Program, information processing method, information processing apparatus, and model generation method

Abstract

To provide a program etc. for determining abnormalities in an exhaust system of a semiconductor manufacturing device with high accuracy.SOLUTION: A program causes a computer to execute the processes of: acquiring process data including at least two of the type and flow rate of gas supplied into a semiconductor manufacturing device, the pressure inside a chamber, or the valve opening degree of an automatic pressure control device; inputting the process data into a learning model that has been trained to output an estimated pressure value inside the semiconductor manufacturing device when the process data is input, and outputting the estimated pressure value; and determining whether an abnormal state exists on the basis of the estimated pressure value.SELECTED DRAWING: Figure 1

Description

[Technical field]

[0001] The present technology relates to a program, an information processing method, an information processing device, and a model generation method related to processing by a semiconductor manufacturing device. [Background technology]

[0002] Conventionally, monitoring methods for detecting abnormalities in semiconductor manufacturing equipment have been proposed. For example, the monitoring method described in Patent Document 1 determines abnormalities based on the gas temperature in a vacuum pump portion of the semiconductor manufacturing equipment. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Unexamined Patent Publication No. 5-9759 Summary of the Invention [Problem to be solved by the invention]

[0004] The program described in Patent Document 1 judges an abnormality in the exhaust system based only on the gas temperature, and therefore has low accuracy in judging an abnormality in the exhaust system.

[0005] The present disclosure has been made in consideration of the above circumstances, and has an object to provide a program or the like for determining an abnormality in an exhaust system of a semiconductor manufacturing apparatus with high accuracy. [Means for solving the problem]

[0006] A program according to an embodiment of the present disclosure acquires process data including at least two of the type and flow rate of gas supplied into a semiconductor manufacturing equipment, the pressure inside a chamber, or the valve opening degree of an automatic pressure control device, inputs the process data into a learning model that has been trained to output an estimated pressure value inside the semiconductor manufacturing equipment when the process data is input, outputs the estimated pressure value, and causes a computer to execute a process of determining whether an abnormal state exists based on the estimated pressure value.

[0007] A program according to an embodiment of the present disclosure acquires an actual pressure value measured by a vacuum gauge in the semiconductor manufacturing equipment, and determines whether an abnormal state exists based on the estimated pressure value and the actual pressure value.

[0008] A program according to an embodiment of the present disclosure acquires process data in an idle state of the device before starting a manufacturing process for supplying gas into a semiconductor device.

[0009] In a program according to an embodiment of the present disclosure, the process data includes an operating state of an exhaust pump.

[0010] A program according to an embodiment of the present disclosure determines that an abnormal state exists when a difference between the estimated pressure value and the actually measured pressure value is equal to or greater than a predetermined value.

[0011] A program according to one embodiment of the present disclosure continuously obtains the difference between the estimated pressure value and the actual measured pressure value, creates time series data that corresponds to the difference and time, and displays the time series data on a display unit.

[0012] A program according to an embodiment of the present disclosure predicts the time when the difference will be equal to or greater than a predetermined value, based on the time-series data.

[0013] An information processing method according to one embodiment of the present disclosure acquires process data including at least two of the type and flow rate of gas supplied into a semiconductor manufacturing equipment, the pressure inside a chamber, or the valve opening degree of an automatic pressure control device, inputs the process data into a learning model that has been trained to output an estimated pressure value inside the semiconductor manufacturing equipment when the process data is input, outputs the estimated pressure value, and determines whether an abnormal state is present based on the estimated pressure value.

[0014] An information processing device according to one embodiment of the present disclosure includes a process data acquisition unit that acquires process data including at least two of the type and flow rate of gas supplied into a semiconductor manufacturing equipment, the pressure inside a chamber, or the valve opening degree of an automatic pressure control device, an estimated pressure value output unit that inputs the process data into a learning model that has been trained to output an estimated pressure value inside the semiconductor manufacturing equipment when the process data is input, and outputs the estimated pressure value, and an abnormality determination unit that determines whether an abnormal state exists based on the estimated pressure value.

[0015] A model generation method according to one embodiment of the present disclosure acquires a pressure value of a vacuum gauge in a semiconductor manufacturing equipment, acquires process data including at least two of the type and flow rate of gas supplied to the semiconductor manufacturing equipment, the pressure in a chamber, or the valve opening of an automatic pressure control device, and training data including the acquired pressure value, and generates a learning model that outputs an estimated pressure value when process data including at least two of the type and flow rate of gas supplied to the semiconductor manufacturing equipment, the pressure in a chamber, or the valve opening of an automatic pressure control device is input based on the acquired training data. Effect of the Invention

[0016] A program according to an embodiment of the present disclosure can determine with high accuracy whether a semiconductor manufacturing device is in an abnormal state. [Brief description of the drawings]

[0017] [Figure 1] FIG. 1 is a schematic diagram showing a schematic configuration of a semiconductor manufacturing apparatus. [Diagram 2] FIG. 1 is a block diagram showing a configuration of an information processing device. [Diagram 3] FIG. 2 is a block diagram showing a configuration of a server device. [Figure 4] FIG. 1 is an explanatory diagram showing a learning model. [Diagram 5] FIG. 2 is an explanatory diagram showing a time-series data table. [Figure 6] FIG. 2 is a schematic diagram illustrating an example of a screen displayed by an information processing device. [Figure 7] 1 is a flowchart showing a procedure for generating a learning model. [Figure 8] 10 is a flowchart showing a procedure of a process performed by an information processing device. [Figure 9] FIG. 11 is a block diagram showing a configuration of an information processing device according to a second embodiment. [Figure 10] 11 is a schematic diagram showing an example of a screen displayed by an information processing device according to the second embodiment. FIG. [Figure 11] 10 is a flowchart showing the procedure of processing performed by an information processing device according to the second embodiment. [Figure 12] FIG. 2 is a schematic diagram showing a schematic configuration of a second semiconductor manufacturing apparatus. [Figure 13] 13 is a flowchart showing the procedure for generating a second learning model. [Figure 14] 11 is a table for comparing process data and device states in an equipment start-up state, an equipment idle state, and a manufacturing process state. [Figure 15] 11 is a graph showing an example of fluctuations in the pressure inside the chamber and the APC opening degree. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0018] (Embodiment 1) Hereinafter, specific examples of the program, information processing device, and the like according to the first embodiment will be described with reference to the drawings.

