Substrate processing apparatus, substrate processing method, learning data generation method, learning method, learning device, learning completed model generation method, and learning completed model

Abstract

A substrate processing apparatus (100) of the present application includes a substrate holding section (120), a liquid supply section (130), a substrate information acquisition section (22a), a liquid processing condition information acquisition section (22b), and a control section (22). The substrate information acquisition section (22a) acquires substrate information including hardening layer thickness information or ion implantation condition information, the hardening layer thickness information indicating a thickness of a hardening layer in a resist layer of a substrate to be processed, and the ion implantation condition information indicating a condition of ion implantation for forming the hardening layer in the resist layer. The liquid processing condition information acquisition section (22b) acquires liquid processing condition information indicating a liquid processing condition for the substrate to be processed from a learning completion model based on the substrate information. The control section (22) controls the substrate holding section (120) and the liquid supply section (130) in accordance with the liquid processing condition information to process the substrate to be processed with a liquid.

Description

Technical Field

[0001] This invention relates to a substrate processing apparatus, a substrate processing method, a method for generating learning data, a learning method, a learning apparatus, a method for generating a learning completion model, and a learning completion model. Background Technology

[0002] The substrate processing apparatus is suitable for use in the manufacture of semiconductor devices, etc. During substrate manufacturing, there are cases where the substrate's properties are modified by ion implantation after forming a resist layer with a predetermined pattern. In this case, the resist layer is stripped using a chemical solution after ion implantation. It is known that if ions are implanted into the resist layer during substrate manufacturing, a hardened layer or a modified layer will form on the resist layer (see Patent Document 1).

[0003] Patent Document 1 describes a semiconductor substrate cleaning agent containing an added hydrocarbon ester of sulfuric acid and hydrogen peroxide solution for stripping and removing the resist layer. According to Patent Document 1, this semiconductor substrate cleaning agent has cleaning power comparable to SPM cleaning agents when stripping the resist layer, while reducing damage to the semiconductor substrate.

[0004] Existing technical documents

[0005] Patent documents

[0006] Patent Document 1: Japanese Patent Application Publication No. 2012-67254 Summary of the Invention

[0007] The problem that the invention aims to solve

[0008] However, in the method of Patent Document 1, there are instances where the hardened layer in the resist layer of the substrate cannot be properly removed. For example, if the amount of reagent used is too small relative to the ion-implanted resist layer, the resist layer cannot be sufficiently removed. Conversely, if the amount of reagent used is too large relative to the ion-implanted resist layer, it leads to increased costs.

[0009] The present invention was made in view of the above-mentioned problems, and its object is to provide a substrate processing apparatus, a substrate processing method, a method for generating learning data, a learning method, a learning apparatus, a method for generating a learning completion model, and a learning completion model that can appropriately remove the hardened layer in the resist layer of a substrate.

[0010] (Technical means to solve the problem)

[0011] According to one aspect of the present invention, a substrate processing apparatus includes: a substrate holding unit that holds a substrate to be processed having a resist layer having a hardened layer formed thereon so that it is rotatable; a solution supply unit that supplies a solution to the substrate to be processed; a substrate information acquisition unit that acquires substrate information including hardened layer thickness information or ion implantation condition information regarding the substrate to be processed; wherein the hardened layer thickness information indicates the thickness of the hardened layer; and the ion implantation condition information indicates the ion implantation conditions under which the hardened layer is formed in the resist layer; a solution processing condition information acquisition unit that acquires solution processing condition information indicating solution processing conditions for the substrate to be processed based on the substrate information and a self-learning completion model; and a control unit that controls the substrate holding unit and the solution supply unit based on the solution processing condition information acquired by the solution processing condition information acquisition unit to process the substrate to be processed using the solution. The aforementioned learning completion model is constructed by performing machine learning on the learning data. In this learning data, substrate information, including hardened layer thickness information or ion implantation condition information, chemical treatment condition information, and processing result information are established for the learning object substrate having a hardened resist layer. Here, the hardened layer thickness information represents the thickness of the hardened layer; the ion implantation condition information represents the ion implantation conditions for forming the hardened layer in the resist layer; the chemical treatment condition information represents the conditions for processing the learning object substrate with the chemical solution; and the processing result information represents the result of processing the learning object substrate with the chemical solution.

[0012] In one embodiment, the substrate processing apparatus further includes a storage unit for storing the learned model.

[0013] In one embodiment, for each of the processing target substrate and the learning target substrate, the hardened layer thickness information includes hardened layer height information representing the height of the hardened layer or hardened layer width information representing the width of the hardened layer.

[0014] In one embodiment, for each of the above-mentioned processing target substrate and the above-mentioned learning target substrate, the drug solution processing condition information includes information indicating any one of the following: concentration of the drug solution, temperature of the drug solution, supply amount of the drug solution, spray pattern of the drug solution, and rotation speed of the substrate when the drug solution is supplied.

[0015] In one embodiment, the information representing the concentration of the above-mentioned drug solution represents the concentration profile of the drug solution as the concentration changes over time.

[0016] In one embodiment, the information representing the temperature of the above-mentioned medicinal solution represents the temperature distribution of the above-mentioned medicinal solution as the temperature changes over time.

[0017] According to another aspect of the present invention, a substrate processing method includes: holding a substrate to be processed having a resist layer having a hardened layer formed in a rotatable manner; obtaining substrate information regarding the substrate to be processed, including hardened layer thickness information or ion implantation condition information; wherein the hardened layer thickness information represents the thickness of the hardened layer; and the ion implantation condition information represents the ion implantation conditions for forming the hardened layer in the resist layer; obtaining, based on the substrate information, solution processing condition information representing solution processing conditions of the substrate to be processed using a self-learning completion model; and processing the substrate to be processed using a solution based on the solution processing conditions of the solution processing condition information. In the step of obtaining the above-mentioned chemical treatment condition information, the learning completion model is constructed by performing machine learning on the learning data. In the learning data, substrate information including hardened layer thickness information or ion implantation condition information, chemical treatment condition information, and processing result information are established for the learning object substrate having a hardened resist layer. The hardened layer thickness information represents the thickness of the hardened layer; the ion implantation condition information represents the ion implantation conditions for forming the hardened layer in the resist layer; the chemical treatment condition information represents the chemical treatment conditions performed on the learning object substrate; and the processing result information represents the result of the chemical treatment performed on the learning object substrate.

[0018] According to another aspect of the present invention, a method for generating learning data includes: a step of obtaining substrate information, including hardened layer thickness information or ion implantation condition information, from time-series data output by a substrate processing apparatus that processes a learning substrate having a hardened resist layer. The hardened layer thickness information indicates the thickness of the hardened layer; the ion implantation condition information indicates the conditions for ion implantation that form the hardened layer in the resist layer. A step of obtaining chemical treatment condition information from the time-series data, the chemical treatment condition information indicating the conditions under which the learning substrate is processed using a chemical solution in the substrate processing apparatus. A step of obtaining processing result information from the time-series data, the processing result information indicating the result of processing the learning substrate using a chemical solution in the substrate processing apparatus. And a step of establishing an association between the substrate information, the chemical treatment condition information, and the processing result information regarding the learning substrate and storing them as learning data in a storage unit.

[0019] According to another aspect of the present invention, the learning method includes: the step of obtaining learning data generated according to the above-described method for generating learning data; and the step of inputting the learning data into a learning program and performing machine learning on the learning data.

[0020] According to another aspect of the present invention, the learning device includes: a storage unit that stores learning data generated according to the above-described method for generating learning data; and a learning unit that inputs the learning data into a learning program and performs machine learning on the learning data.

[0021] According to another aspect of the present invention, the method for generating a learning completion model includes: a step of obtaining learning data generated according to the above-described method for generating learning data; and a step of generating a learning completion model, which is constructed by performing machine learning on the above-described learning data.

[0022] According to another aspect of the present invention, the learning completion model is constructed by machine learning based on the learning data generated by the above-described method for generating learning data.

[0023] According to another aspect of the present invention, a substrate processing apparatus includes: a substrate holding unit that holds a substrate having a hardened resist layer formed thereon so that it can be rotated; a solution supply unit that supplies a solution to the substrate; a storage unit that stores a conversion table that associates substrate information with solution processing condition information, wherein the substrate information includes hardened layer thickness information or ion implantation condition information indicating ion implantation conditions; and the solution processing condition information indicating the conditions for solution processing; a substrate information acquisition unit that acquires substrate information including hardened layer thickness information or ion implantation condition information regarding the substrate; wherein the hardened layer thickness information indicates the thickness of the hardened layer; and the ion implantation condition information indicates the ion implantation conditions for forming the hardened layer; a solution processing condition information acquisition unit that, based on the substrate information, uses the conversion table to acquire solution processing condition information indicating the solution processing conditions for the substrate; and a control unit that controls the substrate holding unit and the solution supply unit based on the solution processing condition information acquired by the solution processing condition information acquisition unit to process the substrate using the solution.

[0024] Invention Effects

[0025] According to the present invention, the hardened layer in the resist layer of the substrate to be processed can be appropriately removed. Attached Figure Description

[0026] Figure 1 This is a schematic diagram of a substrate processing learning system equipped with the substrate processing apparatus of this embodiment.

[0027] Figure 2 This is a schematic diagram of a substrate processing system equipped with the substrate processing apparatus of this embodiment.

[0028] Figure 3 This is a schematic diagram of the substrate processing apparatus of this embodiment.

[0029] Figure 4 This is a block diagram of a substrate processing system equipped with the substrate processing apparatus of this embodiment.

[0030] Figure 5 (a) is a flowchart of the substrate processing method of this embodiment. Figure 5 (b) is a flowchart of the chemical solution treatment in the substrate processing method of this embodiment.

[0031] Figure 6 (a) to (c) are schematic diagrams illustrating the formation process of the hardened layer formed on the substrate processed in the substrate processing apparatus of this embodiment.

[0032] Figure 7 (a) to (d) are schematic diagrams illustrating the formation process of the hardened layer formed on the substrate processed in the substrate processing apparatus of this embodiment.

[0033] Figure 8 This is a block diagram of a substrate processing system and a learning data generation apparatus equipped with the substrate processing apparatus of this embodiment.

[0034] Figure 9 This is a flowchart illustrating the learning data generation method of this embodiment.

[0035] Figure 10 This is a block diagram of the learning data generation apparatus and the learning apparatus of this embodiment.

[0036] Figure 11 This is a flowchart illustrating the learning method and the method for generating the learned model in this embodiment.

[0037] Figure 12 This is a diagram representing the learning data input into the learning device of this embodiment.

[0038] Figure 13 This is a diagram representing the learning data input into the learning device of this embodiment.

[0039] Figure 14 This is a diagram representing the learning data input into the learning device of this embodiment.

