Learning device, information processing device, substrate processing device, substrate processing system, learning method, and processing condition determination method

Abstract

To provide a learning device that is suitable for performing machine learn of a condition which is changed in accordance with the lapse of time for processing a substrate.SOLUTION: A learning device 200 comprises: an examination data acquisition part 261 that acquires a first processing amount indicating a difference in film thickness between before and after processing of a coated film after the processing of the coated film by driving a substrate processing apparatus for performing the processing of the coating film by supplying a processing liquid to a substrate on which the coating film is formed under a processing condition including a fluctuation condition that is fluctuated in accordance with the lapse of time; and a model generation part that generates a learning model that estimates a second processing amount indicating the difference in film thickness between before and after the processing of the coated film for the coated film formed on the substrate to which the processing of the coated film is performed by the substrate processing apparatus by performing the machine learning of learning data including the fluctuation condition and the first processing amount corresponded to the processing condition. The learning model comprises a first convolution neural network.SELECTED DRAWING: Figure 4

Description

Technical Field

[0001] The present invention relates to a learning device, an information processing device, a substrate processing device, a substrate processing system, a learning method, and a processing condition determination method, and to a learning device that generates a learning model for simulating a process according to processing conditions by a substrate processing device, an information processing device that determines processing conditions using the learning model, a substrate processing device including the information processing device, a substrate processing system including the information processing device and the learning device, a learning method executed by the learning device, and a processing condition determination method executed by the information processing device.

Background Art

[0002] In the semiconductor manufacturing process, there is a cleaning process. In the cleaning process, the film thickness of the film formed on the substrate is adjusted by an etching process in which a chemical solution is supplied to the substrate. In this film thickness adjustment, it is important to perform the etching process so that the surface of the substrate becomes uniform, or to flatten the surface of the substrate by the etching process. When discharging the etching solution from the nozzle to a part of the substrate, it is necessary to move the nozzle in the radial direction with respect to the substrate.

[0003] Patent Document 1 describes a liquid processing device capable of performing an etching process on a substrate by discharging an etching solution from a nozzle to the substrate. Patent Document 1 describes an example in which, while performing the etching process on the central region of the substrate, in order to make the in-plane temperature distribution of the wafer uniform, the etching solution is discharged while repeatedly reciprocating the etching nozzle between a first position on the central side through which the discharged etching solution passes through the center of the wafer and a second position on the peripheral side of the wafer closer to the center than this central side position.

[0004] The etching process is a complex process in which the processing amount of the film to be processed changes depending on the difference in the operation of moving the nozzle. Further, the processing amount of the film processed by the etching process is determined after processing the substrate. Therefore, the work of setting the operation of moving the nozzle requires trial and error by a technician. Considerable cost and time are required until the optimal operation of the nozzle is determined. [Prior art documents] [Patent Documents]

[0005] [Patent Document 1] Japanese Patent Publication No. 2015-103656 [Overview of the Initiative] [Problems that the invention aims to solve]

[0006] On the other hand, it is desirable to make the nozzle movement more complex. The nozzle movement represents time-series data indicating the position that changes over time. Making the nozzle movement more complex shortens the sampling interval, thus increasing the dimensionality of the time-series data. In general, as the dimensionality of the training data increases, the amount of data required for machine learning increases exponentially. Therefore, as the dimensionality of the training data increases, it becomes difficult to optimize the learning model obtained by machine learning. Also, since etching is a complex process, there may not be just one nozzle movement suitable for the target processing volume, but rather multiple such movements may exist.

[0007] One of the objectives of the present invention is to provide a learning device, a learning method, and a substrate processing system suitable for machine learning the conditions that change over time when processing a substrate.

[0008] Another object of the present invention is to provide an information processing device, a substrate processing device, a substrate processing system, and a method for determining processing conditions that can present multiple processing conditions for the processing results of a complex process for processing a substrate. [Means for solving the problem]

[0009] A learning device according to one aspect of the present invention includes: an experimental data acquisition unit that acquires a first processing amount indicating the difference in film thickness before and after processing a substrate processing device, which processes a film by supplying a processing liquid to a substrate on which a film has been formed, under processing conditions that include fluctuating conditions that change over time; and a model generation unit that uses machine learning to acquire a learning model that estimates a second processing amount indicating the difference in film thickness before and after processing a film formed on a substrate before it is processed by the substrate processing device, wherein the learning model includes a first convolutional neural network.

[0010] An information processing apparatus according to another aspect of the present invention is an information processing apparatus for managing a substrate processing apparatus, wherein the substrate processing apparatus processes a coating by supplying a processing liquid to a substrate on which a coating has been formed, using processing conditions that include fluctuating conditions that change over time, and a processing condition determination unit that determines processing conditions for driving the substrate processing apparatus using a learning model that estimates a second processing amount indicating the difference in film thickness before and after processing for a coating formed on a substrate before it is processed by the substrate processing apparatus, wherein the learning model includes a first convolutional neural network and is an inference model that has been machine-learned using learning data including fluctuating conditions included in the processing conditions in which the substrate processing apparatus processes a coating and a first processing amount indicating the difference in film thickness before and after processing for a coating formed on a substrate that has been processed by the substrate processing apparatus, and the processing condition determination unit provides the learning model with provisional fluctuating conditions and determines the processing conditions including the provisional fluctuating conditions as the processing conditions for driving the substrate processing apparatus if the second processing amount estimated by the learning model satisfies an acceptable condition.

[0011] A substrate processing system according to yet another aspect of the present invention is a substrate processing system for managing a substrate processing apparatus, comprising a learning device and an information processing device, wherein the substrate processing apparatus processes a coating by supplying a processing liquid to a substrate on which a coating has been formed, under processing conditions that include fluctuating conditions that change over time, the learning device includes an experimental data acquisition unit that acquires a first processing amount indicating the difference in film thickness before and after processing the coating after driving the substrate processing apparatus under the processing conditions, and a model generation unit that generates a learning model that uses machine learning on learning data including fluctuating conditions and the first processing amount corresponding to the processing conditions to estimate a second processing amount indicating the difference in film thickness before and after processing the coating for a coating formed on a substrate before it is processed by the substrate processing apparatus, the learning model includes a first convolutional neural network, and the information processing device includes a processing condition determination unit that uses the learning model generated by the learning device to determine processing conditions for driving the substrate processing apparatus, wherein the processing condition determination unit gives a provisional fluctuating condition to the learning model generated by the learning device and determines processing conditions including the provisional fluctuating condition as processing conditions for driving the substrate processing apparatus if the second processing amount estimated by the learning model satisfies an acceptable condition.

[0012] A learning method according to yet another aspect of the present invention involves having a computer perform the following steps: driving a substrate processing apparatus, which processes a film by supplying a processing solution to a substrate on which a film has been formed, under processing conditions that include variable conditions that change over time, and then obtaining a first processing amount that indicates the difference in film thickness before and after processing the film; and using machine learning on learning data including the variable conditions and the first processing amount corresponding to the processing conditions to generate a learning model that estimates a second processing amount that indicates the difference in film thickness before and after processing the film for a film formed on a substrate before it is processed by the substrate processing apparatus, wherein the learning model includes a first convolutional neural network.