[0019] 1 is a schematic diagram showing the schematic configuration of a semiconductor manufacturing apparatus. The semiconductor manufacturing apparatus 1 according to this embodiment is an apparatus that performs an etching process, a film formation process, or the like by performing a plasma process on a semiconductor (including MEMS (Micro Electro Mechanical Systems)) substrate. The semiconductor manufacturing apparatus 1 includes an information processing apparatus 30, and the information processing apparatus 30 exchanges data and learning models with a server apparatus 5.

[0020] The semiconductor manufacturing equipment 1 includes a chamber 11, a sample stage 13, a gas supply source (Gas No. 1 to No. 6), an MFC (Mass Flow Controller) 71, a gas valve 71a, a vacuum gauge 72, an APC (Automatic Pressure Controller) 73, a first pump 74, a second pump 75, a vacuum gauge 76, an exhaust pipe 77, a pressure / flow meter 78, and a thermometer 79.

[0021] Gas valves 71a are provided in the respective flow paths from gas supply sources associated with different types of gas (Gas No. 1 to No. 6 types) to chamber 11. MFCs 71 are provided in the respective flow paths between the gas supply sources and gas valves 71a, and control the flow rates of the respective gases (Gas No. 1 to No. 6 flow rates).

[0022] A substrate S is placed on a sample stage 13 provided in the chamber 11. With the substrate S electrostatically attracted to an electrostatic chuck (not shown) provided on the sample stage 13, gas is supplied from a gas supply source into the chamber, and a plasma process such as an etching process or a film formation process is performed on the substrate S by a plasma source (not shown). A vacuum gauge 72 measures the pressure inside the chamber 11 (internal chamber pressure).

[0023] The semiconductor manufacturing equipment 1 performs heating by a heater (not shown) that heats the wall surface inside the chamber 11. Also, He gas is supplied to the back surface of the substrate S placed on the sample stage 13 to perform cooling. A pressure / flow meter 78 controls the pressure and measures the flow rate of the He gas supplied to the back surface of the substrate S. A thermometer 79 measures the heater temperature.

[0024] The APC 73 adjusts the opening of a valve (APC opening) provided therein, thereby controlling the pressure inside the chamber 11 (internal chamber pressure) measured by the vacuum gauge 72 to be a set pressure. The first pump 74 is a pump for exhausting reaction products, such as a turbo molecular pump. The second pump 75 is a dry pump, such as a pump for assisting the first pump 74. By operating the first pump 74 and the second pump 75, reaction products inside the chamber 11 are exhausted and the pressure inside the chamber 11 is reduced. Hereinafter, the first pump 74, the second pump 75 and the exhaust pipe 77 are collectively referred to as an exhaust system. When the processing of the substrate S is completed, the semiconductor manufacturing apparatus 1 fully opens the valve of the APC to exhaust the residual gas and He gas from within the chamber.

[0025] In this way, the semiconductor manufacturing apparatus 1 is equipped with various devices and measuring instruments, and the information processing device 30 can obtain various process data related to the manufacturing process from these devices and measuring instruments. In the illustrated example, the information processing device 30 can obtain the following data in the process data: types of gas No. 1 to No. 6 from the gas supply source, flow rates of gas No. 1 to No. 6 from the MFC 71, pressure inside the chamber from the vacuum gauge 72, APC opening from the APC 73, motor speed of the first pump 74 (first pump speed) from the first pump 74, motor speed of the second pump 75 (second pump speed) from the second pump 75, He gas pressure and flow rate from the pressure and flow meter 78, and heater temperature from the thermometer 79. In this embodiment, the operating states of the first pump 74 and the second pump 75 are the first pump speed and the second pump speed. The first pump speed and the second pump speed are controlled to be constant speeds.

[0026] The vacuum gauge 76 measures the pressure value (actual pressure value) of the exhaust pipe 77 between the first pump 74 and the second pump 75. The information processing device 30 is capable of acquiring the actual pressure value from the vacuum gauge 76.

[0027] 2 is a block diagram showing the configuration of an information processing device. The information processing device 30 according to this embodiment is configured to include a processing unit 31, a storage unit 32, a communication unit 33, a display unit 34, an operation unit 35, and a card slot 36. In this embodiment, the information processing device 30 is configured as one device, but the information processing device 30 may be configured to realize the functions of the information processing device 30 by multiple devices performing distributed processing. Also, the information processing device 30 may be configured to be incorporated in the semiconductor manufacturing device 1.

[0028] The information processing device 30 judges whether the semiconductor manufacturing device 1 is in an abnormal state. The semiconductor manufacturing device 1 supplies gas into the chamber 11 and performs substrate processing. If the semiconductor manufacturing device 1 continues manufacturing semiconductors, reaction products (deposits) of the gas supplied into the chamber 11 will clog the piping, or the exhaust capacity will decrease due to deterioration of the first pump 74 or the second pump 75, and the reproducibility of the substrate processing (plasma processing) will be impaired. The information processing device 30 outputs the pressure value (estimated pressure value) of the exhaust system that should be indicated by the vacuum gauge 76 when it is assumed that all the exhaust systems are normal, using the learning model 32b. If the semiconductor manufacturing device 1 is in an abnormal state, the exhaust capacity of the exhaust system will decrease, so the actual pressure value measured by the vacuum gauge 76 will be higher than the estimated pressure value. The information processing device 30 judges whether the semiconductor manufacturing device 1 is in an abnormal state based on the difference between the actual pressure value and the estimated pressure value, and notifies the user of the semiconductor manufacturing device 1 if it is judged to be in an abnormal state. The information processing device 30 may determine whether the semiconductor manufacturing equipment 1 is in an abnormal state based on the first pump rotation speed or the second pump rotation speed. If the reaction products (deposits) of the gas supplied into the chamber 11 clog the piping, the first pump or the second pump may be unable to maintain its rotation speed and may have to slow down. The information processing device 30 can determine that the semiconductor manufacturing equipment 1 is in an abnormal state when the first pump rotation speed or the second pump rotation speed becomes lower than a predetermined rotation speed.

