Computer program, information processing method, and information processing device

Abstract

Provided are a computer program, an information processing method, and an information processing device that can be expected to assist with processing which is performed in response to the condition of a substrate processing device.　A computer program according to this embodiment causes a computer to execute processing to acquire information relating to substrate processing that is performed by a substrate processing device, determine whether an automatic response to the condition of the substrate processing is possible on the basis of the acquired information, and if it has been determined that an automatic response is not possible, output the information relating to the substrate processing and a suggestion for a response to the condition, receive a determination regarding the content of the response to the condition, and output information relating to the content of the response to the condition for which the determination was received.

Description

Computer program, information processing method, and information processing device

[0001] The present disclosure relates to a computer program, an information processing method, and an information processing device.

[0002] Patent Literature 1 proposes a substrate processing system in which a computing device connected to multiple process chambers receives a first measurement value generated by a sensor in a first process chamber during or after a process is performed in the first process chamber, processes the first measurement value using a first trained machine learning model, and determines whether maintenance of the first process chamber is required. In this substrate processing system, the computing device receives a second measurement value generated by the sensor when an adjustment process is performed in the first process chamber after maintenance of the first process chamber is performed, and determines whether the first process chamber is ready for restart using a second trained machine learning model.

[0003] US Patent Application Publication No. 2022 / 0214662

[0004] The present disclosure provides a computer program, an information processing method, and an information processing apparatus that are expected to support processing to deal with situations in a substrate processing apparatus.

[0005] A computer program according to one embodiment acquires information about substrate processing performed by a substrate processing apparatus, determines based on the acquired information whether an automatic response to the substrate processing situation is possible, and if it is determined that an automatic response is not possible, outputs information about the substrate processing and a response plan for the situation, accepts a decision on the response to the situation, and outputs information about the response plan that has been decided upon.

[0006] According to the present disclosure, it is expected that support for processing to deal with the situation of the substrate processing apparatus can be provided.

[0007] FIG. 1 is a schematic diagram for explaining an overview of an information processing system according to the present embodiment. FIG. 2 is a block diagram showing an example of a configuration of an information processing device according to the present embodiment. FIG. 3 is a schematic diagram showing an example of a configuration of a learning model according to the present embodiment. FIG. 4 is a flowchart showing an example of the procedure of automatic response processing performed by the information processing device according to the present embodiment. FIG. 5 is a schematic diagram showing an example of a message exchange with a user performed in the response decision processing. FIG. 6 is a flowchart showing an example of the procedure of the response decision processing performed by the information processing device according to the present embodiment. FIG. 7 is a schematic diagram showing an example of a message exchange in the response decision processing. FIG. 8 is a schematic diagram showing an example of a message exchange in the response decision processing. FIG. 9 is a schematic diagram for explaining the flow of data in the response decision processing. FIG. 10 is a schematic diagram for explaining the flow of data in the response decision processing.

[0008] Specific examples of information processing systems according to embodiments of the present disclosure will be described below with reference to the drawings. Note that the present disclosure is not limited to these examples, but is defined by the claims, and is intended to include all modifications within the meaning and scope of the claims.

[0009] [First Embodiment] <System Overview> FIG. 1 is a schematic diagram illustrating an overview of an information processing system according to this embodiment. The information processing system according to this embodiment is configured to include an information processing apparatus 1 and a substrate processing apparatus 3. The substrate processing apparatus 3 is an apparatus such as a process chamber that performs substrate processing, such as etching, on substrates such as semiconductor wafers. The information processing apparatus 1 outputs a control command to the substrate processing apparatus 3 to control the operation of the substrate processing apparatus 3 based on, for example, settings related to the substrate processing input by a user. The substrate processing apparatus 3 performs substrate processing in accordance with the control command provided by the information processing apparatus 1. The substrate processing apparatus 3 includes one or more sensors (not shown) that measure, for example, the temperature or pressure in the process chamber, and outputs measurement values ​​measured by the sensors during the substrate processing to the information processing apparatus 1. The information processing apparatus 1 determines the status or state of the substrate processing apparatus 3 or the substrate processing based on the sensor measurements provided by the substrate processing apparatus 3, and performs substrate processing by providing a control command to the substrate processing apparatus 3 according to the determined status or state.

[0010] In the information processing system according to this embodiment, the information processing apparatus 1 can automatically respond to the status of the substrate processing apparatus 3 or the substrate processing, which is determined based on sensor measurement values ​​obtained from the substrate processing apparatus 3, based on a predetermined procedure (rule, program, etc.). Alternatively, the information processing apparatus 1 may automatically respond to the status of the substrate processing using a learning model that has been previously subjected to machine learning, so-called AI (Artificial Intelligence). The automatic response process performed by the information processing apparatus 1 is a technology that may be called APC (Advanced Process Control), and this process may be realized by any method.

[0011] For example, when substrate processing is being performed normally by the substrate processing apparatus 3, the information processing apparatus 1 can continue to perform this substrate processing. Also, when an abnormality occurs in the substrate processing by the substrate processing apparatus 3, the information processing apparatus 1 can automatically stop the substrate processing if it is in progress, or can automatically request or schedule maintenance or repair of the substrate processing apparatus 3 based on the cause of the abnormality.

[0012] However, there may be cases where, for example, the sensor measurement values ​​obtained from the substrate processing apparatus 3 are outside the normal range but are not severe enough to stop substrate processing or to require maintenance of the substrate processing apparatus 3. In the information processing system according to this embodiment, when such a situation arises where it is difficult to determine an automatic response, the information processing apparatus 1 inquires of the user about what response to take. At this time, the information processing apparatus 1 provides the user with information such as a display of the sensor measurement values ​​obtained from the substrate processing apparatus 3 or information about similar cases in the past. The information processing apparatus 1 according to this embodiment also provides the user with information about possible or recommended responses to the current situation. Based on this information, the user decides on a response, and the information processing apparatus 1 accepts the user's response decision and performs processing related to the decided response.

[0013] When the information processing apparatus 1 according to this embodiment takes a user-determined action, it stores, in a correspondence DB (database) 5, for example, a sensor measurement value obtained from the substrate processing apparatus 3 and information about the user-determined action in association with each other. If a similar situation occurs thereafter (a similar sensor measurement value is obtained), the information processing apparatus 1 reads out the information about the action stored in the correspondence DB 5 and automatically implements the user-determined action for the similar sensor measurement value. The automatic action of the information processing apparatus 1 based on the information in the correspondence DB 5 may be performed, for example, by rule-based processing, or may be performed using a learning model that has undergone machine learning.

[0014] For example, the information processing apparatus 1 is provided with a learning model that has been machine-learned in advance to determine whether an automatic response is possible or to determine an automatic response method based on sensor measurement values ​​obtained from the substrate processing apparatus 3. The information processing apparatus 1 can re-learn this learning model based on the information stored in the correspondence DB 5. By using the re-learned learning model, the information processing apparatus 1 can take an automatic response based on the information in the correspondence DB 5.

[0015] Furthermore, when providing information to a user, the information processing device 1 according to this embodiment exchanges messages in natural language with the user using a trained large language model (LLM). The information processing device 1 can, for example, read and display information requested by the user using a natural language message from the correspondence DB 5. Furthermore, when presenting one or more countermeasures based on sensor measurement values ​​to the user, the information processing device 1 can explain the contents of the countermeasures to the user using natural language messages. The user collects necessary information by exchanging messages in natural language with the information processing device 1, selects one of the countermeasures proposed by the information processing device 1 or selects their own countermeasure, and inputs the selected countermeasure into the information processing device 1. The information processing device 1 controls the operation of the substrate processing device 3 or schedules maintenance according to the countermeasure selected by the user.

[0016] The information processing device 1 according to the present embodiment can exchange messages in natural language using a large-scale model as described above. However, when accepting input from a user regarding a response decision, information input in non-natural language is used instead of a message in natural language. The information processing device 1 accepts the user's response decision by, for example, displaying multiple options regarding response plans and accepting one selection from the options. Alternatively, for example, the information processing device 1 may accept the user's response decision by accepting a numerical input from the user or an input of table data including multiple numerical values. Because information input in natural language may contain ambiguous content, accepting information regarding the user's response decision by inputting information in non-natural language is expected to prevent the information processing device 1 from performing processing in accordance with the user's decision from performing processing different from what the user intended.