[0040] Figure 15 (a) is a schematic diagram of a substrate with a resist layer that is the object of processing by the substrate processing apparatus of this embodiment, wherein a hardening layer is formed in the resist layer. Figure 15 (b) is a diagram showing the drug treatment conditions obtained based on substrate information.

[0041] Figure 16 This is a diagram representing the learning data input into the learning device of this embodiment.

[0042] Figure 17 This is a schematic diagram of the substrate processing apparatus of this embodiment.

[0043] Figure 18 This is a diagram representing the learning data input into the learning device of this embodiment.

[0044] Figure 19 This is a block diagram of the substrate processing apparatus of this embodiment.

[0045] Figure 20 This is a diagram showing the conversion table in the substrate processing apparatus of this embodiment. Detailed Implementation

[0046] The following description, with reference to the accompanying drawings, outlines the substrate processing apparatus, substrate processing method, method for generating learning data, learning method, learning apparatus, method for generating a completed learning model, and embodiments of the completed learning model of the present invention. Furthermore, in the drawings, the same or equivalent parts are labeled with the same reference numerals and will not be described again. Additionally, in this specification, for ease of understanding of the present invention, mutually orthogonal X, Y, and Z directions are sometimes described. Typically, the X and Y directions are parallel to the horizontal direction, and the Z direction is parallel to the vertical direction.

[0047] First, refer to Figure 1 The substrate processing learning system 200 equipped with the substrate processing apparatus 100 of this embodiment will be described. First, referring to... Figure 1 The substrate processing learning system 200 will be described.

[0048] Figure 1 This is a schematic diagram of the substrate processing learning system 200. (As shown...) Figure 1 As shown, the substrate processing learning system 200 includes a substrate processing apparatus 100, a substrate processing apparatus 100L, a learning data generation apparatus 300, and a learning apparatus 400.

[0049] The substrate processing apparatus 100 processes the substrate to be processed. Here, the substrate to be processed has a resist layer with a hardened layer formed on it, and the substrate processing apparatus 100 processes the resist layer of the substrate to be processed using a chemical solution. Furthermore, the substrate processing apparatus 100 can also perform processing on the substrate to be processed other than chemical treatment. The substrate processing apparatus 100 is a monolithic processing apparatus that processes substrates one by one. Typically, the substrate to be processed is roughly in the shape of a circular plate.

[0050] The substrate processing apparatus 100L processes a learning target substrate. Here, the learning target substrate has a resist layer with a hardened layer, and the substrate processing apparatus 100L processes the resist layer of the learning target substrate using a chemical solution. Furthermore, the substrate processing apparatus 100L can also perform processing on the learning target substrate other than chemical treatment. The structure of the learning target substrate is the same as that of the processing target substrate. The substrate processing apparatus 100L is a monolithic processing apparatus that processes processing target substrates one by one. Typically, the processing target substrate is roughly in the shape of a circular plate. The structure of the substrate processing apparatus 100L is the same as that of the substrate processing apparatus 100. The substrate processing apparatus 100L may also be the same object as the substrate processing apparatus 100. For example, the same substrate processing apparatus may be used to process a learning target substrate before processing the processing target substrate. Alternatively, the substrate processing apparatus 100L may be another product having the same structure as the substrate processing apparatus 100.

[0051] In the following description of this specification, the learning target substrate may be referred to as "learning target substrate WL" and the processing target substrate may be referred to as "processing target substrate Wp". However, when it is not necessary to distinguish between the learning target substrate WL and the processing target substrate Wp in the description, both the learning target substrate WL and the processing target substrate Wp may be referred to as "substrate W".

[0052] The substrate W is, for example, a semiconductor wafer, a substrate for a liquid crystal display device, a substrate for a plasma display, a substrate for a field emission display (FED), a substrate for an optical disc, a substrate for a magnetic disk, a substrate for a magneto-optical disc, a substrate for a photomask, a ceramic substrate, or a substrate for a solar cell.

[0053] The substrate processing apparatus 100L outputs time-series data TDL. Time-series data TDL represents the time-varying changes of physical quantities within the substrate processing apparatus 100L. Specifically, it represents the time-series changes of a physical quantity (value) that varies over a specified period. For example, time-series data TDL may represent the time-varying changes of physical quantities relating to the processing performed by the substrate processing apparatus 100L on the learning target substrate. Alternatively, time-series data TDL may represent the time-varying changes of physical quantities relating to the characteristics of the learning target substrate processed by the substrate processing apparatus 100L. Alternatively, time-series data TDL may also include data representing the manufacturing processes prior to processing the learning target substrate using the substrate processing apparatus 100L.

[0054] Furthermore, the values ​​represented in the time series data TDL can also be values ​​directly measured in the measuring device. Alternatively, the values ​​represented in the time series data TDL can be values ​​that have been processed from values ​​directly measured in the measuring device. Alternatively, the values ​​represented in the time series data TDL can be values ​​that have been processed from values ​​measured in multiple measuring devices.

[0055] The learning data generation apparatus 300 generates learning data LD based on time series data TDL or at least a portion of time series data TDL. The learning data generation apparatus 300 outputs the learning data LD. The learning data LD includes substrate information of the learning target substrate WL, chemical treatment condition information indicating the treatment conditions performed on the learning target substrate WL, and processing result information indicating the result of the chemical treatment performed on the learning target substrate WL. Furthermore, the substrate information of the learning target substrate WL includes substrate information of the learning target substrate WL measured before the chemical treatment.

[0056] The learning device 400 performs machine learning on the learning data LD to generate a learned model LM. The learning device 400 outputs the learned model LM.

[0057] The substrate processing apparatus 100 outputs time-series data TD. Time-series data TD represents the time-varying changes of a physical quantity within the substrate processing apparatus 100. Specifically, time-series data TD represents the time-varying changes of a physical quantity (value) that changes sequentially over a specified period. For example, time-series data TD may represent the time-varying changes of a physical quantity relating to the processing performed by the substrate processing apparatus 100 on the substrate to be processed. Alternatively, time-series data TD may represent the time-varying changes of a physical quantity relating to the characteristics of the substrate to be processed by the substrate processing apparatus 100.

[0058] Furthermore, the values ​​represented in the time series data TD may be values ​​directly measured in the measuring equipment. Alternatively, the values ​​represented in the time series data TD may be values ​​that have been processed from values ​​directly measured in the measuring equipment. Alternatively, the values ​​represented in the time series data TD may be values ​​that have been processed from values ​​measured in multiple measuring equipment. Alternatively, the time series data TD may also include data representing the manufacturing process prior to processing the substrate to be processed using the substrate processing apparatus 100.

[0059] The object used in the substrate processing apparatus 100 corresponds to the object used in the substrate processing apparatus 100L. Therefore, the structure of the object used in the substrate processing apparatus 100 is the same as that of the object used in the substrate processing apparatus 100L. Furthermore, in the time series data TD, the physical quantities of the object used in the substrate processing apparatus 100 correspond to the physical quantities of the object used in the substrate processing apparatus 100L. Therefore, the physical quantities of the object used in the substrate processing apparatus 100L are the same as those of the object used in the substrate processing apparatus 100.

[0060] Substrate information Cp for the substrate Wp to be processed is generated based on time-series data TD. The substrate information Cp for the substrate Wp to be processed corresponds to the substrate information for the learning substrate WL. The substrate information Cp for the substrate Wp to be processed contains substrate information about the substrate Wp to be processed. The substrate information for the substrate Wp to be processed can also be information obtained by measuring the substrate Wp to be processed before it is treated with the chemical solution. Alternatively, the substrate information for the substrate Wp to be processed can also be information about the treatment performed on the substrate Wp to be processed before chemical treatment.

[0061] The self-learning model LM outputs chemical treatment condition information Rp based on the substrate information Cp of the substrate Wp to be processed. The chemical treatment condition information Rp represents the chemical treatment conditions suitable for the substrate Wp to be processed in the substrate processing apparatus 100.

[0062] The above is for reference only. Figure 1 As explained, according to this embodiment, the learning device 400 performs machine learning. Therefore, a highly accurate learning completion model LM can be generated from the very complex and large-scale time-series data TDL. Furthermore, the substrate information Cp from the time-series data TD is input to the learning completion model LM, and the learning completion model LM outputs liquid treatment condition information Rp representing the liquid treatment conditions. Therefore, liquid treatment can be performed based on the substrate Wp to be treated.

[0063] Secondly, refer to Figure 2 The substrate processing system 10 equipped with the substrate processing apparatus 100 of this embodiment will be described. Figure 2 This is a schematic top view of the substrate processing system 10.

[0064] The substrate processing system 10 processes the substrate W. The substrate processing system 10 includes multiple substrate processing apparatuses 100. The substrate processing apparatuses 100 process the substrate W by means of at least one of etching, surface treatment, property imparting, forming a processing film, removing at least a portion of the film, and cleaning.

[0065] like Figure 1As shown, in addition to multiple substrate processing devices 100, the substrate processing system 10 also includes a fluid cabinet 32, a fluid tank 34, multiple loading ports LP, an indexing robot IR, a central robot CR, and a control device 20. The control device 20 controls the loading ports LP, the indexing robot IR, and the central robot CR.

[0066] Each loading port LP stacks and stores multiple substrates W. An indexing robot IR transports substrates W between the loading port LP and the central robot CR. Alternatively, a temporary mounting stage (path) for placing substrates W can be provided between the indexing robot IR and the central robot CR, allowing indirect transfer of substrates W between the two robots via the mounting stage. The central robot CR transports substrates W between the indexing robot IR and the substrate processing apparatus 100. Each substrate processing apparatus 100 sprays liquid onto the substrates W and processes them. The liquid includes a chemical solution and / or a rinsing solution. Alternatively, the liquid may also include other processing liquids. A fluid tank 32 is used to collect the liquid. Furthermore, the fluid tank 32 can also collect gas.

[0067] Specifically, multiple substrate processing devices 100 are configured to surround a central robotic arm CR in a top-view arrangement, forming multiple towers TW (in... Figure 2 There are four towers (TW). Each tower (TW) contains multiple substrate processing devices 100 stacked on top of each other (in... Figure 1 There are three substrate processing units 100. Fluid tanks 34 correspond to multiple towers TW. Liquid in the fluid tank 32 is supplied to all substrate processing units 100 included in the tower TW corresponding to any one of the fluid tanks 34. In addition, gas in the fluid tank 32 is supplied to all substrate processing units 100 included in the tower TW corresponding to any one of the fluid tanks 34.

[0068] The control device 20 controls various operations of the substrate processing system 10. The control device 20 includes a control unit 22 and a storage unit 24. The control unit 22 has a processor. For example, the control unit 22 may have a central processing unit (CPU). Alternatively, the control unit 22 may also have a general-purpose arithmetic unit.

[0069] Storage unit 24 stores data and computer programs. The data includes procedure data. The procedure data contains information representing multiple procedures. Each procedure specifies the processing content and order for the substrate W.