[0029] The processing unit 31 is configured using an arithmetic device such as a CPU (Central Processing Unit), an MPU (Micro-Processing Unit), or a GPU (Graphics Processing Unit). The processing unit 31 can perform various processes by reading and executing programs stored in the storage unit 32. In this embodiment, the processing unit 31 reads and executes a program 32a and a learning model 32b stored in the storage unit 32 to perform various processes such as a process of outputting an estimated pressure value in the semiconductor manufacturing equipment 1, a process of determining whether or not the semiconductor manufacturing equipment 1 is in an abnormal state, and a process of creating time-series data.

[0030] The storage unit 32 is configured using a large-capacity storage device such as a hard disk or an SSD (Solid State Drive). The storage unit 32 stores various programs executed by the processing unit 31 and various data required for the processing of the processing unit 31. In this embodiment, the storage unit 32 stores a program 32a executed by the processing unit 31, a learning model 32b used in the processing performed by the execution of the program 32a, and a database 32c that records data output from the processing unit 31.

[0031] In this embodiment, the program 32a (program product) is provided in a form recorded on a recording medium 99 such as a memory card, and the information processing device 30 reads the program 32a from the recording medium 99 and stores it in the storage unit 32. However, the program 32a may be written in the storage unit 32, for example, during the manufacturing stage of the information processing device 30. For example, the program 32a may be read from the recording medium 99 by a writing device and written to the storage unit 32 of the information processing device 30. The program 32a may be provided in a form distributed via a network.

[0032] The learning model 32b is a learning model that outputs an estimated pressure value based on the process data acquired from the semiconductor manufacturing equipment 1. In this embodiment, the learning model 32b outputs an estimated pressure value when process data including at least two of the types of gases No. 1 to No. 6 and the flow rates of gases No. 1 to No. 6, the pressure inside the chamber, or the APC opening degree measured by various measuring instruments of the semiconductor manufacturing equipment 1 is input. The process data may include at least one of the first pump rotation speed or the second pump rotation speed, and may include the heater temperature or the He gas pressure and flow rate. Although only one learning model 32b is illustrated in FIG. 2, a plurality of learning models 32b may be stored in the storage unit 32. The learning model 32b may be stored in a cloud provided by an external server. Details of the learning model 32b will be described later.

[0033] The database 32c records the time series data created by the processing unit 31. The time series data will be described in detail later. The data recording unit may record the process data, the measured pressure value, and the estimated pressure value in addition to the time series data.

[0034] The communication unit 33 is connected to the semiconductor manufacturing equipment 1 via a communication cable or the like, and transmits and receives various data to and from the semiconductor manufacturing equipment 1 via this communication cable. In this embodiment, the communication unit 33 receives and acquires process data measured by each measuring device of the semiconductor manufacturing equipment 1 through communication via the communication cable. Note that the communication unit 33 may be configured to perform wireless communication with the semiconductor manufacturing equipment 1.

[0035] The display unit 34 is configured with a liquid crystal display or the like, and displays various images and characters based on the processing of the processing unit 31. The operation unit 35 accepts operations by the user and notifies the processing unit 31 of the accepted operations. For example, the operation unit 35 is an input device such as a mouse and a keyboard, and these input devices may be configured to be removable from the information processing device 30. Furthermore, for example, the operation unit 35 may accept operations by the user via an input device such as a touch panel provided on the surface of the display unit 34.

[0036] The card slot 36 allows a recording medium 99 such as a memory card to be attached and detached, and reads and writes data from and to the attached recording medium 99. The card slot 36 reads data from the recording medium 99 and provides it to the processing unit 31, and writes data provided from the processing unit 31 to the recording medium 99. In this embodiment, a learning model 32b that has been machine-learned in advance by the server device 5 is recorded on the recording medium 99 and provided to the information processing device 30, and the information processing device 30 reads the learning model 32b recorded on the recording medium 99 into the card slot 36 and stores it in the storage unit 32. Note that in this embodiment, data is exchanged between the information processing device 30 and the server device 5 via the recording medium 99, but this is not limiting. For example, the information processing device 30 and the server device 5 may exchange data via communication such as a LAN (Local Area Network) or the Internet.

[0037] The storage unit 32 may be an external storage device connected to the information processing device 30. The information processing device 30 may be a multi-computer including a plurality of computers, or may be a virtual machine virtually constructed by software. The information processing device 30 may not have a user interface such as the display unit 34 and the operation unit 35. In this case, for example, the information processing device 30 may be configured such that an administrator operates the information processing device 30 via another device.

[0038] In the processing unit 31 of the information processing device 30 according to this embodiment, a process data acquiring unit 31a, an actual pressure value acquiring unit 31b, an estimated pressure value output unit 31c, an abnormality determining unit 31d, a display processing unit 31e, a time series data creating unit 31f, a data recording unit 31g, and the like are realized as software functional units by the processing unit 31 reading and executing a program 32a stored in a storage unit 32. The processing unit 31 also has a counter function that starts from the time when the semiconductor manufacturing apparatus 1 starts manufacturing semiconductors. Note that, instead of the counter function, a clock function that measures actual time may be provided.

[0039] The process data acquiring unit 31a performs a process of acquiring process data from the semiconductor manufacturing equipment 1 by communicating with the semiconductor manufacturing equipment 1 via the communication unit 33. In this embodiment, the process data includes data measured by various measuring instruments included in the semiconductor manufacturing equipment 1 as described above.

[0040] The actual pressure value acquiring unit 31 b communicates with the semiconductor manufacturing equipment 1 via the communication unit 33 to acquire the actual pressure value measured by the vacuum gauge 76 included in the semiconductor manufacturing equipment 1 .