[0013] A method for determining processing conditions according to yet another aspect of the present invention is a method for determining processing conditions executed by a computer that manages a substrate processing apparatus, wherein the substrate processing apparatus processes a coating by supplying a processing liquid to a substrate on which a coating has been formed, using processing conditions that include fluctuating conditions that change over time, and a process for determining processing conditions for driving the substrate processing apparatus using a learning model that estimates a second processing amount that indicates the difference in film thickness before and after processing of a coating formed on a substrate before it is processed by the substrate processing apparatus, wherein the learning model includes a first convolutional neural network and is an inference model that has been machine-learned using training data that includes fluctuating conditions included in the processing conditions in which the substrate processing apparatus processes a coating and a first processing amount that indicates the difference in film thickness before and after processing of a coating formed on a substrate that has been processed by the substrate processing apparatus, and the process for determining processing conditions includes providing a provisional fluctuating condition to the learning model and determining the processing condition that includes the provisional fluctuating condition as the processing condition for driving the substrate processing apparatus if the second processing amount estimated by the learning model satisfies an acceptable condition. [Effects of the Invention]

[0014] We can provide a learning device, a learning method, and a substrate processing system suitable for machine learning to process substrate conditions that change over time.

[0015] This invention provides an information processing device, a substrate processing device, a substrate processing system, and a method for determining processing conditions that can present multiple processing conditions for the processing results of a complex process for processing substrates. [Brief explanation of the drawing]

[0016] [Figure 1] This is a diagram illustrating the configuration of a substrate processing system according to one embodiment of the present invention. [Figure 2] This figure shows an example of the configuration of an information processing device. [Figure 3] This figure shows an example of the configuration of a learning device. [Figure 4] This figure shows an example of the functional configuration of a substrate processing system. [Figure 5] It is a diagram showing an example of film thickness characteristics. [Figure 6] It is a diagram for explaining a learning model. [Figure 7] It is a flowchart showing an example of the flow of learning processing. [Figure 8] It is a flowchart showing an example of the flow of processing condition determination processing. [Figure 9] It is a flowchart showing an example of the flow of additional learning processing. [Figure 10] It is a first diagram for explaining a learning model according to another embodiment. [Figure 11] It is a second diagram for explaining a learning model according to another embodiment.

Mode for Carrying Out the Invention

[0017] Hereinafter, a substrate processing system according to an embodiment of the present invention will be described in detail with reference to the drawings. In the following description, the substrate refers to a semiconductor substrate (semiconductor wafer), a substrate for FPD (Flat Panel Display) such as a liquid crystal display device or an organic EL (Electro Luminescence) display device, a substrate for an optical disk, a substrate for a magnetic disk, a substrate for a magneto-optical disk, a substrate for a photomask, a ceramic substrate, a substrate for a solar cell, or the like.

[0018] 1. Overall Configuration of Substrate Processing System FIG. 1 is a diagram for explaining the configuration of a substrate processing system according to an embodiment of the present invention. The substrate processing system 1 in FIG. 1 includes an information processing device 100, a learning device 200, and a substrate processing device 300. The learning device 200 is, for example, a server, and the information processing device 100 is, for example, a personal computer.

[0019] The learning device 200 and the information processing device 100 are used to manage the substrate processing device 300. Note that the substrate processing device 300 managed by the learning device 200 and the information processing device 100 is not limited to one unit; multiple substrate processing devices 300 may be managed.

[0020] In the substrate processing system 1 according to this embodiment, the information processing device 100, the learning device 200, and the substrate processing device 300 are connected to each other by wired or wireless communication lines or a communication network. The information processing device 100, the learning device 200, and the substrate processing device 300 are each connected to a network and are capable of sending and receiving data to and from each other. The network may be, for example, a local area network (LAN) or a wide area network (WAN). Alternatively, the network may be the internet. Furthermore, the information processing device 100 and the substrate processing device 300 may be connected by a dedicated communication network. The network connection may be wired or wireless.

[0021] Furthermore, the learning device 200 does not necessarily need to be connected to the substrate processing device 300 and the information processing device 100 by a communication line or communication network. In this case, data generated by the substrate processing device 300 may be passed to the learning device 200 via a recording medium. Alternatively, data generated by the learning device 200 may be passed to the information processing device 100 via a recording medium.

[0022] The substrate processing apparatus 300 is equipped with a display device, an audio output device, and an operating unit (not shown). The substrate processing apparatus 300 is operated according to predetermined processing conditions (processing recipe) of the substrate processing apparatus 300.

[0023] 2. Overview of substrate processing equipment The substrate processing apparatus 300 comprises a control device 10 and a plurality of substrate processing units WU. The control device 10 controls the plurality of substrate processing units WU. The plurality of substrate processing units WU process the substrate by supplying a processing solution to the substrate W on which a film has been formed. The processing solution includes an etching solution, and the substrate processing units WU perform an etching process. The etching solution is a chemical solution. Examples of etching solutions include hydrofluoric acid (a mixture of hydrofluoric acid (HF) and nitric acid (HNO3)), hydrofluoric acid, buffered hydrofluoric acid (BHF), ammonium fluoride, HFEG (a mixture of hydrofluoric acid and ethylene glycol), or phosphoric acid (H3PO4).

[0024] The substrate processing unit WU comprises a spin chuck SC, a spin motor SM, a nozzle 311, and a nozzle moving mechanism 301. The spin chuck SC holds the substrate W horizontally. The spin motor SM has a first rotation axis AX1. The first rotation axis AX1 extends in the vertical direction. The spin chuck SC is attached to the upper end of the first rotation axis AX1 of the spin motor SM. When the spin motor SM rotates, the spin chuck SC rotates around the first rotation axis AX1. The spin motor SM is a stepping motor. The substrate W held by the spin chuck SC rotates around the first rotation axis AX1. Therefore, the rotation speed of the substrate W is the same as the rotation speed of the stepping motor. If an encoder that generates a rotation speed signal indicating the rotation speed of the spin motor is provided, the rotation speed of the substrate W may be obtained from the rotation speed signal generated by the encoder. In this case, a motor other than a stepping motor can be used as the spin motor.

[0025] The nozzle 311 supplies etching solution to the substrate W. The nozzle 311 receives etching solution from an etching solution supply unit (not shown) and discharges the etching solution toward the rotating substrate W.

[0026] The nozzle moving mechanism 301 moves the nozzle 311 in a substantially horizontal direction. Specifically, the nozzle moving mechanism 301 includes a nozzle motor 303 having a second rotation axis AX2 and a nozzle arm 305. The nozzle motor 303 is positioned so that the second rotation axis AX2 is aligned substantially vertically. The nozzle arm 305 has a linear longitudinal shape. One end of the nozzle arm 305 is attached to the upper end of the second rotation axis AX2 such that the longitudinal direction of the nozzle arm 305 is different from that of the second rotation axis AX2. The nozzle 311 is attached to the other end of the nozzle arm 305 so that its discharge port faces downward.

[0027] When the nozzle motor 303 operates, the nozzle arm 305 rotates in the horizontal plane about the second rotation axis AX2. As a result, the nozzle 311 attached to the other end of the nozzle arm 305 moves horizontally (rotates) about the second rotation axis AX2. The nozzle 311 discharges etching solution toward the substrate W while moving horizontally. The nozzle motor 303 is, for example, a stepping motor.

[0028] The control device 10 includes a CPU (Central Processing Unit) and memory, and the CPU controls the entire substrate processing device 300 by executing a program stored in the memory. The control device 10 controls the spin motor SM and the nozzle motor 303.

[0029] The learning device 200 receives experimental data from the substrate processing device 300, uses the experimental data to train a learning model, and outputs the trained learning model to the information processing device 100.

[0030] The information processing device 100 uses a pre-trained model to determine the processing conditions for the substrate that the substrate processing device 300 is scheduled to process. The information processing device 100 outputs the determined processing conditions to the substrate processing device 300.

[0031] Figure 2 shows an example of the configuration of an information processing device. Referring to Figure 2, the information processing device 100 consists of a CPU 101, RAM (random access memory) 102, ROM (read-only memory) 103, storage device 104, operation unit 105, display device 106, and input / output interface 107. The CPU 101, RAM 102, ROM 103, storage device 104, operation unit 105, display device 106, and input / output interface 107 are connected to a bus 108.

