Learning data generation method and learning completion model generation method

Abstract

The invention provides a learning data generation method for additional learning and a learning completion model generation method. The substrate processing apparatus includes a nozzle, a moving mechanism, a storage unit, and a control unit. The moving mechanism moves the nozzle. The storage unit stores a learned model. The learning completion model is generated by learning, as learning data, learning target speed information indicating a moving speed of the nozzle and a processing amount acquired by moving the nozzle at a speed based on the learning target speed information and performing processing on the learning target substrate. The control unit inputs a target amount of the processing amount to the learning-completed model, thereby outputting the processing speed information from the learning-completed model. The control unit controls the movement mechanism such that the nozzle moves at a speed based on the processing-time speed information when processing is performed on the substrate to be processed. The processing-time speed information indicates a moving speed of the nozzle.

Description

[0001] Related divisional applications

[0002] This patent application is a divisional application of the invention patent application with application number 202011559598.X and titled "Substrate processing device, method, system and method for generating learning data". The filing date of the original application is December 25, 2020. Technical Field

[0003] The present invention relates to a method for generating learning data for additional learning and a method for generating a learned model. Background Art

[0004] There is known a substrate processing device that uses a wafer with a film formed on the surface as a processing object, and adjusts the film thickness of the film and removes foreign matter by liquid processing. As one of such substrate processing devices, there is a single-chip substrate processing device including a nozzle for supplying an etching processing liquid to the wafer surface (for example, see Patent Document 1). When performing an etching process, the substrate processing device of Patent Document 1 moves the nozzle based on a predetermined speed profile while ejecting the etching processing liquid from the nozzle.

[0005] [Prior art literature]

[0006] [Patent Document]

[0007] Patent Document 1: Japanese Patent Application Publication No. 2019-54104 Summary of the invention

[0008] [Problems to be solved by the invention]

[0009] However, in a substrate processing apparatus that moves a nozzle based on a predetermined velocity profile, for example, it is necessary to redevelop the velocity profile in accordance with a change in a pre-processing performed on a wafer. This is because when the pre-processing is changed, the film thickness of the coating before etching or the deviation of the film thickness of the coating before etching will change. This places a burden on operators involved in the velocity profile development.

[0010] The present invention has been made in view of the above-mentioned problems, and an object of the present invention is to provide a substrate processing apparatus, a substrate processing method, a substrate processing system, and a method for generating learning data that can reduce the burden on an operator.

[0011] [Technical means to solve the problem]

[0012] According to one embodiment of the present invention, a substrate processing device supplies a processing liquid to a substrate to perform processing on the substrate. The substrate processing device includes a nozzle, a moving mechanism, a storage unit, and a control unit. The nozzle supplies the processing liquid to the substrate. The moving mechanism moves at least one of the nozzle and the substrate. The storage unit stores a learned model. The control unit uses the learned model to control the moving mechanism. The learned model is generated by learning learning object speed information and processing volume as learning data, the learning object speed information indicating at least one of the moving speed of the nozzle and the moving speed of the learning object substrate, or indicating the relative moving speed of the nozzle and the learning object substrate, and the processing volume is obtained by moving at least one of the nozzle and the learning object substrate at a speed based on the learning object speed information and performing the processing on the learning object substrate. The processing volume indicates the amount of processing of the substrate by the processing. The control unit inputs the target amount of the processing volume into the learned model, thereby outputting the processing speed information from the learned model. When the processing is performed on the processing target substrate, the control unit controls the moving mechanism so that at least one of the nozzle and the processing target substrate moves at a speed based on the processing speed information. The processing speed information indicates at least one of the moving speed of the nozzle and the moving speed of the processing target substrate, or indicates the relative moving speed between the nozzle and the processing target substrate.

[0013] In one embodiment, the learning object speed information indicates at least one of the moving speed of the nozzle set for each of the nozzle positions that divide the moving interval of the nozzle movement into a plurality of intervals, and the moving speed of the learning object substrate set for each of the substrate positions that divide the moving interval of the learning object substrate movement into a plurality of intervals, or the learning object speed information indicates the relative moving speed of the nozzle and the learning object substrate for each of the nozzle positions or each of the substrate positions. The processing speed information indicates at least one of the moving speed of the nozzle set for each of the nozzle positions and the moving speed of the processing object substrate set for each of the substrate positions, or the processing speed information indicates the relative moving speed of the nozzle and the processing object substrate for each of the nozzle positions or each of the substrate positions.

[0014] In one embodiment, the substrate processing device further includes a substrate holding portion and a substrate rotating portion. The substrate holding portion holds the substrate horizontally. The substrate rotating portion rotates the substrate and the substrate holding portion integrally around a central axis extending in the vertical direction. The relative movement speed indicates the relative movement speed between the surface of the rotating substrate and the nozzle.

[0015] In one embodiment, the control unit learns the processing speed information and the processing volume as learning data for additional learning, thereby generating a learning completion model after additional learning, and the processing volume is obtained by moving the nozzle and at least one of the processing object substrate at a speed based on the processing speed information and performing the processing on the processing object substrate.

[0016] In one embodiment, the control unit determines whether the processing amount can be used for the additional learning.

[0017] In one embodiment, the substrate processing apparatus further includes a measuring device. The measuring device measures a value of a measurement object. The storage unit stores a recipe for controlling the execution of the processing. The recipe indicates a set value for the measurement object. When the difference between the value of the measurement object and the set value is greater than a predetermined value, the control unit performs different error notifications depending on whether the processing amount can be used for the additional learning.

[0018] In one embodiment, the substrate processing apparatus further includes a processing amount detection unit for detecting the processing amount.

[0019] In one embodiment, the storage unit stores the learning object speed information. The control unit controls the moving mechanism so that at least one of the nozzle and the learning object substrate moves at a speed based on the learning object speed information, and supplies the processing liquid from the nozzle to the learning object substrate, thereby performing the processing on the learning object substrate. The control unit generates the learning data including the processing amount and the learning object speed information, and the processing amount is obtained by performing the processing on the learning object substrate. The control unit classifies the learning data into categories based on a predetermined classification rule. The control unit learns the learning data for each of the categories and generates the learned model. When performing the processing on the processing object substrate, the control unit selects one of the categories based on a predetermined selection rule. The control unit inputs the target amount into the learned model corresponding to the selected category, thereby outputting the processing speed information from the learned model.

[0020] In one embodiment, the classification rule stipulates that the learning data is classified into the categories according to the number of processed wafers of the learning target substrate. The number of processed wafers of the learning target substrate indicates the number of processed wafers counted from the start of use of the processing liquid used in the processing of the learning target substrate. The selection rule stipulates that one of the categories is selected according to the number of processed wafers of the processing target substrate. The number of processed wafers of the processing target substrate indicates the number of processed wafers counted from the start of use of the processing liquid used in the processing of the processing target substrate.

[0021] In one embodiment, the control unit obtains the batch number during learning when performing the processing on the learning object substrate. The batch number during learning is the batch number counted from the start of use of the processing liquid used in the processing. The classification rule stipulates that the learning data is classified into the categories according to the batch number during learning. The control unit obtains the batch number during processing when performing the processing on the processing object substrate. The batch number during processing is the batch number counted from the start of use of the processing liquid used in the processing. The selection rule stipulates that one of the categories is selected according to the batch number during processing.

[0022] In one embodiment, the substrate processing apparatus further includes an input unit. The input unit inputs instructions for selecting reference learning data from the learning data contained in each of the categories for each of the categories. The control unit selects, from the learning data contained in each of the categories, the learning data whose correlation with the reference learning data satisfies a predetermined condition for each of the categories. The control unit learns the selected learning data for each of the categories to generate the learned model.

[0023] According to another embodiment of the present invention, a substrate processing method is a method for supplying a processing liquid to a processing target substrate from a nozzle to perform processing on the processing target substrate. The substrate processing method includes: a step of obtaining a target amount of processing volume, the processing volume indicating the amount of processing on the processing target substrate by the processing; an output step of inputting the target amount into a learned model, thereby outputting processing speed information from the learned model; and a processing step of moving at least one of the nozzle and the processing target substrate at a speed based on the processing speed information, and supplying the processing liquid to the processing target substrate. The learned model is generated by learning the learning target speed information and the processing volume as learning data, the learning target speed information indicating at least one of the moving speed of the nozzle and the moving speed of the learning target substrate, or indicating the relative moving speed of the nozzle and the learning target substrate, and the processing volume is obtained by moving at least one of the nozzle and the learning target substrate at a speed based on the learning target speed information and performing the processing on the learning target substrate. The processing speed information indicates at least one of a moving speed of the nozzle and a moving speed of the processing target substrate, or indicates a relative moving speed between the nozzle and the processing target substrate.

[0024] In one embodiment, the learning object speed information indicates at least one of the moving speed of the nozzle set for each of the nozzle positions that divide the moving interval of the nozzle movement into a plurality of intervals, and the moving speed of the learning object substrate set for each of the substrate positions that divide the moving interval of the learning object substrate movement into a plurality of intervals, or the learning object speed information indicates the relative moving speed of the nozzle and the learning object substrate for each of the nozzle positions or each of the substrate positions. The processing speed information indicates at least one of the moving speed of the nozzle set for each of the nozzle positions and the moving speed of the processing object substrate set for each of the substrate positions, or the processing speed information indicates the relative moving speed of the nozzle and the processing object substrate for each of the nozzle positions or each of the substrate positions.

[0025] In one embodiment, in the processing step, the processing target substrate is rotated around a central axis extending in the vertical direction while being kept horizontal. The relative movement speed between the nozzle and the processing target substrate represents the relative movement speed between the surface of the rotating processing target substrate and the nozzle. The relative movement speed between the nozzle and the processing target substrate represents the relative movement speed between the surface of the rotating processing target substrate and the nozzle.

[0026] In one embodiment, the substrate processing method also includes: a learning data generating step, which generates learning data for additional learning based on the processing speed information output in the output step and the processing amount of the processing object substrate processed in the processing step; and a step of learning the learning data for the additional learning to generate a learning completion model after the additional learning.

[0027] In one embodiment, the learning data generating step includes a step of determining whether the processing amount can be used for the additional learning.

[0028] In one embodiment, the substrate processing method also includes: a step of determining whether the difference between the value of the measured object and a pre-determined set value for the measured object is greater than a specified value; and when the difference between the value of the measured object and the set value is greater than a specified value, a step of performing different error notifications depending on whether the processing amount can be used for the additional learning.

[0029] In one embodiment, the substrate processing method further includes: a step of moving at least one of the nozzle and the learning object substrate at a speed based on the learning object speed information, and supplying the processing liquid to the learning object substrate to perform the processing on the learning object substrate; a step of acquiring the processing amount of the learning object substrate; a step of generating the learning data, the learning data including the processing amount of the learning object substrate and the learning object speed information; and a learning step of classifying the learning data into categories based on a predetermined classification rule, and generating the learned model by learning the learning data for each category. In the output step, one of the categories is selected based on a predetermined selection rule, and the target amount is input into the learned model corresponding to the selected category, thereby outputting the processing speed information from the learned model.

[0030] In one embodiment, the classification rule stipulates that the learning data is classified into the categories according to the number of processed wafers of the learning target substrate. The number of processed wafers of the learning target substrate indicates the number of processed wafers counted from the start of use of the processing liquid used in the processing of the learning target substrate. The selection rule stipulates that one of the categories is selected according to the number of processed wafers of the processing target substrate. The number of processed wafers of the processing target substrate indicates the number of processed wafers counted from the start of use of the processing liquid used in the processing of the processing target substrate.

[0031] In one embodiment, in the step of performing the processing on the learning object substrate, the batch number during learning is obtained. The batch number during learning is the batch number counted from the start of use of the processing liquid used in the processing. The classification rule stipulates that the learning data is classified into the categories according to the batch number during learning. In the step of performing the processing on the processing object substrate, the batch number during processing is obtained. The batch number during processing is the batch number counted from the start of use of the processing liquid used in the processing. The selection rule stipulates that one of the categories is selected according to the batch number during processing.

[0032] In one embodiment, the substrate processing method further includes: for each of the categories, selecting reference learning data from the learning data included in each of the categories; and for each of the categories, selecting the learning data whose correlation with the reference learning data satisfies a predetermined condition from the learning data included in each of the categories. In the learning step, for each of the categories, the selected learning data is learned to generate the learned model.

[0033] According to another embodiment of the present invention, a substrate processing system includes a substrate processing device and an information processing device. The substrate processing device supplies a processing liquid to a substrate to perform processing on the substrate. The information processing device sends processing speed information to the substrate processing device. The information processing device includes a communication unit, a storage unit, and a processing unit. The communication unit of the information processing device receives a target amount of the amount of processing of the processing target substrate by the processing from the substrate processing device. The communication unit of the information processing device sends the processing speed information to the substrate processing device. The storage unit stores a learning completed model. The processing unit inputs the target amount obtained from the substrate processing device via the communication unit into the learning completed model, thereby outputting the processing speed information from the learning completed model. The substrate processing device includes a communication unit, a nozzle, a moving mechanism, and a control unit. The communication unit of the substrate processing device sends the target amount to the information processing device. The communication unit of the substrate processing device receives the processing speed information from the information processing device. The nozzle supplies the processing liquid to the substrate. The moving mechanism moves at least one of the nozzle and the substrate. The control unit controls the moving mechanism based on the processing speed information. The learned model is generated by learning the learning object speed information and the processing amount as learning data, the learning object speed information indicating at least one of the moving speed of the nozzle and the moving speed of the learning object substrate, or indicating the relative moving speed of the nozzle and the learning object substrate, and the processing amount is obtained by moving at least one of the nozzle and the learning object substrate at a speed based on the learning object speed information and performing the processing on the learning object substrate. The processing amount indicates the amount of processing of the learning object substrate by the processing. When performing the processing on the processing object substrate, the control unit controls the moving mechanism so that at least one of the nozzle and the processing object substrate moves at a speed based on the processing speed information. The processing speed information indicates at least one of the moving speed of the nozzle and the moving speed of the processing object substrate, or indicating the relative moving speed of the nozzle and the processing object substrate.

[0034] According to another embodiment of the present invention, a learning device includes a storage unit and a learning unit. The storage unit stores a learning program. The learning unit learns learning data based on the learning program. The learning unit learns speed information and processing volume as learning data, the speed information indicating at least one of a moving speed of a nozzle and a moving speed of a substrate, or indicating a relative moving speed between the nozzle and the substrate, and the processing volume is obtained by moving at least one of the nozzle and the substrate at a speed based on the speed information and supplying a processing liquid to the substrate from the nozzle. The processing volume indicates the amount of the substrate processed by the supply of the processing liquid.

