JP7399783B2 - Substrate processing device, substrate processing method, learning data generation method, learning method, learning device, learned model generation method, and learned model - Google Patents

Description

The present invention relates to a substrate processing apparatus, a substrate processing method, a learning data generation method, a learning method, a learning device, a learned model generation method, and a learned model.

2. Description of the Related Art A substrate processing apparatus that processes a substrate is suitably used for manufacturing semiconductor substrates and the like. If a substrate is charged when it is processed in a substrate processing apparatus, particles may be reattached to the substrate or damage may occur to the substrate due to discharge. For this reason, suppressing the charging of the substrate is being considered (see Patent Document 1).

In the substrate processing method of Patent Document 1, when supplying a liquid containing pure water with a relatively high specific resistance to a substrate, the liquid is first supplied while rotating the substrate at a relatively high rotational speed, and then the liquid is supplied while the substrate is being rotated at a relatively high rotational speed. Supply liquid while rotating the substrate at a low rotation speed. This suppresses the rate of increase in the surface potential of the substrate.

Japanese Patent Application Publication No. 2014-203906

However, in the substrate processing method of Patent Document 1, although it is possible to suppress the charging of the substrate during processing with a liquid containing pure water, it is not possible to eliminate the charge from the charged substrate. For example, a substrate that has been subjected to ion implantation or dry etching may become electrically charged, and if a processing solution is supplied to such a substrate, particles may re-adhere to the substrate or damage may occur to the substrate due to electrical discharge. . Further, when static electricity is removed from a substrate according to a recipe, the static electricity removal process may not be appropriate because the charging status of the substrate varies greatly due to minute differences between substrates.

The present invention has been made in view of the above problems, and its objects are to provide a substrate processing apparatus, a substrate processing method, a learning data generation method, a learning method, a learning device, and a learning data generation method that can appropriately remove static electricity from a substrate to be processed. The object of the present invention is to provide a model generation method and a trained model.

According to one aspect of the present invention, a substrate processing apparatus includes: a substrate holding section that rotatably holds a substrate to be processed; a charge distribution measuring section that measures a charge distribution of the substrate to be processed; and a charge discharging section for removing static electricity from the substrate to be processed. a board information acquisition unit that acquires board information including charge distribution information indicating the charge distribution of the processing target board; and a board information acquisition unit that obtains charge removal processing conditions for the processing target board from a learned model based on the board information. a static elimination processing condition information acquisition section that acquires static elimination processing condition information indicating static elimination processing condition information; and the substrate holding section so as to eliminate static electricity from the substrate to be processed based on the static elimination processing condition information acquired by the static elimination processing condition information acquisition section. and a control section that controls the static eliminator. The learned model includes board information including charge distribution information indicating the charge distribution of the learning target board, static elimination processing condition information indicating the conditions under which the learning target board was subjected to static elimination processing, and the results of static elimination processing of the learning target board. It is constructed by performing machine learning on learning data associated with the processing result information shown in FIG.

In one embodiment, the substrate processing apparatus further includes a storage unit that stores the learned model.

In one embodiment, for each of the processing target substrate and the learning target substrate, the substrate information further includes information indicating a film type or film thickness of the substrate.

In one embodiment, a static eliminating liquid is supplied to each of the processing target substrate and the learning target substrate.

In one embodiment, for each of the processing target board and the learning target board, the static elimination processing condition information includes a concentration of the static eliminating liquid, a temperature of the static eliminating liquid, a supply amount of the static eliminating liquid, and a discharge pattern of the static eliminating liquid. , and information indicating either the rotation speed of the substrate at the time of supplying the static eliminator.

In one embodiment, each of the processing target substrate and the learning target substrate is irradiated with ultraviolet rays.

According to another aspect of the present invention, a substrate processing method includes the steps of: rotatably holding a substrate to be processed; obtaining substrate information including charge distribution information indicating charge distribution of the substrate to be processed; a step of acquiring static electricity removal processing condition information indicating a static electricity removal processing condition of the processing target substrate from the learned model based on board information; and a step of removing static electricity from the processing target board according to the static electricity removal processing conditions of the static electricity removal processing condition information. include. In the step of acquiring the static elimination processing condition information, the learned model includes board information including charge distribution information indicating the charge distribution of the learning target board, and static elimination processing condition information indicating the conditions under which the learning target board was subjected to static elimination processing. , is constructed by performing machine learning on learning data associated with processing result information indicating the result of static elimination processing on the learning target board.

According to still another aspect of the present invention, the method for generating learning data includes generating charge distribution information indicating the charge distribution of the learning target substrate from time-series data output from a substrate processing apparatus that processes the learning target board. acquiring, from the time-series data, static elimination processing condition information indicating the conditions under which the learning target substrate was static-eliminated in the substrate processing apparatus; acquiring processing result information indicating the result of static electricity removal on the learning target board, and storing the board information, the static elimination processing condition information, and the processing result information about the learning target board in a storage unit as learning data in association with each other; and the step of doing so.

According to yet another aspect of the present invention, a learning method includes the steps of acquiring learning data generated according to the learning data generation method described above, and inputting the learning data into a learning program. and performing machine learning on the learning data.

According to yet another aspect of the present invention, the learning device includes a storage unit that stores learning data generated according to the learning data generation method described above, and a storage unit that stores the learning data generated according to the learning data generation method described above. and a learning section that performs machine learning on the learning data.

According to still another aspect of the present invention, a method for generating a trained model includes the steps of: acquiring training data generated according to the method for generating training data described above; and applying machine learning to the training data. and generating a trained model constructed by performing the following steps.

According to yet another aspect of the present invention, the learned model is constructed by subjecting learning data generated according to the above-described learning data generation method to machine learning.

According to still another aspect of the present invention, a substrate processing apparatus includes: a substrate holding section that rotatably holds a substrate; a charge distribution measuring section that measures a charge distribution of the substrate; and a static eliminator that removes static electricity from the substrate. , a storage unit that stores a conversion table in which board information including charge distribution information is associated with charge removal processing condition information indicating conditions for charge removal processing; and a storage unit that stores board information including charge distribution information indicating the charge distribution of the substrate. a board information acquisition unit that acquires, a static elimination process condition information acquisition unit that uses the conversion table based on the board information to acquire static electricity removal process condition information indicating a static electricity removal process condition of the board; and a static electricity removal process condition information acquisition unit that acquires the static electricity removal process condition information. and a control section that controls the substrate holding section and the static elimination section so as to eliminate static electricity from the substrate based on the static elimination processing condition information acquired in the static elimination processing section.

According to the present invention, it is possible to appropriately eliminate static electricity from a substrate to be processed.

FIG. 1 is a schematic diagram of a substrate processing learning system including a substrate processing apparatus according to the present embodiment. 1 is a schematic diagram of a substrate processing system including a substrate processing apparatus of this embodiment. FIG. 1 is a schematic diagram of a substrate processing apparatus of this embodiment. FIG. 1 is a block diagram of a substrate processing system including a substrate processing apparatus of this embodiment. (a) is a flow diagram of the substrate processing method of this embodiment, and (b) is a flow diagram of static elimination processing in the substrate processing method of this embodiment. FIG. 3 is a diagram showing the charge distribution of a substrate measured in the substrate processing apparatus of the present embodiment. (a) to (d) are schematic diagrams showing static elimination processing in the substrate processing apparatus of this embodiment. FIG. 1 is a block diagram of a substrate processing apparatus and a learning data generation apparatus according to the present embodiment. FIG. 2 is a flow diagram showing a learning data generation method according to the present embodiment. FIG. 1 is a block diagram of a learning data generation device and a learning device according to the present embodiment. FIG. 2 is a flow diagram showing a learning method and a learned model generation method according to the present embodiment. FIG. 2 is a diagram showing learning data input to the learning device of this embodiment. FIG. 2 is a diagram showing learning data input to the learning device of this embodiment. (a) is a diagram showing the charge distribution of the substrate to be processed measured in the substrate processing apparatus of the present embodiment, and (b) is a diagram showing the static elimination processing conditions acquired based on the charge distribution of the substrate to be processed. It is a diagram. FIG. 2 is a diagram showing learning data input to the learning device of this embodiment. FIG. 1 is a schematic diagram of a substrate processing apparatus of this embodiment. FIG. 2 is a diagram showing learning data input to the learning device of this embodiment. FIG. 1 is a schematic diagram of a substrate processing apparatus of this embodiment. FIG. 2 is a diagram showing learning data input to the learning device of this embodiment. FIG. 1 is a schematic diagram of a substrate processing apparatus of this embodiment. FIG. 2 is a diagram showing learning data input to the learning device of this embodiment. (a) is a schematic diagram of a substrate holder in the substrate processing apparatus of the present embodiment, and (b) to (d) are schematic top views of the substrate holder. FIG. 2 is a diagram showing learning data input to the learning device of this embodiment. FIG. 1 is a block diagram of a substrate processing apparatus according to the present embodiment. FIG. 3 is a diagram showing a conversion table in the substrate processing apparatus of the present embodiment.

Hereinafter, embodiments of a substrate processing apparatus, a substrate processing method, a learning data generation method, a learning method, a learning device, a learned model generation method, and a learned model according to the present invention will be described with reference to the drawings. . In addition, in the drawings, the same or corresponding parts are given the same reference numerals and the description will not be repeated. Furthermore, in this specification, in order to facilitate understanding of the invention, an X direction, a Y direction, and a Z direction that are perpendicular to each other may be described. Typically, the X and Y directions are parallel to the horizontal direction, and the Z direction is parallel to the vertical direction.

First, with reference to FIG. 1, a substrate processing learning system 200 including a substrate processing apparatus 100 of this embodiment will be described. First, a substrate processing learning system 200 will be explained with reference to FIG.

FIG. 1 is a schematic diagram of a substrate processing learning system 200. As shown in FIG. 1, 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.

The substrate processing apparatus 100 processes a substrate to be processed. Here, the substrate processing apparatus 100 removes static electricity from the substrate to be processed. Note that the substrate processing apparatus 100 may perform processing other than static elimination on the substrate to be processed. The substrate processing apparatus 100 is a single-wafer type that processes target substrates one by one. Typically, the substrate to be processed is approximately disk-shaped.

The substrate processing apparatus 100L processes a learning target substrate. Here, the substrate processing apparatus 100L removes static electricity from the learning target substrate. Note that the substrate processing apparatus 100L may perform processing other than static elimination on the learning target substrate. The configuration of the learning target board is the same as the configuration of the processing target board. The substrate processing apparatus 100L is a single-wafer type that processes target substrates one by one. Typically, the substrate to be processed is approximately disk-shaped. The configuration of the substrate processing apparatus 100L is the same as that of the substrate processing apparatus 100. The substrate processing apparatus 100L may be the same as the substrate processing apparatus 100. For example, the same substrate processing apparatus may have processed a learning target substrate in the past and may subsequently process a processing target substrate. Alternatively, the substrate processing apparatus 100L may be another product having the same configuration as the substrate processing apparatus 100.

In the following description of this specification, a learning target substrate may be described as a "learning target substrate WL", and a processing target substrate may be described as a "processing target substrate Wp". Further, when there is no need to distinguish between the learning target substrate WL and the processing target substrate Wp, the learning target substrate WL and the processing target substrate Wp may be referred to as "substrate W".

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 disk, a substrate for a magnetic disk, a substrate for a magneto-optical disk, or a photomask. substrate, ceramic substrate, or solar cell substrate.

