Substrate processing apparatus, method for determining sign of failure, and method for learning

Abstract

To provide a substrate processing apparatus, a method for determining a sign of failure, and a method for learning capable of detecting a sign of failure of a control valve.SOLUTION: The substrate processing apparatus 100 includes: a pipe 31 through which a processing liquid passes; an open-close valve 33 for opening and closing a flow path of the pipe 31; an electric control valve 34 for adjusting the flow rate of the processing liquid passing through the flow path; a vibration sensor 51 for detecting vibration of the control valve 34; and a determination unit 55a for determining presence or absence of a sign of failure of the control valve 34 on the basis of the vibration detection result of the vibration sensor 51 while the control valve 34 is being driven.SELECTED DRAWING: Figure 4

Description

[Technical Field]

[0001] The present invention relates to a substrate processing apparatus, a method for determining a failure symptom, and a learning method. [Background technology]

[0002] Conventionally, there has been known a substrate processing apparatus that includes a pipe through which a processing liquid passes, an on-off valve that opens and closes the flow path of the pipe, an electrically operated regulating valve that regulates the flow rate of the processing liquid passing through the flow path of the pipe, and a flow meter that detects the flow rate of the processing liquid passing through the pipe (see, for example, Patent Document 1). Patent Document 1 describes a substrate processing apparatus that includes a pipe through which a processing liquid flows, an on-off valve that opens and closes the flow path of the pipe, an regulating valve that regulates the flow rate of the processing liquid flowing through the pipe, and a flow meter that detects the flow rate of the processing liquid flowing through the pipe.

[0003] In such substrate processing apparatuses, the flow rate of the processing liquid is adjusted within an error range of, for example, a few ml / min or less, and therefore the adjusting valve is feedback-controlled so that the detection result of the flow meter matches the target flow rate. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] Japanese Patent Application Publication No. 2023-46224 Summary of the Invention [Problem to be solved by the invention]

[0005] However, in conventional substrate processing apparatuses, the adjusting valve deteriorates over time due to feedback control for adjusting the flow rate. As the adjusting valve deteriorates over time, the flow rate of the processing liquid increases slowly and the flow rate varies greatly (also known as hunting). In this case, it is possible to know, for example, that an abnormality (failure) has occurred in the adjusting valve based on the detection results of the flow meter.

[0006] However, with the conventional method, it is not possible to detect an abnormality in the adjusting valve until an abnormality in the flow rate of the processing liquid flowing through the piping occurs. As a result, a flow rate abnormality may suddenly occur during substrate processing, requiring the substrate processing to be unintentionally interrupted, resulting in a problem of reduced productivity.

[0007] The present invention has been made in consideration of the above-mentioned problems, and its purpose is to provide a substrate processing apparatus, a method for determining signs of failure, and a learning method that are capable of detecting signs of failure in an adjustment valve. [Means for solving the problem]

[0008] After extensive research, the inventors of the present invention have found that vibrations occur in the adjusting valve before abnormalities occur in the flow rate of the treatment liquid passing through the piping. Specifically, they have found that, although flow rate abnormalities do not occur when vibrations occur in the adjusting valve, continued use after vibrations occur in the adjusting valve will cause the adjusting valve to break down, resulting in flow rate abnormalities.

[0009] The present invention is based on new findings by the inventors of the present application, and according to a first aspect of the present invention, a substrate processing apparatus includes a pipe, an on-off valve, an electric adjustment valve, a vibration sensor, and a determination unit. A processing liquid passes through the pipe. The on-off valve opens and closes a flow path of the pipe. The adjustment valve adjusts the flow rate of the processing liquid passing through the flow path. The vibration sensor detects vibration of the adjustment valve. The determination unit determines whether or not there is a sign of a failure of the adjustment valve based on the vibration detection result of the vibration sensor when the adjustment valve is operating.

[0010] In one embodiment, the substrate processing apparatus includes a flow meter and a memory unit. The flow meter detects the flow rate of the processing liquid passing through the flow path. The memory unit stores a trained model. The trained model is generated by machine learning based on a flow rate detection result of the flow meter during training and a vibration detection result of the vibration sensor during training. The determination unit inputs the vibration detection result of the vibration sensor into the trained model to obtain an estimated flow rate of the processing liquid passing through the flow path. The determination unit determines whether or not there is a sign of a failure of the adjustment valve based on the estimated flow rate.

[0011] In one embodiment, the determination unit calculates a difference between the estimated flow rate and the flow rate detection result of the flow meter when the vibration sensor detects vibration of the regulating valve, and determines whether or not there is a sign of a malfunction of the regulating valve based on the difference.

[0012] In one embodiment, the determining unit determines that there is a sign of a malfunction of the adjusting valve when the difference exceeds a threshold value a predetermined number of times or more within a predetermined period of time.

[0013] In one embodiment, the substrate processing apparatus includes a plurality of processing units for processing substrates. The piping, the on-off valve, the adjustment valve, the vibration sensor, the trained model, and the flow meter are each provided for each processing unit. The substrate processing apparatus further includes a common piping connected to the plurality of piping and supplying the processing liquid to the plurality of piping, and a pressure gauge for detecting the pressure within the common piping. The trained model is generated by machine learning based on the flow rate detection result during training, the vibration detection result during training, and the pressure detection result of the pressure gauge during training. The determination unit inputs the vibration detection result and the pressure detection result of the pressure gauge when the vibration sensor detects vibration of the adjustment valve into the trained model to obtain the estimated flow rate.

[0014] In one embodiment, the regulating valve comprises a motorized needle valve.

[0015] In one embodiment, the apparatus further includes a notification unit that, when the determination unit determines that there is a sign of a malfunction in the adjusting valve, issues a notification to prompt replacement of the adjusting valve and / or to prompt inspection of the substrate processing apparatus.

[0016] According to a second aspect of the present invention, a method for determining signs of failure includes the steps of opening or closing a flow path of a pipe, driving an electric adjustment valve that adjusts the flow rate of a processing liquid passing through the flow path, detecting vibration of the adjustment valve, and determining whether or not there are signs of failure of the adjustment valve based on the vibration detection results obtained by detecting vibration of the adjustment valve while the adjustment valve is being driven.

[0017] In one embodiment, the method for determining a failure symptom includes, prior to the determining step, inputting the vibration detection result into a trained model to acquire an estimated flow rate of the treatment liquid passing through the flow path. The trained model has been generated by machine learning based on a training flow rate detection result that detects the flow rate of the treatment liquid passing through the flow path and a training vibration detection result that detects vibration of the adjustment valve. In the determining step, the presence or absence of a failure symptom of the adjustment valve is determined based on the estimated flow rate.

[0018] According to a third aspect of the present invention, a learning method includes the steps of acquiring learning data generated according to a learning data generation method, and performing machine learning by inputting the learning data into a learning program. The learning data generation method includes the steps of acquiring a flow rate detection result during learning from a flow meter that detects a flow rate of a treatment liquid passing through a flow path in a pipe, acquiring a vibration detection result during learning from a vibration sensor that detects vibrations of an electric regulating valve that regulates the flow rate of the treatment liquid passing through the flow path, and storing the flow rate detection result during learning from the flow meter and the vibration detection result during learning from the vibration sensor while the regulating valve is operating associatively as learning data in a storage unit. [Effects of the Invention]

[0019] According to the present invention, it is possible to provide a substrate processing apparatus capable of detecting a symptom of a failure in an adjusting valve, a method for determining a symptom of a failure, and a learning method. [Brief explanation of the drawings]

