Temperature correction information calculation device, semiconductor manufacturing device, program, and temperature correction information calculation method

Abstract

To provide a technique for easily adjusting a set temperature for heat treatment of an object to be processed.SOLUTION: The present disclosure is a temperature correction information calculation device for a semiconductor manufacturing apparatus for forming a film by heat treatment on an object to be processed at a set temperature corrected according to the cumulative film thickness accumulated on the inner wall of the semiconductor manufacturing apparatus. The temperature correction information calculation device includes: a model storage unit for storing a model for generating temperature correction information in which a temperature correction value is associated with the cumulative film thickness; a learning determination unit for determining whether or not to update the model when a film formation result by heat treatment is obtained; a model learning unit for updating the model on the basis of the film formation result on the object to be processed when the learning determination unit determines to update the model; and a temperature correction information generating unit for generating the temperature correction information using the model that is updated by the model learning unit and correcting the set temperature on the basis of the temperature correction information.SELECTED DRAWING: Figure 17

Description

[Technical Field]

[0001] The present disclosure relates to a temperature correction information calculation device, a semiconductor manufacturing device, a program, and a temperature correction information calculation method. [Background technology]

[0002] In semiconductor manufacturing processes, for example, thermal processing systems are used to perform film deposition processes on semiconductor wafers. In thermal processing systems, the process conditions controlled by the semiconductor manufacturing equipment, such as the set temperature, pressure, and gas flow rate, are determined by a process recipe corresponding to the process. When the semiconductor manufacturing equipment repeatedly performs thermal processing, a film is formed on the semiconductor wafer, but deposits adhere to the inner wall surfaces of the semiconductor manufacturing equipment. If the cumulative thickness of this deposit increases, the furnace temperature drops, making it impossible to deposit a film of the desired thickness on the semiconductor wafer, even if the semiconductor manufacturing equipment is controlled at the set temperature according to the process recipe.

[0003] Therefore, a heat treatment system that corrects the set temperature in accordance with the accumulated film thickness is known (see, for example, Patent Document 1). Patent Document 1 discloses a technique for generating temperature correction information in which the accumulated film thickness is associated with the correction amount of the set temperature. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] Japanese Patent Application Laid-Open No. 2008-218558 Summary of the Invention [Problem to be solved by the invention]

[0005] The present disclosure provides a technique that can easily adjust the set temperature for heat treatment of an object to be treated. [Means for solving the problem]

[0006] The present disclosure provides a temperature correction information calculation device for a semiconductor manufacturing device that forms a film on a workpiece by heat treatment at a set temperature corrected according to a cumulative film thickness accumulated on an inner wall of the semiconductor manufacturing device, the device comprising: a model storage unit that stores a model for generating temperature correction information in which a temperature correction value corresponds to the cumulative film thickness; a learning and determination unit that determines whether or not to update the model when a film formation result from the heat treatment is obtained; a model learning unit that updates the model based on the film formation result on the workpiece when the learning and determination unit determines to update the model; and a temperature correction information generation unit that generates the temperature correction information using the model updated by the model learning unit and corrects the set temperature using the temperature correction information, the model including a first table that defines, for each zone in which an individual heater is arranged, a fluctuation amount of the temperature of the workpiece when the set temperature differs by 1°C; The set temperature is corrected based on the first table. a second table that defines the amount of change in film thickness for each zone when the temperature differs by 1°C, and when the learning and determining unit determines that the model should be updated, the model learning unit increases the value in the first table as the difference between the monitored film thickness, which is the film formation result, and the target film thickness increases. setting The amount of temperature fluctuation of the object to be treated when the temperature differs by 1°C is increased, and in the second table setting Increase the amount of film thickness variation when the temperature differs by 1°C. [Effects of the Invention]

[0007] A technique can be provided that can easily adjust the set temperature for heat treatment of an object to be treated. [Brief explanation of the drawings]

[0008] [Figure 1] An example of a schematic diagram of a heat treatment system [Figure 2] An example of a schematic cross-sectional view of semiconductor manufacturing equipment [Figure 3] An example of a divided zone [Figure 4] An example of the control unit configuration [Figure 5] An example of a hardware configuration diagram of a temperature correction information calculation device [Figure 6] 1 is a functional block diagram illustrating an example of a functional configuration of a temperature correction information calculation device, divided into blocks. [Figure 7] An example of a temperature correction table [Figure 8] An example of a diagram for schematically explaining a model stored in a model storage unit [Figure 9] An example of the film formation results when the temperature correction table is used to perform temperature correction according to the cumulative film thickness. [Figure 10] FIG. 1 is an example of a flowchart illustrating a method for generating a temperature correction table. [Figure 11] An example of a diagram explaining cumulative film thickness data and the monitored film thickness formed on a semiconductor wafer [Figure 12] An example of a diagram explaining the relationship between the model and the temperature correction table [Figure 13] An example of a diagram that explains when the calculated model and temperature correction table are not appropriate [Figure 14] An example of a diagram illustrating the unimproved film formation results [Figure 15] An example diagram explaining how to determine whether the learning function is on or off [Figure 16] An example of a flowchart showing the procedure for switching on and off the learning function of the model by the learning determination unit. [Figure 17] An example of a flowchart illustrating a method for determining whether to turn on or off the learning function of a model by the learning determination unit. [Figure 18] An example diagram listing the specific details of when to turn on or off the model's learning function [Figure 19] An example diagram illustrating the potential benefits of automatically switching the model learning function on and off DETAILED DESCRIPTION OF THE INVENTION

[0009] Hereinafter, non-limiting exemplary embodiments of the present disclosure will be described with reference to the accompanying drawings. As an example of an embodiment for carrying out the present disclosure, a heat treatment system and a temperature correction information calculation method performed by the heat treatment system will be described with reference to the drawings.

[0010] [Outline of temperature correction for cumulative film thickness] The thermal processing system uses a temperature correction table to correct the set temperature set in the process recipe. The temperature correction table is generated by a model, which will be described later. The thermal processing system has the function of learning the film formation results and updating the model. However, it was not easy for the user to decide whether to turn the model learning function on or off.

[0011] The heat treatment system of this embodiment can automatically switch the model learning function on and off, taking into consideration the presence or absence of disturbance and the accuracy of the model, as will be described in detail below.

