Information processing device, information processing method, and program

Abstract

To improve the prediction accuracy of the temperature inside a processing container.SOLUTION: There is provided an information processing device comprising: a model acquisition unit which acquires a trained model for predicting the temperature inside a processing container of a first substrate processing device; a data acquisition unit which acquires sensor data used to measure the temperature inside a processing container of a second substrate processing device; an optimization unit which optimizes parameters indicative of the difference in temperature responsiveness between the first substrate processing device and the second substrate processing device on the basis of the trained model and the sensor data; and a prediction unit which predicts the temperature inside the processing container of the second substrate processing device on the basis of the trained model and parameters.SELECTED DRAWING: Figure 4

Description

[Technical Field]

[0001] The present disclosure relates to an information processing device, an information processing method, and a program. [Background technology]

[0002] There are known techniques for predicting the temperature inside a processing vessel for processing a workpiece. For example, Patent Document 1 discloses a method for calibrating a heat treatment apparatus that includes a processing vessel for accommodating the workpiece, multiple heaters, and multiple temperature sensors, stores a thermal model for estimating the temperature of the workpiece inside the processing vessel from the output of the temperature sensors, predicts the temperature of the workpiece inside the processing vessel from the output of the temperature sensors using the thermal model, and controls the heaters based on the predicted temperature to perform heat treatment on the workpiece. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Application Laid-Open No. 2005-026397 Summary of the Invention [Problem to be solved by the invention]

[0004] The present disclosure provides techniques for improving the accuracy of predicting the temperature inside a process vessel. [Means for solving the problem]

[0005] According to one aspect of the present disclosure, there is provided an information processing apparatus comprising: a model acquisition unit that acquires a trained model that predicts the temperature inside a processing vessel provided in a first substrate processing apparatus; a data acquisition unit that acquires sensor data measuring the temperature inside a processing vessel provided in a second substrate processing apparatus; an optimization unit that optimizes a parameter indicating a difference in temperature responsiveness between the first substrate processing apparatus and the second substrate processing apparatus based on the trained model and the sensor data; and a prediction unit that predicts the temperature inside the processing vessel provided in the second substrate processing apparatus based on the trained model and the parameter. [Effects of the Invention]

[0006] In one aspect, the accuracy of predicting the temperature inside the processing vessel can be improved. [Brief explanation of the drawings]

[0007] [Figure 1] 1 is a block diagram showing an example of an overall configuration of a substrate processing system; [Figure 2] 1 is a schematic cross-sectional view showing an example of a substrate processing apparatus. [Figure 3] FIG. 2 is a block diagram illustrating an example of a hardware configuration of a computer. [Figure 4] FIG. 2 is a block diagram illustrating an example of a functional configuration of a prediction device. [Figure 5] 10 is a flowchart illustrating an example of an optimization process. [Figure 6] 10 is a flowchart illustrating an example of a prediction process. [Figure 7] FIG. 10 is a diagram illustrating an example of an operation flow of the prediction device. [Figure 8] FIG. 10 is a diagram illustrating an example of a search process using Bayesian optimization. [Figure 9] FIG. 10 is a diagram illustrating an example of a comparison result of prediction errors before and after optimization. [Figure 10] FIG. 10 is a diagram illustrating an example of prediction accuracy before optimization. [Figure 11] FIG. 10 is a diagram illustrating an example of prediction accuracy after optimization. DETAILED DESCRIPTION OF THE INVENTION

[0008] Hereinafter, embodiments of the present disclosure will be described with reference to the drawings. In the drawings, the same components are denoted by the same reference numerals, and redundant explanations may be omitted.

[0009] [Embodiment] One embodiment of the present disclosure is a substrate processing system for processing substrates, which are an example of objects to be processed. In this embodiment, the substrate processing system includes a substrate processing apparatus that heat-treats semiconductor wafers, which are an example of substrates, in a processing chamber. The substrate processing system also includes a prediction apparatus that predicts a temperature (hereinafter also referred to as an "in-furnace temperature") in the processing chamber where the semiconductor wafers are heat-treated.