[0041] The estimated pressure value output unit 31c reads out the learning model 32b from the storage unit 32, inputs the process data acquired by the process data acquisition unit 31a to the learning model 32b, and outputs the estimated pressure value.

[0042] The abnormality determination unit 31d determines whether the semiconductor manufacturing equipment 1 is in an abnormal state based on the actual pressure value acquired by the actual pressure value acquisition unit 31b and the estimated pressure value output by the estimated pressure value output unit 31c. Specifically, the difference between the actual pressure value and the estimated pressure value is calculated, and if the calculated difference is equal to or greater than a preset value (threshold value), the semiconductor manufacturing equipment 1 is determined to be in an abnormal state. The abnormality determination unit 31d may determine that the semiconductor manufacturing equipment 1 is in an abnormal state if the difference is equal to or greater than the threshold value for a certain period of time or if a difference equal to or greater than the threshold value is calculated a certain number of times or more.

[0043] The display processing unit 31e performs processing to display various images, characters, and the like on the display unit 34. In this embodiment, the display processing unit 31e displays, for example, the determination result of the abnormality determination unit 31d. The display processing unit 31e may also display the acquired process data, the actual pressure value, the estimated pressure value, or time-series data recorded in a database 32c of the storage unit 32 described later.

[0044] The time-series data creation unit 31f performs a process of creating time-series data. Specifically, the time-series data is the difference between the actual pressure value and the estimated pressure value, which is made to correspond to the elapsed time from the time when the semiconductor manufacturing equipment 1 starts manufacturing the semiconductor. The time-series data may include the actual pressure value and the estimated pressure value. The difference may also be made to correspond to the actual time.

[0045] The data recording unit 31g records the time series data created by the time series data creating unit in the database 32c of the storage unit 32. Every time the time series data creating unit 31f creates time series data, the data recording unit 31g registers the time series data in a time series data table of the database 32c. The time series data table will be described in detail later.

[0046] 3 is a block diagram showing the configuration of a server device 5 according to the present embodiment. The server device 5 according to the present embodiment is configured to include a processing unit 51, a storage unit 52, a communication unit 53, and a card slot 54. Note that in the present embodiment, the processing is described as being performed by one server device 5, but the processing may be distributed among a plurality of server devices 5. Furthermore, the processing performed by the server device 5 may be performed by an information processing device 30.

[0047] The processing unit 51 is configured using an arithmetic processing device such as a CPU, an MPU, a GPU, etc. The processing unit 51 reads out and executes a server program 52a stored in the storage unit 52 to perform various processes such as a learning process of the learning model 32b used by the information processing device 30.

[0048] The storage unit 52 is configured using a large-capacity storage device such as a hard disk or SSD. The storage unit 52 stores various programs executed by the processing unit 51 and various data necessary for the processing of the processing unit 51. In this embodiment, the storage unit 52 stores a server program 52a executed by the processing unit 51. In this embodiment, the server program 52a is provided in a form recorded on a recording medium 99 such as a memory card, and the server device 5 reads the server program 52a from the recording medium 99 and stores it in the storage unit 52. However, the server program 52a may be written in the storage unit 52, for example, during the manufacturing stage of the server device 5. For example, the server program 52a may be read from the recording medium 99 by a writing device and written to the storage unit 52 of the server device 5. The server program 52a may be provided in a form of distribution via a network.

[0049] The communication unit 53 communicates with various devices via a network N including an in-house LAN, a wireless LAN, the Internet, etc. The communication unit 53 transmits data provided from the processing unit 51 to other devices, and provides data received from other devices to the processing unit 51.

[0050] The card slot 54 allows a recording medium 99 such as a memory card to be attached and detached, and reads and writes data from and to the attached recording medium 99. The card slot 54 reads data recorded in the recording medium 99 and provides it to the processing unit 51, and writes data provided from the processing unit 51 to the recording medium 99. In this embodiment, the server device 5 writes the learning model 32b that has been subjected to the learning process to the recording medium 99 through the card slot 56, and the server device 5 provides the learning model 32b to the information processing device 30 of the semiconductor manufacturing apparatus 1. In this embodiment, log data 52b including a plurality of process data of the semiconductor manufacturing apparatus 1 when the apparatus is normal and a pressure value measured by the vacuum gauge 76 when the exhaust system is normal, which are recorded in the information processing device 30, are written to the recording medium 99 and provided, and the server device 5 reads and acquires the log data 52b from the recording medium 99 through the card slot 54, and uses it as training data for the learning process.

[0051] The storage unit 52 may be an external storage device connected to the server device 5. The server device 5 may be a multi-computer including a plurality of computers, or may be a virtual machine virtually constructed by software. The server device 5 is not limited to the above configuration, and may include, for example, an operation unit that accepts operation input, or a display unit that displays images.

[0052] In addition, in the processing unit 51 of the server device 5 according to this embodiment, a learning processing unit 51a and the like are realized as software functional units by the processing unit 51 reading and executing the server program 52a stored in the storage unit 52. Note that these functional units are functional units related to the process of generating the learning model 32b, and illustration and description of other functional units are omitted.

[0053] The learning processing unit 51a performs a process of generating the learning model 32b by machine learning the training data. In the first learning stage of the learning model 32b, that is, in the stage of first generating the learning model 32b, the learning processing unit 51a performs the learning process using training data created in advance by an administrator or developer of the system. The training data used at this time can be created using data accumulated from the operation of the semiconductor manufacturing equipment so far.

[0054] 4 is an explanatory diagram showing a learning model. The learning model 32b according to this embodiment is generated by machine learning using, for example, a neural network. However, machine learning may be performed by a method other than the neural network. For example, various machine learning methods such as LSTM (Long Short Term Memory), Transformer, SVM (Support Vector Machine), or k-nearest neighbor method may be adopted.