[0012] [Overall configuration of a heat treatment system including semiconductor manufacturing equipment] Fig. 1 shows an example of a schematic configuration diagram of a heat treatment system. As shown in Fig. 1, heat treatment system 1 of this embodiment includes a plurality of semiconductor manufacturing apparatuses 2 (21 to 2n in Fig. 1), a host computer 3, a temperature correction information calculation apparatus 4, and networks 5 and 6 interconnecting these. Heat treatment system 1 also includes, for example, a measurement apparatus 60 that measures the state (such as the state of a film) of a target object (hereinafter referred to as a semiconductor wafer) processed by semiconductor manufacturing apparatus 2.

[0013] The semiconductor manufacturing equipment 2 may include various types of equipment depending on the process. For example, there are film formation equipment that performs a process of forming a thin film on a semiconductor wafer, an oxidation equipment that performs an oxidation process of oxidizing the surface region of a semiconductor wafer, and a diffusion equipment that performs a process of diffusing (doping) impurities into the surface region of a semiconductor wafer. Below, as an example of a semiconductor manufacturing equipment, a batch-type vertical heat treatment equipment shown in FIG. 2 will be described. In addition, in this embodiment, a film formation process will be described as an example of processing on a semiconductor wafer.

[0014] Fig. 2 is an example of a schematic cross-sectional view of a semiconductor manufacturing apparatus 2. As shown in Fig. 2, the semiconductor manufacturing apparatus 2 includes a substantially cylindrical reaction tube 11. The reaction tube 11 is arranged so that its longitudinal direction is oriented vertically. The reaction tube 11 is made of a material having excellent heat resistance and corrosion resistance, such as quartz.

[0015] An exhaust pipe 12 for exhausting gases inside the reaction tube 11 is airtightly connected to the upper side of the reaction tube 11. The exhaust pipe 12 is provided with a pressure adjusting unit 13 consisting of a valve, a vacuum pump, etc., and adjusts the pressure inside the reaction tube 11 to a desired level (degree of vacuum).

[0016] A substantially cylindrical manifold 14 is provided below the reaction tube 11. The upper end of the manifold 14 is airtightly joined to the lower end of the reaction tube 11.

[0017] A lid body 15 is disposed below the manifold 14 (reaction tube 11). The lid body 15 is configured to be movable up and down by a boat elevator 16. When the lid body 15 is raised by the boat elevator 16, the lower side (furnace opening portion) of the manifold 14 (reaction tube 11) is closed, and when the lid body 15 is lowered by the boat elevator 16, the lower side (furnace opening portion) of the reaction tube 11 is opened.

[0018] A wafer boat 18 is provided above the lid 15 via a heat-retaining cylinder (thermal insulator) 17. The wafer boat 18 is a wafer holder that accommodates (holds) objects to be processed, such as semiconductor wafers W, and in this embodiment, is configured to accommodate a plurality of semiconductor wafers W, for example, 150 semiconductor wafers W, spaced at predetermined intervals in the vertical direction. The semiconductor wafers W are then loaded into the reaction tube 11 by placing the semiconductor wafers W in the wafer boat 18 and raising the lid 15 using the boat elevator 16.

[0019] A heater unit 19 made of, for example, a resistance heating element is provided around the reaction tube 11 so as to surround the reaction tube 11. The heater unit 19 heats the inside of the reaction tube 11 to a predetermined temperature, and as a result, the semiconductor wafer W is heated to the predetermined temperature.

[0020] The heater section 19 has heaters 191 to 195 arranged in, for example, five stages, and power is supplied to each of the heaters 191 to 195 independently from power controllers 196 to 200, and they can be controlled independently. In this way, the inside of the reaction tube 11 is divided into five zones (ZONE 1 to ZONE 5) by the heaters 191 to 195 as shown in Fig. 3. The number of zones is not limited to five, and it may be two or more, and it is not necessary to divide into zones.

[0021] The manifold 14 is provided with a plurality of gas supply pipes for supplying gases into the reaction tube 11. In this embodiment, three gas supply pipes 20 to 22 are provided. A source gas and a carrier gas for film formation are supplied to the gas supply pipes 20 to 22 via flow rate adjusting units 23 to 25 each including a mass flow controller (MFC) or the like for adjusting the gas flow rate.

[0022] Five temperature sensors (thermocouples) (not shown) are arranged in a vertical row on the inner wall of the reaction tube 11. These temperature sensors are covered with quartz pipes or the like to prevent metal contamination of the semiconductor wafers W, and are arranged in each zone shown in FIG.

[0023] The semiconductor manufacturing apparatus 2 is equipped with a control unit 50 for controlling process parameters such as the temperature, gas flow rate, and pressure of the process atmosphere in the reaction tube 11. The control unit 50 receives output signals from a temperature sensor, a pressure sensor, etc. (not shown), and outputs control signals to power controllers 196-200 of heaters 191-195, the pressure adjustment unit 13, and the flow rate adjustment units 23-25.

[0024] [Example of hardware configuration of control unit] 4 is a diagram showing an example of the configuration of the control unit 50. The control unit 50 includes a recipe storage unit 51, a ROM 52, a RAM 53, an I / O port 54, a CPU 55, a communication unit 56, and a bus 57 interconnecting these components.

[0025] The recipe storage unit 51 stores process recipes that define control procedures according to the type of film formation process performed by the semiconductor manufacturing equipment 2. The process recipe is process information prepared for each process that an operator actually performs. The process recipe specifies the temperature changes in each section, the pressure changes within the reaction tube 11, the timing and amount of gas supply start and stop, and other information from the time when the semiconductor wafers W are loaded into the reaction tube 11 to the time when the processed semiconductor wafers W are unloaded. This process recipe can specify the set film thickness for the heat treatment and the set temperature for each section of the equipment. In a typical batch-type heat treatment apparatus, one process recipe is prepared for all semiconductor wafers W. However, in this embodiment, a predetermined process recipe is prepared for each zone shown in FIG. 3 to ensure uniform processing results for the semiconductor wafers W.

[0026] This process recipe also includes an optimization value calculation recipe that corrects the set temperature of each part of the equipment to an optimal temperature (optimized value) based on a temperature correction table, and sets this corrected temperature as the set temperature. Therefore, when semiconductor manufacturing equipment 2 performs heat treatment based on the optimization value calculation recipe, optimized values ​​for the set temperatures of zones 1 to 5 are calculated, and the optimized values ​​become the set temperatures.

[0027] Therefore, the set temperature is corrected by the semiconductor manufacturing equipment 2, but the temperature correction information calculation device 4 may provide the semiconductor manufacturing equipment 2 with a process recipe in which the set temperature has been corrected based on a temperature correction table. Alternatively, the set temperature may be corrected by any device as long as the semiconductor manufacturing equipment 2 can ultimately execute the process recipe at the corrected set temperature.