[0020] [Figure 1] FIG. 1 is a schematic plan view of a substrate processing apparatus according to a first embodiment. [Figure 2] FIG. 2 is a schematic view of a processing unit in the substrate processing apparatus. [Figure 3] FIG. 3 is a cross-sectional view showing the configuration of an adjustment valve. [Figure 4] FIG. 2 is a diagram schematically illustrating the configuration of the periphery of a processing unit according to the first embodiment. [Figure 5] FIG. 1 is a block diagram of a substrate processing apparatus according to a first embodiment. [Figure 6] 10 is a diagram showing an example of a vibration detection result (vibration waveform) detected by a vibration sensor when a regulating valve with no signs of failure is driven. FIG. [Figure 7] FIG. 10 is a diagram showing an example of a flow rate detection result detected by a flow meter when an adjustment valve that shows no signs of failure is driven so as to maintain a constant flow rate of the treatment liquid. [Figure 8] 10 is a diagram showing an example of a vibration detection result (vibration waveform) detected by a vibration sensor when a regulation valve showing signs of failure is driven. FIG. [Figure 9] FIG. 10 is a diagram showing an example of an estimated flow rate obtained when the vibration detection results (vibration waveform) of an adjusting valve showing signs of failure are input to a trained model. [Figure 10] 10 is a flowchart showing a method for determining whether or not there is a symptom of a failure in an adjusting valve in the substrate processing apparatus of the first embodiment. [Figure 11] FIG. 10 is a diagram illustrating an example of a difference between a flow rate detection result and an estimated flow rate. [Figure 12] FIG. 10 is a diagram illustrating another example of the difference between the flow rate detection result and the estimated flow rate. [Figure 13] FIG. 1 is a diagram showing a configuration of a substrate processing learning system. [Figure 14] FIG. 1 is a block diagram of a learning data generation device. [Figure 15] 4 is a flowchart showing a method for generating learning data according to the first embodiment. [Figure 16] FIG. 2 is a block diagram of a learning device. [Figure 17] 4 is a flowchart showing a learning method according to the first embodiment. [Figure 18] FIG. 10 is a schematic plan view of a substrate processing apparatus according to a second embodiment. DETAILED DESCRIPTION OF THE INVENTION

[0021] Hereinafter, embodiments of the present invention will be described with reference to the drawings. However, the present invention is not limited to the following embodiments and can be implemented in various forms without departing from the spirit of the present invention. Note that duplicated explanations may be omitted where appropriate. Furthermore, in the drawings, identical or equivalent parts are designated by the same reference symbols and explanations will not be repeated. In this specification, to facilitate understanding of the invention, mutually orthogonal X-axis, Y-axis, and Z-axis may be described. In this embodiment, the X-axis and Y-axis are parallel to the horizontal direction, and the Z-axis is parallel to the vertical direction.

[0022] (First embodiment) A substrate processing apparatus 100 according to a first embodiment of the present invention will be described with reference to Figures 1 to 17. Figure 1 is a schematic plan view of the substrate processing apparatus 100 according to the first embodiment.

[0023] 1, the substrate processing apparatus 100 processes a substrate W. The substrate processing apparatus 100 processes the substrate W by performing at least one of etching, surface treatment, property imparting, treatment film formation, removal of at least a portion of a film, and cleaning on the substrate W.

[0024] The substrate W is used as a semiconductor substrate. The substrate W includes a semiconductor wafer. For example, the substrate W has a substantially circular disk shape. Here, the substrate processing apparatus 100 processes the substrates W one by one.

[0025] The substrate processing apparatus 100 includes a plurality of processing units 10, a fluid cabinet 110, a plurality of fluid boxes 120, a plurality of load ports LP, an indexer robot IR, a center robot CR, and a control device 101.

[0026] Each load port LP accommodates a plurality of stacked substrates W. The indexer robot IR transports the substrates W between the load port LP and the center robot CR. The center robot CR transports the substrates W between the indexer robot IR and the processing units 10. Each processing unit 10 supplies a processing liquid to the substrate W to process the substrate W. The fluid cabinet 110 accommodates the processing liquid. The processing liquid includes, for example, a chemical liquid, a rinse liquid, a removal liquid, and / or a water repellent agent. The fluid cabinet 110 accommodates the processing liquid. The fluid cabinet 110 may also accommodate a gas.

[0027] The multiple processing units 10 form multiple towers TW (four towers TW in FIG. 1) arranged to surround the center robot CR in a plan view. Each tower TW includes multiple processing units 10 stacked one above the other (three processing units 10 in FIG. 1). Each fluid box 120 corresponds to one of the multiple towers TW. The processing liquid in the fluid cabinet 110 is supplied to all of the processing units 10 included in the tower TW corresponding to the fluid box 120 via one of the fluid boxes 120.

[0028] The controller 101 controls the operation of each part of the substrate processing apparatus 100. For example, the controller 101 controls the load port LP, the indexer robot IR, and the center robot CR.

[0029] The control device 101 includes a control unit 102 and a storage unit 104. The control unit 102 has a processor. The control unit 102 has, for example, a central processing unit (CPU). Alternatively, the control unit 102 may have a general-purpose computer.

[0030] The storage unit 104 stores data and computer programs. The data includes recipe data. The recipe data includes information indicating a plurality of recipes. Each of the plurality of recipes defines the processing content and processing procedure for the substrate W.

[0031] The storage unit 104 includes a main storage device and an auxiliary storage device. The main storage device is, for example, a semiconductor memory. The auxiliary storage device is, for example, a semiconductor memory and / or a hard disk drive. The storage unit 104 may include removable media. The control unit 102 executes computer programs stored in the storage unit 104 to perform substrate processing operations.

[0032] Next, the processing unit 10 in the substrate processing apparatus 100 of the first embodiment will be described with reference to Fig. 2. Fig. 2 is a schematic diagram of the processing unit 10 in the substrate processing apparatus 100.

[0033] The processing unit 10 includes a chamber 12 and a substrate holder 20. The chamber 12 accommodates a substrate W. The substrate holder 20 holds the substrate W.

[0034] The chamber 12 is generally box-shaped and has an internal space. The chamber 12 accommodates the substrates W. Here, the substrate processing apparatus 100 is a single-wafer type that processes the substrates W one by one, and the chamber 12 accommodates the substrates W one by one. The substrates W are accommodated in the chamber 12 and processed within the chamber 12. The chamber 12 accommodates at least a part of the substrate holder 20 and at least a part of the processing liquid supply unit 30, which will be described later.

[0035] The substrate holding unit 20 holds the substrate W. The substrate holding unit 20 holds the substrate W horizontally so that the upper surface (front surface) Wa of the substrate W faces upward and the lower surface (back surface) Wb of the substrate W faces vertically downward. The substrate holding unit 20 also rotates the substrate W while holding it. The substrate holding unit 20 rotates the substrate W while holding it.

[0036] For example, the substrate holding unit 20 may be of a clamping type that clamps the edge of the substrate W. Alternatively, the substrate holding unit 20 may have any mechanism that holds the substrate W from the lower surface Wb. For example, the substrate holding unit 20 may be of a vacuum type.

[0037] For example, the substrate holder 20 includes a spin base 21, a chuck member 22, a shaft 23, an electric motor 24, and a housing 25. The chuck member 22 is provided on the spin base 21. The chuck member 22 chucks the substrate W. Typically, the spin base 21 is provided with a plurality of chuck members 22.

[0038] The shaft 23 is a hollow shaft. The shaft 23 extends vertically along the rotation axis Ax. The spin base 21 is coupled to the upper end of the shaft 23. The substrate W is placed above the spin base 21.

[0039] The spin base 21 is disk-shaped and supports the substrate W horizontally. The shaft 23 extends downward from the center of the spin base 21. The electric motor 24 applies a rotational force to the shaft 23. The electric motor 24 rotates the shaft 23 in a rotational direction, thereby rotating the substrate W and the spin base 21 around the rotation axis Ax. The housing 25 surrounds the shaft 23 and the electric motor 24.

[0040] The substrate processing apparatus 100 further includes a processing liquid supply unit 30. The processing liquid supply unit 30 supplies the processing liquid to the substrate W. Typically, the processing liquid supply unit 30 supplies the processing liquid to the upper surface Wa of the substrate W.

[0041] The processing liquid may include a so-called chemical liquid. The chemical liquid may include, for example, hydrofluoric acid. For example, the hydrofluoric acid may be heated to 40°C or higher and 70°C or lower, or may be heated to 50°C or higher and 60°C or lower. However, the hydrofluoric acid does not have to be heated. The chemical liquid may also include water or phosphoric acid.

[0042] The chemical solution may also include hydrogen peroxide, SC1 (ammonia-hydrogen peroxide mixture), SC2 (hydrochloric acid-hydrogen peroxide mixture), or aqua regia (a mixture of concentrated hydrochloric acid and concentrated nitric acid).

[0043] The processing liquid supply unit 30 includes a pipe 31, a nozzle 32, an opening / closing valve 33, an adjustment valve 34, and a flow meter 35. The nozzle 32 discharges the processing liquid onto the upper surface Wa of the substrate W. For example, the nozzle 32 may discharge the processing liquid onto the central portion of the substrate W, or onto a region between the central portion and the peripheral edge portion of the substrate W. The nozzle 32 has a discharge port and discharges the processing liquid from the discharge port.

