Film thickness determination method and training method of film thickness estimation model
By converting data of the initial reflection spectrum data of the film thickness and processing the global optimization algorithm, the model trained by the simulation spectrum training set is solved, and the film thickness measurement range is small, the accuracy is low and the stability is poor, achieving higher measurement accuracy degree and stability.
Patent Information
- Application Number
- CN202410974292.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-19
- Publication Date
- 2024-12-06
AI Technical Summary
The existing methods of determining film thickness using reflection spectrum have problems such as small measurement range, low accuracy of measurement results, and poor stability.
By obtaining the initial reflection spectrum data of the film to be tested, the data conversion is performed to obtain the target reflection spectrum data, and input it to the film thickness estimation model. The film thickness is determined based on the film thickness interval using a global optimization algorithm based on the progressive non-convexity solver. This model is trained based on the simulation spectrum training set associated with the film thickness. By reasonably setting the number and diversity of the simulation spectrum sample data, it covers a larger film thickness range, thereby improving the accuracy and stability of the estimation model.
The measurement range and accuracy of film thickness are improved, the stability of measurement results is enhanced, local optimal solutions are avoided, and higher accuracy is achieved.
Smart Images

Figure CN119089763A_ABST
Abstract
Description
Technical Field
[0001] The present disclosure relates to the fields of optical measurement and deep learning technology, and more specifically, to a method for determining film thickness and a method for training a film thickness estimation model. Background Art
[0002] With the development of nano-manufacturing technology, micro-nano structures are gradually widely used in semiconductor manufacturing, optical device manufacturing, nano-materials, biomedicine and other fields. Among them, thin films are an important component of micro-nano structures, and film thickness is a key indicator affecting the performance and yield of micro-nano devices, so it is necessary to measure the film thickness with high precision. For example, based on the principle of interference effect, the reflection spectrum can be used to determine the thickness of the film to be measured by reflecting light on the upper and lower surfaces of the film to be measured to form interference.
[0003] In the process of realizing the concept of the present disclosure, the inventors found that there are at least the following problems in the related art: the related method of determining the thickness of the film to be measured using reflection spectroscopy has problems such as small measurement range, low measurement result accuracy, and poor stability. Summary of the invention
[0004] In view of this, the present disclosure provides a method for determining film thickness, and a method and device for training a film thickness estimation model.
[0005] One aspect of the present disclosure provides a method for determining film thickness, comprising: obtaining initial reflection spectrum data of a film to be measured, wherein the initial reflection spectrum data is associated with the film thickness of the film to be measured; performing data conversion on the initial reflection spectrum data to obtain target reflection spectrum data; inputting the target reflection spectrum data into a film thickness estimation model to obtain a film thickness range of the film to be measured, wherein the film thickness estimation model is obtained by training a candidate film thickness estimation model based on a simulation spectrum training set associated with the film thickness; and determining the film thickness of the film to be measured based on the film thickness range based on a preset optimization algorithm, wherein the preset optimization algorithm includes a global optimization algorithm based on an asymptotically non-convex solver.
[0006] According to an embodiment of the present disclosure, the above-mentioned method of determining the film thickness of the film to be tested based on a preset optimization algorithm and according to the film thickness range includes: determining the initial value of the film thickness based on the film thickness range; determining the objective function value based on the initial value of the film thickness based on a preset objective function; and adjusting the parameters of the global optimization algorithm based on the objective function value based on a preset parameter adjustment strategy until a predetermined end condition is met to obtain the film thickness of the film to be tested.
[0007] According to an embodiment of the present disclosure, the preset objective function includes a residual term and a constraint term. The above-mentioned determination of the objective function value based on the preset objective function and according to the initial value of the film thickness includes: determining the first sub-objective function value according to the residual term weight and the initial value of the film thickness; determining the second sub-objective function value according to the constraint term control parameter and the residual term weight; and determining the objective function value according to the first sub-objective function value and the second sub-objective function value.
[0008] According to an embodiment of the present disclosure, the above-mentioned preset parameter adjustment strategy is based on the objective function value, and the parameters of the global optimization algorithm are adjusted until the predetermined end condition is met to obtain the film thickness of the film to be measured, which includes: when the predetermined end condition is not met, iteratively performing the following operations: based on the preset parameter adjustment strategy, according to the objective function value, adjusting the control parameters and the constraint item weights to obtain adjusted control parameters and adjusted constraint item weights; updating the objective function according to the adjusted control parameters and the adjusted constraint item weights to obtain an updated objective function; obtaining an updated objective function value according to the current objective function value and the updated objective function; and in response to satisfying the predetermined end condition, determining the updated objective function value as the film thickness of the film to be measured.
[0009] According to an embodiment of the present disclosure, the initial reflection spectrum data includes M initial reflection light data points, where M is a positive integer, and each initial reflection light data point is used to characterize the light intensity information of the reflected light of the film to be tested under the selected light source wavelength. The above-mentioned data conversion of the initial reflection spectrum data to obtain the target reflection spectrum data includes: normalizing the M initial reflection light data points according to the light source wavelength and light intensity information of each of the M initial reflection light data points to obtain M normalized reflection light data points corresponding to the M initial reflection light data points; determining the feature dimension and data length corresponding to the initial reflection spectrum data according to the M normalized reflection light data points; and determining the target reflection spectrum data according to a preset batch size, feature dimension and data length.
[0010] According to an embodiment of the present disclosure, the film thickness interval includes an interval lower limit value and an interval upper limit value. The above-mentioned determination of the initial value of the film thickness based on the film thickness interval includes: determining the interval middle value of the film thickness interval based on the interval lower limit value and the interval upper limit value; and determining the interval middle value as the initial value of the film thickness.
[0011] Another aspect of the present disclosure provides a method for training a film thickness estimation model, comprising: based on the principle of interference effect, according to the correlation between film thickness and reflected light, obtaining reflection spectrum sample data of each of N sample films, where N is a positive integer; processing the reflection spectrum sample data of each of the N sample films to obtain a simulation spectrum training set associated with the film thickness; and training a candidate film thickness estimation model according to the simulation spectrum training set to obtain a film thickness estimation model, wherein the candidate film thickness estimation model includes a residual network model.
[0012] According to an embodiment of the present disclosure, the residual network model includes a convolutional layer, at least one residual module, at least one batch normalization layer, at least one activation function layer, an average pooling layer and a fully connected layer.
[0013] According to an embodiment of the present disclosure, the above-mentioned processing of the reflection spectrum samples of each of the N sample films to obtain a simulation spectrum training set associated with the film thickness includes: based on preset training set parameters, constructing a first intermediate training set according to the reflection spectrum samples of each of the N sample films, wherein the preset training set parameters include the training set category, the film thickness interval length and the number of training samples; adding random noise to the first intermediate training set to obtain a second intermediate training set; and normalizing the second intermediate training set to obtain a simulation spectrum training set.