[0017] The information processing device 1 may accept not only inputs related to the user's response decisions but also various other inputs in non-natural language from the user. For example, when displaying past sensor measurement values ​​stored in the correspondence DB 5, the information processing device 1 may accept from the user, in non-natural language, conditions for which sensor measurement values ​​to display. In this case, the information processing device 1 accepts various condition inputs from the user in non-natural language, such as identification information of the substrate processing device 3 to be displayed, substrate identification information, sensor identification information, or the date and time of substrate processing, and retrieves and displays information that matches the accepted conditions from the correspondence DB 5. Note that the acceptance of inputs in non-natural language is not limited to the above, and various inputs may be accepted in non-natural language. For inputs that have a significant impact on the system, such as information for which ambiguity due to the use of a large-scale language model is undesirable, it is preferable to accept inputs in non-natural language and have the information processing device 1 process the accepted inputs without using a large-scale language model.

[0018] In this embodiment, the information processing apparatus 1 and the substrate processing apparatus 3 are described as separate apparatuses, but this is not limiting. The information processing apparatus 1 and the substrate processing apparatus 3 may be an integrated apparatus. Furthermore, the information processing apparatus 1 or the substrate processing apparatus 3 may be configured by combining multiple apparatuses. Furthermore, multiple substrate processing apparatuses 3 may be connected to one information processing apparatus 1, and the multiple substrate processing apparatuses 3 may be installed in different companies, factories, etc.

[0019] <Device Configuration> Fig. 2 is a block diagram showing an example configuration of an information processing device 1 according to this embodiment. The information processing device 1 according to this embodiment can be realized by installing a predetermined application program or the like in a general-purpose information processing device such as a personal computer or a server computer. However, the information processing device 1 may also be a dedicated information processing device that controls the substrate processing device 3. The information processing device 1 according to this embodiment is configured to include a processing unit 11, a storage unit 12, a communication unit 13, a display unit 14, an operation unit 15, etc. Note that, although this embodiment will be described assuming that processing is performed by a single information processing device 1, the processing of the information processing device 1 may be distributed among multiple devices.

[0020] The processing unit 11 is configured using an arithmetic processing device such as a CPU (Central Processing Unit), an MPU (Micro-Processing Unit), a GPU (Graphics Processing Unit) or a quantum processor, a ROM (Read Only Memory), a RAM (Random Access Memory), etc. The processing unit 11 reads and executes a program 12a stored in the storage unit 12 to perform various processes such as determining whether an automatic response is possible to the situation of the substrate processing apparatus 3, and accepting a user's decision on a response when an automatic response is not possible.

[0021] The storage unit 12 is configured using a large-capacity storage device such as a hard disk or an SSD (Solid State Drive). The storage unit 12 stores various programs executed by the processing unit 11 and various data required for the processing of the processing unit 11. In this embodiment, the storage unit 12 stores a program 12a executed by the processing unit 11. The storage unit 12 also includes a learning model storage unit 12b that stores information related to a trained learning model used in processing performed by the information processing device 1. The storage unit 12 also includes a correspondence DB 5 that stores sensor measurement values ​​acquired from the substrate processing device 3 and information related to responses taken thereto.

[0022] In this embodiment, the program (computer program, program product) 12a is provided in a form recorded on a recording medium 99 such as a memory card or an optical disc, and the information processing device 1 reads the program 12a from the recording medium 99 and stores it in the storage unit 12. However, the program 12a may also be written to the storage unit 12, for example, during the manufacturing stage of the information processing device 1. Alternatively, the program 12a may be distributed by a remote server device or the like and acquired by the information processing device 1 via communication. For example, the program 12a may be read from the recording medium 99 by a writing device and written to the storage unit 12 of the information processing device 1. The program 12a may be provided in a form distributed via a network or in a form recorded on the recording medium 99.

[0023] The learning model storage unit 12b stores information about learning models that have been previously subjected to machine learning. The information about the learning models may include, for example, information indicating the configuration of the learning models and information such as values ​​of internal parameters determined by machine learning. In the present embodiment, the learning model storage unit 12b stores information about learning models, such as a learning model for determining whether an automatic response to the situation of the substrate processing apparatus 3 is possible, and a large-scale language model for exchanging messages with a user in natural language. In the present embodiment, the information processing apparatus 1 stores information about these learning models, and performs processing such as predictions using these learning models, but this is not limited to this. Information about these learning models may be stored in a device other than the information processing apparatus 1, and this device may perform processing such as predictions using the learning models, and the information processing apparatus 1 may obtain the processing results from this device. Furthermore, the machine learning processing of these learning models may be performed by the information processing apparatus 1 or by a device other than the information processing apparatus 1.

[0024] The correspondence DB 5 is a database that stores various information regarding responses made by the information processing apparatus 1 to the substrate processing apparatus 3. The information processing apparatus 1 stores, in the correspondence DB 5, for example, sensor measurement values ​​acquired from the substrate processing apparatus 3 in association with information such as the date and time when the substrate processing was performed and identification information of the substrate that underwent the substrate processing. The information processing apparatus 1 also stores information indicating the status of the substrate processing apparatus 3 or the substrate processing determined, for example, by a learning model, in association with the information such as the sensor measurement values. The information indicating the status may be, for example, "normal," "abnormal," or "difficult to determine," or may be, for example, "maintenance required," "maintenance not required," or "difficult to determine." Note that the information indicating the status may be other than those described above and may be determined appropriately based on, for example, the type of status of the substrate processing apparatus 3 for which the information processing system is to automatically respond.

[0025] The information processing device 1 also stores information indicating what kind of response was taken in response to these situations, etc., in the response DB 5. In this embodiment, when a situation such as "difficult to judge" is determined by the learning model based on the sensor measurement values ​​acquired from the substrate processing device 3, the information processing device 1 accepts a response decision made by the user and stores information indicating the content of the determined response in the response DB 5. The information indicating the content of the response may be, for example, forcibly stopping the substrate processing, continuing the substrate processing, performing maintenance, or not performing maintenance. Note that the information indicating the content of the response may be other than the above and may be any response that the user can take with respect to the substrate processing device 3. The information processing device 1 may also store information such as a history of messages exchanged in natural language by the user when deciding on a response in the response DB 5 together with the information indicating the content of the response.

[0026] In this embodiment, the information processing device 1 is configured to include the correspondence DB 5, but this is not limiting. The correspondence DB 5 may be included in a device different from the information processing device 1. In this case, the information processing device 1 communicates with the device that includes the correspondence DB 5, and reads information from the correspondence DB 5 and writes information to the correspondence DB 5.

[0027] The communication unit 13 transmits and receives data to and from the substrate processing apparatus 3 via a wired or wireless network. In the present embodiment, the information processing apparatus 1 acquires sensor measurement value data from the substrate processing apparatus 3 and issues control commands to the substrate processing apparatus 3 by communicating with the substrate processing apparatus 3 via the communication unit 13. The communication unit 13 receives data transmitted from the substrate processing apparatus 3 and provides the received data to the processing unit 11. Note that in the present embodiment, data is exchanged between the substrate processing apparatus 3 and the information processing apparatus 1 via communication, but this is not limited thereto, and data may be exchanged via a recording medium such as a memory card. The communication unit 13 may also communicate with an apparatus other than the substrate processing apparatus 3.

[0028] The display unit 14 is configured using a liquid crystal display or the like, and displays various images, characters, etc. based on processing by the processing unit 11. The operation unit 15 accepts user operations and notifies the processing unit 11 of the accepted operations. For example, the operation unit 15 accepts user operations via input devices such as mechanical buttons or a touch panel provided on the surface of the display unit 14. Furthermore, for example, the operation unit 15 may be input devices such as a mouse and a keyboard, and these input devices may be configured to be detachable from the information processing device 1.

[0029] The storage unit 12 may be an external storage device connected to the information processing device 1. The information processing device 1 may be a multi-computer including multiple computers, or may be a virtual machine virtually constructed by software. The information processing device 1 is not limited to the above configuration, and may not include, for example, the display unit 14 and the operation unit 15.

[0030] In the information processing apparatus 1 according to this embodiment, the processing unit 11 reads and executes the program 12a stored in the storage unit 12, thereby realizing a data acquisition unit 11a, an automatic response determination unit 11b, a response plan generation unit 11c, a decision acceptance unit 11d, a control processing unit 11e, a display processing unit 11f, etc. as software functional units in the processing unit 11. Note that in the figure, functional units related to processing for taking a response according to the status of the substrate processing apparatus 3 are shown as functional units of the processing unit 11, and functional units related to other processing are not shown.