The substrate processing apparatus 100L outputs time series data TDL. The time-series data TDL is data indicating temporal changes in physical quantities in the substrate processing apparatus 100L. The time series data TDL indicates time changes in physical quantities (values) that change over time over a predetermined period. For example, the time-series data TDL is data indicating a change over time in a physical quantity regarding a process performed on a learning target substrate by the substrate processing apparatus 100L. Alternatively, the time series data TDL is data indicating changes over time in physical quantities regarding the characteristics of the learning target substrate processed by the substrate processing apparatus 100L.

Note that the value shown in the time series data TDL may be a value directly measured with a measuring device. Alternatively, the value shown in the time series data TDL may be a value obtained by processing a value directly measured by a measuring device. Alternatively, the value shown in the time series data TDL may be calculated from values measured by a plurality of measuring devices.

The learning data generation device 300 generates learning data LD based on the time series data TDL or at least a part of the time series data TDL. The learning data generation device 300 outputs learning data LD. The learning data LD includes board information of the learning target board WL, static elimination processing condition information indicating the processing conditions of the static elimination process performed on the learning target board WL, and static elimination processing performed on the learning target board WL. and processing result information indicating the results. Further, the board information of the learning target board WL includes charge distribution information indicating the charge distribution of the learning target board WL measured before the learning target board WL is subjected to static elimination processing.

The learning device 400 generates a learned model LM by performing machine learning on the learning data LD. Learning device 400 outputs learned model LM.

The substrate processing apparatus 100 outputs time series data TD. The time series data TD is data indicating changes in physical quantities in the substrate processing apparatus 100 over time. The time-series data TD indicates time-based changes in physical quantities (values) that change over time over a predetermined period. For example, the time-series data TD is data indicating changes over time in physical quantities regarding processing performed by the substrate processing apparatus 100 on the substrate to be processed. Alternatively, the time series data TD is data indicating changes over time in physical quantities regarding the characteristics of the substrate to be processed processed by the substrate processing apparatus 100.

Note that the value shown in the time series data TD may be a value directly measured with a measuring device. Alternatively, the value shown in the time series data TD may be a value obtained by processing a value directly measured by a measuring device. Alternatively, the value shown in the time series data TD may be calculated from values measured by a plurality of measuring devices.

The object used by the substrate processing apparatus 100 corresponds to the object used by the substrate processing apparatus 100L. Therefore, the structure of the object used by the substrate processing apparatus 100 is the same as the structure of the object used by the substrate processing apparatus 100L. Furthermore, in the time series data TD, the physical quantity of the object used by the substrate processing apparatus 100 corresponds to the physical quantity of the object used by the substrate processing apparatus 100L. Therefore, the physical quantity of the object used by the substrate processing apparatus 100L is the same as the physical quantity of the object used by the substrate processing apparatus 100.

Substrate information Cp of the processing target substrate Wp is generated from the time series data TD. The substrate information Cp of the processing target substrate Wp corresponds to the substrate information of the learning target substrate WL. The substrate information Cp of the processing target substrate Wp includes charge distribution information indicating the charge distribution of the processing target substrate Wp measured before the processing target substrate Wp is subjected to the static elimination process.

Based on the substrate information Cp of the target substrate Wp, the learned model LM outputs static elimination processing condition information Rp indicating the static elimination processing conditions suitable for the target substrate Wp in the substrate processing apparatus 100.

As described above with reference to FIG. 1, according to this embodiment, the learning device 400 performs machine learning. Therefore, a trained model LM with high accuracy can be generated from the time series data TDL which is extremely complex and has a huge amount of analysis targets. Further, board information Cp including charge distribution information from the time series data TD is input to the learned model LM, and the learned model LM is caused to output static elimination processing condition information Rp indicating the static elimination processing conditions. Therefore, the static electricity removal process can be performed depending on the substrate Wp to be processed.

Next, with reference to FIG. 2, a substrate processing system 10 including the substrate processing apparatus 100 of this embodiment will be described. FIG. 2 is a schematic plan view of the substrate processing system 10.

The substrate processing system 10 processes a substrate W. The substrate processing system 10 includes a plurality of substrate processing apparatuses 100. The substrate processing apparatus 100 processes the substrate W to perform at least one of etching, surface treatment, imparting properties, forming a treated film, removing at least a portion of the film, and cleaning.

As shown in FIG. 2, the substrate processing system 10 includes, in addition to a plurality of substrate processing apparatuses 100, a fluid cabinet 32, a fluid box 34, a plurality of load ports LP, an indexer robot IR, and a center robot CR. , and a control device 20. The control device 20 controls the load port LP, indexer robot IR, and center robot CR.

Each of the load ports LP accommodates a plurality of stacked substrates W. The indexer robot IR transports the substrate W between the load port LP and the center robot CR. Note that an installation stand (path) on which the substrate W is temporarily placed is provided between the indexer robot IR and the center robot CR, and a path is provided between the indexer robot IR and the center robot CR via the installation stand. It is also possible to adopt an apparatus configuration in which the substrate W is transferred indirectly. The center robot CR transports the substrate W between the indexer robot IR and the substrate processing apparatus 100. Each of the substrate processing apparatuses 100 processes the substrate W by discharging a liquid onto the substrate W. The liquid includes a static eliminating liquid, a chemical liquid, and/or a rinsing liquid. Alternatively, the liquid may include other processing liquids. Fluid cabinet 32 contains liquid. Note that the fluid cabinet 32 may contain gas.

Specifically, the plurality of substrate processing apparatuses 100 form a plurality of towers TW (four towers TW in FIG. 2) arranged so as to surround the center robot CR in plan view. Each tower TW includes a plurality of substrate processing apparatuses 100 (three substrate processing apparatuses 100 in FIG. 1) stacked one above the other. Each fluid box 34 corresponds to a plurality of towers TW. The liquid in the fluid cabinet 32 is supplied via one of the fluid boxes 34 to all the substrate processing apparatuses 100 included in the tower TW corresponding to the fluid box 34. Further, the gas in the fluid cabinet 32 is supplied via any one of the fluid boxes 34 to all the substrate processing apparatuses 100 included in the tower TW corresponding to the fluid box 34.

The control device 20 controls various operations of the substrate processing system 10. Control device 20 includes a control section 22 and a storage section 24. The control unit 22 has a processor. The control unit 22 includes, for example, a central processing unit (CPU). Alternatively, the control unit 22 may include a general-purpose computing machine.

The storage unit 24 stores data and computer programs. The data includes recipe data. The recipe data includes information indicating multiple recipes. Each of the plurality of recipes defines processing contents and processing procedures for the substrate W.

The 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. The storage unit 24 may include removable media. The control unit 22 executes a computer program stored in the storage unit 24 to perform a substrate processing operation.

Although FIG. 2 shows one control device 20 for the substrate processing system 10, the control device 20 may be provided for each substrate processing apparatus 100. However, in that case, the substrate processing system 10 preferably includes a separate control device that controls the plurality of substrate processing apparatuses 100 and devices other than the substrate processing apparatus 100.

Note that the storage unit 24 stores a computer program with a predetermined procedure. The substrate processing apparatus 100 operates according to procedures defined in a computer program.

Next, with reference to FIG. 3, the substrate processing apparatus 100 of this embodiment will be described. FIG. 3 is a schematic diagram of the substrate processing apparatus 100 of this embodiment. Although the configuration of the substrate processing apparatus 100 will be described here, the substrate processing apparatus 100L also has the same configuration as the substrate processing apparatus 100.

The substrate processing apparatus 100 processes a substrate W. The substrate processing apparatus 100 includes a chamber 110 , a substrate holding section 120 , a charge distribution measuring section 130 , a static eliminating section 140 , a chemical solution supply section 150 , and a rinsing liquid supply section 160 . Chamber 110 accommodates substrate W. The substrate holding section 120 holds the substrate W. The substrate holding unit 120 rotatably holds the substrate W. The charge distribution measurement unit 130 measures the distribution of the amount of charge charged on the substrate W or the potential distribution of the upper surface Wa of the substrate W. The static eliminator 140 removes static from the substrate W. In FIG. 3, the static eliminator 140 supplies a static eliminator liquid to the substrate W to eliminate static electricity from the substrate W.

The chamber 110 is approximately box-shaped with an internal space. Chamber 110 accommodates substrate W. Here, the substrate processing apparatus 100 is a single-wafer type that processes substrates W one by one, and the chamber 110 accommodates one substrate W at a time. The substrate W is accommodated within the chamber 110 and processed within the chamber 110. The chamber 110 accommodates at least a portion of each of the substrate holding section 120 , the charge distribution measuring section 130 , the static eliminating section 140 , the chemical solution supply section 150 , and the rinsing liquid supply section 160 .

The substrate holding section 120 holds the substrate W. The substrate holding unit 120 holds the substrate W horizontally so 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. Further, the substrate holding unit 120 rotates the substrate W while holding the substrate W.

For example, the substrate holder 120 may be a clamping type that clamps an end of the substrate W. Alternatively, the substrate holding section 120 may have any mechanism for holding the substrate W from the back surface Wb. For example, the substrate holding section 120 may be of a vacuum type. In this case, the substrate holding unit 120 holds the substrate W horizontally by adsorbing the center portion of the back surface Wb of the substrate W, which is a non-device forming surface, to the upper surface. Alternatively, the substrate holder 120 may combine a clamping type in which a plurality of chuck pins are brought into contact with the peripheral end surface of the substrate W and a vacuum type.

For example, the substrate holder 120 includes a spin base 121, a chuck member 122, a shaft 123, and an electric motor 124. The chuck member 122 is provided on the spin base 121. The chuck member 122 chucks the substrate W. Typically, the spin base 121 is provided with a plurality of chuck members 122.

Shaft 123 is a hollow shaft. The shaft 123 extends vertically along the rotation axis Ax. A spin base 121 is coupled to the upper end of the shaft 123. The back surface of the substrate W contacts the spin base 121, and the substrate W is placed above the spin base 121.

The spin base 121 has a disk shape and supports the substrate W horizontally. The shaft 123 extends downward from the center of the spin base 121. Electric motor 124 provides rotational force to shaft 123. The electric motor 124 rotates the substrate W and the spin base 121 around the rotation axis Ax by rotating the shaft 123 in the rotational direction. Here, the direction of rotation is counterclockwise.

The charge distribution measuring section 130 measures the charge distribution of the substrate W. The charge distribution measurement unit 130 measures the amount of charge on the surface of the substrate W. For example, the charge distribution measuring section 130 includes a non-contact surface electrometer. The charge distribution measuring section 130 may be a vibrating capacitance electrometer.

The static eliminator 140 removes static from the substrate. Here, the static eliminator 140 supplies a static eliminator to the substrate W. By the static eliminator 140 supplying the static eliminator liquid to the substrate W, it is possible to suppress the occurrence of discharge in the substrate W even if the substrate W is in a charged state.

The static eliminating liquid is preferably a liquid with relatively low specific resistance. In one example, a liquid having a lower specific resistance than the chemical liquid is used as the static eliminating liquid.

For example, carbonated water is preferably used as the static eliminating liquid. Alternatively, water may be used as the static eliminating liquid. Alternatively, SC1 (ammonia hydrogen peroxide solution mixture), SC2 (hydrochloric acid hydrogen peroxide solution mixture), etc. may be used as the static eliminating liquid. By using a liquid with relatively low specific resistance as the static eliminating liquid, the static charge on the substrate W can be smoothly removed while suppressing the discharge that occurs on the substrate W.

The static eliminator 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 discharges the static eliminating liquid toward the upper surface Wa of the substrate W. Piping 144 is coupled to nozzle 142. Nozzle 142 is provided at the tip of piping 144 . A static eliminating liquid is supplied to the pipe 144 from a supply source. Valve 146 is provided in piping 144 . Valve 146 opens and closes the flow path within piping 144 .

