Computer program, information processing method, and information processing device

Abstract

Provided is a computer program, an information processing method, and an information processing apparatus that are expected to be able to estimate a state of substrate processing based on spectrum data measured at the time of substrate processing. The computer program according to the present embodiment causes a computer to execute the following processing: acquiring spectrum data measured with respect to substrate processing, separating the acquired spectrum data into line spectrum data with respect to a plurality of reaction species, and predicting at least one of an electron density, an electron temperature, and an electron energy distribution of each reaction species based on the separated line spectrum data.

Description

Technical Field

[0001] This disclosure relates to computer programs, information processing methods, and information processing devices. Background Technology

[0002] Patent Document 1 proposes a film thickness measurement method. This method stores information in a storage unit that indicates the relationship between the absorbance spectrum of a substrate film and the film thickness. It measures the absorbance spectrum of a substrate after substrate treatment and, based on the stored information, derives the film thickness of the film existing on the surface of the substrate after substrate treatment from the measured absorbance spectrum.

[0003] Patent documents

[0004] Patent document 1: Japanese Patent Application Publication No. 2023-169638. Summary of the Invention

[0005] This disclosure provides a computer program, information processing method, and information processing apparatus that are expected to be able to infer the state of substrate processing based on spectral data measured during substrate processing.

[0006] According to a computer program of one embodiment, a computer performs the following processing: acquiring spectral data measured with respect to substrate processing, separating the acquired spectral data into line spectral data for a plurality of reaction species, and predicting at least one of electron density, electron temperature, and electron energy distribution for each reaction species based on the separated line spectral data.

[0007] According to this disclosure, it is possible to infer the state of substrate processing based on spectral data measured during substrate processing. Attached Figure Description

[0008] Figure 1 This is a schematic diagram used to illustrate the general outline of the information processing system involved in this embodiment.

[0009] Figure 2 This is a block diagram illustrating a structural example of the information processing apparatus according to this embodiment.

[0010] Figure 3 This is a schematic diagram illustrating a structural example of a separation model used in spectral separation processing.

[0011] Figure 4 This is a schematic diagram illustrating an example of machine learning using a separation model.

[0012] Figure 5 This is a schematic diagram of an example of a table showing the types of spectral widths used to determine the magnitude of the electronic temperature that influences the prediction of a reaction species.

[0013] Figure 6 This is a schematic diagram illustrating an example of a table used to determine the magnitude of the electron temperature corresponding to the natural width and the Stark width.

[0014] Figure 7 This is a schematic diagram used to illustrate line spectral data and Doppler width as an example.

[0015] Figure 8 This is a flowchart illustrating an example of the steps involved in the electronic temperature magnitude prediction processing performed by the information processing apparatus according to this embodiment.

[0016] Figure 9 This is a schematic diagram illustrating a structural example of a reaction species model.

[0017] Figure 10 This is a flowchart illustrating an example of the steps involved in the electron density prediction processing performed by the information processing apparatus according to this embodiment.

[0018] Figure 11 This is a schematic diagram representing an example of time series data. Detailed Implementation

[0019] The following description, with reference to the accompanying drawings, will illustrate specific examples of the information processing system according to the embodiments of this disclosure. It should be noted that this disclosure is not limited to these examples, but rather to all modifications shown in the claims and intended to include those equivalent in meaning and scope to the claims.

[0020] <System Overview>

[0021] Figure 1This is a schematic diagram illustrating the general structure of the information processing system according to this embodiment. The information processing system according to this embodiment comprises an information processing apparatus 1 and a substrate processing apparatus 3. The substrate processing apparatus 3 is, for example, an apparatus that generates plasma within a chamber housing a semiconductor substrate (wafer) and performs various substrate processing operations such as CVD (Chemical Vapor Deposition) or etching to shape the substrate. The substrate processing apparatus 3 has a measuring instrument within the chamber that measures the plasma emission spectrum, performs measurements repeatedly at predetermined intervals, and sends the spectral data obtained through the measurements to the information processing apparatus 1. The spectral data is data related to light emitted from the excited state of atoms or molecules within the plasma. Depending on the type of atom, the type of molecule, or the plasma state, the wavelength of the emitted light varies; utilizing this property, the spectral data can be used for evaluating time variations in specific processes and information about the medium within the chamber. However, because the measuring instrument located at a specific position within the chamber of the substrate processing apparatus 3 measures the spectrum of all light within its sensitivity range, the spectral data includes data related to phenomena other than those being evaluated.

[0022] The information processing device 1 acquires spectral data measured by the measuring instrument of the substrate processing device 3, analyzes and evaluates the substrate processing status based on the acquired spectral data, and controls the operation of the substrate processing device 3 according to the analysis and evaluation results. The information processing device 1 can be an integrated device with the substrate processing device 3 or a separate device. Furthermore, there can be multiple substrate processing devices 3 that perform evaluation and control of substrate processing by the information processing device 1.

[0023] As described above, the spectral data obtained by the measuring instrument of the substrate processing apparatus 3 is data containing multiple spectra resulting from the mixing of multiple reaction species (atomic types and molecular types, etc.). Therefore, the information processing apparatus 1 according to this embodiment performs spectral separation processing to separate the spectral data obtained from the substrate processing apparatus 3 into spectral data for each reaction species (hereinafter referred to as line spectral data). In this embodiment, the information processing apparatus 1 uses a learning model generated in advance through machine learning to separate the spectral data obtained from the substrate processing apparatus 3 into line spectral data for each specified reaction species and spectral data other than the specified reaction species (hereinafter referred to as background spectral data). The learning model accepts the spectral data as input and performs machine learning in advance to separate the spectral data into multiple line spectral data and background spectral data.

[0024] Next, the information processing apparatus 1 according to this embodiment performs processing to predict the order of magnitude (number of bits) of the electron temperature for each separated line spectrum data and background spectrum data. The information processing apparatus 1 predicts the order of magnitude of the electron temperature based on the spectral width of the line spectrum data of the reaction. The information processing apparatus appropriately uses, for example, Doppler broadening (Doppler width), natural width, or Stark broadening (Stark width) as the spectral width of each reaction according to the substrate processing conditions, etc., and can calculate the spectral width of each reaction. The information processing apparatus 1 predicts the order of magnitude of the electron temperature for each reaction by performing a prescribed calculation based on the calculated spectral width or by referring to a prescribed table. In addition, the information processing apparatus 1 predicts the order of magnitude of the electron temperature by performing fitting assuming that the background spectrum data follows, for example, a Boltzmann distribution.

[0025] Furthermore, when searching for the optimal value in the prediction of electron density, which is to be processed in the next stage, the information processing device 1 uses the magnitude of the predicted electron temperature to limit the search range. This allows the information processing device 1 to perform searches and other processing efficiently, potentially reducing the workload. However, the information processing device 1 may also predict the electron density for the next stage without predicting the magnitude of the electron temperature.