[0055] The input layer included in the learning model 32b according to this embodiment has a plurality of neurons that accept various types of process data input, and passes the input process data to the intermediate layer. The intermediate layer has a plurality of neurons that extract features of the process data, and passes the extracted features to the output layer. The output layer has neurons that output an estimated pressure value, and outputs the estimated pressure value based on the features output from the intermediate layer. The process data input to the learning model 32b may be time-series data.

[0056] In the machine learning of the learning model 32b according to this embodiment, normal training data can be used, in which a plurality of process data in normal times is used as input information, and the pressure value measured by the vacuum gauge 76 when the exhaust system is normal is used as output information. When normal process data is input, machine learning is performed using a neural network so that the same value as the pressure value measured by the vacuum gauge 76 in normal times is output. The normal process data and pressure value are, for example, the process data and pressure value during a trial run after installation or maintenance of the semiconductor manufacturing equipment 1. The learning model 32b is trained using normal training data that combines a large number of process data and pressure values.

[0057] Machine learning of the learning model 32b is not limited to the first time. If necessary, the learning model 32b may be retrained using new training data, or new training data may be added to the previous training data and retrained. In this case, the server device 50 may periodically transmit the retrained learning model 32b to the information processing device 30, and the processing unit 31 of the information processing device 30 may read out the retrained learning model 32b.

[0058] In this embodiment, the learning process of the learning model 32b is performed by the server device 5 (see FIG. 3). The server device 5 generates the learning model 32b by performing machine learning with a neural network using the above-mentioned training data prepared in advance. The learning model 32b generated by the server device 5 is provided to the information processing device 30 via a recording medium 99. The learning model 32b may be learned in the information processing device 30.

[0059] FIG. 5 is an explanatory diagram showing a time series data table. The processing unit 31 of the information processing device 30 acquires the process data and the measured pressure value every 10 seconds, for example, and outputs the estimated pressure value. The processing unit 31 calculates the difference between the acquired measured pressure value and the estimated pressure value, and registers the difference in the database 32c of the storage unit 32 in association with the elapsed time from the start of semiconductor manufacturing to the acquisition of the process data and the measured pressure value. This process is performed sequentially, and a time series data table is created in the database 32c. Note that, as shown in FIG. 5, the measured pressure value and the estimated pressure value may be further registered in the time series data table in association with the elapsed time.

[0060] Fig. 6 is a schematic diagram showing an example of a screen displayed by an information processing device. A time-series data table is displayed on the display unit 34 of the information processing device 30. The information processing device 30 may display a graph based on the time-series data table as shown in Fig. 6. The graph displayed on the display unit 34 shows, for example, actual pressure values, estimated pressure values, and differences, with the vertical axis representing pressure values ​​and the horizontal axis representing elapsed time. The graph also shows a threshold value that is a criterion for determining whether or not there is an abnormality in the difference. (e.g., 100 Pa), and it can be confirmed that the difference exceeds this threshold value, and thus the information processing device 30 has determined that the semiconductor manufacturing equipment 1 is in an abnormal state. In this embodiment, the actual measured pressure value is shown by a dotted line, the estimated pressure value by a dashed line, and the difference by a solid line. Furthermore, when the processing unit 31 determines that an abnormal state exists, a warning message for notifying the user that an abnormal state exists is displayed on the display unit 34. When the processing unit 31 determines that an abnormal state exists, the information processing device 30 may notify the user that an abnormal state exists by changing the background color of the screen, turning on a lamp, or outputting a sound.

[0061] FIG. 7 is a flowchart showing the procedure for generating a learning model. The information processing device 30 acquires process data from the semiconductor manufacturing device 1 under normal conditions (S1), and further acquires a pressure value measured by the vacuum gauge 76 of the semiconductor manufacturing device 1 (S2). The information processing device 30 associates the process data with the pressure value to generate training data (S3). The information processing device 30 determines whether the created training data is a certain number or more that is sufficient for generating the learning model 32b (S4), and if the number of training data is less than the certain number (S4: NO), changes the conditions such as the type of gas to be flowed into the chamber 11 or the flow rate of the gas (S5), and acquires the process data again (S1). If the number of training data is a certain number or more (S4: YES), the information processing device 30 transmits the training data to the server device 5 (S6), and causes the server device 5 to generate the learning model 32b (S7).

[0062] 8 is a flowchart showing the procedure of the process performed by the information processing device. The information processing device 30 causes the semiconductor manufacturing device 1 to start manufacturing semiconductors (S11). The information processing device 30 acquires process data from the semiconductor manufacturing device 1 (S12), and further acquires an actual pressure value from the vacuum gauge 76 of the semiconductor manufacturing device 1 (S13). The information processing device 30 inputs the acquired process data to the learning model 32b (S14), and outputs an estimated pressure value (S15). The information processing device 30 calculates the difference between the acquired actual pressure value and the output estimated pressure value (S16), and creates time-series data by associating the calculated difference with time (S17). The information processing device 30 determines whether the difference in the time-series data is equal to or greater than a threshold (S18). If the difference is less than the threshold (S18: NO), the information processing device 30 returns the process to S12. If the difference is equal to or greater than the threshold (S18: YES), the information processing device 30 displays a warning message on the display unit 34 (S19) and determines whether to end the process (S20). For example, if an instruction to end the process is input via the operation unit 35 (S20: YES), the information processing device 30 ends the process. If the process is not to be ended (S20: NO), the information processing device 30 returns the process to S12 and acquires the process data again.

[0063] According to the program and processing by the information processing device 30 of this embodiment, it is possible to judge an abnormality in the semiconductor manufacturing equipment 1 with high accuracy. This enables the user of the semiconductor manufacturing equipment 1 to deal with the abnormal state. In addition, by visualizing the transition of the difference between the actual pressure value and the estimated pressure value, the user can infer the cause of the abnormality.

[0064] (Embodiment 2) When the difference between the measured pressure value and the estimated pressure value is not equal to or greater than a predetermined value (threshold value), the information processing device 30 may predict the time when the difference will be equal to or greater than the predetermined value. A program and an information processing device according to the second embodiment will be described below with reference to the drawings. Among the configurations according to the second embodiment, the same configurations as those in the first embodiment are denoted by the same reference numerals, and detailed description thereof will be omitted.