[0014] According to an embodiment of the present disclosure, a method for determining the correlation between film thickness and reflected light includes: determining the electric field vector of the reflected light on the upper surface of the preset film based on the principle of interference effect and a characteristic matrix of the preset film, wherein the preset film includes at least one dielectric layer, the characteristic matrix is used to describe the refractive index and thickness of each of the at least one dielectric layer, and the electric field vector is used to characterize the amplitude information and phase information of the reflected light; and establishing the correlation between film thickness and reflected light based on the characteristic matrix, as well as the amplitude information and phase information of the reflected light.
[0015] Another aspect of the present disclosure provides a device for determining film thickness, including: an acquisition module, used to acquire initial reflection spectrum data of a film to be tested, wherein the initial reflection spectrum data is associated with the film thickness of the film to be tested; a conversion module, used to perform data conversion on the initial reflection spectrum data to obtain target reflection spectrum data; an input module, used to input the target reflection spectrum data into a film thickness estimation model to obtain a film thickness range of the film to be tested, wherein the film thickness estimation model is obtained by training a candidate film thickness estimation model based on a simulation spectrum training set associated with the film thickness; and a determination module, used to determine the film thickness of the film to be tested based on the film thickness range based on a preset optimization algorithm, wherein the preset optimization algorithm includes a global optimization algorithm based on an asymptotically non-convex solver.
[0016] Another aspect of the present disclosure provides a training device for a film thickness estimation model, including: a sample module, which is used to obtain reflection spectrum sample data of each of N sample films based on the principle of interference effect and the correlation between film thickness and reflected light, where N is a positive integer; a processing module, which is used to process the reflection spectrum sample data of each of the N sample films to obtain a simulation spectrum training set associated with the film thickness; and a training module, which is used to train a candidate film thickness estimation model based on the simulation spectrum training set to obtain a film thickness estimation model, wherein the candidate film thickness estimation model includes a residual network model.
[0017] According to the embodiments of the present disclosure, by using the film thickness estimation model, the film thickness interval of the film to be measured can be obtained based on the initial reflection spectrum data of the film to be measured. By using the global optimization algorithm based on the asymptotic non-convexity solver, the film thickness of the film to be measured can be determined according to the film thickness interval. Since the film thickness estimation model is obtained by training the candidate film thickness estimation model based on the simulation spectrum training set associated with the film thickness, the number and diversity of the simulation spectrum sample data can be reasonably set so that the simulation spectrum training set can cover a larger film thickness range as much as possible, thereby helping to improve the accuracy, stability and range of the film thickness estimation model. In addition, since the global optimization algorithm based on the asymptotic non-convexity solver is used to solve the film thickness of the film to be measured, it is effectively avoided to fall into the local optimal solution in the process of solving the film thickness, thereby further improving the accuracy of the film thickness. Therefore, the film thickness determination method provided by the embodiments of the present disclosure can at least partially solve the technical problems existing in the related art such as small measurement range, low measurement result accuracy, poor stability, etc. BRIEF DESCRIPTION OF THE DRAWINGS
[0018] The above and other objects, features and advantages of the present disclosure will become more apparent through the following description of the embodiments of the present disclosure with reference to the accompanying drawings, in which:
[0019] Figure 1 A schematic diagram of the interference effect principle according to an embodiment of the present disclosure is schematically shown;
[0020] Figure 2 A diagram schematically shows an application scenario of a method for determining film thickness, a method for training a film thickness estimation model, and an apparatus according to an embodiment of the present disclosure;
[0021] Figure 3 A flow chart of a method for determining film thickness according to an embodiment of the present disclosure is schematically shown;
[0022] Figure 4 The schematic diagram of the principle of the progressive non-convex solver according to the embodiment of the present disclosure is shown;
[0023] Figure 5 exemplarily shows initial reflection spectrum data according to an embodiment of the present disclosure;
[0024] Figure 6 A flowchart of a method for training a film thickness estimation model according to an embodiment of the present disclosure is schematically shown;
[0025] Figure 7 The residual network model according to the embodiment of the present disclosure is schematically shown;
[0026] Figure 8 exemplarily shows a normalized reflectance spectrum according to an embodiment of the present disclosure;
[0027] Fig. 9 The structure block diagram of the device for determining the film thickness according to the embodiment of the present disclosure is schematically shown;
[0028] Fig.10 The structural block diagram of a training device for a film thickness estimation model according to an embodiment of the present disclosure is schematically shown. DETAILED DESCRIPTION
[0029] Hereinafter, embodiments of the present disclosure will be described with reference to the accompanying drawings. However, it should be understood that these descriptions are exemplary only and are not intended to limit the scope of the present disclosure. In the following detailed description, for ease of explanation, many specific details are set forth to provide a comprehensive understanding of the embodiments of the present disclosure. However, it is apparent that one or more embodiments may also be implemented without these specific details. In addition, in the following description, descriptions of known structures and technologies are omitted to avoid unnecessary confusion of the concepts of the present disclosure.
[0030] The terms used herein are only for describing specific embodiments and are not intended to limit the present disclosure. The terms "comprise", "include", etc. used herein indicate the existence of the features, steps, operations and / or components, but do not exclude the existence or addition of one or more other features, steps, operations or components.
[0031] All terms (including technical and scientific terms) used herein have the meanings commonly understood by those skilled in the art unless otherwise defined. It should be noted that the terms used herein should be interpreted as having a meaning consistent with the context of this specification and should not be interpreted in an idealized or overly rigid manner.
[0032] When using expressions such as "at least one of A, B, and C, etc.", they should generally be interpreted according to the meaning of the expression commonly understood by those skilled in the art (for example, "a system having at least one of A, B, and C" should include but is not limited to a system having A alone, B alone, C alone, A and B, A and C, B and C, and / or A, B, C, etc.).
[0033] In the embodiments of the present disclosure, the collection, updating, analysis, processing, use, transmission, provision, disclosure, storage, etc. of the data involved (for example, including but not limited to user personal information) are in compliance with the provisions of relevant laws and regulations, are used for legitimate purposes, and do not violate public order and good morals. In particular, necessary measures are taken for user personal information to prevent illegal access to user personal information data and maintain the security of user personal information, network security, and national security.
[0034] In the embodiments of the present disclosure, the user's authorization or consent is obtained before obtaining or collecting the user's personal information.
[0035] With the development of nano-manufacturing technology, micro-nano structures are gradually widely used in semiconductor manufacturing, optical device manufacturing, nano-materials, biomedicine and other fields. Among them, thin films are an important component of micro-nano structures, and film thickness is a key indicator affecting the performance and yield of micro-nano devices, so it is necessary to measure the film thickness with high precision. For example, based on the principle of interference effect, the reflection spectrum can be used to determine the thickness of the film to be measured by reflecting light on the upper and lower surfaces of the film to be measured to form interference.