[0070] Storage unit 24 includes a main storage device and an auxiliary storage device. The main storage device is, for example, a semiconductor memory. The auxiliary storage device is, for example, a semiconductor memory and / or a hard disk drive. Storage unit 24 may also include a removable medium. Control unit 22 executes the computer program stored in storage unit 24 to perform board processing operations.

[0071] The storage unit 24 stores a computer program in a pre-defined order. The board processing apparatus 100 operates in the order specified in the computer program.

[0072] In addition, Figure 2 Although the substrate processing system 10 is represented as having one control device 20, it may also have a control device 20 in each substrate processing device 100. However, in this case, it is preferable that the substrate processing system 10 has other control devices for controlling multiple substrate processing devices 100 and devices other than substrate processing devices 100.

[0073] Next, refer to Figure 3 The substrate processing apparatus 100 of this embodiment will be described. Figure 3 This is a schematic diagram of the substrate processing apparatus 100 of this embodiment. Furthermore, although the configuration of the substrate processing apparatus 100 is described here, the substrate processing apparatus 100L also has the same configuration as the substrate processing apparatus 100.

[0074] The substrate processing apparatus 100 processes the substrate W. The substrate processing apparatus 100 includes a chamber 110, a substrate holding section 120, a chemical supply section 130, and a rinsing solution supply section 140. The chamber 110 houses the substrate W. The substrate holding section 120 holds the substrate W. The substrate holding section 120 holds the substrate W so that it can rotate.

[0075] The chamber 110 is roughly box-shaped with an internal space. The chamber 110 houses the substrate W. Here, the substrate processing apparatus 100 is a monolithic processing apparatus that processes substrates W one by one, with each substrate W being housed in the chamber 110. The substrate W is housed within the chamber 110 and processed within it. At least a portion of each of the substrate holding section 120, the chemical supply section 130, and the rinsing liquid supply section 140 are housed within the chamber 110.

[0076] The substrate holding portion 120 holds the substrate W. The substrate holding portion 120 holds the substrate W horizontally such that the upper surface Wa of the substrate W faces upward and the back surface (lower surface) Wb of the substrate W faces vertically downward. In addition, the substrate holding portion 120 rotates the substrate W while holding the substrate W.

[0077] For example, the substrate holding portion 120 may also be a clamping type holding portion that clamps the end of the substrate W. Alternatively, the substrate holding portion 120 may also have any mechanism for holding the substrate W from the back side Wb. For example, the substrate holding portion 120 may also be a vacuum type holding portion. In this case, the substrate holding portion 120 holds the substrate W horizontally by adsorbing the central portion of the back side Wb of the substrate W (the non-device forming surface) onto the upper surface. Alternatively, the substrate holding portion 120 may also be a combination of a clamping type holding portion that contacts the peripheral end face of the substrate W with a vacuum type holding portion.

[0078] For example, the substrate holding portion 120 includes a rotating base 121, a clamping member 122, a shaft 123, and an electric motor 124. The clamping member 122 is disposed on the rotating base 121. The clamping member 122 clamps the substrate W. Typically, a plurality of clamping members 122 are disposed on the rotating base 121.

[0079] Shaft 123 is a hollow shaft. Shaft 123 extends vertically along the rotation axis Ax. A rotating substrate 121 is attached to the upper end of shaft 123. The back side of substrate W contacts the rotating substrate 121, and substrate W is placed on top of rotating substrate 121.

[0080] The rotating base 121 is a circular plate that horizontally supports the substrate W. A shaft 123 extends downward from the center of the rotating base 121. An electric motor 124 applies rotational force to the shaft 123. The electric motor 124 rotates the substrate W and the rotating base 121 around the rotation axis Ax by rotating the shaft 123 in the rotational direction. Here, the rotation direction is counterclockwise.

[0081] The solution supply unit 130 supplies solution to the substrate W. Thus, the substrate W is treated with the solution.

[0082] For example, the solution may contain hydrofluoric acid (hydrogen fluoride aqueous solution: HF). Alternatively, the solution may be a liquid containing at least one of sulfuric acid, acetic acid, nitric acid, hydrochloric acid, citric acid, buffered hydrofluoric acid (BHF), dilute hydrofluoric acid (DHF), ammonia, dilute ammonia, hydrogen peroxide solution, organic base (e.g., TMAH: tetramethylammonium hydroxide), surfactant, and preservative. Furthermore, the solution may be a mixture of the above liquids. Examples of solutions made by mixing these liquids include SPM (sulfuric acid-hydrogen peroxide solution mixture), SC1 (ammonia-hydrogen peroxide solution mixture), and SC2 (hydrochloric acid-hydrogen peroxide solution mixture).

[0083] The liquid supply unit 130 includes a nozzle 132, a pipe 134, and a valve 136. The nozzle 132 faces the upper surface Wa of the substrate W and sprays liquid medicine towards the upper surface Wa of the substrate W. The pipe 134 is connected to the nozzle 132. The nozzle 132 is located at the top end of the pipe 134. Liquid medicine is supplied from a supply source to the pipe 134. The valve 136 is located in the pipe 134. The valve 136 opens and closes the flow path within the pipe 134.

[0084] The liquid supply unit 130 also includes a nozzle moving unit 138. The nozzle moving unit 138 moves the nozzle 132 between a spraying position and a retracted position. When the nozzle 132 is in the spraying position, the nozzle 132 is located above the substrate W. When the nozzle 132 is in the spraying position, the nozzle 132 sprays liquid towards the upper surface Wa of the substrate W. When the nozzle 132 is in the retracted position, the nozzle 132 is located further radially outward from the substrate W than the substrate W.

[0085] The nozzle moving part 138 includes an arm 138a, a rotating shaft 138b, and a moving mechanism 138c. The arm 138a extends in a generally horizontal direction. A nozzle 132 is mounted at the top end of the arm 138a. The arm 138a is connected to the rotating shaft 138b. The rotating shaft 138b extends in a generally vertical direction. The moving mechanism 138c rotates the rotating shaft 138b about a rotation axis in the generally vertical direction, causing the arm 138a to rotate in a generally horizontal direction. As a result, the nozzle 132 moves in a generally horizontal direction. For example, the moving mechanism 138c includes an arm swing motor that rotates the rotating shaft 138b about the rotation axis. The arm swing motor is, for example, a servo motor. Additionally, the moving mechanism 138c raises and lowers the rotating shaft 138b in the generally vertical direction, causing the arm 138a to rise and fall. As a result, the nozzle 132 moves in the generally vertical direction. For example, the moving mechanism 138c includes a ball screw mechanism and an arm lifting motor that provides driving force to the ball screw mechanism. The arm lifting motor is, for example, a servo motor.

[0086] The rinsing solution supply unit 140 supplies rinsing solution to the substrates W. The rinsing solution may also include any of the following: deionized water (DIW), carbonated water, electrolyzed water, ozone water, ammonia water, hydrochloric acid water with a dilution concentration (e.g., about 10 ppm to 100 ppm), or reduced water (hydrogen water).

[0087] The rinsing fluid supply unit 140 includes a nozzle 142, a pipe 144, and a valve 146. The nozzle 142 faces the upper surface Wa of the substrate W and sprays rinsing fluid towards the upper surface Wa of the substrate W. The pipe 144 is connected to the nozzle 142. The nozzle 142 is located at the top end of the pipe 144. Rinsing fluid is supplied from a supply source to the pipe 144. The valve 146 is located on the pipe 144. The valve 146 opens and closes the flow path within the pipe 144.

[0088] The substrate processing apparatus 100 also includes a cup-shaped body 180. The cup-shaped body 180 collects liquid spilled from the substrate W. The cup-shaped body 180 can rise vertically upwards until it reaches the side of the substrate W. Alternatively, the cup-shaped body 180 can also descend vertically downwards from the side of the substrate W.

[0089] The control device 20 controls various operations of the substrate processing apparatus 100. The control unit 22 controls the substrate holding unit 120, the liquid supply unit 130, the rinsing liquid supply unit 140, and / or the cup-shaped body 180. In one example, the control unit 22 controls the electric motor 124, valves 136 and 146, the moving mechanism 138c, and / or the cup-shaped body 180.

[0090] In the substrate processing apparatus 100 of this embodiment, the substrate W can be subjected to chemical treatment and rinsing treatment.

[0091] Next, refer to Figures 1-4 The substrate processing apparatus 100 of this embodiment will be described. Figure 4 This is a block diagram of a substrate processing system 10 equipped with a substrate processing apparatus 100.

[0092] like Figure 4 As shown, the control device 20 controls various operations of the substrate processing system 10. The control device 20 controls the indexing robot IR, the central robot CR, the substrate holding unit 120, the chemical supply unit 130, the rinsing fluid supply unit 140, and the cup-shaped body 180. Specifically, the control device 20 controls the indexing robot IR, the central robot CR, the substrate holding unit 120, the chemical supply unit 130, the rinsing fluid supply unit 140, and the cup-shaped body 180 by sending control signals to them.

[0093] Specifically, the control unit 22 controls the indexing robot IR and transfers the substrate W through the indexing robot IR.

[0094] The control unit 22 controls the central robot CR, which transfers the substrate W. For example, the central robot CR picks up the unprocessed substrate W and moves it into any one of the multiple chambers 110. Alternatively, the central robot CR picks up the processed substrate W from the chamber 110 and moves it out.

[0095] The control unit 22 controls the substrate holding unit 120 to control the start of the substrate W's rotation, the change of its rotation speed, and the stop of its rotation. For example, the control unit 22 can control the substrate holding unit 120 to change its rotational speed. Specifically, the control unit 22 can change the rotational speed of the substrate W by changing the rotational speed of the electric motor 124 of the substrate holding unit 120.

[0096] The control unit 22 can control valve 136 of the liquid supply unit 130 and valve 146 of the flushing fluid supply unit 140, switching the states of valves 136 and 146 to open and closed states, respectively. Specifically, by controlling valves 136 and 146, the control unit 22 can open valves 136 and 146, allowing the liquid medicine and flushing fluid flowing in pipes 134 and 144 to flow to nozzles 132 and 142. Conversely, by controlling valves 136 of the liquid supply unit 130 and valve 146 of the flushing fluid supply unit 140, the control unit 22 can close valves 136 and 146, stopping the flow of liquid medicine and flushing fluid flowing in pipes 134 and 144 to nozzles 132 and 142.

[0097] Furthermore, the control unit 22 controls the movement of the arm 138a in the horizontal and / or vertical directions via the moving mechanism 138c. Thus, the control unit 22 can move the nozzle 132 mounted at the tip of the arm 138a on the upper surface Wa of the substrate W. Additionally, the control unit 22 can move the nozzle 132 mounted at the tip of the arm 138a between an ejection position and a retraction position. The substrate processing apparatus 100 of this embodiment is suitable for use in forming semiconductor devices.