[0032] RAM 102 is used as the working area for CPU 101. ROM 103 stores the system program. Storage device 104 includes a storage medium such as a hard disk or semiconductor memory and stores the program. The program may also be stored in ROM 103 or other external storage devices.

[0033] A CD-ROM 109 is removable from the storage device 104. The recording medium for storing the program executed by the CPU 101 is not limited to the CD-ROM 109, but may also be an optical disc (MO (Magnetic Optical Disc) / MD (Mini Disc) / DVD (Digital Versatile Disc)), an IC card, an optical card, a mask ROM, an EPROM (Erasable Programmable ROM), or other semiconductor memory. Furthermore, the CPU 101 may download a program from a computer connected to the network and store it in the storage device 104, or a computer connected to the network may write a program to the storage device 104, load the program stored in the storage device 104 into the RAM 102, and execute it on the CPU 101. The program referred to here includes not only programs that can be directly executed by the CPU 101, but also source programs, compressed programs, encrypted programs, etc.

[0034] The operation unit 105 is an input device such as a keyboard, mouse, or touch panel. The user can give predetermined instructions to the information processing device 100 by operating the operation unit 105. The display device 106 is a display device such as a liquid crystal display device, and displays a GUI (Graphical User Interface) or the like to receive instructions from the user. The input / output I / F 107 is connected to a network.

[0035] Figure 3 shows an example of the configuration of a learning device. Referring to Figure 3, the learning device 200 consists of a CPU 201, RAM 202, ROM 203, storage device 204, operation unit 205, display device 206, and input / output I / F 207. The CPU 201, RAM 202, ROM 203, storage device 204, operation unit 205, display device 206, and input / output I / F 207 are connected to a bus 208.

[0036] RAM202 is used as the working area for CPU201. ROM203 stores the system program. Storing device204 includes a storage medium such as a hard disk or semiconductor memory and stores the program. The program may also be stored in ROM203 or other external storage devices. A CD-ROM209 is removable from storage device204.

[0037] The control unit 205 is an input device such as a keyboard, mouse, or touch panel. The input / output interface 207 is connected to a network.

[0038] 3. Functional configuration of the substrate processing system Figure 4 shows an example of the functional configuration of a substrate processing system. Referring to Figure 4, the control device 10 of the substrate processing apparatus 300 controls the substrate processing unit WU to process the substrate W according to the processing conditions. The processing conditions are the conditions for processing the substrate W during a predetermined processing time. The processing time is the time determined for processing the substrate. In this embodiment, the processing time is the time while the nozzle 311 is discharging the etching solution onto the substrate W.

[0039] In this embodiment, the processing conditions include the temperature of the etching solution, the concentration of the etching solution, the flow rate of the etching solution, the rotation speed of the substrate W, and the relative position between the nozzle 311 and the substrate W. The processing conditions also include variable conditions that change over time. In this embodiment, the variable condition is the relative position between the nozzle 311 and the substrate W. The relative position is indicated by the rotation angle of the nozzle motor 303. The processing conditions also include fixed conditions that do not change over time. In this embodiment, the fixed conditions are the temperature of the etching solution, the concentration of the etching solution, the flow rate of the etching solution, and the rotation speed of the substrate W.

[0040] The learning device 200 trains a learning model with training data and generates an inference model that predicts the etching profile from the processing conditions. Hereinafter, the inference model generated by the learning device 200 will be referred to as the predictor.

[0041] The learning device 200 includes an experimental data acquisition unit 261, a predictor generation unit 265, and a predictor transmission unit 267. The functions of the learning device 200 are realized by the CPU 201 of the learning device 200, which executes a learning program stored in the RAM 202.

[0042] The experimental data acquisition unit 261 acquires experimental data from the substrate processing apparatus 300. The experimental data includes the processing conditions used when the substrate processing apparatus 300 actually processes the substrate W, and the film thickness characteristics of the coating formed on the substrate W before and after processing. The film thickness characteristics are indicated by the film thickness at multiple different positions in the radial direction of the substrate W.

[0043] Figure 5 shows an example of film thickness characteristics. Referring to Figure 5, the horizontal axis shows the radial position of the substrate, and the vertical axis shows the film thickness. The origin of the horizontal axis indicates the center of the substrate. The film thickness formed on the substrate W before processing by the substrate processing apparatus 300 is shown by the solid line. The film thickness formed on the substrate W is adjusted by the substrate processing apparatus 300 supplying etching solution according to processing conditions. The film thickness formed on the substrate W after processing by the substrate processing apparatus 300 is shown by the dotted line.

[0044] The difference between the film thickness formed on the substrate W before processing by the substrate processing apparatus 300 and the film thickness formed on the substrate W after processing by the substrate processing apparatus 300 is the processing amount (etching amount). The processing amount indicates the thickness of the film reduced by the process of supplying the etching solution by the substrate processing apparatus 300. The radial distribution of the processing amount is called the etching profile. The etching profile is shown by the processing amount at multiple different locations in the radial direction of the substrate W.

[0045] Furthermore, it is desirable that the film thickness formed by the substrate processing apparatus 300 be uniform across the entire surface of the substrate W. For this reason, a target film thickness is defined for the processing performed by the substrate processing apparatus 300. The target film thickness is indicated by a dashed line. The deviation characteristic is the difference between the film thickness formed on the substrate W after processing by the substrate processing apparatus 300 and the target film thickness. The deviation characteristic includes the difference at each of several locations in the radial direction of the substrate W.

[0046] Returning to Figure 4, the predictor generation unit 265 receives experimental data from the experimental data acquisition unit 261. The predictor generation unit 265 generates a predictor by supervising a neural network using the training data.

[0047] Specifically, the training data includes input data and ground truth data. The input data includes variable conditions included in the processing conditions of the experimental data and fixed conditions other than the variable conditions of the processing conditions included in the experimental data. The ground truth data includes etching profiles. The etching profile is the difference between the film thickness characteristics of the film before processing included in the experimental data and the film thickness characteristics of the film after processing included in the experimental data. This etching profile included in the ground truth data is an example of the first processing amount. The predictor generation unit 265 inputs the input data into the training model that will form the basis of the predictor and determines the parameters of the training model so that the difference between the output of the training model and the ground truth data is small. The predictor generation unit 265 generates a trained model as a predictor by incorporating the parameters set in the trained training model. The predictor is an inference program that incorporates the parameters set in the trained model. The predictor generation unit 265 transmits the predictor to the information processing device 100.

[0048] Figure 6 illustrates the learning model. Referring to Figure 6, the learning model consists of layers A through C arranged in this order from the input side to the output side (from the upper layer to the lower layer). Layer A contains the first convolutional neural network CNN1, layer B contains the fully connected neural network NN, and layer C contains the second convolutional neural network CNN2.

[0049] The first convolutional neural network (CNN1) receives the variable conditions as input. The fully connected neural network (NN) receives the output of the first convolutional neural network (CNN1) and the fixed conditions as input. The second convolutional neural network (CNN2) receives the output of the fully connected neural network (NN) as input.

[0050] The first convolutional neural network CNN1 includes multiple layers. In this embodiment, the first convolutional neural network CNN1 includes three layers. Within the first convolutional neural network CNN1, the first layer L1, the second layer L2, and the third layer L3 are arranged in this order from the input side (upper layer side) to the output side (lower layer side). In this embodiment, the case of including three layers as multiple layers is described, but it may include three or more layers.