[0010] Conventionally, a prediction model for predicting furnace temperatures has been constructed based on sensor data indicating sensor values ​​measured by a temperature sensor installed in a reference substrate processing apparatus (hereinafter also referred to as a "basic apparatus"). Since differences exist between individual substrate processing apparatuses, predicting furnace temperatures for a substrate processing apparatus other than the basic apparatus (hereinafter also referred to as a "target apparatus") may result in reduced prediction accuracy. Hereinafter, a prediction model constructed using sensor data from the basic apparatus will be referred to as a "basic model." The basic apparatus is an example of a first substrate processing apparatus. The target apparatus is an example of a second substrate processing apparatus.

[0011] Since various changes occur in substrate processing apparatuses over time, even if a prediction model had sufficient prediction accuracy at a certain point in time, the prediction accuracy may decrease over time. Furthermore, in recent years, substrate processing apparatuses have become longer in lifespan, and errors due to aging tend to increase.

[0012] The conventional technology builds a prediction model based on sensor data when the temperature is stable, and does not assume the prediction of furnace temperatures other than when the temperature is stable. Furthermore, it does not consider the impact of changes in the substrate processing apparatus over time on prediction accuracy. Because the conventional technology requires a special process for setting up the substrate processing apparatus, the cost of building a prediction model is high, and it is difficult to reconstruct the model after the substrate processing apparatus has started operating.

[0013] Furthermore, in recent years, process recipes executed in substrate processing equipment have become more complex, making it important to save energy throughout the entire process recipe. If the furnace temperature can be predicted with high accuracy throughout the entire process recipe, optimal temperature settings can be achieved, improving energy efficiency. Even if the prediction accuracy improves immediately after constructing a prediction model and when the temperature is stable, it is difficult to improve the prediction accuracy of the entire process recipe.

[0014] The present embodiment aims to improve the accuracy of prediction of the furnace temperature by correcting the predicted value of the furnace temperature output by the basic model using a parameter that indicates the difference in temperature response between the basic device and the target device.

[0015] According to one aspect, the present embodiment predicts the furnace temperature in consideration of the difference in temperature response between the basic equipment and the target equipment, thereby enabling accurate prediction of the furnace temperature of the target equipment. According to another aspect, the present embodiment enables parameters to be optimized if there is sensor data measured when a process recipe is executed, thereby enabling easy tracking of changes over time in the target equipment.

[0016] <System configuration> The overall configuration of a substrate processing system according to this embodiment will be described with reference to Fig. 1. Fig. 1 is a block diagram showing an example of the overall configuration of a substrate processing system.

[0017] 1, the substrate processing system 100 includes substrate processing apparatuses 120a1 to 120a3 and controllers 121a1 to 121a3 in a factory a. The substrate processing apparatuses 120a1 to 120a3 and the controllers 121a1 to 121a3 are connected to each other by wire or wirelessly.

[0018] The substrate processing system 100 also includes substrate processing apparatuses 120b1 and 120b2 and controllers 121b1 and 121b2 in a factory b. The substrate processing apparatuses 120b1 and 120b2 and the controllers 121b1 and 121b2 are connected to each other by wire or wirelessly.

[0019] The substrate processing system 100 also includes substrate processing apparatuses 120c1 and 120c2 and controllers 121c1 and 121c2 in a factory c. The substrate processing apparatuses 120c1 and 120c2 and the controllers 121c1 and 121c2 are connected to each other by wire or wirelessly.

[0020] Substrate processing apparatuses 120a1-120a3, substrate processing apparatuses 120b1-120b2, and substrate processing apparatuses 120c1-120c2 are connected to host apparatuses 110a, 110b, and 110c, respectively, via networks N1-N3. Each substrate processing apparatus performs substrate processing under the control of each control device based on instructions from host apparatuses 110a, 110b, and 110c. Host apparatuses 110a, 110b, and 110c are connected to server apparatus 150 via network N4, such as the Internet.

[0021] In the following description, the substrate processing apparatuses 120a1 to 120a3, 120b1, 120b2, 120c1, and 120c2 are also collectively referred to as substrate processing apparatus 120. The control apparatuses 121a1 to 121a3, 121b1, 121b2, 121c1, and 121c2 are also collectively referred to as control apparatus 121. The host apparatuses 110a, 110b, and 110c are also collectively referred to as host apparatus 110.

[0022] It is assumed that the substrate processing apparatuses 120a1 to 120a3, the substrate processing apparatuses 120b1 and 120b2, and the substrate processing apparatuses 120c1 and 120c2 store various data that they manage within themselves.