[0098] In the substrate processing apparatus 100 of this embodiment, the storage unit 24 stores the learning completion model LM and the control program PG. The substrate processing apparatus 100 operates according to the sequence specified in the control program PG.

[0099] Additionally, the control unit 22 includes a substrate information acquisition unit 22a and a solution treatment condition information acquisition unit 22b. The substrate information acquisition unit 22a acquires substrate information of the substrate Wp to be processed. This substrate information includes hardened layer thickness information or ion implantation condition information; the hardened layer thickness information indicates the thickness of the hardened layer on the substrate Wp, and the ion implantation condition information indicates the ion implantation conditions performed on the substrate Wp. Furthermore, the substrate information acquisition unit 22a can also acquire other information besides hardened layer thickness information or ion implantation condition information from the storage unit 24 as substrate information.

[0100] The Learning Completion Model (LM) generates chemical treatment condition information based on substrate information. Typically, if substrate information is input to the LM, chemical treatment condition information corresponding to the substrate information is output. In one example, if hardened layer thickness information or ion implantation condition information is input to the LM, chemical treatment condition information corresponding to the hardened layer thickness information or ion implantation condition information is output.

[0101] The drug processing condition information acquisition unit 22b acquires the drug processing condition information through a self-learning completion model LM. The drug processing condition information acquisition unit 22b acquires the drug processing condition information in a manner corresponding to the substrate information of the substrate Wp to be processed.

[0102] The control unit 22 controls the substrate holding unit 120 and the liquid supply unit 130 according to the liquid treatment conditions indicated by the liquid treatment condition information.

[0103] The substrate processing system 10 preferably also includes a display unit 42, an input unit 44, and a communication unit 46.

[0104] Display unit 42 displays images. Display unit 42 is, for example, a liquid crystal display or an organic electroluminescent display.

[0105] The input unit 44 is an input device used to input various information to the control unit 22. For example, the input unit 44 is a keyboard and pointing device or a touch panel.

[0106] The communication unit 46 is connected to a network to communicate with external devices. In this embodiment, the network includes, for example, the Internet, a local area network (LAN), the public telephone network, and a short-range wireless network. The communication unit 46 is a communication device, such as a network interface controller.

[0107] Furthermore, the substrate processing system 10 preferably also includes sensors 50. Typically, multiple sensors 50 detect the state of various parts of the substrate processing system 10. For example, at least a portion of the sensors 50 detects the state of various parts of the substrate processing apparatus 100.

[0108] The storage unit 24 stores the output results from the sensor 50 and the control parameters of the control program as time series data TD. Typically, the time series data is stored separately for each substrate W.

[0109] During the processing of each substrate W, sensor 50 detects the physical quantity of the object used by the substrate processing apparatus 100 from the start to the end of processing, and outputs a detection signal representing the physical quantity to control unit 22. Then, during the processing of each substrate W, control unit 22 establishes a correlation between the physical quantity and time and stores it as time series data TD in storage unit 24. This physical quantity is represented by the detection signal output from sensor 50 during the period from the start to the end of processing of substrate W.

[0110] The control unit 22 acquires time-series data TD from the sensor 50 and stores the time-series data TD in the storage unit 24. In this case, the control unit 22 associates the time-series data TD with batch identification information, substrate identification information, processing sequence information, and batch interval information and stores it in the storage unit 24. Batch identification information is information used to identify batches (e.g., batch number). A batch represents a processing unit of substrate W. One batch is composed of a predetermined number of substrates W. Substrate identification information is information used to identify substrate W. Processing sequence information is information indicating the order in which the predetermined number of substrates W constituting one batch are processed. Batch interval information is information indicating the time interval from the end of processing a batch to the beginning of processing the next batch.

[0111] Next, refer to Figures 1-5 The substrate processing method of the substrate processing apparatus 100 of this embodiment will be described. Figure 5 (a) is a flowchart of the substrate processing method in the substrate processing apparatus 100 of this embodiment. Figure 5 (b) is a flowchart of the chemical solution treatment in the substrate processing method of this embodiment.

[0112] like Figure 5 As shown in (a), in step S10, the substrate Wp to be processed is moved into the substrate processing apparatus 100. The moved substrate Wp is mounted in the substrate holding section 120. Typically, the substrate Wp to be processed is moved into the substrate processing apparatus 100 by a central robot CR.

[0113] In step S20, the substrate Wp to be processed is treated with a chemical solution. The chemical solution supply unit 130 supplies the chemical solution to the substrate Wp. The chemical solution is sprayed from the nozzle 132 of the chemical solution supply unit 130 onto the upper surface Wa of the substrate Wp. The chemical solution covers the upper surface Wa of the substrate Wp. Thus, the substrate Wp is treated with the chemical solution. Furthermore, while the substrate Wp is being treated with the chemical solution, the substrate Wp rotates via the substrate holding unit 120. The rotation of the substrate Wp can continue until just before the substrate Wp is about to be removed.

[0114] In step S30, the substrate Wp to be processed is rinsed with a rinsing solution. The rinsing solution supply unit 140 supplies rinsing solution to the substrate Wp. The rinsing solution is sprayed from the nozzle 142 of the rinsing solution supply unit 140 onto the upper surface Wa of the substrate Wp. The rinsing solution covers the upper surface Wa of the substrate Wp. Thus, the substrate Wp is processed using the rinsing solution.

[0115] In step S40, the substrate Wp to be processed is detached from the substrate holding section 120 and removed. Typically, the substrate Wp to be processed is removed from the substrate processing apparatus 100 by a central robot CR. As described above, the substrate Wp to be processed can be subjected to chemical treatment.

[0116] In the substrate processing apparatus 100 of this embodiment, such as Figure 5 (b) indicates that the substrate Wp to be processed is treated with a chemical solution.

[0117] In step S22, substrate information of the substrate Wp to be processed is obtained. The substrate information acquisition unit 22a acquires the substrate information of the substrate Wp to be processed. The substrate information includes hardened layer thickness information or ion implantation condition information of the substrate Wp to be processed.

[0118] For example, the control unit 22 obtains information on the thickness of the hardened layer or ion implantation conditions of the substrate Wp to be processed from the storage unit 24. Furthermore, the thickness of the hardened layer in the substrate Wp to be processed can also be measured within the substrate processing system 10 or the substrate processing apparatus 100. Alternatively, the thickness of the hardened layer in the substrate Wp to be processed can also be measured outside the substrate processing system 10 or the substrate processing apparatus 100.

[0119] Alternatively, ion implantation of the substrate Wp to be processed can be performed within the substrate processing system 10 or the substrate processing apparatus 100. Furthermore, ion implantation of the substrate Wp to be processed can also be performed outside the substrate processing system 10 or the substrate processing apparatus 100. In addition, the control unit 22 can also obtain substrate information other than hardened layer thickness information or ion implantation condition information.

[0120] In step S24, the substrate information of the target substrate Wp is input into the learning completion model LM. Details will be described below. The learning completion model LM is constructed from learning data, which includes: substrate information of the target substrate WL, chemical treatment condition information indicating the treatment conditions performed on the target substrate WL, and processing result information indicating the result of the chemical treatment performed on the target substrate WL. The learning completion model LM outputs chemical treatment condition information Rp corresponding to the substrate information of the target substrate Wp.

[0121] In step S26, the self-learning completion model LM acquires the drug treatment condition information. The drug treatment condition information acquisition unit 22b acquires the drug treatment condition information corresponding to the substrate information from the self-learning completion model LM.

[0122] In step S28, based on the chemical treatment condition information, the substrate holding unit 120 and the chemical supply unit 130 perform chemical treatment on the substrate Wp to be treated. Figure 3 In the substrate processing apparatus 100 shown, the solution supply unit 130 supplies solution to the substrate Wp to be processed according to solution processing condition information. As described above, the substrate Wp to be processed can be processed with solution.

[0123] According to this embodiment, a learning completion model LM constructed through machine learning is used to obtain chemical treatment condition information corresponding to substrate information. Chemical treatment is performed according to the chemical treatment conditions indicated in the chemical treatment condition information. The substrate information includes hardened layer thickness information or ion implantation condition information of the substrate Wp to be treated. According to this embodiment, chemical treatment can be appropriately performed in accordance with the thickness of the hardened layer in the substrate Wp to be treated.

[0124] Next, refer to Figure 6 The hardened layer formed on the substrate W will be described, and the substrate W is the object of the substrate processing method of this embodiment. Figure 6 (a)~ Figure 6 (c) is a schematic diagram illustrating the formation of a hardened layer in the resist layer R of the substrate W.

[0125] like Figure 6 As shown in (a), a resist layer R is formed on the upper surface of the substrate W. The resist layer R extends in a vertical direction relative to the main surface of the substrate W. The resist layer R is patterned into a predetermined shape.

[0126] like Figure 6 As shown in (b), ions are implanted into the substrate W. Here, ions are implanted in a direction parallel to the normal direction of the main surface of the substrate W. Through ion implantation, the surface properties of the substrate W are altered. At this time, the surface of the resist layer R is also altered to form a hardened layer Rc. In addition, the interior of the resist layer R remains unchanged and remains as an inner layer Ri. The hardened layer Rc is harder than the inner layer Ri.

[0127] like Figure 6 As shown in (c), the hardened layer Rc has a specified thickness d. Specifically, the thickness d of the hardened layer Rc includes the height dt and the width dw of the hardened layer Rc. In the case where ions are isotropically implanted into the anti-corrosion layer R, the height dt and the width dw are approximately equal.

[0128] On the other hand, if ions are implanted anisotropically into the anti-corrosion layer R, the height dt will differ from the width dw. For example, as... Figure 6 As shown in (b), if ions are injected from a direction parallel to the normal direction of the main surface of the substrate W, the height dt becomes greater than the width dw.

[0129] Next, refer to Figure 7 The hardened layer formed on the substrate W will be described, and the substrate W is the object of the substrate processing method of this embodiment. Figure 7 (a)~ Figure 7 (d) is a schematic diagram illustrating the formation of the hardened layer in the resist layer R of the substrate W.

[0130] like Figure 7 As shown in (a), a resist layer R is formed on the upper surface of the substrate W. The resist layer R extends in a vertical direction relative to the main surface of the substrate W.

[0131] like Figure 7 As shown in (b), ions are implanted into the substrate W. Here, ions are implanted from the normal direction relative to the main surface of the substrate W in a direction inclined in one direction (leftward relative to the paper plane). Through ion implantation, the surface properties of the substrate W are altered. At this time, the surface of the resist layer R is also altered to form a hardened layer Rc. In this case, the left side of the hardened layer Rc is relatively thicker than the right side of the resist layer R, while the right side of the hardened layer Rc is relatively thinner.