The static eliminator 140 further includes a nozzle moving section 148. The nozzle moving unit 148 moves the nozzle 142 between the ejection position and the retreat position. When the nozzle 142 is in the discharge position, the nozzle 142 is located above the substrate W. When the nozzle 142 is in the discharge position, the nozzle 142 discharges the static eliminating liquid toward the upper surface Wa of the substrate W. When the nozzle 142 is in the retracted position, the nozzle 142 is located radially outside of the substrate W.

Nozzle moving section 148 includes an arm 148a, a rotation shaft 148b, and a moving mechanism 148c. Arm 148a extends substantially horizontally. A nozzle 142 is attached to the tip of the arm 148a. Arm 148a is coupled to pivot shaft 148b. The rotation shaft 148b extends substantially vertically. The moving mechanism 148c rotates the rotation shaft 148b around a rotation axis along a substantially vertical direction, and rotates the arm 148a along a substantially horizontal plane. As a result, the nozzle 142 moves along a substantially horizontal plane. For example, the moving mechanism 148c includes an arm swing motor that rotates the rotation shaft 148b around the rotation axis. The arm swing motor is, for example, a servo motor. Furthermore, the moving mechanism 148c moves the rotation shaft 148b up and down along the substantially vertical direction, and moves the arm 148a up and down. As a result, the nozzle 142 moves along the substantially vertical direction. For example, the moving mechanism 148c includes a ball screw mechanism and an arm lift motor that provides driving force to the ball screw mechanism. The arm lifting motor is, for example, a servo motor.

The chemical liquid supply unit 150 supplies a chemical liquid to the substrate W. Thereby, the substrate W is treated with the chemical solution.

A liquid with relatively high specific resistance may be used as the chemical liquid. For example, the chemical solution includes hydrofluoric acid (hydrogen fluoride water: HF). Alternatively, the chemical solution may include sulfuric acid, acetic acid, nitric acid, hydrochloric acid, citric acid, buffered hydrofluoric acid (BHF), diluted hydrofluoric acid (DHF), ammonia water, diluted ammonia water, hydrogen peroxide water, organic alkali (for example, TMAH: It may be a liquid containing at least one of tetramethylammonium hydroxide, a surfactant, and a corrosion inhibitor. Further, the chemical solution may be a mixture of the above-mentioned solutions. For example, examples of chemical solutions that are a mixture of these include SPM (sulfuric acid/hydrogen peroxide/water mixture), SC1 (ammonia/hydrogen peroxide/water mixture), SC2 (hydrochloric acid/hydrogen peroxide/water mixture), and the like.

Chemical liquid supply section 150 includes a nozzle 152, piping 154, and valve 156. The nozzle 152 faces the upper surface Wa of the substrate W and discharges a chemical liquid toward the upper surface Wa of the substrate W. Piping 154 is coupled to nozzle 152. Nozzle 152 is provided at the tip of piping 154. A chemical solution is supplied to the pipe 154 from a supply source. A valve 156 is provided in the piping 154. Valve 156 opens and closes the flow path within piping 154.

The rinsing liquid supply unit 160 supplies a rinsing liquid to the substrate W. The rinsing liquid supplied from the rinsing liquid supply unit 160 includes deionized water (DIW), carbonated water, electrolyzed ionized water, ozone water, ammonia water, hydrochloric acid water with a diluted concentration (for example, about 10 ppm to 100 ppm), Alternatively, it may contain either reduced water (hydrogen water).

Rinse liquid supply section 160 includes a nozzle 162, piping 164, and valve 166. The nozzle 162 faces the upper surface Wa of the substrate W and discharges a rinse liquid toward the upper surface Wa of the substrate W. Piping 164 is coupled to nozzle 162. Nozzle 162 is provided at the tip of piping 164. A rinsing liquid is supplied to the pipe 164 from a supply source. A valve 166 is provided in the piping 164. Valve 166 opens and closes the flow path within piping 164.

The substrate processing apparatus 100 further includes a cup 180. The cup 180 collects liquid scattered from the substrate W. The cup 180 can rise vertically upward to the side of the substrate W. Further, the cup 180 may descend vertically from the side of the substrate W.

The control device 20 controls various operations of the substrate processing apparatus 100. The control section 22 controls the substrate holding section 120 , the charge distribution measuring section 130 , the static eliminating section 140 , the chemical solution supply section 150 , the rinsing liquid supply section 160 , and/or the cup 180 . In one example, the controller 22 controls the electric motor 124, the valves 146, 156, 166, the moving mechanism 148c, and/or the cup 180.

In the substrate processing apparatus 100 of this embodiment, the substrate W can be subjected to static elimination processing, chemical solution processing, and rinsing processing.

Next, the substrate processing apparatus 100 of this embodiment will be described with reference to FIGS. 1 to 4. FIG. 4 is a block diagram of the substrate processing system 10 including the substrate processing apparatus 100.

As shown in FIG. 4, the control device 20 controls various operations of the substrate processing system 10. The control device 20 controls the indexer robot IR, the center robot CR, the substrate holding section 120, the charge distribution measuring section 130, the static eliminator 140, the chemical solution supply section 150, the rinsing liquid supply section 160, and the cup 180. Specifically, the control device 20 sends control signals to the indexer robot IR, the center robot CR, the substrate holding section 120, the charge distribution measuring section 130, the static eliminating section 140, the chemical solution supply section 150, the rinsing solution supply section 160, and the cup 180. By transmitting , the indexer robot IR, center robot CR, substrate holding section 120, charge distribution measuring section 130, static eliminator 140, chemical solution supply section 150, rinsing liquid supply section 160, and cup 180 are controlled.

Specifically, the control unit 22 controls the indexer robot IR to deliver the substrate W by the indexer robot IR.

The control unit 22 controls the center robot CR and delivers the substrate W by the center robot CR. For example, the central robot CR receives an unprocessed substrate W and carries the substrate W into one of the plurality of chambers 110. Moreover, the center robot CR receives the processed substrate W from the chamber 110 and carries out the substrate W.

The control unit 22 controls the substrate holding unit 120 to start the rotation of the substrate W, change the rotation speed, and stop the rotation of the substrate W. For example, the control unit 22 can control the substrate holding unit 120 to change the rotation speed of the substrate holding unit 120. Specifically, the control section 22 can change the rotation speed of the substrate W by changing the rotation speed of the electric motor 124 of the substrate holding section 120.

The control unit 22 controls the charge distribution measuring unit 130 to measure the charge distribution of the substrate W.

The control unit 22 controls the valve 146 of the static elimination unit 140, the valve 156 of the chemical solution supply unit 150, and the valve 166 of the rinsing liquid supply unit 160, and changes the states of the valves 146, 156, and 166 between an open state and a closed state. Can be switched. Specifically, the control unit 22 controls the valves 146, 156, 166 to open the valves 146, 156, 166, thereby opening the pipes 144, 154, 164 toward the nozzles 142, 152, 162. Static eliminator, chemical solution, and rinsing solution flowing inside can be passed through. The control unit 22 also controls the valve 146 of the static eliminator 140, the valve 156 of the chemical solution supply unit 150, and the valve 166 of the rinsing liquid supply unit 160 to close the valves 146, 156, and 166, thereby closing the nozzle. It is possible to stop the static eliminating liquid, chemical liquid, and rinse liquid flowing in the pipes 144, 154, and 164 toward the pipes 142, 152, and 162.

Further, the control unit 22 controls the moving mechanism 148c to move the arm 148a in the horizontal direction and/or the vertical direction. Thereby, the control unit 22 can move the nozzle 142 attached to the tip of the arm 148a on the upper surface Wa of the substrate W. Further, the control unit 22 can move the nozzle 142 attached to the tip of the arm 148a between the discharge position and the retracted position. The substrate processing apparatus 100 of this embodiment is suitably used for forming semiconductor devices.

In the substrate processing apparatus 100 of this embodiment, the storage unit 24 stores the learned model LM and the control program PG. The substrate processing apparatus 100 operates according to a procedure defined in a control program PG.

Further, the control unit 22 includes a board information acquisition unit 22a and a static elimination processing condition information acquisition unit 22b. The substrate information acquisition section 22a acquires substrate information of the processing target substrate Wp from the charge distribution measurement section 130. The substrate information includes charge distribution information. Note that the substrate information acquisition section 22a may acquire information other than the charge distribution information from the storage section 24 as the substrate information.

The learned model LM generates static elimination processing condition information based on the board information. Typically, when board information is input to the learned model LM, static elimination processing condition information corresponding to the board information is output. In one example, when charge distribution information is input to the learned model LM, charge removal processing condition information corresponding to the charge distribution information is output.

The static elimination processing condition information acquisition unit 22b acquires static elimination processing condition information from the learned model LM. The static elimination processing condition information acquisition unit 22b acquires static elimination processing condition information corresponding to the substrate information of the processing target substrate Wp from the learned model LM.

The control unit 22 controls the substrate holding unit 120 and the static elimination unit 140 according to the static elimination processing conditions indicated in the static elimination processing condition information.

Preferably, the substrate processing system 10 further includes a display section 42, an input section 44, and a communication section 46.

The display unit 42 displays images. The display section 42 is, for example, a liquid crystal display or an organic electroluminescent display.

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

The communication unit 46 is connected to a network and communicates with external devices. In this embodiment, the network includes, for example, the Internet, a LAN (Local Area Network), a public telephone network, and a short-range wireless network. The communication unit 46 is a communication device, for example, a network interface controller.

Furthermore, it is preferable that the substrate processing system 10 further includes a sensor 50. Typically, multiple sensors 50 sense the condition of each portion of substrate processing system 10. For example, at least a portion of the sensor 50 detects the state of each part of the substrate processing apparatus 100.

The storage unit 24 stores output results from the sensor 50 and control parameters based on the control program as time-series data. Typically, time series data is stored separately for each substrate W.

The sensor 50 detects a physical quantity of an object used by the substrate processing apparatus 100 during a period from the start of processing of the substrate W to the end of processing for each processing of one substrate W, and sends a detection signal indicating the physical quantity to the control unit. Output to 22. Then, the control unit 22 associates the physical quantity indicated by the detection signal output from the sensor 50 during the period from the start of processing of the substrate W to the end of processing with time for each processing of one substrate W, and generates time series data. It is stored in the storage unit 24 as a TD.

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 causes the storage unit 24 to store the time series data TD in association with lot identification information, substrate identification information, processing order information, and lot interval information. The lot identification information is information for identifying a lot (for example, a lot number). A lot indicates a processing unit of substrates W. One lot is composed of a predetermined number of substrates W. The board identification information is information for identifying the board W. The processing order information is information indicating the order of processing on a predetermined number of substrates W constituting one lot. The lot interval information is information indicating the time interval from the end of processing for a lot to the start of processing for the next lot.

Next, a substrate processing method using the substrate processing apparatus 100 of this embodiment will be described with reference to FIGS. 1 to 5. FIG. 5(a) is a flowchart of a substrate processing method in the substrate processing apparatus 100 of this embodiment, and FIG. 5(b) is a flowchart of static elimination processing in the substrate processing method of this embodiment.

As shown in FIG. 5A, in step S10, a substrate to be processed Wp is carried into the substrate processing apparatus 100. The carried-in processing target substrate Wp is mounted on the substrate holding section 120. Typically, a substrate to be processed Wp is carried into the substrate processing apparatus 100 by a central robot CR.