[0044] The nozzle 32 is preferably configured to be movable relative to the substrate W. The nozzle 32 can move horizontally and / or vertically in accordance with a movement mechanism controlled by the control unit 102. Note that in this specification, the movement mechanism is omitted to avoid overly complicating the drawings. The nozzle 32 is connected to a pipe 31. A processing liquid is supplied to the pipe 31 from a supply source.

[0045] The on-off valve 33 is provided in the pipe 31 and opens and closes the flow path in the pipe 31. Specifically, the on-off valve 33 includes a valve body (not shown) having a valve seat provided therein, a valve element that opens and closes the valve seat, and an actuator (not shown) that moves the valve element between an open position and a closed position.

[0046] 3 is a cross-sectional view showing the configuration of the regulating valve 34. The regulating valve 34 is provided in the pipe 31 and regulates the flow rate of the processing liquid passing through the flow path of the pipe 31. The regulating valve 34 is an electrically operated regulating valve. In this embodiment, the regulating valve 34 is a motor needle valve.

[0047] 3, specifically, the adjustment valve 34 includes a needle 341, a stepping motor 342, a valve body 343, a case 344, a connecting member 345, and a sealing member 346. The valve body 343 has an internal flow path 343a through which the treatment liquid passes. The valve body 343 also has an inlet 343b and an outlet 343c formed at one end and the other end of the internal flow path 343a, respectively. The inlet 343b and the outlet 343c are connected to the pipe 31.

[0048] The needle 341 adjusts the opening degree of the internal flow path 343a. In other words, the needle 341 adjusts the flow rate of the treatment liquid passing through the internal flow path 343a. The stepping motor 342 moves the needle 341 relative to the valve body 343. Specifically, the stepping motor 342 moves the needle 341 in the axial direction.

[0049] The case 344 is fixed to the valve body 343 and houses the stepping motor 342 and a connecting member 345. The connecting member 345 connects the motor shaft 342a of the stepping motor 342 to the needle 341. The sealing member 346 seals the gap between the case 344 and the needle 341. The sealing member 346 is, for example, an O-ring. The sealing member 346 prevents the processing liquid from the internal flow path 343a from leaking into the case 344.

[0050] The adjusting valve 34 also has an origin (reference position) sensor 347. The origin sensor 347 is a sensor for positioning the needle 341 at the origin position (reference position). Specifically, the origin sensor 347 has a light-emitting unit 347a having a light-emitting element and a light-receiving unit 347b having a light-receiving element. The light-emitting unit 347a and the light-receiving unit 347b are arranged on the opposite side of the motor shaft 342a from the needle 341. The motor shaft 342a moves in the axial direction when the stepping motor 342 is driven, and the position where the motor shaft 342a blocks the light from the light-emitting unit 347a toward the light-receiving unit 347b is the origin position (reference position) of the motor shaft 342a and the needle 341.

[0051] The stepping motor 342 moves the needle 341 between an open position and a closed position. The stepping motor 342 moves the needle 341 in the axial direction, thereby changing the degree of opening of the internal flow path 343a in the valve body 343. The degree of opening of the internal flow path 343a is adjusted by the number of drive pulses input to the stepping motor 342 after the needle 341 is positioned at the home position.

[0052] 2, the flow meter 35 detects the flow rate of the treatment liquid passing through the pipe 31. The flow meter 35 is not particularly limited, but may be, for example, an ultrasonic flow meter. The flow rate detection result of the flow meter 35 is used to control (feedback control) the adjustment valve 34.

[0053] The substrate processing apparatus 100 further includes a cup 80. The cup 80 collects the processing liquid splashed from the substrate W. The cup 80 moves up and down. For example, the cup 80 moves up vertically to the side of the substrate W during the period in which the processing liquid supply unit 30 supplies the processing liquid to the substrate W. In this case, the cup 80 collects the processing liquid splashed from the substrate W due to the rotation of the substrate W. Furthermore, when the period in which the processing liquid supply unit 30 supplies the processing liquid to the substrate W ends, the cup 80 moves down vertically from the side of the substrate W.

[0054] As described above, the control device 101 includes the control unit 102 and the memory unit 104. The control unit 102 controls the substrate holder 20, the processing liquid supply unit 30, and / or the cup 80. In one example, the control unit 102 controls the electric motor 24, the opening / closing valve 33, and the adjusting valve 34.

[0055] 2, the processing liquid supply part 30 is capable of supplying one type of processing liquid to the substrate W, but the processing liquid supply part 30 may be capable of supplying multiple types of processing liquid to the substrate W. For example, the processing liquid supply part 30 may include multiple pipes 31, nozzles 32, on-off valves 33, and adjustment valves 34.

[0056] Next, the processing unit 10 of the first embodiment will be further described with reference to Fig. 4. Fig. 4 is a diagram schematically showing the configuration of the processing unit 10 of the first embodiment.

[0057] As shown in FIG. 4, the substrate processing apparatus 100 includes a detector 41, a controller 42, and a controller 43.

[0058] The detector 41 detects the flow rate of the processing liquid based on the flow rate detection result of the flow meter 35. Then, the detector 41 generates a detection signal. The detection signal indicates the detected value of the flow rate of the processing liquid. The detector 41 outputs the detection signal to the controller 42. More specifically, the detector 41 detects the flow rate of the processing liquid according to a predetermined sampling period. Therefore, the detector 41 outputs the detection signal to the controller 42 according to the predetermined sampling period. Note that the flow meter 35 and the detector 41 may form a flow rate sensor.

[0059] The controller 42 drives the stepping motor 342 of the regulating valve 34 to control the opening of the flow path of the pipe 31. Specifically, the controller 42 includes an arithmetic circuit and a driver circuit. The arithmetic circuit executes PID control in accordance with a predetermined sampling period. The driver circuit generates a drive current that drives the stepping motor 342 of the regulating valve 34.

[0060] More specifically, the controller 42 performs PID control on the detection signal based on the detection signal input from the detector 41 (flow rate sensor) and a preset target value. Then, the controller 42 generates a drive current (pulse signal) for driving the stepping motor 342 of the adjustment valve 34 based on the PID-controlled detection signal, and inputs the drive current to the stepping motor 342 of the adjustment valve 34. This causes the stepping motor 342 of the adjustment valve 34 to rotate based on the drive current. As a result, the needle 341 moves, and the opening of the flow path is adjusted so that the flow rate of the treatment liquid passing through the pipe 31 becomes the target value. The target value is set in the controller 42 by the control unit 102.

[0061] The controller 43 drives the on-off valve 33 to open or close the flow path of the pipe 31. Specifically, the controller 43 includes a driver circuit. The driver circuit generates a drive current for driving the on-off valve 33.

[0062] Specifically, the controller 43 generates a drive current (pulse signal) for driving the on-off valve 33 based on the on-off signal input from the control unit 102, and inputs the drive current to the on-off valve 33. As a result, the on-off valve 33 opens and closes the flow path.

[0063] The substrate processing apparatus 100 also includes a vibration sensor 51 and a determination device 55. The vibration sensor 51 detects vibrations of the regulating valve 34. In this embodiment, the vibration sensor 51 is attached to the regulating valve 34. The vibration sensor 51 generates a detection signal corresponding to the detected vibration and outputs it to the determination device 55. In this embodiment, the detection signal indicates the detected acceleration. The vibration sensor 51 is, for example, an acceleration sensor, a velocity sensor, or a displacement sensor. In this embodiment, the vibration sensor 51 is an acceleration sensor. The acceleration sensor has, for example, a piezoelectric element.

[0064] The determination device 55 includes a determination unit 55a and a storage unit 55b. The determination unit 55a has a processor. The determination unit 55a has, for example, a central processing unit (CPU). Alternatively, the determination unit 55a may have a general-purpose computer.

[0065] The memory unit 55b stores data and computer programs. The data includes, for example, vibration detection results of the vibration sensor 51. The memory unit 55b also stores a trained model M. The trained model M is a model generated by machine learning based on the flow rate detection results of the flow meter 35 during learning and the vibration detection results of the vibration sensor 51 during learning. The trained model M is provided for each recipe. Details of the trained model M will be described later.

[0066] The storage unit 55b includes a main storage device and an auxiliary storage device. The main storage device is, for example, a semiconductor memory. The auxiliary storage device is, for example, a semiconductor memory and / or a hard disk drive. The storage unit 55b may include removable media. The determination unit 55a executes a computer program stored in the storage unit 55b to determine whether or not there is a sign of a malfunction in the adjusting valve 34.