[0132] like Figure 7 As shown in (c), ions are implanted into the substrate W. Here, the ions are implanted in a direction inclined to the right relative to the plane of the paper, from the normal direction relative to the main surface of the substrate W. Through ion implantation, the hardened layer Rc, which is altered on the surface of the resist layer R, expands.

[0133] like Figure 7 As shown in (d), the hardened layer Rc has a specified thickness d. Specifically, the thickness d of the hardened layer Rc includes the height dt and the width dw of the hardened layer Rc. As described above, the hardened layer Rc can be formed.

[0134] In addition, Figure 7 (b) and Figure 7 In (c), since ion implantation is performed from the left and right oblique directions of the resist layer R, the widths of the left and right sides of the hardened layer Rc can be made approximately equal. Furthermore, since ion implantation is performed from the oblique direction of the resist layer R, not only the height dt of the hardened layer Rc can be increased, but also the width dw of the hardened layer Rc can be increased.

[0135] As described above, the thickness d, height dt, and / or width dw of the hardened layer Rc can also be measured using measuring equipment. Alternatively, the thickness d, height dt, and / or width dw of the hardened layer Rc can be roughly defined by the characteristics of the resist layer R (composition, thickness, width, etc.) and / or the ion implantation conditions (ion type, acceleration energy, implantation amount, and implantation direction, etc.). Therefore, the thickness d, height dt, and / or width dw of the hardened layer Rc can be specified according to the ion implantation conditions.

[0136] Reference Figure 1 As described above, the learning model LM is generated based on the learning data LD, which in turn is generated based on the time series data TDL of the substrate processing device 100L.

[0137] Next, refer to Figure 8 The generation of learning data LD is explained. Figure 8 This is a block diagram of a substrate processing system 10L including a substrate processing apparatus 100L and a learning data generation apparatus 300. Here, the learning data generation apparatus 300 is communicatively connected to the substrate processing apparatus 100L. Figure 8 In the substrate processing system 10L equipped with substrate processing apparatus 100L, the control unit 22L does not have a substrate information acquisition unit 22a and a chemical solution processing condition information acquisition unit 22b, and the storage unit 24L does not store the learning completion model LM but stores the test procedure TR. Otherwise, it is similar to... Figure 4 The block diagram of the substrate processing system 10 shown is the same, and repeated descriptions are omitted to avoid being too lengthy.

[0138] The substrate processing system 10L includes multiple substrate processing devices 100L, an indexing robot IRL, a central robot CRL, a control device 20L, a display unit 42L, an input unit 44L, a communication unit 46L, and a sensor 50L. The substrate processing devices 100L, IRL, CRL, 20L, 42L, 44L, and 46L are integrated with... Figure 4 The substrate processing system 10 shown has the same configuration as the substrate processing apparatus 100, indexing robot IR, central robot CR, control device 20, display unit 42, input unit 44 and communication unit 46.

[0139] Additionally, the substrate processing apparatus 100L includes a substrate holding section 120L, a chemical supply section 130L, a rinsing solution supply section 140L, and a cup-shaped body 180L. Preferably, the chamber 110L, substrate holding section 120L, chemical supply section 130L, rinsing solution supply section 140L, and cup-shaped body 180L are integrated with... Figure 3 and Figure 4The substrate holding part 120, the liquid supply part 130, the rinsing liquid supply part 140 and the cup-shaped body 180 shown have the same configuration.

[0140] The control device 20L includes a control unit 22L and a storage unit 24L. The storage unit 24L stores a control program PGL. The substrate processing device 100L operates according to the sequence specified in the control program PGL.

[0141] In addition, the storage unit 24L stores multiple test procedures TR. These multiple test procedures TR contain procedures with different chemical treatment conditions. Therefore, when the control unit 22L processes the learning target substrate WL according to the test procedures TR, different chemical treatments are applied to different learning target substrates WL.

[0142] The storage unit 24L stores time-series data (TDL) of the learning target substrate WL. The time-series data TDL represents the time-varying changes of physical quantities within the substrate processing apparatus 100L. The time-series data TDL represents multiple physical quantities detected by the sensor 50L. The time-series data TDL may also include data representing the manufacturing process prior to processing the learning target substrate WL using the substrate processing apparatus 100L. Furthermore, the time-series data TDL includes substrate information, chemical treatment condition information, and processing result information. The substrate information includes hardened layer thickness information or ion implantation condition information; the chemical treatment condition information indicates the chemical treatment conditions performed on the learning target substrate WL; and the processing result information indicates the result of the chemical treatment performed on the learning target substrate WL.

[0143] The learning data generation device 300 is communicatively connected to the substrate processing device 100L. The learning data generation device 300 communicates with at least a portion of the time-series data TDL of the substrate processing device 100L.

[0144] The learning data generation apparatus 300 includes a control unit 320, a display unit 342, an input unit 344, and a communication unit 346. The learning data generation apparatus 300 can communicate with the control unit 20L of multiple substrate processing apparatuses 100L via the communication unit 346. The display unit 342, the input unit 344, and the communication unit 346 have the same configuration as the display unit 42, the input unit 44, and the communication unit 46.

[0145] The control device 320 includes a control unit 322 and a storage unit 324. The storage unit 324 stores the control program PG3. The learning data generation device 300 operates according to the sequence specified in the control program PG3.

[0146] The control unit 322 receives at least a portion of the time-series data TDL from the substrate processing apparatus 100L and stores the received time-series data TDL in the storage unit 324. The storage unit 324 stores at least a portion of the time-series data TDL of the learning target substrate WL. The time-series data TDL is transmitted from the substrate processing apparatus 100L to the learning data generation apparatus 300 via the communication unit 46L and the communication unit 346. The control unit 322 stores at least a portion of the transmitted time-series data TDL in the storage unit 324. The time-series data TDL stored in the storage unit 324 includes substrate information, chemical treatment condition information, and processing result information of the time-series data TDL.

[0147] The control unit 322 obtains the substrate information, chemical treatment condition information, and processing result information of the learning target substrate WL from the time-series data TDL in the storage unit 324. Furthermore, the control unit 322 summarizes the substrate information, chemical treatment condition information, and processing result information of the learning target substrate WL to generate learning data LD, and the storage unit 324 stores the learning data LD.

[0148] Next, refer to Figure 8 and Figure 9 The method for generating learning data in this embodiment will be explained. Figure 9 This is a flowchart of the method for generating learning data according to this embodiment. The generation of learning data is performed in the learning data generation apparatus 300.

[0149] like Figure 9 As shown, in step S110, time-series data TDL of the learning target substrate WL is acquired. Typically, the learning data generation apparatus 300 receives at least a portion of the time-series data TDL of the learning target substrate WL from the substrate processing apparatus 100L. The storage unit 324 stores the received time-series data TDL.

[0150] In step S112, substrate information is extracted from the time-series data TDL of the learning target substrate WL stored in the storage unit 324. The substrate information includes hardened layer thickness information or ion implantation condition information. The control unit 322 obtains the substrate information of the learning target substrate WL from the time-series data TDL in the storage unit 324.

[0151] In step S114, the drug treatment condition information of the learning target substrate WL is extracted from the time-series data TDL of the learning target substrate WL stored in the storage unit 324. The control unit 322 obtains the drug treatment condition information of the learning target substrate WL from the time-series data TDL in the storage unit 324.

[0152] In step S116, the processing result information of the learning target substrate WL is extracted from the time-series data TDL of the learning target substrate WL stored in the storage unit 324. The control unit 322 obtains the processing result information of the learning target substrate WL from the time-series data TDL in the storage unit 324.

[0153] In step S118, the substrate information, liquid treatment condition information and processing result information of the learning target substrate WL are associated to generate learning data LD, and the storage unit 324 stores the learning data LD for each of the multiple learning target substrates WL.

[0154] In this embodiment, the generated learning data includes substrate information, chemical treatment condition information, and processing result information that are mutually associated for each learning target substrate WL. This learning data is suitable for use in learning processing.

[0155] In addition, Figure 8 In this embodiment, the learning data generation device 300 can be communicatively connected to one substrate processing device 100L, but this embodiment is not limited to this. The learning data generation device 300 can also be communicatively connected to multiple substrate processing devices 100L.

[0156] In addition, in reference Figure 8 and Figure 9 In the description, the time-series data TDL generated by the substrate processing apparatus 100L is transmitted to the learning data generation apparatus 300 via the communication unit 46L and the communication unit 346; however, this embodiment is not limited to this. Alternatively, the control device 320 of the learning data generation apparatus 300 may be assembled into the control device 20 of the substrate processing system 10 equipped with the substrate processing apparatus 100L, so that the time-series data TDL is not transmitted over the network, but the learning data LD is generated within the substrate processing system 10 based on the time-series data TDL.

[0157] Next, refer to Figure 10 The generation of the learning completion model LM in this implementation method will be explained. Figure 10 This is a schematic diagram of the learning data generation apparatus 300 and the learning device 400 according to this embodiment. The learning data generation apparatus 300 and the learning device 400 can communicate with each other.

[0158] The learning device 400 is communicatively connected to the learning data generation device 300. The learning device 400 receives learning data LD from the learning data generation device 300. The learning device 400 performs machine learning based on the learning data LD and generates a completed learning model LM.

[0159] The learning device 400 includes a control unit 420, a display unit 442, an input unit 444, and a communication unit 446. The display unit 442, input unit 444, and communication unit 446 have the ability to communicate with... Figure 4 The display unit 42, input unit 44, and communication unit 46 of the substrate processing system 10 shown have the same configuration.

[0160] The control device 420 includes a control unit 422 and a storage unit 424. The storage unit 424 stores a control program PG4. The learning device 400 operates according to the sequence specified in the control program PG4.

[0161] Storage unit 424 stores learning data LD. The learning data LD is transmitted from learning data generation device 300 to learning device 400 via communication unit 346. Control unit 422 stores the transmitted learning data LD in storage unit 424. In the learning data LD stored in storage unit 424, substrate information, drug treatment condition information, and processing result information of time series data TDL are mutually correlated.

[0162] Storage unit 424 stores the learning program LPG. The learning program LPG is a program used to execute machine learning algorithms, which are used to find certain rules from multiple learning data LDs and generate a learning completion model LM that represents the found rules. Control unit 422 executes the learning program LPG of storage unit 424 to perform machine learning on the learning data LDs and adjust the parameters of the inference program to generate the learning completion model LM.

[0163] Machine learning algorithms are not particularly limited as long as they have a teacher to learn from; for example, they can be decision trees, nearest neighbor algorithms, simple Bayesian classifiers, support vector machines, or neural networks. Therefore, a learned model (LM) can include decision trees, nearest neighbor algorithms, simple Bayesian classifiers, support vector machines, or neural networks. Backpropagation can also be used in the machine learning process that generates the learned model (LM).

