WO2023013436A1

WO2023013436A1 - Prediction method, prediction program, prediction device, learning method, learning program, and learning device

Info

Publication number: WO2023013436A1
Application number: PCT/JP2022/028438
Authority: WO
Inventors: 倫太郎樋口; 光則中森
Original assignee: 東京エレクトロン株式会社
Priority date: 2021-08-05
Filing date: 2022-07-22
Publication date: 2023-02-09
Also published as: TW202310033A; JPWO2023013436A1

Abstract

This prediction method predicts an etching rate for an input condition, which has been input, by using a model that outputs the etching rate on the basis of feature amounts pertaining to a film and a chemical liquid in the etching of a substrate. In addition, the prediction method provides the predicted etching rate.

Description

Prediction method, prediction program, prediction device, learning method, learning program, and learning device

The present disclosure relates to a prediction method, a prediction program, a prediction device, a learning method, a learning program, and a learning device.

In substrate etching, chemical liquid screening is performed to understand the etching characteristics of unknown film types. According to chemical liquid screening, it is possible to grasp which chemical liquid is used and under what conditions to perform etching appropriately.

There is also a technique for generating correction data for correcting the settings of the substrate processing apparatus using a model generated by machine learning based on the setting conditions of the substrate processing apparatus for etching the substrate and the etching execution results. Are known.

JP 2020-4817 A

When performing chemical screening on film types for which there is absolutely no knowledge, it may take an enormous amount of time to select chemicals and set conditions.

The present disclosure provides a prediction method, a prediction program, a prediction device, a learning method, a learning program, and a learning device that can efficiently predict etching conditions.

A prediction method according to an embodiment is a prediction method executed by a computer, and uses a model that outputs an etching rate based on feature amounts related to a film and a chemical solution in etching a substrate to estimate an etching rate for input conditions. and a providing step of providing the etch rate predicted by the predicting step.

According to the present disclosure, etching conditions can be predicted efficiently.

FIG. 1 is a diagram explaining a model. FIG. 2 is a block diagram showing a configuration example of the learning device according to the embodiment. FIG. 3 is a flowchart showing the flow of learning processing for the first model. FIG. 4 is a flowchart showing the flow of learning processing for the second model. FIG. 5 is a block diagram illustrating a configuration example of a prediction device according to the embodiment; FIG. 6 is a flowchart showing the flow of prediction processing using the first model. FIG. 7 is a flowchart showing the flow of prediction processing using the second model. FIG. 8 is a flowchart showing the flow of search processing. FIG. 9 is a diagram showing an example of experimental data. FIG. 10 is a diagram showing an example of experimental data. FIG. 11 is a diagram illustrating a method of calculating activation energy. FIG. 12 is a diagram illustrating an example of a computer that executes programs.

Below, embodiments of the prediction method, prediction program, prediction device, learning method, learning program, and learning device will be described in detail based on the drawings. Note that the disclosed technology is not limited by the following embodiments.

In substrate etching, the etching rate varies depending on the film and chemical conditions. Conversely, in order to perform etching at a desired etching rate, it is necessary to set conditions suitable for the etching rate.

For example, the learning device of the embodiment generates a model representing the relationship between various conditions and the etching rate using a machine learning technique. Furthermore, the prediction device of the embodiment uses the generated model to predict the etching rate or various conditions other than the etching rate.

The model of this embodiment will be described using FIG. FIG. 1 is a diagram explaining a model.

As shown in FIG. 1, an experimental data DB 121 is created by accumulating the results of etching experiments.

For example, the experimental data DB 121 includes film types, deposition conditions, chemical liquid types, processing conditions, and etching rates (ER).

The first model is generated by learning the experimental data DB 121. In the example of FIG. 1, the first model is expressed as ER=f(X). Here, X is a condition group included in the experimental data DB 121 and is a feature amount of the first model.

Furthermore, by adding material properties, which are film properties, and chemical liquid properties, which are chemical properties, to the experimental data DB 121, an experimental data DB 121a is obtained.