In step S20, the processing target substrate Wp is neutralized. The static eliminator 140 removes static from the processing target substrate Wp. Specifically, the static eliminating liquid is discharged from the nozzle 142 of the static eliminating unit 140 onto the upper surface Wa of the processing target substrate Wp. The static eliminating liquid covers the upper surface Wa of the processing target substrate Wp. Thereby, the substrate W is neutralized by the static eliminating liquid. Note that the substrate W is rotated by the substrate holder 120 when static electricity is removed from the substrate Wp to be processed. The rotation of the processing target substrate Wp may continue until immediately before the processing target substrate Wp is transported out.

In step S30, the processing target substrate Wp is processed with a chemical solution. The chemical supply unit 150 supplies a chemical to the processing target substrate Wp. The chemical liquid is discharged from the nozzle 152 of the chemical liquid supply unit 150 onto the upper surface Wa of the processing target substrate Wp. The chemical solution covers the upper surface Wa of the substrate Wp to be processed. Thereby, the processing target substrate Wp is processed with the chemical solution.

In step S40, the processing target substrate Wp is rinsed with a rinsing liquid. The rinsing liquid supply unit 160 supplies a rinsing liquid to the processing target substrate Wp. The rinsing liquid is discharged from the nozzle 162 of the rinsing liquid supply unit 160 onto the upper surface Wa of the processing target substrate Wp. The rinsing liquid covers the upper surface Wa of the processing target substrate Wp. Thereby, the processing target substrate Wp is processed with the rinsing liquid.

In step S50, the processing target substrate Wp is detached from the substrate holding unit 120 and the processing target substrate Wp is carried out. Typically, the processing target substrate Wp is carried out from the substrate processing apparatus 100 by the central robot CR. In the manner described above, the target substrate Wp can be subjected to static elimination processing and chemical solution processing. Note that the chemical solution used for chemical treatment may have a relatively high specific resistance. In this case, it is preferable to perform static elimination treatment before chemical treatment. Thereby, even if the substrate to be processed Wp is charged and carried in, it is possible to suppress the generation of discharge in the substrate to be processed Wp during the chemical treatment. However, the substrate W may be charged by the chemical treatment. In this case, static elimination processing may be performed after chemical treatment.

In the substrate processing apparatus 100 of this embodiment, as shown in FIG. 5(b), the target substrate Wp is subjected to static elimination processing.

In step S22, substrate information of the processing target substrate Wp is acquired. The substrate information acquisition unit 22a acquires substrate information of the processing target substrate Wp. The substrate information includes charge distribution information indicating the charge distribution of the processing target substrate Wp. For example, the control unit 22 controls the charge distribution measurement unit 130, so that the charge distribution measurement unit 130 measures the charge distribution of the processing target substrate Wp. Note that the control unit 22 may acquire substrate information other than charge distribution information. For example, the control unit 22 may acquire substrate information of the processing target substrate Wp from the storage unit 24.

In step S24, the substrate information of the processing target substrate Wp is input into the learned model LM. Although the details will be described later, the trained model LM includes board information of the learning target board WL, static elimination processing condition information indicating the processing conditions of the static elimination process performed on the learning target board WL, and It is constructed from learning data including processing result information indicating the result of the performed static elimination processing. The learned model LM outputs static elimination processing condition information Rp corresponding to the substrate information of the processing target substrate Wp.

In step S26, static elimination processing condition information is acquired from the learned model LM. The static elimination processing condition information acquisition unit 22b acquires static elimination processing condition information corresponding to board information from the learned model LM.

In step S28, the substrate holder 120 and the static eliminator 140 perform the static neutralization process on the processing target substrate Wp according to the static neutralization process condition information. In the substrate processing apparatus 100 shown in FIG. 3, the static eliminator 140 supplies the static eliminator liquid to the processing target substrate Wp according to the static elimination processing condition information. In the manner described above, the target substrate Wp can be subjected to static elimination processing.

According to the present embodiment, static elimination processing condition information corresponding to substrate information including charge distribution information of the processing target substrate Wp is acquired from the learned model LM constructed by machine learning, and The static elimination process is executed according to the static elimination process conditions. Although the charging distribution of the processing target substrate Wp varies greatly depending on the characteristics of the processing target substrate Wp and slight differences in the processing before static elimination, according to the present embodiment, depending on the charging state of the processing target substrate Wp, Static elimination processing can be performed appropriately.

Next, the charged state of the substrate W in the substrate processing method of this embodiment will be explained with reference to FIG. FIG. 6 is a schematic diagram showing the charge distribution of the substrate W. FIG. 6A shows the charge distribution of the substrate measured by the charge distribution measuring section 130. Here, the potential of the substrate W is divided into eight stages and shown for each region.

In the substrate W, a region R1 indicates a region of the substrate W with the lowest potential, and a region R2 indicates a region of the substrate W with the next lowest potential after the region R1. Regions R3 to R7 indicate regions of lower potential in the substrate W in order. Region R8 indicates a region of the substrate W with the highest potential.

In FIG. 6, the lowest potential region of the substrate W is located at the upper end of the substrate W and to the left of the center of the substrate W. In FIG. 6, the potential of the substrate W tends to be low on the upper side and high on the lower side. However, the charge distribution of the substrate W shown in FIG. 6 is merely an example, and the charge distribution of the substrate W varies greatly depending on the type and thickness of the substrate W and slight differences in the previous processing.

Next, the static elimination process in the substrate processing method of this embodiment will be explained with reference to FIGS. 1 to 7. FIGS. 7(a) to 7(d) are schematic diagrams of the static elimination process in the substrate processing method of this embodiment.

As shown in FIG. 7A, the potential of the processing target substrate Wp differs depending on the position. FIG. 7A shows the distribution of charges Q for each position on the processing target substrate Wp. In FIG. 7(a), the larger the charge Q is, the larger the amount of charge is at the corresponding position on the substrate Wp to be processed, and the smaller the charge Q is, the smaller the amount of charge is at the corresponding position on the substrate Wp to be processed. ing. In FIG. 7A, there are regions of relatively high potential at the left end and center of the substrate Wp to be processed, while the potential is relatively low at the right end of the substrate Wp to be processed. The charge distribution measurement unit 130 measures the charge distribution of the processing target substrate Wp.

As shown in FIG. 7(b), static elimination of the processing target substrate Wp is started. The static eliminator 140 discharges a static eliminator onto a region of the processing target substrate Wp that has a relatively low potential. Here, the nozzle 142 of the static eliminator 140 is located above the right end of the processing target substrate Wp, and the static eliminating liquid is discharged onto the right end of the processing target substrate Wp. In a region of the processing target substrate Wp where discharging of the static eliminating liquid is started, if the potential of the processing target substrate Wp is high, there is a risk that discharge will occur, but if the potential is low, the occurrence of discharge can be suppressed.

Thereafter, as shown in FIG. 7C, the nozzle 142 of the static eliminator 140 is scanned over the processing target substrate Wp. Here, the area of the substrate Wp to be processed, where the static eliminator is discharged from the nozzle 142 of the static eliminator 140, is scanned from the edge of the substrate Wp to the process toward the center. In this way, the nozzle 142 of the static eliminator 140 scans the substrate Wp to be processed. The charge on the substrate Wp to be processed is removed by the static eliminating liquid.

As shown in FIG. 7D, by performing the static elimination process on the substrate Wp to be processed, the charge Q of the substrate Wp to be processed is removed, and the amount of charge on the substrate Wp to be processed can be reduced. By removing static electricity from the substrate Wp to be processed as described above, it is possible to suppress the occurrence of discharge in the substrate Wp to be processed.

Note that in the explanation with reference to FIGS. 6 and 7, the potential of the substrate W is positive with respect to the ground, and the supply of the static eliminating liquid is started to a region of the substrate W where the potential is low. However, if the substrate W has a negative charge with respect to the ground, supply of the static eliminating liquid is started to the region of the substrate W that is farthest in potential from the ground.

As described above with reference to FIG. 1, the learned model LM is generated from the learning data LD, and the learning data LD is generated from the time series data TDL of the substrate processing apparatus 100L.

Next, generation of the learning data LD will be explained with reference to FIG. FIG. 8 is a block diagram of a substrate processing system 10L including a substrate processing apparatus 100L and a learning data generation device 300. Here, the learning data generation device 300 is communicably connected to the substrate processing device 100L. In the substrate processing system 10L including the substrate processing apparatus 100L in FIG. 8, the control unit 22L does not have the substrate information acquisition unit 22a and the static elimination processing condition information acquisition unit 22b, and the storage unit 24L stores the learned model LM. The block diagram of the substrate processing system 10 shown in FIG. 4 is the same as that of the substrate processing system 10 except that the test recipe TR is stored without any redundancy, and redundant description will be omitted to avoid redundancy.

The substrate processing system 10L includes a plurality of substrate processing apparatuses 100L, an indexer robot IRL, a center robot CRL, a control device 20L, a display section 42L, an input section 44L, a communication section 46L, and a sensor 50L. The substrate processing apparatus 100L, indexer robot IRL, center robot CRL, control device 20L, display section 42L, input section 44L, and communication section 46L are the substrate processing apparatus 100, indexer robot IR of the substrate processing system 10 shown in FIG. , the center robot CR, the control device 20, the display section 42, the input section 44, and the communication section 46.

Further, the substrate processing apparatus 100L includes a substrate holding section 120L, a charge distribution measuring section 130L, a static eliminating section 140L, a chemical solution supply section 150L, a rinsing liquid supply section 160L, and a cup 180L. The chamber 110L, the substrate holding section 120L, the charge distribution measuring section 130L, the static eliminating section 140L, the chemical solution supply section 150L, the rinsing liquid supply section 160L, and the cup 180L are the substrate holding section 120 and the charge distribution measurement section shown in FIGS. 3 and 4. 130, static eliminator 140, chemical solution supply section 150, rinse liquid supply section 160, and cup 180.

The control device 20L includes a control section 22L and a storage section 24L. The storage unit 24L stores a control program PGL. The substrate processing apparatus 100L operates according to the procedure defined in the control program PGL.

Furthermore, the storage unit 24L stores a plurality of test recipes TR. The plurality of test recipes TR include recipes with different static elimination processing conditions. Therefore, when the control unit 22L processes the learning target substrate WL according to the test recipe TR, different static elimination processes are performed on different learning target substrates WL.

The storage unit 24L stores time-series data TDL of the learning target board WL. The time-series data TDL is data indicating temporal changes in physical quantities in the substrate processing apparatus 100L. The time series data TDL indicates a plurality of physical quantities detected by the sensor 50L. Note that the time series data TDL includes charge distribution information of the learning target board WL measured by the charge distribution measurement unit 130L, static elimination processing condition information indicating the conditions under which the learning target board WL was subjected to static elimination processing, and static elimination processing condition information on the learning target board WL. Contains processing result information indicating the results.

The learning data generation device 300 is communicably connected to the substrate processing device 100L. The learning data generation device 300 communicates at least part of the time series data of the substrate processing device 100L.

The learning data generation device 300 includes a control device 320, a display section 342, an input section 344, and a communication section 346. The learning data generation device 300 can communicate with the control devices 20L of the plurality of substrate processing devices 100L via the communication unit 346. The display section 342, the input section 344, and the communication section 346 have the same configuration as the display section 42, the input section 44, and the communication section 46.

Control device 320 includes a control section 322 and a storage section 324. The storage unit 324 stores a control program PG3. The learning data generation device 300 operates according to the procedure defined in the control program PG3.