[0067] Next, the substrate processing apparatus 100 of the first embodiment will be further described with reference to Fig. 5. Fig. 5 is a block diagram of the substrate processing apparatus 100 of the first embodiment.

[0068] As shown in FIG. 5, the control unit 102 controls the indexer robot IR to transfer the substrate W by the indexer robot IR.

[0069] The control unit 102 controls the center robot CR to transfer the substrate W by the center robot CR. For example, the center robot CR receives an unprocessed substrate W and carries the substrate W into one of the processing units 10. The center robot CR also receives a processed substrate W from the processing unit 10 and carries the substrate W out.

[0070] The control unit 102 controls the substrate holding unit 20 to start rotation of the substrate W, change the rotation speed, and stop rotation of the substrate W. For example, the control unit 102 can control the substrate holding unit 20 to change the rotation speed of the substrate holding unit 20. Specifically, the control unit 102 can change the rotation speed of the substrate W by changing the rotation speed of the electric motor 24 of the substrate holding unit 20.

[0071] The control unit 102 controls the on-off valve 33 to switch the state of the on-off valve 33 between an open state and a closed state. Specifically, the control unit 102 switches the on-off valve 33 to an open state or a closed state by sending a control signal (on-off signal) to the controller 43. This allows the processing liquid in the pipe 31 to pass or not to pass.

[0072] Furthermore, the control unit 102 can control the controller 42 to adjust the opening degree of the adjustment valve 34. Specifically, the control unit 102 adjusts the opening degree of the adjustment valve 34 by sending a target flow rate signal indicating a target value of the flow rate of the treatment liquid to the controller 42.

[0073] The determining unit 55a of the determining device 55 determines whether or not there is a sign of a failure in the adjusting valve 34 based on the vibration detection result of the vibration sensor 51 while the adjusting valve 34 is being driven.

[0074] In this embodiment, the determination unit 55a inputs the vibration detection result of the vibration sensor 51 into the trained model M to obtain an estimated flow rate of the treatment liquid passing through the piping 31. Then, the determination unit 55a determines whether or not there is a sign of a malfunction of the adjustment valve 34 based on the estimated flow rate.

[0075] In this embodiment, the determination unit 55a calculates the difference between the estimated flow rate and the flow rate detection result of the flow meter 35. Then, the determination unit 55a determines whether or not there is a sign of a failure in the regulating valve 34 based on the difference.

[0076] When the determination unit 55a determines that there is a symptom of a malfunction of the adjusting valve 34, it transmits a malfunction symptom signal to the control unit 102. In this embodiment, the substrate processing apparatus 100 further includes a notification unit 190. The control unit 102 transmits a malfunction symptom signal indicating that there is a symptom of a malfunction of the adjusting valve 34 to the notification unit 190. As a result, the notification unit 190 issues a notification to the user urging them to replace the adjusting valve 34 and / or inspect the substrate processing apparatus 100. As a result, for example, the user replaces the adjusting valve 34 with a new one. Note that the determination unit 55a may transmit the malfunction symptom signal directly to the notification unit 190 without going through the control unit 102. The notification unit 190 is not particularly limited, and may include, for example, at least one of a display device having a display panel, a speaker that emits sound, and a light-emitting element that emits light.

[0077] Next, with reference to Figs. 6 to 8, the vibration detection result (vibration waveform) detected by the vibration sensor 51 and the flow rate detection result detected by the flow meter 35 will be described. Fig. 6 is a diagram showing an example of the vibration detection result (vibration waveform) detected by the vibration sensor 51 when the adjustment valve 34, which shows no signs of failure, is driven. Fig. 7 is a diagram showing an example of the flow rate detection result detected by the flow meter 35 when the adjustment valve 34, which shows no signs of failure, is driven so that the flow rate of the treatment liquid becomes constant. Fig. 8 is a diagram showing an example of the vibration detection result (vibration waveform) detected by the vibration sensor 51 when the adjustment valve 34, which shows signs of failure, is driven.

[0078] When an adjusting valve 34 that shows no signs of failure (for example, a brand new adjusting valve 34) is driven, the vibration detection result (vibration waveform) detected by the vibration sensor 51 is as shown in FIG. 6. Specifically, for example, when the needle 341 of the adjusting valve 34 is moved from a predetermined position (for example, a fully closed position) to the origin position, an operation start torque is applied at time Ta1 when the needle 341 starts to move, and therefore the vibration (acceleration) increases. Thereafter, the magnitude of the vibration (acceleration) falls within a substantially constant range between the predetermined position (for example, the fully closed position) and the origin position. Then, at time Ta2 when the needle 341 stops moving, an operation stop torque is applied, and therefore the vibration (acceleration) increases.

[0079] Furthermore, when an adjustment valve 34 that shows no signs of failure is driven so as to maintain a constant flow rate of the processing liquid, the flow rate detection result detected by the flow meter 35 is as shown in FIG. 7. Specifically, for example, the on-off valve 33 is changed from a closed state to an open state (times Tb1 to Tb2). Then, the adjustment valve 34 that shows no signs of failure is feedback-controlled so as to maintain a constant flow rate of the processing liquid (F1) (times Tb2 to Tb3). Thereafter, the on-off valve 33 is changed from an open state to a closed state (times Tb3 to Tb4). In this case, the feedback control of the adjustment valve 34 maintains a substantially constant flow rate of the processing liquid.

[0080] On the other hand, when an adjusting valve 34 showing signs of failure is driven, the vibration detection result (vibration waveform) detected by the vibration sensor 51 is as shown in FIG. 8. Specifically, for example, when the needle 341 of the adjusting valve 34 is moved from a predetermined position (e.g., the fully closed position) to the home position, a start torque is applied at the time Ta1 when the needle 341 starts moving, causing the vibration (acceleration) to increase. Thereafter, the magnitude of the vibration (acceleration) remains within a substantially constant range between the predetermined position (e.g., the fully closed position) and the home position, but may suddenly increase. In other words, while the needle 341 is being moved at a constant speed, the vibration (acceleration) may suddenly increase. Note that the magnitude of the suddenly increased vibration (acceleration) is approximately the same as the magnitude of the vibration (acceleration) at the time Ta1 when the needle 341 starts moving or at the time Ta2 when the needle 341 stops moving.

[0081] Furthermore, when an adjustment valve 34 showing signs of failure is driven so as to maintain a constant flow rate of the treatment liquid, the flow rate detection result detected by the flow meter 35 is as shown in Fig. 7. In other words, even when an adjustment valve 34 showing signs of failure is used, the same flow rate detection result can be obtained as when an adjustment valve 34 showing no signs of failure is used.

[0082] Next, with reference to Fig. 9, an estimated flow rate obtained by inputting the vibration detection result of the vibration sensor 51 into the trained model M will be described. Fig. 9 is a diagram showing an example of an estimated flow rate obtained when the vibration detection result (vibration waveform) of the regulating valve 34 showing signs of failure is input into the trained model M.

[0083] When the vibration detection result (see FIG. 6) of the regulating valve 34, which shows no signs of failure, is input to the trained model M, the estimated flow rate shown in FIG. 7 is output. In other words, it is estimated that the estimated flow rate of the processing liquid becomes approximately constant due to the feedback control (times Tb2 to Tb3).

[0084] On the other hand, when the vibration detection result (see FIG. 8) of the adjusting valve 34 showing signs of failure is input to the trained model M, the estimated flow rate as shown in FIG. 9 is output. Specifically, even if feedback control (times Tb2 to Tb3) is performed, it is estimated that variations will occur in the estimated flow rate of the treatment liquid.

[0085] Next, a method for determining whether or not there is a sign of a malfunction of the regulating valve 34 using the substrate processing apparatus 100 of the first embodiment will be described with reference to Figures 10 to 12. Figure 10 is a flowchart showing a method for determining whether or not there is a sign of a malfunction of the regulating valve 34 using the substrate processing apparatus 100 of the first embodiment. Figure 11 is a diagram showing an example of the difference between the flow rate detection result and the estimated flow rate. Figure 12 is a diagram showing another example of the difference between the flow rate detection result and the estimated flow rate. In this embodiment, the determination of whether or not there is a sign of a malfunction of the regulating valve 34 is performed while a substrate W is being processed using a certain recipe.

