Substrate processing system, substrate processing device, processing method, method for manufacturing semiconductor device, learning model generation system, learning model generation method, and program

The substrate processing system addresses the challenge of uniform film thickness and quality in plasma processing by using a learning model to adjust plasma generation power, resulting in enhanced film deposition quality.

WO2026234831A1PCT designated stage Publication Date: 2026-11-12KOKUSAI DENKI KK
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
PCT/JP2025/027271
Authority / Receiving Office
WO · WO
Patent Type
Applications
Current Assignee / Owner
Filing Date
2025-07-31
Publication Date
2026-11-12

AI Technical Summary

Technical Problem

Existing substrate processing technologies using plasma face challenges in achieving uniform film thickness and quality, particularly in plasma substrate processing.

Method used

A substrate processing system equipped with a processing container, gas supply unit, plasma generation unit, acquisition unit, and control unit that utilizes a learning model to correct plasma generation power based on substrate processing status, ensuring uniform film thickness and improved quality.

Benefits of technology

The system enhances the uniformity and quality of film deposition on substrates by dynamically adjusting plasma generation parameters, leading to improved substrate processing outcomes.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure JP2025027271_12112026_PF_FP_ABST
    Figure JP2025027271_12112026_PF_FP_ABST
Patent Text Reader

Abstract

Provided is a substrate processing system comprising: a processing container configured to process a substrate in at least one processing step; a gas supply unit configured to supply a processing gas to the processing container; a plasma generation unit configured to generate plasma in the processing container using power supplied from a high-frequency power source in the at least one processing step; an acquisition unit configured to acquire the processing status of the substrate; and a control unit configured to be capable of controlling the gas supply unit and the plasma generation unit and that is capable of using a learning model. The learning model indicates the relation between the processing status of the substrate and the magnitude of a reflected wave generated by the plasma generation unit. The control unit uses the learning model to acquire a correction value for correcting the value of the power supplied to the plasma generation unit based on the processing status of the substrate acquired by the acquisition unit.
Need to check novelty before this filing date? Find Prior Art

Description

Substrate processing system, substrate processing apparatus, processing method, semiconductor device manufacturing method, learning model generation system, learning model generation method, and program

[0001] This disclosure relates to a substrate processing system, a substrate processing apparatus, a processing method, a semiconductor device manufacturing method, a learning model generation system, a learning model generation method, and a program.

[0002] A technique for obtaining a uniform film thickness in substrate processing using plasma is sometimes provided (see, for example, Patent Document 1).

[0003] International Publication No. 2016 / 147296

[0004] This disclosure provides a technology that can improve the quality of substrates processed in plasma substrate processing.

[0005] According to one aspect of the present disclosure, a technology is provided comprising: a processing container for processing a substrate in at least one processing step; a gas supply unit for supplying a processing gas to the processing container; a plasma generation unit for generating plasma in the processing container using power supplied from a high-frequency power supply in the at least one processing step; an acquisition unit for acquiring the processing status of the substrate; and a control unit configured to control the gas supply unit and the plasma generation unit, and to be able to utilize a learning model, wherein the learning model shows the relationship between the processing status of the substrate and the magnitude of the reflected wave generated by the plasma generation unit, and the control unit is configured to acquire a correction value that corrects the value of the power supplied to the plasma generation unit based on the processing status of the substrate acquired by the acquisition unit using the learning model.

[0006] According to this disclosure, it is possible to improve the quality of substrates processed in substrate processing using plasma.

[0007] This is a schematic diagram of the vertical processing furnace of the substrate processing apparatus preferably used in this embodiment, showing the processing furnace portion in a vertical cross-section. This is a schematic diagram of the vertical processing furnace of the substrate processing apparatus preferably used in this embodiment, showing the processing furnace portion in a cross-sectional view along line A-A in Figure 1. This is a schematic diagram of the controller of the substrate processing apparatus preferably used in this embodiment. This is a diagram showing the overall processing flow in this embodiment. This is a diagram showing the details of the processing flow in this embodiment, illustrating in detail the steps included in the first reaction gas supply step P508 in Figure 4.

[0008] [Embodiment] An embodiment of the present disclosure will be described below with reference to the drawings. The drawings used in the following description are all schematic, and the dimensional relationships and ratios of each element shown in the drawings do not necessarily correspond to those of reality. Furthermore, the dimensional relationships and ratios of each element do not necessarily correspond between multiple drawings.

[0009] (Configuration) An embodiment of the substrate processing apparatus 100 of this disclosure will be described using Figures 1 and 2. This disclosure is not limited to the substrate processing apparatus 100 according to this embodiment, but can also be suitably applied to substrate processing apparatus 100 having a single-wafer, Hot Wall type, or Cold Wall type processing furnace. Even when using these substrate processing apparatuses, each processing can be performed using the same processing procedures and processing conditions as in the embodiment and modified examples of this disclosure, and the same effects as in the embodiment and modified examples of this disclosure can be obtained. The substrate processing apparatus 100 is an example of a "substrate processing system" in this disclosure. Furthermore, the substrate processing apparatus 100 is an example of a "learning model generation system" in this disclosure.

[0010] (Processing Chamber) As shown in Figure 1, the processing furnace 202 is equipped with reaction tubes 203 that constitute the reaction vessel (processing vessel). The reaction tubes 203 are made of, for example, quartz (SiO 2It is made of a heat-resistant material such as ) or silicon carbide (SiC) and is formed in a cylindrical shape with a closed upper end and an open lower end. A processing chamber 201 is formed in the hollow part of the reaction tube 203 and is configured to accommodate wafers 200 as substrates in a horizontal position and aligned in multiple stages vertically by a boat 217, which will be described later.

[0011] Below the reaction tube 203, a seal cap 219 is provided as a furnace opening cover capable of airtightly closing the lower end opening of the reaction tube 203. The seal cap 219 is configured to abut the lower end of the reaction tube 203 from the vertically downward side. The seal cap 219 is made of a metal such as stainless steel and is formed in a disc shape. An O-ring 220 is provided on the upper surface of the seal cap 219 as a sealing member that abuts the lower end of the reaction tube 203. On the side of the seal cap 219 opposite the processing chamber 201, a rotating mechanism 267 is installed to rotate a boat 217, which will be described later as a substrate holder. The rotation shaft 255 of the rotating mechanism 267 passes through the seal cap 219 and is connected to the boat 217. The rotating mechanism 267 is configured to rotate the wafer 200 by rotating the boat 217. The seal cap 219 is configured to be raised and lowered vertically by a boat elevator 115, which is installed vertically outside the reaction tube 203 as a lifting mechanism. The boat elevator 115 is configured to allow the boat 217 to be moved in and out of the processing chamber 201 by raising and lowering the seal cap 219.

[0012] (Substrate Holder) The boat 217, which serves as a substrate holder, is made of a heat-resistant material such as quartz or silicon carbide, and is configured to support multiple wafers 200 in a horizontal position, aligned at their centers, and in multiple stages. A heat insulating member 218 made of a heat-resistant material such as quartz or silicon carbide is provided at the bottom of the boat 217, and is configured to prevent heat from the heater 207 (described later) from being transferred to the seal cap 219.

[0013] (Heating device) The processing furnace 202 has a heater 207 as a heating device (heating mechanism). The heater 207 is cylindrical and is mounted vertically by being supported concentrically with the reaction tube 203 on a heater base (not shown) which serves as a holding plate. The heater 207 also functions as an activation mechanism that activates the gas with heat, as will be described later.

[0014] (Gas Supply Section) At the bottom of the reaction tube 203 within the processing chamber 201, a first nozzle 233a, which serves as a first gas inlet, and a second nozzle 233b, which serves as a second gas inlet, are provided so as to penetrate the lower side wall of the reaction tube 203 or the side wall surface of a manifold (not shown). The first nozzle 233a and the second nozzle 233b are connected to the first gas supply pipe 232a and the second gas supply pipe 232b, respectively. The second gas supply pipe 232b is connected to the third gas supply pipe 232c. Thus, the reaction tube 203 is equipped with two nozzles 233a and 233b and three gas supply pipes 232a, 232b, and 232c, and is configured to supply multiple types of gas, in this case three types, into the processing chamber 201.