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 board 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 causes the storage unit 324 to store at least a portion of the transmitted time series data TDL. The time series data TDL stored in the storage unit 324 includes substrate information, static elimination processing condition information, and processing result information of the time series data TDL.

The control unit 322 acquires substrate information including charge distribution information of the learning target substrate WL, static elimination processing condition information, and processing result information from the time series data TDL in the storage unit 324. Furthermore, the control unit 322 generates learning data LD by combining the board information, static elimination processing condition information, and processing result information of the learning target board WL, and the storage unit 324 stores the learning data LD.

Next, a method for generating learning data according to this embodiment will be described with reference to FIGS. 8 and 9. FIG. 9 is a flow diagram of a method for generating learning data according to this embodiment. Generation of learning data is performed in the learning data generation device 300.

As shown in FIG. 9, in step S110, time series data TDL of the learning target board WL is acquired. Typically, the learning data generation device 300 receives at least a portion of the time series data TDL of the learning target substrate WL from the substrate processing device 100L. The storage unit 324 stores the received time series data TDL.

In step S112, board information is extracted from the time series data TDL of the learning target board WL stored in the storage unit 324. The board information includes charge distribution information of the learning target board WL. The control unit 322 acquires the board information of the learning target board WL from the time series data TDL in the storage unit 324.

In step S114, static elimination processing condition information for the learning target board WL is extracted from the time series data TDL of the learning target board WL stored in the storage unit 324. The control unit 322 acquires static elimination processing condition information for the learning target board WL from the time series data TDL in the storage unit 324.

In step S116, processing result information of the learning target board WL is extracted from the time series data TDL of the learning target board WL stored in the storage unit 324. The control unit 322 acquires processing result information of the learning target board WL from the time series data TDL in the storage unit 324.

In step S118, the board information, static elimination processing condition information, and processing result information of the learning target board WL are associated and generated as learning data LD, and the storage unit 324 stores the learning data LD for each of the plurality of learning target boards WL. do.

In the present embodiment, the generated learning data includes board information, static elimination processing condition information, and processing result information that are associated with each other for each learning target board WL. Such learning data is suitably used for learning processing.

Note that in FIG. 8, the learning data generation device 300 is communicatively connected to one substrate processing device 100L, but the present embodiment is not limited to this. The learning data generation device 300 may be communicably connected to a plurality of substrate processing devices 100L.

Furthermore, in the explanation with reference to FIGS. 8 and 9, the time series data TDL generated by the substrate processing apparatus 100L was transmitted to the learning data generation apparatus 300 via the communication unit 46L and the communication unit 346, but in this embodiment The format is not limited to this. The control device 320 of the learning data generation device 300 is incorporated in the control device 20 of the substrate processing system 10 including the substrate processing device 100L, so that the time-series data TDL can be stored within the substrate processing system 10 without being transferred over the network. The learning data LD may be generated from the time series data TDL.

Next, with reference to FIG. 10, generation of the trained model LM of this embodiment will be described. FIG. 10 is a schematic diagram of the learning data generation device 300 and the learning device 400 of this embodiment. The learning data generation device 300 and the learning device 400 can communicate with each other.

The learning device 400 is communicably connected to the learning data generation device 300. The learning device 400 receives the learning data LD from the learning data generating device 300. The learning device 400 performs machine learning based on the learning data LD and generates a learned model LM.

The learning device 400 includes a control device 420, a display section 442, an input section 444, and a communication section 446. The display section 442, the input section 444, and the communication section 446 have the same configuration as the display section 42, the input section 44, and the communication section 46 of the substrate processing system 10 shown in FIG.

Control device 420 includes a control section 422 and a storage section 424. Note that the storage unit 424 stores the control program PG4. The learning device 400 operates according to the procedure defined in the control program PG4.

The storage unit 424 stores learning data LD. The learning data LD is transmitted from the learning data generation device 300 to the learning device 400 via the communication unit 346 and the communication unit 446. The control unit 422 causes the storage unit 424 to store the transmitted learning data LD. In the learning data LD stored in the storage unit 424, the board information, static elimination processing condition information, and processing result information of the time series data TDL are associated with each other.

The storage unit 424 stores the learning program LPG. The learning program LPG is a program for executing a machine learning algorithm for finding a certain rule from a plurality of learning data LD and generating a learned model LM expressing the found rule. The control unit 422 executes the learning program LPG in the storage unit 424, performs machine learning on the learning data LD, adjusts the parameters of the inference program, and generates the learned model LM.

The machine learning algorithm is not particularly limited as long as it is supervised learning, such as a decision tree, nearest neighbor method, naive Bayes classifier, support vector machine, or neural network. Thus, the trained model LM includes a decision tree, a nearest neighbor method, a naive Bayes classifier, a support vector machine, or a neural network. In machine learning that generates the learned model LM, an error backpropagation method may be used.

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

The control unit 422 includes an acquisition unit 422a and a learning unit 422b. The acquisition unit 422a acquires the 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 in the storage unit 424, and generates a learned model LM from the learning data LD.

The learning unit 422b performs machine learning on the plurality of learning data LD based on the learning program LPG. As a result, a certain rule is found from among the plurality of learning data LD, and a learned model LM is generated. That is, the learned model LM is constructed by performing machine learning on the learning data LD. The storage unit 424 stores the learned model LM.

Note that, typically, the learned model LM is transferred to the control device 20 in the substrate processing system 10, and the storage unit 24 stores the learned model LM. In this case, as described above with reference to FIG. Obtain static elimination processing conditions from completed model LM.

However, this embodiment is not limited to this. The storage unit 24 does not store the learned model LM, and the static elimination processing condition information acquisition unit 22b may acquire the static elimination processing conditions from outside the substrate processing system 10. For example, the static elimination processing condition information acquisition unit 22b transmits the board information of the processing target substrate Wp to the learned model LM of the learning device 400 via the communication unit 46 and the communication unit 446, and The static elimination processing condition information output in the learned model LM may be received from the learning device 400.

Next, a learning method and a learned model generation method in the learning device 400 of this embodiment will be explained with reference to FIGS. 1 to 11. FIG. 11 is a flow diagram of the learning method and learned model of this embodiment. Learning of the learning data LD and generation of the learned model LM are performed in the learning device 400.

As shown in FIG. 11, in step S122, the acquisition unit 422a of the learning device 400 acquires a plurality of learning data LD from the storage unit 424. In the learning data LD, the board information, static elimination processing condition information, and processing result information of the time series data TDL are associated with each other.

Next, in step S124, the learning unit 422b performs machine learning on the plurality of learning data LD based on the learning program LPG.

Next, in step S126, the learning unit 422b determines whether the machine learning of the learning data LD is finished. Whether or not to end machine learning is determined according to predetermined conditions. For example, machine learning ends when a predetermined number or more of learning data LD are machine learned.

If machine learning does not end (No in step S126), the process returns to step S122. In that case, machine learning is repeated. On the other hand, if the machine learning does not end (Yes in step S126), the process proceeds to step S128.

In step S128, the learning unit 422b outputs a model (one or more functions) to which a plurality of latest parameters (coefficients), that is, a plurality of learned parameters (coefficients) are applied, as a learned model LM. The storage unit 424 stores the learned model LM.

As described above, the learning method is completed and the trained model LM is generated. According to this embodiment, the learned model LM can be generated by performing machine learning on the learning data LD.

Note that in FIG. 10, the learning device 400 is communicatively connected to one learning data generation device 300, but the present embodiment is not limited to this. The learning device 400 may be communicatively connected to a plurality of learning data generation devices 300.

Furthermore, in the explanation with reference to FIGS. 10 and 11, the learning data LD generated by the learning data generation device 300 was transmitted to the learning device 400 via the communication unit 346 and the communication unit 446, but in this embodiment is not limited to this. The control device 420 of the learning device 400 is incorporated in the control device 320 of the learning data generation device 300, and the learning data LD is stored within the learning data generation device 300 without being transferred over the network. A learned model LM may be generated from the LD.

Furthermore, in the explanation with reference to FIGS. 8 to 11, 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, and the learning data generation apparatus 300 generates learning data. Although the learning data LD generated by the device 300 was transmitted to the learning device 400 via the communication unit 346 and the communication unit 446, the present embodiment is not limited thereto. The control device 320 of the learning data generation device 300 and the control device 420 of the learning device 400 are incorporated in the control device 20 of the substrate processing system 10L, and the time series data TDL and the learning data LD are transferred through the network. Instead, the learned model LM may be generated from the time series data TDL in the substrate processing system 10L via the learning data LD.

Next, an example of the learning data LD will be described with reference to FIG. 12. FIG. 12 is a diagram showing an example of the learning data LD.

FIG. 12 shows board information of the learning target board WL, static elimination processing conditions of the learning target board WL, and static elimination processing results of the learning target board WL. Here, the board information is charge distribution information indicating the charge distribution of the learning target board WL. The charge distribution information is generated by measuring the learning target substrate WL by the charge distribution measurement unit 130L before subjecting the learning target substrate WL to static elimination processing. The static elimination processing condition information indicates, for example, the concentration of the static eliminating liquid, the temperature of the static eliminating liquid, the supply amount of the static eliminating liquid, the discharge pattern of the static eliminating liquid, and the rotation speed of the learning target substrate WL when supplying the static eliminating liquid. The processing result shows the result of the learning target board WL after the static elimination processing. In FIG. 12, the learning data LD includes learning data LD1 to learning data LD1000.

The learning data LD1 indicates board information, static elimination processing conditions, and processing results for a certain learning target board WL1. Here, in the learning data LD1, the board information indicates the charge distribution of the learning target board WL1. Furthermore, Lp1 indicates the conditions for the static elimination process performed on the learning target board WL1.

The processing result information indicates the processing result of the static elimination process for the learning target board WL1. The processing result may be determined based on whether or not discharge has occurred on the learning target board WL1. Alternatively, the processing result may be determined based on whether or not a characteristic abnormality is discovered in the learning target board WL1. In the learning data LD1, the result of the learning target board that was subjected to the static elimination process was good, so it is indicated as ○.

The learning data LD2 to LD1000 are generated corresponding to the learning target boards WL2 to WL1000. As shown in the learning data LD2 to LD1000, the charge distribution of the learning target substrate WL typically differs from substrate to substrate. Note that the static elimination processing conditions performed on the learning target board WL may be the same or different. The static elimination processing result changes depending on the charging distribution of the learning target substrate WL and the processing conditions.

In the learning data LD shown in FIG. 12, the number of data is 1000, but the present embodiment is not limited to this. The number of data may be smaller than 1000 pieces or larger than 1000 pieces. However, it is preferable that the number of data is as large as possible.

In addition, in the above explanation with reference to FIG. 12, when the processing result is good, it is shown as ○, and when the processing result is not good, it is shown as ×, and the processing results of learning data LD1 to LD1000 are binarized. However, this embodiment is not limited to this. The processing result may be classified into three or more values. Alternatively, the processing results may be classified into any value between the lowest value and the highest value. For example, the processing result may be quantified in consideration of the amount of static elimination liquid used (supply amount), the time required for static elimination processing, etc., in addition to the characteristics of the learning target substrate WL.

Note that in the learning data LD, it is preferable that the static elimination processing conditions include a plurality of items. For example, the static elimination processing conditions may include the concentration, temperature, and supply amount of the static elimination liquid, the discharge pattern of the static elimination liquid to the learning target substrate, and the rotation speed of the substrate holding unit 120 during static elimination.

Next, with reference to FIG. 13, the learning data LD used in the learning method of this embodiment will be explained. FIG. 13 is a diagram showing an example of the learning data LD.