[0031] The data acquisition unit 11a communicates with the substrate processing apparatus 3 via the communication unit 13 to acquire data on sensor measurement values ​​measured by the substrate processing apparatus 3 during substrate processing, etc. The data acquisition unit 11a acquires data on sensor measurement values ​​at a predetermined cycle, such as once per second, while the substrate processing apparatus 3 is performing substrate processing, and stores the acquired data in the correspondence DB 5. The data acquisition unit 11a may repeatedly acquire data on the sensor measurement values ​​of the substrate processing apparatus 3 and store the sensor measurement values ​​in the correspondence DB 5 as time-series data. Alternatively, the data acquisition unit 11a may acquire time-series data on the sensor measurement values ​​all at once, for example, after the substrate processing is completed, rather than repeatedly acquiring data at a predetermined cycle. In this case, the data acquisition unit 11a may acquire data via a recording medium, etc., rather than via communication.

[0032] In this embodiment, the information processing apparatus 1 acquires and processes time-series sensor measurement data measured by a sensor in the substrate processing apparatus 3. However, the data acquired by the information processing apparatus 1 is not limited to time-series sensor measurement data. For example, the data may be measurement data obtained by measuring values ​​relating to the quality of substrates processed by the substrate processing apparatus 3 using a separate measuring device. Furthermore, the information processing apparatus 1 may acquire various data, such as setting values ​​related to the substrate processing performed by the substrate processing apparatus 3 (so-called recipe setting values), identification information for the substrate processing apparatus 3, or identification information for the substrates subjected to the substrate processing. The data acquired by the information processing apparatus 1 may be any data obtained directly or indirectly related to the substrate processing apparatus 3 or the substrate processing.

[0033] The automatic response determination unit 11b performs processing to determine whether or not an automatic response can be made in accordance with the status of the substrate processing apparatus 3 or the substrate processing, based on data such as sensor measurement values ​​acquired by the data acquisition unit 11a. In the present embodiment, the automatic response determination unit 11b makes the determination using a learning model stored in the learning model storage unit 12b. The learning model used by the automatic response determination unit 11b may be a learning model that has been machine-learned in advance so as to accept, for example, sensor measurement values ​​or time-series sensor measurement values ​​as input and output a determination result as to whether or not an automatic response is possible in accordance with the status of the substrate processing apparatus 3 or the substrate processing, etc. Each time the data acquisition unit 11a acquires sensor measurement value data or time-series data for a predetermined period of time, the automatic response determination unit 11b inputs the acquired data into the learning model and acquires a determination result output by the learning model in response to the data.

[0034] The response plan generator 11c performs a process of generating information on one or more response plans to be proposed to the user when the automatic response determination unit 11b determines that an automatic response is not possible. For example, the response plan generator 11c searches for one or more similar data from the data stored in the response DB 5 based on the sensor measurement data acquired by the data acquisition unit 11a, acquires information on responses implemented for the similar data, and sets the responses as response plans to be proposed to the user. When multiple similar data exist and multiple responses have been implemented, the response plan generator 11c can select one response from the multiple responses according to appropriate conditions, such as adopting the response in the similar data with the highest similarity, adopting the response in the most recent similar data, or adopting the response implemented most frequently, and set this response as the response plan to be proposed to the user. Alternatively, the response plan generator 11c may adopt multiple response plans and propose multiple response plans to the user by ranking them based on, for example, similarity. The similarity between two pieces of data can be obtained, for example, by treating data including multiple measurement values ​​as vectors and calculating the cosine similarity between the two vectors.

[0035] When the automatic response determination unit 11b determines that an automatic response cannot be made to the situation of the substrate processing apparatus 3 or the substrate processing, the decision receiving unit 11d performs a process of receiving information about the response decided to be made to this situation from the user. For example, the decision receiving unit 11d displays the sensor measurement data acquired by the data acquisition unit 11a or a graph of the data, and also displays one or more response plans generated by the response plan generation unit 11c to provide information to the user. At this time, the decision receiving unit 11d outputs the information to be provided to the user as a message in natural language using the large-scale language model stored in the learning model storage unit 12b.

[0036] Furthermore, the decision receiving unit 11d displays and provides information such as sensor measurement value data and countermeasures to the user, and then exchanges messages in natural language with the user using a large-scale language model, i.e., engages in a chat. At this time, the user can request various information from the information processing device 1 via chat, such as display of past similar data as a basis for deciding on a countermeasure. The decision receiving unit 11d interprets the natural language message provided by the user using the large-scale language model, reads requested data from, for example, the correspondence DB 5 in response to the user's request, and displays the data in the form of a graph or table, etc., as needed, along with a natural language message explaining the data. Note that the decision receiving unit 11d may obtain not only information stored in the correspondence DB 5 but also information obtained from other devices through communication via a network, such as information widely available on the Internet, and display the obtained information as a message by aggregating or summarizing it using the large-scale language model.

[0037] In this embodiment, message exchange between the information processing device 1 and the user is performed, for example, by inputting characters using an operation unit 15 such as a keyboard and outputting characters to a display unit 14 such as a display, but this is not limited thereto. Message exchange between the information processing device 1 and the user may also be performed, for example, by voice input / output. In this case, the information processing device 1 performs a process of converting voice data to character data and / or a process of converting character data to voice data. The information processing device 1 acquires a natural language message spoken by the user using a microphone or the like, converts the acquired voice data of the message into character data, and inputs the character data into a large-scale language model. The information processing device 1 converts the message of character data output by the large-scale language model into voice data and outputs the voice from a speaker or the like. The information processing device 1 may also be equipped with a large-scale language model capable of directly accepting input and output of voice data. In this case, the information processing device 1 can handle voice data without converting voice data to character data.

[0038] Based on the information provided by the information processing device 1, the user decides how to respond to the status of the substrate processing device 3 or the substrate processing, and issues a natural language message, such as "Please stop substrate processing" or "No maintenance is required," to the information processing device 1. The decision receiving unit 11d interprets these natural language messages using a large-scale language model to determine that the user has decided on a response, and displays a natural language message confirming the content of the decided response. For example, in response to a message from the user saying "Please stop substrate processing," the decision receiving unit 11d can display a confirmation message such as "Substrate processing will be stopped. Are you sure?". The decision receiving unit 11d also displays, together with the confirmation message, options that constitute a final decision on how to respond to the status of the substrate processing device 3 or the substrate processing, such as two options, for example, "YES" or "NO," and accepts a selection operation for one of these options to accept the user's decision on the response.

[0039] Furthermore, when the user approves the response plan generated by the response plan generating unit 11c, for example, when the user inputs a message such as "I would like to accept that response plan" in response to the displayed response plan, the decision accepting unit 11d may, for example, redisplay the response plan and display a message such as "Is this response plan OK?", and may display two options, "YES" and "NO," and accept an operation to select an option from the user. When multiple response plans are displayed, the decision accepting unit 11d may, for example, display a message such as "Which of response plans 1 to 3 would you like to adopt?", and accept an operation from the user to select one of the three response plans using a pull-down menu or the like.

[0040] In the above example, the user's final decision on the response is received using options, but this is not limited to this. For example, when the user decides to take an action such as changing the gas injection amount or temperature setting for the substrate processing apparatus 3, the information processing apparatus 1 receives input of numerical values ​​such as the gas injection amount or temperature, and receives the final decision on the response from the user. Furthermore, for example, when there are multiple numerical values ​​to be set, the information processing apparatus 1 may receive input of table data that associates setting items with setting values, and receive the final decision on the response from the user. The method of receiving the decision on the response by the decision receiving unit 11d is not limited to the above-mentioned options, numerical input, or table data input, and any other method may be used. It is preferable that the decision receiving unit 11d receive the final decision on the response using non-natural language.