[0035] According to another embodiment of the present invention, a learning method includes the following steps: acquiring learning data; and learning the learning data based on a learning program. The learning data represents speed information and a processing amount, and the processing amount is obtained by moving at least one of a nozzle and a substrate at a speed based on the speed information and supplying a processing liquid to the substrate from the nozzle. The speed information represents at least one of a moving speed of the nozzle and a moving speed of the substrate, or represents a relative moving speed between the nozzle and the substrate. The processing amount represents an amount of the substrate processed by supplying the processing liquid.

[0036] According to another embodiment of the present invention, a method for generating a learned model includes the following steps: acquiring learning data; and generating a learned model by learning the learning data based on a learning program. The learning data represents speed information and a processing amount, and the processing amount is obtained by moving at least one of a nozzle and a substrate at a speed based on the speed information and supplying a processing liquid from the nozzle to the substrate. The speed information represents at least one of a moving speed of the nozzle and a moving speed of the substrate, or represents a relative moving speed between the nozzle and the substrate. The processing amount represents an amount of the substrate processed by supplying the processing liquid.

[0037] According to another embodiment of the present invention, a method for generating learning data includes: a step of moving at least one of a nozzle and a substrate at a speed based on speed information, and supplying a processing liquid from the nozzle to the substrate to perform processing on the substrate; a step of obtaining a processing amount, wherein the processing amount indicates an amount of processing of the substrate by the processing; and a step of generating learning data based on the speed information and the processing amount. The speed information indicates at least one of a moving speed of the nozzle and a moving speed of the substrate, or indicates a relative moving speed of the nozzle and the substrate.

[0038] [Effects of the Invention]

[0039] According to the substrate processing apparatus, substrate processing method, substrate processing system, and learning data generation method of the present invention, it is possible to reduce the burden on operators. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 It is a schematic diagram of the substrate processing apparatus in Embodiment 1 of the present invention.

[0041] Figure 2 This is a schematic diagram of a processing unit in Embodiment 1 of the present invention.

[0042] Figure 3 It is a plan view showing the scanning process in the first embodiment of the present invention.

[0043] Figure 4 This is a diagram showing scanning speed information in Embodiment 1 of the present invention.

[0044] Figure 5 This is a graph showing an example of the moving speed of the first nozzle in the first embodiment of the present invention.

[0045] Figure 6 It is a plan view showing the thickness measurement process in the first embodiment of the present invention.

[0046] Figure 7 It is a schematic diagram of the etching liquid supply unit in Embodiment 1 of the present invention.

[0047] Figure 8 This is a block diagram of the control device in Embodiment 1 of the present invention.

[0048] Fig. 9 This is a flowchart showing a method for generating learning data in Embodiment 1 of the present invention.

[0049] Fig.10 This is a flowchart showing the etching process in the first embodiment of the present invention.

[0050] Fig.11 This is a flowchart showing the process of generating learning data in Embodiment 1 of the present invention.

[0051] Fig.12 This is a flowchart showing the abnormality determination process in the first embodiment of the present invention.

[0052] Fig.13 This is a diagram showing an example of a learning data set table in Embodiment 1 of the present invention.

[0053] Fig.14 This is a block diagram of the control device in Embodiment 1 of the present invention.

[0054] Fig.15This is a flowchart showing a method for generating a learned model in the first embodiment of the present invention.

[0055] Fig.16 This is a block diagram of the control device in Embodiment 1 of the present invention.

[0056] Fig.17 This is a flowchart showing a substrate processing method in Embodiment 1 of the present invention.

[0057] Fig.18 This is a flowchart showing a substrate processing method in Embodiment 1 of the present invention.

[0058] Fig.19 This is a diagram showing an example of the additional learning data set table in Embodiment 1 of the present invention.

[0059] Fig. 20 It is a diagram showing a substrate processing apparatus and a learning apparatus in Embodiment 2 of the present invention.

[0060] Fig.21 This is a flowchart showing a method for generating a learned model in the second embodiment of the present invention.

[0061] Fig. 22 It is a diagram showing a substrate processing system in accordance with a third embodiment of the present invention.

[0062] Fig.23 This is a flowchart showing the processing executed by the control device of the substrate processing apparatus and the processing executed by the processing unit of the learning device.

[0063] Fig.24 It is a schematic diagram of a processing unit in Embodiment 4 of the present invention.

[0064] Fig.25 It is a schematic diagram showing the structure of a substrate processing apparatus in Embodiment 4 of the present invention.

[0065] Fig.26 This is a flowchart showing a method for generating learning data in Embodiment 4 of the present invention.

[0066] Fig. 27 This is a flowchart showing the process of generating learning data in Embodiment 4 of the present invention.

[0067] Fig.28 This is a diagram showing an example of a learning data set table in Embodiment 4 of the present invention.

[0068] Fig.29 This is a flowchart showing a method for generating a learned model in Embodiment 4 of the present invention.

[0069] Fig.30This is a flowchart showing a part of a substrate processing method in Embodiment 4 of the present invention.

[0070] Fig.31 This is a flowchart showing the process of acquiring speed information in the fourth embodiment of the present invention.

[0071] Fig.32 This is a flowchart showing a method for generating learning data in Embodiment 5 of the present invention.

[0072] Fig.33 This is a flowchart showing an example of counting processing in Embodiment 5 of the present invention.

[0073] Fig.34 This is a diagram showing an example of a learning data set table in Embodiment 5 of the present invention.

[0074] Fig.35 This is a flowchart showing a part of the substrate processing method in Embodiment 5 of the present invention.

[0075] [Explanation of Symbols]

[0076] 1: Processing unit

[0077] 2: Chamber

[0078] 3: Rotating chuck

[0079] 4: Etching liquid supply unit

[0080] 5: Rotating motor part

[0081] 6: Nozzle moving mechanism

[0082] 8: Thickness measurement unit

[0083] 9: Probe moving mechanism

[0084] 41: First nozzle

[0085] 100: Substrate processing device

[0086] 101: Control Devices

[0087] 102: Control Department

[0088] 103: Storage

[0089] 136: Learning the model

[0090] 200: Learning device

[0091] 201: Processing Department

[0092] 202: Storage

[0093] TG: Object

[0094] W: substrate DETAILED DESCRIPTION

[0095] Hereinafter, embodiments of the present invention will be described with reference to the accompanying drawings. However, the present invention is not limited to the following embodiments and can be implemented in various ways without departing from the scope of the present invention. In addition, for the parts where the description is repeated, the description is sometimes appropriately omitted. Moreover, in the drawings, the same reference numerals are marked for the same or equivalent parts and the description is not repeated.

[0096] [Implementation Method 1]

[0097] Reference Figure 1 A substrate processing apparatus 100 according to this embodiment will be described. Figure 1 Schematic diagram of a substrate processing apparatus 100 according to this embodiment. Figure 1 1 is a schematic plan view of a substrate processing apparatus 100. The substrate processing apparatus 100 is a single-wafer type apparatus that processes substrates W one by one. In the present embodiment, the substrate W is a semiconductor wafer. The substrate W is substantially disk-shaped.

[0098] like Figure 1 As shown, the substrate processing apparatus 100 includes a plurality of processing units 1, a fluid cabinet 100A, a plurality of fluid boxes 100B, a plurality of load ports LP, an indexer robot IR, a center robot CR and a control device 101.

[0099] Each load port LP stacks and stores a plurality of substrates W. The handling robot IR transfers the substrates W between the load port LP and the central robot CR. The central robot CR transfers the substrates W between the handling robot IR and the processing unit 1. Each processing unit 1 supplies a processing liquid to the substrate W to process the substrate W. The fluid chamber 100A stores the processing liquid.

[0100] The plurality of processing units 1 form a plurality of towers TW ( Figure 1 Each tower TW includes a plurality of processing units 1 stacked up and down ( Figure 1 There are three processing units 1 in the figure. The fluid boxes 100B correspond to the plurality of towers TW. The processing liquid in the fluid chamber 100A is supplied to all the processing units 1 included in the tower TW corresponding to the fluid box 100B via any one of the fluid boxes 100B.

[0101] The control device 101 controls the operation of each unit of the substrate processing apparatus 100. For example, the control device 101 controls the load port LP, the loading and unloading robot IR, and the central robot CR.

[0102] In this embodiment, the control device 101 functions as a learning device. Specifically, the control device 101 performs machine learning. The control device 101 of this embodiment further performs additional learning. Machine learning and additional learning are, for example, any of taught learning, untaught learning, semi-taught learning, reinforcement learning, and deep learning.

[0103] In addition, in general, the concept of machine learning includes additional learning, but for convenience, the term "machine learning" and the term "additional learning" are used separately to explain this embodiment. In this embodiment, "machine learning" means machine learning performed before the manufacture of semiconductor products. That is, "machine learning" means machine learning performed in the manufacturing stage of the substrate processing device 100. On the other hand, "additional learning" means machine learning performed in the manufacturing stage of semiconductor products. That is, "additional learning" means machine learning performed when adjusting the substrate processing device 100.

[0104] Then, refer to Figure 2 The processing unit 1 of this embodiment will be described. Figure 2 Schematic diagram of the processing unit 1 of this embodiment. In detail, Figure 2 is a schematic cross-sectional view of the processing unit 1 .

[0105] The processing unit 1 processes the object constituting the substrate W by using the processing liquid. Hereinafter, the object to be processed by the processing liquid is recorded as "object TG". The object TG is, for example, a substrate body (for example, a substrate body containing silicon), or a substance formed on the surface of the substrate body. The substance formed on the surface of the substrate body is, for example, a substance of the same material as the substrate body (for example, a layer containing silicon), or a substance of a different material from the substrate body (for example, a silicon oxide film, a silicon nitride film, or a resist). The "substance" may also constitute a film.

[0106] In this embodiment, the processing liquid includes an etching liquid, and the processing unit 1 performs etching processing. The object TG is processed (etched) by the etching liquid. The etching liquid is a chemical liquid. The etching liquid is, for example, hydrofluoric nitric acid (a mixture of hydrofluoric acid (HF) and nitric acid (HNO3)), hydrofluoric acid, buffered hydrofluoric acid (BHF), ammonium fluoride, HFEG (a mixture of hydrofluoric acid and ethylene glycol), or phosphoric acid (H3PO4).

[0107] like Figure 2As shown, the processing unit 1 includes a chamber 2, a spin chuck 3, a rotation motor unit 5, a nozzle moving mechanism 6, a thickness measuring unit 8, a probe moving mechanism 9, and a plurality of guards 10 ( Figure 2 The substrate processing apparatus 100 includes two protective covers 10, a first nozzle 41 and a second nozzle 71. The substrate processing apparatus 100 includes an etching liquid supply unit 4 and a rinse liquid supply unit 7. The etching liquid supply unit 4 includes a first supply pipe 42, and the rinse liquid supply unit 7 includes a second supply pipe 72.

[0108] The chamber 2 has a substantially box shape and accommodates the substrate W, the spin chuck 3, the spin motor unit 5, the nozzle moving mechanism 6, a plurality of protective covers 10, the thickness measuring unit 8, the probe moving mechanism 9, the first nozzle 41, the second nozzle 71, a portion of the first supply pipe 42, and a portion of the second supply pipe 72.

[0109] The rotary chuck 3 holds the substrate W horizontally. The rotary chuck 3 is an example of a substrate holding portion. Specifically, the rotary chuck 3 has a plurality of chuck members 32 and a rotary base 33. The plurality of chuck members 32 are arranged on the rotary base 33 along the periphery of the substrate W. The plurality of chuck members 32 hold the substrate W in a horizontal posture. The rotary base 33 is roughly disk-shaped and supports the plurality of chuck members 32 in a horizontal posture.

[0110] The rotary motor unit 5 rotates the substrate W and the rotary chuck 3 integrally around the first rotation axis AX1. The first rotation axis AX1 extends in the up-down direction. In the present embodiment, the first rotation axis AX1 extends in a substantially vertical direction. The first rotation axis AX1 is an example of a central axis, and the rotary motor unit 5 is an example of a substrate rotating unit. In detail, the rotary motor unit 5 rotates the rotary base 33 around the first rotation axis AX1. Therefore, the rotary base 33 rotates around the first rotation axis AX1. As a result, the substrate W held by the plurality of chuck members 32 provided on the rotary base 33 rotates around the first rotation axis AX1.

[0111] Specifically, the rotary motor unit 5 includes a motor body 51, a shaft 53, and an encoder 55. The shaft 53 is coupled to the rotary base 33. The motor body 51 rotates the shaft 53. As a result, the rotary base 33 rotates.

[0112] The encoder 55 is an example of a measuring device that measures a value of a measurement object, and the measurement object of the encoder 55 is the rotation speed of the substrate W. The encoder 55 generates a signal indicating the rotation speed of the substrate W. Specifically, the encoder 55 generates a rotation speed signal indicating the rotation speed of the motor body 51 .

[0113] The first nozzle 41 supplies the etching liquid to the substrate W. Specifically, the first nozzle 41 sprays the etching liquid toward the rotating substrate W. The etching liquid supply unit 4 supplies the etching liquid to the first nozzle 41. Specifically, the first nozzle 41 is connected to one end of the first supply pipe 42. The etching liquid is supplied to the first nozzle 41 via the first supply pipe 42. The first supply pipe 42 is a tubular member through which the etching liquid flows.

[0114] The nozzle moving mechanism 6 moves the first nozzle 41. The nozzle moving mechanism 6 is an example of a moving mechanism. In the present embodiment, the nozzle moving mechanism 6 moves the first nozzle 41 in a substantially horizontal direction. In detail, the nozzle moving mechanism 6 rotates the first nozzle 41 around the second rotation axis AX2 along the substantially vertical direction. The first nozzle 41 sprays the etching liquid toward the substrate W while moving (rotating). The first nozzle 41 is sometimes referred to as a scanning nozzle.

[0115] Specifically, the nozzle moving mechanism 6 has a nozzle arm 61, a first rotating shaft 63 and a first driving unit 65. The nozzle arm 61 extends in a substantially horizontal direction. The first nozzle 41 is arranged at the front end of the nozzle arm 61. The nozzle arm 61 is coupled to the first rotating shaft 63. The first rotating shaft 63 extends in a substantially vertical direction. The first driving unit 65 rotates the first rotating shaft 63 around the second rotating axis AX2, and rotates the nozzle arm 61 along a substantially horizontal plane around the first rotating shaft 63. As a result, the first nozzle 41 moves along a substantially horizontal plane. In detail, the first nozzle 41 revolves around the first rotating shaft 63 around the second rotating axis AX2. The first driving unit 65 includes, for example, a stepping motor.