As shown in FIG. 13, the learning data LD includes learning data LD1 to LD1000. The learning data LD indicates board information of the learning target board WL, static elimination processing conditions of the learning target board WL, and processing result information of the learning target board WL. Here, the static elimination processing conditions include the concentration of the static eliminator, the temperature of the static eliminator, the supply amount of the static eliminator, the discharge pattern of the static eliminator to the learning target board, and the rotation speed of the learning target board when supplying the static eliminator. .

The concentration of the static eliminator indicates the concentration of the static eliminator used for the learning target substrate WL. The temperature of the static eliminator and the temperature of the static eliminator used for the learning target board WL are shown. The supply amount of the static eliminating liquid indicates the supplied amount of the static eliminating liquid used for the learning target board WL. The discharge pattern of the static eliminator onto the learning target substrate indicates a path (movement route of the nozzle 142) along which the static eliminator was ejected onto the learning target substrate WL. The rotational speed of the learning target substrate indicates the rotational speed of the learning target substrate WL when supplying the static eliminating liquid. Further, in the learning data LD, when the result of the learning target board subjected to static elimination processing is good, it is indicated as "〇", and when the result of static elimination processing of the learning target board is not good, it is indicated as "x".

In the learning data LD1, Lc1 indicates the concentration of the static eliminator used for the learning target board WL1, and Lt1 indicates the temperature of the static eliminator used for the learning target board WL1. Further, Ls1 indicates the supply amount of the static eliminator used for the learning target board WL1, Le1 represents the discharge pattern of the static eliminator for the learning target board WL1, and Lv1 represents the static eliminator used for the learning target board WL1. The rotation speed of the learning target substrate WL1 when supplying liquid is shown. In the learning data LD1, the processing result of the learning target board subjected to the static elimination process was good, so the processing result is "O".

The same applies to the learning data LD2 to LD1000. As shown in the learning data LD2 to LD1000, the charge distribution of the learning target substrate WL typically differs from substrate to substrate. Note that the static elimination processing conditions performed on the learning target board WL may be the same or different. For example, at least some of the plurality of items of the static elimination processing conditions may be the same or different. Alternatively, all of the plurality of items of the static elimination processing conditions may be the same or different. The static elimination processing result varies depending on the charging distribution of the learning target substrate WL and the static elimination processing conditions.

Next, the static elimination process in the substrate processing apparatus 100 of this embodiment will be described with reference to FIGS. 1 to 14. FIG. 14(a) shows the charge distribution of the processing target substrate Wp, and FIG. 14(b) shows the static elimination processing condition information Rp generated in the learned model LM.

As shown in FIG. 14(a), the processing target substrate Wp has a charge distribution. The charge distribution of the processing target substrate Wp is measured by the charge distribution measuring section 130. When the control unit 22 inputs charge distribution information indicating the charge distribution of the processing target substrate Wp to the learned model LM, the learned model LM generates charge removal processing condition information Rp corresponding to the charge distribution information.

FIG. 14(b) is a diagram showing static elimination processing condition information Rp. The static elimination processing condition information Rp includes the concentration of the static eliminating liquid, the temperature of the static eliminating liquid, the supply amount of the static eliminating liquid, the discharge pattern of the static eliminating liquid to the learning target substrate, and the rotation speed of the target substrate when supplying the static eliminating liquid. .

In the static elimination processing condition information Rp, Rc indicates the concentration of the static eliminating liquid used for the processing target substrate Wp, and Rt indicates the temperature of the static eliminating liquid used for the processing target substrate Wp. Further, Rs indicates the supply amount of the static eliminator used for the processing target substrate Wp, Re indicates the discharge pattern of the static eliminator used for the processing target substrate Wp, and Rv indicates the supply amount of the static eliminator used for the processing target substrate Wp. 3 shows the rotational speed of the processing target substrate Wp when supplying the static eliminating liquid.

In the explanation with reference to FIGS. 13 and 14, the static elimination processing conditions include the concentration of the static eliminating liquid, the temperature of the static eliminating liquid, the supply amount of the static eliminating liquid, the discharge pattern of the static eliminating liquid, and the rotation speed of the substrate to be processed. Although there are two items, the present embodiment is not limited thereto. The static elimination processing conditions may include any one or more of these five items. Alternatively, the static elimination processing conditions may be a combination of one or more of these five items and another item. Alternatively, the static elimination processing conditions may include one or more items different from these five items.

Note that in the above description with reference to FIGS. 12 to 14, the board information of the learning target board WL is charge distribution information indicating the charge distribution of the learning target board WL, but the present embodiment is not limited to this. . The board information of the learning target board WL may include information other than charge distribution information.

Next, with reference to FIG. 15, the learning data LD used in the learning method of this embodiment will be explained. FIG. 15 is a diagram showing an example of the learning data LD. Note that the learning data LD in FIG. 15 is similar to that shown in FIG. 13 except that the substrate information includes information indicating the film thickness and film type of the learning target substrate WL in addition to the charge distribution of the learning target substrate WL. This is similar to the learning data LD described above, and duplicate descriptions will be omitted to avoid redundancy.

As shown in FIG. 15, the learning data LD includes learning data LD1 to LD1000. The learning data LD indicates board information of the learning target board WL, static elimination processing conditions of the learning target board WL, and processing result information of the learning target board WL. Here, the substrate information of the learning target substrate WL includes the charge distribution of the learning target substrate WL, the film thickness and the film type of the learning target substrate WL.

Further, the static elimination processing conditions include the concentration of the static eliminating liquid, the temperature of the static eliminating liquid, the supply amount of the static eliminating liquid, the discharge pattern of the static eliminating liquid to the learning target substrate, and the rotation speed of the learning target substrate when supplying the static eliminating liquid. Note that in FIG. 15, Lc indicates the concentration of the static eliminator used for the learning target substrate WL, and Lt indicates the temperature of the static eliminator used for the learning target substrate WL. Further, Ls indicates the supply amount of the static eliminator used for the learning target substrate WL, and Le indicates the discharge pattern of the static eliminator for the learning target substrate WL. Further, Lv indicates the rotational speed of the learning target substrate WL when the static eliminating liquid is supplied to the learning target substrate WL.

The film thickness of the learning target substrate WL indicates the thickness of a film existing on the surface. Note that the film thickness of the learning target substrate WL may indicate the thickness of two or more films existing on the surface.

The film type of the learning target substrate WL indicates the type of film present on the surface (upper surface Wa). For example, the film type of the learning target substrate WL is that the film present on the surface is a silicon oxide film, a silicon nitride film, or a mixture thereof. When the film thickness of the learning target substrate WL indicates the thickness of two or more films present on the surface, the film type may indicate the types of the two or more target films.

In the learning data LD1, Ld1 indicates the thickness of the film present on the surface of the learning target substrate WL1. Furthermore, Lk1 indicates the type of film present on the surface of the learning target substrate WL1.

The same applies to the learning data LD2 to LD1000. As shown in the learning data LD2 to LD1000, the charge distribution of the learning target substrate WL typically differs from substrate to substrate. Further, even if the charge distribution of the learning target substrate WL is almost the same, the degree of static elimination processing varies greatly depending on the thickness and/or film type of the film on the surface of the learning target substrate WL. Therefore, it is preferable that the substrate information includes the film thickness and/or film type in addition to the charge distribution. Note that the substrate information may include a combination of one of the film thickness and film type and other items. Alternatively, the substrate information may include one or more items other than the film thickness and film type.

Note that in the substrate processing apparatus 100 shown in FIG. 3, the static eliminating unit 140 supplies the substrate W with the static eliminating liquid in a predetermined state from the supply source, but the present embodiment is not limited to this. The concentration of the static eliminating liquid may be changed as appropriate. Furthermore, in the substrate processing apparatus 100 shown in FIG. 3, the static eliminator 140 supplies the static eliminator liquid to the upper surface Wa of the substrate W, but the present embodiment is not limited to this. The static eliminating liquid may be supplied not only to the upper surface Wa of the substrate W but also to the back surface Wb of the substrate W.

Next, the substrate processing apparatus 100 of this embodiment will be described with reference to FIG. 16. FIG. 16 is a schematic diagram of the substrate processing apparatus 100 of this embodiment. Note that the substrate processing apparatus 100 in FIG. 16 is the same as the substrate processing apparatus 100 described above with reference to FIG. They are the same, and duplicate descriptions will be omitted to avoid redundancy.

In the substrate processing apparatus 100 of this embodiment shown in FIG. 16, the static eliminator 140 further includes a mixing unit 149, a nozzle 149a, a pipe 149b, and a valve 149c in addition to the nozzle 142, the pipe 144, and the valve 146. Mixing unit 149 is connected to a supply source. Here, pure water and carbon dioxide are supplied to the mixing unit 149 from a supply source.

The mixing unit 149 generates carbonated water as a static eliminator by mixing pure water and carbon dioxide. Note that the mixing ratio of pure water and carbon dioxide is changed according to instructions from the control unit 22. Therefore, the concentration of the static eliminating liquid can be changed by the mixing unit 149. Mixing unit 149 is connected to piping 144 .

The nozzle 149a faces the back surface Wb of the substrate W and discharges the static eliminating liquid toward the back surface Wb of the substrate W. Piping 149b is coupled to nozzle 149a. The piping 149b passes through a hollow hole in the shaft 123 of the substrate holder 120. The nozzle 149a is provided at the tip of the pipe 149b. Piping 149b communicates a position between valve 146 and mixing unit 149 in piping 144 and nozzle 149a. Static eliminating liquid is supplied from the mixing unit 149 to the pipe 149b. Valve 149c is provided in piping 149b. Valve 149c opens and closes the flow path within piping 149b.

According to this embodiment, since the mixing unit 149 can change the concentration of the static eliminator, the static eliminator 140 can supply the substrate W with static eliminators having different concentrations. Furthermore, the static eliminator 140 can supply the static eliminator not only to the top surface Wa of the substrate W but also to the back surface Wb.

Typically, the back surface Wb of the substrate W is a non-device forming surface, so even if discharge occurs on the back surface Wb, it may have little effect on the characteristics of the substrate. In this case, it is preferable that the substrate processing apparatus 100 supplies the static eliminating liquid to the back surface Wb of the substrate W before supplying the static eliminating liquid to the upper surface Wa of the substrate W.

Next, with reference to FIG. 17, the learning data LD used in the learning method of this embodiment will be explained. FIG. 17 is a diagram showing an example of the learning data LD. The learning data LD shown in FIG. 17 is suitably used to generate the trained model LM of the substrate processing apparatus 100 shown in FIG. 16. Note that the learning data LD in FIG. 17 is similar to the learning data LD described above with reference to FIG. 13, except that the static elimination processing condition has two values for at least one item of one learning data. , and duplicate descriptions are omitted to avoid redundancy.

As shown in FIG. 17, the learning data LD includes learning data LD1 to LD1000. The static elimination processing conditions include the concentration of the static eliminating liquid, the temperature of the static eliminating liquid, the supply amount of the static eliminating liquid, the discharge pattern of the static eliminating liquid to the learning target substrate, and the rotation speed of the learning target substrate when supplying the static eliminating liquid. In FIG. 17, Lc represents the concentration of the static eliminator used for the learning target substrate WL, and Lt represents the temperature of the static eliminator used for the learning target substrate WL. Further, Ls indicates the supply amount of the static eliminator used for the learning target substrate WL, and Le indicates the ejection pattern and ejection timing of the static eliminator for the learning target substrate WL. Further, Lv indicates the rotational speed of the learning target substrate WL when the static eliminating liquid is supplied to the learning target substrate WL.