[0036] In the process of realizing the concept of the present disclosure, the inventors found that there are at least the following problems in the related art: the related method of determining the thickness of the film to be measured using reflection spectroscopy has problems such as small measurement range, low measurement result accuracy, and poor stability.
[0037] Figure 1 The schematic diagram of the interference effect principle according to an embodiment of the present disclosure is schematically shown.
[0038] like Figure 1 As shown, the light emitted by the light source 101 passes through the optical fiber 103 and the lens group 104 in sequence, and then vertically incident on the film to be tested 105. Among them, a part of the light is directly reflected on the upper surface 105_1 of the film to be tested 105, and a part of the light enters the film to be tested 105 and refracts, and then refracts again after being reflected by the lower surface 105_2 of the film to be tested, and finally emerges from the upper surface 105_1 of the film to be tested. The two parts of the reflected light can pass through the lens group 104 and the optical fiber 103 in sequence and be collected by the spectrometer 102. The light reflected on the upper surface 105_1 of the film to be tested and the light reflected on the lower surface 105_2 of the film to be tested will interfere with each other to form interference fringes of light and dark. The light and dark changes of these interference fringes reflect the periodic changes of light intensity, and the light intensity changes are related to the thickness and refractive index of the film to be tested. Based on this, the reflection spectrum formed by the two coherent lights can be used to determine the film thickness of the film to be tested 105 according to the wavelength and phase changes of the light waves emitted by the light source 101.
[0039] According to an embodiment of the present disclosure, the thickness of the film to be measured can be determined according to the reflection spectrum of the film to be measured by the frequency extraction method or the nonlinear fitting method. However, the inventors have found in practice that the frequency extraction method (such as the Fast Fourier Transform, FFT) often requires a large amount of zero padding, and has problems such as a small measurement range and poor stability of the measurement results. The nonlinear fitting method (such as the least squares model) often requires a given initial value for the solution, and has problems such as improper selection of the initial value, easy to fall into local optimization, and low accuracy of the measurement results.
[0040] In order to at least partially solve the technical problems existing in the related art, the present disclosure provides a method for determining film thickness, a method for training a film thickness estimation model, and a device. The film thickness determination method includes: obtaining initial reflection spectrum data of the film to be measured, wherein the initial reflection spectrum data is associated with the film thickness of the film to be measured; performing data conversion on the initial reflection spectrum data to obtain target reflection spectrum data; inputting the target reflection spectrum data into the film thickness estimation model to obtain the film thickness interval of the film to be measured, wherein the film thickness estimation model is obtained by training a candidate film thickness estimation model based on a simulation spectrum training set associated with the film thickness; and determining the film thickness of the film to be measured based on the film thickness interval based on a preset optimization algorithm, wherein the preset optimization algorithm includes a global optimization algorithm based on an asymptotically non-convex solver.
[0041] Figure 2 The application scenario diagram of the film thickness determination method, film thickness estimation model training method and device according to the embodiments of the present disclosure is schematically shown.
[0042] like Figure 2 As shown, the application scenario 200 according to this embodiment may include a first terminal device 201, a second terminal device 202, a third terminal device 203, a network 204, and a server 205. The network 204 is used to provide a medium for a communication link between the first terminal device 201, the second terminal device 202, the third terminal device 203, and the server 205. The network 204 may include various connection types, such as wired, wireless communication links, or optical fiber cables, etc.
[0043] The user may use the first terminal device 201, the second terminal device 202, and the third terminal device 203 to interact with the server 205 through the network 204 to receive or send messages, etc. Various communication client applications may be installed on the first terminal device 201, the second terminal device 202, and the third terminal device 203, such as shopping applications, web browser applications, search applications, instant messaging tools, email clients, social platform software, etc. (only for example).
[0044] The first terminal device 201, the second terminal device 202, and the third terminal device 203 may be various electronic devices having display screens and supporting web browsing, including but not limited to smart phones, tablet computers, laptop computers, desktop computers, and the like.
[0045] The server 205 may be a server that provides various services, such as a background management server (only as an example) that provides support for websites browsed by users using the first terminal device 201, the second terminal device 202, and the third terminal device 203. The background management server may analyze and process the received data such as user requests, and feed back the processing results (such as web pages, information, or data obtained or generated according to user requests) to the terminal device.
[0046] It should be noted that the film thickness determination method and the film thickness estimation model training method provided in the embodiment of the present disclosure can generally be executed by the server 205. Accordingly, the film thickness determination device and the film thickness estimation model training device provided in the embodiment of the present disclosure can generally be set in the server 205. The film thickness determination method and the film thickness estimation model training method provided in the embodiment of the present disclosure can also be executed by a server or server cluster that is different from the server 205 and can communicate with the first terminal device 201, the second terminal device 202, the third terminal device 203 and / or the server 205. Accordingly, the film thickness determination device and the film thickness estimation model training device provided in the embodiment of the present disclosure can also be set in a server or server cluster that is different from the server 205 and can communicate with the first terminal device 201, the second terminal device 202, the third terminal device 203 and / or the server 205. Alternatively, the film thickness determination method and the film thickness estimation model training method provided in the embodiment of the present disclosure may also be performed by the first terminal device 201, the second terminal device 202 or the third terminal device 203, or may also be performed by other terminal devices different from the first terminal device 201, the second terminal device 202 or the third terminal device 203. Accordingly, the film thickness determination device and the film thickness estimation model training device provided in the embodiment of the present disclosure may also be set in the first terminal device 201, the second terminal device 202 or the third terminal device 203, or may be set in other terminal devices different from the first terminal device 201, the second terminal device 202 or the third terminal device 203.
[0047] For example, the initial reflection spectrum data may be originally stored in any one of the first terminal device 201, the second terminal device 202, or the third terminal device 203 (for example, the first terminal device 201, but not limited thereto), or stored in an external storage device and may be imported into the first terminal device 201. Then, the first terminal device 201 may locally execute the film thickness determination method provided by the embodiment of the present disclosure, or send the initial reflection spectrum data to other terminal devices, servers, or server clusters, and the other terminal devices, servers, or server clusters that receive the initial reflection spectrum data may execute the film thickness determination method provided by the embodiment of the present disclosure.
[0048] For example, the reflection spectrum sample data may be originally stored in any one of the first terminal device 201, the second terminal device 202 or the third terminal device 203 (for example, the first terminal device 201, but not limited thereto), or stored in an external storage device and may be imported into the first terminal device 201. Then, the first terminal device 201 may locally execute the training method for the film thickness estimation model provided in the embodiment of the present disclosure, or send the reflection spectrum sample data to other terminal devices, servers, or server clusters, and the other terminal devices, servers, or server clusters that receive the reflection spectrum sample data may execute the training method for the film thickness estimation model provided in the embodiment of the present disclosure.