[0023] The prediction device 140 is connected to the substrate processing apparatuses 120 including the substrate processing apparatus 120a1, and thereby continuously acquires accumulated data accumulated in each of the substrate processing apparatuses 120. The example in Fig. 2 shows a state in which the prediction device 140 is connected to the substrate processing apparatus 120a1, but this is not limiting. In the present embodiment, the case in which the prediction device 140 is connected to the substrate processing apparatus 120a1 will be described in detail below.

[0024] The substrate processing system 100 shown in Fig. 1 is one example, and it goes without saying that there are various system configuration examples depending on the application and purpose. The classification of devices such as the host device 110, substrate processing device 120, control device 121, prediction device 140, and server device 150 shown in Fig. 1 is one example. For example, the number of factories, the number of host devices 110, the number of substrate processing device 120, the number of control devices 121, the number of prediction devices 140, etc. are one example and are not limited to these.

[0025] For example, the substrate processing system 100 can have various configurations, such as a configuration in which at least two of the host device 110, substrate processing apparatus 120, control device 121, prediction device 140, and server device 150 are integrated together, or a configuration in which they are further divided. For example, the control device 121 may be configured to collectively control a plurality of substrate processing apparatuses 120, or may be provided in a one-to-one correspondence with the substrate processing apparatus 120, or may be integrated with the substrate processing apparatus 120.

[0026] The prediction device 140 may be realized by the host device 110 or by the server device 150. In this case, the prediction device 140 is unnecessary. The prediction device 140 may also be realized by the control device 121. The prediction device 140 may also be realized by a control device that collectively controls a plurality of control devices 121.

[0027] <Substrate processing equipment> An example of a substrate processing apparatus according to this embodiment will be described with reference to Fig. 2. Fig. 2 is a schematic cross-sectional view showing a vertical heat treatment apparatus, which is an example of a substrate processing apparatus.

[0028] The vertical heat treatment apparatus 120 in this embodiment is a substrate treatment apparatus that accommodates a large number of semiconductor wafers W, which are an example of objects to be treated, at one time and performs heat treatment such as oxidation, diffusion, low-pressure CVD (Chemical Vapor Deposition), etc. As shown in Fig. 2, the vertical heat treatment apparatus 120 includes a treatment vessel 10, a gas supply unit 20, an exhaust unit 30, a heating unit 40, a cooling unit 50, a control device 121, etc.

[0029] The processing vessel 10 has a substantially cylindrical shape. The processing vessel 10 includes an inner tube 11, an outer tube 12, a manifold 13, an injector 14, a gas outlet 15, a lid 16, etc. The inner tube 11 has a substantially cylindrical shape. The outer tube 12 has a substantially cylindrical shape with a ceiling, and the inner tube 11 and the outer tube 12 form a double-tube structure. The inner tube 11 and the outer tube 12 are made of a heat-resistant material such as quartz.

[0030] The manifold 13 has a substantially cylindrical shape. The manifold 13 supports the lower ends of the inner tube 11 and the outer tube 12. The manifold 13 is made of, for example, stainless steel. The injector 14 passes through the manifold 13 and extends horizontally into the inner tube 11, and then bends in an L shape inside the inner tube 11 and extends upward. The base end of the injector 14 is connected to the gas introduction pipe 24, and the tip is open. The injector 14 discharges the process gas (hereinafter simply referred to as "gas") introduced via the gas introduction pipe 24 into the inner tube 11 from the opening at the tip. There may be multiple injectors 14.

[0031] The gas outlet 15 is formed in the manifold 13. The processing gas is exhausted by the exhaust unit 30 through the gas outlet 15. The lid 16 airtightly closes the opening at the lower end of the manifold 13. The lid 16 is made of, for example, stainless steel. A wafer boat (substrate holder) 18 is placed on the lid 16 via a heat-insulating tube 17. The heat-insulating tube 17 and the wafer boat 18 are made of, for example, a heat-resistant material such as quartz.

[0032] The wafer boat 18 holds a plurality of semiconductor wafers W in a substantially horizontal position at predetermined intervals in the vertical direction. The wafer boat 18 is loaded into the processing vessel 10 by the lifting mechanism 19 raising the lid 16, and is accommodated in the processing vessel 10. The wafer boat 18 is unloaded from the processing vessel 10 by the lifting mechanism 19 lowering the lid 16.