[0015] The first gas supply pipe 232a is equipped with a mass flow controller (hereinafter referred to as MFC) 241a, which is a flow control unit (flow control section), and a valve 243a, which is an on-off valve, in order from the upstream direction. Furthermore, a first inert gas supply pipe 232d is connected to the first gas supply pipe 232a downstream of the valve 243a. This first inert gas supply pipe 232d is equipped with an MFC 241d, which is a flow control unit (flow control section), and a valve 243d, which is an on-off valve, in order from the upstream direction. Furthermore, the first nozzle 233a described above is connected to the downstream end of the first gas supply pipe 232a. The first nozzle 233a is provided in the annular space between the inner wall of the reaction tube 203 and the wafer 200, rising upward in the direction of wafer 200 stacking, along the upper part of the inner wall of the reaction tube 203 from the lower part. That is, the first nozzle 233a is provided to the side of the wafer arrangement region where the wafers 200 are arranged. The first nozzle 233a is configured as an L-shaped long nozzle. A gas supply port 248a for supplying gas is provided on the side of the first nozzle 233a. The gas supply port 248a opens so as to face the center of the reaction tube 203, making it possible to supply gas toward the wafer 200. Multiple gas supply ports 248a are provided extending from the bottom to the top of the reaction tube 203, each having the same opening area and being provided at the same opening pitch.

[0016] The first gas supply system mainly consists of the first gas supply pipe 232a, MFC 241a, and valve 243a. The first nozzle 233a may also be considered as part of the first gas supply system. Alternatively, the first inert gas supply system mainly consists of the first inert gas supply pipe 232d, MFC 241d, and valve 243d. The first inert gas supply system also functions as a purge gas supply system.

[0017] The second gas supply pipe 232b is equipped with an MFC 241b and a valve 243b, which is an on / off valve, in order from the upstream direction. Downstream of the valve 243b on the second gas supply pipe 232b, the second inert gas supply pipe 232e is connected. The second inert gas supply pipe 232e is equipped with an MFC 241e and a valve 243e, which is an on / off valve, in order from the upstream direction. The second nozzle 233b is connected to the downstream end of the second gas supply pipe 232b. The second nozzle 233b is located in the buffer chamber 237, which is a gas dispersion space.

[0018] The buffer chamber 237 is located in the annular space between the inner wall of the reaction tube 203 and the wafer 200, extending from the lower to the upper part of the inner wall of the reaction tube 203, and is provided along the wafer 200 stacking direction. In other words, the buffer chamber 237 is located to the side of the wafer arrangement region. A gas supply port 248c is provided at the end of the wall of the buffer chamber 237 adjacent to the wafer 200 for supplying gas. The gas supply port 248c opens so as to face the center of the reaction tube 203, making it possible to supply gas toward the wafer 200. Multiple gas supply ports 248c are provided extending from the lower to the upper part of the reaction tube 203, each having the same opening area and being provided at the same opening pitch.

[0019] The second nozzle 233b is located at the end of the buffer chamber 237 opposite to the end where the gas supply port 248c is provided, and is positioned to rise upward in the wafer 200 stacking direction, along the upper part of the inner wall of the reaction tube 203 from the lower part. In other words, the second nozzle 233b is located to the side of the wafer arrangement region. The second nozzle 233b is configured as an L-shaped long nozzle. A gas supply port 248b for supplying gas is provided on the side of the second nozzle 233b. The gas supply port 248b opens so as to face the center of the buffer chamber 237. Similar to the gas supply port 248c of the buffer chamber 237, multiple gas supply ports 248b are provided extending from the lower to the upper part of the reaction tube 203.

[0020] The second gas supply system is mainly composed of the second gas supply pipe 232b, MFC 241b, and valve 243b. The second nozzle 233b and buffer chamber 237 may also be considered as part of the second gas supply system. Alternatively, the second inert gas supply system is mainly composed of the second inert gas supply pipe 232e, MFC 241e, and valve 243e. The second inert gas supply system also functions as a purge gas supply system.

[0021] The third gas supply pipe 232c is equipped with an MFC 241c and a valve 243c, which is an on / off valve, in order from the upstream direction. Furthermore, a third inert gas supply pipe 232f is connected downstream of the valve 243c of the third gas supply pipe 232c and upstream of the connection point with the second gas supply pipe. This third inert gas supply pipe 232f is equipped with an MFC 241f and a valve 243f, which is an on / off valve, in order from the upstream direction. The downstream end of the third gas supply pipe 232c is connected downstream of the valve 243b of the second gas supply pipe 232b.

[0022] The third gas supply system is mainly composed of the third gas supply pipe 232c, MFC 241c, and valve 243c. Alternatively, the downstream portion of the connection between the second gas supply pipe 232b and the third gas supply pipe 232c, the second nozzle 233b, and the buffer chamber 237 may also be considered as part of the third gas supply system. Furthermore, the third inert gas supply system is mainly composed of the third inert gas supply pipe 232f, MFC 241f, and valve 243f. The third inert gas supply system also functions as a purge gas supply system.

[0023] From the first gas supply pipe 232a, a raw material gas containing a first predetermined element, such as silicon (Si) as the first predetermined element (silicon-containing gas), is supplied into the processing chamber 201 via the MFC 241a, valve 243a, and first nozzle 233a. That is, the first gas supply system is configured as a raw material gas supply system (silicon-containing gas supply system). At the same time, an inert gas may be supplied from the first inert gas supply pipe 232d into the first gas supply pipe 232a via the MFC 241d and valve 243d. For example, the silicon-containing gas may be hexachlorodisilane (Si). 2 Cl6 ) gas, monochlorosilane (SiH 3 Cl) gas, dichlorosilane (SiH 2 Cl 2 ) gas, bis(tert-butylamino)silane (SiH 2 [NH(C 4 H 9 )] 2 ) gas and the like can be used. One or more of these can be used as the silicon-containing gas. Further, as the inert gas, for example, rare gases such as helium (He) gas, neon (Ne) gas, argon (Ar) gas, krypton (Kr) gas, xenon (Xe) gas, and nitrogen (N 2 ) gas and the like can be used. One or more of these can be used as the inert gas.

[0024] From the second gas supply pipe 232b, as the first reaction gas containing the second predetermined element, for example, a source gas (nitrogen-containing gas) containing nitrogen (N) atoms as the second predetermined element is supplied into the processing chamber 201 via the MFC 241b, the valve 243b, the second nozzle 233b, and the buffer chamber 237. That is, the second gas supply system is configured as a reaction gas supply system. At this time, at the same time, the inert gas may be supplied from the second inert gas supply pipe 232e into the second gas supply pipe 232b via the MFC 241e and the valve 243e. Note that as the nitrogen-containing gas, for example, diazene (N 2 H 2 ) gas, ammonia (NH 3 ) gas, hydrazine (N 2 H 4 ) gas, N 3 H 8 gas and the like can be used. One or more of these can be used as the nitrogen-containing gas.

[0025] From the third gas supply pipe 232c, a raw material gas (oxygen-containing gas) containing oxygen (O) atoms as the third predetermined element, as a second reaction gas containing a third predetermined element, is supplied into the processing chamber 201 via the MFC 241c, valve 243c, second gas supply pipe 232b, second nozzle 233b, and buffer chamber 237. In other words, the third gas supply system is configured as an oxygen-containing gas supply system. At the same time, an inert gas may be supplied from the third inert gas supply pipe 232f into the third gas supply pipe 232c via the MFC 241f and valve 243f. Examples of oxygen-containing gases include nitric oxide (NO) gas and nitrogen dioxide (NO) gas. 2 ) gas, oxygen (O 2 ) gas, ozone (O 3 ) gases, etc. can be used. One or more of these can be used as oxygen-containing gases.

[0026] In this embodiment, the first reaction gas and the second reaction gas are supplied into the processing chamber 201 (buffer chamber 237) from the same nozzle. However, they may be supplied into the processing chamber 201 from separate nozzles and separate buffer chambers. That is, a dedicated nozzle and buffer chamber may be provided for the first reaction gas, and a dedicated nozzle and buffer chamber may be provided for the second reaction gas. Alternatively, a buffer chamber housing the first nozzle 233a may be provided.