[0028] The ROM 52 is a recording medium that is made up of an EEPROM, a flash memory, a hard disk, etc., and stores the operating program of the CPU 55 and the like.

[0029] The RAM 53 functions as a work area for the CPU 55. The RAM 53 stores, for example, the number of times a process has been executed in the semiconductor manufacturing equipment 2. The cumulative film thickness of the deposits adhering to the inside of the semiconductor manufacturing equipment 2 can be determined based on the number of times the process has been executed and the set film thickness stored in the recipe storage unit 51.

[0030] The I / O port 54 supplies measurement signals relating to temperature, pressure, and gas flow rate to the CPU 55, and also outputs control signals output by the CPU 55 to each unit (power controllers 196-200, flow rate adjustment units 23-25, and pressure adjustment unit 13). In addition, an operation panel 58 through which an operator operates the semiconductor manufacturing equipment 2 is connected to the I / O port 54.

[0031] The CPU 55 constitutes the core of the control unit 50, executes the operating program stored in the ROM 52, and controls the operation of the semiconductor manufacturing equipment 2 in accordance with the process recipe stored in the recipe memory unit 51 in accordance with instructions from the operation panel 58.

[0032] A communication unit 56 performs communication between the semiconductor manufacturing equipment 2, the host computer 3, and the temperature correction information calculation device 4 via LANs 5 and 6. A bus 57 transfers information between the various units.

[0033] The host computer 3 is a device that controls the entire semiconductor manufacturing apparatus 2, and executes processes to instruct each semiconductor manufacturing apparatus 2 to perform heat treatment, etc. The host computer 3 has a known configuration and is not shown in the drawings.

[0034] [Example of hardware configuration of temperature correction information calculation device] FIG. 5 shows an example of the hardware configuration of the temperature correction information calculation device 4. The temperature correction information calculation device 4 executes a process to generate a temperature correction table according to the accumulated film thickness (number of film formations) in order to reduce the difference between the furnace temperature and the set temperature caused by the accumulated film thickness in each of the semiconductor manufacturing devices 2. The temperature correction information calculation device 4 also updates the model used to generate the temperature correction table according to the film formation results. The model will be described in detail later.

[0035] The temperature correction information calculation device 4 has a CPU (Central Processing Unit) 501, a ROM (Read Only Memory) 502, and a RAM (Random Access Memory) 503. The CPU 501, the ROM 502, and the RAM 503 form a so-called computer. The temperature correction information calculation device also has an auxiliary storage device 504, an operation device 505, a display device 506, an I / F (Interface) device 507, and a drive device 508. The hardware components of the temperature correction information calculation device 4 are connected to each other via a bus 509.

[0036] The CPU 501 executes various programs installed in the auxiliary storage device 504 .

[0037] The ROM 502 is a non-volatile memory that functions as a main storage device and stores various programs, data, etc. that are required for the CPU 501 to execute various programs installed in the auxiliary storage device 504.

[0038] The RAM 503 is a volatile memory such as a dynamic random access memory (DRAM) or a static random access memory (SRAM), and functions as a main storage device. The RAM 503 provides a working area into which various programs installed in the auxiliary storage device 504 are expanded when executed by the CPU 501.

[0039] The auxiliary storage device 504 is a nonvolatile large-capacity storage device that stores various programs. The auxiliary storage device 504 may be a nonvolatile large-capacity storage medium such as a hard disk drive (HDD) or a solid state drive (SSD).

[0040] The operation device 505 is an input device used by an administrator to input various instructions to the temperature correction information calculation device 4. The display device 506 is a display device that displays internal information of the temperature correction information calculation device 4 and information acquired from the outside.

[0041] The I / F device 507 is a connection device that connects to the LAN 6 and communicates with the control unit 50 of the semiconductor manufacturing equipment 2. The I / F device 507 also communicates with the measuring equipment 60 and the host computer 3.

[0042] Drive device 508 is a device for loading a recording medium. Recording media include media that record information optically, electrically, or magnetically, such as CD-ROMs, flexible disks, and magneto-optical disks. Recording media may also include semiconductor memories that record information electrically, such as ROMs and flash memories.

[0043] The various programs to be installed in the auxiliary storage device 504 are installed, for example, by setting a distributed recording medium in the drive device 508 and reading the various programs recorded on the recording medium by the drive device 508. Alternatively, the various programs to be installed in the auxiliary storage device 504 may be installed by downloading them from a predetermined server.

[0044] [Functions of the temperature correction information calculation device] Next, the information held by the temperature correction information calculation device 4 and its functional configuration will be described with reference to Fig. 6. Fig. 6 is an example of a functional block diagram explaining the functional configuration held by the temperature correction information calculation device 4 in blocks.

[0045] The temperature correction information calculation device 4 has an acquisition unit 61, a temperature correction information generation unit 62, a model learning unit 63, and a learning determination unit 64. These functions of the temperature correction information calculation device 4 are functions or means realized by the CPU 501 shown in FIG. 5 executing a program loaded from the auxiliary storage device 504 to the RAM 503.

[0046] The acquiring unit 61 acquires log information relating to the heat treatment process of the semiconductor wafer W executed based on the process recipe from the control unit 50 of the semiconductor manufacturing equipment 2. The acquiring unit 61 can also acquire the monitored film thickness of the semiconductor wafer W monitored from the measuring device 60. The log information indicates the number of times the process has been executed in the semiconductor manufacturing equipment 2, and therefore the cumulative film thickness of the deposits adhering to the interior of the semiconductor manufacturing equipment 2 can be identified based on this number of times the process has been executed and the set film thickness stored in the process recipe.

[0047] The temperature correction information generating unit 62 generates a temperature correction table using a model, which will be described later, for the relationship between the set temperature included in the log information and the monitored film thickness measured by the measuring device 60 .

[0048] The model learning unit 63 updates the model stored in the model storage unit 72. Updating the model means changing the model to one that more accurately represents the relationship between the set temperature and film thickness in the semiconductor manufacturing equipment 2, based on the film formation result and the target film thickness. In this sense, updating can be called learning, and this term is also used in this embodiment.

[0049] The learning determination unit 64 determines whether updating the model is appropriate based on the film formation results and the like, and switches the model learning unit 63 to update the model (ON) or not (OFF).

[0050] The temperature correction information calculation device 4 also has a storage unit 70 formed in the RAM 503 or the auxiliary storage device 504. The storage unit 70 has a temperature correction table storage unit 71 and a model storage unit 72.