The second model is generated by learning the experimental data DB 121a. In the example of FIG. 1, the second model is expressed as ER=g(Y,Z). where Y is the physical property of the material. Also, Z is the physical property of the chemical solution.

The model may be a linear or nonlinear regression model. For example, the first model is a regression model with X as an explanatory variable and ER as an objective variable. Also, for example, the second model is a regression model with Y and Z as explanatory variables and ER as an objective variable.

In the explanation below, explanatory variables may be referred to as feature quantities. Also, X, Y and Z may be vectors.

In addition, each of the film types, deposition conditions, chemical liquid types, processing conditions, material physical properties, and chemical liquid physical properties shown in FIG. 1 may include a plurality of feature quantities. For example, processing conditions may include temperature and processing time.

Also, each feature amount may be a qualitative variable or a quantitative variable. For example, film type, chemical liquid type, etc. may be qualitative variables. Also, the film formation conditions and processing conditions may include quantitative variables such as temperature and concentration. In addition, the material physical properties and chemical liquid physical properties may include quantitative variables.

Here, in the embodiment, it is assumed that the learning device generates the model. Also, the prediction device performs prediction processing using the model generated by the learning device. Note that the learning device and the prediction device may be realized by the same device.

[Learning device]
The configuration of the learning device will be described with reference to FIG. FIG. 2 is a block diagram showing a configuration example of the learning device according to the embodiment.

As shown in FIG. 2, the learning device 10 has an I/F (interface) section 11, a storage section 12 and a control section 13.

The I/F unit 11 is an interface for exchanging data between other devices. For example, the I/F unit 11 is a NIC (Network Interface Card). Furthermore, the I/F section 11 may be connected to input/output devices such as a mouse, keyboard, display, and speaker.

The storage unit 12 is implemented by, for example, a RAM (Random Access Memory), a semiconductor memory device such as flash memory, or a storage device such as a hard disk or optical disk.

The storage unit 12 stores an experimental data DB 121, material information 122, chemical solution information 123, and model information 124.

The experimental data DB 121 is a database created by accumulating the results of etching experiments, as explained in FIG.

The material information 122 is information that associates the film type with the material properties of the film. Material physical properties include, for example, atomic composition, crystallinity, binding energy, and the like. Here, the crystallinity is, for example, a value indicating whether the film is crystalline or amorphous. The crystallinity may also be an index related to the crystal structure of the film, such as plane orientation and defect density.

The chemical liquid information 123 is information that associates chemical liquid types and chemical liquid physical properties. Chemical properties include, for example, ORP (Oxidation-Reduction Potential), pH, dielectric constant, and the like.

The physical properties of the material information 122 and the chemical information 123 may be measured values or information obtained from literature or the like.

The model information 124 is information for constructing a model. The model information 124 is generated and updated by learning processing executed by the learning device 10 .

In this embodiment, the model information 124 is assumed to be parameters such as coefficients of the first model and the second model shown in FIG.

The control unit 13 can execute programs stored in an internal storage device using the RAM as a work area, for example, by a CPU (Central Processing Unit), MPU (Micro Processing Unit), GPU (Graphics Processing Unit), etc. It is realized by

Also, the control unit 13 may be realized by an integrated circuit such as ASIC (Application Specific Integrated Circuit) and FPGA (Field Programmable Gate Array).

The control unit 13 has an acquisition unit 131, a prediction unit 132, and an update unit 133. Note that the internal configuration of the control unit 13 is not limited to the configuration described here, and may be another configuration as long as it performs the information processing described later.

The acquisition unit 131 acquires a value indicating the physical properties of the film and a value indicating the physical properties of the chemical solution based on the type of film and the type of chemical solution included in the input conditions.

The acquisition unit 131 acquires the material properties corresponding to the film type in the experimental data DB 121 from the material information 122 . Further, the acquisition unit 131 acquires the chemical solution physical properties corresponding to the chemical solution type in the experimental data DB 121 from the chemical solution information 123 .

Here, the material properties are an example of values that indicate the physical properties of the film. Also, the physical properties of the chemical solution are examples of values indicating the physical properties of the chemical solution.