[0041] The control processing unit 11e controls the operation of the substrate processing apparatus 3 by outputting control commands. The control processing unit 11e operates the substrate processing apparatus 3 according to, for example, conditions (recipe) preset by a user, and performs substrate processing, such as etching, on substrates such as wafers. When the automatic response determination unit 11b determines that automatic response is possible based on sensor measurement data obtained from the substrate processing apparatus 3, the control processing unit 11e controls the operation of the substrate processing apparatus 3 according to a predetermined procedure, algorithm, program, or the like. For this purpose, the information processing apparatus 1 stores in advance in the storage unit 12, for example, processing procedures to be performed by the control processing unit 11e as an automatic response in response to sensor measurement values ​​or the substrate processing status determined therefrom. Based on this information stored in the storage unit 12, for example, when it is determined from the sensor measurement data that substrate processing is being performed normally, the control processing unit 11e can continue the substrate processing. Furthermore, when it is determined from the sensor measurement data that an obvious abnormality (an abnormality that does not require user inquiry) has occurred, for example, the control processing unit 11e can cause the substrate processing apparatus 3 to perform a predetermined emergency action, such as forcibly stopping the operation of the substrate processing apparatus 3. In addition, information such as automatic response procedures stored in the memory unit 12 can be modified or added based on the content of the response decided upon, for example, when it is determined that automatic response is not possible and the user decides on a response.

[0042] If the automatic response determination unit 11b determines that an automatic response is not possible based on the sensor measurement value data, the control processing unit 11e can perform control such as stopping the operation of the substrate processing apparatus 3 or changing the temperature setting of the substrate processing apparatus 3, based on the user's response decision accepted by the decision accepting unit 11d. Furthermore, the control processing unit 11e may not only directly control the operation of the substrate processing apparatus 3, but also communicate with a device that manages a maintenance schedule for the substrate processing apparatus 3 and perform various processes related to the substrate processing apparatus 3, such as requesting maintenance or determining the date and time for maintenance.

[0043] The display processing unit 11f performs processing to display various information on the display unit 14, such as displaying the sensor measurement data acquired by the data acquisition unit 11a, displaying a chat screen for exchanging messages with users by the decision receiving unit 11d, and displaying the operating status of the substrate processing apparatus 3.

[0044] Note that message exchange between the information processing device 1 and the user may be performed by voice chat rather than by character (text) chat. In this case, the information processing device 1 acquires the user's voice message via a device such as a microphone, and outputs the voice message based on the output of a large-scale language model via a device such as a speaker. The information processing device 1 may also perform mutual conversion between voice data and text data. Furthermore, the information processing device 1 may, for example, accept a message from the user via voice and display the message to the user in text, or may, for example, accept a message from the user via text and output the message to the user via voice.

[0045] <Automatic Response Processing> The following description will be given taking as an example a case where the information processing apparatus 1 performs an automatic response related to maintenance of the substrate processing apparatus 3. However, the automatic response performed by the information processing apparatus 1 is not limited to a process related to maintenance, and may be any process.

[0046] In this example, each time the substrate processing apparatus 3 performs substrate processing on one substrate, the information processing apparatus 1 determines whether maintenance of the substrate processing apparatus 3 is necessary based on sensor measurement values ​​and takes automatic action. If the information processing apparatus 1 determines that maintenance of the substrate processing apparatus 3 is necessary, it stops substrate processing on the substrate processing apparatus 3 and automatically schedules maintenance of the substrate processing apparatus 3 by, for example, transmitting information to a system that manages a maintenance schedule. If the information processing apparatus 1 determines that maintenance is not necessary, it automatically performs substrate processing on the next substrate.

[0047] The information processing device 1 uses a learning model that has been machine-learned in advance to determine whether maintenance of the substrate processing device 3 is required based on sensor measurement values ​​acquired from the substrate processing device 3. FIG. 3 is a schematic diagram showing an example of the configuration of a learning model according to this embodiment. The learning model 7 used by the information processing device 1 according to this embodiment is a learning model that has been machine-learned in advance to accept, for example, sensor measurement values ​​of the substrate processing device 3 as input and output a classification result into either "maintenance required" or "maintenance not required." The sensor measurement value data input to the learning model 7 may be a vector or matrix containing multiple measurement values, or may be time-series data. The two output values ​​of the learning model 7 are, for example, numerical values ​​ranging from 0 to 1, and the sum of the two output values ​​is 1. The information processing device 1 can compare the two output values ​​of the learning model 7 and use the larger value as the determination result of whether maintenance is required.

[0048] Furthermore, the information processing device 1 determines that it is difficult to determine whether maintenance is required when the difference between the two output values ​​of the learning model 7 is equal to or smaller than a threshold value (e.g., 0.4), or when the larger value does not exceed a threshold value (e.g., 0.7), etc. In other words, the information processing device 1 is unable to automatically respond to the situation of the substrate processing device 3 and can determine that it is necessary to request the user to decide on a response.

[0049] The learning model 7 can be generated in advance by performing so-called supervised machine learning processing using learning data (teacher data) that associates, for example, sensor measurement values ​​of the substrate processing apparatus 3 with flag information indicating whether or not maintenance of the substrate processing apparatus 3 was required for these sensor measurement values. Note that the machine learning processing for generating the learning model 7 may be performed by the information processing apparatus 1 or by a device different from the information processing apparatus 1.

[0050] The learning model 7 may have various configurations, such as a deep neural network (DNN), a convolutional neural network (CNN), a support vector machine (SVM), a decision tree, or a random forest. The configuration of these learning models and the machine learning process for generating the learning models are existing technologies, and therefore detailed explanations thereof will be omitted.

[0051] The illustrated configuration of the learning model 7 is merely an example and is not limiting. The learning model 7 may be configured to output, for example, a single value indicating the degree to which maintenance is required. In this case, if the output value of the learning model 7 is in the range of 0 to 1, the information processing device 1 may determine, for example, that maintenance is not required for an output value of 0.0 to 0.3, that it is difficult to determine for an output value of 0.3 to 0.7, and that maintenance is required for an output value of 0.7 to 1.0. The learning model 7 may also be configured to output three values: maintenance required, maintenance not required, and difficult to determine. In this case, the information processing device 1 may determine the largest output value of the three values ​​output by the learning model 7 as the determination result. Furthermore, the data input to the learning model 7 is not limited to sensor measurement values, but may also include input information to the substrate processing device 3, such as setting information (recipe) related to the substrate processing performed by the substrate processing device 3.

[0052] Furthermore, the learning model 7 may be configured to output, in addition to two output values ​​related to the need for maintenance, a value evaluating the reliability of these output values. The evaluated reliability value may be calculated by the information processing device 1 rather than output by the learning model 7. The evaluated reliability value may be calculated, for example, based on the inverse of the variance of the output values ​​of the learning model 7, or may be calculated by an approximate calculation based on a technique such as Laplace approximation, which is a Bayesian modeling-related technique. The information processing device 1 determines whether the evaluated reliability value exceeds a threshold, and if it does not exceed the threshold, determines that it is difficult to determine whether maintenance is necessary. In this way, various methods may be adopted for the configuration of the learning model 7 and the method for determining the difficulty of determination.

[0053] In the present embodiment, the information processing device 1 is configured to determine whether maintenance is necessary and whether automatic response is possible using the learning model 7, but the present invention is not limited to this. For example, the information processing device 1 may be configured to determine whether maintenance is necessary and whether automatic response is possible based on a comparison between a sensor measurement value acquired from the substrate processing device 3 and a predetermined threshold value.

[0054] 4 is a flowchart showing an example of the procedure of the automatic correspondence processing performed by the information processing apparatus 1 according to the present embodiment. The control processing unit 11e of the processing unit 11 of the information processing apparatus 1 according to the present embodiment performs substrate processing, such as etching, on a substrate such as a wafer, in accordance with a substrate processing setting (recipe) previously determined by, for example, a user (step S1). The data acquisition unit 11a of the processing unit 11 periodically acquires sensor measurement values ​​by repeatedly communicating with the substrate processing apparatus 3, or acquires all sensor measurement values ​​measured during the substrate processing after the substrate processing for one substrate is completed (step S2). The data acquisition unit 11a stores the sensor measurement values ​​acquired from the substrate processing apparatus 3 in the correspondence DB 5 together with information such as the date and time the substrate processing was performed and the identification information of the substrate that underwent the substrate processing (step S3).

[0055] The automatic response determination unit 11b of the processing unit 11 inputs the sensor measurement values ​​acquired in step S2 into the learning model 7 stored in the learning model storage unit 12b, and acquires the output of the learning model 7 indicating whether maintenance is required, thereby determining whether maintenance is required by the learning model 7 (step S4). The automatic response determination unit 11b determines whether automatic response is possible for maintenance of the substrate processing apparatus 3 based on the output value of the learning model 7 (step S5). At this time, the automatic response determination unit 11b can determine whether automatic response is possible based on whether the larger of the two values ​​output by the learning model 7 exceeds a predetermined threshold.