[0027] (Plasma Generation Section) As shown in Figure 2, within the buffer chamber 237, a first rod-shaped electrode 269, which is a first electrode having an elongated structure as a plasma generation section, and a second rod-shaped electrode 270, which is a second electrode acting as a counter electrode to the first electrode, are arranged from the bottom to the top of the reaction tube 203 along the stacking direction of the wafer 200. Each of the first rod-shaped electrode 269 and the second rod-shaped electrode 270 is provided parallel to the second nozzle 233b. Each of the first rod-shaped electrode 269 and the second rod-shaped electrode 270 is protected by an electrode protection tube 275, which is a protective tube that protects each electrode from the top to the bottom. Either the first rod-shaped electrode 269 or the second rod-shaped electrode 270 is connected to a high-frequency power supply 273 via a matching unit 272, and the other is connected to ground, which is a reference potential. As a result, plasma is generated in the plasma generation region 224 between the first rod-shaped electrode 269 and the second rod-shaped electrode 270. The plasma source, which functions as a plasma generator (plasma generation unit), is mainly composed of a first rod-shaped electrode 269, a second rod-shaped electrode 270, an electrode protection tube 275, and a matching unit 272. The plasma source functions as an activation mechanism that activates gas with plasma, as will be described later. A high-frequency power supply 273 may also be included in the plasma source.

[0028] The electrode protection tube 275 is structured to allow the first rod-shaped electrode 269 and the second rod-shaped electrode 270 to be inserted into the buffer chamber 237 while being isolated from the atmosphere of the buffer chamber 237. However, if the inside of the electrode protection tube 275 is the same atmosphere as the outside air (atmosphere), the first rod-shaped electrode 269 and the second rod-shaped electrode 270 inserted into the electrode protection tube 275 will be oxidized by the heat from the heater 207. Therefore, an inert gas purging mechanism is provided inside the electrode protection tube 275 to prevent oxidation of the first rod-shaped electrode 269 or the second rod-shaped electrode 270 by filling or purging the inside of the electrode protection tube 275 with an inert gas such as nitrogen and keeping the oxygen concentration sufficiently low.

[0029] The plasma generation unit described above is not limited to using capacitively coupled plasma (CCP), but may use any of the following: inductively coupled plasma (ICP), electron cyclotron resonance plasma (ECR plasma), helicon wave excited plasma (HWP), or surface wave plasma (SWP).

[0030] (Gas Exhaust Means) The reaction tube 203 is provided with an exhaust pipe 231 for exhausting the atmosphere inside the processing chamber 201. A vacuum pump 246, which serves as a vacuum exhaust device, is connected to the exhaust pipe 231 via a pressure sensor 245, which serves as a pressure detector (pressure detection unit) for detecting the pressure inside the processing chamber 201, and an APC (Auto Pressure Controller) valve 244, which serves as a pressure regulator (pressure adjustment unit). The APC valve 244 is an on / off valve that can be opened and closed to start and stop vacuum exhaust in the processing chamber 201, and the pressure can be adjusted by adjusting the valve opening. By operating the vacuum pump 246 and adjusting the valve opening of the APC valve 244 based on the pressure information detected by the pressure sensor 245, the processing chamber 201 can be vacuum exhausted to a predetermined pressure (vacuum level). The exhaust system mainly consists of the exhaust pipe 231, the APC valve 244, and the pressure sensor 245. The vacuum pump 246 may also be included in the exhaust system.

[0031] (Control device) As shown in Figure 3, the control unit 121, which is the control unit (control device), is connected to MFCs 241a, 241b, 241c, 241d, 241e, 241f, valves 243a, 243b, 243c, 243d, 243e, 243f, pressure sensor 245, APC valve 244, vacuum pump 246, heater 207, temperature sensor 263, rotating mechanism 267, boat elevator 115, high-frequency power supply 273, matching unit 272, etc. The control unit 121 controls various operations such as adjusting the flow rate of various gases by MFCs 241a, 241b, 241c, 241d, 241e, and 241f, opening and closing valves 243a, 243b, 243c, 243d, 243e, and 243f, opening and closing of APC valve 244 and pressure adjustment operation based on pressure sensor 245, temperature adjustment operation of heater 207 based on temperature sensor 263, starting and stopping vacuum pump 246, adjusting the rotation speed of rotary mechanism 267, and raising and lowering boat 217 by boat elevator 115, as well as power supply control of high-frequency power supply 273 and impedance control by matching unit 272.

[0032] Furthermore, the control unit 121 is configured as a computer equipped with a CPU (Central Processing Unit) 121a, RAM (Random Access Memory) 121b, storage unit 121c, and I / O port 121d. The RAM 121b, storage unit 121c, prediction unit 121f, and I / O port 121d are configured to exchange data with the CPU 121a via an internal bus. An input / output device 122, configured as, for example, a touch panel, is connected to the control unit 121.

[0033] The storage unit 121c is configured by, for example, a flash memory, an HDD (Hard Disk Drive), an SSD (Solid State Drive), or the like. As shown in FIG. 3, a learning model 130 and a program 132 are stored in the storage unit 121c in a readable manner. More specifically, the program 132 includes a control program for controlling the operation of the substrate processing apparatus 100, a process recipe in which procedures, conditions and the like of substrate processing such as thin film formation to be described later are described, and the like. The process recipe is combined so as to cause the control unit 121 to execute each procedure in a substrate processing step such as a thin film formation step to be described later and to obtain a predetermined result, and functions as a program. Hereinafter, the process recipe, the control program, and the like are collectively referred to simply as a program (program product). Note that when the term "program" is used in the present specification, the case may include only the process recipe alone, the case may include only the control program alone, or the case may include both. The learning model 130 functions as a program that corrects values included in the process recipe in a substrate processing step to be described later. Further, the RAM 121b is configured as a memory area (work area) in which programs, data, and the like read by the CPU 121a are temporarily held.

[0034] The I / O port 121d is connected to the above-described MFCs 241a to 241f, valves 243a to 243f, a pressure sensor 245, an APC valve 244, a vacuum pump 246, a temperature sensor 263, a heater 207, a matching unit 272, a high-frequency power supply 273, a rotation mechanism 267, a boat elevator 115, and the like.

[0035] The CPU 121a is configured to read and execute a control program from the storage unit 121c, and to read a process recipe from the storage unit 121c in response to input of an operation command from the input / output device 122 or the like. The CPU 121a is configured to control the flow rate adjustment operation of various gases by the MFCs 241a to 241f, the opening and closing operation of the valves 243a to 243f, the opening and closing operation of the APC valve 244, the pressure adjustment operation by the APC valve 244 based on the pressure sensor 245, the start and stop of the vacuum pump 246, the temperature adjustment operation of the heater 207 based on the temperature sensor 263, the rotation and rotation speed adjustment operation of the boat 217 by the rotation mechanism 267, the elevating operation of the boat 217 by the boat elevator 115, the impedance adjustment operation by the matcher 272, the power supply of the high-frequency power source 273, and the like, so as to conform to the content of the read process recipe.

[0036] Note that when the CPU 121a executes the control operations for the various configurations described above, feedback control or sequence control is executed as an example. That is, when executing these control operations, the CPU 121a executes acquiring the status of the various configurations described above and operating the various configurations.

[0037] As an example shown in FIG. 3, the CPU 121a can acquire the status of various control operations including the operations of the MFCs 241a to 241f, various sensors including the pressure sensor 245 and the temperature sensor 263, the operation of the matcher 272, and the power supplied to the high-frequency power source 273. In other words, these statuses that can be acquired by the CPU 121a are an example of the "substrate processing status" in the present disclosure.

[0038] The inference unit 121f executes an operation of performing inference for generating the learning model 130 in a substrate processing step to be described later. The inference unit 121f may be provided as a hardware configuration separate from the CPU 121a, or may be configured by a functional configuration of the CPU 121a (that is, an operation by executing the program 132). In the present embodiment, the inference unit 121f will be described as an operation of the CPU 121a.

