Abnormal discharge determination method, abnormal discharge determination system, and substrate processing apparatus

Abstract

To provide a technique of determining the presence / absence of the occurrence of abnormal discharge in a plasma processing chamber with a machine-learned determination model.SOLUTION: An abnormal discharge determination method for determining the presence / absence of the occurrence of abnormal discharge in a plasma processing chamber with a machine-learned determination model comprises the steps of: acquiring a difference image between a first image and a second image obtained by imaging the inside of the plasma processing chamber at different timings; converting the difference image into a binary image in white and black; extracting a contour from a white area of the binary image; creating a plurality of feature amounts input to a determination model based on the contour; and determining the presence / absence of the occurrence of abnormal discharge in the plasma processing chamber based on the output of the determination model to which the plurality of feature amounts are input.SELECTED DRAWING: Figure 3

Description

[Technical Field]

[0001] The present disclosure relates to an abnormal discharge determination method, an abnormal discharge determination system, and a substrate processing apparatus. [Background technology]

[0002] For example, in a substrate processing apparatus such as an FPD (Flat Panel Display) manufacturing apparatus that uses high density plasma CVD (Chemical Vapor Deposition), an abnormal discharge may occur inside the plasma processing chamber. Patent Document 1 describes a method for detecting an abnormal discharge in a plasma processing apparatus. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Application Laid-Open No. 2001-319922 Summary of the Invention [Problem to be solved by the invention]

[0004] The present disclosure provides a technique for determining whether or not an abnormal discharge has occurred inside a plasma processing chamber using a machine-learned determination model. [Means for solving the problem]

[0005] One aspect of the present disclosure is an abnormal discharge determination method for determining whether or not an abnormal discharge has occurred inside a plasma processing chamber using a machine-learned determination model, the method comprising: (a) obtaining a difference image between a first image and a second image taken inside the plasma processing chamber at different times; (b) converting the difference image into a binary image of white and black; (c) extracting contours from white areas of the binary image; (d) creating a plurality of feature amounts to be input into the determination model based on the contours; and (e) determining whether or not an abnormal discharge has occurred inside the plasma processing chamber based on the output of the determination model to which the plurality of feature amounts have been input. [Effects of the Invention]

[0006] According to the present disclosure, it is possible to provide a technique for determining whether or not an abnormal discharge has occurred inside a plasma processing chamber using a machine-learned determination model. [Brief explanation of the drawings]

[0007] [Figure 1] 1 is an explanatory diagram of an example of an abnormal discharge determination system according to an embodiment of the present invention; [Figure 2] FIG. 2 is a diagram illustrating a hardware configuration of an example of a computer. [Figure 3] 1 is a functional configuration diagram of an example of an abnormal discharge determination system according to an embodiment of the present invention. [Figure 4] 10 is a flowchart of an example of a teacher data creation process. [Figure 5] FIG. 10 is an explanatory diagram of an example of a process for generating a difference image. [Figure 6] 10 is a flowchart illustrating an example of a process for creating a plurality of feature amounts. [Figure 7] FIG. 10 is an explanatory diagram of an example of a process for converting a differential image into a binary image. [Figure 8] FIG. 10 is an explanatory diagram of an example of processing for a binary image. [Figure 9] FIG. 10 is an explanatory diagram of an example of processing for extracting a contour from a white area. [Figure 10]FIG. 10 is an explanatory diagram of an example of an original image before binarization to which contours extracted from white areas of the binarized image are applied. [Figure 11] 10 is a flowchart illustrating an example of a process for determining the occurrence of abnormal discharge. [Figure 12] 1 is a functional configuration diagram of an example of an abnormal discharge determination system according to an embodiment of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0008] DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS In the following, embodiments of the present invention will be described with reference to the accompanying drawings. In this specification and drawings, illustrations and descriptions of parts that are not necessary for the description of the present embodiment will be omitted as appropriate.

[0009] Fig. 1 is an explanatory diagram of an example of an abnormal discharge determination system according to this embodiment. The abnormal discharge determination system in Fig. 1 has a configuration in which a substrate processing apparatus 1 such as an FPD manufacturing apparatus and a server apparatus 20 are communicably connected via a communication path.