[0033] The gas supply unit 20 includes a gas source 21, an IGS (Integrated Gas System) 22, an external pipe 23, and a gas introduction pipe 24. The gas source 21 is a supply source of processing gas and includes, for example, a film formation gas source, a cleaning gas source, and a purge gas source. The IGS 22 is an integrated circuit of gas pipes, and integrates a group of pipes connected to the film formation gas source, cleaning gas source, purge gas source, etc. of the gas source 21. A flow rate control unit is installed within the IGS 22 and controls the flow rate of gas flowing through each pipe. The flow rate control unit includes, for example, a mass flow controller and an opening / closing valve.

[0034] The IGS 22 is connected to an external pipe 23. The external pipe 23 is connected to a gas introduction pipe 24. A heater (not shown) is wrapped around the outer periphery of the external pipe 23 to heat the external pipe 23. The gas introduction pipe 24 is connected to the processing vessel 10 and introduces gas into the processing vessel 10. That is, the flow rate of the processing gas from the gas source 21 is controlled by a flow rate control unit in the IGS 22, and the processing gas is heated as it flows through the external pipe 23 and flows into the gas introduction pipe 24, and is supplied from the gas introduction pipe 24 into the processing vessel 10 via the injector 14. The injector 14 functions as a gas inlet for the processing vessel 10.

[0035] A gas piping joint 82 connected to the gas introduction pipe 24 is provided near the gas inlet of the processing chamber 10. A temperature sensor 80 is configured to pass through the joint 82. The temperature sensor 80 is configured to measure the temperature of the gas in the gas introduction pipe 24. The temperature sensor 80 transmits the measured temperature to the control device 121. A second heater 81 is also provided in the gas introduction pipe 24, and the second heater 81 is configured to heat the gas in the gas introduction pipe 24.

[0036] The exhaust unit 30 includes an exhaust device 31, an exhaust pipe 32, and a pressure controller 33. The exhaust device 31 is, for example, a vacuum pump such as a dry pump or a turbo molecular pump. The pressure controller 33 is installed in the exhaust pipe 32 and controls the pressure inside the processing vessel 10 by adjusting the conductance of the exhaust pipe 32. The pressure controller 33 is, for example, an automatic pressure control valve.

[0037] The heating unit 40 includes a thermal insulator 41, a first heater 42, and an outer skin 43. The thermal insulator 41 has a substantially cylindrical shape and is disposed around the outer tube 12. The thermal insulator 41 is primarily composed of silica and alumina. The first heater 42 has a linear shape and is disposed in a spiral or serpentine shape around the inner periphery of the thermal insulator 41. The first heater 42 is configured to enable temperature control by dividing the processing vessel 10 into multiple zones in the height direction. The outer skin 43 is disposed to cover the outer periphery of the thermal insulator 41. The outer skin 43 maintains the shape of the thermal insulator 41 and reinforces it. The outer skin 43 is formed of a metal such as stainless steel. A water-cooled jacket may be disposed around the outer skin 43 to suppress thermal influence of the heating unit 40 on the outside. The heating unit 40 heats the inside of the processing vessel 10 by heat generated by the first heater 42.

[0038] The cooling unit 50 supplies a cooling fluid toward the processing vessel 10 to cool the semiconductor wafer W in the processing vessel 10. The cooling fluid may be, for example, air. The cooling unit 50 supplies the cooling fluid toward the processing vessel 10 when, for example, rapidly lowering the temperature of the semiconductor wafer W after heat treatment. The cooling unit 50 has a fluid flow path 51, an outlet 52, a distribution flow path 53, a flow rate adjustment unit 54, and a heat exhaust port 55.

[0039] A plurality of fluid flow paths 51 are formed in the height direction between the insulating material 41 and the outer skin 43. The fluid flow paths 51 are, for example, flow paths formed around the outside of the insulating material 41. Blowing holes 52 are formed from each fluid flow path 51, penetrating the insulating material 41, and blow out the cooling fluid into the space between the outer pipe 12 and the insulating material 41. The distribution flow paths 53 are provided outside the outer skin 43, and distribute and supply the cooling fluid to each fluid flow path 51. The flow rate adjustment unit 54 is interposed in the distribution flow paths 53, and adjusts the flow rate of the cooling fluid supplied to the fluid flow paths 51.