[0039] The control unit 121 can be configured by installing the above-mentioned program stored in an external storage device (for example, a magnetic disk such as an HDD, an optical disk such as a CD or DVD, a semiconductor memory such as a USB memory or SSD) 123 onto a computer. The storage unit 121c and the external storage device 123 are configured as computer-readable recording media. Hereinafter, these will be collectively referred to simply as recording media. In this specification, the term recording media may include only the storage unit 121c, only the external storage device 123, or both. The program (program product) may be provided to the computer using communication means such as the Internet or a dedicated line, without using the external storage device 123.

[0040] (Peripheral configuration) A temperature sensor 263 is installed inside the reaction tube 203 as a temperature detector. The system is configured to adjust the amount of power supplied to the heater 207 based on the temperature information detected by the temperature sensor 263 so that the temperature inside the processing chamber 201 reaches a desired temperature distribution. The temperature sensor 263 is configured in an L-shape, similar to the first nozzle 233a and the second nozzle 233b, and is installed along the inner wall of the reaction tube 203.

[0041] (Substrate Processing Method, Semiconductor Device Manufacturing Method) Next, an example of a method for forming a silicon nitride (SiON) film as an insulating film on a substrate as one step in the manufacturing process of a semiconductor device (device) using the processing furnace of the substrate processing apparatus 100 described above will be explained. In the following explanation, the operation of each part constituting the substrate processing apparatus 100 is controlled by the control unit 121.

[0042] As shown in Figure 4, the processing sequence of this embodiment involves a nitride film formation step P503 in which a silicon nitride (SiN) film is formed on the surface of the wafer 200, and an oxidation step P514 in which an oxidizing treatment is performed on the wafer 200 on which the SiN film has been formed to form a SiON film, in succession. This forms, for example, the gate insulating film of a MOS type field-effect transistor.

[0043] In this specification, the term "wafer" may refer to "the wafer itself" or "a laminate (assembly) of a wafer and a predetermined layer or film formed on its surface," meaning that the term "wafer" may include the predetermined layer or film formed on the surface. Furthermore, in this specification, the term "surface of the wafer" may refer to "the surface (exposed surface) of the wafer itself" or "the surface of a predetermined layer or film formed on the wafer, that is, the outermost surface of the wafer as a laminate."

[0044] Therefore, when the Specified Act states "supply a predetermined gas to a wafer," it may mean "supply a predetermined gas to the surface (exposed surface) of the wafer itself," or it may mean "supply a predetermined gas to layers, films, etc., formed on the wafer, i.e., to the outermost surface of the wafer as a laminate." Furthermore, when the Specified Act states "form a predetermined layer (or film) on a wafer," it may mean "form a predetermined layer (or film) on the surface (exposed surface) of the wafer itself," or it may mean "form a predetermined layer (or film) on layers, films, etc., formed on the wafer, i.e., on the outermost surface of the wafer as a laminate."

[0045] Furthermore, the term "substrate" used herein has the same meaning as the term "wafer" used herein.

[0046] Next, the substrate processing method in this disclosure will be described with reference to Figures 4 and 5. The substrate processing method has a plurality of steps, as shown in Figure 4. The plurality of steps shown in Figures 4 and 5 are processed, for example, by the CPU 121a reading and executing the program 132.

[0047] (Loading Process P500) First, in loading process P500, multiple wafers 200 are loaded into the boat 217 (wafer charging). Next, the boat elevator 115 is driven based on the control of the control unit 121 to raise the boat 217. As a result, as shown in Figures 1 and 2, the boat 217 holding the multiple wafers 200 is loaded into the processing chamber 201 (boat loading). At this time, the seal cap 219 closes the lower end of the reaction tube 203 via the O-ring 220. As a result, the processing chamber 201 is hermetically sealed.

[0048] (Pressure adjustment and heating process P502) Once the boat 217 has been brought into the processing chamber 201, in the pressure adjustment and heating process P502, the atmosphere inside the processing chamber 201 is exhausted so that the pressure inside the processing chamber 201 reaches a predetermined pressure (for example, 10 to 100 Pa). Specifically, while exhausting with a vacuum pump 246, the valve opening of the APC valve 244 is feedback controlled based on the pressure information detected by the pressure sensor 245 to set the pressure inside the processing chamber 201 to a predetermined level. The processing chamber 201 is also heated by a heater 207 so that it reaches a predetermined temperature. Specifically, the amount of power supplied to the heater 207 is controlled based on the temperature information detected by the temperature sensor 263 to set the temperature inside the processing chamber 201 to a predetermined level (for example, 300 to 600°C). Then, the rotation mechanism 267 is activated to start rotating the wafer 200 that has been brought into the processing chamber 201. The rotation of the wafer 200 will continue until the nitride film formation process P503 and the oxidation process P514, which will be described later, are completed.

[0049] In this specification, numerical ranges such as "10 to 100 Pa" mean that the lower and upper limits are included within that range. Therefore, for example, "10 to 100 Pa" means "10 Pa or more and 100 Pa or less." The same applies to other numerical ranges.

[0050] (Nitride film formation process P503) In nitride film formation process P503, two or more types of processing gases are alternately supplied onto the wafer one at a time under predetermined film formation conditions (temperature, time, etc.), and a SiN film is formed on the surface of the wafer 200 using the raw material gas and the first reaction gas by an alternating supply method, which is a method of forming a thin film in units of less than one atomic layer to several atomic layers by utilizing the reaction of the processing gases on the wafer 200. Film thickness control in this alternating supply method can be achieved by controlling the number of alternating supply cycles of the processing gases.

[0051] In the nitride film formation process P503, first, in the raw material gas supply process P504, raw material gas is supplied from the first nozzle 233a into the processing chamber 201 to form a silicon (Si)-containing layer with a thickness of less than one atomic layer to several atomic layers on the surface of the wafer 200. Then, in the raw material gas exhaust process P506, an inert gas is supplied as a purge gas into the processing chamber 201 to exhaust the raw material gas. Subsequently, in the first reaction gas supply process P508, the first reaction gas is plasma-excited to generate active species (activated first reaction gas), which are supplied into the processing chamber 201 to form a SiN layer on the wafer 200. Then, in the first reaction gas exhaust process P510, an inert gas is supplied as a purge gas into the processing chamber 201 to exhaust the first reaction gas. Then, the raw material gas supply process P504, the raw material gas exhaust process P506, the first reaction gas supply process P508, and the first reaction gas exhaust process P510 are considered as one cycle, and this cycle is repeated a predetermined number of times to form a SiN film of the desired thickness.

[0052] For convenience, this film deposition sequence may also be shown as follows in this specification. The same notation will be used in the following descriptions of variations and other embodiments.

[0053] (Raw material gas → First reaction gas) × n ⇒ SiN (n is an integer of 1 or more)

[0054] Furthermore, the nitride film formation process P503 will be explained in more detail.

[0055] (Raw material gas supply process P504) First, in the raw material gas supply process P504, with valve 243a closed, the supply of raw material gas is started from the first gas supply pipe 232a while adjusting the flow rate to a predetermined level using the MFC 241a.

[0056] When the pressure inside the processing chamber 201 reaches a predetermined level, the valve of the APC valve 244 is closed to temporarily stop the exhaust of gas from the processing chamber 201. Then, valve 243a is opened, and gas is supplied into the buffer chamber 237 (i.e., into the processing chamber 201). At this time, because the valve of the APC valve 244 is closed, the pressure inside the processing chamber 201 rises rapidly. The raw material gas supplied into the processing chamber 201 generates a silicon (Si)-containing layer on the outermost surface of the wafer 200, with a thickness of less than one atomic layer to several atomic layers, as the first layer. The raw material gas that did not contribute to the formation of the Si-containing layer flows down inside the processing chamber 201 and is exhausted from the exhaust pipe 231.