[0049] It should be understood that Figure 2 The number of terminal devices, networks and servers in the embodiment is only for illustration. Any number of terminal devices, networks and servers may be provided according to implementation requirements.
[0050] Figure 3 The flowchart of the method for determining the film thickness according to the embodiment of the present disclosure is schematically shown.
[0051] like Figure 3 As shown, the method 300 includes operations S310 to S340.
[0052] In operation S310, initial reflection spectrum data of the film to be tested is obtained, wherein the initial reflection spectrum data is associated with the film thickness of the film to be tested;
[0053] In operation S320, data conversion is performed on the initial reflection spectrum data to obtain target reflection spectrum data;
[0054] In operation S330, the target reflection spectrum data is input into a film thickness estimation model to obtain a film thickness range of the film to be measured, wherein the film thickness estimation model is obtained by training a candidate film thickness estimation model based on a simulation spectrum training set associated with the film thickness;
[0055] In operation S340 , based on a preset optimization algorithm, the film thickness of the film to be tested is determined according to the film thickness range, wherein the preset optimization algorithm includes a global optimization algorithm based on an asymptotically non-convex solver.
[0056] According to the embodiments of the present disclosure, the initial reflection spectrum data may be understood as interference spectrum data generated by the interference effect and collected by the spectrometer, and the initial reflection spectrum data is associated with the film thickness of the film to be measured.
[0057] According to one embodiment of the present disclosure, data conversion may be performed on the initial reflection spectrum data to convert the initial reflection spectrum data into a form suitable for input into a film thickness estimation model to obtain target reflection spectrum data.
[0058] According to an embodiment of the present disclosure, the target reflection spectrum data can be input into a film thickness estimation model to obtain a film thickness interval of the film to be measured. On this basis, the film thickness of the film to be measured can be determined based on the film thickness interval based on a preset optimization algorithm. Among them, the film thickness estimation model is obtained by training a candidate film thickness estimation model based on a simulation spectrum training set associated with the film thickness. The preset optimization algorithm includes a global optimization algorithm based on an asymptotically non-convex solver.
[0059] According to an embodiment of the present disclosure, the simulation spectrum training set may include a plurality of simulation spectrum sample data. Preferably, the number and diversity of the simulation spectrum sample data may be reasonably set so that the simulation spectrum training set can cover as large a range of film thickness as possible, thereby facilitating the generalization ability of the film thickness estimation model, and further contributing to improving the accuracy, stability and range of the film thickness estimation model. Among them, the range of the film thickness estimation model can be understood as the range of film thickness that the film thickness estimation model can accurately estimate.
[0060] According to an embodiment of the present disclosure, a Gradual Non-Convex (GNC) solver can be used to process non-convex optimization problems, wherein a non-convex optimization problem can be understood as a situation where the objective function is not a convex function, that is, the objective function may have multiple local optimal solutions rather than a global optimal solution. By introducing constraint terms, the Gradual Non-Convex solver can temporarily convert the non-convex optimization problem into a convex optimization problem for solution in the early stage of optimization, and then gradually restore the convex optimization problem to the original non-convex optimization problem by gradually adjusting the objective function and constraint conditions, thereby guiding the solution process to converge to the global optimal solution and finally obtaining the global optimal solution.
[0061] According to the embodiments of the present disclosure, by using the film thickness estimation model, the film thickness interval of the film to be measured can be obtained based on the initial reflection spectrum data of the film to be measured. By using the global optimization algorithm based on the asymptotic non-convexity solver, the film thickness of the film to be measured can be determined according to the film thickness interval. Since the film thickness estimation model is obtained by training the candidate film thickness estimation model based on the simulation spectrum training set associated with the film thickness, the number and diversity of the simulation spectrum sample data can be reasonably set so that the simulation spectrum training set can cover a larger film thickness range as much as possible, thereby helping to improve the accuracy, stability and range of the film thickness estimation model. In addition, since the global optimization algorithm based on the asymptotic non-convexity solver is used to solve the film thickness of the film to be measured, it is effectively avoided to fall into the local optimal solution in the process of solving the film thickness, thereby further improving the accuracy of the film thickness. Therefore, the film thickness determination method provided by the embodiments of the present disclosure can at least partially solve the technical problems existing in the related art such as small measurement range, low measurement result accuracy, poor stability, etc.
[0062] Reference below Figure 4~Figure 5 , combined with specific embodiments Figure 3 The method shown is further explained.
[0063] According to an embodiment of the present disclosure, based on a preset optimization algorithm, according to the film thickness range, determining the film thickness of the film to be tested includes: determining an initial value of the film thickness according to the film thickness range; based on a preset objective function, determining an objective function value according to the initial value of the film thickness; and based on a preset parameter adjustment strategy, adjusting the parameters of the global optimization algorithm according to the objective function value until a predetermined end condition is met to obtain the film thickness of the film to be tested.
[0064] According to an embodiment of the present disclosure, the initial value of the film thickness can be used as the initial solution in the above-mentioned global optimization algorithm solution process. The objective function value can be determined based on the preset objective function and the initial value of the film thickness. On this basis, the parameters of the global optimization algorithm can be adjusted based on the preset parameter adjustment strategy and the objective function value to guide the solution process to converge to the global optimal solution until the predetermined end condition is met, and the film thickness of the film to be measured is obtained.
[0065] It can be understood that, since the film thickness interval defines the upper and lower limits of the actual film thickness, and since the initial film thickness value is determined based on the film thickness interval, the initial film thickness value can be closer to the actual film thickness than a randomly selected initial value.
[0066] According to the embodiments of the present disclosure, in the process of solving the film thickness of the film to be tested based on a preset optimization algorithm, since the initial value of the film thickness is determined based on the film thickness range, it is closer to the actual film thickness than the randomly set initial value of the film thickness. Therefore, improper selection of the initial value can be effectively avoided, which helps to obtain the global optimal solution more quickly and accurately.
[0067] According to an embodiment of the present disclosure, wherein the preset objective function includes a residual term and a constraint term, based on the preset objective function and according to the initial value of the film thickness, determining the objective function value includes: determining the first sub-objective function value according to the residual term weight and the initial value of the film thickness; determining the second sub-objective function value according to the constraint term control parameter and the residual term weight; and determining the objective function value according to the first sub-objective function value and the second sub-objective function value.
[0068] According to one embodiment of the present disclosure, the preset objective function may be as follows:
[0069]
[0070] In formula (1)~(2), is the residual term, is the residual term weight, is the constraint control parameter, is a constant, x is the initial value of the film thickness, is the value of the first sub-objective function, is the value of the second sub-objective function.