[0116] The second nozzle 71 supplies the eluent to the substrate W. In detail, the second nozzle 71 sprays the eluent toward the rotating substrate W. The eluent supply unit 7 supplies the eluent to the second nozzle 71. In detail, the eluent is supplied to the second nozzle 71 via the second supply pipe 72. The second supply pipe 72 is a tubular member for the eluent to flow. The eluent is, for example, deionized water, carbonated water, electrolytic ion water, hydrogen water, ozone water, or hydrochloric acid water of a diluted concentration (for example, about 10ppm to 100ppm). The second nozzle 71 sprays the eluent in a stationary state. The second nozzle 71 is sometimes referred to as a fixed nozzle. In addition, the second nozzle 71 may also be a scanning nozzle.

[0117] Each of the protective covers 10 has a substantially cylindrical shape. The plurality of protective covers 10 blocks the etching liquid and the rinse liquid discharged from the substrate W.

[0118] The thickness measuring unit 8 measures the thickness of the object TG in a non-contact manner, and generates a thickness detection signal indicating the thickness of the object TG. The thickness detection signal is input to the control device 101 .

[0119] The thickness measuring unit 8 measures the thickness of the object TG by, for example, a spectroscopic interference method. Specifically, the thickness measuring unit 8 includes an optical probe 81, a signal line 83, and a thickness measuring device 85. The optical probe 81 has a lens. The signal line 83 connects the optical probe 81 and the thickness measuring device 85. The signal line 83 includes, for example, an optical fiber. The thickness measuring device 85 has a light source and a light receiving element. The light emitted by the light source of the thickness measuring device 85 is emitted to the object TG via the signal line 83 and the optical probe 81. The light reflected by the object TG is received by the light receiving element of the thickness measuring device 85 via the optical probe 81 and the signal line 83. The thickness measuring device 85 analyzes the light received by the light receiving element and calculates the thickness of the object TG. The thickness measuring device 85 generates a thickness detection signal representing the calculated thickness of the object TG.

[0120] The probe moving mechanism 9 moves the optical probe 81 in a substantially horizontal direction. Specifically, the probe moving mechanism 9 rotates the optical probe 81 around a third rotation axis AX3 in a substantially vertical direction. The optical probe 81 emits light toward the substrate W while moving (rotating). Therefore, the thickness detection signal indicates the thickness distribution of the object TG.

[0121] Specifically, the probe moving mechanism 9 has a probe arm 91, a second rotating shaft 93 and a second driving unit 95. The probe arm 91 extends in a substantially horizontal direction. The optical probe 81 is arranged at the front end of the probe arm 91. The probe arm 91 is coupled to the second rotating shaft 93. The second rotating shaft 93 extends in a substantially vertical direction. The second driving unit 95 rotates the second rotating shaft 93 around the third rotating axis AX3, thereby rotating the probe arm 91 along a substantially horizontal plane around the second rotating axis 93. As a result, the optical probe 81 moves along a substantially horizontal plane. In detail, the optical probe 81 revolves around the second rotating shaft 93 around the third rotating axis AX3. The second driving unit 95 includes, for example, a stepping motor.

[0122] In this embodiment, the thickness measuring unit 8 is an example of a processing amount detection unit, and is used to detect the processing amount. The processing amount indicates the amount of the substrate W processed by the processing unit 1. In detail, the processing amount indicates the amount of the substrate W processed by the etching liquid (processing liquid) supplied from the first nozzle 41 to the substrate W. In this embodiment, the processing amount indicates the etching amount. The etching amount indicates the difference between the thickness of the object TG before the etching process and the thickness of the object TG after the etching process.

[0123] When executing machine learning, the control device 101 calculates the etching amount based on the thickness detection signal input from the thickness measuring unit 8 (thickness measuring device 85), thereby acquiring the etching amount. More specifically, the control device 101 acquires the distribution of the etching amount (distribution of the processing amount). The control device 101 uses the acquired etching amount to generate learning data (learning data set) for machine learning.

[0124] Furthermore, the control device 101 calculates the etching amount based on the thickness detection signal input from the thickness measuring unit 8 (thickness measuring device 85) when manufacturing the semiconductor product, thereby acquiring the etching amount. The control device 101 determines whether to generate learning data for additional learning based on the acquired etching amount when manufacturing the semiconductor product. When the control device 101 determines to generate learning data for additional learning, it generates learning data for additional learning based on the acquired etching amount.

[0125] Furthermore, the control device 101 receives a rotation speed signal from the encoder 55. The rotation speed of the substrate W affects the etching amount. The control device 101 determines whether an abnormality has occurred in the substrate processing device 100 based on the rotation speed signal. Figure 8 As described, a recipe 131 for controlling each part of the substrate processing apparatus 100 is stored, and the recipe 131 indicates a set value of the rotation speed of the motor body 51. The set value of the rotation speed of the motor body 51 is an example of a set value for a measurement object. The control device 101 controls the processing performed by the processing unit 1 with reference to the recipe 131. When the difference between the rotation speed indicated by the rotation speed signal and the set value of the rotation speed of the motor body 51 included in the recipe 131 is greater than a predetermined value, the control device 101 determines that the rotation speed of the substrate W presents an abnormal value. When the control device 101 determines that the rotation speed of the substrate W presents an abnormal value, it determines that an abnormality has occurred in the substrate processing apparatus 100.

[0126] Next, refer to Figure 3 The scanning process of the substrate W by the first nozzle 41 will be described. Figure 3 FIG. 2 is a plan view showing the scanning process of this embodiment. Figure 3 As shown in FIG. 1 , the scanning process refers to a process in which the first nozzle 41 ejects the processing liquid onto the object TG while moving in a manner such that the processing liquid forms an arc-shaped trajectory TJ1 relative to the liquid landing position on the surface of the object TG when viewed from above. The trajectory TJ1 passes through the center portion CT of the substrate W. The center portion CT represents the portion of the substrate W through which the first rotation axis AX1 passes. The scanning process is performed while the substrate W is rotating.

[0127] In this embodiment, the first nozzle 41 moves from the first position X1 to the ninth position X9 while ejecting the etching liquid toward the rotating substrate W. Each position X1 to X9 included in the first position X1 to the ninth position X9 is included in the trajectory TJ1. The interval from the first position X1 to the ninth position X9 represents the movement interval of the first nozzle 41.

[0128] Among the first position X1 to the ninth position X9, the first position X1 indicates the starting position of the ejection of the processing liquid (etching liquid), and the ninth position X9 indicates the stopping position of the ejection of the processing liquid (etching liquid). The moving speed of the first nozzle 41 at the first position X1 is 0 mm / s, and the moving speed of the first nozzle 41 at the ninth position X9 is 0 mm / s. Therefore, the first position X1 is the starting position of the scanning process, and the ninth position X9 is the ending position of the scanning process. Moreover, the first position X1 is the starting position of the movement of the first nozzle 41, and the ninth position X9 is the ending position of the movement of the first nozzle 41. In addition, in the following description, the moving speed of the first nozzle 41 during the scanning process is sometimes recorded as "scanning speed".

[0129] The first nozzle 41 passes through each intermediate position between the first position X1 and the ninth position X9 (positions X2 to X8 from the second position X2 to the eighth position X8) during scanning. Each intermediate position is an example of a nozzle position, and the movement section of the first nozzle 41 is divided into a plurality of sections.

[0130] Next, refer to Figure 4 The scanning speed information is described below. The scanning speed information indicates a set value of the moving speed of the first nozzle 41 during scanning processing (a set value of the scanning speed). Figure 4 is a diagram showing scanning speed information of this embodiment. Scanning speed information is an example of speed information, learning object speed information, and processing speed information. Figure 4 Indicates reference Figure 3 The relationship between the positions X1 to X9 included in the movement section of the first nozzle 41 and the setting value of the scanning speed is described.

[0131] Figure 4 In the figure, the upper column indicates each position X1 to X9 included in the moving interval of the first nozzle 41, and the lower column indicates the set value of the scanning speed. Each position X1 to X9 included in the moving interval of the first nozzle 41 is defined by the radial position of the substrate W. Specifically, the upper column indicates the starting position of the moving interval of the first nozzle 41 (the moving starting position of the first nozzle 41), the ending position of the moving interval of the first nozzle 41 (the moving ending position of the first nozzle 41), and multiple intermediate positions between the starting position and the ending position of the moving interval of the first nozzle 41 (multiple positions passed by the first nozzle 41).

[0132] like Figure 4 As shown, the scanning speed information indicates the setting value of the scanning speed for each of the positions X1 to X9 included in the moving interval of the first nozzle 41. Hereinafter, each of the positions X1 to X9 included in the moving interval of the first nozzle 41 is sometimes described as a "speed setting position". In the present embodiment, the scanning speed information indicates nine speed setting positions.

[0133] Specifically, each speed setting position corresponds to the reference Figure 3 The positions X1 to X9 from the first position X1 to the ninth position X9 described above are as follows. Figure 3 As described, the scanning speed set at the start position (first position X1) of the movement interval of the first nozzle 41 is 0 [mm / s], and the scanning speed set at the end position (ninth position X9) of the movement interval of the first nozzle 41 is 0 [mm / s].

[0134] Reference Figure 1 and Figure 2 The control device 101 described here refers to the scanning speed information based on the scanning speed information. Figure 2 The nozzle moving mechanism 6 (first drive unit 65) is controlled. As a result, the first nozzle 41 moves along the reference line so that the scanning speed at each speed setting position becomes the scanning speed specified by the scanning speed information. Figure 3 The illustrated trajectory TJ1 moves.

[0135] Next, refer to Figure 5 The moving speed (scanning speed) of the first nozzle 41 during the scanning process will be described. Figure 5 This is a graph showing an example of the moving speed of the first nozzle 41 in this embodiment.

[0136] Figure 5 In FIG. 1 , the vertical axis represents the scanning speed [mm / s], and the horizontal axis represents the radial position [mm] of the substrate W. Figure 5 As shown, the scanning speed at the start of the scanning process is 0 [mm / s]. Also, the scanning speed at the end of the scanning process is 0 [mm / s].

[0137] As reference Figure 3 and Figure 4 As described above, the scanning speed is set for each of the positions X1 to X9 (each speed setting position) from the first position X1 to the ninth position X9. Figure 5 As shown in FIG. 1 , between adjacent speed setting positions, the scanning speed continuously changes from the scanning speed set for one speed setting position to the scanning speed set for another speed setting position. Figure 4As shown, the scanning speed Y3 is set for the third position X3, and the scanning speed Y4 is set for the fourth position X4. Therefore, the scanning speed of the first nozzle 41 continuously changes from the scanning speed Y3 to the scanning speed Y4 during the period when the first nozzle 41 moves from the third position X3 to the fourth position X4.

[0138] Next, refer to Figure 6 The thickness measurement process performed by the thickness measuring unit 8 will be described. Figure 6 2 is a plan view showing the thickness measurement process of this embodiment. Figure 6 As shown in FIG. 1 , the so-called thickness measurement process refers to a process in which the optical probe 81 measures the thickness of the object TG while moving in a manner that forms an arc-shaped trajectory TJ2 with respect to the thickness measurement position of the object TG when viewed from above. The trajectory TJ2 passes through the edge portion EG of the substrate W and the center portion CT of the substrate W. The edge portion EG indicates the peripheral portion of the substrate W. The thickness measurement process is performed while the substrate W is rotating.

[0139] Specifically, the optical probe 81 emits light toward the object TG while moving between the center portion CT and the edge portion EG of the substrate W in a top view. As a result, the thickness of the object TG is measured at each measurement position included in the trajectory TJ2. Each measurement position corresponds to each radial position of the substrate W. Therefore, the thickness distribution of the object TG in the radial direction RD of the substrate W is measured by the thickness measurement process.

[0140] Next, refer to Figure 7 The etching liquid supply unit 4 of this embodiment will be described. Figure 7 Schematic diagram of the etching liquid supply unit 4 of this embodiment. Figure 7 As shown, the etching liquid supply unit 4 has a reference Figure 2 In addition to the first supply pipe 42 described above, the first supply pipe 42 further includes a temperature sensor 421 , a concentration sensor 422 , a valve 423 , a mixing valve 424 , a flow meter 425 , and a heater 426 .

[0141] The temperature sensor 421 is an example of a measuring device that measures the value of a measurement object, and the measurement object of the temperature sensor 421 is the temperature of the etching liquid flowing through the first supply pipe 42. Hereinafter, the temperature of the etching liquid flowing through the first supply pipe 42 is described as "the temperature of the etching liquid". The temperature sensor 421 generates a temperature signal indicating the temperature of the etching liquid. The temperature of the etching liquid affects the etching amount.

[0142] The temperature signal is input to the reference Figure 1 and Figure 2 The control device 101 described above determines whether an abnormality has occurred in the substrate processing apparatus 100 based on the temperature signal. Figure 8 The recipe 131 described here indicates a set value of the temperature of the etching liquid. The set value of the temperature of the etching liquid is an example of a set value for the measurement object. When the difference between the temperature indicated by the temperature signal and the set value of the temperature of the etching liquid contained in the recipe 131 becomes greater than a predetermined value, the control device 101 determines that the temperature of the etching liquid has shown an abnormal value. When the control device 101 determines that the temperature of the etching liquid has shown an abnormal value, it determines that an abnormality has occurred in the substrate processing device 100.

[0143] The concentration sensor 422 is an example of a measuring device that measures the value of a measurement object, and the measurement object of the concentration sensor 422 is the concentration of the etching component contained in the etching liquid flowing through the first supply pipe 42. Hereinafter, the concentration of the etching component contained in the etching liquid flowing through the first supply pipe 42 is described as "the concentration of the etching liquid". The concentration sensor 422 generates a concentration signal indicating the concentration of the etching liquid. The concentration of the etching liquid affects the etching amount.

[0144] The concentration signal is input to the reference Figure 1 and Figure 2 The control device 101 described above determines whether an abnormality has occurred in the substrate processing apparatus 100 based on the concentration signal. Figure 8 The recipe 131 described here indicates a set value of the concentration of the etching solution. The set value of the concentration of the etching solution is an example of a set value for the measurement object. When the difference between the concentration indicated by the concentration signal and the set value of the concentration of the etching solution contained in the recipe 131 becomes greater than a predetermined value, the control device 101 determines that the concentration of the etching solution has an abnormal value. When the control device 101 determines that the concentration of the etching solution has an abnormal value, it determines that an abnormality has occurred in the substrate processing device 100.