[0026] Next, the information processing apparatus 1 according to this embodiment performs processing to predict the electron density of each reaction species based on the separated line spectral data and background spectral data of each reaction species and the predicted electron temperature magnitude. The information processing apparatus 1 pre-stores a reaction species model, which takes the electron temperature and electron density as input and the luminescence intensity as output for each reaction species. The information processing apparatus 1 can predict the electron density and electron temperature of each reaction species by searching for the input electron temperature and electron density so that the difference between the output of the reaction species model and the line spectral data of each reaction species becomes less than a threshold. In addition, the information processing apparatus 1 can limit the search range of electron temperature by utilizing the magnitude of the electron temperature predicted in the previous processing during this search.

[0027] The information processing device 1 stores the predicted results of electron temperature, electron density, and electron temperature obtained through these processes into the prediction result DB (database) 5. The information processing device 1 repeatedly acquires spectral data from the substrate processing device 3 and makes the aforementioned predictions based on the acquired spectral data at a predetermined period, storing and accumulating time-series data of the repeated prediction results in the prediction result DB5. Using the time-series data stored in the prediction result DB5, the information processing device 1 predicts the values ​​of electron density and electron temperature at the next opportune moment based on changes in past values. By feeding back information such as the predicted values ​​based on the time-series data or the range calculated based on those predicted values ​​to the aforementioned prediction processing of electron density and electron temperature, the information processing device 1 can further limit the search range for electron density and electron temperature.

[0028] The information processing apparatus 1 according to this embodiment can adjust settings such as pressure, gas flow rate, or power during substrate processing performed by the substrate processing apparatus 3 based on the predicted electron density and electron temperature of each reaction species, and control the substrate processing. Furthermore, the information processing apparatus 1 can display the predicted electron density and electron temperature of each reaction species, as well as information such as the time-series changes of the predicted results, on a display or the like for the user.

[0029] <Device Structure>

[0030] Figure 2 This is a block diagram illustrating a structural example of the information processing apparatus 1 according to this embodiment. The information processing apparatus 1 according to this embodiment can be implemented, for example, by installing a specified application program on a general-purpose information processing device such as a personal computer or server computer. The information processing apparatus 1 according to this embodiment is configured with a processing unit 11, a storage unit 12, a communication unit 13, a display unit 14, and an operation unit 15. Furthermore, in this embodiment, processing is described as being performed in a single information processing apparatus 1, but the processing by the information processing apparatus 1 can also be performed by multiple devices.

[0031] The processing unit 11 is constructed using computing devices such as CPU (Central Processing Unit), MPU (Micro-Processing Unit), GPU (Graphics Processing Unit), or quantum processor, as well as ROM (Read Only Memory) and RAM (Random Access Memory). The processing unit 11 performs various processes by reading and executing program 12a stored in the storage unit 12, including processing the acquisition of spectral data from the substrate processing device 3, separating the spectral data into line spectral data for each reaction species, predicting the electron temperature magnitude of each reaction species, and predicting the electron density and electron temperature of each reaction species.

[0032] The storage unit 12 is configured using a high-capacity storage device such as a hard disk or an SSD (Solid State Drive). The storage unit 12 stores various programs executed by the processing unit 11, as well as various data required for processing by the processing unit 11. In this embodiment, the storage unit 12 stores program 12a executed by the processing unit 11. Furthermore, the storage unit 12 includes a model information storage unit 12b and a prediction result DB5. The model information storage unit 12b stores information related to the model used in processes such as separation processing to separate line spectrum data and prediction of electron density. The prediction result DB5 stores and accumulates prediction results for electron density and electron temperature.

[0033] In this embodiment, the program (computer program, program product) 12a is provided in the form of a recording medium 99 such as a memory card or optical disc, and the information processing device 1 reads the program 12a from the recording medium 99 and stores it in the storage unit 12. However, the program 12a may also be written into the storage unit 12, for example, during the manufacturing stage of the information processing device 1. Furthermore, the program 12a may be a program sent by a remote server device or the like, which the information processing device 1 obtains through communication. For example, the program 12a may also be a program read from the recording medium 99 and written into the storage unit 12 of the information processing device 1 by a writing device. The program 12a may be provided by means of distribution via a network or by means of being recorded in the recording medium 99.

[0034] The model information storage unit 12b stores information related to various models, such as physical models representing the characteristics of reaction species and learning models generated through simulation and machine learning. This model information may include, for example, information representing the model structure and predetermined internal parameter values. In this embodiment, the model information storage unit 12b stores information related to separation models, such as line spectrum data and background spectrum data obtained from the substrate processing device 3, which separate spectral data into multiple reaction species. Furthermore, the model information storage unit 12b stores information related to reaction species models that predict luminescence intensity for each reaction species based on inputs of electron temperature and electron density. Additionally, the model information storage unit 12b stores information related to prediction models that predict values ​​for the next timing based on time-series data such as electron density and electron temperature stored in the prediction result DB5.

[0035] Furthermore, in this embodiment, although one or more models storing information in the model information storage unit 12b are configured to be pre-generated in the information processing device 1 through machine learning or the like, this is not a limitation. This model-related information can also be generated in a device different from the information processing device 1 and stored in the model information storage unit 12b. Additionally, model-related information can also be stored in a device different from the information processing device 1.

[0036] The prediction result DB5 stores and accumulates information such as electron density and electron temperature for each reaction species, predicted based on the spectral data obtained from the substrate processing apparatus 3. The prediction result DB5 stores information such as the date and time of obtaining the spectral data from the substrate processing apparatus 3, the identification information of the substrate being processed, the obtained spectral data, the separated line spectral data and background spectral data, and the predicted electron temperature magnitude, electron density and electron temperature, etc.

[0037] The communication unit 13 is connected to the substrate processing device 3 via a communication line, for example, and performs data transmission and reception between the communication unit 13 and the substrate processing device 3. In this embodiment, the communication unit 13 receives spectral data sent from the substrate processing device 3 and provides it to the processing unit 11. In addition, the communication unit 13 sends control commands and other data provided by the processing unit 11 to the substrate processing device 3.

[0038] The display unit 14 is configured using a liquid crystal display or the like, and displays various images and text based on the processing of the processing unit 11. The display unit 14 displays various information such as spectral data acquired from the substrate processing apparatus 3, separated line spectrum data and background spectrum data, and electron density and electron temperature of each reaction species.

[0039] The operation unit 15 accepts user operations and notifies the processing unit 11 of the accepted operations. For example, the operation unit 15 accepts user operations via input devices such as mechanical buttons or a touch panel provided on the surface of the display unit 14. Alternatively, the operation unit 15 may be an input device such as a mouse and keyboard, and these input devices may be detachable from the information processing device 1.