[0086] In step S1, the control unit 102 controls the controller 43 to change the on-off valve 33 from a closed state to an open state.

[0087] Next, in step S2, the control unit 102 controls the controller 42 to adjust the opening of the regulating valve 34. Specifically, the control unit 102 transmits a target flow rate signal to the controller 42. The controller 42 also receives a flow rate detection result from the flow meter 35. Then, based on the target flow rate signal and the flow rate detection result, the controller 42 drives the regulating valve 34 so that the flow rate of the processing liquid flowing through the pipe 31 matches the target flow rate. At this time, the stepping motor 342 of the regulating valve 34 moves the needle 341 in the axial direction.

[0088] Next, in step S3, the determination unit 55a receives the vibration detection result of the regulating valve detected by the vibration sensor 51. The determination unit 55a also receives the flow rate detection result detected by the flow meter .

[0089] Next, in step S4, the determination unit 55a inputs the vibration detection result to the trained model M. At this time, the determination unit 55a may input factor information affecting the control performance, which will be described later, control parameter information, and time-series data (including the vibration detection result) to the trained model M.

[0090] Next, in step S5, the determination unit 55a acquires the estimated flow rate from the trained model M. Specifically, if there are no signs of a malfunction in the regulating valve 34, the estimated flow rate will be approximately constant, for example, as shown between times Tb2 and Tb3 in Fig. 7. On the other hand, if there are signs of a malfunction in the regulating valve 34, the estimated flow rate will vary more, for example, as shown between times Tb2 and Tb3 in Fig. 9.

[0091] Next, in step S6, the determination unit 55a calculates the difference between the flow rate detection result received in step S3 and the estimated flow rate acquired in step S5.

[0092] Specifically, when there is no sign of a malfunction in the regulating valve 34, both the flow rate detection result and the estimated flow rate are substantially constant, as shown, for example, between times Tb2 and Tb3 in Fig. 7. Therefore, the difference between the flow rate detection result and the estimated flow rate is constant at substantially zero, as shown, for example, in Fig. 11.

[0093] On the other hand, if there are signs of a malfunction in the regulating valve 34, the flow rate detection result will be approximately constant, but the estimated flow rate will vary greatly. As a result, the difference between the flow rate detection result and the estimated flow rate will increase, for example, as shown in Figure 12. Specifically, the average value of the difference will increase, or the variation in the difference will increase.

[0094] Next, in step S7, the determination unit 55a determines whether or not there is a sign of a malfunction of the adjusting valve 34 based on the difference. For example, the determination unit 55a determines that there is a sign of a malfunction when the difference exceeds a threshold value a predetermined number of times or more within a predetermined period. The threshold value is a value greater than the difference shown in FIG. 11. The predetermined number of times may be one. However, in order to suppress erroneous determination due to noise, the predetermined number of times is preferably multiple times.

[0095] Furthermore, for example, the determining unit 55a may determine that there is a sign of a malfunction when the average value of the differences exceeds a threshold value, which is a value greater than the differences shown in FIG.

[0096] If the determining unit 55a determines in step S7 that there is a symptom of a malfunction, the process proceeds to step S8.

[0097] Next, in step S8, the presence of a malfunction symptom is notified. Specifically, the determination unit 55a transmits a malfunction symptom signal to the control unit 102. The control unit 102 transmits the malfunction symptom signal to the notification unit 190. As a result, the notification unit 190 notifies the user that there is a malfunction symptom of the adjusting valve 34. Note that the determination unit 55a may transmit the malfunction symptom signal to the notification unit 190 without going through the control unit 102. Thereafter, the processing proceeds to step S9.

[0098] On the other hand, if the determining unit 55a determines in step S7 that there is no sign of a malfunction, the process proceeds to step S9.

[0099] Next, in step S9, the control unit 102 determines whether or not a predetermined time has elapsed since the on-off valve 33 was opened.

[0100] If the control unit 102 determines in step S9 that the predetermined time has not elapsed since the on-off valve 33 was opened, the process returns to step S2.

[0101] On the other hand, if the control unit 102 determines in step S9 that the predetermined time has elapsed since the on-off valve 33 was opened, the process proceeds to step S10.

[0102] Next, in step S10, the control unit 102 controls the controller 43 to change the on-off valve 33 from the open state to the closed state.

[0103] In this manner, the determination of whether or not there is a sign of a malfunction in the adjusting valve 34 is completed.

[0104] In this embodiment, as described above, the determination unit 55a determines whether or not there is a sign of a malfunction of the adjusting valve 34 based on the vibration detection result of the vibration sensor 51 while the adjusting valve 34 is driven. Therefore, it is possible to detect a sign of a malfunction of the adjusting valve 34. In other words, it is possible to detect a sign of an abnormality in the adjusting valve 34 before an abnormality occurs in the flow rate of the processing liquid flowing through the pipe 31, and it is possible to replace the adjusting valve 34 with a new one. Therefore, it is possible to prevent a sudden flow rate abnormality from occurring, and it is possible to prevent unintentional interruptions of substrate processing. As a result, it is possible to prevent a decrease in productivity.

[0105] As described above, the determination unit 55a inputs the vibration detection result of the vibration sensor 51 into the trained model M to obtain the estimated flow rate of the processing liquid passing through the piping 31, and determines whether or not there is a sign of a malfunction in the regulating valve 34 based on the estimated flow rate. Therefore, it is possible to easily determine whether or not there is a sign of a malfunction in the regulating valve 34.

[0106] Furthermore, as described above, the determination unit 55a calculates the difference between the estimated flow rate and the flow rate detection result, and determines, based on the difference, whether or not there is a sign of a malfunction in the regulating valve 34. Therefore, if there is no sign of a malfunction in the regulating valve 34, a difference such as that shown in Fig. 11 is obtained, whereas if there is a sign of a malfunction in the regulating valve 34, a difference such as that shown in Fig. 12 is obtained. Therefore, it is possible to more easily determine whether or not there is a sign of a malfunction in the regulating valve 34.

[0107] Furthermore, as described above, the determining unit 55a determines that there is a sign of a malfunction when the difference exceeds the threshold value a predetermined number of times or more within a predetermined period of time, thereby making it possible to prevent erroneous determinations caused by noise in the detection results.

[0108] As described above, the adjusting valve 34 has a motor needle valve. Since a motor needle valve is likely to show signs of vibration before it fails, applying the present invention to an adjusting valve 34 having a motor needle valve is particularly effective.

[0109] Furthermore, as described above, the device is provided with a notification unit 190 that notifies the user of the presence of a failure symptom when the determination unit 55a determines that there is a failure symptom in the adjusting valve 34. Therefore, the user can be notified of the presence of a failure symptom in the adjusting valve 34. Therefore, the user can easily know that there is a failure symptom in the adjusting valve 34.

[0110] Next, the substrate processing learning system 1000 will be described with reference to Figures 13 to 17. Figure 13 is a diagram showing the configuration of the substrate processing learning system 1000.

[0111] As shown in FIG. 13, the substrate processing learning system 1000 includes a substrate processing apparatus 100, a learning data generating device 400, and a learning device 500.

[0112] The substrate processing apparatus 100 processes the substrate W with a processing liquid in order to generate learning data. Hereinafter, for ease of understanding, the time when the learning data is generated may be referred to as "during learning." Furthermore, the term "during learning" may be used to refer to the results detected by each part of the substrate processing apparatus 100 when generating the learning data.

[0113] The substrate processing apparatus 100 transmits at least the learning flow rate detection result detected by the flowmeter 35 during learning and the learning vibration detection result detected by the vibration sensor 51 during learning to the learning data generation device 400. Specifically, the substrate processing apparatus 100 transmits at least the time-series data of the learning flow rate detection result and the learning vibration detection result to the learning data generation device 400. In this embodiment, the substrate processing apparatus 100 transmits information on factors affecting control performance, control parameter information, and the time-series data to the learning data generation device 400. In this embodiment, the learning flow rate detection result and the learning vibration detection result do not include the detection results of the flowmeter 35 and the vibration sensor 51 when an adjustment valve 34 showing signs of failure is used. In other words, in this embodiment, the learning flow rate detection result and the learning vibration detection result are acquired using an adjustment valve 34 showing no signs of failure (e.g., a new adjustment valve 34).

[0114] Factors that affect control performance include, for example, the production recipe number, the position of the needle 341, the target flow rate value, the flow rate detection result of the flow meter 35, the primary (upstream) pressure value of the adjustment valve 34, the temperature of the processing liquid, and the opening / closing signal of the opening / closing valve 33.