[0145] The valve 423 is arranged in the first supply piping 42. The valve 423 switches the supply of the etching liquid to the first nozzle 41 and stops the supply. In detail, when the valve 423 is opened, the etching liquid is sprayed from the first nozzle 41 toward the substrate W. On the other hand, when the valve 423 is closed, the spraying of the etching liquid stops. In addition, the valve 423 controls the flow rate of the etching liquid flowing downstream of the valve 423 in the first supply piping 42. In detail, the flow rate of the etching liquid flowing downstream of the valve 423 is adjusted corresponding to the opening of the valve 423. Therefore, the spraying flow rate of the etching liquid is adjusted corresponding to the opening of the valve 423. The valve 423 is, for example, a motor valve.

[0146] The mixing valve 424 is disposed in the first supply pipe 42. When the mixing valve 424 is opened, pure water flows into the first supply pipe 42, and the concentration of the etching solution is diluted.

[0147] The flow meter 425 is an example of a measuring device that measures the value of a measurement object, and the measurement object of the flow meter 425 is the ejection flow rate of the etching solution. Specifically, the measurement object of the flow meter 425 is the flow rate of the etching solution flowing through the first supply pipe 42. The flow meter 425 generates an ejection flow rate signal indicating the ejection flow rate of the etching solution. The ejection flow rate of the etching solution affects the etching amount.

[0148] The discharge flow rate signal is input to the reference Figure 1 and Figure 2 The control device 101 described above determines whether an abnormality has occurred in the substrate processing apparatus 100 based on the ejection flow rate signal. Figure 8 The recipe 131 described here indicates a set value for the ejection flow rate of the etching liquid. The set value for the ejection flow rate of the etching liquid is an example of a set value for the measurement object. When the difference between the ejection flow rate indicated by the ejection flow rate signal and the set value for the ejection flow rate of the etching liquid contained in the recipe 131 becomes greater than a predetermined value, the control device 101 determines that the ejection flow rate of the etching liquid presents an abnormal value. When the control device 101 determines that the ejection flow rate of the etching liquid presents an abnormal value, it determines that an abnormality has occurred in the substrate processing device 100.

[0149] The heater 426 heats the etching liquid flowing through the first supply pipe 42 .

[0150] Next, refer to Figure 8 The control device 101 will be described. Figure 8 1 is a block diagram of the control device 101 of this embodiment. Figure 8 1 represents the control device 101 before machine learning is performed. Figure 8 As shown, the control device 101 includes a control unit 102 , a storage unit 103 , an input unit 104 , and a display unit 105 .

[0151] The control unit 102 has a processor. For example, the control unit 102 has a central processing unit (CPU) or a microprocessor (MPU). Alternatively, the control unit 102 has a general-purpose computing unit. The control unit 102 may also have a neural network processing unit (NCU).

[0152] The storage unit 103 stores data and computer programs. The storage unit 103 has a main storage device. The main storage device is, for example, a semiconductor memory. The storage unit 103 may also have an auxiliary storage device. The auxiliary storage device is, for example, a semiconductor memory and / or a hard disk drive. The storage unit 103 may also have a removable medium. The control unit 102 controls the operation of each unit of the substrate processing device 100 based on the data and computer programs stored in the storage unit 103. Furthermore, the control unit 102 performs machine learning and additional learning based on the data and computer programs stored in the storage unit 103.

[0153] Specifically, the storage unit 103 stores a recipe 131, a control program 132, an estimation program 133, and a learning program 134. The recipe 131 specifies the processing content and processing flow of the substrate W. In addition, the recipe 131 indicates various setting values. For example, the recipe 131 indicates a reference Figure 4 Scan speed information described.

[0154] The control unit 102 controls the operation of each unit of the substrate processing apparatus 100 based on the recipe 131 and the control program 132. The estimation program 133 is used to refer to the Fig.15 Generation of the learned model described above The learning program 134 is a program for executing an algorithm for finding a certain rule from a learning data set and generating a model (learned model) that expresses the rule.

[0155] The storage unit 103 stores the learning object speed information #A1 to the learning object speed information #An when the machine learning is executed. Here, "n" represents a positive integer. The learning object speed information #A1 to the learning object speed information #An are respectively stored in the reference Figure 4 The scanning speed information described above also indicates scanning speed information. Figure 4 The setting contents of the scanning speeds set in the speed setting positions described above are different from each other.

[0156] The input unit 104 receives input from the operator and outputs information indicating the input result to the control unit 102. For example, the input unit 104 receives input of learning object speed information #A1 to learning object speed information #An. The input unit 104 includes, for example, a touch panel and a pointing device. The touch panel is, for example, arranged on the display surface of the display unit 105. The input unit 104 and the display unit 105 constitute, for example, a graphical user interface.

[0157] The display unit 105 displays various information. In this embodiment, the display unit 105 displays, for example, various error screens and various setting screens (input screens). The display unit 105 has, for example, a liquid crystal display or an organic electroluminescence (EL) display.

[0158] In addition, the control device 101 may also have an interface for acquiring the learning object speed information #A1 to the learning object speed information #An from a removable memory. The removable memory includes, for example, a Universal Serial Bus (USB) memory. Alternatively, the control device 101 may also have a drive for acquiring the learning object speed information #A1 to the learning object speed information #An from a removable medium. The removable medium includes, for example, an optical disc such as a Compact Disc (CD).

[0159] Next, refer to Figure 1 , Figure 2 , Figure 8 and Fig. 9 A method of generating learning data executed by the substrate processing apparatus 100 will be described. Fig. 9 : is a flowchart showing a method for generating learning data in this embodiment. Fig. 9 The following is a diagram showing the processing executed by the control unit 102 when generating the learning data. The method for generating the learning data according to the present embodiment includes the processing of steps S1 to S9.

[0160] Fig. 9 The process shown is started by the operator operating the input unit 104. When the learning data (learning data set) is generated, a plurality of substrates W as learning objects are accommodated in at least one of the plurality of load ports LP. In addition, as the learning object substrates W, the same type of substrates W as the substrates W used in the manufacture of semiconductor products are used.

[0161] When generating learning data (learning data set), the control unit 102 first selects one of the plurality of processing units 1 as the learning target processing unit 1. Fig. 9 The described process is sequentially performed on the plurality of processing units 1 included in the substrate processing apparatus 100 .

[0162] When the learning object processing unit 1 is determined, the control unit 102 selects one of the learning object speed information #A1 to the learning object speed information #An, and obtains one of the learning object speed information #A1 to the learning object speed information #An (step S1). Hereinafter, one of the learning object speed information #A1 to the learning object speed information #An is recorded as "learning object speed information #A". For example, the control unit 102 selects the learning object speed information #A in the order of the identification numbers respectively assigned to the learning object speed information #A1 to the learning object speed information #An from small to large.

[0163] After selecting one of the learning object speed information #A1 to the learning object speed information #An, the control unit 102 controls the handling robot IR and the central robot CR to carry the learning object substrate W into the chamber 2 of the learning object processing unit 1 (step S2). The control unit 102 holds the learning object substrate W carried into the chamber 2 on the spin chuck 3.

[0164] When the learning object substrate W is held on the spin chuck 3, the control unit 102 causes the thickness measuring unit 8 to measure the thickness distribution of the object TG contained in the learning object substrate W (step S3). The thickness distribution of the object TG measured here means the thickness distribution of the object TG before the etching process. Hereinafter, the thickness distribution of the object TG before the etching process is sometimes described as "thickness distribution before the process".

[0165] After the thickness distribution before processing is measured, the control unit 102 controls the operation of each unit of the substrate processing apparatus 100 to perform etching processing on the learning target substrate W (step S4). Figure 3 to Figure 5 As described above, the operations of the various parts of the substrate processing apparatus 100 are controlled so that the first nozzle 41 supplies etching liquid to the learning target substrate W while moving (spinning) at a speed based on the learning target speed information #A, thereby performing etching processing on the learning target substrate W.

[0166] After the etching process is executed, the control unit 102 causes the thickness measuring unit 8 to measure the thickness distribution of the object TG contained in the learning object substrate W (step S5). The thickness distribution of the object TG measured here means the thickness distribution of the object TG after the etching process. Hereinafter, the thickness distribution of the object TG after the etching process is sometimes described as "thickness distribution after the process". Fig.10 As described above, the thickness distribution after treatment means the thickness distribution after the drying treatment.

[0167] After measuring the thickness distribution after processing, the control unit 102 releases the spin chuck 3 from holding the learning object substrate W, and causes the handling robot IR to carry out the learning object substrate W from the chamber 2 (step S6). Subsequently, the control unit 102 controls the handling robot IR and the central robot CR to carry the learning object substrate W to one of the plurality of loading ports LP.

[0168] Next, the control unit 102 acquires the processing amount (etching amount) (step S7 ). Specifically, the control unit 102 calculates the difference between the thickness distribution before the processing and the thickness distribution after the processing to acquire the etching amount.

[0169] When the processing amount (etching amount) is acquired, the control unit 102 generates learning data based on the learning target speed information #A and the processing amount (etching amount) (step S8 ). The learning data represents the learning target speed information #A and the processing amount (etching amount). The learning data is stored in the storage unit 103 .

[0170] When the learning data is generated, the control unit 102 determines whether an abnormality has occurred in the substrate processing apparatus 100 (step S9), and ends the Fig. 9 For example, the control unit 102 refers to Figure 2 and Figure 7 As described above, it is determined whether or not at least one of the rotation speed of the substrate W, the temperature of the etching solution, the concentration of the etching solution, and the ejection flow rate of the etching solution has an abnormal value.

[0171] The control unit 102 repeatedly executes Fig. 9 The processing shown is performed until all of the learning object speed information #A1 to the learning object speed information #An are selected. As a result, a learning data set is generated. The learning data set includes a plurality of learning data. The learning data set is stored in the storage unit 103. In addition, at least one of the processes of step S7 to step S9 may be performed between the process of step S5 and the process of step S6.

[0172] Next, refer to Figure 2 , Figure 8 and Fig.10 The etching process (step S4) will be described. Fig.10 : is a flowchart showing the etching process of this embodiment.

[0173] like Fig.10 As shown in FIG. 1 , after the spin chuck 3 holds the learning object substrate W, the learning object substrate W is processed (step S41). Specifically, the control unit 102 causes the rotation motor unit 5 to rotate the substrate W held by the spin chuck 3. Then, the control unit 102 Fig. 9As described above, the nozzle moving mechanism 6 and the etching liquid supply unit 4 are controlled so that the etching liquid is supplied from the first nozzle 41 toward the training target substrate W. As a result, the training target substrate W is etched.

[0174] When the etching of the object substrate W is completed, the control unit 102 controls the eluent supply unit 7 to supply the eluent to the object substrate W, thereby removing the etching liquid from the object substrate W (step S42). Specifically, the etching liquid is flushed to the outside of the object substrate W by the eluent, and is discharged to the periphery of the object substrate W. As a result, the liquid film of the eluent on the object substrate W is replaced by the liquid film of the eluent.

[0175] After replacing the etching liquid with the rinse liquid, the control unit 102 controls the rotary motor unit 5 to dry the learning target substrate W (step S43). Fig.10 The processing shown is completed. Specifically, the control unit 102 increases the rotation speed of the object substrate W compared to the rotation speed during the etching process and the rinsing process. As a result, a large centrifugal force is applied to the rinsing liquid on the object substrate W, and the rinsing liquid attached to the object substrate W is thrown around the object substrate W. In this way, the rinsing liquid is removed from the object substrate W, and the object substrate W is dried. In addition, the control unit 102 stops the rotation of the substrate W by the rotation motor unit 5, for example, after a specified time has passed since the high-speed rotation of the object substrate W was started.

[0176] Next, refer to Figure 2 , Figure 8 and Fig.11 The process of generating learning data (step S8) will be described. Fig.11 : is a flowchart showing the process of generating learning data in this embodiment.

[0177] like Fig.11 As shown, the control unit 102 generates learning data using the learning target speed information #A and the processing amount (etching amount) (step S81). The control unit 102 determines whether the generated learning data (processing amount) can be used for machine learning (step S82).

[0178] Specifically, the control unit 102 determines whether the processing amount (etching amount) is greater than a specified value. If the processing amount (etching amount) is not greater than the specified value, the control unit 102 determines that the learning data (processing amount) can be used for machine learning. On the other hand, if the processing amount (etching amount) is greater than the specified value, the control unit 102 determines that the learning data (processing amount) cannot be used for machine learning. For example, the control unit 102 determines that the learning data (processing amount) cannot be used for machine learning when at least one point of the processing amount (etching amount) distribution is greater than the specified value.

[0179] In addition, the control unit 102 may also determine that the learning data (processing amount) cannot be used for machine learning when the processing amount (etching amount) distribution is above a specified value in multiple places, and may also determine that the learning data (processing amount) cannot be used for machine learning when the range of the processing amount (etching amount) above the specified value is above a predetermined range.

[0180] When it is determined that the learning data (processing amount) can be used for machine learning (Yes in step S82), the control unit 102 assigns a first flag to the learning data (step S83), and ends the process. Fig.11 Specifically, the control unit 102 associates the learning data with the first flag and stores it in the storage unit 103. The first flag indicates that the learning data is used.

[0181] On the other hand, when it is determined that the learning data (processing amount) cannot be used for machine learning (No in step S82), the control unit 102 assigns a second tag to the learning data (step S84), and ends the process. Fig.11 Specifically, the control unit 102 associates the learning data with the second label and stores them in the storage unit 103. The second label indicates that the learning data is not used. Therefore, the learning data to which the second label is assigned is excluded from the learning data set used for machine learning.

[0182] Next, refer to Figure 2 , Figure 8 and Fig.12 The abnormality determination process (step S9 ) for determining whether an abnormality has occurred in the substrate processing apparatus 100 will be described. Fig.12 : is a flowchart showing the abnormality determination process in this embodiment.

[0183] like Fig.12 As shown, the control unit 102 determines whether an abnormality occurs in the substrate processing apparatus 100 (step S91). When the control unit 102 determines that no abnormality occurs in the substrate processing apparatus 100 (No in step S91), the process ends. Fig.12 Processing shown.

[0184] On the other hand, when the control unit 102 determines that an abnormality has occurred in the substrate processing apparatus 100 (Yes in step S91), different error notifications are performed depending on whether the learning data (processing volume) can be used for machine learning. In this embodiment, the control unit 102 determines whether the learning data is associated with the first tag or the second tag (step S92).

[0185] If the first mark is associated with the learning data (A in step S92), the control unit 102 causes the display unit 105 to display the first error screen (step S93), and the process ends. Fig.12The first error screen indicates that an abnormality has occurred in the substrate processing apparatus 100. Specifically, the first error screen indicates the measurement object (monitoring object) in which an abnormality has occurred.