[0040] Furthermore, the storage unit 12 may be an external storage device connected to the information processing device 1. Additionally, the information processing device 1 may be a multi-computer comprising multiple computers, or a virtual machine constructed via software. Furthermore, the information processing device 1 is not limited to the above-described structure; for example, it may not include a display unit 14 and an operation unit 15.

[0041] Furthermore, in the information processing apparatus 1 according to this embodiment, the processing unit 11 reads and executes the program 12a stored in the storage unit 12, which serves as a software-level functional unit. The processing unit 11 implements a spectral data acquisition unit 11a, a spectral separation unit 11b, an electron temperature magnitude prediction unit 11c, an electron density prediction unit 11d, a time series data prediction unit 11e, a display processing unit 11f, and a control processing unit 11g, among others. In this figure, the functional units of the processing unit 11 are shown as those related to the process of predicting the electron density and electron temperature of each reaction species based on the spectral data acquired from the substrate processing apparatus 3; functional units for other processes are omitted from the illustration.

[0042] The spectral data acquisition unit 11a communicates with the substrate processing apparatus 3 via the communication unit 13, and performs processing to acquire spectral data measured by measuring instruments or sensors when the substrate processing apparatus 3 performs substrate processing. In this embodiment, the spectral data is data representing the correspondence between the wavelength and intensity of light measured within the cavity of the substrate processing apparatus 3, and is provided to the information processing apparatus 1, for example, as a one-dimensional array. The spectral data acquisition unit 11a establishes a correspondence between the spectral data acquired from the substrate processing apparatus 3 and information such as date and time and substrate identification information, and stores it in the prediction result DB5.

[0043] The spectral separation unit 11b performs the process of separating the spectral data acquired by the spectral data acquisition unit 11a into multiple line spectral data for each reaction species and background spectral data other than these reaction species. In this embodiment, the spectral separation unit 11b uses a separation model pre-generated through machine learning to separate the spectral data into multiple line spectral data and background spectral data. Information related to the separation model is stored in the model information storage unit 12b. The spectral separation unit 11b stores the information related to the line spectral data and background spectral data separated from the spectral data in a corresponding manner with the original spectral data in the prediction result DB5.

[0044] The electron temperature magnitude prediction unit 11c performs a process to predict the magnitude (number of bits) of the electron temperature for each reaction species based on the line spectrum data of each reaction species separated by the spectral separation unit 11b. In this embodiment, the electron temperature magnitude prediction unit 11c uses, for example, any one of Doppler width, natural width, or Stark width as the spectral width for each reaction species, depending on the substrate processing conditions, etc. When using Doppler width, the electron temperature magnitude prediction unit 11c calculates the width of the line spectrum data of the reaction species and uses it as the Doppler width, calculates the electron temperature of the reaction species using a prescribed calculation formula related to the Doppler width, and uses this number of bits as the predicted result of the electron temperature magnitude.

[0045] Furthermore, in this embodiment, the electron temperature magnitude corresponding to the natural width or Stark width of the reactant is pre-calculated, for example, as a table mapping reactants to electron temperature magnitudes, and is pre-stored in the storage unit 12 by the information processing device 1. When the natural width is used as the spectral width, the electron temperature magnitude prediction unit 11c reads the table related to the electron temperature magnitude of the natural width from the storage unit 12, and predicts the electron temperature magnitude corresponding to the natural width of the reactant's spectrum by obtaining the electron temperature magnitude corresponding to the reactant to be predicted from the table. Similarly, when the Stark width is used as the spectral width, the electron temperature magnitude prediction unit 11c reads the table of electron temperature magnitudes related to the Stark width from the storage unit 12, and predicts the electron temperature magnitude corresponding to the Stark width of the reactant's spectrum by obtaining the electron temperature magnitude corresponding to the reactant to be predicted from the table.

[0046] Furthermore, the electron temperature magnitude prediction unit 11c predicts the electron temperature magnitude for background spectral data separated from the spectral data. The electron temperature magnitude prediction unit 11c assumes, for example, that the background spectral data follows a Boltzmann distribution, and predicts the electron temperature magnitude by fitting the separated background spectral data to a Boltzmann distribution. The electron temperature magnitude prediction unit 11c stores the information corresponding to the predicted electron temperature magnitude in the prediction result DB5, establishing a correspondence between the predicted electron temperature magnitude and the original spectral data.

[0047] The electron density prediction unit 11d uses a pre-stored reaction type model for each reaction type to predict the electron density and electron temperature of each reaction type. The reaction type model takes the electron density and electron temperature of the reaction type as input and outputs the luminescence intensity (spectrum) of the light from that reaction type. In this embodiment, the reaction type model is given as a pre-determined logical operation formula based on the theoretical equation (rate equation) for determining the plasma composition. However, the reaction type model can also be given as, for example, a learning model (surrogate model) generated by machine learning using simulated data, or a model generated by other methods.

[0048] The electron density prediction unit 11d inputs appropriate electron density and electron temperature into the reaction species model and acquires the spectrum output by the model. The electron temperature input to the reaction species model is a value on the order of the predicted electron temperature for each reaction species. The electron density prediction unit 11d calculates the difference between the spectrum output by the reaction species model and the line spectrum data separated for that reaction species. If the calculated difference is greater than a threshold, the electron density prediction unit 11d appropriately corrects the electron density and electron temperature input to the reaction species model until the difference between the spectrum output by the model and the line spectrum data for that reaction species is less than the threshold. This process of correcting the electron density and electron temperature is repeated to search for the optimal solution. If the difference is less than the threshold, the electron density prediction unit 11d considers the current electron density and electron temperature as the optimal solution and outputs them as predicted values. Furthermore, the electron density prediction unit 11d stores the predicted electron density and electron temperature in the prediction result DB5, establishing a correspondence between the predicted electron density and electron temperature and the original spectral data.

[0049] The time series data prediction unit 11e processes multiple electron density and electron temperature predictions repeatedly predicted by the electron density prediction unit 11d and stored in the prediction result DB5 as time series data, and performs processing to predict the electron density and electron temperature for the next or even later time period based on past electron density and electron temperature. In the time series data prediction unit 11e, a learning model pre-generated through machine learning, such as an RNN (Recurrent Neural Network) or LSTM (Long Short-Term Memory), is used to predict time series data. The time series data prediction unit 11e sequentially inputs the electron density or electron temperature data of the time series into the learning model and obtains the predicted values ​​of electron density or electron temperature sequentially output by the learning model, thereby obtaining the predicted values ​​for the next time period. The time series data prediction unit 11e can provide the predicted values ​​of electron density and electron temperature to the electron density prediction unit 11d, which uses the provided predicted values, for example, as initial values ​​for a search, to search for the optimal electron density and electron temperature.