[0071] According to an embodiment of the present disclosure, in the solving process of the progressive non-convexity solver, the original non-convex optimization problem can be temporarily converted into a convex optimization problem by introducing a constraint term, and the solution of the convex optimization problem is obtained, and the constraint term control parameters and the residual term weights are gradually changed in subsequent iterations until the convex optimization problem is restored to the original non-convex optimization problem, thereby obtaining a solution to the non-convex problem. The solution obtained in each iteration is used as the initial solution of the subsequent iteration.
[0072] It can be understood that in each round of iterative solution mentioned above, the asymptotically non-convex solver can ensure that there is always an initial value that is sufficiently close to the true global optimal solution. Therefore, the global optimization algorithm based on the asymptotically non-convex solver of the embodiment of the present disclosure can effectively avoid local optimal solutions.
[0073] According to an embodiment of the present disclosure, based on a preset parameter adjustment strategy, according to the objective function value, the parameters of the global optimization algorithm are adjusted until a predetermined end condition is met, and the film thickness of the film to be measured is obtained, including: when the predetermined end condition is not met, iteratively performing the following operations: based on the preset parameter adjustment strategy, according to the objective function value, adjusting the control parameters and the constraint item weights to obtain adjusted control parameters and adjusted constraint item weights; updating the objective function according to the adjusted control parameters and the adjusted constraint item weights to obtain an updated objective function; obtaining an updated objective function value according to the current objective function value and the updated objective function; and in response to satisfying the predetermined end condition, determining the updated objective function value as the film thickness of the film to be measured.
[0074] Figure 4 The schematic diagram shows a principle diagram of a progressive non-convex solver according to an embodiment of the present disclosure.
[0075] As an example, Figure 4 As shown, curve 403 can represent the original non-convex optimization problem. In the early stage of the solution, the curve 403 can be converted into curve 401 by adjusting the constraint control parameters and the residual weights, so as to temporarily convert the non-convex optimization problem into a convex optimization problem, thereby facilitating the solution. In curve 401, the initial value of the film thickness can be used as the initial value for the solution, and the global optimal solution (i.e., point A) of curve 401 can be easily solved. In the next iteration, the constraint control parameters and the residual weights can be further adjusted to convert curve 401 into 402 to increase the non-convexity of the objective function. In curve 402, point A can be used as the initial value for the solution to solve the global optimal solution (i.e., point B) of curve 402. Since the global optimal solution of curve 401 is used as the initial value for solving the global optimal solution of curve 402, even if curve 402 increases the non-convexity, the global optimal solution can still be found. Through multiple iterations of solving, the constraint control parameters and the residual weights are continuously adjusted according to the objective function value until the original curve 403 is restored, so that the global optimal solution (ie, point C) of the non-convex optimization problem can be solved.
[0076] According to an embodiment of the present disclosure, the preset parameter adjustment strategy may include: allowing a looser constraint strength to avoid local optimality in the initial stage of the solution, and strengthening the constraint strength as the iteration progresses to ensure the solution accuracy. Furthermore, in the solution process, based on an adaptive adjustment mechanism (for example, based on a learning rate), the constraint control parameters and residual item weights can be dynamically adjusted according to the objective function value, the convergence speed and stability can be optimized, the solution process is guided to converge to the global optimal point, and finally the global optimal solution is calculated to obtain the film thickness of the film to be tested.
[0077] As an example, Table 1 shows the film thickness determined by the film thickness determination method provided by the embodiment of the present disclosure for different types and thicknesses of films to be tested. As shown in Table 1, the film thickness determination method provided by the embodiment of the present disclosure has a wide measurement range, high measurement result accuracy, and good measurement result stability in practical applications.
[0078]
[0079] Table 1
[0080] According to an embodiment of the present disclosure, the initial reflection spectrum data includes M initial reflection light data points, where M is a positive integer, and each initial reflection light data point is used to characterize the light intensity information of the reflected light of the film to be tested under the selected light source wavelength. The initial reflection spectrum data is converted to obtain the target reflection spectrum data, including: normalizing the M initial reflection light data points according to the light source wavelength and light intensity information of each of the M initial reflection light data points to obtain M normalized reflection light data points corresponding to the M initial reflection light data points; determining the feature dimension and data length corresponding to the initial reflection spectrum data according to the M normalized reflection light data points; and determining the target reflection spectrum data according to a preset batch size, feature dimension and data length.
[0081] Figure 5 The initial reflection spectrum data according to an embodiment of the present disclosure is exemplarily shown.
[0082] like Figure 5 As shown, the initial reflection spectrum data includes M initial reflection light data points, wherein for each initial reflection light data point, the horizontal coordinate of the initial reflection light data point is the wavelength of the light source, and the vertical coordinate of the initial reflection light data point is the intensity of the reflection light of the film to be tested at the aforementioned wavelength of the light source.
[0083] According to an embodiment of the present disclosure, the M initial reflected light data points can be normalized according to the light source wavelength and light intensity information of each of the M initial reflected light data points to obtain M normalized reflected light data points corresponding to the M initial reflected light data points. On this basis, the feature dimension (Feat Dim) and data length (Data Length) corresponding to the initial reflected light spectrum data can be determined according to the M normalized reflected light data points to obtain normalized reflected light spectrum data X' with a dimension of (Feat Dim, Data Length). Furthermore, the two-dimensional array X' can be converted into a three-dimensional tensor I according to the preset batch size, feature dimension and data length to adapt to the input requirements of the film thickness estimation model.
[0084] Exemplarily, the normalization formula may be as follows:
[0085]
[0086] In formula (3), is the normalized reflected light data point, is the initial reflected light data point.
[0087] For example, the formula for converting a two-dimensional array X' into a three-dimensional tensor I may be as follows:
[0088]
[0089] In formula (4), X' is the normalized reflectance spectrum data, and I is the target reflectance spectrum data.
[0090] According to an embodiment of the present disclosure, the target reflection spectrum data I may be input into a film thickness estimation model to output a film thickness interval (L, U).
[0091] According to an embodiment of the present disclosure, the film thickness interval includes an interval lower limit value and an interval upper limit value. According to the film thickness interval, determining the initial value of the film thickness includes: determining the interval middle value of the film thickness interval according to the interval lower limit value and the interval upper limit value; and determining the interval middle value as the initial value of the film thickness.
[0092] According to an embodiment of the present disclosure, determining the middle value of the film thickness interval as the initial value of the film thickness can reduce the initial search range, avoid instability that may be caused by extreme values, and help the algorithm converge to the optimal solution more quickly.
[0093] For example, if the upper limit of the film thickness interval (L, U) is U and the lower limit is L, then the middle value of the film thickness interval can be determined to be (L+U) / 2. Further, the initial value of the film thickness can be determined to be (L+U) / 2.
[0094] Figure 6 The flowchart of the method for training a film thickness estimation model according to an embodiment of the present disclosure is schematically shown.
[0095] like Figure 6 As shown, the method 600 includes operations S610 to S630.