[0010] The substrate processing apparatus 1 includes a plasma processing chamber 10. Inside the plasma processing chamber 10, a plasma-based process is performed on a glass substrate or the like for manufacturing an FPD. The plasma-based process is, for example, an etching process or a film formation process. The abnormal discharge determination system according to this embodiment determines whether or not an abnormal discharge has occurred inside the plasma processing chamber 10 during plasma processing, as described below.

[0011] A window 18 is provided on the sidewall of plasma processing chamber 10. Window 18 is made of a highly light-transmitting material (e.g., quartz glass) so that images can be taken from outside plasma processing chamber 10. Substrate processing apparatus 1 also includes an imaging unit 14, such as a camera, that takes images of the inside of plasma processing chamber 10 through window 18. Imaging unit 14 makes it possible to take images of the inside of plasma processing chamber 10 while plasma is being generated.

[0012] The imaging unit 14 captures images of the interior of the plasma processing chamber 10 during plasma processing and outputs the captured image data. The captured image data output from the imaging unit 14 is recorded in at least one of the recording unit 16 of the substrate processing apparatus 1 and the recording unit 24 of the server device 20. The captured image data output from the imaging unit 14 may be a moving image (hereinafter simply referred to as a moving image) or a still image (hereinafter simply referred to as an image).

[0013] The control unit 12 of the substrate processing apparatus 1 or the control unit 22 of the server apparatus 20 processes the imaging data output from the imaging unit 14 to create a determination model for determining whether or not abnormal discharge has occurred inside the plasma processing chamber 10, as described below. The control unit 12 of the substrate processing apparatus 1 or the control unit 22 of the server apparatus 20 also determines whether or not abnormal discharge has occurred inside the plasma processing chamber 10 by using a machine-learned determination model, which will be described below. The distribution unit 26 notifies the substrate processing apparatus 1 of information on whether or not abnormal discharge has occurred inside the plasma processing chamber 10, etc., from the server apparatus 20.

[0014] The control unit 12 of the substrate processing apparatus 1 is a computer that controls the operation of the entire apparatus. A program that controls the operation of the entire apparatus is stored in the recording unit 16 of the control unit 12. By executing the program, the control unit 12 sends control signals to each unit for performing substrate processing in the substrate processing apparatus 1, thereby controlling the operation of each unit. The program is installed in the control unit 12 from a storage medium such as a hard disk, compact disk, magneto-optical disk, memory card, or flexible disk.

[0015] The control unit 12 of the substrate processing apparatus 1 and the control unit 22 of the server apparatus 20 display a screen for receiving information input from an operator and a screen for outputting information such as results to an operator. The control unit 12 of the substrate processing apparatus 1 may be built into the substrate processing apparatus 1 as shown in Fig. 1, or may be provided outside the substrate processing apparatus 1 and connected to the substrate processing apparatus 1 via a communication path so as to be able to communicate with the substrate processing apparatus 1.

[0016] The communication path may be a wired or wireless communication path, and may be a path for exchanging various signals inside and outside the computer. The communication path may utilize a network such as a local area network (LAN).

[0017] The control unit 12 of the substrate processing apparatus 1 is realized by, for example, a computer 500 having the hardware configuration shown in Fig. 2. The server device 20 is also realized by, for example, a computer 500 having the hardware configuration shown in Fig. 2. Fig. 2 is a diagram showing the hardware configuration of an example computer.

[0018] 2 includes an input device 501, an output device 502, an external I / F (interface) 503, a RAM (random access memory) 504, a ROM (read only memory) 505, a CPU (central processing unit) 506, a communication I / F 507, and an HDD (hard disk drive) 508, all of which are interconnected by a bus B. The input device 501 and the output device 502 may be connected and used when necessary.

[0019] The input device 501 is a keyboard, mouse, touch panel, etc., and is used by an operator or the like to input various operation signals. The output device 502 is a display or the like, and displays the results of processing by the computer 500. The communication I / F 507 is an interface that connects the computer 500 to a network or the like. The HDD 508 is an example of a non-volatile storage unit that stores programs and data.

[0020] The external I / F 503 is an interface with an external device. The computer 500 can read and / or write data from and to a recording medium 503a such as an SD (Secure Digital) memory card via the external I / F 503. The ROM 505 is an example of a non-volatile semiconductor memory (storage unit) that stores programs and data. The RAM 504 is an example of a volatile semiconductor memory (storage unit) that temporarily stores programs and data.