[0050] The display processing unit 11f performs the processing of displaying various texts and images on the display unit 14. In this embodiment, the display processing unit 11f generates charts from, for example, spectral data acquired from the substrate processing apparatus 3, line spectral data separated from the spectral data, or background spectral data, and displays them on the display unit 14. Furthermore, the display processing unit 11f displays information about the electron density and electron temperature predicted by the electron density prediction unit 11d on the display unit 14. Additionally, the display processing unit 11f can also display various other information on the display unit 14.

[0051] The control processing unit 11g communicates with the substrate processing apparatus 3 via the communication unit 13, issuing various commands or instructions to the substrate processing apparatus 3, thereby controlling the substrate processing performed by the substrate processing apparatus 3. In this embodiment, the control processing unit 11g adjusts the pressure, gas flow rate, or power settings of the substrate processing apparatus 3 based on information such as electron density and electron temperature predicted by the electron density prediction unit 11d, and controls the substrate processing. For example, the control processing unit 11g can determine whether the substrate processing is normal or abnormal based on the predicted electron density, and perform control processing such as stopping the substrate processing when an abnormality is found. Furthermore, for example, the control processing unit 11g can adjust the settings of the substrate processing apparatus 3 to reduce / increase the gas flow rate related to a certain reaction when the electron density of a specific reaction is higher / lower than the normal range. The control processing unit 11g can also control various other substrate processing methods besides those described above.

[0052] <Spectral Separation Processing>

[0053] The information processing apparatus 1 in this embodiment uses a learning model, namely a separation model, generated in advance through machine learning to perform spectral separation processing on the spectral data obtained from the substrate processing apparatus 3. Figure 3 This is a schematic diagram illustrating a structural example of a separation model used in spectral separation processing. The separation model 21 involved in this embodiment is a learning model that takes spectral data acquired from the substrate processing apparatus 3 as input and performs machine learning beforehand to separate the spectral data into line spectral data of multiple reaction species and background spectral data. In the illustrated example, the separation model 21 separates the line spectral data of N reaction species (N being a natural number) (line spectral data of reaction species 1, reaction species 2, ..., reaction species N) and one background spectral data from the original spectral data. The separation model 21 can, for example, employ a neural network structure, but it can also be a learning model with other structures.

[0054] In this embodiment, the spectral data input to the separation model 21 is data relating the wavelength and intensity of light, for example, provided as a one-dimensional array. The line spectral data and background spectral data for each reaction species output by the separation model 21 are parameters of a function obtained by approximating the separated spectral data using a predefined function. This function can be various types, such as the Gaussian function, Lorentz function, Voigt function, Boltzmann function, or exponential function.

[0055] For example, when a single separated spectral data point is represented by three parameters of a defined function, separation model 21 outputs (N+1)×3 values. In this case, the data output by separation model 21 can be represented as an (N+1)×3 matrix. Furthermore, different functions can be used for each reaction, in which case the number of parameters for each reaction can also be different.

[0056] Figure 4This is a schematic diagram illustrating an example of machine learning for separating model 21. In the machine learning process of generating separating model 21, information processing device 1 uses a spectral data reproduction unit 22 and an error calculation unit 23. The spectral data reproduction unit 22 receives multiple line spectral data and background spectral data output from separating model 21 as input, and outputs spectral data that reproduces and synthesizes the multiple line spectral data and background spectral data. That is, the spectral data reproduction unit 22 performs the process of restoring the line spectral data and background spectral data separated by separating model 21 to the original spectral data. The reproduced spectral data output by the spectral data reproduction unit 22 is data in the same format as the spectral data input to separating model 21, for example, a one-dimensional array. Based on the input data, the spectral data reproduction unit 22 reproduces each line spectral data and background spectral data using a predefined function, and by synthesizing these reproduced data, it can reproduce the original spectral data.

[0057] The error calculation unit 23 calculates the error between the reproduced spectral data reproduced by the spectral data reproduction unit 22 and the original spectral data input to the separation model 21. The error calculation unit 23 calculates the error using functions such as loss functions or error functions, for example, calculating the mean squared error or mean absolute error. Based on the error calculated by the error calculation unit 23, the information processing device 1 updates the parameters of the separation model 21, for example, using the error backpropagation method, thereby enabling machine learning of the separation model 21.

[0058] The information processing device 1 uses a sufficient amount of spectral data pre-measured in the substrate processing device 3 as learning data to perform machine learning on the separation model 21 using the method described above. According to this method, no correct answer labels need to be attached to the spectral data obtained through measurement; machine learning of the separation model 21 can be performed through so-called unsupervised learning. The information processing device 1 stores information such as the internal parameters of the separation model 21 obtained through machine learning in the model information storage unit 12b. During spectral separation processing, the information processing device 1 reads information such as the internal parameters from the model information storage unit 12b to construct the separation model 21, inputs the spectral data obtained from the substrate processing device 3 into the separation model 21, and can separate the spectral data into multiple line spectral data and background spectral data by acquiring the data output by the separation model 21. Furthermore, although it is assumed that the information processing device 1 performs machine learning on the separation model 21 in this embodiment, it is not limited to this. Machine learning on the separation model 21 can also be performed in other devices, and the information processing device 1 can acquire information such as the internal parameters of the separation model 21 obtained through machine learning from that device and store it in the model information storage unit 12b.

[0059] <Predictive Processing of Electron Temperature Scale>

[0060] The information processing apparatus 1 in this embodiment performs a process to predict the electron temperature magnitude of each reaction species based on multiple line spectrum data and background spectrum data separated by spectral separation processing. In this embodiment, based on the substrate processing conditions performed in the substrate processing apparatus 3, it is assumed that the electron temperature of the reaction species is significantly affected by any one of the three spectral widths: Doppler width, natural width, and Stark width. The information processing apparatus 1 determines, based on the substrate processing conditions, which spectral width to use to predict the electron temperature magnitude for each reaction species.

[0061] Figure 5 This is a schematic diagram of an example of a table used to determine the types of spectral widths that influence the prediction of the magnitude of electron temperature in a reaction. Figure 5 The table showing the determination of spectral width types establishes a correspondence between N reaction types 1 to N and substrate processing conditions A, B, C, etc., and stores spectral width types (Doppler width, natural width, Stark width) as factors of electron temperature magnitude. When the information processing device 1 acquires spectral data from the substrate processing device 3, it acquires information related to the substrate processing conditions and, by referring to the table shown, determines which spectral width to predict the electron temperature magnitude for each reaction type.