[0186] If the second mark is associated with the learning data (B in step S92), the control unit 102 causes the display unit 105 to display a second error screen (step S94), and the process ends. Fig.12 The second error screen corresponds to a situation where an abnormality occurs in the substrate processing apparatus 100, and indicates that the processing amount (etching amount) is greater than a specified value. Therefore, the second error screen can notify the operator of the possibility that the reason why the processing amount (etching amount) is greater than the specified value is not on the learning object substrate W side but on the substrate processing apparatus 100 side.

[0187] Next, refer to Fig.13 The learning data set table TB10 for managing the learning data set will be described. Fig.13 1 is a diagram showing an example of the learning data set table TB10 according to the present embodiment.

[0188] In this embodiment, the control unit 102 stores the learning data in the storage unit 103 in a table format. Fig.13 As shown, the learning data set table TB10 includes a speed information field TB11 , a processing amount field TB12 , and a flag field TB13 .

[0189] In the speed information column TB11, the learning object speed information #A is stored. In the processing amount column TB12, the processing amount (etching amount) is stored. In the mark column TB13, the mark is stored. Fig.13 As shown, the learning target speed information #A, the processing amount (etching amount), and the flag are stored in the learning data set table TB10 in association with each other.

[0190] In the mark column TB13, "OK" indicates a first mark and "NG" indicates a second mark. In other words, "OK" indicates that the corresponding learning data can be used for machine learning, and "NG" indicates that the corresponding learning data cannot be used for machine learning.

[0191] Next, refer to Fig.14 The control device 101 after the learning data set 135 is generated will be described. Fig.14 is a block diagram of the control device 101 of this embodiment. Fig.14 The control device 101 after the learning data set 135 is generated is shown.

[0192] like Fig.14 As shown in FIG. 1 , the storage unit 103 stores a learning data set 135. The learning data set 135 includes a plurality of learning data. Fig.13 As described above, the control unit 102 stores a plurality of learning data in the learning data set table TB10. After generating the learning data set 135, the control unit 102 may or may not erase the learning object speed information #A1 to the learning object speed information #An from the storage unit 103.

[0193] The control unit 102 performs learning on the learning data set 135 (a plurality of learning data) to generate learned parameters. The control unit 102 generates a learned model based on the learned parameters.

[0194] Next, refer to Figure 1 , Figure 2 , Fig.13 , Fig.14 and Fig.15 A method of generating a learned model executed by the substrate processing apparatus 100 will be described. Fig.15 : is a flowchart showing a method for generating a learning-completed model in this embodiment. Fig.15 The following is a process executed by the control unit 102 when generating a learned model. The method for generating a learned model according to the present embodiment includes the processes of step S11 to step S14.

[0195] First, the control unit 102 acquires the learning data set 135 from the storage unit 103 (step S11). At this time, the control unit 102 acquires the learning data set 135 to which the first label is assigned from the plurality of learning data.

[0196] When the control unit 102 acquires the learning data set 135, it learns the learning data set 135 (a plurality of learning data) based on the learning program 134 (step S12). Here, the so-called learning refers to finding a certain rule from the learning data set 135. In this embodiment, the learning data represents the relationship between the scanning speed information (the scanning speed at each speed setting position) and the processing amount (etching amount). The control unit 102 learns the learning data set 135 to find a certain rule between the scanning speed information and the processing amount (etching amount).

[0197] The control unit 102 generates learned parameters by learning the learning data set 135 (step S13). More specifically, the learned parameters are output from the learning program 134. The learned parameters are parameters (coefficients) acquired based on the results of learning using the learning data set 135.

[0198] Next, the control unit 102 generates a learned model based on the acquired learned parameters (step S14). Fig.15The processing shown is completed. In detail, the control unit 102 incorporates the learned parameters into the inference program 133, thereby generating a learned model. The inference program 133 is a program that outputs a certain result relative to the input by applying the incorporated learned parameters, and the learned model is a model that includes the learned parameters. In this embodiment, the "input" is the target amount of processing volume (etching amount) and the "output" is the scanning speed information. In other words, the target amount of processing volume (etching amount) is the explanatory variable, and the scanning speed information output from the learned model is the target variable. In the following, the target amount of processing volume (etching amount) is sometimes recorded as "target processing volume" or "target etching amount". In addition, the scanning speed information output from the learned model is sometimes recorded as "speed information during processing".

[0199] Next, refer to Figure 2 and Fig.16 Next, the control device 101 used in the manufacture of semiconductor products will be described. Fig.16 is a block diagram of the control device 101 of this embodiment. Fig.16 A control device 101 is shown during the manufacture of semiconductor products.

[0200] like Fig.16 As shown, during the manufacture of semiconductor products, the storage unit 103 stores the learned model 136. After the learned model 136 is generated, the control unit 102 may or may not erase the learning data set 135 from the storage unit 103.

[0201] The control unit 102 controls the nozzle moving mechanism 6 based on the learned model 136 when manufacturing semiconductor products. In detail, the control unit 102 inputs the target processing amount (target etching amount) as an explanatory variable into the learned model 136, thereby outputting the processing speed information as the target variable from the learned model 136. When manufacturing semiconductor products, the control unit 102 controls the nozzle moving mechanism 6 so that the first nozzle 41 moves at a speed based on the processing speed information. Specifically, when the first nozzle 41 performs processing (etching) on ​​the substrate W, it moves (rotates) at a speed based on the processing speed information. In addition, in the following description, the substrate W processed when manufacturing the semiconductor product is sometimes described as a "processing object substrate W".

[0202] Next, refer to Figure 1 , Figure 2 , Fig.16 , Fig.17 and Fig.18 A substrate processing method performed by the substrate processing apparatus 100 will be described. Fig.17 and Fig.18is a flow chart showing a substrate processing method in this embodiment. Fig.17 and Fig.18 1 and 2 show the processes executed by the control unit 102 when etching the processing target substrate W. The substrate processing method of the present embodiment includes the processes of step S21 to step S32.

[0203] Fig.17 and Fig.18 The process shown is started by an operator operating the input unit 104. A plurality of substrates W to be processed are accommodated in at least one of the plurality of load ports LP.

[0204] When etching the processing target substrate W, first, the control unit 102 obtains the target thickness of the target object TG (step S21 ). Specifically, the control unit 102 obtains the target thickness distribution of the target object TG. The target thickness is input to the control unit 102 by the operator operating the input unit 104 .

[0205] When the target thickness is obtained, the control unit 102 controls the handling robot IR and the central robot CR to carry the processing target substrate W into the chamber 2 of one of the plurality of processing units 1 (step S22 ). The control unit 102 holds the processing target substrate W carried into the chamber 2 on the spin chuck 3 .

[0206] When the processing target substrate W is held on the spin chuck 3, the control unit 102 causes the thickness measuring unit 8 to measure the thickness distribution of the object TG included in the processing target substrate W (step S23). In other words, the thickness distribution before processing is measured.

[0207] Next, the control unit 102 obtains the target processing amount (target etching amount) as an explanatory variable based on the thickness distribution before processing and the target thickness distribution (step S24). Specifically, the control unit 102 calculates the difference between the thickness distribution before processing and the target thickness distribution to obtain the target processing amount (target etching amount).

[0208] Next, the control unit 102 inputs the target processing amount (target etching amount) into the learned model 136 and outputs the processing speed information from the learned model 136 to thereby acquire the processing speed information as a target variable (step S25 ).

[0209] When the processing speed information is obtained, the control unit 102 controls the operation of each unit of the substrate processing apparatus 100 to perform etching processing on the processing target substrate W (step S26). Figures 3 to 5As described above, the operations of the various parts of the substrate processing apparatus 100 are controlled so that the first nozzle 41 is moved (rotated) at a speed based on the speed information during processing, while the etching liquid is supplied from the first nozzle 41 to the processing target substrate W, thereby performing etching processing on the processing target substrate W. Fig.10 The processing target substrate W is etched in the same processing as described above.

[0210] After the etching process is executed, the control unit 102 causes the thickness measuring unit 8 to measure the thickness distribution of the object TG included in the process target substrate W (step S27). In other words, the thickness distribution after the process is measured.

[0211] After measuring the thickness distribution after processing, the control unit 102 releases the spin chuck 3 from holding the processing target substrate W, and causes the handling robot IR to carry out the processing target substrate W from the chamber 2 (step S28). Then, the control unit 102 controls the handling robot IR and the central robot CR to carry the processing target substrate W to one of the plurality of load ports LP.

[0212] When the substrate W to be processed is transported, Fig.18 As shown, the control unit 102 acquires the processing amount (etching amount) (step S29). Specifically, the control unit 102 calculates the difference between the thickness distribution before the processing and the thickness distribution after the processing to acquire the etching amount.

[0213] When the control unit 102 acquires the processing amount (etching amount), it determines whether the processing amount (etching amount) matches the target processing amount (target etching amount) (step S30 ).

[0214] If the control unit 102 determines that the processing amount (etching amount) is inconsistent with the target processing amount (target etching amount) (No in step S30), the process moves to step S31. If the control unit 102 determines that the processing amount (etching amount) is consistent with the target processing amount (target etching amount) (Yes in step S30), the process moves to step S32. In addition, the control unit 102 may also determine whether the processing amount (etching amount) is consistent with the target processing amount (target etching amount) within the allowable range.

[0215] In step S31, the control unit 102 generates learning data for additional learning based on the processing speed information and the processing amount (etching amount). Fig.11 The same processing as described above generates learning data for additional learning. When the learning data for additional learning is generated, the process proceeds to step S32.

[0216] In step S32, the control unit 102 determines whether an abnormality has occurred in the substrate processing apparatus 100 (step S32), and ends Fig.17 and Fig.18 Specifically, by referring to Fig.12 The same processing as the processing described above is used to determine whether an abnormality has occurred in the substrate processing apparatus 100 .

[0217] During the manufacture of semiconductor products, the control unit 102 repeatedly executes Fig.17 and Fig.18 As a result, a learning data set for additional learning is generated. The learning data set for additional learning is stored in the storage unit 103. In addition, at least one of the processes of step S29 to step S32 may also be performed between step S26 and step S27. Hereinafter, the learning data set for additional learning is sometimes described as "data set for additional learning". Moreover, the learning data for additional learning is sometimes described as "data for additional learning".

[0218] Next, refer to Fig.19 The additional learning data set table TB20 for managing the additional learning data set will be described. Fig.19 1 is a diagram showing an example of the additional learning data set table TB20 according to the present embodiment.

[0219] In this embodiment, the control unit 102 stores the additional learning data in the storage unit 103 in a table format. Fig.19 As shown, the additional learning data set table TB20 includes a speed information column TB21 , a processing amount column TB22 , and a flag column TB23 .

[0220] The speed information column TB21 stores the processing speed information #B. The processing amount column TB22 stores the processing amount (etching amount). Fig.13 The flag column TB13 described above similarly stores flags. Fig.19 As shown, the processing speed information #B, the processing amount (etching amount), and the flag are stored in association with each other in the additional learning data set table TB20.

[0221] When generating the additional learning data set, the control unit 102 generates a learned model in the same manner as when executing machine learning. Fig.15 Next, a method for generating a learned model after additional learning is described.

[0222] First, the control unit 102 acquires an additional learning data set from the storage unit 103 (step S11). At this time, the control unit 102 acquires additional learning data to which the first label is given from a plurality of additional learning data.

[0223] When the control unit 102 acquires the additional learning data set, it learns the additional learning data set (a plurality of additional learning data) based on the learning program 134 (step S12 ).

[0224] The control unit 102 generates new learned parameters by learning the additional learning data set (step S13 ). More specifically, the new learned parameters are output from the learning program 134 .

[0225] Next, the control unit 102 generates a learned model after additional learning based on the new learned parameters (step S14). Fig.15 Specifically, the control unit 102 incorporates new learned parameters into the estimation program 133 to thereby generate a learned model after additional learning.

[0226] Above, refer to Figures 1 to 19 Embodiment 1 of the present invention is described. According to this embodiment, the burden on operators involved in the development of scanning speed information can be reduced. Specifically, a learning model 136 is generated using various learning object substrates W with different pre-processing, so that even if the pre-processing is changed during the manufacturing stage of the semiconductor product, the operator does not need to develop new scanning speed information.

[0227] Moreover, according to the present embodiment, even if there is a deviation in the thickness of the object TG before processing, the learned model 136 will output scanning speed information for setting the thickness of the object TG after processing to the target thickness. Therefore, it is not necessary to develop a plurality of scanning speed information according to the thickness deviation of the object TG before processing. Moreover, even if there is a deviation in the thickness of the object TG before processing, the thickness distribution after processing can be stably set to the target thickness distribution. For example, the surface of the substrate W can be made substantially parallel to the horizontal plane. Alternatively, the thickness of a specific area in the object TG can be set to the target thickness.

[0228] Furthermore, according to the present embodiment, during the manufacture of semiconductor products, even if an error occurs that the thickness of the object TG after processing does not reach the target thickness, since the substrate processing apparatus 100 performs additional learning, it is still possible to adjust the learned parameters during the manufacture of the semiconductor product after the error occurs, thereby setting the thickness after processing to the target thickness. Furthermore, since the substrate processing apparatus 100 performs additional learning, the operator does not need to develop new scanning speed information in response to the occurrence of the error. Furthermore, even if the pre-processing is changed, since the learned parameters are adjusted by additional learning, the operator does not need to develop new scanning speed information.

[0229] In addition, the factors that affect the processing volume (etching volume) include, in addition to the scanning speed, the rotation speed of the substrate W, the temperature of the processing liquid, the concentration of the processing liquid, and the ejection flow rate of the processing liquid. However, the control unit 102 does not actively perform processing (control) for changing the values ​​of the other factors when generating learning data for machine learning and processing the processing object substrate W. In other words, the control unit 102 adjusts the scanning speed in order to set the processing volume as the target volume, and on the other hand, does not adjust (change) the values ​​of the other factors. Therefore, when generating learning data and processing the processing object substrate W, the values ​​of the other factors present approximately the same values ​​except when an abnormality occurs. In this embodiment, the speed information during processing is selected as the target variable from the multiple factors that affect the processing volume because when the homogenization of the processing volume distribution is set as the evaluation index, the correlation between the scanning speed and the evaluation index is stronger than the correlation between the other factors other than the scanning speed and the evaluation index. According to this embodiment, in order to generate learning data, it is sufficient to create a data set of the learning object speed information #A (scanning speed information), so the burden on the operator involved in the development of the learning completed model or scanning speed information can be reduced.