[0021] The CPU 506 is a computing device that controls the entire computer 500 and realizes its functions by reading programs and data from storage units such as the ROM 505 and HDD 508 onto the RAM 504 and executing the processes.

[0022] The control unit 12 of the substrate processing apparatus 1 and the control unit 22 of the server apparatus 20 shown in Fig. 1 can realize the various functions of Fig. 3 or 12 by causing a computer 500 having the hardware configuration shown in Fig. 2 to execute processes in accordance with programs. The functional configuration of Fig. 3 is an example in which the teacher data creation process, judgment model creation process, and abnormal discharge occurrence determination process according to this embodiment are performed by the server apparatus 20. The functional configuration of Fig. 12 is an example in which the server apparatus 20 performs the teacher data creation process and judgment model creation process, and the substrate processing apparatus 1 performs the abnormal discharge occurrence determination process.

[0023] Here, the functional configuration of Fig. 3 will be described, and the functional configuration of Fig. 12 will be described later. Fig. 3 is a functional configuration diagram of an example of an abnormal discharge determination system according to this embodiment. The server device 20 shown in Fig. 3 has an imaging data receiving unit 50, a teacher data creating unit 52, a determination model creating unit 54, an abnormal discharge occurrence determination unit 56, an abnormal discharge occurrence notifying unit 58, an operation accepting unit 60, and an output control unit 62.

[0024] The recording unit 24 of the server device 20 records the photographed data 70, the teacher data 72, and the determination model 74. The recording unit 24 may be realized by the HDD 508, or may be realized by a recording device connected so as to be able to communicate via a network or the like.

[0025] The photographed data 70 is a video or an image captured by the photographing unit 14 of the substrate processing apparatus 1 of the interior of the plasma processing chamber 10. The training data 72 is training data used for supervised machine learning of a judgment model for determining whether or not abnormal discharge has occurred inside the plasma processing chamber 10. The judgment model 74 is a judgment model for determining whether or not abnormal discharge has occurred inside the plasma processing chamber 10. Multiple types of judgment models 74 may be prepared using existing methods such as gradient boosting decision trees or random forests, and a highly accurate judgment model may be adopted.

[0026] The imaging data receiving unit 50 receives imaging data 70 of the interior of the plasma processing chamber 10 captured by the imaging unit 14 of the substrate processing apparatus 1, and records the data in the recording unit 24. The training data creating unit 52 processes the imaging data 70 to create training data 72 to be used for supervised machine learning of a determination model 74. The determination model creating unit 54 uses the training data 72 to generate a machine-learned determination model 74.

[0027] The abnormal discharge occurrence determination unit 56 uses the machine-learned determination model 74 to determine whether or not abnormal discharge has occurred inside the plasma processing chamber 10 from the imaging data 70 of the interior of the plasma processing chamber 10 captured by the imaging unit 14 of the substrate processing apparatus 1. The abnormal discharge occurrence notification unit 58 performs processing to notify the abnormal discharge occurrence determination unit 56 of the presence or absence of abnormal discharge occurring inside the plasma processing chamber 10. The processing to notify the presence or absence of abnormal discharge occurring inside the plasma processing chamber 10 may be, for example, a notification to a preset contact address (e.g., an email address) of the operator, or a notification to the control unit 12 of the substrate processing apparatus 1.

[0028] An operation reception unit 60 receives various operations from an operator via the input device 501. An output control unit 62 controls the output of various screens and the like on the output device 502, and controls notifications of whether or not abnormal discharge has occurred inside the plasma processing chamber 10.

[0029] The teacher data creation unit 52 of the server device 20 according to this embodiment performs teacher data creation processing, for example, according to the procedure shown in Fig. 4. Fig. 4 is a flowchart of an example of the teacher data creation processing.