[0051] Fig. 7 is an example of a temperature correction table. The temperature correction table shows the relationship between the cumulative film thickness of deposits adhering inside the equipment and the temperature correction amount for each temperature (set temperature) inside the semiconductor manufacturing equipment 2. In the temperature correction table of Fig. 7, the temperature inside the semiconductor manufacturing equipment 2 is set to a predetermined temperature, and a temperature correction value is determined for each zone when deposits of a predetermined cumulative film thickness have adhered inside the equipment.

[0052] Using Figure 7 as an example, a semiconductor wafer is initially formed with a film thickness of tk1 (nm) before the process recipe is executed. tk1 is, for example, 100 (nm), but it varies depending on the process recipe. The process recipe includes six film formation processes. These six film formation processes constitute one cycle of heat treatment. A film is formed with a fixed film thickness Δtk in each film formation process. The cumulative film thickness increases as tk2 = tk1 + Δtk, tk3 = tk2 + Δtk, ..., until a film thickness of tk6 + Δtk is obtained by the sixth process. Note that one cycle of heat treatment can include any number of film formation processes, and the film thickness formed in one film formation process varies depending on the process recipe.

[0053] When the semiconductor manufacturing equipment 2 corrects the set temperature of zone 1, for example, when the cumulative film thickness is tk2 [nm], the corrected set temperature is calculated as "set temperature of the process recipe + 0.1"°C.

[0054] The temperature correction table is an example of information that stores temperature correction values ​​for correcting the set temperature, and the storage format of the temperature correction values ​​is not limited to a table or tabular format. For example, the temperature correction table may be in a functional format or a graphical format.

[0055] If the set temperature and cumulative film thickness differ from the values ​​defined in the temperature correction table, the semiconductor manufacturing equipment 2 can perform interpolation to accommodate the difference.

[0056] Fig. 8 is a diagram for schematically explaining the models stored in the model storage unit 72. A model is prepared for each set temperature, but Fig. 8 shows a model for an arbitrary set temperature.

[0057] In the model of FIG. 8(a), the amount of temperature fluctuation of the semiconductor wafer W when the set temperature differs by 1°C is determined for each zone. In other words, the temperature of the semiconductor wafer W may not be exactly the same as the set temperature due to factors such as the temperature distribution in the reaction tube 11. The model of FIG. 8(a) calculates the actual temperature of the semiconductor wafer W when the set temperature differs by 1°C. The model of FIG. 8(a) represents the amount of temperature change in each slot when the temperature in each zone is increased by +1°C. In other words, changing the temperature in one zone also changes the temperatures in the other zones. This temperature calculated by the model of FIG. 8(a) is used when referencing the model of FIG. 8(b).

[0058] In the model of Figure 8(b), the amount of film thickness variation when the temperature differs by 1°C is determined for each zone. A film is formed on the semiconductor wafer W to a thickness that corresponds to the actual temperature of the semiconductor wafer W during heat treatment (assuming other parameters such as gas type and pressure are constant) and the heat treatment time. The model of Figure 8(b) represents the correspondence between the temperature of the actual semiconductor wafer W and film thickness according to various findings. For example, in Zone 1, the amount of film thickness variation when the temperature differs by 1°C (this temperature has been corrected in Figure 8(a)) is A [nm].

[0059] An example of generating a temperature correction table for zone 1 using the model in Fig. 8(b) will be described below. For example, the following evaluation function is used to generate the temperature correction table.

[0060] Evaluation function J = f (residual error from target film thickness, model showing temperature and film thickness change, temperature change) The optimum value is the combination of temperature change amounts that minimizes the evaluation function J within the constraint range, with the constraint conditions that the heater power does not saturate and the temperature constraint range specified by the user is not exceeded. The method for generating the temperature correction table will be described in detail later.

[0061] Therefore, since the two models in Figure 8 are referenced when generating the temperature correction table, it is desirable that they be as accurate as possible. For this reason, the temperature correction information calculation device 4 has a function for learning the model. The model learning unit 63 updates (learns) the model based on the difference between the monitored film thickness and the target film thickness. The idea is that the greater the difference between the monitored film thickness, which is the film formation result, and the target film thickness, the greater the amount of temperature fluctuation of the semiconductor wafer W when the temperature differs by 1°C in the model of Figure 8(a), and the greater the amount of film thickness fluctuation when the temperature differs by 1°C in the model of Figure 8(b).

[0062] More specifically, the model learning unit 63 updates the two models by applying the difference between the monitored film thickness and the target film thickness to an extended Kalman filter or the like. The model update method is not a feature of the present application, so details are omitted. In the following, unless otherwise specified, the models are assumed to include the two models in FIG. 8.

[0063] [Outline of temperature compensation function according to cumulative film thickness] FIG. 9 is a diagram illustrating an example of a film formation result when a film formation process is performed by performing temperature correction according to the accumulated film thickness using the temperature correction table.

[0064] (1) First, heat treatment is performed by the semiconductor manufacturing equipment 2 without a temperature correction table. In Figure 9, this corresponds to heat treatment "before temperature correction." Heat treatment "before temperature correction" is called the first heat treatment (it includes six film formation processes). The film thickness formed in one film formation process is Δtk [nm]. The semiconductor manufacturing equipment 2 performs the film formation process in several steps to obtain a film with a film thickness set in the process recipe (for example, tk6 + Δtk [nm]). Figure 9(a) shows the film formation result obtained in the film formation process of the first heat treatment.

[0065] In FIG. 9(a), the horizontal axis represents the cumulative film thickness and the vertical axis represents the monitor film thickness. The monitor film thickness varies around Δtk. In FIG. 9(a), one cycle includes six film formation processes. The monitor film thickness formed in one film formation process is shown against the cumulative film thickness accumulated over six film formation processes. The multiple graphs in FIG. 9(a) represent the monitor film thickness of semiconductor wafers W at different vertical positions in the wafer boat 18. In other words, because the gas concentration differs depending on the height within the reaction tube 11, semiconductor wafers W at different heights are selected in advance for monitoring. The semiconductor wafers W whose monitor film thickness is measured are semiconductor wafers W that represent each zone.

[0066] As the number of film formation processes increases, the cumulative film thickness on the inner wall surface of the semiconductor manufacturing equipment also increases. If the cumulative film thickness increases, even if the same processing temperature is used, the temperature inside the furnace will decrease, making it impossible to form a thin film of the desired thickness on the semiconductor wafer W. In Figure 9(a), the monitored film thickness before temperature correction tends to gradually increase as the cumulative film thickness increases.