[0065] FIG. 9 is a block diagram showing the configuration of an information processing device according to the second embodiment. The information processing device 30 according to this embodiment includes a prediction unit 31h in addition to the functional units according to the first embodiment. The prediction unit 31h predicts the time when the difference between the measured pressure value and the estimated pressure value becomes equal to or greater than a threshold value based on the time series data recorded in the database 32c of the storage unit 32. Specifically, for example, an approximation function that represents the relationship between the time and the difference of the time series data is obtained by the least squares method, the maximum likelihood method, or the like, and predicts how many seconds remain until the difference becomes equal to or greater than a threshold value and the semiconductor manufacturing device 1 becomes abnormal based on the approximation function. The information processing device 30 may predict the time when the abnormal state occurs using a learning model such as seq2seq, LSTM, or Transformer. In this case, time series data for a plurality of periods is used as the training data for generating the learning model.

[0066] FIG. 10 is a schematic diagram showing an example of a screen displayed by the information processing device according to the second embodiment. When the information processing device 30 according to the present embodiment determines that the semiconductor manufacturing device 1 is not in an abnormal state, the difference between the measured pressure value and the estimated pressure value becomes equal to or exceeds a threshold value, and displays a predicted time until the semiconductor manufacturing device 1 becomes in an abnormal state. As shown in FIG. 10, a predicted transition of the graph showing the difference may be displayed by a dashed line. When the information processing device 30 determines that the semiconductor manufacturing device 1 is in an abnormal state, it displays a warning message notifying the user that the device is in an abnormal state, as in the first embodiment (see FIG. 6). When the amount of change in the difference and the transition of the difference is small and the time at which the difference will become equal to or exceeds the threshold value cannot be predicted, the predicted time is not displayed.

[0067] 11 is a flowchart showing the procedure of the process performed by the information processing device according to the second embodiment. Steps S21 to S30 are the same as steps S11 to S20 in the first embodiment. If the difference between the actual pressure value and the estimated pressure value is less than the threshold (S28: NO), the information processing device 30 judges whether or not it is possible to predict the time until the semiconductor manufacturing device 1 becomes abnormal (S31). If it is possible to predict the time (S31: YES), it predicts the time until the difference between the actual pressure value and the estimated pressure value becomes equal to or greater than the threshold and the semiconductor manufacturing device 1 becomes abnormal (S32), displays the predicted time on the display unit 34 (S33), and returns the process to S22. If it is impossible to predict the time (S31: NO), the information processing device 30 returns the process to S22.

[0068] According to the program and the processing by the information processing device 30 of this embodiment, when there is a sign that the semiconductor manufacturing equipment 1 will enter an abnormal state, it is possible to predict the time when the abnormal state will occur. This enables the user of the semiconductor manufacturing equipment 1 to prepare for or avoid the abnormal state.

[0069] (Embodiment 3) The server device 5 according to the third embodiment generates a second learning model by fine-tuning a learning model 32b (first learning model) that is learned by first training data including a pressure value (first pressure value) acquired from a vacuum gauge 76 of a semiconductor manufacturing device 1 (first semiconductor manufacturing device). When the semiconductor manufacturing device is installed at a usage location, the type of pump or the layout or length of the exhaust pipe is changed depending on the usage location. The information processing device according to the present embodiment uses a second learning model generated by fine-tuning the first learning model to output an estimated pressure value of the semiconductor manufacturing device in which the type of pump or the layout or length of the exhaust pipe has been changed.

[0070] FIG. 12 is a schematic diagram showing a schematic configuration of the second semiconductor manufacturing apparatus. The second learning model is used to output an estimated pressure value of the second semiconductor manufacturing apparatus 1B, which is different from the semiconductor manufacturing apparatus 1 (first semiconductor manufacturing apparatus). For example, the second semiconductor manufacturing apparatus 1B is different from the first semiconductor manufacturing apparatus 1 in the type of the first pump 74 and the second pump 75, or the layout or length of the exhaust pipe 77 as shown in FIG. 12, so that the learning model 32b may not be able to output a correct estimated pressure value. The server device 5 fine-tunes the learning model 32b based on second training data in which a plurality of process data of the second semiconductor manufacturing apparatus 1B in normal operation is used as input information, and the pressure value (second pressure value) measured by the vacuum gauge 76 of the second semiconductor manufacturing apparatus 1B when the exhaust system is normal is used as output information, thereby generating the second learning model. The second training data used in generating the second learning model is less in quantity than the training data (first training data) used in generating the learning model 32b, but the second training data may be equal to or greater than the quantity of the first training data.

[0071] 13 is a flow chart showing the procedure of generating the second learning model. The information processing device 30B (see FIG. 12) of the second semiconductor manufacturing equipment 1B acquires process data from the second semiconductor manufacturing equipment 1B (S41), and further acquires a pressure value from the vacuum gauge 76 of the second semiconductor manufacturing equipment 1B (S42). The information processing device 30B creates second training data by associating the process data with the pressure value (S43). The information processing device 30B determines whether the number of second training data is equal to or greater than a certain number required for fine tuning (S44), and if the number of second training data is less than the certain number (S44: NO), changes the conditions such as the type of gas to be flowed into the chamber 11 of the second semiconductor manufacturing equipment 1B or the flow rate of the gas (S45), and acquires the process data again (S41). If the number of second training data is equal to or greater than a certain number (S44: YES), the information processing device 30B transmits the second training data to the server device 5 (S46), causes the server device 5 to read the first learning model (S47), and generates a second learning model by fine tuning (S48).

[0072] According to the model generation method of this embodiment, it is possible to quickly provide a learning model that outputs estimated pressure values ​​for semiconductor manufacturing equipment having different pump types or exhaust pipe layouts or lengths.