[0164] For example, a neural network consists of an input layer, one or more intermediate layers, and an output layer. Specifically, neural networks are deep neural networks (DNNs), recurrent neural networks (RNNs), or convolutional neural networks (CNNs) used for deep learning. For example, a deep neural network consists of an input layer, multiple intermediate layers, and an output layer.

[0165] The control unit 422 includes an acquisition unit 422a and a learning unit 422b. The acquisition unit 422a acquires learning data LD from the storage unit 424. The learning unit 422b performs machine learning on the learning data LD by executing the learning program LPG of the storage unit 424, and generates a learning completed model LM based on the learning data LD.

[0166] Learning unit 422b performs machine learning on multiple learning data LDs based on the learning program LPG. As a result, certain rules are identified from the multiple learning data LDs to generate a completed learning model LM. That is, the completed learning model LM is constructed by performing machine learning on the learning data LDs. Storage unit 424 stores the completed learning model LM.

[0167] Then, typically, the learned model LM is transferred to the substrate processing system 10, and the storage unit 24 stores the learned model LM. In this case, as shown in the reference... Figure 4 As described above, the storage unit 24 of the control device 20 in the substrate processing system 10 stores the learning completion model LM, and the drug processing condition information acquisition unit 22b acquires the drug processing conditions from the learning completion model LM stored in the storage unit 24.

[0168] However, this embodiment is not limited to this. Alternatively, the storage unit 24 may not store the learning completion model LM, and the liquid treatment condition information acquisition unit 22b may acquire the liquid treatment conditions from outside the substrate processing system 10. For example, the liquid treatment condition information acquisition unit 22b may also send the substrate information of the substrate Wp to be processed to the learning completion model LM of the learning device 400 via the communication unit 46 and the communication unit 446, and receive the liquid treatment condition information output from the learning completion model LM from the learning device 400 via the communication unit 446 and the communication unit 46.

[0169] Next, refer to Figures 1 to 11 The learning method in the learning device 400 of this embodiment will be described. Figure 11 This is a flowchart of the learning method in this embodiment. The learning of the learning data LD and the generation of the learning completion model LM are performed in the learning device 400.

[0170] like Figure 11 As shown, in step S122, the acquisition unit 422a of the learning device 400 acquires multiple learning data LDs from the storage unit 424. In the learning data LDs, substrate information, drug treatment condition information, and processing result information of the learning target substrate WL are mutually associated.

[0171] Next, in step S124, the learning unit 422b performs machine learning on multiple learning data LDs based on the learning program LPG.

[0172] Next, in step S126, the learning unit 422b determines whether the machine learning of the learning data LD has ended. Whether to end the machine learning is determined according to pre-defined conditions. For example, if machine learning is performed on a predetermined number or more of the learning data LD, then the machine learning ends.

[0173] If the machine learning process has not ended (No in step S126), the process returns to step S122. In this case, the machine learning is repeated. On the other hand, if the machine learning process has ended (Yes in step S126), the process proceeds to step S128.

[0174] In step S128, the learning unit 422b outputs the model (one or more functions) with the latest multiple parameters (coefficients), i.e., multiple learned parameters (coefficients), as the learned completed model LM. The storage unit 424 stores the learned completed model LM.

[0175] As described above, the learning method concludes, generating a completed learning model LM. According to this embodiment, by performing machine learning on the learning data LD, a completed learning model LM can be generated.

[0176] In addition, Figure 10 In this embodiment, the learning device 400 can be communicatively connected to one learning data generation device 300, but this embodiment is not limited to this. The learning device 400 can also be communicatively connected to multiple learning data generation devices 300.

[0177] In addition, in reference Figure 10 and Figure 11 In the description, the learning data LD generated by the learning data generation device 300 is transmitted to the learning device 400 via the communication unit 346 and the communication unit 446, but this embodiment is not limited to this. Alternatively, the control device 420 of the learning device 400 may be assembled into the control device 320 of the learning data generation device 300, and the learning data LD may not be transmitted over the network, but a learning completion model LM may be generated within the learning data generation device 300 based on the learning data LD.

[0178] Furthermore, in reference Figures 8-11In the description, the time-series data TDL generated by the substrate processing apparatus 100L is transmitted to the learning data generation apparatus 300 via communication unit 46L and communication unit 346, and the learning data LD generated by the learning data generation apparatus 300 is transmitted to the learning apparatus 400 via communication unit 346 and communication unit 446. However, this embodiment is not limited to this. Alternatively, the control device 320 of the learning data generation apparatus 300 and the control device 420 of the learning apparatus 400 may be assembled into the control device 20 of the substrate processing system 10L, without transmitting the time-series data TDL and the learning data LD over the network, and instead generating the learning completion model LM within the substrate processing system 10L based on the time-series data TDL and the learning data LD.

[0179] Secondly, refer to Figure 12 This example illustrates the use of learning data LD. Figure 12 This is a diagram representing an example of learning data LD.

[0180] Figure 12 This section displays the substrate information, chemical treatment conditions, and processing results of the learning substrate WL. Here, the substrate information refers to the hardened layer thickness of the learning substrate WL. The thickness of the hardened layer can be obtained by measuring the hardened layer within the resist layer of the learning substrate WL. The chemical treatment conditions information indicates the conditions under which the learning substrate WL was treated. These conditions may include, for example, the concentration of the chemical solution, the temperature of the chemical solution, the supply rate of the chemical solution, and the rotational speed of the learning substrate WL during chemical supply. The processing results indicate the final result of the learning substrate WL after chemical treatment. Figure 12 In this context, the learning data LD includes learning data LD1 to learning data LD1000.

[0181] The learning data LD1 represents the substrate information, chemical treatment conditions, and treatment results of a learning target substrate WL1. Here, in the learning data LD1, Ld1 represents the thickness of the hardened layer of the learning target substrate WL1. Additionally, Lp1 represents the chemical treatment conditions performed on the learning target substrate WL1.

[0182] The processing result information indicates the processing result of the chemical treatment on the learning target substrate WL1. The processing result can also be determined by whether any abnormalities in the characteristics of the learning target substrate WL1 are found. In the learning data LD1, if the result of the chemical treatment on the learning target substrate is good, it is represented as ○. On the other hand, if the result of the chemical treatment on the learning target substrate is not good, it is represented as ×.

[0183] The learning data LD2 to LD1000 are generated corresponding to the learning target substrates WL2 to WL1000. The chemical treatment conditions performed on the learning target substrates WL can be the same or different. The treatment results vary greatly depending on the thickness of the hardened layer of the learning target substrate WL.

[0184] exist Figure 12 The learning data LD shown contains 1000 data points, but this embodiment is not limited to this. The number of data points can be less than 1000 or more. However, it is preferable to have as many data points as possible.

[0185] Furthermore, in the learning data LD, the substrate information preferably includes multiple items. For example, the hardened layer thickness information may also include hardened layer height information indicating the height of the hardened layer and hardened layer width information indicating the width of the hardened layer.

[0186] Figure 13 This is a graph representing an example of learning data LD. For example... Figure 13 As shown, the learning data LD includes substrate information of the learning target substrate WL, liquid treatment condition information of the learning target substrate WL, and processing result information of the learning target substrate WL. Figure 13 The learning data LD, except that the substrate information of the learning target substrate WL includes hardened layer height and hardened layer width information, is similar to the reference data. Figure 12 The learning data LD is the same, and repeated records have been omitted to avoid being too lengthy.

[0187] Here, the substrate information refers to the height and width of the hardened layer of the learning object substrate WL. The height and width of the hardened layer can be obtained by measuring the hardened layer within the resist layer of the learning object substrate WL. Figure 13 In this context, the learning data LD includes learning data LD1 to LD1000.

[0188] The learning data LD1 represents the substrate information, chemical treatment conditions, and treatment results of a learning target substrate WL1. Here, in the learning data LD1, Ldt1 represents the height of the hardened layer of the learning target substrate WL1, and Ldw1 represents the width of the hardened layer of the learning target substrate WL1.

[0189] The learning data LD2 to LD1000 are generated corresponding to the learning target substrates WL2 to WL1000. The processing results vary significantly depending on the height and width of the hardened layer of the learning target substrate WL. Preferably, the learning data LD has parameters that greatly contribute to the variation of the chemical treatment results of the learning target substrate WL.

[0190] In addition, in reference Figure 12 and Figure 13 In the above description, a good processing result is represented by ○, and a bad processing result is represented by ×. The processing results of the learning data LD1 to LD1000 are binarized, but this embodiment is not limited to this. The processing results may also be classified into multiple values ​​of 3 or more. Alternatively, the processing results may be classified into any value between the minimum and maximum values. For example, in addition to the characteristics of the learning target substrate WL, the processing results may also be quantified considering the amount of liquid used (supply) or the time required for liquid treatment.

[0191] Furthermore, in the learning data LD, the solution processing conditions preferably include multiple items. For example, the solution processing conditions may also include the concentration of the solution, temperature, supply amount, the solution spray pattern on the learning target substrate, and the rotation speed of the substrate holding part 120 during solution processing.

[0192] Next, refer to Figure 14 The learning data LD used in the learning method of this embodiment will be explained. Figure 14 This is a diagram representing an example of learning data LD.

[0193] like Figure 14 As shown, the learning data LD includes learning data LD1 to LD1000. The learning data LD represents the substrate information of the learning target substrate WL, the chemical treatment conditions of the learning target substrate WL, and the processing result information of the learning target substrate WL. Here, the chemical treatment conditions include the concentration of the chemical solution, the temperature of the chemical solution, the supply amount of the chemical solution, the chemical spray pattern on the learning target substrate, and the rotation speed of the learning target substrate when supplying the chemical solution.

[0194] The concentration of the solution indicates the concentration of the solution used on the learning substrate WL. The temperature of the solution indicates the temperature of the solution used on the learning substrate WL. The supply amount of the solution indicates the supply amount of the solution used on the learning substrate WL. The spray pattern of the solution on the learning substrate indicates the path of spraying the solution onto the learning substrate WL (the operating path of the nozzle 132). The rotation speed of the learning substrate indicates the rotation speed of the learning substrate WL when the solution is supplied. Furthermore, in the learning data LD, a "○" indicates a good result when the learning substrate treated with the solution is good, and a "×" indicates a bad result when the learning substrate treated with the solution is not good.

[0195] In the learning data LD1, Lc1 represents the concentration of the solution used on the learning substrate WL1, and Lt1 represents the temperature of the solution used on the learning substrate WL1. Additionally, Ls1 represents the supply amount of the solution used on the learning substrate WL1, Le1 represents the spray pattern of the solution onto the learning substrate WL1, and Lv1 represents the rotational speed of the learning substrate WL1 when the solution is supplied. In the learning data LD1, since the processing result of the learning substrate WL1 treated with the solution is good, the processing result is "○".