[0051] The first layer L1, the second layer L2, and the third layer L3 each include a convolutional layer and a pooling layer. The convolutional layer has multiple filters. Multiple filters are applied in the convolutional layer. The pooling layer compresses the output of the convolutional layer. The number of filters in the convolutional layer of the second layer L2 is set to twice the number of filters in the convolutional layer of the first layer L1. The number of filters in the convolutional layer of the third layer L3 is set to twice the number of filters in the convolutional layer of the second layer L2. Therefore, as many features as possible can be extracted from the fluctuating conditions. Here, the fluctuating conditions include the relative position of the nozzle with respect to the substrate W, which fluctuates over time. Since the first convolutional neural network CNN1 extracts features using multiple filters, it extracts more features that include a time element regarding the change in the relative position of the nozzle with respect to the substrate W. Note that this example shows the number of filters in the second layer L2 convolutional layer being twice the number of filters in the first layer L1 convolutional layer, but it does not have to be twice. The number of filters in the second layer L2 convolutional layer only needs to be greater than the number of filters in the first layer L1 convolutional layer. Also, the number of filters in the third layer L3 convolutional layer does not have to be twice the number of filters in the second layer L2 convolutional layer. The number of filters in the third layer L3 convolutional layer only needs to be greater than the number of filters in the second layer L2 convolutional layer.

[0052] A fully connected neural network (NN) has multiple layers. In the example in Figure 6, the NN has two layers: an input layer (ba) and an output layer (bb). In the example in Figure 6, each layer contains multiple nodes. In the example in Figure 6, the ba layer has 5 nodes and the bb layer has 4 nodes, but the number of nodes is not limited to these. The number of nodes in the ba layer is set to be equal to the sum of the number of output nodes and the number of fixed conditions in the first convolutional neural network (CNN1). The number of nodes in the bb layer is set to be equal to the number of input nodes in the second convolutional neural network (CNN2). The outputs of the nodes in the ba layer are connected to the inputs of the nodes in the bb layer. Parameters include coefficients that weight the outputs of the nodes in the ba layer. One or more hidden layers may be provided between the ba and bb layers.

[0053] The second convolutional neural network (CNN2) includes multiple layers. In this embodiment, the second convolutional neural network (CNN2) includes three layers. In the second convolutional neural network (CNN2), the fourth layer L4, the fifth layer L5, and the sixth layer L6 are arranged in this order from the input side (upper layer side) to the output side (lower layer side). In this embodiment, the case of including three layers as multiple layers is described, but it may include three or more layers.

[0054] The fourth layer L4, the fifth layer L5, and the sixth layer L6 each include a convolutional layer and a pooling layer. The convolutional layer has multiple filters. Multiple filters are applied in the convolutional layer. The pooling layer compresses the output of the convolutional layer. The number of filters in the convolutional layer of the fifth layer L5 is set to half the number of filters in the convolutional layer of the fourth layer L4. Similarly, the number of filters in the convolutional layer of the sixth layer L6 is set to half the number of filters in the convolutional layer of the fifth layer L5. Therefore, as many features as possible can be extracted from the etching profile. The etching profile is represented by the difference in film thickness E[n] before and after processing at multiple positions P[n] (n is an integer of 1 or more) in the radial direction of the substrate W. Therefore, the multiple processing amounts in the etching profile fluctuate with changes in position in the radial direction of the substrate W. The second convolutional neural network (CNN2) extracts features using multiple filters, so as the processing load changes, it extracts more features that include the radial position element of the substrate W. Here, we show an example where the number of filters in the fifth layer L5 convolutional layer is set to half the number of filters in the fourth layer L4 convolutional layer, but it does not have to be half. The number of filters in the fifth layer L5 convolutional layer can be less than the number of filters in the fourth layer L4 convolutional layer. Also, the number of filters in the sixth layer L6 convolutional layer does not have to be half the number of filters in the fifth layer L5 convolutional layer. The number of filters in the sixth layer L6 convolutional layer can be less than the number of filters in the fifth layer L5 convolutional layer.

[0055] When variable and fixed conditions, which are input data, are given to the learning model, the learning model predicts an etching profile. This etching profile predicted by the learning model is an example of the second processing quantity. The difference between the etching profile predicted by the learning model and the correct data, the etching profile file, is calculated as the error. The learning model then learns to reduce this error. For example, the learning model uses backpropagation to update the values ​​of multiple filters in the first convolutional neural network CNN1, the weight parameters defined by multiple nodes in the fully connected neural network NN, and the multiple filters in the second convolutional neural network CNN2.

[0056] Returning to Figure 4, the information processing device 100 includes a processing condition determination unit 151, a predictor receiving unit 155, a prediction unit 159, an evaluation unit 161, and a processing condition transmission unit 163. The functions of the information processing device 100 are realized by the CPU 101 of the information processing device 100, which executes a processing condition determination program stored in the RAM 102. The predictor receiving unit 155 receives predictors transmitted from the learning device 200 and outputs the received predictors to the prediction unit 159.

[0057] The processing condition determination unit 151 determines the processing conditions for the substrate W to be processed by the substrate processing apparatus 300, and outputs the variable conditions and fixed conditions included in the processing conditions to the prediction unit 159.

[0058] The prediction unit 159 estimates the etching profile from the variable conditions and fixed conditions. Specifically, the prediction unit 159 inputs the variable conditions and fixed conditions received from the processing condition determination unit 151 into the predictor, and outputs the etching profile output by the predictor to the evaluation unit 161.

[0059] The evaluation unit 161 evaluates the etching profile input from the prediction unit 159 and outputs the evaluation result to the processing condition determination unit 151. Specifically, the evaluation unit 161 acquires the film thickness characteristics of the substrate W that the substrate processing apparatus 300 is scheduled to process before processing. The evaluation unit 161 calculates the predicted film thickness characteristics after etching from the etching profile input from the prediction unit 159 and the film thickness characteristics of the substrate W before processing, and compares it with the target film thickness characteristics. If the comparison result satisfies the evaluation criteria, the processing conditions determined by the processing condition determination unit 151 are output to the processing condition transmission unit 163. For example, the evaluation unit 161 calculates the deviation characteristics and determines whether the deviation characteristics satisfy the evaluation criteria. The deviation characteristics are the difference between the film thickness characteristics of the substrate W after etching and the target film thickness characteristics. The evaluation criteria can be arbitrarily determined. For example, the evaluation criteria may be that the maximum difference in the deviation characteristics is less than or equal to a threshold, or that the average difference is less than or equal to a threshold.

[0060] The processing condition transmission unit 163 transmits the processing conditions determined by the processing condition determination unit 151 to the control device 10 of the substrate processing apparatus 300. The substrate processing apparatus 300 processes the substrate W according to the processing conditions.

[0061] If the evaluation result does not meet the evaluation criteria, the evaluation unit 161 outputs the evaluation result to the processing condition determination unit 151. The evaluation result includes the film thickness characteristics predicted after the etching process or the difference between the film thickness characteristics predicted after the etching process and the target film thickness characteristics.

[0062] The processing condition determination unit 151 determines new processing conditions for the prediction unit 159 to predict, in response to the evaluation results input from the evaluation unit 161. The processing condition determination unit 151 selects one of several pre-prepared variable conditions using experimental design, pairwise method, or Bayesian estimation, and determines the processing conditions, which include the selected variable condition and fixed conditions, as new processing conditions for the prediction unit 159 to predict.

[0063] The processing condition determination unit 151 may use Bayesian estimation to search for processing conditions. If the evaluation unit 161 outputs multiple evaluation results, there will be multiple pairs of processing conditions and evaluation results. From the trends of the etching profiles in each of the multiple pairs, the system searches for processing conditions that result in a uniform film thickness or processing conditions that minimize the difference between the film thickness characteristics predicted after etching and the target film thickness characteristics.