[0056] If it is determined that an automatic response is not possible (S5: NO), the decision receiving unit 11d of the processing unit 11 performs a response determination process to receive a decision from the user as to what response to take in response to the situation of the substrate processing apparatus 3 (step S6), and the process proceeds to step S7. Details of the response determination process performed in step S6 will be described later.

[0057] If automatic response is possible (S5: YES), or if the user's response decision has been accepted in step S6, the decision accepting unit 11d determines whether maintenance of the substrate processing apparatus 3 is necessary based on the output value of the learning model 7 or the user's decision (step S7). If it is determined that maintenance is not necessary (S7: NO), the control processing unit 11e returns to step S1 and performs substrate processing on the next substrate. If it is determined that maintenance is necessary (S7: YES), the control processing unit 11e stops the substrate processing that was being performed (step S8). The processing unit 11 schedules maintenance of the substrate processing apparatus 3, for example, by sending information indicating the status of the substrate processing apparatus 3 to a system that manages information related to maintenance of the substrate processing apparatus 3 and requesting maintenance (step S9).

[0058] After maintenance of the substrate processing apparatus 3 is performed, the processing unit 11 acquires the maintenance results of the substrate processing apparatus 3, for example, from a system that manages information about maintenance, or by input from the user (step S10). The maintenance results may include various information, such as whether maintenance was necessary for the substrate processing apparatus 3 or what kind of maintenance was performed on the substrate processing apparatus 3. The processing unit 11 stores the information about the maintenance results acquired in step S10 in the correspondence DB 5 (step S11), and then ends the process. The information processing apparatus 1 is expected to use the information stored in the correspondence DB 5 to, for example, re-learn the learning model 7, thereby improving the accuracy of the learning model 7 in determining whether maintenance is necessary. The information processing apparatus 1 is also expected to present the information stored in the correspondence DB 5 to the user, thereby assisting the user in deciding what action to take.

[0059] <Response Determination Process> Fig. 5 is a schematic diagram showing an example of a message exchange with a user performed in the response determination process. When the information processing apparatus 1 according to this embodiment determines that an automatic response cannot be performed based on the sensor measurement values ​​acquired from the substrate processing apparatus 3, the information processing apparatus 1 displays a chat screen on the display unit 14 for exchanging messages with the user. For example, on this chat screen, the information processing apparatus 1 displays a title string "Maintenance Determination Chat" at the top of the screen, and below that, messages from the information processing apparatus 1 to the user and messages from the user to the information processing apparatus 1 are displayed in chronological order from top to bottom. The information processing apparatus 1 also displays the user's messages near the left side of the screen and the information processing apparatus 1's messages near the right side of the screen.

[0060] When the information processing apparatus 1 determines that it is difficult to determine whether maintenance is required, it displays a message on the chat screen, such as "Suspicious behavior was observed at OES 777.0 nm. Please determine whether maintenance is required. See the URL for details," based on the sensor measurement values ​​acquired from the substrate processing apparatus 3, and also displays a graph such as a scatter plot or line graph generated based on the sensor measurement values ​​that form the basis of this message. In practice, the "URL" in the message is embedded in the form of a link to a URL (Uniform Resource Locator) such as a website that provides detailed information.

[0061] Furthermore, the information processing device 1 searches for information containing sensor measurement values ​​similar to the currently acquired sensor measurement value from the information stored in the correspondence DB 5. If similar sensor measurement values ​​are stored in the correspondence DB 5, the information processing device 1 acquires information about what measures have been taken in response to these similar sensor measurement values ​​from the correspondence DB 5. Based on the information acquired from the correspondence DB 5, the information processing device 1 generates and displays a message indicating a proposed measure, such as, for example, "Similar behavior has occurred in a similar process in the past, and since there was no impact on the finished product, maintenance is deemed unnecessary."

[0062] After the information processing device 1 displays these messages, the user can input a message in natural language using, for example, a keyboard or a touch panel software keyboard, to request the information processing device 1 to display various information and have the information processing device 1 implement the response plan that the user has decided on. In this example, the user inputs the message "Show me the data for that process" to request the information processing device 1 to display the data. In response to this message from the user, the information processing device 1 displays a response message saying "Please check the URL," which includes the URL of a website or the like that displays the data requested by the user. At this time, the information processing device 1 may also display the data requested by the user on the chat screen.

[0063] The user can exchange messages with the information processing device 1 one or more times to collect information that will be used to determine how to respond to the situation of the substrate processing device 3. Having decided on a response based on the information provided by the information processing device 1, the user inputs a message such as, for example, "No maintenance required, OK." Upon receiving such a natural language message regarding the response decision, the information processing device 1 displays a message confirming the user's decision, such as, "We will continue processing without performing maintenance. Please approve.", along with a button labeled "Approve" and a button labeled "Cancel." The user can select either the "Approve" or "Cancel" option for the confirmation message from the information processing device 1 by clicking or touching either the "Approve" or "Cancel" button displayed as a two-choice option on the chat screen. In this example, "Approve" is selected.

[0064] When the information processing device 1 according to the present embodiment receives a user's decision on what action to take in response to the status of the substrate processing device 3 or the substrate processing, the information processing device 1 accepts the final decision in non-natural language, thereby accurately accepting a decision that eliminates ambiguity that may be contained in a natural language message. In this example, the information processing device 1 presents a binary option of approval or cancellation to accept the user's final decision on the action, but the acceptance of action decisions based on non-natural language is not limited to this. For example, when there are multiple possible action plans, the information processing device 1 may number these multiple action plans and present them to the user, and accept the user's decision on the action by inputting or selecting a number, etc. The information processing device 1 may present three or more options and accept the selection of one of them.

[0065] For example, the information processing apparatus 1 may accept input of numerical values ​​related to settings (temperature settings or flow rate settings) related to substrate processing in the substrate processing apparatus 3, and thereby accept a decision on an action to change the set value to the input value. Furthermore, when there are multiple setting values ​​to be changed, the information processing apparatus 1 may display a table in which names of multiple settings and their corresponding setting values ​​are associated, and accept input of multiple setting values ​​into the table, thereby accepting a decision on an action to change the multiple setting values. In this case, for example, a user may create table data in which names of settings and their corresponding setting values ​​are associated as a separate file, and the information processing apparatus 1 may read this file to accept a decision on an action to change the multiple setting values. The information processing apparatus 1 may accept a decision on an action based on non-natural language in various ways other than those described above.

[0066] Fig. 6 is a flowchart showing an example of the procedure of the response decision process performed by the information processing apparatus 1 according to this embodiment. The process of this flowchart is the process performed in step S6 of the flowchart shown in Fig. 4. The response plan generator 11c of the processor 11 of the information processing apparatus 1 according to this embodiment searches for similar data from the data stored in the correspondence DB 5 based on the sensor measurement values ​​acquired from the substrate processing apparatus 3 (step S21), and acquires the similar data and data related thereto from the correspondence DB 5.

[0067] Next, the response plan generator 11c generates a response plan for the substrate processing apparatus 3 or the status of the substrate processing, etc., based on the data acquired from the response DB 5 (step S22). At this time, the response plan generator 11c inputs the current sensor measurement value and the data acquired from the response DB 5 into the large-scale language model to instruct the generation of a response plan, and can generate a response plan by acquiring natural language information output by the large-scale language model. The response plan generator 11c displays the sensor measurement value data acquired from the substrate processing apparatus 3 and the response plan in natural language generated by the response plan generator 11c on the display unit 14 (step S23).

[0068] The decision accepting unit 11d of the processing unit 11 accepts a message in natural language input by a user on the chat screen (step S24). The decision accepting unit 11d inputs the message accepted in step S24 into a large-scale language model and interprets it to determine whether the message is an instruction to respond to the status of the substrate processing apparatus 3 or the substrate processing (step S25). If the message is not an instruction to respond (S25: NO), the decision accepting unit 11d generates a response to the user's message accepted in step S24 using the large-scale language model (step S26). The decision accepting unit 11d displays the natural language response message generated in step S26 on the chat screen (step S27), and returns to step S24.