[0096] In operation S610, based on the interference effect principle and the correlation between the film thickness and the reflected light, the reflection spectrum sample data of each of N sample films is obtained, where N is a positive integer;
[0097] In operation S620, the reflection spectrum sample data of each of the N sample films are processed to obtain a simulation spectrum training set associated with the film thickness;
[0098] In operation S630, a candidate film thickness estimation model is trained according to the simulation spectrum training set to obtain a film thickness estimation model, wherein the candidate film thickness estimation model includes a residual network model.
[0099] According to the embodiments of the present disclosure, since the film thickness estimation model is trained based on a simulation spectrum training set, and since the simulation spectrum training set is obtained by processing the reflection spectrum sample data of each of N sample films, the number and diversity of the reflection spectrum sample data can be reasonably set so that the simulation spectrum training set can cover as large a film thickness range as possible, which is beneficial to improving the generalization ability of the film thickness estimation model, and further helps to improve the accuracy, stability and range of the film thickness estimation model.
[0100] Figure 7 The residual network model according to an embodiment of the present disclosure is schematically shown.
[0101] According to an embodiment of the present disclosure, the residual network model includes a convolutional layer, at least one residual module, at least one batch normalization layer, at least one activation function layer, an average pooling layer and a fully connected layer.
[0102] As an example, Figure 7 As shown, the residual network model can include an input layer, a convolution layer, a batch normalization layer, an activation function layer, a residual block group layer, a batch normalization layer, an activation function layer, an average pooling layer, and a fully connected output layer in sequence. Specifically, after the input data is extracted through the convolution layer to realize preliminary feature extraction, it is sent to the batch normalization layer, and the normalized feature distribution is used to accelerate the convergence speed and improve the generalization. Then, it passes through the activation function layer (for example, the ReLU activation function can be selected) to increase the nonlinear processing capability. After that, the data continues to be passed to the multiple residual blocks of the residual block group layer. Each residual block directly adds the input to the convolution output through a skip connection to promote the effective propagation of deep features. By learning the residual between input and output through residual blocks, the gradient vanishing and explosion problems in training can be avoided. In a specific residual block, downsampling operations are implemented to increase feature channels and expand the receptive field of the model. The output after all residual blocks passes through batch normalization and activation function layers again. Then the features of each channel are subjected to global average pooling operations to reduce the feature dimension to avoid overfitting. The fully connected output layer maps the final features to the category prediction output.
[0103] According to an embodiment of the present disclosure, a simulation spectrum training set can be input into a residual network model to extract features from the spectrum data. The input simulation spectrum training set contains multiple classes, each of which corresponds to a preset film thickness range. The predicted probability distribution of each class can be calculated, and the one with the highest probability is taken as the classification result, and the code i used to represent the classification result is output. The model parameters of the residual network model can be adjusted according to the output code i until the preset end condition is met to obtain a film thickness estimation model. Furthermore, the initial values of the upper and lower bounds of the film thickness (L, U) can be determined according to the predefined code-film thickness range conversion algorithm.
[0104] According to an embodiment of the present disclosure, the reflection spectrum samples of each of N sample films are processed to obtain a simulation spectrum training set associated with the film thickness, including: based on preset training set parameters, a first intermediate training set is constructed according to the reflection spectrum samples of each of the N sample films, wherein the preset training set parameters include the training set category, the film thickness interval length and the number of training samples; random noise is added to the first intermediate training set to obtain a second intermediate training set; and the second intermediate training set is normalized to obtain a simulation spectrum training set.
[0105] As an example, Table 2 exemplarily shows the preset training parameters of the first intermediate training set.
[0106]
[0107] Table 2
[0108] According to an embodiment of the present disclosure, random noise may be added to the first intermediate training set to obtain a second intermediate training set. It should be noted that the method for normalizing the second intermediate training set may refer to the method for normalizing the M initial reflected light data points. For example, based on the above formula (3), the second intermediate training set may be normalized to obtain a simulated spectrum training set (such as Figure 8 as shown).
[0109] According to an embodiment of the present disclosure, a method for determining the correlation between film thickness and reflected light includes: determining the electric field vector of the reflected light on the upper surface of the preset film based on the principle of interference effect and a characteristic matrix of the preset film, wherein the preset film includes at least one dielectric layer, the characteristic matrix is used to describe the refractive index and thickness of each of the at least one dielectric layer, and the electric field vector is used to characterize the amplitude information and phase information of the reflected light; and establishing the correlation between film thickness and reflected light based on the characteristic matrix, as well as the amplitude information and phase information of the reflected light.
[0110] According to one embodiment of the present disclosure, the correlation between the film thickness and the reflected light can be expressed as:
[0111]
[0112] In formula (5), is the combined amplitude of the reflected light, is the phase thickness of the film layer, is the wavelength, n is the refractive index, d is the thickness, , are the Fresnel reflection coefficients at the top and bottom boundaries of the film, respectively, is the amplitude of the light source.
[0113] Based on the above-mentioned film thickness determination method, the present disclosure also provides a film thickness determination device. Fig. 9 The structure block diagram of the device for determining the film thickness according to an embodiment of the present disclosure is schematically shown.
[0114] like Fig. 9 As shown, the film thickness determination device 900 may include an acquisition module 910 , a conversion module 920 , an input module 930 , and a determination module 940 .
[0115] The acquisition module 910 is used to acquire initial reflection spectrum data of the film to be measured, wherein the initial reflection spectrum data is associated with the film thickness of the film to be measured.
[0116] The conversion module 920 is used to perform data conversion on the initial reflection spectrum data to obtain target reflection spectrum data.
[0117] The input module 930 is used to input the target reflection spectrum data into the film thickness estimation model to obtain the film thickness range of the film to be measured, wherein the film thickness estimation model is obtained by training the candidate film thickness estimation model based on the simulation spectrum training set associated with the film thickness.
[0118] The determination module 940 is used to determine the film thickness of the film to be tested according to the film thickness range based on a preset optimization algorithm, wherein the preset optimization algorithm includes a global optimization algorithm based on an asymptotically non-convex solver.
[0119] According to an embodiment of the present disclosure, the determination module 940 may include an initial value determination submodule, an objective function value determination submodule, and a parameter adjustment submodule.
[0120] The initial value determination submodule is used to determine the initial value of the film thickness according to the film thickness range.
[0121] The objective function value determination submodule is used to determine the objective function value based on a preset objective function and an initial value of the film thickness.
[0122] The parameter adjustment submodule is used to adjust the parameters of the global optimization algorithm based on a preset parameter adjustment strategy and the objective function value until a predetermined end condition is met to obtain the film thickness of the film to be measured.
[0123] According to an embodiment of the present disclosure, the initial value determination submodule may include an intermediate value determination unit and an initial value determination unit.
[0124] The intermediate value determination unit is used to determine the interval intermediate value of the film thickness interval according to the interval lower limit value and the interval upper limit value.