[0057] (Raw material gas exhaust process P506) In the raw material gas exhaust process P506, after a predetermined time has elapsed, valve 243a is closed to stop the supply of raw material gas into the buffer chamber 237 (i.e., into the processing chamber 201). At this time, APC valve 244 is left open, and the vacuum pump 246 discharges any unreacted raw material gas remaining in the processing chamber 201 or raw material gas that has contributed to the formation of the first layer from the processing chamber 201. At this time, after closing valve 243a, valve 243d is opened, and an inert gas is supplied into the processing chamber 201 as a purge gas, which enhances the effect of discharging raw material gas and reaction by-products remaining in the processing chamber 201.

[0058] At this time, it is not necessary to completely discharge the gas remaining in the processing chamber 201, nor is it necessary to completely purge the processing chamber 201. If only a small amount of gas remains in the processing chamber 201, it will not adversely affect the subsequent first reaction gas supply step P508. The flow rate of the inert gas supplied to the processing chamber 201 does not need to be large; for example, supplying an amount of inert gas roughly equal to the volume of the reaction tube 203 (processing chamber 201) allows for purging to the extent that it does not adversely affect the first reaction gas supply step P508. In this way, by not completely purging the processing chamber 201, the purging time can be shortened and throughput can be improved. The consumption of inert gas can also be kept to the absolute minimum.

[0059] (First reaction gas supply process P508) Once the raw material gas exhaust process P506 is completed, in the first reaction gas supply process P508, a rare gas is supplied to the buffer chamber 237 as an excitation gas. Specifically, the flow rate is adjusted by the MFC 241e, and the valve 243e is opened to supply the rare gas to the buffer chamber 237. Then, by supplying high-frequency power from the high-frequency power supply 273 to the first rod-shaped electrode 269 and the second rod-shaped electrode 270, a rare gas plasma is generated in the plasma generation region 224 within the buffer chamber 237, and the rare gas is activated.

[0060] Next, the first reaction gas is supplied into the buffer chamber 237. Specifically, the valve 232b is opened while adjusting the flow rate using the MFC 241b to be within the range of, for example, 1 to 10 slm, and the first reaction gas is supplied into the buffer chamber 237. The first reaction gas supplied into the buffer chamber 237 collides with the activated noble gas and is indirectly activated (indirect excitation). Alternatively, the first reaction gas may be directly activated (direct excitation), similar to the second reaction gas described later.

[0061] The details of the procedure for activating the noble gas as plasma in this embodiment will be described later.

[0062] The activated first reaction gas is ejected from the gas supply port 248c towards the wafer 200 along with the activated noble gas and supplied into the processing chamber 201.

[0063] The activated first reaction gas supplied into the processing chamber 201 reacts with at least a portion of the first layer, i.e., the Si-containing layer, formed on the surface of the wafer 200 in the raw material gas supply process P504. As a result, the first layer is plasma-nitrided and transformed (modified) into a second layer containing Si and N, i.e., a silicon nitride layer (SiN layer), and a SiN film is formed. At this time, the first reaction gas and noble gas that did not contribute to deposition on the surface of the wafer 200 flow down through the processing chamber 201 and are exhausted from the exhaust pipe 231. In this way, the noble gas functions as an excitation gas that indirectly activates the first reaction gas, and also functions as a carrier gas that promotes the supply of the activated first reaction gas into the processing chamber 201.

[0064] (First reaction gas exhaust process P510) After a predetermined time has elapsed, in the first reaction gas exhaust process P510, the supply of high-frequency power from the high-frequency power supply 273 to the first rod-shaped electrode 269 and the second rod-shaped electrode 270 is stopped. Also, the valve provided in the gas supply pipe 232b is closed to stop the supply of the first reaction gas into the buffer chamber 237. Then, using the same processing procedure as in the raw material gas exhaust process P506, any unreacted first reaction gas or reaction by-products that have contributed to the formation of the second layer remaining in the processing chamber 201 are discharged from the processing chamber 201. At this time, it is not necessary to completely discharge the gas remaining in the processing chamber 201, which is the same as in the raw material gas exhaust process P506.

[0065] (Performed a predetermined number of times) The cycle consisting of the raw material gas supply process P504, the raw material gas exhaust process P506, the first reaction gas supply process P508, and the first reaction gas exhaust process P510 is performed a predetermined number of times as defined in program 132. By performing these cycles, a SiN film of the desired thickness is formed on the wafer 200. For example, it is preferable to make the thickness of the SiN layer formed when the above cycle is performed once smaller than the desired thickness, and to repeat the above cycle multiple times until the thickness of the SiN film formed by stacking the SiN layers reaches the desired thickness.

[0066] Here, the procedure for activating the noble gas as plasma in this embodiment will be described in more detail with reference to Figure 5. Note that the words "calculate," "correct," and "set" defined in steps S600 to S624 in Figure 5 do not include execution. That is, the operations based on these words are performed in later steps.

[0067] First, in step S600, the CPU 121a acquires the current processing status of the substrate. Specifically, the CPU 121a acquires various conditions in the substrate processing process, such as the size of the substrate, the type of substrate, the amount of inert gas introduced, the amount of the first reaction gas introduced, the number of times the nitride film formation process P503 is executed, and the temperature and pressure inside the processing furnace 202. In addition, the CPU 121a may acquire other information as part of the "substrate processing status," such as the processing time, the power value supplied from the high-frequency power supply 273, and the number of product substrates. Furthermore, the CPU 121a may acquire other values ​​such as the time until impedance matching, the value of the traveling wave or reflected wave, the frequency of the high frequency when impedance matching is performed, the conductance value when impedance matching is performed, and the input and output impedance of the matching unit 272. In other words, step S600 is an example of an "acquisition process" in this disclosure. After that, the CPU 121a proceeds to step S602.

[0068] In step S602, the CPU 121a acquires the learning model 130. More specifically, the CPU 121a reads the learning model 130 from the memory unit 121c and makes it executable by temporarily storing it in the RAM 121b. If the learning model 130 is not stored in the memory unit 121c at the start of the first reaction gas supply step P508, step S602 is omitted. After that, the CPU 121a proceeds to step S604.

[0069] In step S604, the CPU 121a calculates a correction value based on the learning model 130. More specifically, the CPU 121a inputs various values ​​related to the processing status of the substrate acquired in step S600 into the learning model 130 acquired in step S602, and calculates a correction value using the learning model 130. If the learning model 130 is not stored in the storage unit 121c at the start of the first reaction gas supply step P508, step S604 is omitted. After that, the CPU 121a proceeds to step S606.

[0070] In step S606, the CPU 121a corrects the power output of the high-frequency power supply 273 to generate plasma. More specifically, the CPU 121a corrects the power output of the high-frequency power supply 273 to generate plasma in the substrate processing step, as included in program 132, based on the correction value calculated in step S604. That is, the CPU 121a corrects the power output of the high-frequency power supply 273 to a value different from the power output of the high-frequency power supply 273 as defined in the recipe included in program 132.

[0071] In this process, "correcting" also includes cases where the correction value calculated in step S604 is substantially equal to the power value output by the high-frequency power supply 273 as defined in the recipe included in program 132. If the learning model 130 is not stored in the memory unit 121c at the start of the first reaction gas supply step P508, step S606 is omitted. After that, the CPU 121a proceeds to step S608.

[0072] In step S608, the CPU 121a determines whether the corrected power value exceeds a predetermined threshold. More specifically, in step S606, the CPU 121a determines whether the corrected power value exceeds the upper limit (i.e., the limited value) of the power output by the high-frequency power supply 273, as included in program 132. The "predetermined threshold" in this process is set as needed, based on the density of the plasma generated to process the substrate, the specifications of the high-frequency power supply 273, etc.

[0073] If the CPU 121a determines that step S608 is positive, it proceeds to step S610. If the CPU 121a determines that step S608 is negative, it proceeds to step S614. If the learning model 130 is not stored in the memory unit 121c at the start of the first reaction gas supply step P508 (i.e., step S606 is omitted), the CPU 121a determines that it is negative.

[0074] In step S610, the CPU 121a readjusts the corrected power value so that it is below a predetermined threshold. In other words, in step S606, if the corrected power value exceeds a predetermined threshold, the CPU 121a readjusts the power value so that it is below the predetermined threshold. In other words, the CPU 121a determines a correction value so that the power supplied by the high-frequency power supply 273 to the first rod-shaped electrode 269 and the second rod-shaped electrode 270 is below a predetermined threshold. After that, the CPU 121a proceeds to step S612.