[0062] In this embodiment, the electron temperature order is predefined for the natural width and Stark width among the three spectral widths. Figure 6 This is a schematic diagram illustrating an example of a table used to determine the magnitude of the electron temperature corresponding to the natural width and the Stark width. Figure 6 The table showing the electron temperature magnitude determination establishes a correspondence between N reaction species 1 to N and their natural widths and Stark widths, storing the electron temperature magnitudes. Information processing device 1, based on... Figure 5 When the table shown uses natural width or Stark width as the spectral width, refer to... Figure 6 The table shown determines the electron temperature magnitude corresponding to the reaction species and the natural width or Stark width.

[0063] in addition, Figure 5 and Figure 6 The information stored in the table shown was determined in advance by the designer or manager of the information processing system involved in this embodiment through spectral observation and simulation, and is pre-stored in the storage unit 12 of the information processing device 1. Furthermore, in this example, [the following is used:] Figure 5 and Figure 6 The two tables shown are not limited to these; they can also be combined into one table.

[0064] Furthermore, information processing device 1 is based on Figure 5The table shown uses Doppler width as the spectral width to predict the order of magnitude of electron temperature based on the line spectral data separated from the spectral data obtained from the substrate processing device 3. Figure 7 This is a schematic diagram illustrating an example of line spectral data and Doppler width, and it is a graph that charts the line spectral data with wavelength λ on the horizontal axis and spectral intensity on the vertical axis. In the line spectral data shown in this figure, the wavelength where the spectral intensity reaches its maximum is defined as λ0. In this embodiment, the width of the wavelength range that includes wavelength λ0 and whose spectral intensity reaches more than half of the maximum value is defined as the Doppler width ΔλD. The information processing device 1 can calculate the Doppler width ΔλD of each separated line spectral data.

[0065] Between the Doppler width ΔλD of the reaction species and the electron temperature Tg, the following formula (1) holds. Furthermore, in formula (1), λ0 is the center wavelength of the spectrum, such as... Figure 7 As shown. Furthermore, in formula (1), c is the speed of light (299792458 m / s), mg is the mass of the reactant, k is the Boltzmann constant, and ln2 is the natural logarithm of 2. These values ​​are given in advance as constants.

[0066] [Formula 1]

[0067]

[0068] Information processing device 1 calculates the electron temperature Tg using the aforementioned formula (1) based on the Doppler width ΔλD calculated from the line spectral data. However, in this embodiment, information processing device 1 does not utilize the calculated electron temperature Tg itself, but instead obtains the number of bits in the calculated electron temperature Tg and uses it as a predicted value of the electron temperature order of magnitude. For in Figure 5 The spectral width factor in the table is set to all reaction species with Doppler width. The information processing device 1 predicts the electron temperature magnitude based on the Doppler width ΔλD calculated from the line spectral data and the above formula (1).

[0069] Furthermore, the information processing device 1 predicts the magnitude of the electron temperature based on background spectral data separated from the spectral data acquired by the substrate processing device 3. The background spectrum represents the offset component after removing the line spectral signal at a specific wavelength from the input emission signal. In this embodiment, the information processing device 1 assumes that the background spectral data follows a Boltzmann distribution, calculates the braking radiation cross-section based on the momentum transfer caused by collisions between molecules / atoms used in the process and electrons, and predicts the electron temperature or electron energy distribution consistent with the offset signal of the background spectrum based on the emissivity of each wavelength obtained using the braking radiation cross-section.

[0070] The information processing device 1 calculates the electron temperature Te by fitting the background spectral data to the above formula (1), obtains the number of bits of the calculated electron temperature Te, and uses it as a predicted value of the electron temperature magnitude related to the background spectral data.

[0071] Figure 8 This is a flowchart illustrating an example of the steps in the electron temperature magnitude prediction processing performed by the information processing apparatus 1 according to this embodiment. The electron temperature magnitude prediction unit 11c of the processing unit 11 of the information processing apparatus 1 according to this embodiment acquires line spectral data of multiple reaction species separated by the spectral separation process (step S1). Furthermore, the electron temperature magnitude prediction unit 11c acquires information related to the conditions of the substrate processing being performed when the spectral data is acquired from the substrate processing apparatus 3 (step S2). Additionally, the substrate processing conditions may be, for example, information indicating which process is being performed in the substrate processing, such as etching or film deposition.

[0072] The electron temperature magnitude prediction unit 11c selects the target reaction for electron temperature magnitude prediction from multiple reaction types from which the line spectrum data has been separated (step S3). Based on the reaction type selected in step S3 and the substrate processing conditions obtained in step S2, the electron temperature magnitude prediction unit 11c refers to... Figure 5 The table showing the type of spectral width is used (step S4). Based on the results of the reference type determination table, the electron temperature magnitude prediction unit 11c determines whether the spectral width that has a significant impact on the electron temperature of the target reaction species is a Doppler width (step S5).

[0073] When it is determined to be a Doppler width (S5: Yes), the electron temperature magnitude prediction unit 11c, based on the line spectrum data of the reaction species of the electron temperature magnitude prediction object obtained in step S1, performs... Figure 7 The Doppler width is calculated using the method shown (step S6). The electron temperature magnitude prediction unit 11c calculates the electron temperature Tg using formula (1) based on the Doppler width ΔλD calculated in step S6, and calculates the electron temperature magnitude for the target reaction species by obtaining the number of bits of the calculated electron temperature Tg (step S7), and proceeds the process to step S10.

[0074] When it is determined that it is not a Doppler width (S5: No), that is, when it is determined that the spectral width that has a greater impact on the electron temperature of the target reaction species is the natural width or the Stark width, the electron temperature magnitude prediction unit 11c determines the type of spectral width (natural width or Stark width) determined by the table based on the reaction species selected in step S3 and the type referenced in step S4. Figure 6The electron temperature magnitude determination table shown is used to obtain the electron temperature magnitude corresponding to the reaction species and spectral width type (step S9), and the process is then advanced to step S10.

[0075] The electron temperature magnitude prediction unit 11c determines whether the electron temperature magnitude prediction for all reaction species based on the multiple line spectral data acquired in step S1 has been completed (step S10). If the prediction for all reaction species has not been completed (S10: No), the electron temperature magnitude prediction unit 11c returns the process to step S3, selects other reaction species, and repeats the same process.

[0076] When the prediction of all reaction species has been completed (S10: Yes), the electron temperature magnitude prediction unit 11c acquires the background spectral data separated from the spectral data through spectral separation processing (step S11). The electron temperature magnitude prediction unit 11c calculates the electron temperature Te by fitting the background spectral data acquired in step S11 to the above formula (1), and calculates the electron temperature magnitude of the background spectral data by acquiring the number of bits of the calculated electron temperature Te (step S12), and ends the processing.