[0064] Specifically, the processing condition determination unit 151 searches for processing conditions that minimize the objective function. The objective function is a function that indicates the uniformity of the film thickness or a function that indicates the agreement between the film thickness characteristics and the target film thickness characteristics. For example, the objective function is a function in which the difference between the film thickness characteristics predicted after etching and the target film thickness characteristics is expressed as a parameter. Here, the parameter is the corresponding variation condition. The corresponding variation condition is the variation condition used by the predictor to estimate the etching profile. The processing condition determination unit 151 selects a variation condition, which is a parameter determined by the search, from among a plurality of variation conditions, and determines a new processing condition that includes the selected variation condition and a fixed condition.

[0065] Figure 7 is a flowchart showing an example of the learning process flow. The learning process is performed by the CPU 201 of the learning device 200, which executes the learning program stored in the RAM 202.

[0066] Referring to Figure 7, the CPU 201 of the learning device 200 acquires experimental data. The CPU 201 controls the input / output I / F 107 to acquire experimental data from the substrate processing device 300 (step S11). The experimental data may also be acquired by reading experimental data recorded on a recording medium such as a CD-ROM 209 with the storage device 104. Multiple experimental data sets are acquired here. The experimental data includes processing conditions and the film thickness characteristics of the coating formed on the substrate W before and after processing. The film thickness characteristics are indicated by the film thickness of the coating formed on the substrate W at multiple different positions in the radial direction of the substrate W.

[0067] In the next step S12, experimental data to be processed is selected, and the process proceeds to step S13. In step S13, the variable conditions, fixed conditions, and etching profile included in the experimental data are set as training data. The etching profile is the difference between the film thickness characteristics of the film before processing included in the experimental data and the film thickness characteristics of the film after processing included in the experimental data. The training data includes input data and ground truth data. In this embodiment, the variable conditions and fixed conditions included in the experimental data are set as input data, and the etching profile is set as ground truth data.

[0068] In the next step, S14, the CPU 201 trains the learning model and proceeds to step S15. Input data is fed into the learning model, and filters and parameters are determined to minimize the error between the learning model's output and the ground truth data. This adjusts the filters and parameters of the learning model.

[0069] In step S15, it is determined whether the adjustment is complete. Training data to be used to evaluate the learning model is prepared in advance, and the performance of the learning model is evaluated using the evaluation training data. If the evaluation result meets the predetermined evaluation criteria, it is determined that the adjustment is complete. If the evaluation result does not meet the evaluation criteria (NO in step S15), the process returns to step S12, but if the evaluation result meets the evaluation criteria (YES in step S15), the process proceeds to step S16.

[0070] If the process returns to step S12, in step S12, experimental data that was not selected for processing from the experimental data acquired in step S11 is selected. In the loop from step S12 to step S15, the CPU 201 trains the learning model using multiple training data. This adjusts the filters and parameters of the learning model to appropriate values. In step S16, the learning parameters of the trained model are stored. In step S17, the trained model is set as the predictor, the predictor is sent to the information processing device 100, and the process ends. The CPU 201 controls the input / output I / F 107 and sends the predictor to the information processing device 100.

[0071] Figure 8 is a flowchart showing an example of the processing condition determination process. The processing condition determination process is a process executed by the CPU 101 of the information processing device 100 when the CPU 101 executes a processing condition determination program stored in the RAM 102.

[0072] Referring to Figure 8, the CPU 101 of the information processing device 100 selects one of several pre-prepared variation conditions (step S21) and proceeds to step S22. One of several pre-prepared variation conditions is selected using experimental design, pairwise method, or Bayesian estimation, etc.

[0073] In step S22, the etching profile is estimated from the variable and fixed conditions using a predictor, and the process proceeds to step S23. The variable and fixed conditions are input to the predictor, and the etching profile output by the predictor is obtained. In step S23, the film thickness characteristics after processing are compared with the target film thickness characteristics. The film thickness characteristics after processing the substrate W are calculated from the film thickness characteristics of the substrate W to be processed by the substrate processing apparatus 300 before processing and the etching profile estimated in step S22. Then, the film thickness characteristics after processing are compared with the target film thickness characteristics. Here, the difference between the film thickness characteristics after processing the substrate W and the target film thickness characteristics is calculated.

[0074] In step S24, it is determined whether the comparison result meets the evaluation criteria. If the comparison result meets the evaluation criteria (YES in step S24), the process proceeds to step S25; otherwise, the process returns to step S21. For example, the evaluation criteria are met if the maximum difference is less than or equal to the threshold. Also, the evaluation criteria are met if the average difference is less than or equal to the threshold.

[0075] In step S25, processing conditions including the variable conditions selected in step S21 are set as candidates for processing conditions to drive the substrate processing device 300, and the process proceeds to step S26. In step S26, it is determined whether or not a search termination instruction has been received. If a termination instruction is received by the user operating the information processing device 100, the process proceeds to step S27; otherwise, the process returns to step S21. Alternatively, instead of a termination instruction entered by the user, it may be determined whether or not a predetermined number of processing conditions have been set as candidates.

[0076] In step S27, one processing condition is selected from the one or more processing conditions set as candidates, and the process proceeds to step S28. Alternatively, one processing condition may be selected by the user operating the information processing device 100 from the one or more processing conditions set as candidates. This expands the range of selection for the user. Furthermore, the variation condition that results in the simplest nozzle operation may be automatically selected from among the variation conditions included in the multiple processing conditions. The variation condition that results in the simplest nozzle operation can be, for example, the variation condition with the fewest number of speed change points. This makes it possible to present multiple variation conditions for the processing results of complex nozzle operations that process the substrate W. By selecting a variation condition that makes nozzle control easy from among the multiple variation conditions, the control of the substrate processing device 300 becomes easier.

[0077] In step S28, the processing conditions, including the variable conditions determined in step S28, are transmitted to the substrate processing device 300, and the processing ends. The CPU 101 controls the input / output I / F 107 to transmit the processing conditions to the substrate processing device 300. When the substrate processing device 300 receives the processing conditions from the information processing device 100, it processes the substrate W according to those processing conditions.

[0078] 4. Specific Examples In this embodiment, the variable conditions are time-series data sampled with a nozzle operation processing time of 60 seconds and a sampling interval of 0.01 seconds. The variable conditions consist of 6001 values. Therefore, the variable conditions can represent complex nozzle operations. In particular, nozzle operations with a relatively large number of speed change points can be accurately represented by the variable conditions. On the other hand, because the variable conditions have a large number of dimensions, overfitting may occur when the time-series data of the variable conditions is used to train a fully connected neural network model.

[0079] In this embodiment, the predictor generation unit 265 uses a learning model, including the convolutional neural network shown in Figure 6, to learn the variable and fixed conditions. The inventors have experimentally discovered that the desired result can be obtained as an etching profile predicted by the predictor trained on the learning model shown in Figure 6, using the variable and fixed conditions, which consist of 6001 values ​​representing complex nozzle operation.

[0080] Furthermore, in this embodiment, when the processing condition determination unit 151 searches for processing conditions, it searches for processing conditions corresponding to different etching profiles, so that processing conditions corresponding to multiple different etching profiles are selected. For this reason, the processing condition determination unit 151 can efficiently search for processing conditions from among multiple processing conditions in which the target etching profile is predicted.

[0081] Although an example using a sampling interval of 0.01 seconds was explained, the sampling interval is not limited to this. A longer or shorter sampling interval may be used. For example, the sampling interval may be 0.1 seconds or 0.005 seconds.

[0082] 5. Other Embodiments (1) In the embodiment described above, the learning device 200 generates a predictor based on the training data. The learning device 200 may also perform additional training on the predictor. After the predictor is generated, the learning device 200 obtains the film thickness characteristics and processing conditions of the substrate W processed by the substrate processing device 300 before and after processing. The learning device 200 then generates training data from the film thickness characteristics and processing conditions before and after processing, and performs machine learning on the predictor to perform additional training. The configuration of the neural network constituting the predictor is not changed by the additional training, but the parameters are adjusted.