[0069] If the message from the user is an instruction to take a response (S25: YES), the decision accepting unit 11d displays a confirmation message regarding the response selected by the user on the display unit 14 (step S28). The decision accepting unit 11d also accepts the user's selection by displaying, for example, a binary option (step S29). The decision accepting unit 11d determines whether the selection accepted in step S29 is a response decision (step S30). If the selection is not a response decision (S30: NO), the decision accepting unit 11d returns the process to step S24. If the selection is a response decision (S30: YES), the decision accepting unit 11d considers the selection accepted in step S29 to be the user's response decision, stores the content of the selected response and information such as messages exchanged with the user during this process in the response DB 5 (step S31), and terminates the process.

[0070] The information processing apparatus 1 stores in advance in the storage unit 12 information such as conditions or procedures for automatically responding to the status of the substrate processing apparatus 3 or the substrate processing. The storage unit 12 stores conditions that specify thresholds or ranges for sensor measurement values ​​acquired from the substrate processing apparatus 3, for example, and processes to be executed when these conditions are satisfied, in association with each other. The information processing apparatus 1 according to this embodiment can modify or add information such as the conditions or procedures for the automatic response stored in the storage unit 12 based on the information stored in the correspondence DB 5 in step S31.

[0071] <Use of Large-Scale Language Model> The information processing device 1 exchanges messages based on natural language with the user in the above-described response determination process. The information processing device 1 according to the present embodiment exchanges messages based on natural language using a large-scale language model that has been machine-learned in advance. For example, a learning model such as a Transformer with an attention mechanism in a large-scale neural network, a Bidirectional Encoder Representations from Transformers (BERT), or a Generative Pre-trained Transformer (GPT) can be adopted as the large-scale language model. The large-scale language model used by the information processing device 1 may be a widely available, general-purpose large-scale language model that has been trained with information related to substrate processing through fine tuning.

[0072] Furthermore, the information processing apparatus 1 according to this embodiment stores and accumulates in the correspondence DB 5 information such as sensor measurement values ​​acquired from the substrate processing apparatus 3, information regarding responses implemented in response to these sensor measurement values, and a history of message exchanges between the user and the information processing apparatus 1. The information processing apparatus 1 provides the large-scale language model with the sensor measurement values ​​newly acquired from the substrate processing apparatus 3 and the information stored in the correspondence DB 5, thereby causing the large-scale language model to generate response plans suitable for the status of the substrate processing apparatus 3 and presenting one or more response plans to the user. Furthermore, the information processing apparatus 1 uses the large-scale language model to perform processes such as generating a response message to an inquiry message provided by the user and obtaining information necessary for generating the response message from the correspondence DB 5.

[0073] 7 to 9 are schematic diagrams illustrating an example of message exchange during the response determination process. FIG. 7 shows an example of a message exchange at a certain point in time (January 3, 20XX). In this example, the information processing device 1, which determined that automatic response was not possible, displays the message "OES777.0 nm emission intensity has exceeded the upper limit of the Control Limit. Please decide on a response. For details, see the URL." to request the user to decide how to respond to the situation of the substrate processing device 3. The information processing device 1 can generate these messages by, for example, applying the current sensor measurement value information to pre-stored template data. The information processing device 1 also displays this message along with a graph created based on the current sensor measurement value.

[0074] Next, the information processing device 1 displays a message as a countermeasure for the situation of the substrate processing device 3, saying, "The cause appears to be an abnormality in plasma ignition, so we suggest adjusting the ignition step length." The information processing device 1 generates this countermeasure by providing the large-scale language model with the current sensor measurement value data, the similar data stored in the correspondence DB 5, and an instruction (so-called prompt) for creating a countermeasure based on both data. The prompt provided to the large-scale language model at this time can be, for example, "Please create a countermeasure for the situation of the current sensor measurement value by referring to the similar data." Note that the similar data acquired from the correspondence DB 5 includes, in addition to the sensor measurement value, various information such as the content of the countermeasure implemented in response to the sensor measurement value and the content of the message exchange when deciding on the countermeasure.

[0075] In response to the message from the information processing device 1 proposing a countermeasure, the user inputs an inquiry message asking, "Does this affect the finished product?" In response to this message, the information processing device 1 displays a response message saying, "According to past analyses, it is highly likely that it will have an impact. For details, please refer to the past reports at the URL," along with a graph of past similar sensor measurement values, etc. At this time, the information processing device 1 provides the user's inquiry message to the large-scale language model, which generates an answer message based on the sensor measurement values ​​and similar data that have already been input.

[0076] In this example, the user decides on an action based on the response from the information processing device 1 and inputs a message indicating the action, such as "Please modify the Ignition step." The information processing device 1 inputs this message into the large-scale language model, which notifies the information processing device 1 that this message determines the action and provides information such as the specific action. Upon receiving the message indicating the action, the information processing device 1 displays a message saying, "The recipe will be modified as follows. Please approve it." along with a table comparing the values ​​of the Ignition step before and after the modification. The information processing device 1 also displays an "Approve" button and a "Cancel" button to accept the user's decision on the action to modify the recipe based on the options. When the information processing device 1 receives a message from the user to determine the action, it displays the message based on a pre-stored template and displays user interface components such as a table and buttons that are predetermined according to the action determined by the user. In this example, the user selects the "Approve" button, and in response, the information processing device 1 displays a message saying, "The modification has been reflected."

[0077] The information processing device 1 also creates information including the message exchange with the user and the content of the response decided by the user, and stores this information in the correspondence DB 5 as a report. In this example, before storing the report, the information processing device 1 displays a message saying, "A report has been created. Please check the URL and approve it if there are no problems." to request the user to confirm the created report. The information processing device 1 displays an "Approve" button and a "Cancel" button to accept the user's decision on whether to approve the report. If approval is obtained from the user, the information processing device 1 stores the report in the correspondence DB 5 and displays a message such as, "The report has been saved. Problem response is complete." This report is stored in association with the sensor measurement value data, and is also retrieved when the information processing device 1 retrieves the above-mentioned similar data from the correspondence DB 5.

[0078] FIG. 8 shows an example of a message exchange that occurred a few days after the message exchange shown in FIG. 7 (January 9, 20XX). In this example, as in the previous case, the information processing device 1 displays the message "OES777.0 nm emission intensity is...see...," a graph of the sensor measurement values ​​acquired this time from the substrate processing device 3, and a suggested action, "We suggest that the cause is... plasma." However, this time, because the previous report is stored in the correspondence DB 5, the information processing device 1 provides the information from this report to the large-scale language model, thereby displaying an additional message, "According to past analysis, this type of fluctuation is likely to affect the performance. For details, see the past report at the URL." At this time, the prompt provided by the information processing device 1 to the large-scale language model may be the same, but the additional message is generated by providing different similar data along with the prompt.

[0079] In this example, in response to a message from the information processing device 1 proposing a countermeasure, the user inputs a message pointing out an error in the countermeasure, such as "This is likely not due to an ignition abnormality, but rather to an abnormality on the sensor side." The information processing device 1 inputs this message into the large-scale language model, which notifies the information processing device 1 that the message determines a countermeasure different from the countermeasure and that the countermeasure is to replace the sensor. If the information processing device 1 determines that the countermeasure is to replace the sensor, it displays a message saying, "Understood. We propose replacing the sensor and conducting an experiment to isolate the cause. Please approve." It also displays a pull-down menu indicating that the part will be replaced as a countermeasure, as well as "Approve" and "Cancel" buttons requesting approval for the countermeasure. The information processing device 1 displays a message based on a pre-stored template for sensor replacement, and also displays user interface components, such as the message, pull-down menu, and button, defined for sensor replacement. Furthermore, when displaying a pull-down menu, the information processing device 1 displays the pull-down menu so that a value corresponding to the determined countermeasure is selected by default. Although not shown, the pull-down menu also includes a number of selectable actions that can be taken depending on the state of the substrate processing apparatus 3, such as sensor calibration, in addition to part replacement.

[0080] In this example, in response to a request for approval for part replacement from the information processing device 1, the user inputs a message saying, "Replacement is not necessary. Calibration is sufficient." The information processing device 1 inputs this message into the large-scale language model, which then notifies the information processing device 1 that this message determines a change in the response content and that the response content is sensor calibration. When sensor calibration is determined as the response content, the information processing device 1 displays a message saying, "Understood. If the following is acceptable, please approve." It also displays a pull-down menu indicating that sensor calibration will be performed as the response, as well as "Approve" and "Cancel" buttons requesting approval for this response. The information processing device 1 displays a menu based on pre-stored template phrases for sensor calibration, and also displays the above-mentioned messages, pull-down menus, buttons, and other user interface components.