In FIG. 17, one learning data has two values for the items of the static eliminator concentration Lc, the supply amount Ls, and the ejection pattern Le of the static eliminator.

In the substrate processing apparatus 100 shown in FIG. 16, the concentration of the static eliminating liquid can be changed by the mixing unit 149. Therefore, in the learning data LD1, Lcu1 indicates the concentration of the static eliminating liquid supplied from the nozzle 142 to the upper surface Wa of the learning target substrate WL, and Lcd1 indicates the concentration of the static eliminating liquid supplied from the nozzle 149a to the back surface Wb of the learning target substrate WL. Indicates the concentration of static eliminator.

In the learning data LD1, Lsu1 indicates the supply amount of the static eliminating liquid supplied from the nozzle 142 to the upper surface Wa of the learning target substrate WL, and Lsd1 indicates the supply amount of the static eliminating liquid supplied from the nozzle 149a to the back surface Wb of the learning target substrate WL. Indicates the supply amount of static eliminator. Further, Leu1 indicates the discharge pattern and discharge timing of the static eliminating liquid from the nozzle 142 to the upper surface Wa of the learning target substrate WL, and Led1 indicates the discharge timing of the static eliminating liquid from the nozzle 149a to the back surface Wb of the learning target substrate WL. .

The same applies to the learning data LD2 to LD1000. As shown in the learning data LD2 to LD1000, the charge distribution of the learning target substrate WL typically differs from substrate to substrate. Furthermore, the result of the static elimination process on the learning target substrate WL varies greatly depending on the concentration, supply amount, discharge pattern, and discharge timing of the static eliminating liquid supplied from the nozzle 142 and the nozzle 149a for the learning target substrate WL. For this reason, it is preferable that the learning data LD includes items that greatly contribute to fluctuations in the result of static elimination processing of the learning target substrate WL.

In the above description, the static eliminator 140 supplies a so-called static eliminator, but in this embodiment, the static eliminator 140 is not limited to supplying a so-called static eliminator. For example, the static eliminator 140 may supply a processing liquid other than the static eliminator. The static eliminator 140 may supply the processing liquid so that no discharge occurs on the processing target substrate Wp.

Note that in the substrate processing apparatus 100 described above, the static eliminating unit 140 supplies the static eliminating liquid to the substrate W, but the present embodiment is not limited to this. The static eliminator 140 may supply the static eliminator to the chamber 110 toward the substrate W.

Next, the substrate processing apparatus 100 of this embodiment will be explained with reference to FIG. FIG. 18 is a schematic diagram of the substrate processing apparatus 100 of this embodiment. Note that the substrate processing apparatus 100 in FIG. 18 is similar to the substrate processing apparatus 100 described above with reference to FIG. Omit duplicate descriptions to avoid duplication.

In the substrate processing apparatus 100 of this embodiment shown in FIG. 18, the static eliminator 140 supplies the substrate W with a static eliminator. The electricity eliminating body is, for example, a highly humid gas. The charge eliminating body may be generated by adding water vapor to an inert gas.

Static eliminator 140 includes a nozzle 142, piping 144, and valve 146. One end of the piping 144 is connected to the nozzle 142, and the other end of the piping 144 is connected to a supply source of an electric eliminator. The pipe 144 is supplied with an electricity eliminating body from a supply source. Valve 146 is provided in piping 144 . Valve 146 opens and closes the flow path within piping 144 .

Note that the electricity eliminating body may be supplied before the chemical treatment. Alternatively, the charge eliminating body may be supplied during chemical treatment. According to the present embodiment, the charge on the substrate can be easily suppressed by the charge eliminating body.

Next, with reference to FIG. 19, the learning data LD used in the learning method of this embodiment will be explained. FIG. 19 is a diagram showing an example of the learning data LD. The learning data LD shown in FIG. 19 is suitably used to generate the trained model LM of the substrate processing apparatus 100 shown in FIG. 18. Note that the learning data LD in FIG. 19 is the same as the learning data LD described above with reference to FIG. 13, except that the ejection pattern Le is omitted, and duplicate descriptions are omitted to avoid redundancy. do.

As shown in FIG. 19, the learning data LD includes learning data LD1 to LD1000. The static elimination processing conditions include the humidity of the static eliminating body, the temperature of the static eliminating body, the supply amount of the static eliminating body, and the rotation speed of the learning target board when supplying the static neutralizing body. In FIG. 19, Lc indicates the humidity of the electricity eliminator used for the learning target board WL, and Lt indicates the temperature of the electricity eliminator used for the learning target board WL. Further, Ls indicates the supply amount of the charge eliminator used for the learning target board WL, and Lv indicates the rotation speed of the learning target board WL when supplying the charge remover to the learning target board WL. show.

The same applies to the learning data LD2 to LD1000. As shown in the learning data LD2 to LD1000, the charge distribution of the learning target substrate WL typically differs from substrate to substrate. It is preferable that the learning data LD includes items that greatly contribute to fluctuations in the result of static elimination processing of the learning target board WL. Note that, in order to efficiently eliminate static electricity from the substrate W, the static eliminating body may be used in combination with a static eliminating liquid.

Note that in the substrate processing apparatus 100 described above, the static eliminator 140 supplies the static eliminator liquid or the static eliminator to the substrate W, but the present embodiment is not limited thereto. The static eliminator 140 may remove static electricity from the substrate W using ultraviolet light.

Next, the substrate processing apparatus 100 of this embodiment will be described with reference to FIG. 20. FIG. 20 is a schematic diagram of the substrate processing apparatus 100 of this embodiment. Note that the substrate processing apparatus 100 in FIG. 20 is similar to the substrate processing apparatus 100 described above with reference to FIG. 3, except that the static eliminator 140 uses ultraviolet light to neutralize the substrate W, and in order to avoid redundancy, Omit duplicate descriptions.

In the substrate processing apparatus 100 of this embodiment shown in FIG. 20, the static eliminator 140 irradiates ultraviolet rays to eliminate static from the substrate W.

The static eliminator 140 includes a light source 140U and a light source moving unit 148U. The light source 140U emits ultraviolet light. For example, the light source 140U includes an excimer lamp. In one example, the light source 140U includes a quartz tube charged with a discharge gas and a pair of electrodes disposed within the quartz tube.

The light source moving unit 148U moves the light source 140U between the irradiation position and the retreat position. When the light source 140U is in the irradiation position, the light source 140U is located above the substrate W. When the light source 140U is in the irradiation position, the light source 140U irradiates the upper surface Wa of the substrate W with ultraviolet rays. When the light source 140U is in the retracted position, the light source 140U is located on the outer side of the substrate W in the radial direction of the substrate W.

The light source moving unit 148U includes an arm 148a, a rotation shaft 148b, and a moving mechanism 148c. Arm 148a extends substantially horizontally. A light source 140U is attached to the tip of the arm 148a. Arm 148a is coupled to pivot shaft 148b. The rotation shaft 148b extends substantially vertically. The moving mechanism 148c rotates the rotation shaft 148b around a rotation axis along a substantially vertical direction, and rotates the arm 148a along a substantially horizontal plane. As a result, the light source 140U moves along a substantially horizontal plane. For example, the moving mechanism 148c includes an arm swing motor that rotates the rotation shaft 148b around the rotation axis. The arm swing motor is, for example, a servo motor.

Next, with reference to FIG. 21, the learning data LD used in the learning method of this embodiment will be explained. FIG. 21 is a diagram showing an example of the learning data LD. The learning data LD shown in FIG. 21 is suitably used to generate the learned model LM of the substrate processing apparatus 100 shown in FIG. 20. Note that the learning data LD in FIG. 21 is the same as the learning data LD described above with reference to FIG. 13, except that the concentration Lc, temperature Lt, and ejection pattern Le are omitted, and in order to avoid redundancy, Omit duplicate descriptions.

As shown in FIG. 21, the learning data LD includes learning data LD1 to LD1000. The static elimination processing conditions include the irradiation intensity from the light source and the rotation speed of the learning target substrate when irradiating the ultraviolet rays. In FIG. 17, Ls1 indicates the irradiation intensity from the light source used for the learning target substrate WL1, and Lv1 indicates the rotation speed of the learning target substrate WL1 when irradiating ultraviolet rays.

The same applies to the learning data LD2 to LD1000. As shown in the learning data LD2 to LD1000, the charge distribution of the learning target substrate WL typically differs from substrate to substrate. It is preferable that the learning data LD includes items that greatly contribute to fluctuations in the result of static elimination processing of the learning target board WL.

Note that in the substrate processing apparatus 100 described above, the static eliminator 140 applies the static eliminator, the static eliminator, or the ultraviolet light to the substrate W, but the present embodiment is not limited thereto. The substrate holding unit 120 may function as the static eliminating unit 140 to eliminate static from the substrate W.

Next, the substrate processing apparatus 100 of this embodiment will be described with reference to FIG. 22. FIG. 22(a) is a schematic diagram of the substrate processing apparatus 100 of this embodiment, and FIG. 22(b) is a schematic plan view of the substrate holding section 120 in the substrate processing apparatus 100 of this embodiment. Note that the substrate processing apparatus 100 in FIG. 22 is similar to the substrate processing apparatus 100 described above, except that the substrate holder 120 eliminates static electricity from the substrate W, and a duplicate description will be omitted to avoid redundancy.

In the substrate processing apparatus 100 of this embodiment shown in FIG. 22, the substrate holder 120 includes a spin base 121, a chuck member 122, a shaft 123, and an electric motor 124. The chuck member 122 is provided on the spin base 121. The chuck member 122 chucks the substrate W. The spin base 121 is provided with six chuck members 122 (ie, chuck members 122a to 122f).

Here, the chuck member 122 contacts the end of the upper surface Wa of the substrate W to chuck the substrate W. The chuck members 122 are individually driven to rotate in the circumferential direction by a drive mechanism housed within the spin base 121. Here, when the chuck member 122 rotates counterclockwise, the chuck member 122 chucks the substrate W. Conversely, when the chuck member 122 rotates clockwise, the substrate W is detached from the chuck member 122.

In the substrate processing apparatus 100 of this embodiment, the chuck member 122 is made of a conductive material. Further, inside the spin base 121, a wiring 121L separated for each chuck member 122 is provided, and the shaft 123 is provided with a wiring 123L connected to the plurality of wirings 121L of the spin base 121. One end of the wiring 123L is connected to the ground of the substrate processing apparatus 100.

Therefore, when the chuck member 122 rotates and comes into contact with the end of the upper surface Wa of the substrate W, the substrate W can be neutralized. Therefore, the substrate holding section 120 functions as a static eliminator 140. Further, when a specific chuck member 122 among the six chuck members 122a to 122f is rotated and brought into contact with the substrate W, the charges in a specific region of the substrate W are preferentially transferred to the outside of the substrate processing apparatus 100. can. Further, after the static electricity is removed, the remaining chuck member 122 is rotated and brought into contact with the substrate W, so that the chuck member 122 can chuck the substrate W.

22(c) and 22(d) are schematic plan views of the substrate holding section 120 in the substrate processing apparatus 100 of this embodiment. FIG. 22(c) also shows the charge distribution of the substrate W measured by the charge distribution measuring section 130. As shown in FIG. 22(c), there is a low potential portion at the end of the upper right region of the substrate W. In this case, the chuck member 122a located near the end of the upper right region of the substrate W is rotated and brought into contact with the substrate W, while the other chuck members 122b to 122f are not rotated. In this case, the charge on the substrate W flows gently from the upper right side end of the substrate W to the ground via the chuck member 122a and the wirings 121L and 123L connected to the chuck member 122a. It is possible to suppress the generation of electrical discharge during chucking.