[0083] The information obtained from the actual processing of the substrate W by the substrate processing device 300 is used to train the predictor through machine learning, thereby improving the accuracy of the predictor. Furthermore, the amount of training data used to generate the predictor can be minimized.

[0084] Figure 9 is a flowchart showing an example of the flow of the additional learning process. The additional learning process is a process executed by the CPU 201 of the learning device 200, which executes the additional learning program stored in the RAM 202. The additional learning program is part of the learning program.

[0085] Referring to Figure 9, the CPU 201 of the learning device 200 acquires production data (step S31) and proceeds to step S32. The production data includes the processing conditions when the substrate processing device 300 processes the substrate W after the predictor is generated, and the film thickness characteristics of the coating before and after processing. The CPU 201 controls the input / output I / F 107 to acquire production data from the substrate processing device 300. The production data may also be acquired by reading experimental data recorded on a recording medium such as a CD-ROM 209 with the storage device 104.

[0086] In step S32, the variable conditions, the fixed conditions included in the processing conditions of the production data, and the etching profile are set as training data. The etching profile is the difference between the film thickness characteristics of the film before processing included in the production data and the film thickness characteristics of the film after processing included in the production data. The variable conditions and the fixed conditions included in the processing conditions are set as input data. The etching profile is set as the ground truth data.

[0087] In the next step, S33, the CPU 201 further trains the predictor and proceeds to step S34. Input data is fed into the predictor, and filters and parameters are determined to minimize the difference between the predictor's output and the ground truth data. This further refines the predictor's filters and parameters.

[0088] In step S34, it is determined whether the adjustment is complete. The performance of the predictor is evaluated using the training data for evaluation. The adjustment is determined to be complete if the evaluation result meets the predetermined additional training evaluation criteria. The additional training evaluation criteria are higher than the evaluation criteria used when the predictor was generated. If the evaluation result does not meet the additional training evaluation criteria (NO in step S34), the process returns to step S31, but if the evaluation result meets the additional training evaluation criteria (YES in step S34), the process ends.

[0089] (2) The learning device 200 may generate a distillation model that has been trained on a new learning model using distillation data including processing conditions determined by the information processing device 100 and etching profiles predicted by a predictor from those processing conditions. This makes it easier to prepare data for training the new learning model.

[0090] (3) In this embodiment, the training data used to generate the predictor includes variable conditions and fixed conditions in the input data. The present invention is not limited thereto. The input data may include only variable conditions and not include fixed conditions.

[0091] (4) In this embodiment, the relative position between the nozzle 311 and the substrate W was shown as an example of a variable condition, but the present invention is not limited thereto. If at least one of the etching solution temperature, etching solution concentration, etching solution flow rate, and substrate W rotation speed fluctuates over time, these may be used as variable conditions. Furthermore, the variable conditions are not limited to one type, but may include multiple types.

[0092] Figure 10 is the first diagram illustrating a learning model according to another embodiment. Here, we will explain using the case where the flow rate of the etching solution discharged from the nozzle fluctuates over time. In this case, the fluctuating conditions include the flow rate of the etching solution that fluctuates over time. In this case, the learning model shown in Figure 10 is used. The difference between the learning model shown in Figure 10 and the learning model shown in Figure 6 is that the fluctuating conditions input to the first convolutional neural network CNN1 include a position condition indicating the relative position of the nozzle to the substrate that fluctuates over time, and a flow rate condition indicating the flow rate of the etching solution that fluctuates over time. For this reason, the first convolutional neural network CNN1 performs a 2-channel convolution process.

[0093] In this case, the positional condition and flow rate condition represent the relative position of the nozzle to the substrate and the flow rate of the etching solution, respectively, at the same time. Therefore, when training the positional condition and flow rate condition, it is possible to train them while retaining temporal information. Furthermore, since a single first convolutional neural network CNN1 is used, the number of training parameters can be reduced, and overfitting can be suppressed.

[0094] Furthermore, in the learning model, the positional conditions and flow rate conditions may be processed by separate convolutional neural networks. Figure 11 is a second diagram illustrating a learning model according to another embodiment. Referring to Figure 11, a first convolutional neural network CNN1 that processes nozzle conditions and a third convolutional neural network CNN3 that processes flow rate conditions are provided on the input side of the fully connected neural network NN.

[0095] (5) In the above embodiment, the learning model includes a first convolutional neural network CNN1, a fully connected neural network NN, and a second convolutional neural network CNN2, but the present invention is not limited thereto. For example, in the predictor, either or both of the fully connected neural network NN and the second convolutional neural network CNN2 may not be provided.

[0096] (6) Although the present invention has been described using the example of the information processing device 100 and the learning device 200 being separate from the substrate processing device 300, the present invention is not limited thereto. The information processing device 100 may be incorporated into the substrate processing device 300. Furthermore, the information processing device 100 and the learning device 200 may be incorporated into the substrate processing device 300. Also, although the information processing device 100 and the learning device 200 were described as separate devices, they may be configured as a single integrated device.

[0097] Effects in the embodiment of 6. In the learning device 200 of the above embodiment, since the variable conditions are values ​​that change over time, features that take time into account can be extracted by using the first convolutional neural network CNN1. Furthermore, by training the first convolutional neural network CNN1, the number of learning parameters can be reduced, thereby improving the generalization performance of the learning model.

[0098] Furthermore, since the processing load is defined for each of multiple different positions in the radial direction of the substrate, training the second convolutional neural network (CNN2) with these processing loads allows for the extraction of features that take into account the radial positional elements of the substrate. This also reduces the number of training parameters and improves the generalization performance of the trained model.

[0099] Furthermore, a fully connected neural network (NN) is provided between the first convolutional neural network (CNN1) and the second convolutional neural network (CNN2). In this case, the number of outputs of the first convolutional neural network (CNN1) and the number of inputs of the second convolutional neural network (CNN2) can be adjusted by the fully connected neural network (NN). Also, because the number of outputs of the first convolutional neural network (CNN1) and the number of inputs of the second convolutional neural network (CNN2) can be adjusted by the fully connected neural network (NN), machine learning can proceed well even if the number of outputs of the first convolutional neural network (CNN1) and the number of inputs of the second convolutional neural network (CNN2) do not need to be matched. Moreover, since the number of outputs of the first convolutional neural network (CNN1) and the number of inputs of the second convolutional neural network (CNN2) do not need to be matched, machine learning can be performed on training data with a higher number of dimensions. Therefore, machine learning can be performed on variable conditions with a higher number of dimensions. Also, machine learning can be performed on fixed conditions with a higher number of dimensions, and machine learning can be performed on a larger variety of conditions included in the processing conditions for driving the substrate processing device.

[0100] Furthermore, within the first convolutional neural network CNN1, the number of filters increases from the upper layers to the lower layers, making it possible to extract more features related to the variable conditions. Similarly, within the second convolutional neural network CNN2, the number of filters decreases from the upper layers to the lower layers, making it possible to extract more features that take into account the positions of multiple processing loads. As a result, the generalization performance of the learning device 200 can be improved.

[0101] Furthermore, since the learning model includes a first-order convolutional neural network (CNN1), it is possible to generate a learning model with improved generalization performance even when there is a large amount of data with varying conditions.

[0102] 7. Correspondence between each component of the claim and each part of the embodiment Substrate W is an example of a substrate, etching solution is an example of a processing solution, substrate processing apparatus 300 is an example of a substrate processing apparatus, experimental data acquisition unit 261 is an example of an experimental data acquisition unit, predictor is an example of a learning model, and predictor generation unit 265 is an example of a model generation unit. Furthermore, information processing apparatus 100 is an example of an information processing apparatus, variable condition generation unit 251 is an example of a variable condition generation unit, nozzle 311 is an example of a nozzle that supplies processing solution to the substrate, nozzle moving mechanism 301 is an example of a moving unit, and prediction unit 159, evaluation unit 161, and processing condition determination unit 151 are examples of processing condition determination units.