[0073] (Modification) In each of the above-described embodiments, the processing unit 31 of the information processing device 30 determines whether the exhaust system is in an abnormal state by comparing the estimated pressure value with the actual pressure value, but this is not limited to the above. The storage unit 32 of the information processing device 30 may store a range of pressure values ​​of the exhaust system that are expected when the exhaust system is normal, and the processing unit 31 may determine that the exhaust system is in an abnormal state when the estimated pressure value is outside the range of pressure values ​​stored in the storage unit 32.

[0074] In each of the above-described embodiments, the processing unit 31 of the information processing device 30 determines whether the exhaust system is in an abnormal state in a manufacturing process state (during a manufacturing process) in which gas is supplied into the chamber, but this is not limited to the above. The processing unit 31 may determine whether the exhaust system is in an abnormal state in an apparatus idle state before starting a manufacturing process. In this case, the process data includes data or measured values ​​acquired from various devices or measuring instruments in the apparatus idle state.

[0075] (supplementary explanation) The following provides a supplementary explanation of each state of the apparatus. FIG. 14 is a table comparing process data and equipment states in an apparatus startup state, an apparatus idle state, and a manufacturing process state. The apparatus startup state is a state in which the pressure inside the chamber is evacuated to a predetermined reference pressure or less, and the first pump 74 and the second pump 75 are started to evacuate the pressure inside the chamber from atmospheric pressure to the reference pressure or less. The heater is started to increase the temperature from room temperature to a preset temperature. The chiller (lower electrode coolant) that cools the back surface of the substrate S is started to increase or decrease the temperature from room temperature to a preset temperature. The gas valve 71a is closed. The APC opening changes from a closed state of 0% opening to a maximum opening of 100%. The pressure inside the chamber is evacuated from atmospheric pressure to the reference pressure or less and changes. The plasma source is stopped.

[0076] In the idle state of the device, the first pump 74 and the second pump 75 are operating, and the heater and chiller are controlled to be within a preset temperature range. The gas valve 71a is closed, and the APC opening is fixed to the maximum opening (100%). The pressure inside the chamber is in a high vacuum state below the reference pressure, and the plasma source is stopped.

[0077] In the manufacturing process state, the first pump 74 and the second pump 75 are operating, and the heater and chiller are controlled to be within a preset temperature range. The gas valve is opened. The MFC controls the gas flow rate in the chamber to be the flow rate set in the process recipe indicating the substrate processing conditions. The APC opening is controlled and fluctuates so that the pressure in the chamber becomes the pressure set in the process recipe. The pressure in the chamber becomes the pressure set in the process recipe. The plasma source applies high frequency power set in the process recipe.

[0078] 14, the process data fluctuates little in the apparatus idle state. Therefore, the processing unit 31 of the information processing device 30 can determine with high accuracy whether the exhaust system is in an abnormal state by a method of determining whether the estimated pressure value is within the range of the pressure value stored in the storage unit 32. Note that the processing unit 31 may determine that the exhaust system is in an abnormal state when the estimated pressure value is outside the range of the pressure values ​​stored in the storage unit 32 even in the manufacturing process state.

[0079] The following provides a supplementary explanation of the APC opening. Figure 15 is a graph showing an example of fluctuations in the chamber pressure and the APC opening. When the pressure in the chamber falls below a predetermined reference pressure value after performing evacuation to reduce the pressure in the chamber during equipment startup, the semiconductor manufacturing equipment 1 enters an equipment idle state. After that, an instruction to start a semiconductor manufacturing process is input, and the semiconductor manufacturing equipment 1 changes the APC opening to increase the pressure in the chamber to a pressure (set pressure) set in the process recipe, entering a manufacturing process state.

[0080] The horizontal axis of the graph shown in Fig. 15 indicates the elapsed time (seconds) from when a command to start the semiconductor manufacturing process was input, and the vertical axis indicates the APC opening (%) or the pressure inside the chamber (Pa). In the graph of Fig. 15, the change in the pressure inside the chamber is indicated by a solid line, and the change in the APC opening is indicated by a dashed line.

[0081] In the equipment startup state, the APC opening is controlled to be 0% to 100%, and the pressure in the chamber is reduced from atmospheric pressure to the reference pressure or less, and the equipment is in an idle state. When the equipment idle state is shifted to the manufacturing process state, the semiconductor manufacturing equipment 1 reduces the APC opening to, for example, about 5% as shown in FIG. 15, and keeps the pressure in the chamber at the set pressure. In the example shown in FIG. 15, the set pressure is 4 Pa. Also, as shown in the vicinity of 10 seconds to 11 seconds in the graph shown in FIG. 15, the pressure in the chamber may vary from the set pressure. The pressure in the chamber varies, for example, when the plasma source starts applying high-frequency power. When the pressure in the chamber varies, the semiconductor manufacturing equipment 1 controls the APC so that the pressure in the chamber becomes the designated pressure, and varies the APC opening. Therefore, the APC opening included in the process data input to the learning model is variable data.

[0082] The embodiments disclosed herein are illustrative in all respects and should not be considered as limiting. The technical features described in each embodiment can be combined with each other, and the scope of the present invention is intended to include all modifications within the scope of the claims and equivalents to the scope of the claims. In addition, the independent claims and dependent claims described in the claims can be combined with each other in any combination regardless of the reference format. Furthermore, the claims use a format in which a claim references two or more other claims (multiple claim format), but this is not limited to this. A multiple claim (multi-multi claim) that references at least one multiple claim may be described. [Explanation of symbols]

[0083] 1 semiconductor manufacturing equipment 5 server device 11 chamber 13 sample stage 30 information processing device 31 processing unit 31a process data acquisition unit 31b actual pressure value acquisition unit 31c estimated pressure value output unit 31d abnormality determination unit 31e display processing unit 31f time series data creation unit 31g data recording unit 31h prediction unit 32 memory unit 32a program 32b learning model 32c database 33 communication unit 34 display unit 35 operation unit 36 ​​card slot 51 processing unit 51a learning processing unit 52 memory unit 53 communication unit 54 card slot 71 MFC 71a gas valve 72 vacuum gauge 73 APC 74 first pump 75 second pump 76 vacuum gauge 77 exhaust pipe 78 pressure / flow meter 99 recording medium

Claims

1. In a semiconductor manufacturing equipment idle state before starting a manufacturing process for supplying gas into the equipment, process data is acquired that includes at least two of the type and flow rate of gas supplied into the equipment, the pressure inside the chamber, or the valve opening of an automatic pressure control device; inputting the process data into a learning model that has been trained to output an estimated pressure value in the semiconductor manufacturing equipment when the process data is input, and outputting the estimated pressure value; Determine whether an abnormal state exists based on the estimated pressure value. A program that causes a computer to perform a process.