[0077] <Electron Density Prediction Processing>

[0078] The information processing apparatus 1 according to this embodiment performs a process to predict the electron density of each reaction species based on multiple line spectrum data separated by spectral separation processing and the electron temperature magnitude predicted by electron temperature magnitude prediction processing. Furthermore, in this embodiment, the information processing apparatus 1 predicts both the electron density and the electron temperature of each reaction species.

[0079] The information processing apparatus 1 in this embodiment uses a reaction species model pre-generated to reproduce the characteristics of each reaction species to predict electron density and electron temperature. Figure 9 This is a schematic diagram illustrating a structural example of a reaction species model. The reaction species model 31 used in the electron density prediction processing of the information processing apparatus 1 according to this embodiment is a model that accepts the electron temperature and electron density of the reaction species as input and outputs predicted values ​​of the line spectrum data of the reaction species. Furthermore, the reaction species model 31 can accept information related to the reaction species being predicted (reaction species information) as input. For example, if a reaction species model 31 is generated separately for multiple reaction species, no reaction species information input is required. Conversely, in a structure where one reaction species model 31 can predict multiple reaction species, information related to the reaction species being predicted is provided to the reaction species model 31 as reaction species information.

[0080] The reaction species model 31 can be given as a physical model that expresses the physical properties of the reaction species using equations, for example, a model based on a rate equation. In this case, information such as constants or coefficients set in the rate equation is provided to the reaction species model 31 as reaction species information. Alternatively, the reaction species model 31 can be given as a learning model generated through machine learning, such as a neural network. In this case, information corresponding to the input and output of the reaction species model 31 is collected in advance through experiments or simulations, and the reaction species model 31 is pre-generated by using this collected information for machine learning.

[0081] The information processing device 1 appropriately sets and inputs the electron density and electron temperature as inputs to the reaction species model 31, and acquires the line spectrum data output by the reaction species model 31 in response. Furthermore, at this time, the information processing device 1 sets the electron temperature within the range of the electron temperature magnitude predicted in the aforementioned electron temperature magnitude prediction processing. The line spectrum data output by the reaction species model 31 may, for example, be a one-dimensional array of data having multiple intensity values ​​corresponding to the light wavelength, or it may be, for example, parameters representing a function of the line spectrum data, similar to those of the line spectrum data output by the separation model 21.

[0082] Next, the information processing device 1 compares the line spectrum data acquired from the reaction seed model 31 with the line spectrum data separated by the spectral separation process, and calculates the difference (error) between the two line spectrum data. When the calculated error is greater than a predetermined threshold, the information processing device 1 adjusts the electron density and electron temperature input to the reaction seed model 31 to reduce the error, and repeats the acquisition of the line spectrum data output by the reaction seed model 31 and the calculation of the error between it and the line spectrum data obtained by the spectral separation process, until the error becomes less than the threshold. This process is a so-called search for the optimal solution. Existing optimization methods can be applied to update the electron density and electron temperature values ​​input to the reaction seed model 31 in a way that reduces the error.

[0083] When the calculated error is less than a predetermined threshold, the information processing device 1 will input the electron density and electron temperature of the reaction species model 31 as the prediction result of the electron density and electron temperature of the target reaction species.

[0084] In this embodiment, the information processing device 1 predicts electron density and electron temperature, but it is not limited to this; it could also predict electron energy distribution instead of electron temperature. Electron energy distribution can be represented, for example, by an EEDF (Electron Energy Distribution Function). In predicting electron energy distribution, the reaction model 31 can take electron density and electron energy distribution as input and output line spectrum data. Furthermore, the information processing device 1 predicts the magnitude of electron energy distribution instead of predicting the magnitude of electron temperature. Electron temperature and the variance (dispersion) of electron energy distribution are synonymous; for example, by pre-specifying the electron energy distribution function by the user, a prediction of the magnitude of electron energy distribution can be obtained from the prediction of the magnitude of electron temperature. The function of electron energy distribution can be, for example, a Maxwell distribution or a Druyvesteyn distribution. The magnitude of electron energy distribution is the magnitude of the estimated coefficients of these distribution functions. Similar to the case of using electron temperature described above, the information processing device 1 uses the reaction species model 31 and the prediction results of the electron energy distribution magnitude to search for the optimal electron density and electron energy distribution, thereby predicting the electron density and electron energy distribution of each reaction species.

[0085] Furthermore, if the electron density is known experimentally or can be obtained by measurement, the information processing device 1 may use the known electron density value instead of predicting the electron density to predict the electron temperature (or electron energy distribution) using the reaction species model 31.

[0086] Figure 10 This is a flowchart illustrating an example of the electron density prediction processing steps performed by the information processing apparatus 1 according to this embodiment. The electron density prediction unit 11d of the processing unit 11 of the information processing apparatus 1 according to this embodiment acquires line spectrum data separated from the spectral data acquired from the substrate processing apparatus 3 for the reaction species of the electron density prediction target (step S31). Furthermore, the electron density prediction unit 11d acquires reaction species information regarding the reaction species of the prediction target (step S32). Additionally, the reaction species information acquired in step S32 may include constants or coefficients of the calculation formula, for example, if the reaction species model 31 is given by a calculation formula such as a rate equation. If a reaction species model 31 that does not require this information is used, the electron density prediction unit 11d may not acquire the reaction species information.

[0087] The electron density prediction unit 11d appropriately sets the electron density and electron temperature as initial values ​​input to the reaction seed model 31 (step S33). At this time, the electron density prediction unit 11d sets the electron temperature within the range of the electron temperature order of magnitude predicted by the electron temperature order of magnitude prediction processing. The electron density prediction unit 11d inputs the electron density and electron temperature set in step S33 to the reaction seed model 31 and acquires the line spectrum data output by the reaction seed model 31 (step S34). The electron density prediction unit 11d calculates the error between the line spectrum data acquired in step S34 and the line spectrum data acquired in step S31 (step S35).

[0088] The electron density prediction unit 11d determines whether the error calculated in step S35 is less than a predetermined threshold (step S36). When the error is greater than the threshold (S36: No), the electron density prediction unit 11d resets the electron density and electron temperature input to the reaction seed model 31 to reduce the error (step S37) and returns the process to step S34. When the error is less than the threshold (S36: Yes), the electron density prediction unit 11d takes the electron density and electron temperature input to the reaction seed model 31 as the optimal value and stores the electron density and electron temperature as the prediction result in the prediction result DB5 (step S38).