[0067] (2) The temperature correction information generator 62 generates the temperature correction table shown in Fig. 9(b) by using the model of Fig. 8 for the relationship between the set temperature included in the log information and the monitored film thickness obtained from the measurement device 60. The generation of the temperature correction table immediately after the initial heat treatment is called the first calculation process.

[0068] In Figure 9(b), the process corresponding to "calculation" is the generation of a temperature correction table. However, the second and subsequent calculations after the temperature correction table has been generated include updating the model. Since no temperature correction is performed in the first heat treatment (as there is a high possibility that an appropriate monitor film thickness will not be obtained), the model is not updated in the first calculation process.

[0069] (3) Heat treatment is performed by the semiconductor manufacturing equipment 2 using the temperature correction table. Figure 9 corresponds to the heat treatment "after temperature correction." Figure 9(c) shows the film formation results obtained in the film formation process in which the set temperature was corrected using the temperature correction table. Figure 9(c) shows a tendency for the monitored film thickness to stabilize due to temperature correction, even if the cumulative film thickness increases.

[0070] The model learning unit 63 can update the model using the monitored film thickness (3). Whether to update the model is determined when there is a large difference between the target film thickness and the film formation result (when the criteria are not met, as will be described later), and also based on disturbances and the accuracy of the model. Therefore, the model may be updated in the second and subsequent calculation processes.

[0071] [Creating a temperature correction table] FIG. 10 is a flowchart illustrating a method for generating the temperature correction table.

[0072] First, the temperature correction information generating unit 62 extracts the accumulated film thickness data and the monitor film thickness formed on the semiconductor wafer W from the log information acquired by the acquiring unit 61, as shown in FIG. 11(a) (S101).

[0073] Next, the temperature correction information generating unit 62 converts the extracted cumulative film thickness data and film thickness data (cumulative film thickness data - film thickness data) into "cumulative film thickness data - film thickness fluctuation data" as shown in the film thickness fluctuation of FIG. 11(a) (S102).

[0074] Next, the temperature correction information generating unit 62 extracts data on the processing temperature and processing time of the film forming process from the log information (S103).

[0075] Next, the temperature correction information generating unit 62 converts the accumulated film thickness data-film thickness fluctuation data into accumulated film thickness data-wafer temperature fluctuation data using the model (S104). For example, in the film formation process of this embodiment, as shown in the temperature fluctuation of FIG. 11(a), when the wafer temperature decreases by 1°C, the film thickness decreases by 1 nm.

[0076] Next, the temperature correction information generating unit 62 generates temperature correction information (temperature correction table) by optimization using the evaluation function J as described above (S105). The temperature correction table is obtained by converting the graph in FIG. 11(b) into a table.

[0077] The temperature correction table is generated in this manner. Note that the temperature correction table is generated each time, regardless of whether a model is generated (learning on or off). In other words, whether learning is on or off, the temperature correction table will have different values ​​than before the calculation. If the model is not updated (learning off), a new temperature correction table is calculated using the initial model and the most recent deposition results (+ process log). If the model is updated (learning on), a new temperature correction table is calculated using the updated model and the most recent deposition results (+ process log).

[0078] [Relationship between model and temperature correction table] FIG. 12 is a diagram illustrating the relationship between the model and the temperature correction table.

[0079] (i) With an initial model prepared in advance, the temperature correction information generating unit 62 generates a temperature correction table using the model and the film formation results of the initial heat treatment.

[0080] (ii) Furthermore, in the second and subsequent calculation processes (processes for updating the model based on the film formation results obtained when the set temperature is corrected using the temperature correction table), the model learning unit 63 updates the model itself using the film formation results.

[0081] (iii) If a temperature correction table has been generated, the temperature correction information generating unit 62 can update the temperature correction table using the model and the film formation results.

[0082] In this way, the model and the temperature correction table are updated based on the film formation results, so that even if the cumulative film thickness is large, a film thickness close to the target film thickness can be obtained.

[0083] [Issues with updating the model and temperature correction table] The learning function for the model and temperature correction table can be turned on or off at the discretion of the user of the semiconductor manufacturing equipment 2. However, in cases such as those described below, the calculated model and temperature correction table may not be appropriate, making it difficult for the user to determine whether to turn the learning function on or off.

[0084] Figure 13 is a diagram that explains the case where the calculated model and temperature correction table are inappropriate. First, as a premise, the temperature correction value of the temperature correction table is small (approximately less than 0.5°C), so if there is a disturbance (a factor that causes film thickness fluctuations) other than the cumulative film thickness, the learning function may not function correctly. In other words, a situation may arise in which the temperature correction value due to the disturbance is larger than the temperature correction value corresponding to the cumulative film thickness.

[0085] Figure 13(a) shows the film formation results when heat treatment is performed under the influence of both disturbances and cumulative film thickness. Based on this film formation result, the temperature correction information generation unit 62 generates a temperature correction table using a model. In this case, the obtained temperature correction table contains not only the cumulative film thickness but also temperature correction values ​​that suppress the influence of disturbances (Figure 13(b)). The temperature correction information generation unit 62 sends the generated temperature correction table to the control unit 50, and the control unit 50 calculates (corrects) the set temperature using the generated temperature correction table.

[0086] Next, assume that the semiconductor manufacturing equipment 2 performs heat treatment based on the set temperature corrected using the temperature correction table, but that there are no disturbances at this point. In this case, the semiconductor manufacturing equipment 2 should form a film using a temperature correction value that is based only on the cumulative film thickness, but because the temperature correction table also includes a temperature correction value that suppresses the effects of disturbances, the desired film formation results cannot be obtained (FIG. 13(c)).

[0087] The model learning unit 63 learns the model based on the film formation results in Figure 13(c), but it will recognize the deviation in the film formation results due to the absence of disturbance as a deviation in the model and perform erroneous learning. In other words, it is not preferable for the model learning unit 63 to learn the model based on the film formation results in Figure 13(c). If the model learning unit 63 were to learn the model based on the film formation results in Figure 13(c), the temperature correction table may also become inappropriate.

[0088] To prevent this problem, the temperature correction information calculation device 4 can, in principle, turn off the learning function of the model. Turning off the learning function of the model means that the model remains in its initial state and a new temperature correction table is generated based on the film formation results of the initial heat treatment. When learning is turned off, the model is not updated, but a new temperature correction table is generated.

[0089] However, a temperature correction table using a model with low accuracy may not improve the film formation results.