[0103] 8. Summary of Embodiments (Section 1) A learning device according to one aspect of the present invention is: An experimental data acquisition unit acquires a first processing amount that indicates the difference in film thickness before and after processing the film, after driving a substrate processing apparatus that processes the film by supplying a processing liquid to a substrate on which a film has been formed, under processing conditions that include fluctuating conditions that change over time, and The system includes a model generation unit that uses machine learning to generate a learning model that estimates a second processing amount, which indicates the difference in film thickness before and after processing of the coating formed on the substrate before the coating is processed by the substrate processing apparatus, using learning data including the variable conditions and the first processing amount corresponding to the processing conditions. The aforementioned learning model includes a first convolutional neural network.

[0104] According to the learning device described in Section 1, since the variable conditions are values ​​that change over time, features that take time into account can be extracted by using a convolutional neural network. Furthermore, by using a convolutional neural network, the number of learning parameters can be reduced, thereby improving the generalization performance of the learning model. As a result, it becomes possible to provide a learning device suitable for machine learning the conditions that change over time in order to process a substrate.

[0105] (Article 2) In the learning device described in Article 1, The first processing amount and the second processing amount are the difference in film thickness before and after processing of the coating at each of a plurality of different positions in the radial direction of the substrate. The learning model may further include a second convolutional neural network that outputs the first processing amount or the second processing amount.

[0106] According to the learning device described in Section 2, the first and second processing parameters are determined for each of several different positions in the radial direction of the substrate. By training a convolutional neural network with either the first or second processing parameter, features that take into account the radial positional element of the substrate can be extracted. Furthermore, the number of learning parameters can be reduced, and the generalization performance of the learning model can be improved.

[0107] (Article 3) In the learning device described in Article 2, The learning model further includes a fully connected neural network to which the output of the first convolutional neural network and the fixed conditions other than the variable conditions among the processing conditions are input. The second convolutional neural network may receive the output of the fully connected neural network as input.

[0108] According to the learning device described in Section 3, a fully connected neural network is provided between the first convolutional neural network and the second convolutional neural network. In this case, the number of features output from the first convolutional neural network and the number of features input to the second convolutional neural network can be adjusted by the fully connected neural network.

[0109] (Article 4) In the learning device described in Article 2 or Article 3, The number of filters used in each of the multiple layers of the aforementioned first convolutional neural network is such that the number of filters used in the lower layer is double the number of filters used in the upper layer. The number of filters used in each of the multiple layers of the second convolutional neural network may be such that the number of filters used in the lower layer is half the number of filters used in the upper layer.

[0110] According to the learning device described in Section 4, the number of filters increases from the upper layers to the lower layers within the first convolutional neural network, making it possible to extract more features of the variable conditions. Furthermore, within the second convolutional neural network, the number of filters decreases from the upper layers to the lower layers, making it possible to extract more features of multiple processing quantities. As a result, it becomes possible to improve the accuracy of the learning device.

[0111] (Article 5) In the learning device described in any one of paragraphs 1 to 4, The substrate processing apparatus supplies the processing liquid to the substrate by moving a nozzle that supplies the processing liquid to the substrate. The aforementioned fluctuation conditions may include nozzle movement conditions that indicate the relative position of the nozzle with respect to the substrate, which fluctuates over time.

[0112] According to the learning device described in Section 5, the nozzle movement conditions are input to the first convolutional neural network. Therefore, even when there is a large amount of data on nozzle movement conditions, a learning model with improved generalization performance can be generated.

[0113] (Item 6) In the learning device described in Item 5, The aforementioned variable conditions may further include discharge flow rate conditions that indicate the flow rate of the processing liquid discharged from the nozzle, which changes over time.

[0114] According to the learning device described in Section 6, it is possible to generate a learning model with improved generalization performance even when there is a large amount of data on discharge flow rate conditions.

[0115] (Section 7) An information processing device according to another aspect of the present invention is: An information processing device for managing a substrate processing device, The substrate processing apparatus processes the coating by supplying a processing liquid to the substrate on which the coating has been formed, under processing conditions that include fluctuating conditions that change over time. The system includes a processing condition determination unit that determines processing conditions for driving the substrate processing apparatus, using a learning model that estimates a second processing amount indicating the difference in film thickness before and after processing of the coating formed on the substrate before processing the coating by the substrate processing apparatus, The learning model includes a first convolutional neural network and is an inference model that has been machine-trained using training data including the variable conditions included in the processing conditions under which the substrate processing apparatus processed the coating and a first processing amount indicating the difference in film thickness before and after processing of the coating formed on the substrate processed by the substrate processing apparatus. The processing condition determination unit provides the learning model with provisional variation conditions and determines the processing conditions, including the provisional variation conditions, as the processing conditions for driving the substrate processing apparatus if the second processing amount estimated by the learning model satisfies the allowable conditions.

[0116] According to the information processing device described in Section 7, a learning model is given a provisional fluctuation condition that changes over time. When the processing amount estimated by the learning model satisfies the acceptable conditions, the processing conditions including the provisional fluctuation condition are determined as the processing conditions for driving the substrate processing device. Therefore, multiple provisional fluctuation conditions can be determined for a processing amount that satisfies the acceptable conditions. As a result, it becomes possible to present multiple processing conditions for the processing results of a complex process that processes a substrate.

[0117] (Clause 8) The substrate processing apparatus may include the information processing apparatus described in Clause 7.

[0118] According to the substrate processing apparatus described in Section 8, it becomes possible to present multiple processing conditions for the processing results of a complex process that processes substrates.

[0119] (Section 9) A substrate processing system according to another aspect of the present invention is: A substrate processing system for managing substrate processing equipment, Equipped with a learning device and an information processing device, The substrate processing apparatus processes the coating by supplying a processing liquid to the substrate on which the coating has been formed, under processing conditions that include fluctuating conditions that change over time. The learning device includes an experimental data acquisition unit that, after driving the substrate processing apparatus under the processing conditions to process the coating formed on the substrate, acquires a first processing amount indicating the difference in film thickness before and after processing the coating, The system includes a model generation unit that uses machine learning to generate a learning model that estimates a second processing amount, which indicates the difference in film thickness before and after processing of the coating formed on the substrate before the coating is processed by the substrate processing apparatus, using learning data including the variable conditions and the first processing amount corresponding to the processing conditions. The aforementioned learning model includes a first convolutional neural network, The information processing device includes a processing condition determination unit that determines processing conditions for driving the substrate processing device using the learning model generated by the learning device, The processing condition determination unit assigns a provisional variation condition to the learning model generated by the learning device, and determines the processing condition including the provisional variation condition as the processing condition for driving the substrate processing device if the second processing amount estimated by the learning model satisfies the allowable condition.

[0120] The substrate processing system described in Section 9 is suitable for machine learning to process substrates under conditions that change over time, and it is possible to present multiple processing conditions for the processing results of a complex substrate processing process.

[0121] (Section 10) Another learning method according to another aspect of the present invention is: A substrate processing apparatus that processes a coating by supplying a processing liquid to a substrate on which a coating has been formed is driven under processing conditions that include fluctuating conditions that change over time, and after processing the coating, a first processing amount is obtained that indicates the difference in film thickness before and after processing the coating. The process of having a computer perform machine learning on training data including the variable conditions and the first processing amount corresponding to the processing conditions to generate a learning model that estimates a second processing amount that indicates the difference in film thickness before and after processing of the coating formed on the substrate before the coating is processed by the substrate processing apparatus, The aforementioned learning model includes a first convolutional neural network.

[0122] According to the learning method described in Section 10, the learning model includes a convolutional neural network. Therefore, it is possible to provide a learning method suitable for machine learning to process conditions that change over time in order to process the substrate.