[0089] The electron density prediction unit 11d determines whether the prediction of electron density and electron temperature for all reaction types of the prediction target has been completed (step S39). If the prediction for all reaction types has not been completed (S39: No), the electron density prediction unit 11d returns to step S31 and repeats the same process for other reaction types. If the prediction for all reaction types has been completed (S39: Yes), the display processing unit 11f of the processing unit 11 displays the prediction results of electron density and electron temperature on the display unit 14 (step S40). In addition, the control processing unit 11g of the processing unit 11 performs the processing of controlling the operation of the substrate processing device 3 based on the predicted electron density and electron temperature (step S41) and ends the electron density prediction process.

[0090] <Time Series Data Forecasting and Processing>

[0091] The information processing apparatus 1 in this embodiment stores the spectral data acquired from the substrate processing apparatus 3, the line spectral data and background spectral data of each reaction species separated from the spectral data, the electron temperature magnitude predicted based on the line spectral data and background spectral data of each reaction species, and the electron density and electron temperature predicted for each reaction species in a prediction result DB5. This information is repeatedly memorized by the information processing apparatus 1, and is accumulated as time-series data in the prediction result DB5. Based on this time-series data stored in the prediction result DB5, the information processing apparatus 1 performs processing to predict future data.

[0092] Figure 11 This is a schematic diagram illustrating an example of time-series data, showing a graph with time t on the horizontal axis and the predicted electron density on the vertical axis. According to this example graph, it shows the tendency for electron density to increase over time during substrate processing performed by the substrate processing apparatus 3. The information processing apparatus 1 according to this embodiment uses a learning model such as an RNN or LSTM, pre-generated through machine learning, to predict time-series data such as electron density and electron temperature for each reaction species. For example, the learning model outputs a predicted value for the data at the next timing, relative to the data input at a certain timing.

[0093] After predicting electron density and electron temperature, the information processing device 1 inputs the current electron density and electron temperature values, which are the prediction results, into the learning model of the predicted time series data to obtain the predicted values ​​of electron density and electron temperature at the next time step. The information processing device 1 calculates a predetermined range (e.g., ±10% of the predicted values) including the obtained predicted values ​​of electron density and electron temperature as the prediction range.

[0094] For example, the information processing device 1 feeds back the predicted values ​​and prediction ranges of electron density and electron temperature to the electron density prediction process described above. As described above, the electron density prediction process performs the search for the optimal electron density and electron temperature, but the information processing device 1 can also perform operations such as using the electron density and electron temperature predicted based on time series data for the next time period as the initial values ​​for the search, or limiting the search range of electron density and electron temperature to the range predicted based on time series data.

[0095] <Information Provision and Control Processing>

[0096] The information processing apparatus 1 according to this embodiment provides information related to substrate processing to the user by displaying the predicted electron density and electron temperature of each reaction on the display unit 14. The information processing apparatus 1 repeatedly acquires spectral data while the substrate processing apparatus 3 is performing substrate processing, and repeatedly predicts the electron density and electron temperature of each reaction based on the acquired spectral data. The information processing apparatus 1 can display, for example, the predicted changes in electron density and electron temperature over time as a graph in real time. The user can, for example, monitor in real time the changes in electron density and electron temperature of important reactions in substrate processing processes such as etching or film deposition. Furthermore, based on information such as the predicted electron density or electron temperature for a given reaction, the user can estimate, for example, the consumption rate of components installed in the chamber of the substrate processing apparatus 3, and perform component replacement, etc.

[0097] Furthermore, the information processing device 1 controls the substrate processing apparatus 3 based on predicted electron density and electron temperature. The information processing device 1 can, for example, determine whether the predicted electron density and electron temperature are within a normal range, and if they are not within the normal range, control to stop the substrate processing of the substrate processing apparatus 3. Furthermore, for example, if the predicted electron density is higher / lower than the normal range, the information processing device 1 adjusts the settings of the substrate processing apparatus 3 to reduce / increase the flow rate of the gas associated with that reaction. Furthermore, for example, if the electron density or electron temperature of a specified reaction reaches a specified target value, the information processing device 1 can perform control such as advancing the substrate processing process to the next process.

[0098] Providing information to users and controlling substrate processing based on the above-mentioned predictions of electron density and electron temperature is just one example, and is not limited to this. The information processing device 1 can provide various information and control processing based on the prediction results.

[0099] <Summary>

[0100] In the information processing system of this embodiment, as described above, the information processing device 1 acquires spectral data obtained from substrate processing measurements by the substrate processing device 3, separates the acquired spectral data into multiple line spectra for multiple reaction species, and predicts at least one of the electron density, electron temperature, and electron energy distribution for each reaction species based on the separated line spectra. Therefore, the information processing system of this embodiment is expected to be able to infer the state of substrate processing based on spectral data measured during substrate processing. The information processing system predicts at least one of the electron density, electron temperature, and electron energy distribution for each reaction species from the spectral data obtained simultaneously by the measuring instruments installed in the substrate processing device 3, and is expected to perform processing such as providing detailed information about substrate processing to the user, or controlling substrate processing in response to the conditions of each reaction species.

[0101] Furthermore, in the information processing system of this embodiment, the information processing device 1 predicts the electron density of each reaction species and the electron temperature or electron energy distribution based on the line spectral data separated from the spectral data. Therefore, the information processing system of this embodiment is expected to predict more information related to substrate processing and achieve the provision of more information to the user, or control of substrate processing with good processing accuracy.

[0102] Furthermore, in the information processing system of this embodiment, the information processing device 1 uses a separation model 21, which has been pre-machined to accept spectral data as input and output line spectral data of a specified reaction species and background spectral data other than the specified reaction species contained in the input spectral data, to separate the spectral data obtained from the substrate processing device 3 into line spectral data and background spectral data for each reaction species. The separation model 21 performs machine learning based on the error between the spectral data synthesized from the multiple line spectral data and background spectral data output by the separation model 21 relative to the input spectral data and the original spectral data input to the separation model 21. Thus, the information processing system of this embodiment generates the separation model 21 through so-called unsupervised machine learning, and it is expected to use the generated separation model 21 to separate spectral data into line spectral data and background spectral data for each reaction species with high accuracy.

[0103] Furthermore, in the information processing system of this embodiment, the information processing device 1 predicts the electron temperature magnitude of each reaction species based on the separated line spectral data, and predicts the electron density of each reaction species based on the separated line spectral data and the predicted electron temperature magnitude. Additionally, the information processing device 1 calculates the spectral width related to the separated spectral data and predicts the electron temperature magnitude based on the calculated spectral width. Furthermore, when the spectral width, which has a significant impact on the electron temperature magnitude, is the Doppler width, the information processing device 1 calculates the spectral width to predict the electron temperature magnitude; when the spectral width is the natural width or the Stark width, a predefined electron temperature magnitude is used as the prediction result. Therefore, the information processing system of this embodiment predicts the electron temperature magnitude of each reaction species based on the separated line spectral data, and by using this electron temperature magnitude in subsequent prediction processing (electron density prediction processing), it is expected to improve the accuracy and processing speed of subsequent prediction processing.