[0090] FIG. 14 shows an example of a film formation result that does not improve. For ease of explanation, the target film thickness is assumed to be 100 [nm]. Without a temperature correction table (first heat treatment), the film thickness was thinner than the target film thickness (90 [nm]). When the temperature correction information calculation device 4 generated a temperature correction table based on this film formation result (first calculation process) and film formation was performed, the film thickness was thicker than the target film thickness (120 [nm]). Next, when the semiconductor manufacturing equipment 2 performed film formation using the temperature correction table generated by the same model, the film thickness was thinner than the target film thickness (80 [nm]).

[0091] As shown in Figure 14, if a low-precision model is not updated, the film thickness may be too thick or too thin for the target film thickness each time heat treatment is performed, and stable film formation may become difficult even if a temperature correction table generated by a low-precision model is used.

[0092] [Automatic switching of learning function] Therefore, the temperature correction information calculation device 4 of this embodiment controls the on / off of the model learning function as follows. The default setting for the learning function is off (new calculation of the temperature correction table). From the second calculation process onwards, the temperature correction information calculation device 4 automatically determines whether the learning function is on or off.

[0093] 15 is a diagram illustrating an example of a method for determining whether to turn the learning function on or off. The learning determination unit 64 uses the following two criteria (I, II) to determine whether to update the model.

[0094] I. It is assumed that there is a disturbance → OFF One way to determine whether or not a disturbance has occurred is to infer it from the film formation results. For example, if the in-plane distribution of film thickness changes significantly, there is a possibility that a disturbance has occurred in the film formation process. Figure 15(a) shows an example in which the film thickness at the center of the semiconductor wafer W is thicker.

[0095] There are also other ways to infer the temperature from other causes. For example, if hardware such as thermocouples or other temperature sensors is replaced before heat treatment, differences between individual temperature sensors may affect the measured temperature.

[0096] II. There is no disturbance and it can be assumed that the accuracy of the model is low → ON Although the semiconductor manufacturing equipment 2 performs film formation processing by correcting the temperature using the temperature correction table, if the monitored film thickness deviates from the target film thickness, it can be assumed that there is no hope of improvement without model learning.

[0097] 15(b), similar to Fig. 14, the semiconductor manufacturing equipment 2 performed heat treatment by correcting the temperature using the temperature correction table, but the film formation results were either too thick or too thin for the target film thickness with each heat treatment (cycle). In this case, it can be inferred that there is no hope of improvement without model learning (the accuracy of the model is low).

[0098] [Toggle model training on and off] 16 is a flowchart showing the procedure for switching on and off the model learning function by the learning determination unit 64. The process in FIG. 16 is executed following the initial heat treatment.

[0099] When the initial heat treatment (first cycle) is completed, the measurement device 60 measures the monitor film thickness of the semiconductor wafer W to be monitored. The learning and determination unit 64 determines whether the monitor film thickness satisfies a criterion with respect to the target film thickness (S1). The criterion means that for each cumulative film thickness, the difference between the monitor film thickness and the target film thickness is within a threshold value. If the criterion is met, the temperature correction table is not necessary, and the process of FIG. 16 ends.

[0100] If the criteria are not met, the temperature correction information generating unit 62 generates a temperature correction table from the initial model and the film formation results (S2). In this way, if the film formation results do not meet the criteria, the temperature correction table is automatically generated. Since the model learning function is off, the model is not updated.

[0101] The semiconductor manufacturing equipment 2 performs the second cycle of heat treatment using the temperature correction table (S3).

[0102] When the second cycle of heat treatment is completed, the measuring device 60 measures the monitor film thickness of the semiconductor wafer W to be monitored. The learning and judging unit 64 judges whether the monitor film thickness satisfies the standard for the target film thickness (S4). Steps S2 to S4 correspond to the first calculation process.

[0103] If the criteria are met, the temperature correction table generated in step S2 is determined to be appropriate, and the processing of FIG. 16 ends.

[0104] If the criteria are not met, the learning determination unit 64 determines whether to turn on the learning function of the model or leave it off (S5). Details of step S5 will be explained with reference to FIG.

[0105] When the model learning function is off, the temperature correction information generation unit 62 generates a new temperature correction table based on the current model (for example, the model in the initial state) and the film formation result in step S3 or S7. When the model learning function is on, the model learning unit 63 learns (updates) the model, and the temperature correction information generation unit 62 updates the temperature correction table using the learned model and the film formation result in step S3 or S7 (S6).

[0106] Thereafter, the semiconductor manufacturing equipment 2 repeats the heat treatment and determines whether the standard is met (S7, S8). If the standard is not met, the process returns to step S5, allowing the learning determination unit 64 to determine whether to turn on the model learning function.

[0107] Fig. 17 is a flowchart illustrating a method for determining whether to turn on or off the learning function of a model by the learning determination unit 64. Fig. 18 is a diagram listing specific details of the determination of whether to turn on or off the learning function of a model.

[0108] The learning determination unit 64 determines whether there is a disturbance other than the process (S11). As shown in FIG. 18(a), a disturbance other than the process refers to "a (human) operation or the like that is obviously going to affect the process result without the need to check the process result." A disturbance inferred from the process result refers to "a state in which there is no particular abnormality in the apparent conditions, such as the process recipe or the hardware conditions of the equipment, but the actual film formation result predicts that some (unintended) disturbance has occurred." Examples include the following:

[0109] a) There was hardware replacement, such as a temperature sensor replacement, prior to the heat treatment. The log information includes hardware information such as temperature sensor identification information, and the learning determination unit 64 can detect hardware replacement based on changes in this hardware identification information. Individual differences in hardware may affect temperature control, and the deposition results may not be suitable for updating the model.

[0110] b) The user edited the process recipe to set the temperature and performed the heat treatment. More precisely, if the user edits even one of the deposition-related parameters, such as not only the temperature but also the gas flow rate or deposition time, and then performs heat treatment, this is considered a disturbance other than the process. In other words, in the flow of heat treatment A → temperature correction table calculation (calculation) / update → process recipe parameter editing → heat treatment B → model update → temperature correction table calculation (calculation), if learning is performed in a state where the deposition conditions for heat treatment A and heat treatment B are different to begin with (the deposition conditions in the process recipe were changed before heat treatment B), an incorrect model will be learned, as if the deviation in the film thickness was due to a deviation in the model, even though the deviation in the film thickness is actually due to the difference in deposition conditions.

[0111] Since the log information includes the set temperature, etc., when the log information does not match the data in the past calculation history, it is determined that the recipe has been edited.If a user edits a process recipe, the film formation results will include the influence of the temperature set by the user, so there is a risk that the model may not be suitable for updating.