[0075] In step S612, the CPU 121a sets an increase in the time during which power is supplied to the first rod-shaped electrode 269 and the second rod-shaped electrode 270. More specifically, the CPU 121a sets the high-frequency power supply 273 to extend the time during which power is supplied to the first rod-shaped electrode 269 and the second rod-shaped electrode 270 and plasma is generated. The specific value of this increased time is set as appropriate. For example, it is preferable that the CPU 121a determines this value in step S606 based on the magnitude (e.g., difference or ratio) of the corrected power value and the power value reset in step S610. After that, the CPU 121a proceeds to step S614.

[0076] In step S614, the CPU 121a generates plasma. More specifically, the CPU 121a controls the high-frequency power supply 273 to supply power to the first rod-shaped electrode 269 and the second rod-shaped electrode 270, thereby generating a rare gas plasma. In other words, step S614 is an example of the "plasma generation process" in this disclosure.

[0077] The plasma generated in step S614 is used to activate the first reaction gas, as described above, and is then used to process the wafer 200. The CPU 121a then proceeds to step S616.

[0078] In step S616, the CPU 121a measures the magnitude of the reflected wave. More specifically, the CPU 121a measures the power reflected from the first rod-shaped electrode 269 and the second rod-shaped electrode 270 out of the power supplied from the high-frequency power supply 273. In other words, the CPU 121a measures the magnitude of the power supplied from the high-frequency power supply 273 that was not used to generate plasma due to impedance mismatch or the like. Note that any specific method for measuring the reflected wave is acceptable. Then the CPU 121a proceeds to step S618.

[0079] Then, in step S618, the CPU 121a stops the power supply. More specifically, in step S618, the CPU 121a controls the high-frequency power supply 273 to stop supplying power to the first rod-shaped electrode 269 and the second rod-shaped electrode 270. Then the CPU 121a proceeds to step S620.

[0080] In step S620, the CPU 121a estimates the relationship between the current state and the magnitude of the reflected wave. More specifically, the CPU 121a operates as an estimation unit 121f and estimates the relationship between the current state and the magnitude of the reflected wave based on the current processing status of the substrate acquired in step S600 and the magnitude of the reflected wave measured in step S616. Any machine learning method, especially reinforcement learning, can be used for the estimation procedure. In other words, step S620 is an example of the "estimation process" in this disclosure. After creating the estimation result, the CPU 121a proceeds to step S622.

[0081] In step S622, the CPU 121a creates a learning model 130 based on the inference results. More specifically, the CPU 121a creates the learning model 130 based on the relationship between the current state and the magnitude of the reflected wave, which was inferred in step S620. In other words, step S622 is an example of a "generation process" in this disclosure. After creating the learning model 130, the CPU 121a proceeds to step S624.

[0082] In step S624, the CPU 121a records the learning model 130. More specifically, the CPU 121a stores the learning model 130 created in step S622 in the memory unit as shown in Figure 3. In other words, if the learning model 130 is stored in the memory unit, the CPU 121a modifies the learning model 130. After storing the learning model 130 in the memory unit, the CPU 121a completes the procedure for activating a series of noble gases as plasma.

[0083] The procedure for activating the noble gas as plasma, as shown in Figure 5, is performed as part of each step included in the first reaction gas supply step P508. That is, if the first reaction gas exhaust step P510 is performed multiple times as the cycle is executed multiple times, the CPU 121a executes the series of steps shown in Figure 5 in each cycle. In other words, by executing the cycle, the CPU 121a modifies the learning model 130 if it is already stored in the memory unit 121c. That is, the procedure for activating the noble gas as plasma, as shown in Figure 5, is an example of a "learning model generation method" in this disclosure.

[0084] Examples of processing conditions when performing the nitride film formation process P503 include: processing temperature (wafer temperature): 200 to 600°C, processing pressure (pressure inside the processing chamber): 10 to 100 Pa, raw material gas supply flow rate: 1 to 2000 sccm, first reaction gas supply flow rate: 100 to 10000 sccm, and rare gas supply flow rate: 100 to 10000 sccm. By setting each processing condition to a value within its respective range, it is possible to properly carry out the film formation process. In this specification, processing temperature refers to the temperature of the wafer 200 or the temperature inside the processing chamber 201, and processing pressure refers to the pressure inside the processing chamber 201. Also, processing time refers to the time during which the process is continued. These terms are also used in the following explanation.

[0085] (Oxidation process P514) After the CPU 121a has formed a SiN film of the desired thickness in the nitride film formation process P503, in the oxidation process P514, the CPU 121a oxidizes the SiN film by supplying an activated second reaction gas activated by plasma onto the wafer 200 to form a silicon oxynitride (SiON) film.

[0086] First, the second reactive gas is supplied into the buffer chamber 237. Specifically, the valve 243c is opened while adjusting the flow rate using the MFC 241c, and the second reactive gas is supplied into the buffer chamber 237. Then, by supplying high-frequency power from the high-frequency power supply 273 to the first rod-shaped electrode 269 and the second rod-shaped electrode 270, a second reactive gas plasma is generated in the buffer chamber 237, and the second reactive gas is directly activated (directly excited). Alternatively, the second reactive gas may be indirectly activated (indirectly excited), similar to the first reactive gas.

[0087] The activated second reaction gas is ejected from the gas supply port 248c of the buffer chamber 237 towards the wafer 200 along with the inert gas and supplied into the processing chamber 201. By adjusting the flow rate with the MFC 241f and opening the valve 243f, the supply of the inert gas as a carrier gas into the buffer chamber 237 (i.e., into the processing chamber 201) can be accelerated to promote the supply of the activated second reaction gas into the processing chamber 201.

[0088] The activated second reaction gas supplied into the processing chamber 201 reacts with at least a portion of the SiN film. As a result, the SiN film is plasma-oxidized and transformed (modified) into a film containing Si, O, and N, i.e., an SiON film. This results in the formation of an SiON film on the wafer 200. After a predetermined time has elapsed, the supply of high-frequency power from the high-frequency power supply 273 to the first rod-shaped electrode 269 and the second rod-shaped electrode 270 is stopped. The valve provided in the gas supply pipe 232c is also closed to stop the supply of the second reaction gas into the buffer chamber 237. Then, using the same processing procedure as in the raw material gas exhaust process P506, any unreacted or second reaction gas that has contributed to the formation of the SiON film, as well as reaction by-products, remaining in the processing chamber 201 are discharged from the processing chamber 201. At this time, it is not necessary to completely discharge the gas remaining in the processing chamber 201, and it is also possible to supply purge gas into the processing chamber, similar to the raw material gas exhaust process P506.

[0089] The following sequence represents the flow of the SiON film formed by the nitride film formation step P503 and oxidation step P514 described above.

[0090] (Raw material gas → First reaction gas) × n → Second reaction gas ⇒ SiON (n is an integer of 1 or more)

[0091] For example, the processing conditions for the oxidation process P514 are: processing temperature (wafer temperature): 200 to 600°C, processing pressure (pressure inside the processing chamber): 10 to 100 Pa, second reaction gas supply flow rate: 100 to 10000 sccm, inert gas supply flow rate: 100 to 10000 sccm.

[0092] (Atmospheric pressure return process and cooling process P516) Once the oxidation process P514 is complete, the CPU 121a performs the atmospheric pressure return process and cooling process P516 by stopping the rotation of the boat 217 and stopping the rotation of the wafer 200. Then, while returning the pressure in the processing chamber 201 to atmospheric pressure, the wafer 200 is cooled. Specifically, while keeping the valve 243e open and supplying inert gas into the processing chamber 201, the valve opening of the APC valve 244 is feedback controlled based on the pressure information detected by the pressure sensor 245 to raise the pressure in the processing chamber 201 to atmospheric pressure. Then, the amount of current supplied to the heater 207 is controlled to lower the temperature of the wafer 200.