[0196] The same applies to the learning data LD2 to LD1000. Furthermore, the chemical treatment conditions applied to the learning target substrate WL can be the same or different. For example, at least some of the multiple items of the chemical treatment conditions can be the same or different. Alternatively, all of the multiple items of the chemical treatment conditions can be the same or different. The processing result can vary depending on the substrate condition of the learning target substrate WL and the chemical treatment conditions.

[0197] Next, refer to Figures 1 to 15 The liquid treatment in the substrate processing apparatus 100 of this embodiment will be described. Figure 15 (a) shows a schematic diagram of the substrate Wp to be processed. Figure 15 (b) represents the drug treatment condition information Rp generated in the learned model LM.

[0198] like Figure 15 As shown in (a), the substrate Wp to be processed has a resist layer R on which a hardened layer Rc is formed. Here, the height of the hardened layer Rc is dt, and the width of the hardened layer Rc is dw.

[0199] Figure 15 (b) is a diagram representing the drug treatment condition information Rp. The drug treatment condition information Rp includes the concentration of the drug, the temperature of the drug, the supply amount of the drug, the spray pattern of the drug on the learning target substrate, and the rotation speed of the learning target substrate when the drug is supplied.

[0200] In the chemical treatment condition information Rp, Rc represents the concentration of the chemical solution used on the substrate Wp to be treated, and Rt represents the temperature of the chemical solution used on the substrate Wp to be treated. Additionally, Rs represents the supply amount of the chemical solution used on the substrate Wp to be treated, Pe represents the spray pattern of the chemical solution used on the substrate Wp to be treated, and Rv represents the rotational speed of the substrate Wp to be treated when supplying the chemical solution.

[0201] In this case, the control unit 22 controls the substrate holding unit 120 and the liquid supply unit 130 according to the liquid treatment conditions shown in the liquid treatment condition information Rp, so as to process the substrate Wp to be processed using the liquid.

[0202] In addition, in reference Figure 14 and Figure 15 The description of the chemical treatment conditions includes five items: chemical concentration, chemical temperature, chemical supply volume, chemical spray pattern, and rotation speed of the substrate being treated. However, this embodiment is not limited to these. The chemical treatment conditions may also include one or more of these five items. Alternatively, the chemical treatment conditions may be a combination of one or more of these five items with other items. Alternatively, the chemical treatment conditions may include one or more items that are different from these five items.

[0203] In addition, in reference Figures 12-15 In the above description, the substrate information of the learning target substrate WL is information obtained by measuring the thickness of the hardened layer; however, this embodiment is not limited to this. The substrate information of the learning target substrate WL may also include information related to the ion implantation conditions for forming the hardened layer.

[0204] Next, refer to Figure 16 The learning data LD used in the learning method of this embodiment will be explained. Figure 16 This is a graph representing an example of learning data LD. Furthermore, Figure 16 The learning data LD, besides representing the substrate information indicating the ion implantation conditions used to form the hardened layer of the learning object substrate WL, is similar to the reference data. Figure 12 The learning data LD is the same, and repeated records have been omitted to avoid being too lengthy.

[0205] like Figure 16 As shown, the learning data LD includes learning data LD1 to LD1000. The learning data LD represents substrate information of the learning target substrate WL, the chemical treatment conditions of the learning target substrate WL, and the treatment result information of the learning target substrate WL. Here, the substrate information of the learning target substrate WL includes ion implantation condition information indicating the ion implantation conditions used to form the hardened layer of the learning target substrate WL.

[0206] In addition, the ion implantation condition information includes ion type, acceleration energy, implantation amount, and implantation direction. Ion type indicates the type of ions used for ion implantation, and acceleration energy indicates the acceleration energy of the ion type during ion implantation. Furthermore, implantation amount indicates the amount of ion type being implanted, and implantation direction indicates the direction in which ions are implanted onto the target substrate WL.

[0207] In the learning data LD1, Lk1 represents the type of ions used for ion implantation when forming the hardened layer of the learning target substrate WL1, and La1 represents the acceleration energy during ion implantation when forming the hardened layer of the learning target substrate WL1. Additionally, Lu1 represents the ion implantation amount during the formation of the hardened layer of the learning target substrate WL1, and Ld1 represents the direction of ion implantation into the learning target substrate WL during the formation of the hardened layer of the learning target substrate WL1.

[0208] The same applies to the learning data LD2 to LD1000. Thus, the substrate information may not represent the thickness of the hardened layer itself, but rather the ion implantation conditions used to form the hardened layer. Furthermore, the ion implantation condition information may include any one of the following: ion type, acceleration energy, implantation amount, and implantation direction. Additionally, the ion implantation condition information may include any one of the following items in combination with other items: ion type, acceleration energy, implantation amount, and implantation direction. Alternatively, the ion implantation condition information may include more than one item other than ion type, acceleration energy, implantation amount, and implantation direction. Furthermore, the substrate information may also be a combination of hardened layer thickness information and ion implantation condition information.

[0209] In addition, Figure 3 In the substrate processing apparatus 100 shown, the liquid supply unit 130 supplies a liquid in a certain state to the substrate W, but this embodiment is not limited to this. The concentration and temperature of the liquid can also be appropriately changed.

[0210] Next, refer to Figure 17 The substrate processing apparatus 100 of this embodiment will be described. Figure 17 This is a schematic diagram of the substrate processing apparatus 100 according to this embodiment. Furthermore, Figure 17 The substrate processing apparatus 100, in addition to the adjustable concentration and temperature of the solution supplied from the solution supply unit 130, is similar to the reference... Figure 3 The substrate processing apparatus 100 is the same as described above, and repeated descriptions are omitted to avoid being too lengthy.

[0211] exist Figure 17 In the substrate processing apparatus 100 of this embodiment shown, the solution supply unit 130 supplies SPM as a solution to the substrate W. SPM is generated by mixing sulfuric acid and hydrogen peroxide solution. For example, the solution supply unit 130 can vary the concentration (mixing ratio) of the SPM when supplying it to the substrate W. Thus, the substrate W is treated with SPM.

[0212] The liquid supply unit 130 includes a nozzle 132, a pipe 134a, a valve 136a, an adjusting valve 137a, a pipe 134b, a valve 136b, and an adjusting valve 137b. The nozzle 132 faces the upper surface Wa of the substrate W and sprays liquid towards the upper surface Wa of the substrate W.

[0213] Pipe 134a is connected to nozzle 132. Nozzle 132 is located at the top of pipe 134a. Hydrogen peroxide solution is supplied from a supply source to pipe 134a. Valve 136a and regulating valve 137a are located on pipe 134a. Valve 136a opens and closes the flow path within pipe 134a. Regulating valve 137a regulates the flow rate of hydrogen peroxide solution through the flow path within pipe 134a.

[0214] Pipe 134b is connected to nozzle 132. Nozzle 132 is located at the top of pipe 134b. Sulfuric acid is supplied from a supply source to pipe 134b. The temperature of the sulfuric acid may be the same as or different from the temperature of the hydrogen peroxide solution. Valve 136b and regulating valve 137b are located on pipe 134b. Valve 136b opens and closes the flow path within pipe 134b. Regulating valve 137b regulates the flow rate of sulfuric acid through the flow path within pipe 134b.

[0215] According to this embodiment, by adjusting the regulating valves 137a and 137b, the flow rates of hydrogen peroxide solution and sulfuric acid flowing in the pipes 134a and 134b can be changed. Therefore, the concentration and temperature of SPM generated by mixing hydrogen peroxide solution and sulfuric acid can be adjusted.

[0216] In addition, as referenced Figure 6 and Figure 7 As described above, when removing the resist layer R of the substrate W, it is relatively easy to remove the inner layer Ri using the solution, but relatively difficult to remove the hardened layer Rc using the solution. Therefore, it is preferable to control the concentration of SPM to increase its removal capacity when removing the hardened layer Rc, and to control the concentration of SPM to decrease its removal capacity when removing the inner layer Ri.

[0217] For example, when removing the hardened layer Rc, the concentration of SPM is controlled by increasing the ratio of hydrogen peroxide solution in the SPM, while when removing the inner layer Ri, the concentration of SPM is controlled by decreasing the ratio of hydrogen peroxide solution in the SPM. This allows for the removal of the resist layer R while simultaneously suppressing damage to it.

[0218] Next, refer to Figure 18 The learning data LD used in the learning method of this embodiment will be explained. Figure 18 This is a diagram representing an example of learning data LD. Figure 18 The learning data LD is suitable for use in generating Figure 17 The learning completion model LM of the substrate processing apparatus 100 shown. Furthermore... Figure 18In addition to showing the distribution of time-varying physical parameters represented by the value of at least one item in the learning data LD, the learning data LD also differs from the reference data. Figure 1 The learning data LD is the same, and repeated records have been omitted to avoid being too lengthy.

[0219] like Figure 18 As shown, the learning data LD includes learning data LD1 to LD1000. The chemical treatment conditions include the concentration distribution of the chemical solution, the temperature distribution of the chemical solution, the supply amount of the chemical solution, the ejection pattern of the chemical solution onto the learning target substrate, and the rotation speed of the learning target substrate during chemical supply. The concentration distribution represents the time-varying concentration of the SPM during chemical treatment. The temperature distribution represents the time-varying temperature of the SPM used on the learning target substrate WL. Furthermore, the supply amount represents the supply amount of SPM used on the learning target substrate WL, and the ejection pattern represents the ejection pattern and timing of the SPM onto the learning target substrate WL. Additionally, the rotation speed represents the rotation speed of the learning target substrate WL when SPM is supplied to it.

[0220] exist Figure 17 In the substrate processing apparatus 100 shown, the ratio of hydrogen peroxide solution to sulfuric acid in the SPM can be changed by adjusting valves 137a and 137b. Therefore, in Figure 18 In this context, the concentration of the solution represents the change in concentration of the solution applied to the learning substrate WL over time, and the temperature of the solution represents the change in temperature of the solution applied to the learning substrate WL over time.

[0221] In the learning data LD1, Lcp1 represents the concentration distribution of the SPM supplied to the learning substrate WL1, Ltp1 represents the temperature distribution of the SPM supplied to the learning substrate WL1, Ls1 represents the supply amount of SPM to the learning substrate WL1, and Le1 represents the ejection pattern of the SPM to the learning substrate WL1. Additionally, Lv1 represents the rotational speed of the learning substrate WL1 when supplying SPM to it.

[0222] The same applies to the learning data LD2 to LD1000. Regarding the learning substrate WL, the results of the chemical treatment of the learning substrate WL vary significantly due to the concentration and temperature distribution of the SPM. Therefore, it is preferable that the learning data LD has a significant contribution to the variation in the chemical treatment results of the learning substrate WL.