Furthermore, the acquisition unit 131 can obtain the experimental data DB 121a by adding the physical properties of the material and the physical properties of the chemical solution to the experimental data DB 121. As described with reference to FIG. 1, the experimental data DB 121a is used for learning the second model.

The prediction unit 132 predicts the etching rate for the input conditions by using a model that outputs the etching rate based on the feature values related to the film and the chemical solution in the etching of the substrate.

The first model and the second model are examples of models that output the etching rate based on the feature values related to the film and the chemical liquid in the etching of the substrate.

The update unit 133 updates the parameters of the model so that the difference between the ER predicted by the prediction unit 132 and the correct ER is reduced. For example, the updating unit 133 updates the coefficients of the regression model using the method of least squares. The updating unit 133 updates the model information 124 .

For example, the prediction unit 132 inputs the film type, film formation conditions, chemical liquid type, and processing conditions of the experimental data DB 121 into the first model, and outputs ER. Then, the updating unit 133 updates the parameters of the first model so that the difference between the ER output by the first model and the ER of the experimental data DB 121 becomes smaller.

For example, the learning device 10 can obtain a regression model such as formula (1) as the first model.

(1) x _n (n is a positive integer) is a feature amount. Also, the coefficient of _xn is a parameter obtained by learning processing and included in the model information 124 . For example, x ₁ , x ₂ , and x ₃ are feature quantities corresponding to film type, film formation conditions, and chemical solution temperature (one of processing conditions), respectively.

Also, for example, the prediction unit 132 inputs the physical properties of the material and the physical properties of the liquid medicine of the experimental data DB 121a into the second model, and outputs the ER. Then, the update unit 133 updates the parameters of the second model so that the difference between the ER output by the second model and the ER of the experimental data DB 121a is reduced.

For example, the learning device 10 can obtain a regression model such as Equation (2) as the second model.

y _n and z _n (n is a positive integer) in equation (2) are feature quantities. Also, the coefficients of y _n and z _n in equation (2) are parameters obtained by learning processing and included in the model information 124 . For example, y ₁ and y ₂ are feature quantities corresponding to material properties. Also, for example, z ₁ and z ₂ in the formula (2) are feature quantities corresponding to chemical liquid physical properties.

Although the learning device 10 generates a linear regression model in this embodiment, the learning device 10 may generate a nonlinear regression model. Further, the learning device 10 may further update the model information 124 when the experimental data DB 121 is changed.

The flow of learning processing by the learning device 10 will be described using FIGS. 3 and 4. FIG. FIG. 3 is a flowchart showing the flow of learning processing for the first model. FIG. 4 is a flowchart showing the flow of learning processing for the second model.

As shown in FIG. 3, in the learning process of the first model, the learning device 10 first inputs the feature amount obtained from the experimental data into the first model to predict the etching rate (step S101).

Then, the learning device 10 updates the parameters of the first model so that the predicted etching rate approaches the correct etching rate (step S102).

As shown in FIG. 4, in the learning process of the second model, the learning device 10 first acquires material properties corresponding to the type of film and chemical liquid properties corresponding to the type of chemical liquid (step S111).

Next, the learning device 10 inputs the feature quantity including the physical properties of the material and the physical properties of the chemical solution into the second model to predict the etching rate (step S112).

Then, the learning device 10 updates the parameters of the second model so that the predicted etching rate approaches the correct etching rate (step S113).

[Prediction device]
The configuration of the prediction device will be described with reference to FIG. FIG. 5 is a block diagram illustrating a configuration example of a prediction device according to the embodiment;

As shown in FIG. 5, the prediction device 20 has an I/F section 21, a storage section 22 and a control section 23.

The I/F unit 21 is an interface for exchanging data between other devices. For example, the I/F unit 21 is a NIC. Furthermore, the I/F section 21 may be connected to input/output devices such as a mouse, keyboard, display, and speaker.

The storage unit 22 is realized by, for example, a semiconductor memory device such as a RAM, a flash memory, or a storage device such as a hard disk or an optical disk.

The storage unit 22 stores an experimental data DB 121, material information 221, chemical solution information 222, and model information 223.