[0230] [Implementation Method 2]

[0231] Next, refer to Fig. 20 and Fig.21 Embodiment 2 of the present invention will be described. However, matters different from Embodiment 1 will be described, and the description of matters similar to Embodiment 1 will be omitted. Embodiment 2 is different from Embodiment 1 in that a learned model 136 is generated by a learning device 200 .

[0232] Fig. 20 1 is a diagram showing a substrate processing apparatus 100 and a learning apparatus 200 in this embodiment. In this embodiment, the control device 101 of the substrate processing apparatus 100 is as shown in FIG. Figures 1 to 13 As described above, the learning data set 135 is generated. The learning data set 135 is input to the learning device 200 .

[0233] The learning device 200 generates a learned model 136 using the learning data set 135. The learned model 136 is input to the control device 101 of the substrate processing apparatus 100. Specifically, the learning device 200 includes a processing unit 201 and a storage unit 202. The processing unit 201 is an example of a learning unit.

[0234] The processing unit 201 has a processor. The processing unit 201 has, for example, a CPU or an MPU. Alternatively, the processing unit 201 has a general-purpose arithmetic unit. The processing unit 201 may also have an NCU. The storage unit 202 stores data and computer programs. The storage unit 202 has a main storage device. The main storage device is, for example, a semiconductor memory. The storage unit 202 may also have an auxiliary storage device. The auxiliary storage device is, for example, a semiconductor memory and / or a hard disk drive. The storage unit 202 may also have a removable medium. The processing unit 201 performs machine learning based on the data and computer programs stored in the storage unit 202.

[0235] Specifically, the storage unit 202 stores the inference program 133 and the learning program 134 described in Implementation Example 1. The processing unit 201 learns the learning data set 135 (a plurality of learning data) to generate learned parameters, and generates a learned model 136 based on the learned parameters, similarly to the control unit 102 described in Implementation Example 1.

[0236] In addition, the learning data set 135 may be input to the learning device 200 via a removable memory or a removable medium. Similarly, the learned model 136 may be input to the substrate processing device 100 via a removable memory or a removable medium. Specifically, the learning device 200 may have an interface for acquiring the learning data set 135 from the removable memory and saving the learned model 136 to the removable memory, or may have a drive for acquiring the learning data set 135 from the removable medium and saving the learned model 136 to the removable medium.

[0237] Alternatively, the substrate processing apparatus 100 and the learning apparatus 200 may be connected so as to be able to communicate with each other. In this case, the learning data set 135 is sent from the substrate processing apparatus 100 to the learning apparatus 200, and the learning completed model 136 is sent from the learning apparatus 200 to the substrate processing apparatus 100. In addition, the communication method may be either a wired method or a wireless method. The substrate processing apparatus 100 and the learning apparatus 200 may also be connected so as to be able to communicate with each other via a public communication network such as an Internet line.

[0238] Next, refer to Fig. 20 and Fig.21 A method for generating the learned model 136 executed by the learning device 200 will be described. Fig.21 1 is a flowchart showing a method for generating the learned model 136 according to the present embodiment. Fig.21 The following table shows the processing executed by the processing unit 201 when generating the learned model 136. The method for generating the learned model 136 according to the present embodiment includes the processing of steps S200 to S203.

[0239] First, the processing unit 201 acquires the learning data set 135 from the storage unit 202 (step S200). At this time, the processing unit 201 acquires the learning data set 135 to which the first label is assigned from the plurality of learning data.

[0240] When the processing unit 201 acquires the learning data set 135 , it performs learning on the learning data set 135 (a plurality of learning data) based on the learning program 134 (step S201 ).

[0241] The processing unit 201 generates learned parameters by learning the learning data set 135 (step S202 ). More specifically, the learned parameters are output from the learning program 134 .

[0242] Next, the processing unit 201 generates the learned model 136 based on the acquired learned parameters (step S203). Fig.21 Specifically, the processing unit 201 incorporates the learned parameters into the estimation program 133 , thereby generating the learned model 136 .

[0243] Above, refer to Fig. 20 and Fig.21 Embodiment 2 of the present invention is described. According to this embodiment, as in Embodiment 1, the burden on the operator involved in the development of the scanning speed information can be reduced. Moreover, according to this embodiment, it is not necessary to perform machine learning in the substrate processing apparatus 100. Therefore, as the substrate processing apparatus 100, an existing substrate processing apparatus can be used.

[0244] Furthermore, in the present embodiment, the learning data set 135 is input from the substrate processing apparatus 100 to the learning apparatus 200 . However, the learning data set 135 may be input from the substrate processing apparatus 100 to the learning apparatus 200 .

[0245] Furthermore, the learning apparatus 200 may further perform the additional learning described in Embodiment 1. When the learning apparatus 200 performs the additional learning, an additional learning data set or additional learning data is input from the substrate processing apparatus 100 to the learning apparatus 200 .

[0246] In the present embodiment, the learning data set 135 is input to the learning device 200 from the substrate processing apparatus 100, but learned parameters may be input to the learning device 200 from the substrate processing apparatus 100. When learned parameters are input to the learning device 200 from the substrate processing apparatus 100, the learning device 200 (processing unit 201) performs reference Fig.21 The processing of step S203 is described.

[0247] [Implementation method 3]

[0248] Next, refer to Fig.17 , Fig.18 , Fig. 22 and Fig.23 Embodiment 3 of the present invention will be described. However, matters different from Embodiment 1 and Embodiment 2 will be described, and matters similar to Embodiment 1 and Embodiment 2 will be omitted. Embodiment 3 is different from Embodiment 1 and Embodiment 2 in that the learning device 200 outputs processing speed information as a target variable from the learned model 136.

[0249] Fig. 22 is a diagram showing a substrate processing system 300 in this embodiment. Fig. 22 As shown, the substrate processing system 300 includes a substrate processing apparatus 100 and a learning apparatus 200. The learning apparatus 200 is an example of an information processing apparatus. In this embodiment, the processing unit 201 of the learning apparatus 200 generates the learned model 136 as described in Embodiment 2. The processing unit 201 stores the learned model 136 in the storage unit 202.

[0250] like Fig. 22 As shown, the substrate processing apparatus 100 further includes a communication unit 106. The learning apparatus 200 further includes a communication unit 203. The communication unit 106 and the communication unit 203 are, for example, a LAN board or a wireless LAN board. The communication unit 106 and the communication unit 203 are connected to each other so as to be able to communicate with each other via a public communication network such as an Internet line.

[0251] The communication unit 106 of the substrate processing apparatus 100 is controlled by the control device 101 (control unit 102) and transmits the target processing amount (target etching amount) to the communication unit 203 of the learning device 200. Furthermore, the communication unit 106 of the substrate processing apparatus 100 is controlled by the control device 101 (control unit 102) and receives the processing speed information from the communication unit 203 of the learning device 200. As a result, the control device 101 (control unit 102) acquires the processing speed information. The control unit 102 stores the acquired processing speed information in the storage unit 103.

[0252] The communication unit 203 of the learning device 200 is controlled by the processing unit 201, and receives the target processing amount (target etching amount) from the communication unit 106 of the substrate processing apparatus 100. As a result, the processing unit 201 obtains the target processing amount (target etching amount). In addition, the communication unit 203 of the learning device 200 is controlled by the processing unit 201, and sends the processing speed information to the communication unit 106 of the substrate processing apparatus 100.

[0253] Next, refer to Fig. 22 and Fig.23 The processing executed by the control device 101 (control unit 102) and the processing executed by the processing unit 201 will be described. Fig.23 The flowchart shows the process executed by the control device 101 (control unit 102 ) of the substrate processing apparatus 100 and the process executed by the processing unit 201 of the learning apparatus 200 .

[0254] As described in the first embodiment, the control device 101 (control unit 102) of the substrate processing apparatus 100 acquires the target processing amount (target etching amount) ( Fig.17 In step S21 to step S24), the communication unit 106 is controlled to send the target processing amount (target etching amount) to the communication unit 203 of the learning device 200 (step S210).

[0255] The processing unit 201 of the learning device 200 controls the communication unit 203 to receive the target processing amount (target etching amount) (step S211). As a result, the processing unit 201 acquires the target processing amount (target etching amount).

[0256] The processing unit 201 inputs the target processing amount (target etching amount) as an explanatory variable to the learned model 136, thereby outputting the processing speed information as the target variable from the learned model 136. As a result, the processing unit 201 of the learning device 200 acquires the processing speed information (step S212).

[0257] The processing unit 201 of the learning device 200 controls the communication unit 203 to send the processing speed information to the communication unit 106 of the substrate processing device 100 (step S213). The control device 101 (control unit 102) of the substrate processing device 100 controls the communication unit 106 to receive the processing speed information (step S214). As a result, the control device 101 (control unit 102) obtains the processing speed information. Subsequently, the control device 101 (control unit 102) performs the reference operation. Fig.17 and Fig.18 The processing after step S26 is described.

[0258] Above, refer to Fig.17 , Fig.18 , Fig. 22 and Fig.23 Embodiment 3 of the present invention is described. According to this embodiment, the burden on the operator involved in the development of the scanning speed information can be reduced as in Embodiments 1 and 2. Furthermore, according to this embodiment, it is not necessary to store the learned model 136 in the storage unit 103 of the substrate processing apparatus 100.

[0259] [Implementation Method 4]

[0260] Next, refer to Figure 1 , Figure 8 , Fig.14 , Fig.16 as well as Figures 24 to 31 Embodiment 4 of the present invention will be described. However, matters different from Embodiments 1 to 3 will be described, and the description of matters similar to Embodiments 1 to 3 will be omitted. Embodiment 4 is different from Embodiments 1 to 3 in that the learned model 136 is classified into a plurality of categories.

[0261] Fig.24 Schematic diagram of the processing unit 1 of this embodiment. In detail, Fig.24 is a schematic cross-sectional view of the processing unit 1. Fig.24 As shown, the substrate processing apparatus 100 of this embodiment further includes an etching liquid recovery unit 12. The etching liquid recovery unit 12 has a recovery pipe 121. The chamber 2 further accommodates a portion of the recovery pipe 121. In addition, another portion of the recovery pipe 121 is accommodated in the reference Figure 1 In a corresponding one of the fluid tanks 100B described.

[0262] like Fig.24 As shown, the protective cover 10 includes a cup 11. The cup 11 has a protective cover 11a and a liquid receiving portion 11b. The protective cover 11a has a roughly cylindrical shape and receives the etching liquid (processing liquid) scattered from the rotating substrate W. The liquid receiving portion 11b is connected to the lower end of the protective cover 11a. The liquid receiving portion 11b is annular and forms an annular groove. The etching liquid (processing liquid) received by the protective cover 11a flows to the liquid receiving portion 11b due to its own weight. As a result, the etching liquid (processing liquid) is collected in the liquid receiving portion 11b. One end of the recovery pipe 121 is connected to the liquid receiving portion 11b. The etching liquid (processing liquid) collected in the liquid receiving portion 11b flows into the recovery pipe 121 due to its own weight.

[0263] The etching liquid (processing liquid) flowing into the recovery pipe 121 is as shown in FIG. Fig.25 As described above, the etching liquid (processing liquid) is returned to the container (processing liquid container 45 ) storing the etching liquid (processing liquid) and reused for processing (etching) the substrate W.

[0264] Next, refer to Fig.25 The substrate processing apparatus 100 will be further described. Fig.25 Schematic diagram showing the structure of the substrate processing apparatus 100 according to the present embodiment. Fig.25 Indicates reference Figure 1 The internal structure of the fluid box 100B is described. Fig.25 In order to simplify the drawings, reference is omitted. Figure 7The temperature sensor 421, the concentration sensor 422, the mixing valve 424, the flow meter 425 and the heater 426 are described.

[0265] First, refer to Fig.25 The etching liquid supply unit 4 is further described. Fig.25 As shown in FIG. 1 , the etching liquid supply unit 4 further includes a circulation pipe 43, a circulation pump 44, and a processing liquid container 45. The processing liquid container 45 stores the etching liquid (processing liquid). The processing liquid container 45 is replaceably arranged in the fluid box 100B ( Figure 1 ). By replacing the processing liquid container 45, a new processing liquid container 45 storing the etching liquid is arranged in the fluid box 100B. Fig.24 As described above, the etching liquid (processing liquid) flowing into the recovery pipe 121 returns to the processing liquid container 45 and is reused for processing (etching) the substrate W. If the etching liquid is reused continuously, the etching amount will decrease, so the processing liquid container 45 in the fluid box 100B must be replaced appropriately. The processing liquid container 45 is, for example, a tank.

[0266] The circulation pipe 43 and the circulation pump 44 are housed in the fluid box 100B ( Figure 1 ) in. Both ends of the circulation pipe 43 are connected to the treatment liquid container 45. The circulation pump 44 is installed in the circulation pipe 43. The circulation pump 44 is controlled by the control device 101 (control unit 102).

[0267] When the circulation pump 44 is driven, the etching liquid stored in the processing liquid container 45 flows into the circulation pipe 43 from one end thereof and flows in the circulation pipe 43. The etching liquid guided to the other end of the circulation pipe 43 returns to the processing liquid container 45 from the other end thereof.

[0268] The other end of the first supply pipe 42 is connected to the circulation pipe 43. The etching liquid flowing in the circulation pipe 43 flows into the first supply pipe 42. When the valve 423 is open, the etching liquid flowing from the circulation pipe 43 into the first supply pipe 42 is guided to the first nozzle 41 ( Fig.24 When the valve 423 is in a closed state, the flow of the etching liquid flowing from the circulation pipe 43 to the first supply pipe 42 is blocked by the valve 423 .

[0269] Next, refer to Fig.25 The etching liquid recovery unit 12 is further described. Fig.25 As shown, the etching liquid recovery unit 12 further includes a recovery pump 122. The recovery pump 122 is housed in the fluid box 100B ( Figure 1 ).

[0270] The recovery pump 122 is installed in the recovery pipe 121. The recovery pump 122 is controlled by the control device 101 (control unit 102). The recovery pump 122 drives the etching liquid so that the etching liquid flowing from the processing unit 1 (liquid receiving portion 11b) into the recovery pipe 121 flows in the recovery pipe 121. The other end of the recovery pipe 121 is connected to the processing liquid container 45, and the etching liquid is guided to the other end of the recovery pipe 121 by the driving of the recovery pump 122, and is recovered from the other end of the recovery pipe 121 to the processing liquid container 45. Therefore, the etching liquid used for processing (etching) the substrate W can be reused.

[0271] Next, refer to Figure 8 and Fig.26 A method of generating learning data executed by the substrate processing apparatus 100 according to the present embodiment will be described. Fig.26 : is a flowchart showing a method for generating learning data in this embodiment. Fig.26 The processing executed by the control unit 102 when generating the learning data is shown. Fig. 9 In addition to the described steps S1 to S9 , the processes of steps S101 and S102 are also included.