[0040] The heat exhaust port 55 is provided above the plurality of blow-out holes 52, and exhausts the cooling fluid supplied to the space between the outer pipe 12 and the heat insulating material 41 to the outside of the processing vessel 10. The cooling fluid exhausted to the outside of the processing vessel 10 is cooled, for example, by a heat exchanger and supplied again to the distribution flow path 53. However, the cooling fluid exhausted to the outside of the processing vessel 10 may be exhausted without being reused.

[0041] The temperature sensor 60 detects the temperature inside the processing vessel 10. The temperature sensor 60 is provided, for example, in the inner pipe 11. However, the temperature sensor 60 may be provided at any position where it can detect the temperature inside the processing vessel 10, for example, in the space between the inner pipe 11 and the outer pipe 12. The temperature sensor 60 has, for example, multiple temperature measuring units provided at different positions in the height direction corresponding to multiple zones. The multiple temperature measuring units may be, for example, thermocouples or resistance temperature detectors. The temperature sensor 60 transmits the temperatures detected by the multiple temperature measuring units to the control device 121.

[0042] The control device 121 controls the operation of the vertical heat treatment device 120, thereby controlling the semiconductor process performed in the vertical heat treatment device 120. The control device 121 may be, for example, a computer.

[0043] <Computer> The host device 110, the control device 121, the prediction device 140, and the server device 150 included in the substrate processing system 100 shown in Fig. 1 are realized by a computer having a hardware configuration as shown in Fig. 3. Fig. 3 is a block diagram showing an example of the hardware configuration of a computer.

[0044] 3, the computer 500 includes an input device 501, an output device 502, an external I / F (interface) 503, a RAM (random access memory) 504, a ROM (read only memory) 505, a CPU (central processing unit) 506, a communication I / F 507, and an HDD (hard disk drive) 508, all of which are interconnected by a bus B. The input device 501 and the output device 502 may be connected and used when necessary.

[0045] The input device 501 is a keyboard, mouse, touch panel, etc., and is used by an operator or the like to input various operation signals. The output device 502 is a display, etc., and displays the results of processing by the computer 500. The communication I / F 507 is an interface that connects the computer 500 to a network. The HDD 508 is an example of a non-volatile storage device that stores programs and data.

[0046] The external I / F 503 is an interface with an external device. The computer 500 can read and / or write data from and to a recording medium 503a such as an SD (Secure Digital) memory card via the external I / F 503. The ROM 505 is an example of a non-volatile semiconductor memory (storage device) that stores programs and data. The RAM 504 is an example of a volatile semiconductor memory (storage device) that temporarily stores programs and data.

[0047] The CPU 506 is a computing device that controls the entire computer 500 and realizes its functions by reading programs and data from storage devices such as the ROM 505 and HDD 508 onto the RAM 504 and executing the processes.

[0048] <Functional configuration> The functional configuration of the prediction device in this embodiment will be described with reference to Fig. 4. Fig. 4 is a block diagram showing an example of the functional configuration of the prediction device.

[0049] 4, the prediction device 140 includes a model storage unit 201, a data acquisition unit 210, a model acquisition unit 220, an optimization unit 230, a recipe reception unit 240, and a prediction unit 250. The prediction device 140 functions as the model storage unit 201, the data acquisition unit 210, the model acquisition unit 220, the optimization unit 230, the recipe reception unit 240, and the prediction unit 250 by executing a prediction program installed in advance.

[0050] The model storage unit 201 is realized, for example, by the RAM 504 or the HDD 508 shown in Fig. 3. The data acquisition unit 210, the model acquisition unit 220, the optimization unit 230, the recipe reception unit 240, and the prediction unit 250 are realized, for example, by the CPU 506 shown in Fig. 3 executing a program loaded on the RAM 504.

[0051] A trained prediction model is pre-stored in the model storage unit 201. The trained prediction model includes a base model and correction parameters. However, in the initial state, the correction parameters may not be included in the prediction model. Alternatively, in the initial state, the prediction model may include correction parameters to which predetermined initial values ​​are set.

[0052] The basic model is a machine learning model that predicts the temperature (furnace temperature) inside the processing vessel 10 provided in the substrate processing apparatus 120. The basic model is trained to predict the furnace temperature of the basic apparatus based on sensor data indicating sensor values ​​measured by a temperature sensor 60 provided in the basic apparatus. The basic model receives control information related to the control of the furnace temperature as input and outputs a predicted value of the furnace temperature. The control information related to the control of the furnace temperature may include the power of the heating unit 40 that heats the inside of the processing vessel 10 and the output of the cooling unit 50 that cools the inside of the processing vessel 10.