The material information 221 and the chemical information 222 are the same information as the material information 122 and the chemical information 123 of the learning device 10, respectively.

The model information 223 is information for constructing a learned model. The model information 223 is information similar to the model information 124 of the learning device 10 . However, it is assumed that the model information 223 has been updated (or generated) by the learning device 10 .

In this embodiment, the model information 223 is assumed to be parameters such as coefficients of the first model and the second model shown in FIG.

The control unit 23 is realized, for example, by executing a program stored in an internal storage device using a RAM as a work area by a CPU, MPU, GPU, or the like.

Also, the control unit 23 may be realized by an integrated circuit such as ASIC and FPGA, for example.

The control unit 23 has an acquisition unit 231 , a prediction unit 232 , a search unit 233 and a provision unit 234 . Note that the internal configuration of the control unit 23 is not limited to the configuration described here, and may be another configuration as long as it performs the information processing described later.

The acquisition unit 231 acquires a value indicating the physical properties of the film and a value indicating the physical properties of the chemical solution based on the type of film and the type of chemical solution included in the input conditions.

Here, the prediction device 20 accepts input of some feature amounts as input conditions. Also, the prediction device 20 may receive input of information for specifying a part of the feature amount.

For example, when a film type is input to the prediction device 20, the acquisition unit 231 acquires material properties corresponding to the film type from the material information 221. Further, for example, when a chemical liquid type is input to the prediction device 20 , the acquisition unit 231 acquires chemical liquid properties corresponding to the chemical liquid type from the chemical liquid information 222 .

The prediction unit 232 predicts the etching rate for the input conditions by using a model that outputs the etching rate based on the feature values related to the film and the chemical solution in the etching of the substrate.

For example, the prediction unit 232 predicts the etching rate using a first model that outputs the etching rate based on feature amounts including film type, film deposition conditions, chemical liquid type, and etching processing conditions.

Also, for example, the prediction unit 232 predicts the etching rate using a second model that outputs the etching rate based on the feature amount including the values indicating the physical properties of the material and the physical properties of the chemical solution.

Using the model, the search unit 233 searches for feature amounts related to the film and the chemical based on the designated etching rate. Thereby, the prediction device 20 can predict not only the etching rate but also other feature quantities.

For example, the search unit 233 uses the first model to search for at least one of the film type, film formation conditions, chemical liquid type, and processing conditions from the etching rate, and outputs the search results as recommended conditions.

Also, for example, the searching unit 233 searches for at least one of material properties and chemical liquid properties from the etching rate using the second model, and outputs the search results as recommended conditions.

Further, for example, the search unit 233 searches for the chemical liquid type or the chemical liquid physical properties from the etching rate and film-related conditions (film type, film formation conditions, material physical properties, etc.) using the second model, and the search result is used as a valid method. Output as a drug solution candidate.

The provision unit 234 provides the etching rate predicted by the prediction unit 232. Also, the providing unit 234 provides the feature amount (search result) searched by the searching unit 233 .

For example, the provision unit 234 outputs the prediction result and the search result in a predetermined data format such as text, voice, and image via the I/F unit 21.

For example, the providing unit 234 may transmit the etching rate or the search result to the external device via the network connected to the I/F unit 21. Further, for example, the providing unit 234 may cause the display unit connected to the I/F unit 21 to display the etching rate or the search result.

The flow of prediction processing and search processing by the prediction device 20 will be described using FIGS. 6, 7 and 8. FIG. FIG. 6 is a flowchart showing the flow of prediction processing using the first model. FIG. 7 is a flowchart showing the flow of prediction processing using the second model. FIG. 8 is a flowchart showing the flow of search processing.

As shown in FIG. 6, in the prediction process using the first model, first, the prediction device 20 inputs the feature amount obtained from the input conditions to the first model to predict the etching rate (step S201).

Then, the prediction device 20 provides the predicted etching rate (step S202).

As shown in FIG. 7, in the prediction process using the second model, first, the prediction device 20 acquires material properties according to the type of film and chemical properties according to the type of chemical (step S211).

Next, the prediction device 20 inputs the feature amount including the physical properties of the material and the physical properties of the chemical solution into the second model to predict the etching rate (step S212).