[0112] c) The monitor position was changed. The user can set the position of the semiconductor wafer W whose film thickness is to be monitored (the film thickness of which semiconductor wafer W in the wafer boat 18 the measurement device will monitor). The slot number of the wafer boat 18 whose film thickness is monitored is recorded in the log information. If the monitor position is changed, the film formation results will include the difference in the monitor position, which may make them unsuitable for updating the model.

[0113] d) A certain amount of time, such as six months, has passed since the last model update and before this heat treatment was carried out. The date and time of the heat treatment is recorded in the log information. If a long time has passed between the model update and the heat treatment, the state of the semiconductor manufacturing equipment 2 may change, and the film formation results may not be suitable for updating the model.

[0114] e) The surface condition of the dummy wafer (whether or not it has a coating) is inconsistent. A management device (not shown) manages the history of each semiconductor wafer W, and it is clear that the initial state of the semiconductor wafer W is different. If the surface state of the semiconductor wafer when the model was generated differs from the surface state of the semiconductor wafer in the current film formation process, the film formation results will also change, and there is a risk that the film formation results will not be suitable for updating the model.

[0115] f) The magnification (output) of the top plate heater or cooling fan is extremely high. The magnification of the heater and cooling fan is also included in the log information. It is known that extreme changes in the heater and cooling fan output can affect the deposition results, so the deposition results may not be suitable for updating the model.

[0116] If there is a disturbance other than the process, the model should not be learned, so the learning decision unit 64 decides to turn off the learning function of the model (S16). That is, the temperature correction information generation unit 62 generates a new temperature correction table. Note that the generated temperature correction table is either deleted or treated as not to be used.

[0117] If there is no disturbance other than the process, the learning determination unit 64 determines whether there is a disturbance in the film formation result (process result) (S12). As shown in FIG. 18(b), the disturbance in the film formation result refers to the following.

[0118] g) There is a significant change in the film thickness distribution on the semiconductor wafer W or in the uniformity between semiconductor wafers. The in-plane distribution of the film thickness of the semiconductor wafer W and the film thickness of the semiconductor wafer W to be monitored are measured by the measuring device 60. In this case, there is a possibility that factors other than temperature may have had an effect, and therefore the film formation results may not be suitable for updating the model.

[0119] h) Measurement noise (abnormal points) occurs during film thickness measurement. Measurement noise (abnormal points) refers to the measurement of a film thickness that cannot or hardly ever occur during film deposition. In this case, it can be determined that the monitored film thickness is not reliable enough, and the film deposition results may not be suitable for updating the model.

[0120] If there is a disturbance in the film formation result, the model should not be learned, so the learning determination unit 64 determines to turn off the learning function of the model (S16). That is, the temperature correction information generation unit 62 generates a new temperature correction table. Note that the generated temperature correction table is deleted or treated as not to be used.

[0121] If there is no disturbance in the film formation result, the learning determination unit 64 determines whether the accuracy of the model is sufficient (S13). As shown in Fig. 18(c), the accuracy of the model is insufficient in the following cases.

[0122] i) The deposition results were worse before and after calculating the temperature correction table. For example, in the process of Figure 16, if the film formation results of the heat treatment (S3) after the temperature correction table has been optimized by calculation are poorer than the film formation results of the initial heat treatment, learning is turned off at the timing of the judgment in S5. Regardless of whether learning has been performed or not, if the accuracy of the model is low, the film formation results may not approach the target film thickness even if a temperature correction table optimized by calculation is used. In this case, it becomes necessary to update (learn) the model.

[0123] This phenomenon can occur when there is a discrepancy between the conditions (temperature, pressure, etc.) used when creating the model and the conditions used when the heat treatment was actually performed, or when there is insufficient data available when creating the model.

[0124] If the accuracy of the model is not sufficient, the model should be updated, so the learning determination unit 64 turns on the model learning function (S15). In this case, the temperature correction table is updated with the learned model.

[0125] If the accuracy of the model is sufficient, there is no need to update the model, and the learning determination unit 64 turns off the learning function of the model (S14). That is, the temperature correction information generation unit 62 generates a new temperature correction table. Note that the generated temperature correction table is either deleted or treated as not being used.

[0126] Furthermore, as in the conventional case, regardless of the result of the model learning determination made by the learning determination unit 64, the user can set the model learning function to ON or OFF (S17).

[0127] [Potential effects of automatically switching the model learning function on and off] FIG. 19 is a diagram illustrating the expected effect of automatically switching the model learning function on and off.

[0128] 1. First, if model learning is on even though a disturbance is present, the learning determination unit 64 can automatically control learning to be turned off. This prevents the temperature correction information calculation device 4 from performing erroneous learning and calculating an inappropriate temperature correction table.

[0129] 2. Furthermore, if model learning is turned off even though the model accuracy is poor, the learning determination unit 64 can automatically control learning to be turned on. This allows the temperature correction information calculation device 4 to prevent calculation of the temperature correction table using a model that will result in a film formation result that does not converge to the target film thickness.

[0130] 3. These effects allow the temperature correction information calculation device 4 to reduce the number of heat treatments that tend to be wasted, such as the second and subsequent heat treatments. As a result, the present disclosure can reduce the number of heat treatment cycles performed, the time required for adjustment, and the amount of monitor wafers consumed. In other words, the present disclosure can reduce the cost of adjusting the temperature in heat treatment.

[0131] [Major effects] As described above, the heat treatment system of this embodiment can automatically switch the model learning function on and off, thereby reducing the need for the user to make cumbersome decisions. Furthermore, because the model learning function is automatically switched on and off, the set temperature for heat treatment of semiconductor wafers can be easily adjusted.

[0132] 〔others〕 In this embodiment, the temperature correction information calculation device 4 shown in FIG. 1 performs model learning and generates the temperature correction table, but the semiconductor manufacturing device 2 may have the function of the temperature correction information calculation device 4.

[0133] Furthermore, the temperature correction information calculation device 4 may be a server connected to a network. The semiconductor manufacturing equipment 2 can communicate with the temperature correction information calculation device 4 to obtain the temperature correction table. The temperature correction information calculation device 4 may be located on-premise or in the cloud.

[0134] 1 is just one example, and it goes without saying that there are various system configuration examples depending on the application and purpose. The division of devices such as the host computer 3, semiconductor manufacturing device 2, measurement device 60, and temperature correction information calculation device 4 in FIG. 1 is just one example.

[0135] For example, the heat treatment system 1 can have various configurations, such as a configuration in which at least two of the host computer 3, semiconductor manufacturing equipment 2, measurement equipment 60, and temperature correction information calculation device 4 are integrated, or a configuration in which they are further divided. For example, a temperature correction information calculation device 4 may be provided for each semiconductor manufacturing equipment 2.