[0053] The prediction model corrects the predicted value output by the basic model using a correction parameter. The correction parameter is a parameter that indicates the difference in temperature responsiveness between the basic apparatus and the target apparatus. As an example, the temperature responsiveness may be the temperature change that occurs when the temperature of the processing vessel is controlled based on predetermined control information. The difference in temperature responsiveness may be the ratio of the furnace temperatures when the heating unit 40 and the cooling unit 50 are operated based on the same control information in the basic apparatus and the target apparatus.

[0054] Specifically, the prediction model is defined by equation (1).

[0055]

number

[0056] where k is a time index, y(k) is a predicted value of the furnace temperature at time k, f is a basic model, u(k) is control information related to temperature control at time k, G(u) is a correction term, and g is a correction matrix. The correction matrix g is an example of a correction parameter.

[0057] The predicted value y is a vector including the predicted value of the furnace temperature in each zone. The control information u is a vector including the power of the heating unit 40 and the output of the cooling unit 50 corresponding to each zone. The correction matrix g is a matrix including the correction coefficient for each zone.

[0058] The data acquiring unit 210 acquires sensor data measured in the target apparatus. The sensor data is time-series data including actual values ​​of furnace temperatures measured by a temperature sensor installed in the target apparatus. The data acquiring unit 210 may acquire, together with the sensor data, a process recipe that was executed when the sensor data was generated. The process recipe may be a recipe including multiple processes with different set temperatures for the processing vessel 10.

[0059] The model acquisition unit 220 acquires a trained prediction model. The model acquisition unit 220 may read out a prediction model stored in the model storage unit 201. The trained prediction model includes a basic model and correction parameters.

[0060] The optimization unit 230 optimizes the correction parameters based on the prediction model acquired by the model acquisition unit 220 and the sensor data acquired by the data acquisition unit 210. Specifically, the optimization unit 230 predicts the furnace temperature using the prediction model in accordance with the process recipe acquired by the data acquisition unit 210, and optimizes the correction parameters based on the error between the actual furnace temperature value and the predicted furnace temperature value contained in the sensor data. The error may be an average value for the entire process recipe.

[0061] For example, the optimization unit 230 may optimize the correction parameters by Bayesian optimization. However, the optimization method used by the optimization unit 230 is not limited to Bayesian optimization, and any mathematical optimization method may be used. In Bayesian optimization, for example, the correction parameters may be optimized so as to minimize the absolute mean square error (RMSE). However, the error used for optimization is not limited to the absolute mean square error, and any error may be used.

[0062] The absolute mean square error may be calculated, for example, by equation (2).

[0063]

number

[0064] Here, n is the number of sensor values ​​included in the sensor data, y(i) is the predicted value of the i-th sensor value, and y'(i) is the actual measured value of the i-th sensor value.

[0065] In a substrate processing apparatus 120 in which the processing vessel 10 is divided into multiple zones, the temperature of a certain zone is affected by all of the first heaters 42 provided in the processing vessel 10. Since the prediction model uses the power input to each of the first heaters 42 as an input, the correction parameters for each zone have the same number of elements as the number of first heaters 42. Since it is difficult for a user to sequentially design all of the elements of the correction parameters, the correction parameters can be determined efficiently by using an optimization method based on machine learning, such as Bayesian optimization.

[0066] The optimization unit 230 may iteratively optimize the correction parameters based on sensor data acquired at multiple points in time. In other words, the optimization unit 230 may optimize the correction parameters, which have been optimized based on sensor data measured at a first point in time, based on sensor data measured at a second point in time. However, the second point in time is a point in time later than the first point in time.

[0067] For example, the optimization unit 230 may optimize the correction parameters every time the data acquisition unit 210 acquires sensor data. The optimization unit 230 may optimize the correction parameters every time the target apparatus executes a process recipe, based on the sensor data generated when the process recipe is executed.

[0068] The recipe receiving unit 240 receives an input of a process recipe to be predicted. The process recipe may include a plurality of processes in which the set temperatures of the processing vessel 10 are different.