Then, the prediction device 20 provides the predicted etching rate (step S213).

As shown in FIG. 8, in the search process, first, the prediction device 20 uses the second model to search for the feature amount related to the chemical solution from the input conditions and etching rate related to the film (step S221).

Then, the prediction device 20 provides the searched feature amount (step S222).

[Example 1]
A specific example of the etching rate prediction process and search process will be described with reference to FIG. FIG. 9 is a diagram showing an example of experimental data.

"Film" in Fig. 9 is the type of film. "Depo" in FIG. 9 is a film forming condition. "Chem_1", "Chem_2", "Chem_1_vol ratio", "Chem_2_vol ratio" and "Chem_1_vconc (vol%)" in Fig. 9 are chemical types. "Temp (deg.C)" and "Process time (min)" in FIG. 9 are processing conditions.

The learning device 10 generates a regression model like Equation (3) from the experimental data shown in FIG.

(3) x _n (n is a positive integer) is a feature amount. For example, x ₁ , x ₂ , and x ₃ are respectively “Chem_1_vconc (vol%)” (HF concentration), “Temp (deg.C)” (temperature), “Process time (min)” (processing time). It is also assumed that _xn is normalized to be in the range of 0 to 1.

In this way, the learning device 10 can generate a regression model using some feature amounts of experimental data after performing feature selection by a predetermined method.

Here, it is assumed that (HF concentration, temperature, processing time) = (0.1 vol%, 40°C, 2 min) are input to the prediction device 20 as input conditions.

First, the prediction device 20 normalizes the input conditions as (0.1 vol%, 40°C, 2 min) = (0.38, 0, 0.05). Then, the prediction device 20 inputs the normalized input conditions to the learned regression model (3) and predicts ER=8.5 (nm/min).

Also, when ER=20 is specified, the prediction device 20 obtains (HF concentration, temperature, processing time)=(0.15 vol%, 47.3° C., 2.1 min) through search processing.

Thus, according to this embodiment, the etching rate can be predicted from the input conditions, and each condition can be searched based on the etching rate.

[Example 2]
A specific example of the etching rate prediction process for a new film will be described with reference to FIG. FIG. 10 is a diagram showing an example of experimental data.

"Film" in Fig. 10 is the type of film. "Depo" and "Depo_temp (deg.C)" in FIG. 10 are film forming conditions. "Chem_1", "Chem_2", "Chem_1_vol ratio", "Chem_2_vol ratio", and "Chem_1_vconc (vol%)" in Fig. 10 are chemical liquid types. "Temp (deg.C)" in FIG. 10 is a processing condition.

The learning device 10 generates a regression model like Equation (4) from the experimental data shown in FIG.

(3) x _n (n is a positive integer) is a feature amount. For example, x ₁ and x ₂ respectively correspond to "Depo_temp (deg.C)" (film formation temperature) and "Chem_1_vconc (vol%)" (HF concentration) in FIG. It is also assumed that _xn is normalized to be in the range of 0 to 1.

Here, it is assumed that (film formation temperature, HF concentration) = (1000°C, 1 vol%) is input to the prediction device 20 as input conditions.

First, the prediction device 20 normalizes the input conditions as (1000°C, 1 vol%) = (0.643, 0.069). Then, the prediction device 20 inputs the normalized input conditions to the learned regression model (4) and predicts ER=3.5 (nm/min).

As described above, according to the present embodiment, the etching rate can be predicted even when the film formation temperature of the input condition does not exist in the experimental data (new film).

[Other embodiments]
The providing unit 234 can provide the activation energy calculated from the relationship between the input processing temperature and the etching rate predicted by the prediction unit 232 . A method for calculating the activation energy will be described with reference to FIG. FIG. 11 is a diagram illustrating a method of calculating activation energy.

As shown in FIG. 11, the providing unit 234 plots the reaction (etching) rate constant ln k based on the predicted etching rate or the experimental data etching rate and the reciprocal of the etching processing temperature T.