[0123] (Section 11) A method for determining processing conditions according to another aspect of the present invention is: A method for determining processing conditions, which is executed by a computer that manages a substrate processing apparatus, The substrate processing apparatus processes the coating by supplying a processing liquid to the substrate on which the coating has been formed, under processing conditions that include fluctuating conditions that change over time. The process includes determining processing conditions for driving the substrate processing apparatus, using a learning model that estimates a second processing amount indicating the difference in film thickness before and after processing of the coating formed on the substrate before processing the coating by the substrate processing apparatus, The learning model includes a first convolutional neural network and is an inference model that has been machine-trained using training data including the variable conditions included in the processing conditions under which the substrate processing apparatus processed the coating and a first processing amount indicating the difference in film thickness before and after processing of the coating formed on the substrate processed by the substrate processing apparatus. The process for determining the processing conditions includes providing the learning model with provisional variation conditions and determining the processing conditions, including the provisional variation conditions, as the processing conditions for driving the substrate processing apparatus if the second processing amount estimated by the learning model satisfies the acceptable conditions.

[0124] The substrate condition determination method described in Section 11 provides a method for determining processing conditions that can present multiple processing conditions for the processing results of a complex process that processes a substrate. [Explanation of symbols]

[0125] 1…Substrate processing system, 100…Information processing device, 151…Processing condition determination unit, 155…Predictor receiving unit, 159…Prediction unit, 161…Evaluation unit, 163…Processing condition transmission unit, 200…Learning device, 251…Variation condition generation unit, 261…Experimental data acquisition unit, 265…Predictor generation unit, 267…Predictor transmission unit, 300…Substrate processing device, 301…Nozzle movement mechanism, 303…Nozzle motor, 305…Nozzle arm, 311…Nozzle, AX1…First rotation axis, AX2…Second rotation axis, CNN1…First convolutional neural network, CNN2…Second convolutional neural network, CNN3…Third convolutional neural network, L1~L6…Layer 1~Layer 6, NN…Fully connected neural network, SC…Spin chuck, SM…Spin motor, W…Substrate, WU…Substrate processing unit

Claims

1. An experimental data acquisition unit acquires a first processing amount that indicates the difference in film thickness before and after processing the film, after driving a substrate processing apparatus that processes the film by supplying a processing liquid to a substrate on which a film has been formed, under processing conditions that include fluctuating conditions that change over time, and The system includes a model generation unit that uses machine learning to generate a learning model that estimates a second processing amount, which indicates the difference in film thickness before and after processing of the coating formed on the substrate before the coating is processed by the substrate processing apparatus, using learning data including the variable conditions and the first processing amount corresponding to the processing conditions. The aforementioned learning model is a learning device including a first convolutional neural network.

2. The first processing amount and the second processing amount are the difference in film thickness before and after processing of the coating at each of a plurality of different positions in the radial direction of the substrate. The learning device according to claim 1, further comprising a second convolutional neural network that outputs the first processing amount or the second processing amount, the learning model.

3. The learning model further includes a fully connected neural network to which the output of the first convolutional neural network and the fixed conditions other than the variable conditions among the processing conditions are input. The learning device according to claim 2, wherein the second convolutional neural network is input to the output of the fully connected neural network.

4. The number of filters used in each of the multiple layers of the first convolutional neural network is such that the number of filters used in the lower layer is double the number of filters used in the upper layer. The learning device according to claim 2, wherein the number of filters used in each of the multiple layers of the second convolutional neural network is such that the number of filters used in the lower layer is half the number of filters used in the upper layer.

5. The substrate processing apparatus supplies the processing liquid to the substrate by moving a nozzle that supplies the processing liquid to the substrate. The learning apparatus according to any one of claims 1 to 4, wherein the variable conditions include nozzle movement conditions indicating the relative position of the nozzle with respect to the substrate, which changes over time.

6. The learning device according to claim 5, wherein the variable conditions further include discharge flow rate conditions indicating the flow rate of the processing liquid discharged from the nozzle, which changes over time.

7. An information processing device for managing a substrate processing device, The substrate processing apparatus processes the coating by supplying a processing liquid to the substrate on which the coating has been formed, under processing conditions that include fluctuating conditions that change over time. The system includes a processing condition determination unit that determines processing conditions for driving the substrate processing apparatus, using a learning model that estimates a second processing amount indicating the difference in film thickness before and after processing of the coating formed on the substrate before processing the coating by the substrate processing apparatus, The learning model includes a first convolutional neural network and is an inference model that has been machine-trained using training data including the variable conditions included in the processing conditions under which the substrate processing apparatus processes the coating and a first processing amount indicating the difference in film thickness before and after processing of the coating formed on the substrate processed by the substrate processing apparatus. The processing condition determination unit provides the learning model with a provisional variation condition and determines the processing condition including the provisional variation condition as the processing condition for driving the substrate processing device when the second processing amount estimated by the learning model satisfies an acceptable condition.

8. A substrate processing apparatus comprising the information processing apparatus described in claim 7.

9. A substrate processing system for managing substrate processing equipment, Equipped with a learning device and an information processing device, The substrate processing apparatus processes the coating by supplying a processing liquid to the substrate on which the coating has been formed, under processing conditions that include fluctuating conditions that change over time. The learning device includes an experimental data acquisition unit that, after driving the substrate processing apparatus under the processing conditions to process the coating formed on the substrate, acquires a first processing amount indicating the difference in film thickness before and after processing the coating, The system includes a model generation unit that uses machine learning to generate a learning model that estimates a second processing amount, which indicates the difference in film thickness before and after processing of the coating formed on the substrate before the coating is processed by the substrate processing apparatus, using learning data including the variable conditions and the first processing amount corresponding to the processing conditions. The aforementioned learning model includes a first convolutional neural network, The information processing device includes a processing condition determination unit that determines processing conditions for driving the substrate processing device using the learning model generated by the learning device, A substrate processing system in which the processing condition determination unit provides a provisional variation condition to the learning model generated by the learning device, and determines the processing condition including the provisional variation condition as the processing condition for driving the substrate processing device when the second processing amount estimated by the learning model satisfies an acceptable condition.

10. A substrate processing apparatus that processes a coating by supplying a processing liquid to a substrate on which a coating has been formed is driven under processing conditions that include fluctuating conditions that change over time, and after processing the coating, a first processing amount is obtained that indicates the difference in film thickness before and after processing the coating. The process of having a computer perform machine learning on training data including the variable conditions and the first processing amount corresponding to the processing conditions to generate a learning model that estimates a second processing amount that indicates the difference in film thickness before and after processing of the coating formed on the substrate before the coating is processed by the substrate processing apparatus, The learning model includes a learning method comprising a first convolutional neural network.

11. A method for determining processing conditions, which is executed by a computer that manages a substrate processing apparatus, The substrate processing apparatus processes the coating by supplying a processing liquid to the substrate on which the coating has been formed, under processing conditions that include fluctuating conditions that change over time. The process includes determining processing conditions for driving the substrate processing apparatus, using a learning model that estimates a second processing amount indicating the difference in film thickness before and after processing of the coating formed on the substrate before processing the coating by the substrate processing apparatus, The learning model includes a first convolutional neural network and is an inference model that has been machine-trained using training data including the variable conditions included in the processing conditions under which the substrate processing apparatus processes the coating and a first processing amount indicating the difference in film thickness before and after processing of the coating formed on the substrate processed by the substrate processing apparatus. A method for determining the processing conditions, comprising the process of providing a learning model with a provisional variation condition and determining the processing conditions including the provisional variation condition as the processing conditions for driving the substrate processing apparatus when the second processing amount estimated by the learning model satisfies an acceptable condition.

Images