[0081] In this example, the user operates the "Approve" button, and the information processing device 1 thereby accepts from the user a decision to calibrate the sensor. The information processing device 1 makes a request for sensor calibration to, for example, a system that manages the maintenance of the substrate processing device 3, and displays a message saying "A request for sensor calibration has been issued" based on a pre-stored template.

[0082] When the user decides on a solution that differs from the proposed solution, the information processing device 1 according to this embodiment inquires the user about the reason or basis for this decision. In this example, the information processing device 1 inquires the user by displaying a message based on a pre-stored template, such as "In order to expedite future solutions, please tell us why you determined that a sensor abnormality was the cause." In response, the user enters the message "Because there is no abnormality in Vpp." Although not shown in FIG. 8 , the information processing device 1 creates a report including the reason for the user's decision, asks the user for confirmation, and stores this report in the correspondence DB 5 if the user approves.

[0083] FIG. 9 shows an example of a message exchange that occurred approximately one month after the message exchange shown in FIG. 8 (February 10, 20XX). In this example, as in the previous case, the information processing device 1 displays the message "OES 777.0 nm emission intensity...see...," a graph of the sensor measurement values ​​acquired this time from the substrate processing device 3, and the message "See past analysis...." However, this time, since the previous report is stored in the correspondence DB 5, the information processing device 1 checks for any abnormalities in Vpp and then displays a message suggesting a countermeasure: "Vpp also exhibits unusual behavior. This is likely due to an abnormal plasma ignition rather than an abnormality in the OES sensor. We suggest adjusting the ignition step length." The information processing device 1 also displays a message saying, "The recipe will be modified as follows. Please approve.", along with a comparison of the values ​​of the ignition step before and after the modification in a table format. The information processing device 1 also displays an "Approve" button and a "Cancel" button to accept the user's decision on the recipe modification based on the options. In this example, the user operates the "Approve" button, and the information processing device 1 then changes the Ignition step and displays a message saying "Modifications have been reflected."

[0084] In the present embodiment, if a similar countermeasure proposed to a user in a similar situation is approved multiple times, the information processing device 1 suggests to the user that the countermeasure be automatically implemented in the event of a similar situation occurring in the future. In this example, the information processing device 1 displays a message saying, "Since the initial proposal for the above has been approved multiple times, we propose automating the recipe modification for future OES 777.0 nm emission intensity control limit exceedances. You can check and modify the recipe modification logic via the URL." to suggest changing the conditions for the automatic response. The information processing device 1 displays an "Approve" button and a "Cancel" button to request the user's approval for changing the conditions for the automatic response. In this embodiment, the information processing device 1 displays the above message based on a pre-stored template for the condition change, and also displays user interface components such as pre-defined buttons. In this example, the user does not operate the button, but instead enters the message "I'll consider this later," thereby postponing the change to the conditions for the automatic response.

[0085] If the user approves the change in the conditions for automatic response, the information processing device 1 changes the conditions for automatic response stored in the storage unit 12, for example, and automatically corrects the recipe if a similar situation occurs thereafter. The information processing device 1 may also re-train the learning model that determines whether or not an automatic response is possible so that it determines that an automatic response is possible for this condition.

[0086] 10 and 11 are schematic diagrams for explaining the flow of data in the correspondence determination process. These diagrams illustrate the flow of data exchanged between the information processing device 1 (the main processing executed by the processing unit 11 thereof), the large-scale language model, and the correspondence DB 5 in the example of message exchange shown in FIG. 8. First, the information processing device 1 displays a message saying "OES777.0 nm emission intensity is...see" and a graph of the sensor measurement values ​​based on the sensor measurement values ​​acquired from the substrate processing device 3. This message is not generated using a large-scale language model, but is generated by the information processing device 1 based on a predetermined template.

[0087] Next, the information processing device 1 extracts data similar to the sensor measurement value currently acquired from the substrate processing device 3 from the information stored in the correspondence DB 5. The information processing device 1 inputs a predetermined question (for example, "What should we communicate?") and the similar data extracted from the correspondence DB 5 into the large-scale language model (so-called prompt input), and acquires answer information in natural language from the large-scale language model. Based on the information acquired from the large-scale language model, the information processing device 1 displays the message "See past analysis..." and the response suggestion "We suggest that the cause is... plasma."

[0088] In response, the user inputs a message, "This is an ignition abnormality...high.", which is given to the large-scale language model, which then interprets the content of this message and gives the resulting information to the information processing device 1. Based on the information given by the large-scale language model, the information processing device 1 uses a predetermined template to display the message "Understood...please," a pull-down menu with "Parts replacement" selected, and "Accept" and "Cancel" buttons.

[0089] In response, the message "Enough until replacement..." input by the user is given to the large-scale language model, which then interprets the content of this message and gives the resulting information to the information processing device 1. Based on the information given from the large-scale language model, the information processing device 1 uses a predetermined template to display the message "Understood... please do so," a pull-down menu with "sensor calibration" selected, and "Accept" and "Cancel" buttons.

[0090] When the user operates the "Approve" button, the fact that the user has selected "Approve" is conveyed to the information processing device 1 without going through a large-scale language model. In response, the information processing device 1 displays a message "A sensor calibration request has been issued" and a message "Please let us know what to do next" using a predetermined template.

[0091] In response, the user inputs a message, "Because there is no abnormality in Vpp.", which is provided to the large-scale language model, which then interprets the message and provides the information resulting from that interpretation to the information processing device 1. In response to this information, the information processing device 1 creates a report including information on the response determined by the user and information such as messages exchanged with the user, and stores the report in the response DB 5. Note that in this figure, the process of receiving approval from the user for the created report is not shown.

[0092] <Acquisition of Similar Data> The information processing apparatus 1 according to this embodiment performs a process of acquiring similar data from information stored in the correspondence DB 5 based on sensor measurement values ​​acquired from the substrate processing apparatus 3. To this end, the information processing system according to this embodiment converts various data, such as sensor measurement values ​​acquired from the substrate processing apparatus 3 and natural language messages exchanged with users, into vector data using a pre-generated encoder, and stores the converted vector data in the correspondence DB 5. When searching the correspondence DB 5 for data similar to desired data, the information processing apparatus 1 converts the desired data into vector data using the encoder, calculates the similarity with each vector data stored in the correspondence DB 5, and can determine vector data whose similarity exceeds a threshold as similar. The information processing apparatus 1 can acquire data similar to the desired data from the correspondence DB 5 by restoring vector data determined to be similar using a pre-generated decoder.

[0093] The correspondence DB 5 may store only vector data obtained by converting original data using an encoder, or may store the original data and the vector data in association with each other. Furthermore, RAG (Retrieval Augment Generation) is a technology for performing language processing using a large-scale language model and a database in combination. The information processing system according to this embodiment may use this RAG technology to link the large-scale language model and the database. Since RAG is an existing technology, detailed description thereof will be omitted.

[0094] <Other Application Examples> In the above-described embodiment, an information processing system has been described in which the information processing apparatus 1 automatically performs substrate processing using the substrate processing apparatus 3 according to a predetermined procedure, and if an automatic response is not possible, the information processing system presents a response plan to the user and executes the response determined by the user. This information processing system is intended for situations in which, for example, the substrate processing apparatus 3 installed in a factory or the like is automatically operated to continuously perform substrate processing. However, the application of the technology of the present disclosure is not limited to the above and can be applied to various situations.

[0095] The technology disclosed herein can be applied, for example, to searching for optimal setting values ​​for substrate processing by repeatedly conducting experiments using a substrate processing apparatus 3, so-called substrate processing recipe development. The information processing apparatus 1 performs substrate processing based on initial setting values, for example, provided by a user. Based on sensor measurements obtained during and as a result of the substrate processing, the information processing apparatus 1 estimates optimal setting values, such as setting values ​​that are expected to produce better results or useful information, and performs substrate processing using the estimated setting values. The information processing apparatus 1 automatically and repeatedly predicts setting values ​​and performs substrate processing using the estimated setting values ​​to search for optimal setting values. Estimation of setting values ​​can be performed using techniques such as design of experiments based on an orthogonal array or Bayesian optimization. Design of experiments is a technique for estimating experimental conditions that efficiently collect information to achieve a purpose. Bayesian optimization is a technique for estimating experimental conditions that are likely to produce an optimal value based on a Gaussian process and then repeatedly conducting experiments under the estimated experimental conditions to obtain the optimal value. However, the information processing apparatus 1 may also employ techniques other than design of experiments or Bayesian optimization to estimate setting values.