[0223] In addition, in reference Figures 1 to 18In the above description, although the storage unit 24 of the substrate processing apparatus 100 or the storage unit 424 of the learning device 400 stores a learning completion model LM constructed through machine learning, this embodiment is not limited to this. The storage unit 24 of the substrate processing apparatus 100 or the storage unit 424 of the learning device 400 may also store a transformation table CT instead of the learning completion model LM.

[0224] Next, refer to Figure 19 The substrate processing apparatus 100 of this embodiment will be described. Figure 19 The substrate processing apparatus 100, in addition to storing a conversion table CT in the storage unit 24 to replace the learning completion model LM, has the same characteristics as the reference. Figure 4 The substrate processing apparatus 100 has the same configuration, and its repeated description is omitted to avoid being too lengthy.

[0225] like Figure 19 As shown, in the substrate processing apparatus 100, the storage unit 24 stores a conversion table CT. The conversion table CT establishes a correlation between the substrate information of the substrate Wp to be processed and the chemical processing condition information. The substrate information of the substrate Wp to be processed includes, for example, hardened layer height information and hardened layer width information. Furthermore, the conversion table CT is created based on the substrate information of the substrate WL to be learned, the chemical processing condition information, and the processing result information.

[0226] The substrate information acquisition unit 22a acquires substrate information from the storage unit 24. For example, the substrate information acquisition unit 22a acquires hardened layer height information and hardened layer width information from the storage unit 24.

[0227] The liquid treatment condition information acquisition unit 22b acquires liquid treatment condition information based on the substrate information using the conversion table CT. Typically, the liquid treatment condition information acquisition unit 22b extracts the value corresponding to the substrate information from the conversion table CT, and acquires the liquid treatment condition information based on the relationship between the substrate information and the liquid treatment condition information established in the conversion table CT. Thus, the liquid treatment condition information acquisition unit 22b uses the conversion table CT to acquire liquid treatment condition information corresponding to the substrate information.

[0228] Then, the control unit 22 controls the substrate holding unit 120 and the liquid supply unit 130 according to the liquid treatment conditions indicated by the liquid treatment condition information.

[0229] Figure 20 This is a diagram representing an example of the transformation table CT. For example... Figure 20 As shown, the conversion table CT represents the substrate information and chemical treatment conditions of the substrate Wp to be processed. In the conversion table CT, the substrate information includes information related to the hardening layer. Here, the substrate information of the substrate Wp to be processed includes the hardening layer height and width information.

[0230] The conversion table CT1 represents the chemical treatment conditions corresponding to a certain substrate information. Here, in the conversion table CT1, the substrate information includes the hardening layer height information and hardening layer width information of the substrate Wp to be processed. dt1 represents the height of the hardening layer of a certain substrate Wp to be processed. dw1 represents the width of the hardening layer of a certain substrate Wp to be processed. Rp1 represents the chemical treatment conditions to be performed on the substrate Wp to be processed. Therefore, assuming that the height of the hardening layer of the substrate Wp to be processed is dt1 and the width of the hardening layer is dw1, the substrate processing apparatus 100 performs chemical treatment using the chemical treatment conditions shown by Rp1.

[0231] The same applies to transformation tables CD2 to CD1000. Typically, for transformation tables CT1 to CD1000, at least one of the height and width of the hardened layer is different.

[0232] Furthermore, if the height and width of the hardened layer on the substrate Wp being processed are inconsistent with the values ​​shown in the conversion table CT, the chemical treatment conditions for the substrate Wp can also be determined by linear interpolation of the values ​​of the chemical treatment conditions shown in the conversion table CT. Alternatively, the chemical treatment conditions for the substrate Wp can also be determined by polynomial interpolation of the values ​​of the chemical treatment conditions shown in the conversion table.

[0233] The embodiments of the present invention have been described above with reference to the accompanying drawings. However, the present invention is not limited to the above embodiments and can be implemented in various ways without departing from its spirit. Furthermore, the present invention can also be formed by appropriately combining the various constituent elements disclosed in the above embodiments. For example, it is possible to delete some constituent elements from all the constituent elements shown in the embodiments. Furthermore, it is possible to appropriately combine constituent elements spanning different embodiments. For easier understanding, the various constituent elements are schematically shown on the main body in the drawings, and the thickness, length, number, spacing, etc. of the illustrated constituent elements may differ from the actual figures for ease of drawing. In addition, the material, shape, size, etc., of the constituent elements shown in the above embodiments are merely examples and are not particularly limited; various modifications can be made without substantially departing from the effects of the present invention.

[0234] Industrial availability

[0235] This invention is applicable to substrate processing apparatus, substrate processing method, method for generating learning data, learning method, learning device, method for generating learning completed model, and learning completed model.

[0236] [Explanation of the labels in the attached diagram]

[0237] 10: Substrate Processing System

[0238] 20: Control device

[0239] 22: Control Department

[0240] 22a: Substrate Information Acquisition Unit

[0241] 22b: Information Acquisition Department for Chemical Solution Processing Conditions

[0242] 24: Storage Department

[0243] LM: Learning Complete Model

[0244] 100: Substrate processing apparatus

[0245] 130: Medicine Supply Department

[0246] 140: Fluid Supply Department

[0247] 200: Substrate Processing Learning System

[0248] 300: Learning Data Generation Device

[0249] 400: Learning Device

Claims

1. A substrate processing apparatus comprising: a substrate holding section that holds a processing target substrate having a resist layer in which a hardened layer is formed so as to be rotatable; a chemical liquid supply section that supplies a chemical liquid to the processing target substrate; a substrate information acquisition section that acquires substrate information including hardened layer thickness information or ion implantation condition information regarding the processing target substrate, the hardened layer thickness information indicating a thickness of the hardened layer, the ion implantation condition information indicating a condition of ion implantation in which the resist layer is formed with the hardened layer; a chemical liquid processing condition information acquisition section that acquires chemical liquid processing condition information indicating a chemical liquid processing condition for the processing target substrate from a learning completion model in accordance with the substrate information; and a control section that controls the substrate holding section and the chemical liquid supply section in accordance with the chemical liquid processing condition information acquired by the chemical liquid processing condition information acquisition section so as to process the processing target substrate with the chemical liquid. the learning completion model is constructed by machine learning of learning data in which substrate information including hardened layer thickness information or ion implantation condition information regarding a learning target substrate having a resist layer in which a hardened layer is formed, chemical liquid processing condition information, and processing result information are associated with each other, the hardened layer thickness information indicating a thickness of the hardened layer, the ion implantation condition information indicating a condition of ion implantation in which the resist layer is formed with the hardened layer, the chemical liquid processing condition information indicating a condition of processing the learning target substrate with the chemical liquid, the processing result information indicating a result of processing the learning target substrate with the chemical liquid.

2. The substrate processing apparatus of claim 1, wherein, a storage section that stores the learning completion model.

3. The substrate processing apparatus according to claim 1 or 2, wherein the hardened layer thickness information includes hardened layer height information indicating a height of the hardened layer or hardened layer width information indicating a width of the hardened layer, for each of the processing target substrate and the learning target substrate.

4. The substrate processing apparatus as claimed in claim 1 or 2, wherein the chemical liquid processing condition information includes information indicating any one of a concentration of the chemical liquid, a temperature of the chemical liquid, a supply amount of the chemical liquid, a spray pattern of the chemical liquid, and a substrate rotation speed at the time of supplying the chemical liquid, for each of the processing target substrate and the learning target substrate.

5. The substrate processing apparatus of claim 4, wherein, the information indicating the concentration of the chemical liquid indicates a concentration distribution in which the concentration of the chemical liquid changes with time.

6. The substrate processing apparatus of claim 4, wherein, the information indicating the temperature of the chemical liquid indicates a temperature distribution in which the temperature of the chemical liquid changes with time.

7. A substrate processing method comprising: a step of holding a processing target substrate having a resist layer in which a hardened layer is formed so as to be rotatable; a step of acquiring substrate information including hardening layer thickness information or ion implantation condition information about the above-mentioned processing target substrate; wherein the hardened layer thickness information indicating a thickness of the hardened layer, the ion implantation condition information indicating a condition of ion implantation in which the resist layer is formed with the hardened layer; a step of acquiring chemical liquid processing condition information indicating a chemical liquid processing condition for the processing target substrate from a learning completion model in accordance with substrate information; and a step of processing the processing target substrate with the chemical liquid in accordance with the chemical liquid processing condition of the chemical liquid processing condition information. the learning completion model is constructed by machine learning of learning data in which substrate information including hardened layer thickness information or ion implantation condition information regarding a learning target substrate having a resist layer in which a hardened layer is formed, chemical liquid processing condition information, and processing result information are associated with each other, the hardened layer thickness information indicating a thickness of the hardened layer, the ion implantation condition information indicating a condition of ion implantation in which the resist layer is formed with the hardened layer, the chemical liquid processing condition information indicating a condition of processing the learning target substrate with the chemical liquid, the processing result information indicating a result of processing the learning target substrate with the chemical liquid. In the step of acquiring the chemical liquid processing condition information, the learning completion model is constructed by machine learning on learning data in which substrate information including hardening layer thickness information or ion implantation condition information, chemical liquid processing condition information, and processing result information are associated with respect to a learning target substrate having a resist layer in which a hardening layer is formed; the hardening layer thickness information indicates a thickness of the hardening layer; the ion implantation condition information indicates a condition of ion implantation in which the resist layer is formed with the hardening layer; the chemical liquid processing condition information indicates a condition of chemical liquid processing performed on the learning target substrate; and the processing result information indicates a result of the chemical liquid processing performed on the learning target substrate.

8. A learning data generation method, comprising: The time-series data output from the substrate processing apparatus that processes the learning target substrate having the resist layer in which the hardened layer is formed, is acquired, and the step of acquiring the substrate information including the hardened layer thickness information or the ion implantation condition information is included. the hardening layer thickness information indicates a thickness of the hardening layer; and the ion implantation condition information indicates a condition of ion implantation in which the resist layer is formed with the hardening layer; a step of acquiring chemical liquid processing condition information from the time series data; the chemical liquid processing condition information indicating a condition of processing performed on the learning target substrate by the substrate processing apparatus using a chemical liquid; a step of acquiring processing result information from the time series data; the processing result information indicating a result of processing performed on the learning target substrate by the substrate processing apparatus using a chemical liquid; and a step of storing, as learning data, the substrate information, the chemical liquid processing condition information, and the processing result information in association with respect to the learning target substrate in a storage unit.

9. A learning method, comprising: a step of acquiring learning data generated by the learning data generation method according to claim 8; and a step of inputting the learning data to a learning program and machine learning on the learning data.

10. A learning device, comprising: a storage unit storing learning data generated by the learning data generation method according to claim 8; and a learning unit inputting the learning data to a learning program and machine learning on the learning data.

11. A learning completion model generation method, comprising: a step of acquiring learning data generated by the learning data generation method according to claim 8; and a step of generating a learning completion model constructed by machine learning on the learning data.

Images