Further, as shown in FIG. 22(d), in another substrate W, there are two parts of the substrate W having a low potential at the end of the diagonally upper right region and the end of the diagonally lower right region. In that case, the chuck members 122a and 122c located near the end of the diagonally upper right region and the end of the lower right region of the substrate W are rotated and brought into contact with the substrate W, while the other chuck members 122b and 122d ~122f is not rotated. In this case, the charge on the substrate W is transferred from the edge of the diagonally upper right region and the edge of the lower right region of the substrate W to the ground via the chuck members 122a, 122c and the wirings 121L, 123L connected to the chuck members 122a, 122c. Since it flows slowly, it is possible to suppress the generation of electric discharge when the substrate W is chucked by the chuck member 122.

Next, with reference to FIG. 23, the learning data LD used in the learning method of this embodiment will be explained. FIG. 23 is a diagram showing an example of the learning data LD. The learning data LD shown in FIG. 23 is suitably used to generate the trained model LM of the substrate processing apparatus 100 shown in FIG. 22. Note that the learning data LD in FIG. 23 is the same as the learning data LD described above with reference to FIG. 13, except that the static elimination processing condition has the rotation start pattern Lr, and is duplicated to avoid redundancy. Description is omitted.

As shown in FIG. 23, the learning data LD includes learning data LD1 to LD1000. The static elimination processing conditions include a chuck rotation start pattern.

The learning data LD1 indicates board information, static elimination processing conditions, and processing results for a certain learning target board WL1. Here, in the learning data LD1, the board information indicates the charge distribution of the learning target board WL1. Further, Lr1 indicates a pattern of the chuck member that is first rotated when chucking the learning target substrate WL1.

The same applies to the learning data LD2 to LD1000. As shown in the learning data LD2 to LD1000, the charge distribution of the learning target substrate WL typically differs from substrate to substrate. It is preferable that the learning data LD includes items that greatly contribute to fluctuations in the result of static elimination processing of the learning target board WL.

Note that in the above explanation with reference to FIGS. 1 to 23, the storage unit 24 of the substrate processing apparatus 100 or the storage unit 424 of the learning device 400 stores the trained model LM constructed by machine learning. Embodiments are not limited thereto. The storage unit 24 of the substrate processing apparatus 100 or the storage unit 424 of the learning device 400 may store a conversion table instead of the learned model LM.

Next, the substrate processing apparatus 100 of this embodiment will be explained with reference to FIG. The substrate processing apparatus 100 in FIG. 24 has the same configuration as the substrate processing apparatus 100 described above with reference to FIG. 4, except that the storage unit 24 stores a conversion table CT instead of the learned model LM. To avoid redundancy, duplicate descriptions will be omitted.

As shown in FIG. 24, in the substrate processing apparatus 100, the storage unit 24 stores a conversion table CT. The conversion table CT associates the substrate information of the processing target substrate Wp with the static elimination processing conditions. The substrate information of the processing target substrate Wp includes, for example, the lowest potential value and the lowest potential position of the substrate W. Note that the conversion table CT is created based on the board information of the learning target board WL, the static elimination processing condition information, and the processing result information.

The substrate information acquisition unit 22a identifies the lowest potential of the processing target substrate Wp and its position from the charge distribution of the processing target substrate Wp measured by the charge distribution measurement unit 130.

The static elimination processing condition information acquisition unit 22b acquires static elimination processing condition information from the board information based on the conversion table CT. Typically, the static elimination processing condition information acquisition unit 22b extracts a value corresponding to the board information from the conversion table CT, and based on the relationship between the board information and the static elimination processing condition information associated in the conversion table CT, Obtain static elimination processing condition information. In this way, the static elimination processing condition information acquisition unit 22b uses the conversion table CT to acquire static elimination processing condition information corresponding to the board information.

Thereafter, the control unit 22 controls the substrate holding unit 120 and the static elimination unit 140 according to the static elimination processing conditions indicated in the static elimination processing condition information.

FIG. 25 is a diagram showing an example of the conversion table CT. As shown in FIG. 25, the conversion table CT shows substrate information of the processing target substrate Wp and static elimination processing conditions. Here, in the conversion table CT, the substrate information indicates information regarding the charge distribution of the processing target substrate Wp. Here, the information regarding the charge distribution of the processing target substrate Wp includes the lowest potential position and the lowest potential value of the processing target substrate Wp. The lowest potential position of the processing target substrate Wp is represented by the xy coordinates of the processing target substrate Wp. The static elimination processing condition information indicates the conditions for the static elimination processing to be performed on this processing target substrate Wp.

Conversion table CT1 indicates static elimination processing conditions that correspond to certain board information. Here, in the conversion table CT1, the substrate information includes the lowest potential position and lowest potential value of the processing target substrate Wp. (x1, y1) indicates the lowest potential position of a certain processing target substrate Wp. C1 indicates the lowest potential value of a certain processing target substrate Wp. Rp1 indicates the conditions for the static elimination process to be performed on this substrate Wp to be processed. Therefore, if the lowest potential position of the substrate Wp to be processed is (x1, y1) and the lowest potential value is C1, the substrate processing apparatus 100 performs the static elimination process under the static elimination process conditions indicated by Rp1. .

The same applies to conversion tables CD2 to CD1000. Typically, conversion tables CT1 to CD1000 differ in at least one of the lowest potential position and the lowest potential value.

Note that if the lowest potential value and the position of the substrate to be processed Wp measured by the charge distribution measurement unit 130 do not match the values shown in the conversion table, the static elimination processing conditions for the substrate to be processed Wp are changed as shown in the conversion table CT. It may be determined by linear interpolation of the values of the static electricity removal processing conditions. Alternatively, the static elimination processing conditions for the processing target substrate Wp may be determined by interpolating the values of the static elimination processing conditions shown in the conversion table CT using a polynomial.

The embodiments of the present invention have been described above with reference to the drawings. However, the present invention is not limited to the above-described embodiments, and can be implemented in various forms without departing from the spirit thereof. Moreover, various inventions can be formed by appropriately combining the plurality of components disclosed in the above embodiments. For example, some components may be deleted from all the components shown in the embodiments. Furthermore, components of different embodiments may be combined as appropriate. For ease of understanding, the drawing mainly shows each component schematically, and the thickness, length, number, spacing, etc. of each component shown in the diagram may differ from the actual one for convenience of drawing. may be different. Further, the materials, shapes, dimensions, etc. of each component shown in the above embodiments are merely examples, and are not particularly limited, and various changes can be made without substantially departing from the effects of the present invention. be.

INDUSTRIAL APPLICABILITY The present invention is suitably used for a substrate processing apparatus, a substrate processing method, a learning data generation method, a learning method, a learning device, a learned model generation method, and a learned model.

10 Substrate processing system 20 Control device 22 Control section 22a Substrate information acquisition section 22b Static elimination processing condition information acquisition section 24 Storage section LM Learned model 100 Substrate processing device 130 Charge distribution measurement section 140 Chemical solution supply section 150 Rinse liquid supply section 200 Substrate processing Learning system 300 Learning data generation device 400 Learning device

Claims (13)

a substrate holder that rotatably holds a substrate to be processed;
a charge distribution measuring unit that measures the charge distribution of the substrate to be processed;
a static eliminator that removes static from the substrate to be processed;
a substrate information acquisition unit that acquires substrate information including charge distribution information indicating the charge distribution of the substrate to be processed;
a static elimination processing condition information acquisition unit that acquires static elimination processing condition information indicating static elimination processing conditions of the processing target board from the learned model based on the board information;
a control unit that controls the substrate holding unit and the static elimination unit to eliminate static electricity from the substrate to be processed based on the static elimination process condition information acquired by the static elimination process condition information acquisition unit;
The learned model includes board information including charge distribution information indicating the charge distribution of the learning target board, static elimination processing condition information indicating the conditions under which the learning target board was subjected to static elimination processing, and the results of static elimination processing of the learning target board. A substrate processing apparatus constructed by performing machine learning on learning data associated with processing result information shown in FIG. The substrate processing apparatus according to claim 1, further comprising a storage unit that stores the learned model. 3. The substrate processing apparatus according to claim 1, wherein, for each of the processing target substrate and the learning target substrate, the substrate information further includes information indicating a film type or film thickness of the substrate. The substrate processing apparatus according to any one of claims 1 to 3, wherein a static eliminating liquid is supplied to each of the processing target substrate and the learning target substrate. For each of the processing target board and the learning target board, the static elimination processing condition information includes the concentration of the static eliminating liquid, the temperature of the static eliminating liquid, the supply amount of the static eliminating liquid, the discharge pattern of the static eliminating liquid, and the static eliminating liquid. The substrate processing apparatus according to claim 4, further comprising information indicating one of the rotational speeds of the substrate when supplying the liquid. 4. The substrate processing apparatus according to claim 1, wherein the processing target substrate and the learning target substrate are each irradiated with ultraviolet rays. rotatably holding the substrate to be processed;
acquiring substrate information including charge distribution information indicating the charge distribution of the substrate to be processed;
acquiring static elimination processing condition information indicating static elimination processing conditions for the processing target substrate from the learned model based on the substrate information;
A substrate processing method comprising the step of neutralizing the substrate to be processed in accordance with the static elimination processing conditions of the static elimination processing condition information,
In the step of acquiring the static elimination processing condition information, the learned model includes board information including charge distribution information indicating the charge distribution of the learning target board, and static elimination processing condition information indicating the conditions under which the learning target board was subjected to static elimination processing. , a substrate processing method constructed by performing machine learning on learning data associated with processing result information indicating a result of static elimination processing on the learning target board. acquiring substrate information including charge distribution information indicating charge distribution of the learning target board from time-series data output from a substrate processing apparatus that processes the learning target board;
acquiring, from the time-series data, static elimination processing condition information indicating conditions under which the learning target substrate was neutralized in the substrate processing apparatus;
acquiring processing result information indicating a result of static electricity removal from the learning target substrate in the substrate processing apparatus from the time series data;
A method for generating data for learning, comprising the step of associating the board information, the static elimination processing condition information, and the processing result information with respect to the learning target board and storing them as learning data in a storage unit. Obtaining learning data generated according to the learning data generation method according to claim 8;
A learning method including the step of inputting the learning data into a learning program and performing machine learning on the learning data. A storage unit that stores learning data generated according to the learning data generation method according to claim 8;
A learning device, comprising: a learning section that inputs the learning data into a learning program and performs machine learning on the learning data. Obtaining learning data generated according to the learning data generation method according to claim 8;
A method for generating a trained model, the method comprising: generating a trained model constructed by subjecting the training data to machine learning. A learned model constructed by subjecting learning data generated according to the learning data generation method according to claim 8 to machine learning. a substrate holder that rotatably holds the substrate;
a charge distribution measuring section that measures the charge distribution of the substrate;
a static eliminator that removes static from the substrate;
a storage unit that stores a conversion table in which substrate information including charge distribution information and static elimination processing condition information indicating conditions for static elimination processing are associated;
a substrate information acquisition unit that acquires substrate information including charge distribution information indicating the charge distribution of the substrate;
a static elimination processing condition information acquisition unit that uses the conversion table to acquire static elimination processing condition information indicating static elimination processing conditions for the substrate based on the board information;
A substrate processing apparatus, comprising: a control section that controls the substrate holding section and the static eliminator so as to eliminate static electricity from the substrate based on the static elimination processing condition information acquired by the static elimination processing condition information acquisition section.

Images