[0069] The prediction unit 250 predicts the furnace temperature of the target apparatus based on the optimized prediction model. The optimized prediction model includes the basic model acquired by the model acquisition unit 220 and the correction parameters optimized by the optimization unit 230. The prediction unit 250 may predict the furnace temperature of the target apparatus when the process recipe accepted by the recipe acceptance unit 240 is executed.

[0070] The prediction unit 250 outputs the prediction result of the furnace temperature. The prediction result may include prediction data indicating the predicted value of the furnace temperature output by the prediction model. For example, the prediction unit 250 may display the prediction result on an output device 502 connected to the prediction device 140.

[0071] <Processing Procedure> An information processing method executed by the substrate processing system 100 in this embodiment will be described with reference to Figures 5 and 6. In this embodiment, the information processing method includes an optimization process (see Figure 5) and a prediction process (see Figure 6).

[0072] <Optimization process> 5 is a flowchart showing an example of optimization processing executed by the prediction device 140. The optimization processing is processing for optimizing correction parameters based on sensor data generated by the target device.

[0073] In step S1, the data acquisition unit 210 of the prediction device 140 acquires sensor data generated by the target device and the process recipe executed when the sensor data was generated. The data acquisition unit 210 acquires the sensor data after the target device executes the process recipe at least once. The data acquisition unit 210 may acquire sensor data every time the target device executes the process recipe. The data acquisition unit 210 sends the acquired sensor data and process recipe to the optimization unit 230.

[0074] In step S2, the model acquisition unit 220 of the prediction device 140 reads out a trained prediction model from the model storage unit 201. The trained prediction model includes a base model and correction parameters. The model acquisition unit 220 sends the read out prediction model to the optimization unit 230.

[0075] In step S3, the optimization unit 230 of the prediction device 140 receives the sensor data and the process recipe from the data acquisition unit 210. The optimization unit 230 also receives the prediction model from the model acquisition unit 220.

[0076] The optimization unit 230 acquires control information related to temperature control of the processing vessel based on the process recipe. The optimization unit 230 inputs the acquired control information into a prediction model. The prediction model outputs a predicted value of the furnace temperature based on the input control information. The predicted value output by the prediction model is a value obtained by correcting the predicted value output by the basic model using a correction parameter. The optimization unit 230 generates prediction data indicating the predicted value output by the prediction model.

[0077] In step S4, the optimization unit 230 of the prediction device 140 calculates a prediction error based on the prediction data and the sensor data. Specifically, the optimization unit 230 calculates the absolute mean square error between the prediction data and the sensor data for the entire process recipe for each zone in the processing vessel.

[0078] In step S5, the optimization unit 230 of the prediction device 140 updates the correction parameters based on the prediction errors calculated in step S4. Specifically, the optimization unit 230 designs new correction parameters that are highly likely to reduce the prediction errors by Bayesian optimization. The optimization unit 230 updates the correction parameters included in the prediction model with the new correction parameters.

[0079] In step S6, the optimization unit 230 of the prediction device 140 determines whether a predetermined convergence condition is satisfied. The convergence condition may be, for example, that the number of updates of the correction parameters is equal to or greater than a predetermined number, or that the reduction in the prediction error is equal to or less than a predetermined threshold.

[0080] If the convergence condition is satisfied (YES), the optimization unit 230 proceeds to step S7. On the other hand, if the convergence condition is not satisfied (NO), the optimization unit 230 returns to step S3.

[0081] When the process returns to step S3, the optimization unit 230 executes steps S3 to S6 again based on the prediction model in which the correction parameters have been updated in the immediately preceding step S5. In this way, the optimization unit 230 repeatedly executes the prediction of the furnace temperature and the update of the correction parameters until it determines in step S6 that the convergence condition is satisfied.

[0082] In step S7, the optimization unit 230 of the prediction device 140 stores the prediction model including the optimized correction parameters in the model storage unit 201. As a result, the prediction model stored in the model storage unit 201 is updated to the optimized prediction model.

[0083] <Prediction processing> 6 is a flowchart showing an example of a prediction process executed by the prediction device 140. The prediction process is a process for predicting the furnace temperature of the target device based on a prediction model including optimized correction parameters.

[0084] In step S11, the recipe receiving unit 240 of the prediction apparatus 140 receives an input of a process recipe to be predicted. The recipe receiving unit 240 sends the received process recipe to the prediction unit 250.