[0272] In this embodiment, the control unit 102 increments the processing number counter value (step S101) after the learning object substrate W is moved in (step S2). That is, the control unit 102 increases the processing number counter value by "1". The processing number counter value when the learning data is generated represents the number of processing pieces of the learning object substrate W. The control unit 102 stores the incremented processing number counter value in the storage unit 103.

[0273] Specifically, the control unit 102 counts the number of processed sheets of the learning object substrate W from the start of use of the processing liquid (etching liquid) used in the processing (etching) of the learning object substrate W. More specifically, the control unit 102 counts the number of processed sheets of the learning object substrate W from the start of use of the processing liquid container 45 ( Fig.25 ) in the processing liquid container 45 ( Fig.25 ) When the processing liquid container 45 is replaced with a new one, the processing sheet number counter value is reset. Here, the processing sheet number counter value becomes "0" by being reset.

[0274] When the control unit 102 acquires the processing amount (etching amount) (step S7 ), the control unit 102 associates the acquired processing amount with the processed wafer count counter value stored in the storage unit 103 and stores it in the storage unit 103 (step S102 ).

[0275] In addition, the timing of executing each process of step S101 and step S102 is not limited to Fig.26 The timing for executing each process of step S101 and step S102 is not particularly limited as long as it is before the next substrate W is carried in.

[0276] Next, refer to Figure 8 and Fig. 27 The process of generating learning data (step S8) will be described. Fig. 27 : is a flowchart showing the process of generating learning data in this embodiment.

[0277] In this embodiment, the generation process of learning data refers to Fig.11 In addition to steps S81 to S84 described above, the process of step S110 is also included. After generating the learning data (step S81 ), the control unit 102 classifies the learning data into categories based on a predetermined classification rule. The classification rule is stored in the storage unit 103 .

[0278] In this embodiment, the classification rule stipulates that the learning data is classified into each category according to the number of processed wafers (processed wafer number counter value) of the learning object substrate W. More specifically, the classification rule stipulates that the processed wafer number counter value is divided into a plurality of ranges (a plurality of categories), and the control unit 102 refers to the processed wafer number counter value associated with the learning data to classify the learning data by category.

[0279] In addition, the timing of executing the process of step S110 is not limited to Fig. 27 The timing of executing the process of step S110 is not particularly limited as long as it is after the generation of the learning data.

[0280] Next, refer to Fig.28 The learning data set table TB30 for managing the learning data set 135 will be described. Fig.28 1 is a diagram showing an example of the learning data set table TB30 according to the present embodiment.

[0281] like Fig.28 As shown, the learning data set table TB30 includes a category information column TB31 and a learning data set information column TB32. In each column of the category information column TB31, a range of the processing number counter value is predefined. Fig.13 The control unit 102 associates the learning data set table TB10 corresponding to each column of the category information column TB31 (each range of the processing number counter value) with each column of the learning data set information column TB32.

[0282] In addition, each value specified in each column of the category information field TB31 may be a value (default value) specified in advance in the recipe 131 , or may be a value input by the operator operating the input unit 104 .

[0283] Next, refer to Fig.14 and Fig.29 A method of generating the learned model 136 executed by the substrate processing apparatus 100 according to the present embodiment will be described. Fig.29 1 is a flowchart showing a method for generating the learned model 136 in this embodiment. Fig.29 The following is a diagram showing the processing executed by the control unit 102 when generating the learned model 136. The method for generating the learned model 136 according to the present embodiment includes the processing of steps S111 to S118. Fig.29 The processing shown is performed for each learning data set 135 created for each category.

[0284] First, the control unit 102 obtains one of the plurality of learning data sets 135 from the storage unit 103 (step S111). Fig.15 In step S11 , similarly, the learning data to which the first label is assigned is acquired from a plurality of learning data.

[0285] The control unit 102 causes the display unit 105 to display an input screen for information representing the acquired learning data set 135 (step S112). Hereinafter, the input screen is sometimes described as a "selected instruction input screen". Specifically, on the selected instruction input screen, information representing each learning data contained in the acquired learning data set 135 is displayed. In detail, the information representing the learning data represents the learning object speed information #A and the processing amount (etching amount). Hereinafter, the acquired learning data set 135 is sometimes described as a "current learning data set 135".

[0286] The selected instruction input screen is a screen for the operator to select one of the learning data contained in the current learning data set 135. The operator operates the input unit 104 to input an instruction for selecting one of the learning data displayed on the selected instruction input screen. For example, the operator can also select one of the learning data based on information indicating the processing amount (etching amount). In the following, the learning data selected by the operator is sometimes recorded as "reference learning data".

[0287] like Fig.29As shown, after the display of the selection instruction input screen is displayed (step S112), the control unit 102 determines whether an instruction to select reference learning data is input (step S113). The control unit 102 waits until an instruction to select reference learning data is input (No in step S113).

[0288] When it is determined that an instruction to select reference learning data has been input (Yes in step S113 ), control unit 102 selects one of the learning data included in current learning data set 135 as reference learning data according to the instruction from the operator (step S114 ).

[0289] After the reference learning data is selected, the control unit 102 selects learning data whose correlation with the reference learning data satisfies a predetermined condition from among the learning data included in the current learning data set 135 (step S115 ).

[0290] In this embodiment, the control unit 102 selects the learning data based on the processing amount (etching amount) of the reference learning data and the root mean square error (RMSE) of the processing amount (etching amount) of each learning data contained in the current learning data set 135. In detail, the control unit 102 calculates the value of the root mean square error by the following formula, and selects the learning data whose root mean square error value is above the threshold.

[0291] [Number 1]

[0292]

[0293] In the above formula, “n” represents the number of measurement locations of the thickness measuring unit 8 (thickness measuring device 85), and “y i ” represents the distribution of the processing amount of the benchmark learning data (the processing amount of each measurement site), “y i “(hat)” represents the distribution of the processing amount of the learning data (the processing amount of each measurement site), and i represents the number of learning data included in the current learning data set 135 .

[0294] In addition, in this embodiment, the error (distance) between the processing amount (etching amount) of the reference learning data and the processing amount (etching amount) of each learning data included in the current learning data set 135 is obtained by the root mean square error (RMSE), but it can also be obtained based on other indicators. For example, the mean absolute error (MAE) or the mean square error (MSE) can also be used to obtain it.

[0295] like Fig.29As shown in FIG. 1 , after the learning data is selected (step S115), the control unit 102 generates the learning completion parameters (step S117) by learning the selected learning data (step S116). More specifically, the learning completion parameters are output from the learning program 134. The control unit 102 generates the learning completion model 136 based on the acquired learning completion parameters (step S118). Fig.29 The shown processing ends.

[0296] In this embodiment, for each category, Fig.29 As a result, a learned model 136 is generated for each category.

[0297] Next, refer to Fig.16 and Fig.30 , a substrate processing method performed by the substrate processing apparatus 100 of this embodiment is described. Fig.30 1 is a flow chart showing a part of the substrate processing method in this embodiment. Fig.30 FIG. 1 shows a part of the process performed by the control unit 102 when etching the processing target substrate W. Fig.17 and Fig.18 In addition to the processes of step S21 to step S32 described above, the process of step S120 is also included.

[0298] In this embodiment, the control unit 102 increments the processed wafer number counter value (step S120) after the processing target substrate W is carried in (step S22). The processed wafer number counter value during the processing of the processing target substrate W indicates the number of processed wafers of the processing target substrate W. The control unit 102 stores the incremented processed wafer number counter value in the storage unit 103.

[0299] Specifically, the control unit 102 counts the number of processed substrates W from the start of use of the processing liquid (etching liquid) used in the processing (etching) of the processing target substrate W. More specifically, the control unit 102 counts the number of processed substrates W from the start of use of the processing liquid container 45 ( Fig.25 ) in the processing liquid container 45 ( Fig.25 ) is replaced with a new processing liquid container 45, the processed sheet number counter value is reset. Here, the processed sheet number counter value becomes "0" by being reset.

[0300] Next, refer to Fig.16 and Fig.31 The process of acquiring the speed information (step S25) will be described. Fig.31 : is a flowchart showing the process of acquiring speed information.

[0301] like Fig.31 As shown, after obtaining the target processing amount (target etching amount) ( Fig.30 In step S24), the control unit 102 selects one of the categories based on a predetermined selection rule (step S251). The selection rule is stored in the storage unit 103.

[0302] In this embodiment, the selection rule specifies that one of the plurality of categories is selected according to the number of substrates W to be processed. Figure 26 to Figure 28 In the plurality of ranges (a plurality of divisions) of the processing number counter value described above, the processing number counter value ( Fig.30 The range (division) corresponding to step S120) is determined, and one of the multiple categories is selected.

[0303] The control unit 102 inputs the target processing amount (target etching amount) to the learned model 136 corresponding to the selected category (step S252), thereby outputting the processing speed information from the learned model 136. As a result, the control unit 102 acquires the processing speed information (step S253).

[0304] Above, refer to Figure 1 , Figure 8 , Fig.14 , Fig.16 as well as Figures 24 to 31 Embodiment 4 of the present invention is described. According to this embodiment, as in Embodiments 1 to 3, the burden on operators involved in the development of scanning speed information can be reduced. Moreover, according to this embodiment, processing speed information corresponding to the number of processed substrates W from the start of use of the etching liquid can be obtained. Therefore, the first nozzle 41 can be moved at a more appropriate scanning speed. Therefore, the substrate W can be processed (etched) more appropriately.

[0305] Furthermore, in this embodiment, as shown in FIG. Fig.29 As described above, the learned model 136 is generated using the learning data whose correlation with the reference learning data satisfies the predetermined condition. As a result, more appropriate processing speed information can be acquired. Therefore, the substrate W can be processed (etched) more appropriately.

[0306] In the present embodiment, the learned model 136 is generated using the learning data whose correlation with the reference learning data satisfies the predetermined condition. However, the learned model 136 may be generated for each category. Fig.15 By performing the processing shown, a learned model 136 is generated for each category.

[0307] Furthermore, in the present embodiment, machine learning has been described, but additional learning can also be performed in the same manner.

[0308] [Implementation method 5]

[0309] Next, refer to Figure 1 , Figure 8 , Fig.16 , Fig.24 , Fig.25 , Fig. 27 as well as Figure 31 to Figure 35 Embodiment 5 of the present invention will be described. However, matters different from Embodiments 1 to 4 will be described, and matters similar to Embodiments 1 to 4 will be omitted. Embodiment 5 is different from Embodiments 1 to 4 in that the learned model 136 is classified according to the number of batches. In addition, the number of substrates W contained in one batch is, for example, 25.

[0310] First, refer to Figure 8 and Fig.32 A method of generating learning data executed by the substrate processing apparatus 100 according to the present embodiment will be described. Fig.32 : is a flowchart showing a method for generating learning data in this embodiment. Fig.32 The processing executed by the control unit 102 when generating the learning data is shown. Fig. 9 In addition to the steps S1 to S9 described above, the processes of steps S130 and S140 are also included.

[0311] In this embodiment, the control unit 102 performs a counting process (step S130) after the learning object substrate W is moved in (step S2). By performing the counting process, the control unit 102 obtains the learning batch number, which is the batch number counted from the start of use of the processing liquid (etching liquid) used in the processing of the learning object substrate W. More specifically, the control unit 102 obtains the batch number counted from the start of use of the processing liquid container 45 ( Fig.25 ) is the number of learning batches counted from the start of processing (etching) of the learning target substrate W. The control unit 102 stores the number of learning batches acquired by the counting process in the storage unit 103.

[0312] When the control unit 102 acquires the processing amount (etching amount) (step S7 ), the control unit 102 associates the acquired processing amount with the learning-time batch number stored in the storage unit 103 and stores the acquired processing amount in the storage unit 103 (step S140 ).

[0313] In addition, the timing of executing each process of step S130 and step S140 is not limited to Fig.32The timing for executing each process of step S130 and step S140 is not particularly limited as long as it is before the next substrate W is carried in.

[0314] Next, refer to Figure 8 and Fig.33 An example of counting processing will be described. Fig.33 This is a flowchart showing an example of counting processing. Fig.33 The counting process shown includes steps S131 to S134.

[0315] When the counting process is started, the control unit 102 first increments the processing number counter value (step S131). In the following description, the processing number counter value may be referred to as a "first counter value".

[0316] The control unit 102 determines whether the incremented first counter value is equal to or less than a fixed value (step S132). The fixed value indicates the number of substrates W included in one lot.

[0317] When the incremented first counter value exceeds the fixed value, the control unit 102 determines that the incremented first counter value is not less than the fixed value (No in step S132), and increments the batch number counter value (step S133). Hereinafter, the batch number counter value may be described as a "second counter value". The second counter value when generating the learning data indicates the batch number during learning.

[0318] After the second counter value is incremented, the control unit 102 changes the first counter value to "1" (step S134), and the process ends. Fig.33 Alternatively, when the control unit 102 determines that the first counter value after increment is less than the fixed value (yes in step S132), the counting process is terminated. Fig.33 The counting process shown.

[0319] In addition, the control unit 102 controls the processing liquid container 45 ( Fig.25 ) is replaced with a new processing liquid container 45, the second counter value is changed to "1".

[0320] Next, refer to Figure 8 and Fig. 27 The process of generating learning data (step S8) is described. In the present embodiment, the classification rule stipulates that the learning data is classified into each category according to the number of batches during learning (the second counter value). The control unit 102 refers to the second counter value associated with the learning data and classifies the learning data by category (step S110).

[0321] Next, refer to Fig.34 The learning data set table TB40 for managing the learning data set 135 will be described. Fig.34 1 is a diagram showing an example of the learning data set table TB40 according to the present embodiment.

[0322] like Fig.34 As shown, the learning data set table TB40 includes a category information column TB41 and a learning data set information column TB42. In each column of the category information column TB41, the number of batches during learning is predefined. In this embodiment, the control unit 102 creates a reference for each category. Fig.13 The control unit 102 associates the learning data set table TB10 corresponding to each column (number of batches during learning) of the category information column TB41 with each column of the learning data set information column TB42.

[0323] Next, refer to Fig.16 and Fig.35 A substrate processing method performed by the substrate processing apparatus 100 according to the present embodiment will be described. Fig.35 1 is a flow chart showing a part of the substrate processing method in this embodiment. Fig.35 FIG. 1 shows a part of the process performed by the control unit 102 when etching the processing target substrate W. Fig.17 and Fig.18 In addition to the processes of step S21 to step S32 described above, the process of step S150 is also included.