[0030] In step S100, the teacher data creation unit 52 classifies the images obtained from the imaging data 70 into images of the interior of the plasma processing chamber 10 in a state where no abnormal discharge is occurring (before the occurrence of abnormal discharge) and images of the interior of the plasma processing chamber 10 in a state where an abnormal discharge has occurred (after the occurrence of abnormal discharge). For convenience, even when the occurrence of abnormal discharge is not recorded in the imaging data 70, the "state where no abnormal discharge is occurring" will also be referred to as "before the occurrence of abnormal discharge." Note that if there is a data set in which images of the interior of the plasma processing chamber 10 taken before the occurrence of abnormal discharge and images taken after the occurrence of abnormal discharge are associated with each other, the processing of step S100 may be omitted.

[0031] In step S102, the teacher data creation unit 52 generates a difference image with abnormal discharge occurring and a difference image without abnormal discharge occurring, for example, as shown in Fig. 5. Fig. 5 is an explanatory diagram of an example of the process for generating the difference image. The difference image is an image composed of the difference in brightness value for each pixel of the two images.

[0032] For example, when generating a differential image in the presence of abnormal discharge, the teacher data creation unit 52 generates a differential image between an image of the interior of the plasma processing chamber 10 taken before the occurrence of abnormal discharge and an image taken after the occurrence of abnormal discharge.

[0033] Furthermore, when generating a difference image without abnormal discharge, the teacher data creation unit 52 generates a difference image between an image of the interior of the plasma processing chamber 10 taken before the occurrence of abnormal discharge and another image taken before the occurrence of abnormal discharge. The difference image is an image for extracting an area where there is a change in color.

[0034] In step S104, the training data creation unit 52 creates a plurality of feature amounts for each differential image in accordance with the procedure shown in Fig. 6. Fig. 6 is a flowchart showing an example of processing for creating a plurality of feature amounts.

[0035] In step S200, the training data creation unit 52 converts the difference image into a binary image of white and black using the difference image as the original image, as shown in Fig. 7. Fig. 7 is an explanatory diagram of an example of the process of converting the difference image into a binary image.

[0036] The teacher data creation unit 52 grayscales the differential image and converts it into a grayscale image (0 to 255) as shown in Fig. 7(A), for example. Also, as shown in Fig. 7(B), the teacher data creation unit 52 converts the grayscale image (0 to 255) of Fig. 7(A) into a binary image of black (0) and white (255).

[0037] In step S202, the training data creation unit 52 performs erosion and expansion processes to remove noise from the binary image shown in Fig. 7(B), converting it into a noise-removed binary image shown in Fig. 8(A). Fig. 8 is an explanatory diagram of an example of processing on a binary image.

[0038] In step S204, the training data creation unit 52 performs a synthesis process on the noise-removed binary image shown in Figure 8(A) to expand the discrete white areas of the binary image so that they become continuous, and converts it into the synthesized binary image shown in Figure 8(B).

[0039] In step S206, the training data creation unit 52 performs contour extraction processing to extract contours from white regions on the combined binary image shown in Fig. 8(B), and converts it into a binary image in which contours have been extracted from white regions, for example, as shown in Fig. 9. Fig. 9 is an explanatory diagram of an example of processing to extract contours from white regions.

[0040] In step S208, the training data creation unit 52 applies the contour extracted from the white region shown in Fig. 9 to the original image before binarization, and recognizes the entire original image 1000, a contour 1002, and an area inside the contour 1004, as shown in Fig. 10, for example. The training data creation unit 52 calculates the HSV average value of the area inside the contour 1004 in the original image, the HSV standard deviation of the area inside the contour 1004, the area surrounded by the contour 1002, and the length of the contour 1002. HSV is a color model that expresses color using hue, saturation, and lightness.

[0041] The HSV average value includes a hue average value, a saturation average value, and a lightness average value. The HSV standard deviation includes a hue standard deviation, a saturation standard deviation, and a lightness standard deviation. In step S210, the training data creation unit 52 calculates the HSV average value of the entire 1000 original images before binarization and the HSV standard deviation of the entire 1000 original images before binarization.

[0042] In step S212, the training data creation unit 52 outputs, as multiple feature quantities, the calculated HSV average value within the contour 1004, the HSV standard deviation within the contour 1004, the area of ​​the region surrounded by the contour 1002, the length of the contour 1002, the HSV average value of the entire 1000 of the original image before binarization, and the HSV standard deviation of the entire 1000 of the original image before binarization.