[0104] Furthermore, in the information processing system of this embodiment, the information processing device 1 uses a reaction species model 31 to predict line spectrum data based on the input of the electron density and electron temperature (or electron energy distribution) of the reaction species. By performing a search for electron density and electron temperature, the error between the line spectrum data separated from the spectral data obtained from the substrate processing device 3 and the line spectrum data predicted by the reaction species model 31 becomes less than a threshold, thereby predicting the electron density and electron temperature of each reaction species. In addition, the information processing device 1 can also limit the search range of electron temperature based on the magnitude of the electron temperature predicted from the line spectrum data. Therefore, the information processing system of this embodiment, utilizing the reaction species model 31 generated in unsupervised learning without teacher data (labeled data), is expected to predict the electron density and electron temperature of each reaction species with high accuracy.

[0105] Furthermore, in the information processing system of this embodiment, the information processing device 1 repeatedly performs processes such as acquiring spectral data, separating line spectral data, and predicting electron density. Based on the prediction result of the electron density at a certain moment, it predicts the range of the electron density at the next moment, and based on the separated line spectral data and the predicted electron density range, it predicts the electron density at the next moment and thereafter. Therefore, the information processing system of this embodiment is expected to perform electron density prediction with higher accuracy.

[0106] Furthermore, in the information processing system of this embodiment, the information processing device 1 controls the substrate processing performed by the substrate processing device 3 based on the predicted electron density of each reaction species. Therefore, the information processing system of this embodiment is expected to improve the accuracy of substrate processing.

[0107] The embodiments disclosed herein should be considered exemplary in all respects and not restrictive. The scope of this disclosure is not defined by the foregoing meaning, but by the claims, and is intended to include all modifications within the meaning and scope equivalent to the claims.

[0108] The items described in the various embodiments can be combined with each other. Furthermore, the independent and dependent claims described in the claims, regardless of their referencing form, can be combined with each other in any combination. Further, although the claims use a form referencing two or more other claims (multiple reference form), this is not a limitation. A form describing multiple reference claims that refer to at least one multiple reference claim (multiple multiple reference claims) can also be used.

[0109] Symbol Explanation

[0110] 1. Information processing device (computer)

[0111] 3. Substrate processing apparatus

[0112] 5. Prediction Results (DB)

[0113] 11 Processing Department

[0114] 11a Spectral Data Acquisition Department

[0115] 11b Spectral Separation Section

[0116] 11c electron temperature magnitude prediction unit

[0117] 11d electron density prediction department

[0118] 11e Time Series Data Forecasting Department

[0119] 11f Display Processing Unit

[0120] 11g Control Processing Unit

[0121] 12 Storage Department

[0122] 12a Program (Computer Program)

[0123] 12b Model Information Storage Department

[0124] 13 Ministry of Communications

[0125] 14 Display Section

[0126] 15 Operations Department

[0127] 21. Separation Model (Learning Model)

[0128] 22 Spectral Data Reproduction Department

[0129] 23 Error Calculation Department

[0130] 31. Reaction Species Model (Model)

[0131] 99 Recording media

Claims

1. A computer program, wherein, The computer program causes the computer to perform the following processes: Acquire spectral data measured related to substrate processing; The acquired spectral data is separated into line spectral data for multiple reaction species; Based on the separated line spectral data, at least one of the electron density, electron temperature, and electron energy distribution for each reaction species is predicted.

2. The computer program according to claim 1, wherein, A machine learning model is used, wherein the machine learning takes received spectral data as input and outputs line spectral data of a specified reaction species contained in the input spectral data, as well as background spectral data other than the specified reaction species. The acquired spectral data is input into the learning model to obtain the line spectral data and background spectral data output by the learning model, thereby separating the spectral data into line spectral data for each reaction species.

3. The computer program according to claim 2, wherein, The learning model performs machine learning based on the error between the synthesized spectral data (line spectral data and background spectral data output by the learning model relative to the input spectral data) and the input spectral data.

4. The computer program according to claim 1, wherein, Based on the separated line spectral data, the order of magnitude of the electronic temperature or the order of magnitude of the electronic energy distribution for each reaction species is predicted. Based on the separated line spectral data and the predicted electron temperature or electron energy distribution, the electron density of each reaction species is predicted.

5. The computer program according to claim 4, wherein, Calculate the spectral width associated with the separated line spectrum data, and based on the calculated spectral width, predict the magnitude of the electron temperature or the magnitude of the electron energy distribution.

6. The computer program according to claim 5, wherein, The spectral width is the Doppler width, natural width, or Stark width of the reaction species spectrum.

7. The computer program according to claim 1, wherein, A model that predicts the line spectral data of a reaction species using inputs of the species' electron density and electron temperature or electron energy distribution. The electron density and electron temperature or electron energy distribution of each reaction species are predicted by searching for the following input values, wherein the error between the isolated line spectral data and the line spectral data predicted by the model is less than a threshold value for the electron density and electron temperature or electron energy distribution.

8. The computer program according to claim 7, wherein, Based on the acquired spectral data, the order of magnitude of the electronic temperature or the order of magnitude of the electronic energy distribution for each reaction species is predicted. Based on the predicted magnitude of electron temperature or electron energy distribution, the search range of the electron temperature is limited during the search using the input values ​​of electron density and electron temperature or electron energy distribution of the model.

9. The computer program according to claim 1, wherein, The process of acquiring the spectral data, separating the line spectral data, and predicting at least one of the electron density, electron temperature, and electron energy distribution is repeated. Based on the prediction results of at least one of the electron density, electron temperature, and electron energy distribution at a given moment, predict the range of at least one of the electron density, electron temperature, and electron energy distribution at the next moment and thereafter. Based on the separated line spectral data and the predicted range, at least one of the electron density, electron temperature and electron energy distribution is predicted for the next time step and thereafter.

10. The computer program according to claim 1, wherein, The substrate processing is controlled based on at least one of the electron density, electron temperature, and electron energy distribution corresponding to the predicted spectrum of each reaction species.

11. An information processing method, wherein, The following processes are performed by the information processing device: Acquire spectral data measured related to substrate processing; The acquired spectral data is separated into line spectral data for multiple reaction species; Based on the separated line spectral data, at least one of the electron density, electron temperature, and electron energy distribution for each reaction species is predicted.

12. An information processing apparatus, wherein, The system includes a processing unit that performs the following processing: acquiring spectral data measured regarding substrate processing; The acquired spectral data is separated into line spectral data for multiple reaction species; Based on the separated line spectral data, at least one of the electron density, electron temperature, and electron energy distribution for each reaction species is predicted.

Images