[0125] The initial value determination unit is used to determine the middle value of the interval as the initial value of the film thickness.
[0126] According to an embodiment of the present disclosure, the objective function value determination submodule may include a first determination unit, a second determination unit, and a third determination unit.
[0127] The first determination unit is used to determine the first sub-objective function value according to the residual term weight and the initial value of the film thickness.
[0128] The second determination unit is used to determine the value of the second sub-objective function according to the constraint item control parameter and the residual item weight.
[0129] The third determining unit is used to determine the objective function value according to the first sub-objective function value and the second sub-objective function value.
[0130] According to an embodiment of the present disclosure, the parameter adjustment submodule may include an adjustment unit, an objective function updating unit, an objective function value updating unit, and a film thickness determination unit.
[0131] The adjustment unit is used to adjust the control parameters and constraint item weights based on the preset parameter adjustment strategy and the objective function value to obtain adjusted control parameters and adjusted constraint item weights.
[0132] The objective function updating unit is used to update the objective function according to the adjusted control parameters and the adjusted constraint item weights to obtain an updated objective function.
[0133] The objective function value updating unit is used to obtain an updated objective function value according to the current objective function value and the updated objective function.
[0134] The film thickness determination unit is used to determine the updated objective function value as the film thickness of the film to be measured in response to meeting a predetermined end condition.
[0135] According to an embodiment of the present disclosure, the conversion module 920 may include a first normalization processing submodule, a first determination submodule, and a second determination submodule.
[0136] The first normalization processing submodule is used to perform normalization processing on the M initial reflected light data points according to the light source wavelength and light intensity information of each of the M initial reflected light data points to obtain M normalized reflected light data points corresponding to the M initial reflected light data points.
[0137] The first determination submodule is used to determine the feature dimension and data length corresponding to the initial reflection spectrum data according to the M normalized reflection light data points.
[0138] The second determination submodule is used to determine the target reflection spectrum data according to a preset batch size, feature dimension and data length.
[0139] Based on the above-mentioned film thickness determination method, the present disclosure also provides a training device for a film thickness estimation model. Fig.10 The structural block diagram of a training device for a film thickness estimation model according to an embodiment of the present disclosure is schematically shown.
[0140] like Fig.10 As shown, the training apparatus 1000 for the film thickness estimation model may include a sample module 1010 , a processing module 1020 , and a training module 1030 .
[0141] The sample module 1010 is used to obtain the reflection spectrum sample data of each of N sample films based on the interference effect principle and the correlation between the film thickness and the reflected light, where N is a positive integer.
[0142] The processing module 1020 is used to process the reflection spectrum sample data of each of the N sample films to obtain a simulation spectrum training set associated with the film thickness.
[0143] The training module 1030 is used to train a candidate film thickness estimation model according to the simulation spectrum training set to obtain a film thickness estimation model, wherein the candidate film thickness estimation model includes a residual network model.
[0144] According to an embodiment of the present disclosure, the processing module 1020 may include a construction submodule, a noise submodule and a second normalization submodule.
[0145] The construction submodule is used to construct a first intermediate training set based on preset training set parameters and the reflection spectrum samples of each of the N sample films, wherein the preset training set parameters include the training set category, the film thickness interval length and the number of training samples.
[0146] The noise submodule is used to add random noise to the first intermediate training set to obtain a second intermediate training set.
[0147] The second normalization submodule is used to perform normalization processing on the second intermediate training set to obtain a simulated spectrum training set.
[0148] According to the embodiments of the present invention, any one or more of the modules, submodules, units, and subunits, or at least part of the functions of any one of them can be implemented in one module. According to the embodiments of the present invention, any one or more of the modules, submodules, units, and subunits can be split into multiple modules for implementation. According to the embodiments of the present invention, any one or more of the modules, submodules, units, and subunits can be at least partially implemented as hardware circuits, such as field programmable gate arrays (FPGAs), programmable logic arrays (PLAs), systems on chips, systems on substrates, systems on packages, application specific integrated circuits (ASICs), or can be implemented by hardware or firmware in any other reasonable way of integrating or packaging circuits, or can be implemented in any one of the three implementation methods of software, hardware, and firmware, or in any appropriate combination of any of them. Alternatively, according to the embodiments of the present invention, one or more of the modules, submodules, units, and subunits can be at least partially implemented as computer program modules, and when the computer program modules are run, the corresponding functions can be performed.
[0149] For example, any multiple of the acquisition module 910, the conversion module 920, the input module 930, and the determination module 940 can be combined in one module / unit / sub-unit for implementation, or any one of the modules / units / sub-units can be split into multiple modules / units / sub-units. Alternatively, at least part of the functions of one or more of these modules / units / sub-units can be combined with at least part of the functions of other modules / units / sub-units and implemented in one module / unit / sub-unit. According to an embodiment of the present disclosure, at least one of the acquisition module 910, the conversion module 920, the input module 930, and the determination module 940 can be at least partially implemented as a hardware circuit, such as a field programmable gate array (FPGA), a programmable logic array (PLA), a system on a chip, a system on a substrate, a system on a package, an application specific integrated circuit (ASIC), or can be implemented by hardware or firmware such as any other reasonable way of integrating or packaging the circuit, or by any one of the three implementation methods of software, hardware, and firmware, or by a suitable combination of any of them. Alternatively, at least one of the acquisition module 910 , the conversion module 920 , the input module 930 , and the determination module 940 may be at least partially implemented as a computer program module, and when the computer program module is executed, a corresponding function may be performed.
[0150] For example, any multiple of the sample module 1010, the processing module 1020, and the training module 1030 can be combined in one module / unit / sub-unit for implementation, or any one of the modules / units / sub-units can be split into multiple modules / units / sub-units. Alternatively, at least part of the functions of one or more of these modules / units / sub-units can be combined with at least part of the functions of other modules / units / sub-units and implemented in one module / unit / sub-unit. According to an embodiment of the present disclosure, at least one of the sample module 1010, the processing module 1020, and the training module 1030 can be at least partially implemented as a hardware circuit, such as a field programmable gate array (FPGA), a programmable logic array (PLA), a system on a chip, a system on a substrate, a system on a package, an application specific integrated circuit (ASIC), or can be implemented by hardware or firmware such as any other reasonable way of integrating or packaging the circuit, or implemented in any one of the three implementation methods of software, hardware, and firmware, or in any appropriate combination of any of them. Alternatively, at least one of the sample module 1010, the processing module 1020, and the training module 1030 may be at least partially implemented as a computer program module, and when the computer program module is executed, the corresponding function may be performed.