[0115] The control parameters include, for example, the period of PID control and the position of the needle 341 at the start of control.

[0116] The time series data includes, for example, time series data of the operation amount (movement amount) of the needle 341 calculated for each control cycle based on the control parameters, and time series data of the vibration detection result of the vibration sensor 51.

[0117] The substrate processing apparatus 100 stores information on factors affecting control performance, control parameter information, and time-series data in the storage unit 104, or has a sensor for detecting part of this information.

[0118] The learning data generation device 400 associates the learning-time flow rate detection result with the learning-time vibration detection result to generate learning data 410. In this embodiment, the learning data generation device 400 associates factor information that affects control performance, control parameter information, and time-series data to generate the learning data 410. The learning data generation device 400 transmits the generated learning data 410 to the learning device 500.

[0119] The learning device 500 performs preprocessing on the data included in the learning data 410. For example, the learning device 500 converts the control signals for the on-off valve 33 and the adjustment valve 34 into analog signals. In addition, for example, the learning device 500 removes values ​​that can be determined as noise based on a normal distribution, and interpolates values ​​so that they approach a normal distribution.

[0120] Furthermore, the learning device 500 extracts an influence index for each piece of data included in the learning data 410 on the flow rate (objective variable) of the treatment liquid flowing through the pipe 31. Then, the learning device 500 deletes from the learning data 410 any data included in the learning data 410 that has a small influence index. The method for extracting the influence index is not particularly limited, but may be performed by multivariate analysis, for example. Multivariate analysis includes, for example, simple regression analysis or multiple regression analysis.

[0121] Then, the learning device 500 generates a trained model M by machine learning based on the training data 410. In detail, the learning device 500 generates the trained model M by inputting the training data 410 into a learning program. The learning device 500 transmits the generated trained model M to the substrate processing apparatus 100.

[0122] Next, the device 400 for generating training data will be further described with reference to Fig. 14. Fig. 14 is a block diagram showing the device 400 for generating training data.

[0123] 14, the learning data generation device 400 includes a control device 401. The control device 401 includes a control unit 402 and a storage unit 403.

[0124] The control unit 402 includes a processor. The control unit 402 includes, for example, a CPU or an MPU. Alternatively, the control unit 402 may include a general-purpose computing device. The control unit 402 may further include an NPU.

[0125] The storage unit 403 stores a control program 431 and data. The control program 431 is a computer program. The storage unit 403 has a main storage device. The storage unit 403 may further have an auxiliary storage device. The storage unit 403 may include removable media. The control unit 402 executes various processes based on the control program 431 and data stored in the storage unit 403.

[0126] The storage unit 403 further stores a data generation program 432. The control unit 402 stores the flow rate detection result during learning and the vibration detection result during learning, which are acquired from the substrate processing apparatus 100, in the storage unit 403. In this embodiment, the control unit 402 stores information on factors affecting control performance, control parameter information, and time-series data, which are acquired from the substrate processing apparatus 100, in the storage unit 403.

[0127] The data generation program 432 is a program for associating the flow rate detection results during learning, the vibration detection results during learning, etc., to generate the learning data 410. In this embodiment, the data generation program 432 is a program for associating information on factors that affect control performance, control parameter information, time-series data, etc., to generate the learning data 410.

[0128] The control unit 402 generates learning data 410 by associating the flow rate detection result during learning, the vibration detection result during learning, and the like, and stores the data in the storage unit 403. In this embodiment, the control unit 402 generates learning data 410 by associating information on factors that affect control performance, control parameter information, time-series data, and the like, and stores the data in the storage unit 403.

[0129] The training data generation device 400 further includes a display unit 404 and an input unit 405.

[0130] The display unit 404 displays various screens and various pieces of information. The display unit 404 is, for example, a liquid crystal display or an organic EL display. The display unit 404 may display an input screen for inputting various pieces of data to the control unit 402.

[0131] The input unit 405 receives input from a user and outputs various information to the control unit 402. The input unit 405 includes, for example, a touch panel and a pointing device. The input unit 405 may receive input of an instruction to generate training data 410.

[0132] Next, a method for generating the training data 410 according to the first embodiment will be described with reference to Fig. 14 and Fig. 15. Fig. 15 is a flowchart showing the method for generating the training data 410 according to the first embodiment. The method for generating the training data 410 includes steps S101 and S102.

[0133] 15, in step S101, the learning data generation device 400 (controller 402) acquires a learning-time flow rate detection result and a learning-time vibration detection result. Specifically, the controller 402 acquires the learning-time flow rate detection result and the learning-time vibration detection result from the substrate processing apparatus 100. In this embodiment, the controller 402 acquires factor information that affects control performance, control parameter information, and time-series data. Note that in this embodiment, the controller 402 acquires the learning-time flow rate detection result and the learning-time vibration detection result in one step (step S101), but the learning-time flow rate detection result and the learning-time vibration detection result may also be acquired in separate steps.

[0134] Next, in step S102, the control unit 402 associates the flow rate detection result during learning with the vibration detection result during learning, and stores the associated data in the storage unit 403 as learning data 410. In this embodiment, the control unit 402 associates information on factors that affect control performance, control parameter information, and time-series data, and stores the associated data in the storage unit 403 as learning data 410.

[0135] In this way, the generation of the learning data 410 is completed.

[0136] Next, the learning device 500 will be further described with reference to Fig. 16. Fig. 16 is a block diagram showing the learning device 500. As shown in Fig. 16, the learning device 500 includes a control device 501. The control device 501 includes a control unit 502 and a storage unit 503.

[0137] The control unit 502 includes a processor. The control unit 502 includes, for example, a CPU or an MPU. Alternatively, the control unit 502 may include a general-purpose computing device. The control unit 502 may further include an NPU.

[0138] The storage unit 503 stores a control program 531 and data. The control program 531 is a computer program. The storage unit 503 has a main storage device. The storage unit 503 may further have an auxiliary storage device. The storage unit 503 may include removable media. The control unit 502 executes various processes based on the control program 531 and data stored in the storage unit 503.

[0139] The storage unit 503 further stores a learning program 532. The control unit 502 causes the storage unit 503 to store the learning data 410 acquired from the device 400 for generating learning data.

[0140] The learning program 532 is a program for executing a machine learning algorithm to find certain rules from the learning data 410 and generate a trained model M that expresses the found rules. When the control unit 502 executes the learning program 532, the parameters of the inference program are adjusted by machine learning based on the learning data 410. As a result, the trained model M is generated. The control unit 502 stores the generated trained model M in the memory unit 503.

[0141] The machine learning algorithm is not particularly limited as long as it is supervised learning, and may be, for example, a decision tree, a nearest neighbor method, a naive Bayes classifier, a support vector machine, or a neural network. Therefore, the trained model M includes a decision tree, a nearest neighbor method, a naive Bayes classifier, a support vector machine, or a neural network.

[0142] For example, a neural network includes an input layer, one or more hidden layers, and an output layer. Specifically, the neural network is a deep neural network (DNN), a recurrent neural network (RNN), or a convolutional neural network (CNN), and performs deep learning. For example, a deep neural network includes an input layer, multiple hidden layers, and an output layer.

[0143] Learning device 500 further includes a display unit 504 and an input unit 505 .

[0144] The display unit 504 displays various screens and information. The display unit 504 is, for example, a liquid crystal display or an organic EL display. The display unit 504 may display an input screen for inputting various data to the control unit 502.

[0145] The input unit 505 accepts input from a user and outputs various information to the control unit 502. The input unit 505 includes, for example, a touch panel and a pointing device. The input unit 505 may accept input of an instruction to generate a trained model M.

[0146] Next, the learning method of the first embodiment will be described with reference to Fig. 16 and Fig. 17. Fig. 17 is a flowchart showing the learning method of the first embodiment. The learning method of the learning data 410 includes steps S201 to S205.

[0147] As shown in FIG. 17, in step S201, the control unit 502 acquires the learning data 410 from the storage unit 503.

[0148] Next, in step S202, the control unit 502 performs preprocessing on the learning data 410.

[0149] Next, in step S203, control unit 502 deletes from learning data 410 data with small influence indices among the data included in learning data 410.

[0150] Next, in step S204, the control unit 502 inputs the learning data 410 into the learning program 532 to perform machine learning.

