization objectives, the error proxy model, and the output is the optimal parameter set. The model uses the error proxy model for the calculation of the high process correlation error.

[0044] In order to better illustrate the analysis and optimization method of this embodiment, a typical 4-layer silicon wafer process film structure is used as an example. The silicon wafer film layers are photoresist layer, anti-reflection layer, hard mask layer and substrate layer from top to bottom. The optical parameters of the focus sensor include: numerical aperture NA, incident light angle of incidence, incident light center wavelength, incident light wavelength bandwidth, incident light spectral distribution, polarization coefficient; silicon wafer film layer process parameters include: material and film thickness of each film layer, and the output parameter is a high degree of process correlation error. The machine learning agent model is established using a deep neural network method.

[0045] Figure 5 , Figure 6 and Figure 7 The following are schematic diagrams of how the height process correlation error changes with the thickness of the photoresist, the thickness of the anti-reflective layer and the thickness of the hard mask when other parameters are fixed. It can be clearly seen that the silicon wafer process structure parameters will have a significant impact on the height process correlation error. Figure 8 This is a schematic diagram of how the height process correlation error changes with the incident angle when other parameters are fixed. The optical parameters of the focus sensor will also cause changes in the height process correlation error. Fig. 9 It is a schematic diagram of the change of the height process correlation error with the photoresist thickness when different anti-reflection layer thickness and hard mask thickness are used for the fixed focus sensor optical parameters. The figure shows that for the height process correlation error, there is a complex coupling relationship between the three parameters of photoresist thickness, anti-reflection layer thickness and hard mask thickness. Fig.10 It is a schematic diagram of the variation of the height process correlation error with the photoresist thickness when the silicon wafer film layer process parameters are fixed and different focus sensor optical parameters are used. This figure shows that for the height process correlation error, there is also a complex coupling relationship between the focus sensor optical parameters.

[0046] In order to quantitatively analyze the weight of the influence of the optical parameters of the focus sensor and the process parameters of the silicon wafer film layer on the height process correlation error, a global sensitivity analysis model is used for analysis. Fig.11 It is the result of global sensitivity analysis based on variance method. The figure quantitatively represents the influence weights of all the above parameters, and the relative size between these parameters can be judged through this figure. According to the results of global sensitivity analysis, the design of error proxy model and parameter optimization model can be optimized, and parameters with small influence weights can be ignored, thereby reducing the complexity of the model and algorithm and improving the calculation speed.

[0047] For the parameter optimization process, the design optimization objectives are to minimize the absolute value of the height correlation error and the statistical sum of the height correlation error, and the design constraints are the design ranges of the focus sensor optical parameters and the silicon wafer film process parameters. The above statistical sum is designed by weighting the mean value and standard deviation of the height correlation error.

[0048] Fig.12 This is a Pareto frontier diagram of multi-objective optimization of focus sensor optical parameters and silicon wafer film process parameters using genetic algorithm. In the figure, f1 is the absolute value of the highly process-dependent error, and f2 is the weighted sum of the mean and standard deviation. For the convenience of display, the value of f1 in the figure is enlarged by 1000 times. This figure represents all optimal solution sets that meet the minimum optimization objectives under the above constraints. The most optimized parameter set can be selected from the above solution set.

[0049] This embodiment establishes an error proxy model for highly process-related errors based on machine learning, which in principle can approximate the physical model of the focus sensor and the multi-film layer structure of the silicon wafer with high precision. The training and optimization of the error proxy model through a training data set formed by a large amount of simulation or actual test data training can improve the accuracy of the error proxy model in assessing the actual situation. At the same time, the optimal parameter design can be achieved by combining parameter optimization without relying on the analytical calculation of the physical model, making the parameter analysis and optimization process more feasible and with a faster calculation speed.

[0050] Based on the same inventive concept, the second embodiment of the present disclosure provides an analysis and optimization device for focusing sensor optical parameters and silicon wafer film layer process parameters, and its structural schematic diagram is shown in FIG. Fig.13 As shown, it mainly includes: a proxy model construction module 10, which is used to construct an error proxy model with the optical parameters of the focus sensor and the process parameters of the silicon wafer film layer as input and a highly process-related error as output; a training module 20, which is used to construct a training data set to train the error proxy model until the model converges; a parameter optimization module 30, which is used to use the error proxy model to perform global optimization on the optical parameters of the focus sensor and the process parameters of the silicon wafer film layer according to preset error targets and parameter optimization ranges, so as to obtain an optimal parameter set that meets the preset error targets.

[0051] Specifically, the optical parameters of the focus sensor include at least one of the following parameters: numerical aperture, incident angle of incident light, central wavelength of incident light, wavelength bandwidth of incident light, spectral distribution of incident light, and polarization coefficient of incident light; the silicon wafer film layer process parameters include at least one of the following parameters: film layer structure of the silicon wafer, material type of each film layer, refractive index and extinction coefficient of each film layer, and film layer thickness.

[0052] In some embodiments, the device further includes a sensitivity analysis module 40, which is specifically used to perform a global sensitivity analysis on all parameters, and determine at least one key parameter whose sensitivity index is greater than a preset threshold value among all parameters; and, when the sensitivity analysis module 40 determines the key parameters, the parameter optimization module 30 can use the error proxy model to perform a global optimization on the key parameters according to the preset error target and the key parameter optimization range, and obtain the optimal key parameter set that meets the preset error target. In addition, when the sensitivity analysis module 40 determines the key parameters, the training module 20 can also perform sample screening on the training data set according to the key parameters, and optimize the error proxy model according to the screened training data set.