Then, the providing unit 234 can calculate the activation energy Ea from the slope of the straight line 301 obtained by approximating the plotted points by the formula (5). In addition, A of (5) Formula is a constant and R is a gas constant.

As a result, the prediction device 20 can provide not only the etching rate but also the activation energy.

As described above, the prediction unit 132 of the learning device 10 according to the embodiment uses a model that outputs an etching rate based on feature amounts related to a film and a chemical solution in substrate etching, and uses a model that outputs an etching rate for an input input condition. Predict rates. The updating unit 133 also updates the parameters of the model so that the difference between the etching rate predicted by the prediction unit 132 and the correct etching rate becomes smaller. As a result, according to the embodiment, it is possible to obtain a model that can efficiently predict etching conditions.

The prediction unit 232 of the prediction device 20 of the embodiment predicts the etching rate for the inputted input conditions using a model that outputs the etching rate based on the feature values of the film and the chemical solution in the etching of the substrate. Also, the providing unit 234 provides the etching rate predicted by the predicting unit 232 . As a result, according to the embodiment, etching conditions can be predicted efficiently.

The prediction unit 232 of the prediction device 20 of the embodiment estimates the etching rate using a model that outputs the etching rate based on feature amounts including film type, film deposition conditions, chemical liquid type, and etching processing conditions. Predict. As a result, according to the embodiment, it is possible to easily predict the etching rate using feature amounts similar to experimental data.

The prediction unit 232 of the prediction device 20 of the embodiment predicts the etching rate using a model that outputs the etching rate based on the feature amount including the value indicating the physical properties of the film and the value indicating the physical properties of the chemical solution. As a result, according to the embodiment, it is possible to predict the etching rate with high accuracy using the feature values relating to the physical properties of the film and the chemical solution.

The acquisition unit 231 of the prediction device 20 of the embodiment acquires a value indicating the physical properties of the film and a value indicating the physical properties of the chemical solution based on the type of film and the type of chemical solution included in the input conditions. The prediction unit 232 uses a model to predict the etching rate for the value acquired by the acquisition unit 231 . As a result, according to the embodiment, it is possible to automatically acquire the feature values relating to the physical properties of the film and the chemical solution.

The search unit 233 of the prediction device 20 of the embodiment uses the model to search for feature amounts related to the film and the chemical based on the designated etching rate. Also, the providing unit 234 provides the feature amount searched by the searching unit 233 . As a result, according to the embodiment, conditions other than the etching rate can be obtained by searching.

The prediction unit 232 of the prediction device 20 of the embodiment predicts the etching rate for the input processing temperature using a model that outputs the etching rate based on the feature amount including the etching processing temperature. The providing unit 234 also provides the activation energy calculated from the relationship between the input processing temperature and the etching rate predicted by the prediction unit 232 . As a result, according to the embodiment, it is possible to provide analysis results from a viewpoint different from the etching rate.

The embodiments disclosed this time should be considered illustrative in all respects and not restrictive. The embodiments described above may be omitted, substituted, or modified in various ways without departing from the scope and spirit of the appended claims.

The various processes described in the above embodiments can be realized by executing a prepared program on a computer. Therefore, an example of a computer that executes a program having functions similar to those of the above-described embodiments will be described below. FIG. 12 is a diagram illustrating an example of a computer that executes programs.

As shown in FIG. 12, the computer 1000 has a computer 1010 that executes various arithmetic processes, an input device 1020 that receives data input, and a monitor 1030. The computer 1000 also has an interface device 1040 for connecting with various devices, and a communication device 1050 for connecting with other information processing devices or the like by wire or wirelessly. Computer 1000 also has RAM 1060 and storage device 1070 for temporarily storing various information. Each device 1010 - 1070 is also connected to a bus 1080 .

The storage device 1070 stores programs having functions similar to those of the acquisition unit 231, the prediction unit 232, the search unit 233, and the provision unit 234 shown in FIG. Also, the storage device 1070 stores material information 221 , chemical solution information 222 and model information 223 . The input device 1020 receives input of various information such as operation information from the user of the computer 1000, for example. The monitor 1030 displays various screens such as a display screen to the user of the computer 1000, for example. The interface device 1040 is connected with, for example, a printing device. The communication device 1050 is connected to, for example, a network (not shown) and exchanges various information with other information processing devices.