[0043] Note that the calculated feature amounts including the average HSV value within the contour 1004, the HSV standard deviation within the contour 1004, the area of ​​the region surrounded by the contour 1002, the length of the contour 1002, the average HSV value of the entire 1000 of the original image before binarization, and the HSV standard deviation of the entire 1000 of the original image before binarization are just examples, and for example, the average HSV value of the entire 1000 of the original image before binarization and the HSV standard deviation of the entire 1000 of the original image before binarization may be omitted.

[0044] Returning to step S106 in FIG. 4, the teacher data creating unit 52 creates teacher data 72 in which the plurality of feature quantities created in step S104 are set as explanatory variables and the occurrence or non-occurrence of abnormal discharge is set as a response variable.

[0045] The judgment model creation unit 54 of the server device 20 according to this embodiment uses the teacher data 72 created in step S106 to perform supervised machine learning of the differential image with abnormal discharge and the differential image without abnormal discharge, and generates a machine-learned judgment model 74 for determining whether or not abnormal discharge has occurred inside the plasma processing chamber 10.

[0046] The abnormal discharge occurrence determination unit 56 of the server device 20 according to this embodiment performs the abnormal discharge occurrence determination process, for example, according to the procedure shown in Fig. 11. Fig. 11 is a flowchart of an example of the abnormal discharge occurrence determination process.

[0047] In step S300, the abnormal discharge occurrence determination unit 56 generates a difference image between an image of frame N (first image) and an image of frame N+1 (second image) from the photographic data 70, for example, as shown in Fig. 5. Note that the two image frames from which the difference image is generated do not necessarily have to be adjacent, and for example, a difference image may be generated between an image of frame N and an image of frame N+m (m is an integer of 2 or greater).

[0048] In step S302, the abnormal discharge occurrence determination unit 56 creates a plurality of feature amounts of the difference image in accordance with the procedure shown in FIG.

[0049] In step S200, the abnormal discharge occurrence determination unit 56 converts the difference image into a binary image of white and black, for example, as shown and described in Fig. 7. In step S202, the abnormal discharge occurrence determination unit 56 performs contraction and expansion processes to remove noise from the binary image shown in Fig. 7(B), and converts it into a noise-removed binary image shown in Fig. 8(A).

[0050] In step S204, the abnormal discharge occurrence determination unit 56 performs a synthesis process on the noise-removed binary image shown in Figure 8(A) to expand the discrete white areas of the binary image so that they become continuous, and converts it into the synthesized binary image shown in Figure 8(B).

[0051] In step S206, the abnormal discharge occurrence determination unit 56 performs a contour extraction process on the synthesized binary image shown in Figure 8(B) to extract contours from the white area, and converts it into a binary image in which contours are extracted from the white area, for example, as shown in Figure 9.

[0052] In step S208, the abnormal discharge occurrence determination unit 56 applies the contour extracted from the white region shown in Fig. 9 to the original image before binarization, and recognizes the entire original image 1000, a contour 1002, and an area inside the contour 1004, for example, as shown and described in Fig. 10. The abnormal discharge occurrence determination unit 56 calculates the HSV average value of the area inside the contour 1004 in the original image, the HSV standard deviation of the area inside the contour 1004, the area of ​​the area surrounded by the contour 1002, and the length of the contour 1002. In step S210, the abnormal discharge occurrence determination unit 56 calculates the HSV average value of the entire original image 1000 before binarization and the HSV standard deviation of the entire original image 1000 before binarization.

[0053] In step S212, the abnormal discharge occurrence determination unit 56 outputs, as multiple feature quantities, the calculated HSV average value within the contour 1004, the HSV standard deviation within the contour 1004, the area of ​​the contour 1002, the length of the area surrounded by the contour 1002, the HSV average value of the entire original image 1000 before binarization, and the HSV standard deviation of the entire original image 1000 before binarization. Note that the types of multiple feature quantities created by the abnormal discharge occurrence determination unit 56 in the procedure shown in Fig. 6 are the same as the types of multiple feature quantities created by the above-mentioned training data creation unit 52 in the procedure shown in Fig. 6.

[0054] Returning to step S304 in Figure 11, the abnormal discharge occurrence determination unit 56 inputs the multiple feature amounts generated in step S302 into the machine-learned determination model 74, and determines whether or not abnormal discharge has occurred based on the output from the machine-learned determination model 74.