[0151] In step S205, the control unit 502 generates a model (one or more functions) to which multiple learned parameters (coefficients) are applied as the learned model M. The control unit 502 stores the generated learned model M in the storage unit 503.

[0152] (Second embodiment) Next, a substrate processing apparatus 100 according to a second embodiment of the present invention will be described with reference to Fig. 18. Fig. 18 is a schematic plan view of the substrate processing apparatus 100 according to the second embodiment. In the second embodiment, an example will be described in which the substrate processing apparatus 100 includes a common pipe 600 that supplies a processing liquid to a plurality of processing units 10, and a pressure gauge 730 that detects the pressure inside the common pipe 600. Note that, for simplification of the drawing, Fig. 18 depicts only two of the processing units 10 included in the substrate processing apparatus 100.

[0153] 18, the substrate processing apparatus 100 includes a plurality of processing units 10 and a common pipe 600. The processing units 10 have the same configuration. The pipe 31, the on-off valve 33, the adjustment valve 34, the vibration sensor 51, the flow meter 35, and the trained model M are provided for each processing unit 10. In this embodiment, the pipe 31, the on-off valve 33, the controller 43, the adjustment valve 34, the controller 42, the vibration sensor 51, the determination unit 55a, the flow meter 35, the detector 41, and the trained model M are provided for each processing unit 10.

[0154] The common pipe 600 is connected to a supply source and receives the processing liquid from the supply source. The common pipe 600 supplies the processing liquid to the pipes 31 of the multiple processing units 10. The common pipe 600 has an upstream pipe 600a, a midstream pipe 600b, and a downstream pipe 600c. The upstream pipe 600a is connected to the supply source. The midstream pipe 600b is located downstream of the upstream pipe 600a and is connected to the pipes 31 of the multiple processing units 10. For ease of understanding, the processing unit 10 connected most upstream of the midstream pipe 600b may be referred to as processing unit 10a, and the processing unit 10 connected most downstream of the midstream pipe 600b may be referred to as processing unit 10b. The downstream pipe 600c is located downstream of the midstream pipe 600b. The downstream pipe 600c is connected to the supply source and returns the processing liquid to the supply source.

[0155] The substrate processing apparatus 100 includes an on-off valve 610 and a pressure adjustment valve 620. The on-off valve 610 is provided in the upstream pipe 600a and opens and closes a flow path in the upstream pipe 600a. Specifically, the on-off valve 610 includes, for example, a valve body (not shown) having a valve seat provided therein, a valve element that opens and closes the valve seat, and an actuator (not shown) that moves the valve element between an open position and a closed position.

[0156] The pressure regulating valve 620 is provided in the downstream pipe 600 c and regulates the pressure of the processing liquid in the downstream pipe 600 c. Specifically, the pressure regulating valve 620 includes, for example, a valve body (not shown) with a valve seat provided therein, a valve element that opens and closes the valve seat, and an adjustment spring (not shown) that adjusts the position of the valve element.

[0157] The substrate processing apparatus 100 includes a pipe 710 and a regulator 720. The pipe 710 connects a gas (e.g., air) supply source to the pressure adjustment valve 620. In this embodiment, the regulator 720 is an electropneumatic regulator. The regulator 720 is provided on the pipe 710 and adjusts the air pressure supplied to the pressure adjustment valve 620. The air supplied from the regulator 720 to the pressure adjustment valve 620 acts on an adjustment spring of the pressure adjustment valve 620, thereby adjusting the position of the valve element.

[0158] The substrate processing apparatus 100 includes a pressure gauge 730, a pressure adjustment valve controller 740, and a thermometer 750. The pressure gauge 730 is provided in the upstream pipe 600a and detects the pressure of the processing liquid in the upstream pipe 600a. The pressure detection result (pressure detection signal) of the pressure gauge 730 is transmitted to the pressure adjustment valve controller 740. In this embodiment, the pressure detection result of the pressure gauge 730 is also transmitted to the determination device 55 of each processing unit 10.

[0159] The pressure regulating valve controller 740 drives the regulator 720 to control the air pressure supplied to the pressure regulating valve 620. Specifically, the pressure regulating valve controller 740 includes an arithmetic circuit. The arithmetic circuit executes PID control in accordance with a predetermined sampling period. The pressure regulating valve controller 740 performs PID control on the pressure detection signal based on a pressure detection signal input from the pressure gauge 730 and a preset target pressure value. The pressure regulating valve controller 740 then generates a control signal based on the PID-controlled pressure detection signal and inputs it to the regulator 720.

[0160] The thermometer 750 detects the temperature of the processing liquid in the upstream pipe 600a and the temperature of the processing liquid in the downstream pipe 600c. The temperature detection result of the thermometer 750 is transmitted to the control unit 102. In this embodiment, the temperature detection result of the thermometer 750 is also transmitted to the determination device 55 of each processing unit 10.

[0161] In this embodiment, the trained model M is generated by machine learning based on the flow rate detection results of the flow meter 35 at the time of learning, the vibration detection results of the vibration sensor 51 at the time of learning, and the pressure detection results of the pressure meter 730 at the time of learning. The trained model M may also be generated by machine learning based on the flow rate detection results of the flow meter 35 at the time of learning, the vibration detection results of the vibration sensor 51 at the time of learning, the pressure detection results of the pressure meter 730 at the time of learning, and the temperature detection results of the thermometer 750 at the time of learning. In this embodiment, the trained model M may also be generated by machine learning based on the pressure detection results of the pressure meter 730 at the time of learning and the temperature detection results of the thermometer 750 at the time of learning, in addition to the factor information, control parameter information, and time-series data that affect the control performance described above.

[0162] In the present embodiment, the determination unit 55a inputs the vibration detection result of the vibration sensor 51 and the pressure detection result of the pressure gauge 730 when the regulating valve 34 is operating to the trained model M to obtain an estimated flow rate. The determination unit 55a may input the vibration detection result of the vibration sensor 51, the pressure detection result of the pressure gauge 730 when the regulating valve 34 is operating, and the temperature detection result of the thermometer 750 when the regulating valve 34 is operating to the trained model M to obtain an estimated flow rate. In the present embodiment, the determination unit 55a may generate the trained model M by machine learning based on the pressure detection result of the pressure gauge 730 at the time of learning and the temperature detection result of the thermometer 750 at the time of learning in addition to the factor information, control parameter information, and time-series data that affect the control performance described above.

[0163] In this embodiment, the learning data 410 includes the pressure detection result of the pressure gauge 730 during learning. The learning data 410 may also include the pressure detection result of the pressure gauge 730 during learning and the temperature detection result of the thermometer 750 during learning.

[0164] In this embodiment, as described above, the determination unit 55a acquires an estimated flow rate by inputting the vibration detection result of the vibration sensor 51 and the pressure detection result of the pressure gauge 730 when the adjustment valve 34 is operating into the trained model M. Therefore, for example, the determination unit 55a (hereinafter sometimes referred to as the determination unit 55a of the processing unit 10a) that determines whether or not there is a sign of a malfunction in the adjustment valve 34 of the processing unit 10a can also detect abnormalities in devices around the common pipe 600 (opening / closing valve 610, pressure adjustment valve 620, etc.) or in other processing units 10 (processing unit 10b, etc.).

[0165] Specifically, for example, if there is no sign of a malfunction in the adjustment valve 34 of the processing unit 10a, but an abnormality occurs in the equipment around the common pipe 600 or in another processing unit 10, the pressure of the processing liquid in the common pipe 600 connected to the multiple processing units 10 increases or decreases. In this case, the estimated flow rate obtained by inputting the pressure detection result of the pressure gauge 730 into the trained model M is, for example, as shown in FIG. 9. On the other hand, because the adjustment valve 34 of the processing unit 10a is feedback-controlled, the flow rate detection result of the flow meter 35 is, for example, as shown in FIG. 7. Therefore, the difference between the estimated flow rate and the flow rate detection result of the flow meter 35 is, for example, as shown in FIG. 12. Therefore, the determination unit 55a of the processing unit 10a determines that there is a sign of a malfunction in the adjustment valve 34, or that an abnormality has occurred in any of the equipment related to the system supplying the processing liquid (the equipment around the common pipe 600 and all of the processing units 10). The user then inspects, for example, the equipment around the common pipe 600 and / or the other processing units 10.

[0166] In this embodiment, similarly to the first embodiment, if there is a symptom of a malfunction in the adjusting valve 34 of the processing unit 10a, the determining unit 55a of the processing unit 10a can detect the symptom of the malfunction.