The computer 1010 performs various processes by reading each program stored in the storage device 1070, developing it in the RAM 1060, and executing it. Further, these programs can cause the computer 1000 to function as the acquiring unit 231, the predicting unit 232, the searching unit 233, and the providing unit 234 shown in FIG.

It should be noted that the above program does not necessarily have to be stored in the storage device 1070. For example, the computer 1000 may read and execute a program stored in a storage medium readable by the computer 1000 . Examples of storage media readable by the computer 1000 include portable recording media such as CD-ROMs, DVDs (Digital Versatile Discs), USB (Universal Serial Bus) memories, semiconductor memories such as flash memories, and hard disk drives. . Alternatively, the program may be stored in a device connected to a public line, the Internet, a LAN, etc., and the computer 1000 may read out the program from the device and execute the program.

Here, an example of a computer for realizing the prediction device 20 has been described, but the learning device 10 is also realized by a computer having the same configuration as the computer described here.

REFERENCE SIGNS LIST 10 learning device 11, 21 I/F section 12, 22 storage section 13, 23 control section 20 prediction device 121 experiment data DB
122, 221 Material information 123, 222 Chemical solution information 124, 223 Model information

Claims

A computer implemented prediction method comprising:
a prediction step of predicting an etching rate for input conditions using a model that outputs an etching rate based on feature values relating to films and chemicals in substrate etching;
a providing step of providing the etching rate predicted by the predicting step;
Forecasting methods, including
The prediction step predicts the etching rate using a model that outputs the etching rate based on feature amounts including the type of the film, film deposition conditions, the type of the chemical solution, and etching processing conditions. Item 1. The prediction method according to item 1.
2. The prediction according to claim 1, wherein the prediction step predicts the etching rate using a model that outputs the etching rate based on a feature amount including a value indicating the physical property of the film and a value indicating the physical property of the chemical solution. Method.
further comprising an acquisition step of acquiring a value indicating a physical property of the film and a value indicating the physical property of the chemical solution based on the type of the film and the type of the chemical solution included in the input conditions;
4. The prediction method according to claim 3, wherein the prediction step uses the model to predict the etching rate for the values obtained by the obtaining step.
Further comprising a search step of searching for feature values related to the film and the chemical based on the specified etching rate using the model,
The prediction method according to claim 1 or 2, wherein said providing step provides the feature amount searched for by said searching step.
The prediction step predicts the etching rate for the input processing temperature using a model that outputs the etching rate based on the feature amount including the etching processing temperature,
3. The prediction method according to claim 1, wherein said providing step provides activation energy calculated from a relationship between said input processing temperature and said etching rate predicted by said predicting step.
predicting the etching rate for the input conditions using a model that outputs the etching rate based on the feature values related to the film and the chemical liquid in the etching of the substrate;
A prediction program that causes a computer to execute a process of providing an etching rate predicted by the predicting process.
a prediction unit that predicts an etching rate for input conditions using a model that outputs an etching rate based on feature amounts related to films and chemicals in substrate etching;
a providing unit that provides the etching rate predicted by the predicting unit;
A prediction device having
A computer implemented learning method comprising:
a prediction step of predicting an etching rate for input conditions using a model that outputs an etching rate based on feature values relating to films and chemicals in substrate etching;
an update step of updating parameters of the model so that the difference between the etching rate predicted by the prediction step and the correct etching rate becomes smaller;
How to learn, including
predicting the etching rate for the input conditions using a model that outputs the etching rate based on the feature values related to the film and the chemical liquid in the etching of the substrate;
A learning program that causes a computer to execute a process of updating parameters of the model so that a difference between the etching rate predicted by the predicting process and a correct etching rate becomes smaller.
a prediction unit that predicts an etching rate for input conditions using a model that outputs an etching rate based on feature amounts related to films and chemicals in substrate etching;
an updating unit that updates the parameters of the model so that the difference between the etching rate predicted by the prediction unit and the correct etching rate becomes smaller;
A learning device having