[0055] If it is determined in step S306 that an abnormal discharge has occurred, the abnormal discharge occurrence notification unit 58 performs processing in step S308 to notify the abnormal discharge occurrence determined by the abnormal discharge occurrence determination unit 56 inside the plasma processing chamber 10. Note that if it is determined in step S306 that an abnormal discharge has not occurred, the abnormal discharge occurrence notification unit 58 skips the processing in step S308.

[0056] The abnormal discharge occurrence determination process shown in Fig. 11 may be performed before maintenance or at predetermined intervals by recording image data 70 of the interior of plasma processing chamber 10 captured by imaging unit 14 in recording unit 24. The abnormal discharge occurrence determination process shown in Fig. 11 may also be performed after plasma-based processing in plasma processing chamber 10. The abnormal discharge occurrence determination process shown in Fig. 11 may also be performed in real time on image data 70 of the interior of plasma processing chamber 10 captured by imaging unit 14.

[0057] Fig. 12 is a functional configuration diagram of an example of the abnormal discharge determination system according to this embodiment. The functional configuration in Fig. 12 is an example in which the server device 20 performs the teacher data creation process and the determination model creation process, and the substrate processing apparatus 1 performs the abnormal discharge occurrence determination process.

[0058] 12 includes an imaging data receiving unit 50, a teacher data creating unit 52, a judgment model creating unit 54, an operation accepting unit 60, and an output control unit 62. In addition, the recording unit 24 of the server device 20 records imaging data 70, teacher data 72, and a judgment model 74.

[0059] 12 includes an imaging data transmitting unit 80, a judgment model receiving unit 82, an abnormal discharge occurrence determining unit 84, and an abnormal discharge occurrence notifying unit 86. In addition, the recording unit 16 of the substrate processing apparatus 1 records imaging data 90 and a machine-learned judgment model 92.

[0060] The server device 20 shown in Figure 12 performs processing similar to that of the server device 20 with the functional configuration shown in Figure 3, creating training data 72 from the shooting data 70 and using the training data 72 to generate a machine-learned judgment model 74.

[0061] The imaging data transmission unit 80 of the substrate processing apparatus 1 transmits imaging data 70 of the interior of the plasma processing chamber 10 captured by the imaging unit 14 to the server device 20. The judgment model receiving unit 82 receives the machine-learned judgment model 74 from the server device 20 and records it in the recording unit 16 as a machine-learned judgment model 92. The recording unit 16 also records imaging data 90 of the interior of the plasma processing chamber 10 captured by the imaging unit 14.

[0062] The abnormal discharge occurrence determination unit 84 determines whether or not an abnormal discharge has occurred inside the plasma processing chamber 10 by using the machine-learned determination model 92 and processing similar to that of the abnormal discharge occurrence determination unit 56 described above, from the imaging data 90 of the interior of the plasma processing chamber 10 captured by the imaging unit 14. The abnormal discharge occurrence notification unit 86 performs processing for notifying the operator of the presence or absence of an abnormal discharge occurring inside the plasma processing chamber 10 determined by the abnormal discharge occurrence determination unit 84. The processing for notifying the operator of the presence or absence of an abnormal discharge occurring inside the plasma processing chamber 10 may be, for example, a notification to a preset contact address (e.g., an email address), or may be an alarm output from the substrate processing apparatus 1, for example.

[0063] 3 and 12 are merely examples, and for example, the functional configuration of the server device 20 in FIG. 12 may be integrated into the substrate processing apparatus 1, and the server device 20 may be omitted.

[0064] As described above, this embodiment provides a technique for determining whether or not abnormal discharge has occurred inside plasma processing chamber 10 from an image of the inside of plasma processing chamber 10 captured by imaging unit 14. Therefore, this embodiment allows an operator to notice the occurrence of abnormal discharge without opening the lid of plasma processing chamber 10, and allows parts that need to be replaced to be prepared in advance, thereby shortening the time required for restoration through maintenance work.

[0065] Although the preferred embodiments of the present invention have been described above in detail, the present invention is not limited to the above-described embodiments, and various modifications and substitutions can be made to the above-described embodiments without departing from the scope of the present invention. For example, in this embodiment, the presence or absence of abnormal discharge is determined from an image captured inside the plasma processing chamber 10 during plasma processing, but the present invention is not limited to this. Even if plasma is not being generated, the present invention can be applied to cases where abnormal discharge occurs due to the application of voltage or the supply of high-frequency power.