[0136] The semiconductor manufacturing apparatus 2 disclosed in this specification is not limited to a batch processing apparatus, but can also be applied to either a single-wafer apparatus that processes substrates one by one, or a semi-batch apparatus.

[0137] The substrate processing performed by the semiconductor manufacturing apparatus disclosed in this specification may be applied not only to film formation processing but also to an oxidation apparatus that performs oxidation processing to oxidize the surface region of a semiconductor wafer, a diffusion apparatus that diffuses (dopes) impurities into the surface region of a semiconductor wafer, an annealing apparatus, an etching apparatus, etc.

[0138] The semiconductor manufacturing apparatus 2 disclosed in this specification may be an apparatus that processes a substrate using plasma. [Explanation of symbols]

[0139] 1. Heat Treatment System 2. Semiconductor manufacturing equipment 4 Temperature correction information calculation device 61 Acquisition Department 62 Temperature correction information generation section 63 Model Learning Department 64 Learning Assessment Department

Claims

1. 1. A temperature correction information calculation device for a semiconductor manufacturing device that forms a film on a processing object by heat treatment at a set temperature corrected according to a cumulative film thickness accumulated on an inner wall of the semiconductor manufacturing device, a model storage unit that stores a model for generating temperature correction information in which a temperature correction value is associated with the cumulative film thickness; a learning determination unit that determines whether to update the model when a film formation result by heat treatment is obtained; a model learning unit that updates the model based on the film formation result on the processing target object when the learning determination unit determines that the model should be updated; a temperature correction information generation unit that generates the temperature correction information using the model updated by the model learning unit and corrects the set temperature using the temperature correction information, The model includes a first table that defines, for each zone in which an individual heater is disposed, a variation in temperature of the object to be processed when the set temperature differs by 1°C, and a second table that defines, for each zone, a variation in film thickness when the set temperature differs by 1°C, the set temperature being corrected based on the first table; When the learning determination unit determines that the model should be updated, the model learning unit increases the amount of fluctuation in the temperature of the object to be processed when the set temperature differs by 1°C in the first table, and increases the amount of fluctuation in the film thickness when the set temperature differs by 1°C in the second table, as the difference between the monitor film thickness, which is the film formation result, and the target film thickness increases. Temperature correction information calculation device.

2. 2. The temperature correction information calculation device according to claim 1, wherein the learning determination unit determines not to update the model when it is determined that the film formation result is affected by a disturbance.

3. 3. The temperature correction information calculation device according to claim 2, wherein the learning determination unit determines not to update the model if hardware of the semiconductor manufacturing equipment is replaced before the film formation result is obtained.

4. 4. The temperature correction information calculation device according to claim 2, wherein the learning determination unit determines not to update the model when a parameter related to film formation is edited.

5. The temperature correction information calculation device according to any one of claims 2 to 4, wherein the learning judgment unit judges not to update the model if the position of the workpiece whose film thickness is being monitored is changed before the film formation result is obtained.

6. The temperature correction information calculation device according to any one of claims 2 to 5, wherein the learning judgment unit determines not to update the model if a certain amount of time has passed since the model was last updated until the current heat treatment is performed.

7. The temperature correction information calculation device according to any one of claims 2 to 6, wherein the learning judgment unit determines not to update the model when it is determined that the surface condition of the workpiece is different from the last time the model was updated, based on the history of the workpiece managed by the management device.

8. The temperature correction information calculation device according to any one of claims 2 to 7, wherein the learning determination unit determines not to update the model when an output of a heater or a cooling fan is equal to or greater than a certain value in the heat treatment in which the film formation result was obtained.

9. The temperature correction information calculation device according to any one of claims 2 to 8, wherein the learning judgment unit judges not to update the model when the in-plane distribution of the film thickness of the workpiece or the difference in film thickness between the workpieces does not satisfy a criterion.

10. The temperature correction information calculation device according to any one of claims 2 to 9, wherein the learning judgment unit judges to update the model when the film formation result of the heat treatment after updating the model is farther from the target film thickness compared to the film formation result before updating.

11. A semiconductor manufacturing device comprising the temperature correction information calculation device according to any one of claims 1 to 10.

12. 1. A method for calculating temperature correction information for a semiconductor manufacturing apparatus that heat-treats a processing target at a set temperature corrected according to a cumulative film thickness accumulated on an inner wall of the semiconductor manufacturing apparatus, comprising: a step of determining whether to update a model for generating temperature correction information in which a temperature correction value is associated with the cumulative film thickness when a film formation result by the heat treatment is obtained; updating the model based on the film formation result on the processing object when it is determined that the model should be updated; generating the temperature correction information using the updated model; The model includes a first table that defines, for each zone in which an individual heater is disposed, a variation in temperature of the object to be processed when the set temperature differs by 1°C, and a second table that defines, for each zone, a variation in film thickness when the set temperature differs by 1°C, the set temperature being corrected based on the first table; If it is determined in the determining step that the model should be updated, in the model updating step, the greater the difference between the monitor film thickness, which is the film formation result, and the target film thickness, the greater the amount of variation in the temperature of the object to be processed when the set temperature differs by 1°C in the first table, and the greater the amount of variation in the film thickness when the set temperature differs by 1°C in the second table. Temperature correction information calculation method.

13. a temperature correction information calculation device for a semiconductor manufacturing apparatus that forms a film on a processing object by heat treatment at a set temperature corrected according to a cumulative film thickness accumulated on an inner wall of the semiconductor manufacturing apparatus; a learning determination unit that, when a film formation result by the heat treatment is obtained, determines whether or not to update a model for generating temperature correction information in which a temperature correction value is associated with a cumulative film thickness; a model learning unit that updates the model based on the film formation result on the processing target object when the learning determination unit determines that the model should be updated; the model learning unit generates the temperature correction information using the updated model, and the temperature correction information generation unit corrects the set temperature using the temperature correction information; The model includes a first table that defines, for each zone in which an individual heater is disposed, a variation in temperature of the object to be processed when the set temperature differs by 1°C, and a second table that defines, for each zone, a variation in film thickness when the set temperature differs by 1°C, the set temperature being corrected based on the first table; When the learning determination unit determines that the model should be updated, the model learning unit increases the amount of fluctuation in the temperature of the object to be processed when the set temperature differs by 1°C in the first table, and increases the amount of fluctuation in the film thickness when the set temperature differs by 1°C in the second table, as the difference between the monitor film thickness, which is the film formation result, and the target film thickness increases. program.

Images