2. During a manufacturing process in which gas is supplied into a semiconductor manufacturing device, process data is acquired that includes at least two of the type and flow rate of the gas supplied into the semiconductor manufacturing device, the pressure inside the chamber, or the valve opening of an automatic pressure control device; inputting the process data into a learning model that has been trained to output an estimated pressure value in the semiconductor manufacturing equipment when the process data is input, and outputting the estimated pressure value; Determine whether an abnormal state exists based on the estimated pressure value. A program that causes a computer to perform a process.

3. Acquire process data including at least two of the type and flow rate of gas supplied into semiconductor manufacturing equipment, the pressure inside the chamber, or the valve opening of an automatic pressure control device; inputting the process data into a learning model that has been trained to output an estimated pressure value in an exhaust pipe of the semiconductor manufacturing equipment when the process data is input, and outputting the estimated pressure value in the exhaust pipe of the semiconductor manufacturing equipment; Determine whether an abnormal state exists based on the estimated pressure value. A program that causes a computer to perform a process.

4. Acquire an actual pressure value measured by a vacuum gauge in the semiconductor manufacturing equipment; Determine whether an abnormal state exists based on the estimated pressure value and the actually measured pressure value. The program according to claim 1.

5. The process data includes an operating state of an exhaust pump.

4. The program according to claim 1.

6. If the difference between the actually measured pressure value and the estimated pressure value is equal to or greater than a predetermined value, it is determined that an abnormal state exists. The program according to claim 4.

7. In an idle state of semiconductor manufacturing equipment before starting a manufacturing process for supplying gas into the equipment, process data is acquired that includes at least two of the type and flow rate of gas supplied into the semiconductor manufacturing equipment, the pressure inside the chamber, or the valve opening of an automatic pressure control device; inputting the process data into a learning model that has been trained to output an estimated pressure value in the semiconductor manufacturing equipment when the process data is input, and outputting the estimated pressure value; Determine whether an abnormal state exists based on the estimated pressure value. Information processing methods.

8. During a manufacturing process in which gas is supplied into a semiconductor manufacturing device, process data is acquired that includes at least two of the type and flow rate of the gas supplied into the semiconductor manufacturing device, the pressure inside the chamber, or the valve opening of an automatic pressure control device; inputting the process data into a learning model that has been trained to output an estimated pressure value in the semiconductor manufacturing equipment when the process data is input, and outputting the estimated pressure value; Determine whether an abnormal state exists based on the estimated pressure value. Information processing methods.

9. Acquire process data including at least two of the type and flow rate of gas supplied into semiconductor manufacturing equipment, the pressure inside the chamber, or the valve opening of an automatic pressure control device; inputting the process data into a learning model that has been trained to output an estimated pressure value in an exhaust pipe of a semiconductor manufacturing equipment when the process data is input, and outputting the estimated pressure value in the exhaust pipe of the semiconductor manufacturing equipment; Determine whether an abnormal state exists based on the estimated pressure value. Information processing methods.

10. A process data acquisition unit that acquires process data including at least two of the type and flow rate of gas supplied into the semiconductor manufacturing equipment, the pressure inside the chamber, or the valve opening of an automatic pressure control device, while the equipment is in an idle state before starting a manufacturing process of supplying gas into the semiconductor manufacturing equipment; an estimated pressure value output unit that inputs the process data into a learning model that has been trained to output an estimated pressure value in the semiconductor manufacturing equipment when the process data is input, and outputs the estimated pressure value; an abnormality determination unit that determines whether an abnormal state exists based on the estimated pressure value; An information processing device comprising:

11. A process data acquisition unit that acquires process data including at least two of the type and flow rate of gas supplied into the semiconductor manufacturing equipment, the pressure inside the chamber, or the valve opening of an automatic pressure control device during a manufacturing process of supplying gas into the semiconductor manufacturing equipment; an estimated pressure value output unit that inputs the process data into a learning model that has been trained to output an estimated pressure value in the semiconductor manufacturing equipment when the process data is input, and outputs the estimated pressure value; an abnormality determination unit that determines whether an abnormal state exists based on the estimated pressure value; An information processing device comprising:

12. A process data acquisition unit that acquires process data including at least two of the type and flow rate of gas supplied into semiconductor manufacturing equipment, the pressure inside the chamber, or the valve opening of an automatic pressure control device; an estimated pressure value output unit that inputs the process data into a learning model that has been trained to output an estimated pressure value in an exhaust pipe of a semiconductor manufacturing equipment when the process data is input, and outputs the estimated pressure value in the exhaust pipe of the semiconductor manufacturing equipment; an abnormality determination unit that determines whether an abnormal state exists based on the estimated pressure value; An information processing device comprising:

13. Acquire the pressure value of the vacuum gauge in the semiconductor manufacturing equipment, In an idle state of the semiconductor manufacturing equipment before starting a manufacturing process for supplying gas into the semiconductor manufacturing equipment, training data is acquired, the training data including process data including at least two of the type and flow rate of gas supplied to the semiconductor manufacturing equipment, the pressure inside the chamber, and the valve opening of an automatic pressure control device, and the acquired pressure value; Based on the acquired training data, a learning model is generated that outputs an estimated pressure value when process data including at least two of the type and flow rate of gas supplied to the semiconductor manufacturing equipment, the pressure inside the chamber, or the valve opening of the automatic pressure control device is input. How the model is generated.

Images