[0066] 1 shows an example in which one server device 20 corresponds to one substrate processing apparatus 1, but one server device 20 may correspond to a plurality of substrate processing apparatuses 1. The functions of the server device 20 may be distributed among a plurality of computers, or may be provided in a cloud computer. [Explanation of symbols]

[0067] 1. Substrate processing equipment 10 Plasma Processing Chamber 12, 22 Control section 14 Photography Department 20 Server device 52 Training Data Creation Department 54 Decision Model Creation Department 56 Abnormal discharge occurrence determination unit

Claims

1. An abnormal discharge determination method for determining whether or not an abnormal discharge has occurred inside a plasma processing chamber using a machine-learned determination model, comprising: (a) obtaining a difference image between a first image and a second image taken at different times of the interior of the plasma processing chamber; (b) converting the difference image into a black and white binary image; (c) extracting contours from white areas of the binary image; (d) creating a plurality of feature amounts to be input to the determination model based on the contour; (e) determining whether or not an abnormal discharge has occurred inside the plasma processing chamber based on an output of the determination model to which the plurality of feature quantities have been input; and An abnormal discharge determination method comprising:

2. (f) removing noise from the binary image; (g) combining the discrete white regions of the binary image from which the noise has been removed so that they are continuous; and In (c), the contour is extracted from a white area of ​​the binary image after synthesis. The abnormal discharge determination method according to claim 1.

3. (d) applying the contour extracted in (c) to the difference image to generate the plurality of feature amounts including at least an HSV average value within the contour, an HSV standard deviation within the contour, an area of ​​the region surrounded by the contour, and a length of the contour in the difference image. The abnormal discharge determination method according to claim 1 or 2.

4. The machine-learned determination model is created by supervised machine learning using training data in which the information on the occurrence or non-occurrence of abnormal discharge is used as a target variable and the plurality of feature quantities are used as explanatory variables. The abnormal discharge determination method according to any one of claims 1 to 3.

5. 5. The abnormal discharge determination method according to claim 4, wherein the machine-learned determination model is generated by supervised machine learning using training data, wherein when the plurality of feature quantities created from a difference image between an image of the interior of the plasma processing chamber taken before the abnormal discharge occurred and an image of the interior of the plasma processing chamber taken after the abnormal discharge occurred are used as explanatory variables, the dependent variable is whether the abnormal discharge has occurred, and when the plurality of feature quantities created from a difference image between an image of the interior of the plasma processing chamber taken before the abnormal discharge occurred and another image of the interior of the plasma processing chamber taken before the abnormal discharge occurred are used as explanatory variables, the dependent variable is whether the abnormal discharge has occurred.

6. An abnormal discharge determination system that determines whether or not an abnormal discharge has occurred inside a plasma processing chamber using a machine-learned determination model, A control unit; an imaging unit that images the inside of the plasma processing chamber; a recording unit that records the image of the inside of the plasma processing chamber captured by the imaging unit; and The control unit obtaining a difference image between a first image and a second image taken at different times of the interior of the plasma processing chamber; converting the difference image into a black and white binary image; extracting contours from white areas of the binary image; creating a plurality of feature amounts to be input to the determination model based on the contour; determining whether or not abnormal discharge has occurred inside the plasma processing chamber based on an output of the determination model to which the plurality of feature quantities have been input; An abnormal discharge detection system that controls

7. A substrate processing apparatus having a function of determining whether or not an abnormal discharge has occurred inside a plasma processing chamber using a machine-learned determination model, A control unit; the plasma processing chamber; an imaging unit that images the inside of the plasma processing chamber; and The control unit obtaining a difference image between a first image and a second image taken at different times of the interior of the plasma processing chamber; converting the difference image into a black and white binary image; extracting contours from white areas of the binary image; creating a plurality of feature amounts to be input to the determination model based on the contour; determining whether or not abnormal discharge has occurred inside the plasma processing chamber based on an output of the determination model to which the plurality of feature quantities have been input; A substrate processing apparatus that controls the substrate processing apparatus.

Images