[0093] (Removal process P518) Subsequently, as the removal process P518, the CPU 121a removes the wafer 200 that has been processed from the processing chamber 201 in the reverse order of the loading process P500 described above, and completes the substrate processing process according to this embodiment.

[0094] According to this embodiment, one or more of the following effects can be obtained.

[0095] (Operation and Effects) In the substrate processing apparatus 100 according to this embodiment, the CPU corrects the values ​​of the power supplied to the first rod-shaped electrode 269 and the second rod-shaped electrode 270 using a learning model 130. The learning model 130 is generated based on the processing status of the substrate and the magnitude of the reflected waves of the power supplied to the first rod-shaped electrode 269 and the second rod-shaped electrode 270. Therefore, with this substrate processing apparatus 100, by not using a portion of the power input to the first rod-shaped electrode 269 and the second rod-shaped electrode 270 for plasma generation, it is possible to suppress variations in the amount of plasma generated that are greater than expected. In other words, with this substrate processing apparatus 100, the amount of plasma generated is more likely to be homogenized in each cycle, thus improving the quality of the processed substrate.

[0096] Furthermore, in the substrate processing apparatus 100 according to this embodiment, the CPU corrects the power values ​​input to the first rod-shaped electrode 269 and the second rod-shaped electrode 270 for each cycle that is executed multiple times. Therefore, with the substrate processing apparatus 100 according to this embodiment, the amount of plasma generated is more likely to be homogenized in each cycle compared to when the same correction value is applied to each cycle that is executed multiple times, thus improving the quality of the processed substrate.

[0097] Furthermore, in the substrate processing apparatus 100 according to this embodiment, if the power supplied to the first rod-shaped electrode 269 and the second rod-shaped electrode 270 exceeds a predetermined threshold, the CPU corrects the power value so that the power value becomes less than or equal to the threshold. Therefore, the substrate processing apparatus 100 according to this embodiment can suppress the unintentional oversupply of power to the first rod-shaped electrode 269 and the second rod-shaped electrode 270, which would result in a higher-than-expected amount of generated plasma, thereby suppressing deterioration of film characteristics due to plasma damage.

[0098] Furthermore, in the substrate processing apparatus 100 according to this embodiment, the CPU extends the processing time by extending the time for plasma generation if the power supplied to the first rod-shaped electrode 269 and the second rod-shaped electrode 270 exceeds a predetermined threshold. Therefore, the substrate processing apparatus 100 according to this embodiment can prevent the processing of the substrate from not progressing due to inadvertently supplying too little power to the first rod-shaped electrode 269 and the second rod-shaped electrode 270, resulting in insufficient plasma generation.

[0099] Furthermore, in the substrate processing apparatus 100 according to this embodiment, the learning model 130 is modified in each cycle based on the processing status of the substrate and the magnitude of the reflected waves of the power supplied to the first rod-shaped electrode 269 and the second rod-shaped electrode 270. Therefore, according to the substrate processing apparatus 100 of this embodiment, even if the relationship between the processing status of the substrate and the magnitude of the reflected waves of the power supplied to the first rod-shaped electrode 269 and the second rod-shaped electrode 270 changes due to changes in the configuration used for processing the substrate over time, the quality of the processed substrate can be improved.

[0100] Furthermore, in the substrate processing apparatus 100 according to this embodiment, a learning model 130 is created based on the processing status of the substrate and the magnitude of the reflected waves of the power supplied to the first rod-shaped electrode 269 and the second rod-shaped electrode 270. Therefore, according to the substrate processing apparatus 100 in this embodiment, a learning model 130 can be provided that suppresses changes in the amount of plasma generated that are greater than expected, because a portion of the power input to the first rod-shaped electrode 269 and the second rod-shaped electrode 270 is not used for plasma generation. For example, by providing the learning model 130 generated by this substrate processing apparatus 100 to another substrate processing apparatus 100 of the same type, the amount of plasma generated in that other substrate processing apparatus 100 can be made more uniform in each cycle, thereby improving the quality of the processed substrate.

[0101] (First Modification) In the above description, the substrate processing apparatus 100 stored the learning model 130 in a memory unit. The technology relating to this disclosure is not limited thereto. For example, in step S602, the substrate processing apparatus 100 may download the learning model 130 from another device that can communicate with the substrate processing apparatus 100. Similarly, in step S624, the substrate processing apparatus 100 may upload the learning model 130 to another device that can communicate with the substrate processing apparatus 100. In other words, the specific recording location is not limited as long as the learning model 130 can be used by an external device. The substrate processing apparatus 100 in this modification is another example of the "substrate processing system" and another example of the "learning model generation system" in this disclosure. Furthermore, the procedure performed by the substrate processing apparatus 100 in this modification to activate a noble gas as plasma is another example of the "learning model generation method" in this disclosure.

[0102] (Second Modification) In the above description, the substrate processing apparatus 100 performed both the creation of the learning model 130 and the use of the learning model 130. The technology relating to this disclosure is not limited thereto. For example, the substrate processing apparatus 100 may be configured to perform only the generation of the learning model 130. If the substrate processing apparatus 100 operates without performing the steps S604 and S606 in the above description, it will only perform the generation of the learning model 130. In other words, the substrate processing apparatus 100 in this modification is another example of the "learning model generation system" in this disclosure. Furthermore, the procedure performed by the substrate processing apparatus 100 in this modification to activate a noble gas as plasma is another example of the "learning model generation method" in this disclosure.

[0103] (Third Modification) Alternatively, for example, the substrate processing apparatus 100 may be configured to only utilize the learning model 130. If the substrate processing apparatus 100 operates without performing steps S616, S620, S622, and S624 as described above, it will only utilize the learning model 130. In other words, the substrate processing apparatus 100 in this modification is another example of the "substrate processing system" in this disclosure.

[0104] (Other Variations) In the above description, the learning model 130 was acquired in each cycle, but the technology relating to this disclosure is not limited to this. For example, the substrate processing apparatus 100 may acquire the learning model 130 in step S602 of the first cycle of a cycle that is executed multiple times, and not acquire it in subsequent cycles. In other words, the substrate processing apparatus 100 may not perform modifications to the learning model 130 in each cycle.

[0105] Furthermore, in the above description, the substrate processing apparatus 100 corrected the power value in all of the multiple cycles that were executed. However, the technology relating to this disclosure is not limited to this. That is, the power value may be corrected in only a portion of the multiple cycles that were executed.

[0106] Furthermore, in the above description, the substrate processing apparatus 100 corrected the power value in a process in which multiple cycles were executed. However, the technology of this disclosure is not limited to this. The technology of this disclosure is also applicable in cases where the processing steps of the substrate processing process are executed only once (in other words, in cases without cycles), and also in cases where the power value is corrected. That is, in the technology of this disclosure, correcting the power value is applicable to systems that generate plasma and process a substrate in at least one processing step. Also, in the technology of this disclosure, creating a learning model is applicable to systems that generate plasma and process a substrate in at least one processing step. In these systems, the specific number of processing steps and the number of cycles included in the processing steps are not limited.

[0107] Furthermore, in the above description, the substrate processing apparatus 100 included the amount of noble gas introduced and the amount of the first reaction gas introduced in step S600 as the processing status of the substrate. The technology relating to this disclosure is not limited thereto. For example, in step S600, the "processing status of the substrate" may not include the conditions related to the gas supply process, such as the amount of noble gas introduced and the amount of the first reaction gas introduced.

[0108] Furthermore, in the above description, the substrate processing apparatus 100 determined in step S608 whether the corrected power value exceeded a predetermined threshold. The technology relating to this disclosure is not limited thereto. That is, the procedure in step S608 may be omitted, and no limit may be placed on the corrected power value. Alternatively, for example, the procedure in step S612 may be omitted, and plasma may be generated based on the time set in the recipe.

[0109] Furthermore, in this embodiment, for example, a buffer chamber for generating plasma is provided inside the reaction tube, and a first rod-shaped electrode 269, which is the first electrode, and a second rod-shaped electrode 270, which is the second electrode, are provided inside the buffer chamber. However, the invention is not limited to this configuration, and a buffer chamber 237 may be provided on the side wall inside the reaction tube so as to protrude toward the heater 207, and only a gas supply nozzle may be provided inside the buffer chamber, with the first electrode and the second electrode, which are the plasma generation parts, provided on the outside of the reaction tube so as to sandwich the buffer chamber from the outside.