[0324] In this embodiment, the control unit 102 performs a counting process (step S150) after the processing target substrate W is carried in (step S22). By performing the counting process, the control unit 102 obtains the number of batches during processing, which is the number of batches counted from the start of use of the processing liquid (etching liquid) used in the processing of the processing target substrate W. More specifically, the control unit 102 obtains the number of batches of the processing liquid container 45 ( Fig.25 ) in the etching solution, and the number of processing batches from the start of processing (etching) of the processing target substrate W. More specifically, the control unit 102 executes and refers to Fig.33 The number of lots during processing (second counter value) can be acquired by performing the same processing as described above. The control unit 102 stores the number of lots during processing (second counter value) acquired by the counting processing in the storage unit 103 .

[0325] Next, refer to Fig.16 and Fig.31 The process of obtaining the processing speed information (step S25) is described below. In this embodiment, the selection rule stipulates that one of the multiple categories is selected according to the number of batches during processing. The control unit 102 selects from the reference Figure 32 to Figure 34The number of batches during learning described above is determined as the number of batches during processing stored in the storage unit 103 ( Fig.35 Step S150), thereby selecting one of multiple categories.

[0326] The control unit 102 inputs the target processing amount (target etching amount) to the learned model 136 corresponding to the selected category (step S252), thereby outputting the processing speed information from the learned model 136. As a result, the control unit 102 acquires the processing speed information (step S253).

[0327] In addition, the learned model 136 can be performed for each category. Fig.29 The processing shown in the figure can also be performed for each category. Fig.15 Generated by the processing shown.

[0328] Above, refer to Figure 1 , Figure 8 , Fig.16 , Fig.24 , Fig.25 , Fig. 27 as well as Figure 31 to Figure 35 Embodiment 5 of the present invention is described. According to this embodiment, as in Embodiments 1 to 4, the burden on the operator involved in the development of scanning speed information can be reduced. Moreover, according to this embodiment, processing speed information corresponding to the number of processed substrates W from the start of use of the etching solution can be obtained. Therefore, the substrate W can be processed (etched) more appropriately.

[0329] In addition, in the present embodiment, machine learning is described, but additional learning can also be performed in the same manner.

[0330] The above describes the embodiments of the present invention with reference to the accompanying drawings. However, the present invention is not limited to the embodiments and can be implemented in various ways without departing from the scope of the present invention. Moreover, the multiple constituent elements disclosed in the embodiments can be appropriately changed. For example, a constituent element of all constituent elements shown in a certain embodiment can be added to a constituent element of another embodiment, or some constituent elements of all constituent elements shown in a certain embodiment can be deleted from the embodiment.

[0331] In order to facilitate the understanding of the invention, the drawings schematically illustrate each component as the main body, and in order to facilitate the preparation of the drawings, the thickness, length, number, spacing, etc. of each component illustrated in the drawings are sometimes different from the actual ones. Moreover, the structure of each component shown in the above-mentioned embodiment is an example and is not particularly limited. Of course, various changes can be made within the scope that does not substantially deviate from the effect of the present invention.

[0332] For example, refer to Figures 1 to 35 In the described embodiment, the thickness measuring unit 8 measures the thickness of the object TG, but the thickness measuring unit 8 may also measure the thickness of the substrate W. More specifically, the thickness measuring unit 8 may also measure the thickness of the substrate W to be studied and the thickness of the substrate W to be processed. Here, in the case where a substance is formed on the surface of the substrate body, the thickness of the substrate W represents the thickness of the substrate body and the substance combined. In addition, in the case where the thickness measuring unit 8 measures the thickness of the substrate W, the etching amount represents the difference between the thickness of the substrate W before the etching process and the thickness of the substrate W after the etching process. Moreover, the target thickness (target thickness distribution) represents the target thickness (target thickness distribution) of the substrate W to be processed.

[0333] Moreover, refer to Figures 1 to 35 In the described embodiment, the substrate W is a semiconductor wafer, but the substrate W is not limited to a semiconductor wafer. For example, the substrate W may be a substrate for a liquid crystal display device, a substrate for a field emission display (FED), a substrate for an optical disk, a substrate for a magnetic disk, a substrate for an optical magneto-optical disk, a substrate for a photomask, a ceramic substrate, or a substrate for a solar cell.

[0334] Moreover, refer to Figures 1 to 35 In the described embodiment, the processing liquid is an etching liquid, but the processing liquid is not limited to an etching liquid. The processing liquid can be any liquid as long as it is used to process the substrate W. For example, the processing liquid can also be a removal liquid for removing the object TG. By using the removal liquid, it is possible to perform a process of removing a specific film or a process of removing a specific film mixed with foreign matter. In the case of performing a resist removal process on the substrate W, the removal liquid is, for example, a sulfuric acid / hydrogen peroxide mixture (SPM). The so-called resist removal process refers to a process of removing the resist from the surface of the semiconductor substrate.

[0335] Moreover, refer to Figures 1 to 35 In the described embodiment, the process performed by the substrate processing apparatus 100 is an etching process, but the process performed by the substrate processing apparatus 100 is not limited to the etching process. For example, the process may also be a film forming process. When the substrate processing apparatus 100 performs the film forming process, the processing liquid is, for example, SPM or ozone water. At this time, an oxide film is formed on the substrate W. Moreover, the processing amount indicates the film forming amount.

[0336] Moreover, refer to Figures 1 to 35 In the described embodiment, as a structure for holding the substrate W, a clamping chuck for clamping the substrate W is described. However, as a structure for holding the substrate W, a vacuum chuck may be adopted.

[0337] Moreover, refer to Figures 1 to 35 In the embodiment described above, the error is notified to the operator by displaying an error screen on the display unit 105, but the error may be notified to the operator by sound. In this case, the substrate processing apparatus 100 may include a speaker.

[0338] Moreover, refer to Figures 1 to 35 In the described embodiment, the scanning speed information indicates the set value of the scanning speed for each position included in the moving interval of the first nozzle 41, but the scanning speed information may indicate only one set value of the scanning speed. In this case, the first nozzle 41 moves at a fixed speed from the starting position to the ending position of the moving interval of the first nozzle 41.

[0339] Moreover, refer to Figures 1 to 35 In the described embodiment, the control unit 102 stores the learning data in the storage unit 103 in association with the tag (first tag or second tag), but when the control unit 102 determines that the learning data (processing amount) cannot be used for machine learning or additional learning, the learning data may be deleted from the storage unit 103. For example, the control unit 102 may delete the speed information column TB11 and the processing amount column TB12 corresponding to the learning data assigned the second tag from the learning data set table TB10 when the learning data set 135 is generated or when machine learning is executed. Similarly, the control unit 102 may delete the speed information column TB21 and the processing amount column TB22 corresponding to the additional learning data assigned the second tag from the additional learning data set table TB20 when the additional learning data set is generated or when additional learning is executed.

[0340] Moreover, refer to Figures 1 to 35 In the described embodiment, the control unit 102 determines whether the learning data can be used for machine learning or additional learning after generating the learning data, but the control unit 102 may determine whether the processing volume can be used for machine learning or additional learning before generating the learning data. In this case, when the control unit 102 determines that the processing volume cannot be used for machine learning or additional learning, it does not generate the learning data using the processing volume, but transfers to the next process (next step).

[0341] Moreover, refer to Figures 1 to 35 In the described embodiment, the control unit 102 determines whether the learning data (processing amount) can be used, but the process of determining whether the learning data (processing amount) can be used may be omitted.

[0342] Moreover, refer to Figures 1 to 35In the described embodiment, the training data set 135 and the additional training data set are stored in the storage unit 103 and the storage unit 202 in a table format. However, the training data set 135 and the additional training data set may be stored in a format other than a table format.

[0343] Moreover, refer to Figures 1 to 35 In the described embodiment, the learning object speed information and the speed information during processing specify the moving speed of the first nozzle 41, but the learning object speed information and the speed information during processing may also specify the relative moving speed between the first nozzle 41 and the substrate W. In addition, when the substrate W rotates, the learning object speed information and the speed information during processing specify the relative moving speed between the surface of the rotating substrate W and the first nozzle 41. For example, the relative moving speed represents the sum of the speed component (vector) of the first nozzle 41 and the speed component (vector) of the portion of the rotating substrate W facing the first nozzle 41. Here, the speed component of the substrate W represents the speed in the circumferential direction.

[0344] Moreover, refer to Figures 1 to 35 In the described embodiment, the first nozzle 41 rotates, but the first nozzle 41 may also move linearly.

[0345] Moreover, refer to Figures 1 to 35 In the described embodiment, the first nozzle 41 is a scanning nozzle, but the first nozzle 41 may also be a fixed nozzle. In this case, the processing unit 1 includes a substrate moving mechanism for moving the substrate W instead of the nozzle moving mechanism 6. The learning object speed information and the speed information during processing replace the moving speed of the first nozzle 41 to specify the moving speed of the substrate W. The learning completed model 136 outputs substrate speed information indicating the moving speed of the substrate W. The control unit 102 controls the substrate moving mechanism based on the substrate speed information. The substrate speed information may also indicate the moving speed of the substrate W set for each position (each substrate position) of the moving interval in which the substrate W is moved, which is divided into a plurality of intervals, in the same manner as the scanning speed information. In addition, the substrate W may either rotate or move in a straight line.

[0346] Moreover, refer to Figures 1 to 35 In the described embodiment, only the first nozzle 41 moves during the processing of the substrate W, but the first nozzle 41 and the substrate W may also move. In this case, the processing unit 1 further includes a substrate moving mechanism for moving the substrate W instead of the nozzle moving mechanism 6. The learning object speed information and the processing speed information specify the moving speed of the first nozzle 41 and the moving speed of the substrate W. The learning completed model 136 outputs substrate speed information that specifies the moving speed of the substrate W in addition to the scanning speed information. The control unit 102 performs control of the nozzle moving mechanism 6 based on the scanning speed information and control of the substrate moving mechanism based on the substrate speed information.

[0347] Moreover, refer to Figures 1 to 35 In the described embodiment, the substrate processing apparatus 100 includes the thickness measuring unit 8 , but the thickness measuring unit 8 may be omitted. When the substrate processing apparatus 100 does not include the thickness measuring unit 8 , the thickness is measured by a thickness measuring device outside the substrate processing apparatus 100 .

[0348] Moreover, refer to Figures 1 to 35 In the described embodiment, the target processing volume (target etching amount) is selected as the explanatory variable, but other factors affecting the processing volume can also be added to the explanatory variables. Therefore, in addition to representing the relationship between the processing volume (etching amount) and the scanning speed information, the learning data set 135 can also represent the relationship between the processing volume and other factors affecting the processing volume. At this time, as explanatory variables, the target processing volume and other factors affecting the processing volume are input into the learning completion model 136, and the scanning speed information is output from the learning completion model 136. Other factors affecting the processing volume may include, for example, at least one of the rotation speed of the substrate W, the temperature of the processing liquid, the concentration of the processing liquid, and the ejection flow rate of the processing liquid.

[0349] Moreover, refer to Figures 1 to 35 In the described embodiment, the scanning speed information (processing speed information) is selected as the target variable, but other factors that affect the processing volume may be added to the target variable. Therefore, the learned model 136 may output other factors that affect the processing volume as the target variable in addition to the scanning speed information. For example, the learned model 136 may output at least one of the rotation speed of the substrate W, the temperature of the processing liquid, the concentration of the processing liquid, and the ejection flow rate of the processing liquid in addition to the scanning speed information.

[0350] Moreover, refer to Figures 1 to 35 In the described embodiment, the learning program 134 is a program for executing the following algorithm, that is, learning the learning data set 135 (a plurality of learning data) to generate a learned parameter, and incorporating the learned parameter into the inference program 133, thereby generating a learned model 136, but the learning program 134 may also be a program for executing the following algorithm, that is, learning the learning data set 135 (a plurality of learning data) to generate a learned model including the learned parameter. That is, the process of generating the learned parameter and incorporating the learned parameter into the inference program 133 may also be omitted. Alternatively, the learning program 134 may also be a program for executing the following algorithm, that is, learning the learning data set 135 (a plurality of learning data) to generate a learned model that does not include the learned parameter.

[0351] [Industrial Applicability]

[0352] The present invention is useful in the field of processing substrates.

Claims

1. A method for generating learning data for additional learning, which is a method for generating learning data for additional learning using a substrate processing device, wherein the substrate processing device supplies a processing liquid to a substrate based on an output of a learning model generated by learning the learning data to perform processing on the substrate, the method comprising: The step of calculating the difference between the thickness distribution of the substrate before processing and the target thickness distribution of the substrate to obtain a target amount of processing volume, i.e., a target processing volume, wherein the processing volume represents the amount of processing of the substrate by the processing; A step of performing the processing on the substrate to obtain the processing amount; a step of determining whether the processing volume is consistent with the target processing volume within an allowable range; as well as The step of adding a mark to the processing amount, the mark indicating that the processing amount is used as learning data for the additional learning, if the processing amount and the target processing amount are determined to be inconsistent within the allowable range.

2. A method for generating a learned model, which is a method for generating a learned model used in a substrate processing device, wherein the substrate processing device supplies a processing liquid to a processing target substrate based on an output of the learned model to perform processing on the processing target substrate, the learned model generation method comprising: supplying the processing liquid to a learning object substrate to perform the processing on the learning object substrate; a step of acquiring a processing amount, wherein the processing amount indicates an amount of processing performed on the learning object substrate by the processing; a step of generating learning data, wherein the learning data includes the processing amount of the learning target substrate; as well as a learning step of classifying the learning data into categories based on a predetermined classification rule, and learning the learning data for each category to generate the learned model, wherein The classification rule defines that the learning data is classified into the categories according to the number of processed sheets of the learning target substrates.

3. A method for generating a learned model, which is a method for generating a learned model used in a substrate processing device, wherein the substrate processing device supplies a processing liquid to a processing target substrate based on an output of the learned model to perform processing on the processing target substrate, the learned model generation method comprising: supplying the processing liquid to a learning object substrate to perform the processing on the learning object substrate; a step of acquiring a processing amount, wherein the processing amount indicates an amount of processing performed on the learning object substrate by the processing; a step of generating learning data, wherein the learning data includes the processing amount of the learning target substrate; as well as a learning step of classifying the learning data into categories based on a predetermined classification rule, and learning the learning data for each category to generate the learned model, wherein In the step of executing the process on the learning target substrate, a learning batch number is acquired, the learning batch number being the batch number counted from the start of use of the process liquid used in the process, The classification rule defines classification of the learning data into the categories according to the number of batches during the learning.

Images