[0167] Other configurations, effects, a method for generating the training data 410, and a training method of the second embodiment are the same as those of the first embodiment.

[0168] The embodiments of the present invention have been described above with reference to the drawings. However, the present invention is not limited to the above embodiments and can be implemented in various forms without departing from the spirit of the present invention. Furthermore, the components disclosed in the above embodiments can be modified as appropriate. For example, some of the components shown in one embodiment may be added to the components of another embodiment, or some of the components shown in one embodiment may be deleted from the embodiment.

[0169] The drawings mainly show each component in a schematic manner to facilitate understanding of the invention, and the thickness, length, number, spacing, etc. of each component shown in the drawings may differ from the actual ones due to the convenience of creating the drawings. Furthermore, the configuration of each component shown in the above embodiment is merely an example and is not particularly limited, and it goes without saying that various modifications are possible within a range that does not substantially deviate from the effects of the present invention.

[0170] For example, in the above embodiment, an example has been described in which the determination of the presence or absence of a symptom of a malfunction of the adjusting valve 34 is performed while the substrate W is being processed, but the present invention is not limited to this. The determination of the presence or absence of a symptom of a malfunction of the adjusting valve 34 may also be performed when the substrate W is not being processed.

[0171] Furthermore, in the above embodiment, an example has been described in which the trained model M is used to determine whether or not there is a sign of failure in the adjusting valve 34, but the present invention is not limited to this. For example, the presence or absence of a sign of failure may be determined from vibration detection results such as those shown in Fig. 8 without using the trained model M. In this case, for example, when a substrate W is not being processed or feedback control is not being performed, the needle 341 may be moved to obtain vibration detection results (vibration waveforms) such as those shown in Fig. 8, and whether or not there is a sign of failure may be determined by determining whether or not the magnitude of vibration in the vibration waveform during a period excluding the start time Ta1 and the stop time Ta2 is greater than or equal to a threshold value.

[0172] In the above embodiment, an example has been described in which the presence or absence of signs of a malfunction of the regulating valve 34 is determined based on the difference between the estimated flow rate and the flow rate detection result, but the present invention is not limited to this. For example, the presence or absence of signs of a malfunction of the regulating valve 34 may be determined based on a waveform indicating an estimated flow rate as shown in Fig. 9 without using the flow rate detection result.

[0173] In the above embodiment, an example in which the determination unit 55a and the control unit 102 are provided separately has been described, but the present invention is not limited to this. For example, the control unit 102 may also function as the determination unit 55a. Furthermore, the control unit 102 may perform part of the functions of the determination unit 55a. In these cases, the control unit 102 is an example of the "determination unit" of the present invention.

[0174] In the above embodiment, an example has been described in which the learning device 500 performs preprocessing on the learning data 410 and extracts the influence index, but the present invention is not limited to this. For example, the learning data generation device 400 may perform at least one of preprocessing on the learning data 410 and extracting the influence index.

[0175] In the above embodiment, the learning data generation device 400 and the learning device 500 are provided separately, but the present invention is not limited to this. For example, a single device having the functions of both the learning data generation device 400 and the learning device 500 may be provided.

[0176] Furthermore, in the above embodiment, an example has been described in which the processing liquid from the supply source is supplied directly to the substrate W, but the present invention is not limited to this. For example, a first processing liquid from a first supply source and a second processing liquid from a second supply source may be mixed to generate a mixed processing liquid, and the mixed processing liquid may be supplied to the substrate W. That is, in the present invention, it may be determined whether or not there is a sign of failure in an adjustment valve provided in a pipe through which the processing liquids pass before being mixed. [Industrial Applicability]

[0177] The present invention is useful in the field of substrate processing. [Explanation of symbols]

[0178] 10, 10a, 10b: Processing units 31: Piping 33: Opening and closing valve 34: Adjusting valve 35:Flowmeter 51: Vibration sensor 55a: Judgment section 55b: Storage section 100: Substrate processing apparatus 190: Information Department 600: Common piping 730: Pressure gauge M: Trained model S1: Step (process for opening or closing) S2: Step (driving process) S3: Step (vibration detection process) S5: Step (process for obtaining estimated flow rate) S7: Step (judging process) S101: Step (Step of acquiring flow rate detection results during learning, Step of acquiring vibration detection results during learning) S102: Step (storing step) S201: Step (process for acquiring learning data) S204: Step (machine learning process) W: Substrate

Claims

1. a pipe through which the treatment liquid passes; an on-off valve that opens and closes the flow path of the piping; an electrically operated adjusting valve for adjusting the flow rate of the treatment liquid passing through the flow path; a vibration sensor for detecting vibration of the adjusting valve; Judgment section and Equipped with The substrate processing apparatus, wherein the determining unit determines whether or not there is a sign of a failure of the adjusting valve based on a vibration detection result of the vibration sensor while the adjusting valve is driven.

2. a flow meter for detecting a flow rate of the treatment liquid passing through the flow path; a storage unit that stores a trained model generated by machine learning based on the flow rate detection result of the flow meter during learning and the vibration detection result of the vibration sensor during learning; Equipped with The determination unit The vibration detection result of the vibration sensor is input to the trained model to obtain an estimated flow rate of the treatment liquid passing through the flow path; The substrate processing apparatus according to claim 1 , wherein the presence or absence of a symptom of a failure of the adjusting valve is determined based on the estimated flow rate.

3. The determination unit Calculating a difference between the estimated flow rate and the flow rate detection result of the flow meter when the vibration sensor detects the vibration of the regulating valve; The substrate processing apparatus according to claim 2 , wherein the presence or absence of a symptom of a failure of the adjusting valve is determined based on the difference.

4. The substrate processing apparatus according to claim 3 , wherein the determining unit determines that there is a sign of a failure of the adjusting valve when the difference exceeds a threshold value a predetermined number of times or more within a predetermined period.

5. a plurality of processing units for processing substrates; the piping, the on-off valve, the adjustment valve, the vibration sensor, the trained model, and the flow meter are each provided for each processing unit; The substrate processing apparatus includes: a common pipe connected to the plurality of pipes and supplying the processing liquid to the plurality of pipes; a pressure gauge for detecting the pressure in the common pipe; Furthermore, the trained model is generated by machine learning based on the flow rate detection result at the time of training, the vibration detection result at the time of training, and the pressure detection result at the time of training of the pressure meter; 5. The substrate processing apparatus according to claim 2, wherein the determination unit inputs the vibration detection result and the pressure detection result of the pressure gauge when the vibration sensor detects vibration of the adjustment valve into the learned model to obtain the estimated flow rate.

6. The substrate processing apparatus according to claim 1 , wherein the adjusting valve comprises a motor needle valve.

7. 5. A substrate processing apparatus as described in any one of claims 1 to 4, further comprising an alarm unit that issues an alarm to prompt replacement of the adjustment valve and / or an alarm to prompt inspection of the substrate processing apparatus when the judgment unit determines that there are signs of failure of the adjustment valve.

8. opening or closing a flow path of the piping; driving an electric adjusting valve that adjusts the flow rate of the treatment liquid passing through the flow path; detecting vibration of the regulating valve; a step of determining whether or not there is a sign of a failure of the regulating valve based on a vibration detection result obtained by detecting vibration of the regulating valve while the regulating valve is being driven; A method for determining a failure symptom, including:

9. Prior to the determining step, inputting the vibration detection result into a trained model to obtain an estimated flow rate of the treatment liquid passing through the flow path; the trained model is generated by machine learning based on a learning-time flow rate detection result that detects a flow rate of the treatment liquid passing through the flow path and a learning-time vibration detection result that detects vibration of the adjustment valve, 9. The method for determining a symptom of a failure according to claim 8, wherein in the determining step, it is determined whether or not there is a symptom of a failure of the adjusting valve based on the estimated flow rate.

10. acquiring training data generated according to the training data generation method; a step of performing machine learning by inputting the learning data into a learning program; Including, The learning data generation method includes: acquiring a learning flow rate detection result of a flow meter that detects the flow rate of the treatment liquid passing through a flow path of the piping; acquiring a learning vibration detection result of a vibration sensor that detects vibration of an electric adjustment valve that adjusts the flow rate of the treatment liquid passing through the flow path; a step of associating the flow rate detection result during learning of the flow meter with the vibration detection result during learning of the vibration sensor while the regulating valve is driven, and storing the result as learning data in a storage unit; including, learning methods.

Images