[0110] Furthermore, in the embodiments described above, an example of forming a silicon-based insulating film such as a SiN film was explained. However, this disclosure is not limited to such embodiments. For example, this disclosure is also suitably applicable when forming a film containing metal elements such as titanium (Ti), zirconium (Zr), hafnium (Hf), tantalum (Ta), niobium (Nb), aluminum (Al), molybdenum (Mo), and tungsten (W), i.e., a metallic film, on a wafer 200.

[0111] The order in which each gas is flowed can be changed as appropriate. Even when performing these film formations, the film formation can be carried out under the same processing conditions as in the above-described embodiment, and the same effects as in the above-described embodiment can be obtained.

[0112] In other words, this disclosure can be suitably applied when forming a film containing a predetermined element such as a semiconductor element or a metallic element.

[0113] Furthermore, the above-described embodiment illustrates an example of forming a film on a substrate. However, this disclosure is not limited to such embodiments. For example, this disclosure is suitably applicable to etching processes, annealing processes, and the like. It is also suitably applicable to film modification processes.

[0114] While embodiments of this disclosure have been described above with reference to the attached drawings, it is clear that any person with ordinary skill in the art to which this disclosure belongs could conceive of various modifications or applications within the scope of the technical idea described in the claims, and these too are naturally understood to fall within the technical scope of this disclosure.

Claims

1. A substrate processing system comprising: a processing container for processing a substrate in at least one processing step; a gas supply unit for supplying a processing gas to the processing container; a plasma generation unit for generating plasma in the processing container using power supplied from a high-frequency power supply in the at least one processing step; an acquisition unit for acquiring the processing status of the substrate; and a control unit configured to control the gas supply unit and the plasma generation unit, and to be able to utilize a learning model, wherein the learning model shows the relationship between the processing status of the substrate and the magnitude of the reflected wave generated by the plasma generation unit, and the control unit is configured to acquire a correction value that corrects the value of the power supplied to the plasma generation unit based on the processing status of the substrate acquired by the acquisition unit using the learning model.

2. The substrate processing system according to claim 1, wherein the processing container processes the substrate by repeating the processing step of at least one process as one cycle multiple times.

3. The substrate processing system according to claim 2, wherein the control unit determines an amount to correct the power value for each processing step in the cycle.

4. The substrate processing system according to any one of claims 1 to 3, wherein the control unit determines an amount to correct the value of the power supplied to the plasma generation unit so that the value of the power supplied to the plasma generation unit is less than a predetermined value.

5. The substrate processing system according to claim 4, wherein the control unit extends the processing time when it determines that the power value exceeds the predetermined value.

6. The substrate processing system according to claim 1, wherein the control unit modifies the learning model based on the results of the substrate processing status acquired by the acquisition unit and the magnitude of the reflected wave generated when power is supplied to the plasma generation unit.

7. A substrate processing apparatus comprising: a processing container for processing a substrate in at least one processing step; a gas supply unit for supplying a processing gas to the processing container; a plasma generation unit for generating plasma in the processing container using power supplied from a high-frequency power supply in at least one processing step; an acquisition unit for acquiring the processing status of the substrate; and a control unit configured to control the gas supply unit and the plasma generation unit, and to be able to utilize a learning model, wherein the learning model shows the relationship between the processing status of the substrate and the magnitude of the reflected wave generated by the plasma generation unit, and the control unit is configured to acquire a correction value for correcting the value of the power supplied to the plasma generation unit based on the processing status of the substrate acquired by the acquisition unit using the learning model.

8. The substrate processing apparatus according to claim 7, wherein the processing vessel processes the substrate by repeating the processing step of at least one process as one cycle for a plurality of cycles.

9. A processing method comprising: a plasma generation step of generating plasma using power supplied from a high-frequency power supply; a substrate processing step of processing a substrate with the plasma generated in the plasma generation step in at least one processing step; and an acquisition step of acquiring the processing status of the substrate, wherein the plasma generation step is performed in at least one processing step, and the plasma generation step includes a step of acquiring a correction value that corrects the value of the power supplied in the plasma generation step based on the processing status of the substrate acquired in the acquisition step, using a learning model that shows the relationship between the processing status of the substrate and the magnitude of the reflected wave generated in the plasma generation step.

10. The processing method according to claim 9, wherein in the substrate processing step, the substrate is processed by repeating the one processing step as one cycle multiple times.

11. The processing method according to claim 10, wherein an amount is determined to correct the value of the power for each processing step within the cycle.

12. The processing method according to claim 9 or claim 10, further comprising a gas supply step of supplying a processing gas to the substrate, which is included in the processing status of the substrate.

13. A processing method comprising: a plasma generation step of generating plasma using power supplied from a high-frequency power supply; a substrate processing step of processing a substrate with the plasma generated in the plasma generation step in at least one processing step; and an acquisition step of acquiring the processing status of the substrate, wherein the plasma generation step is performed in at least one processing step, and the plasma generation step includes a step of acquiring a correction value that corrects the value of the power supplied in the plasma generation step based on the processing status of the substrate acquired in the acquisition step, using a learning model that shows the relationship between the processing status of the substrate and the magnitude of the reflected wave generated in the plasma generation step.

14. The method for manufacturing a semiconductor device according to claim 13, wherein in the substrate processing step, the substrate is processed by repeating the processing step as one cycle multiple times.

15. A program that causes a computer to execute a substrate processing device a plasma generation procedure for generating plasma using power supplied from a high-frequency power supply, a substrate processing procedure for processing a substrate with the plasma generated in the plasma generation procedure in at least one processing step, and an acquisition procedure for acquiring the processing status of the substrate, wherein the plasma generation procedure is performed in at least one processing step, and the plasma generation procedure includes a step to acquire a correction value that corrects the value of the power supplied in the plasma generation procedure based on the processing status of the substrate acquired in the acquisition procedure, using a learning model that shows the relationship between the processing status of the substrate and the magnitude of the reflected wave generated in the plasma generation procedure.

16. The program according to claim 15, wherein the substrate processing procedure involves processing the substrate by repeating the processing step multiple times, with each processing step considered as one cycle.

17. A learning model generation system comprising: a processing container for processing a substrate in at least one processing step; a gas supply unit for supplying a processing gas to the processing container; a plasma generation unit for generating plasma in the processing container using power supplied from a high-frequency power supply in at least one processing step; and an estimation unit for estimating the relationship between the processing status of the substrate and the magnitude of the reflected wave generated in the plasma generation unit when power is supplied to the plasma generation unit.

18. The learning model generation system according to claim 17, wherein the estimation unit estimates the situation in which the processing gas has been supplied to the substrate, including this in the processing status of the substrate.

19. A method for generating a learning model, comprising: a plasma generation step of generating plasma in a plasma generation section within a processing vessel that processes a substrate in at least one processing step using power supplied from a high-frequency power supply in the at least one processing step; an acquisition step of acquiring the processing status of the substrate and the magnitude of the reflected wave generated in the plasma generation section when power is supplied in the plasma generation step; an estimation step of estimating the relationship between the processing status of the substrate acquired in the acquisition step and the magnitude of the reflected wave generated in the plasma generation section; and a generation step of generating a learning model based on the results estimated in the estimation step.

20. A program that causes a computer to execute the following steps: a plasma generation procedure in which a plasma is generated in a plasma generation section within a processing vessel that processes a substrate in at least one processing step, using power supplied from a high-frequency power supply in the at least one processing step; an acquisition procedure for acquiring the processing status of the substrate and the magnitude of the reflected wave generated in the plasma generation section when power is supplied in the plasma generation procedure; an estimation procedure for estimating the relationship between the processing status of the substrate acquired in the acquisition procedure and the magnitude of the reflected wave generated in the plasma generation section; and a generation procedure for generating a learning model based on the results estimated in the estimation procedure.