[0096] For example, when it is not possible to predict optimal setting values, such as setting values ​​that are expected to produce better results or useful information, when the reliability of the setting value prediction is low, or when substrate processing using the predicted setting values ​​is difficult, the information processing device 1 according to this embodiment determines that automatic response is not possible. When it determines that automatic response is not possible, the information processing device 1 presents a response plan to the user along with data such as sensor measurement values ​​obtained through previous experiments. The information processing device 1 utilizes a large-scale language model to exchange messages with the user in natural language, appropriately presents information required by the user, and accepts the user's decision on the response. The information processing device 1 performs substrate processing using the response determined by the user, and then automatically searches for optimal setting values.

[0097] <Summary> In the information processing system according to this embodiment having the above configuration, the information processing device 1 acquires sensor measurement values ​​related to the substrate processing performed by the substrate processing device 3, and determines whether an automatic response to the substrate processing situation is possible based on the acquired sensor measurement values. If it is determined that an automatic response is possible, the information processing device 1 performs a predetermined automatic response. If it is determined that an automatic response is not possible, the information processing device 1 outputs the sensor measurement values ​​acquired from the substrate processing device 3 and a response plan to the situation based on the sensor measurement values, accepts a response to the situation from the user, and implements the determined response. As a result, the information processing system according to this embodiment can be expected to support the information processing device 1 in performing response processing according to the situation of the substrate processing device 3 or the substrate processing, etc.

[0098] Furthermore, in the information processing system according to this embodiment, the information processing device 1 stores information relating to the response decided by the user in the response DB 5, and generates a response plan for the status of the substrate processing device 3 or the substrate processing, etc., based on the information stored in the response DB 5. As a result, when a similar situation occurs in the future, the information processing system according to this embodiment can be expected to propose a more appropriate response plan to the user based on the response decided by the user.

[0099] Furthermore, in the information processing system according to this embodiment, the information processing device 1 extracts data similar to the sensor measurement values ​​acquired from the substrate processing device 3 from the information stored in the correspondence DB 5, and generates a countermeasure plan based on the extracted information. As a result, the information processing system according to this embodiment can be expected to effectively utilize the information stored in the correspondence DB 5 to propose more appropriate countermeasures to the user.

[0100] Furthermore, in the information processing system according to this embodiment, the information processing device 1 performs automatic response processing based on information such as conditions or procedures stored in advance in the storage unit 12. When a response decision is received from the user, the information processing device 1 updates, for example, by adding or correcting the information such as the conditions or procedures for the automatic response stored in the storage unit. As a result, the information processing system according to this embodiment is expected to reflect the response decided by the user in subsequent automatic response processing.

[0101] Furthermore, in the information processing system according to this embodiment, the information processing device 1 accepts input of a message in natural language from the user in response to the output response plan. Based on the input message, the information processing device 1 outputs information related to substrate processing, such as previous sensor measurement values, or information related to the updated response plan. This allows the information processing system according to this embodiment to provide various information based on the user's message in natural language, and is expected to assist the user in determining a response.

[0102] Furthermore, in the information processing system according to this embodiment, the information processing device 1 outputs a natural language message in response to a natural language message received as input from a user, and stores the message from the user and the response message in association with each other in the correspondence DB 5. The information processing device 1 outputs subsequent response messages based on the information stored in the correspondence DB 5. As a result, the information processing system according to this embodiment is able to exchange messages in natural language with the user, and is also able to exchange subsequent messages based on previous message exchanges, which is expected to enable more accurate message exchanges with the user.

[0103] In the information processing system according to the present embodiment, the information processing device 1 accepts the decision on the response based on a non-natural language. As a result, the information processing system according to the present embodiment is expected to more reliably implement the response decided by the user compared to when the decision on the response is accepted through message exchange based on a natural language.

[0104] Furthermore, in the information processing system according to this embodiment, the information processing device 1 utilizes a large-scale language model that has been trained in advance to generate countermeasures for the status of the substrate processing device 3 or the substrate processing, based on the sensor measurement values ​​acquired from the substrate processing device 3 and the information stored in the correspondence DB 5. As a result, the information processing system according to this embodiment is expected to provide countermeasures appropriate for the status in a natural language that is easy for the user to understand.

[0105] The embodiments disclosed herein are to be considered as illustrative in all respects and not restrictive. The scope of the present disclosure is defined by the claims, not by the above meaning, and is intended to include all modifications within the meaning and scope of the claims.

[0106] The matters described in each embodiment can be combined with each other. Furthermore, the independent claims and dependent claims described in the claims can be combined with each other in any and all combinations, regardless of the reference format. Furthermore, the claims do not use a format in which a claim references two or more other claims (multiple claim format), but this is not limited to this. They may also be written using a multiple claim format or a format in which multiple claims reference at least one other multiple claim (multi-multi claim format).

[0107] REFERENCE SIGNS LIST 1 Information processing device (computer) 3 Substrate processing device 5 Correspondence DB 7 Learning model 11 Processing unit 11a Data acquisition unit 11b Automatic correspondence determination unit 11c Correspondence plan generation unit 11d Decision reception unit 11e Control processing unit 11f Display processing unit 12 Storage unit 12a Program (computer program) 13 Communication unit 14 Display unit 15 Operation unit

Claims

1. A computer program that causes a computer to perform the following processes: acquire information related to substrate processing performed by a substrate processing apparatus; determine whether or not an automatic response to the status of the substrate processing is possible based on the acquired information; if it is determined that an automatic response is not possible, output information related to the substrate processing and a response plan for the status; accept a decision on the response to the status; and output information related to the response plan that has been decided upon.

2. The computer program according to claim 1, wherein information relating to the response content for which a decision has been accepted is stored in a storage unit, and the response plan is generated based on the information stored in the storage unit.

3. The computer program according to claim 2, further comprising: extracting information similar to the acquired information relating to substrate processing from the information stored in the storage unit; and generating the countermeasure plan based on the extracted information.

4. The computer program of claim 2, which accepts information relating to substrate processing as input, uses a learning model that has been machine-learned to predict whether or not an automatic response to the substrate processing situation is possible, and determines whether or not the response is possible, and re-learns the learning model based on the information stored in the memory unit.

5. The computer program according to claim 1, wherein the automatic response is performed based on conditions or procedures stored in a storage unit, and the conditions or procedures for the automatic response are updated based on the response content for which a decision has been accepted.

6. The computer program according to claim 1, further comprising: receiving input of natural language information for the outputted countermeasure; and outputting information relating to the substrate processing or the updated countermeasure based on the received natural language information.

7. A computer program as claimed in claim 6, which outputs natural language information in response to natural language information that has been input, stores the input natural language information and the output natural language information in a memory unit in association with each other, and outputs subsequent natural language information in response based on the natural language information stored in the memory unit.

8. The computer program of claim 1, wherein the decision is accepted based on a non-natural language.

9. The computer program according to claim 8, wherein the decision is accepted by accepting a selection from options, accepting a numerical input, or accepting an input of table data including a plurality of numerical values.

10. The computer program of claim 1, wherein the response proposals are generated using a large-scale language model.

11. An information processing method in which an information processing device acquires information related to substrate processing performed by a substrate processing device, determines whether or not an automatic response to the substrate processing situation is possible based on the acquired information, and if it is determined that an automatic response is not possible, outputs information related to the substrate processing and a response plan for the situation, accepts a decision on the response to the situation, and outputs information related to the response plan that has been decided upon.

12. An information processing device comprising a processing unit, wherein the processing unit acquires information relating to substrate processing performed by a substrate processing device, determines whether an automatic response to the situation of the substrate processing is possible based on the acquired information, and if it is determined that an automatic response is not possible, outputs information relating to the substrate processing and a response plan for the situation, accepts a decision on the response to the situation, and outputs information relating to the response plan that has been decided upon.

Images