[0053] Specifically, the sensitivity analysis module 40 realizes its function by calling the global sensitivity analysis model, and the global sensitivity analysis model is constructed based on the variance method or the density function method. Specifically, the parameter optimization module 30 realizes its function by calling the parameter optimization model, and the parameter optimization model executes the parameter optimization algorithm, and the parameter optimization algorithm includes at least one of the following methods: genetic algorithm, evolutionary algorithm, particle swarm algorithm. Specifically, the error proxy model is established using a machine learning method, and the machine learning method includes at least one of the following methods: neural network, decision tree, random forest, support vector regression.

[0054] The specific functions and principles implemented by each functional module in this embodiment have been described in detail in the first embodiment and will not be repeated here.

[0055] This embodiment establishes an error proxy model for highly process-related errors based on machine learning, which in principle can approximate the physical model of the focus sensor and the multi-film layer structure of the silicon wafer with high precision. The training and optimization of the error proxy model through a training data set formed by a large amount of simulation or actual test data training can improve the accuracy of the error proxy model in assessing the actual situation. At the same time, the optimal parameter design can be achieved by combining parameter optimization without relying on the analytical calculation of the physical model, making the parameter analysis and optimization process more feasible and with a faster calculation speed.

[0056] Based on the same inventive concept, the third embodiment of the present disclosure provides a storage medium storing a computer program, which, when executed by a processor, implements the steps of the method for analyzing and optimizing the optical parameters of the focus sensor and the silicon wafer film layer process parameters described in the first embodiment of the present disclosure.

[0057] Based on the same inventive concept, the fourth embodiment of the present disclosure provides an electronic device, comprising at least a memory and a processor, wherein a computer program is stored on the memory, and when the processor executes the computer program on the memory, the processor implements the steps of the method for analyzing and optimizing the optical parameters of the focus sensor and the process parameters of the silicon wafer film layer as described in the first embodiment of the present disclosure.

[0058] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present disclosure, rather than to limit them. Although the present disclosure has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein by equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present disclosure.

Claims

1. A method for analyzing and optimizing the optical parameters of a focus sensor and the process parameters of a silicon wafer film layer, characterized in that: include: Construct an error proxy model that takes the focus sensor optical parameters and silicon wafer film layer process parameters as input and takes highly process-related errors as output; Constructing a training data set to train the error proxy model until the model converges; According to the preset error target and parameter optimization range, the error proxy model is used to perform global optimization on the optical parameters of the focus sensor and the process parameters of the silicon wafer film layer to obtain an optimal parameter set that meets the preset error target.

2. The analysis and optimization method according to claim 1, characterized in that: The optical parameters of the focus sensor include at least one of the following parameters: numerical aperture, incident angle of incident light, central wavelength of incident light, wavelength bandwidth of incident light, spectral distribution of incident light, and polarization coefficient of incident light; The silicon wafer film layer process parameters include at least one of the following parameters: the film layer structure of the silicon wafer, the material type of each film layer, the refractive index and extinction coefficient of each film layer, and the film layer thickness.

3. The analysis and optimization method according to claim 1, characterized in that: After the error proxy model is trained by constructing a training data set until the model converges, the method further includes: Performing a global sensitivity analysis on all parameters, and determining at least one key parameter having a sensitivity index greater than a preset threshold among all parameters; According to the preset error target and the key parameter optimization range, the error proxy model is used to perform global optimization on the key parameters to obtain the optimal key parameter set that meets the preset error target.

4. The analysis and optimization method according to claim 3, characterized in that: After performing a global sensitivity analysis on all parameters and determining at least one key parameter whose sensitivity index is greater than a preset threshold value among all parameters, the method further includes: The training data set is sampled and screened according to the key parameters, and the error proxy model is optimized according to the screened training data set.

5. The analysis and optimization method according to claim 3, characterized in that: The step of performing global sensitivity analysis on all parameters is achieved by calling a global sensitivity analysis model, and the global sensitivity analysis model is constructed based on a variance method or a density function method.

6. The analysis and optimization method according to claim 1, characterized in that: The step of using the error proxy model to globally optimize the optical parameters of the focus sensor and the process parameters of the silicon wafer film layer according to the preset error target and the parameter optimization range to obtain the optimal parameter set that meets the preset error target is achieved by calling the parameter optimization model. The parameter optimization model executes a parameter optimization algorithm, and the parameter optimization algorithm includes at least one of the following methods: genetic algorithm, evolutionary algorithm, and particle swarm algorithm.

7. The analysis and optimization method according to any one of claims 1 to 6, characterized in that: The error proxy model is established using a machine learning method, and the machine learning method includes at least one of the following methods: neural network, decision tree, random forest, and support vector regression.

8. An analysis and optimization device for focusing sensor optical parameters and silicon wafer film layer process parameters, characterized in that: include: A proxy model building module is used to build an error proxy model that takes the focus sensor optical parameters and silicon wafer film layer process parameters as input and takes a highly process-related error as output; A training module, used to construct a training data set to train the error proxy model until the model converges; The parameter optimization module is used to use the error proxy model to perform global optimization on the optical parameters of the focus sensor and the process parameters of the silicon wafer film layer according to a preset error target and a parameter optimization range, so as to obtain an optimal parameter set that meets the preset error target.

9. A storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the method for analyzing and optimizing the optical parameters of the focus sensor and the process parameters of the silicon wafer film layer according to any one of claims 1 to 7 are implemented.

10. An electronic device, comprising at least a memory and a processor, wherein a computer program is stored in the memory, wherein: The processor implements the steps of the method for analyzing and optimizing the optical parameters of the focus sensor and the process parameters of the silicon wafer film layer according to any one of claims 1 to 7 when executing the computer program on the memory.

Images