[0151] It should be noted that the film thickness determination device part in the embodiment of the present disclosure corresponds to the film thickness determination method part in the embodiment of the present disclosure, and the description of the film thickness determination device part specifically refers to the film thickness determination method part, which will not be repeated here. The film thickness estimation model training device part in the embodiment of the present disclosure corresponds to the film thickness estimation model training method part in the embodiment of the present disclosure, and the description of the film thickness estimation model training device part specifically refers to the film thickness estimation model method part, which will not be repeated here.
[0152] The flowcharts and block diagrams in the accompanying drawings illustrate the possible implementation architecture, functions and operations of the systems, methods and computer program products according to various embodiments of the present disclosure. In this regard, each box in the flowchart or block diagram may represent a module, a program segment, or a part of a code, and the above-mentioned module, program segment, or a part of the code contains one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box may also occur in an order different from that marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram or flowchart, and the combination of boxes in the block diagram or flowchart, can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions. It can be understood by those skilled in the art that the features recorded in the various embodiments of the present disclosure can be combined and / or combined in a variety of ways, even if such a combination or combination is not explicitly recorded in the present disclosure. In particular, without departing from the spirit and teaching of the present disclosure, the features described in the various embodiments of the present disclosure may be combined and / or combined in a variety of ways. All of these combinations and / or combinations fall within the scope of the present disclosure.
[0153] The embodiments of the present disclosure are described above. However, these embodiments are only for illustrative purposes and are not intended to limit the scope of the present disclosure. Although the embodiments are described above, this does not mean that the measures in the various embodiments cannot be used in combination to advantage. Without departing from the scope of the present disclosure, those skilled in the art may make a variety of substitutions and modifications, which should all fall within the scope of the present disclosure.
Claims
1. A method for determining film thickness, comprising: Acquiring initial reflection spectrum data of the film to be tested, wherein the initial reflection spectrum data is associated with the film thickness of the film to be tested; Performing data conversion on the initial reflection spectrum data to obtain target reflection spectrum data; Inputting the target reflection spectrum data into a film thickness estimation model to obtain a film thickness range of the film to be measured, wherein the film thickness estimation model is obtained by training a candidate film thickness estimation model based on a simulation spectrum training set associated with the film thickness; and Based on a preset optimization algorithm, the film thickness of the film to be tested is determined according to the film thickness range, wherein the preset optimization algorithm includes a global optimization algorithm based on an asymptotically non-convex solver.
2. The method according to claim 1, wherein: The step of determining the film thickness of the film to be measured based on the preset optimization algorithm and according to the film thickness range includes: Determining an initial value of the film thickness according to the film thickness range; Based on a preset objective function, determining an objective function value according to the initial value of the film thickness; and Based on a preset parameter adjustment strategy, the parameters of the global optimization algorithm are adjusted according to the objective function value until a predetermined end condition is met, thereby obtaining the film thickness of the film to be measured.
3. The method according to claim 2, wherein: The preset objective function includes a residual term and a constraint term. The determining of the objective function value based on the preset objective function and according to the initial value of the film thickness includes: Determining a first sub-objective function value according to a residual term weight and an initial value of the film thickness; Determining a second sub-objective function value according to the constraint item control parameter and the residual item weight; and The objective function value is determined according to the first sub-objective function value and the second sub-objective function value.
4. The method according to claim 3, wherein: The step of adjusting the parameters of the global optimization algorithm based on the preset parameter adjustment strategy and according to the objective function value until a predetermined end condition is met to obtain the film thickness of the film to be measured comprises: If the predetermined end condition is not met, the iteration performs the following operations: Based on a preset parameter adjustment strategy, the control parameter and the constraint item weight are adjusted according to the objective function value to obtain an adjusted control parameter and an adjusted constraint item weight; updating the objective function according to the adjusted control parameter and the adjusted constraint item weight to obtain an updated objective function; Obtaining an updated objective function value according to the current objective function value and the updated objective function; and In response to satisfying the predetermined end condition, the updated objective function value is determined as the film thickness of the film to be measured.
5. The method according to claim 1, wherein: The initial reflection spectrum data includes M initial reflection light data points, where M is a positive integer, and each of the initial reflection light data points is used to characterize the light intensity information of the reflected light of the film to be tested under the selected light source wavelength. The data conversion of the initial reflection spectrum data to obtain the target reflection spectrum data includes: Normalizing the M initial reflected light data points according to the light source wavelength and light intensity information of each of the M initial reflected light data points to obtain M normalized reflected light data points corresponding to the M initial reflected light data points one by one; Determining a characteristic dimension and a data length corresponding to the initial reflection spectrum data according to the M normalized reflection light data points; and The target reflection spectrum data is determined according to a preset batch size, the feature dimension and the data length.
6. The method according to claim 2, wherein: The film thickness interval includes a lower limit value and an upper limit value, and determining the initial value of the film thickness according to the film thickness interval includes: Determining a middle value of the film thickness interval according to the interval lower limit value and the interval upper limit value; and The middle value of the interval is determined as the initial value of the film thickness.
7. A method for training a film thickness estimation model, comprising: Based on the interference effect principle and the correlation between film thickness and reflected light, the reflectance spectrum sample data of N sample films are obtained, where N is a positive integer. Processing the reflection spectrum sample data of each of the N sample films to obtain a simulation spectrum training set associated with the film thickness; and According to the simulation spectrum training set, a candidate film thickness estimation model is trained to obtain the film thickness estimation model, wherein the candidate film thickness estimation model includes a residual network model.
8. The method according to claim 7, wherein: The residual network model includes a convolutional layer, at least one residual module, at least one batch normalization layer, at least one activation function layer, an average pooling layer and a fully connected layer.
9. The method according to claim 7, wherein: The processing of the reflection spectrum samples of each of the N sample films to obtain a simulation spectrum training set associated with the film thickness includes: Based on preset training set parameters, a first intermediate training set is constructed according to the reflection spectrum samples of each of the N sample films, wherein the preset training set parameters include a training set category, a film thickness interval length, and a number of training samples; adding random noise to the first intermediate training set to obtain a second intermediate training set; and The second intermediate training set is normalized to obtain the simulated spectrum training set.
10. The method according to claim 7, wherein: The method for determining the correlation between the film thickness and the reflected light includes: Determining an electric field vector of reflected light from an upper surface of a preset film based on the interference effect principle and a characteristic matrix of the preset film, wherein the preset film includes at least one dielectric layer, the characteristic matrix is used to describe the refractive index and thickness of each of the at least one dielectric layer, and the electric field vector is used to characterize amplitude information and phase information of the reflected light; and According to the characteristic matrix, and the amplitude information and phase information of the reflected light, a correlation relationship between the film thickness and the reflected light is established.
Citation Information
Cited By
Intelligent wide-spectrum film thickness real-time monitoring system and method
CN120778013A
Intelligent wide-spectrum film thickness real-time monitoring system and method
CN120778013B
Atomic film thickness measuring method based on quasi Brewster angle spectrum
CN121230633A