[0085] In step S12, the model acquisition unit 220 of the prediction device 140 reads out the optimized prediction model from the model storage unit 201. The optimized prediction model includes the base model and the optimized correction parameters. The model acquisition unit 220 sends the read out prediction model to the prediction unit 250.

[0086] In step S13, the prediction unit 250 of the prediction device 140 receives the process recipe from the recipe receiving unit 240. The prediction unit 250 also receives the optimized prediction model from the model acquisition unit 220.

[0087] The prediction unit 250 acquires setting information related to temperature control of the processing vessel based on the process recipe. The prediction unit 250 inputs the acquired setting information into a prediction model. The prediction model outputs a predicted value of the furnace temperature based on the input setting information. The predicted value output by the prediction model is a value obtained by correcting the predicted value output by the basic model using the optimized correction parameters. The prediction unit 250 generates prediction data indicating the predicted value output by the prediction model.

[0088] In step S14, the prediction unit 250 of the prediction device 140 outputs the prediction result including the prediction data generated in step S13. Specifically, the prediction unit 250 displays the prediction result on the output device 502 connected to the prediction device 140.

[0089] A user of the prediction device 140 can verify the furnace temperature when a process recipe is executed in a target device by referring to the prediction result output from the prediction device 140. The user of the prediction device 140 can use the verification result of the furnace temperature in the design or operation of the target device or process recipe.

[0090] <Operation flow> The operation flow of the prediction device 140 will be described with reference to Fig. 7. Fig. 7 is a diagram showing an example of the operation flow of the prediction device.

[0091] Figure 7 shows the operational flow when a process is executed N times in the target equipment. The process recipes used to execute each process may be the same or different. Each process recipe may be a recipe executed to manufacture a product to be shipped.

[0092] In the first process, a process recipe R-1 is created by the user of the target equipment. Next, based on the process recipe R-1, a temperature prediction P1-1 based on a prediction model and a substrate processing P2-1 using the target equipment are performed. In the temperature prediction P1-1, only a prediction based on the basic model is performed. Next, a correction parameter optimization P3-1 is performed based on the prediction data D1-1 obtained in the temperature prediction P1-1 and the sensor data D2-1 obtained in the substrate processing P2-1. This generates an optimized correction matrix G-1.

[0093] In the second process, temperature prediction P1-2 is performed based on the prediction model and substrate processing P2-2 is performed using the target equipment, based on the process recipe R-2. In the temperature prediction P1-2, the predicted value output by the basic model is corrected using the correction matrix G-1 generated in the first process. Next, correction parameter optimization P3-2 is performed based on the predicted data D1-2 obtained in the temperature prediction P1-2 and the sensor data D2-2 obtained in the substrate processing P2-2. This generates an optimized correction matrix G-2.

[0094] Thereafter, the correction matrix G is optimized in a similar manner each time a process is executed. In the Nth process, temperature prediction P1-N based on the prediction model and substrate processing P2-N using the target equipment are performed based on the process recipe RN. In the temperature prediction P1-N, the predicted value output by the basic model is corrected using the correction matrix G generated in the (N-1)th process. Next, correction parameter optimization P3-N is performed based on the predicted data D1-N obtained in the temperature prediction P1-N and the sensor data D2-N obtained in the substrate processing P2-N. This generates an optimized correction matrix GN.

[0095] The correction matrix G is always optimized based on the most recent sensor data D2 generated by the target device, which can reduce the impact of changes in the target device over time. For example, when the target device is relocated or undergoes maintenance, the operating environment of the target device changes, which can result in a decrease in prediction accuracy. Conventional techniques require a special process to be performed for setup in a new operating environment, but with this method, a correction matrix G optimized for the new operating environment can be obtained by performing the process once.

[0096] <Verification results> The verification results of the prediction device 140 will be described with reference to FIGS.

[0097] Fig. 8 is a diagram showing an example of a search process using Bayesian optimization. Fig. 8 shows a graph showing the transition of prediction error when the correction parameters are optimized using Bayesian optimization. The horizontal axis of Fig. 8 represents the number of iterations of Bayesian optimization, and the vertical axis represents the root mean square error (RMSE_score). EI (Expected Improvement) is used as the acquisition function for Bayesian optimization.

[0098] As shown in Fig. 8, for each of the seven zones in the processing vessel, the prediction error converged after about 20 iterations. Fig. 8 shows that Bayesian optimization is an effective method for optimizing the correction parameters.

[0099] Fig. 9 is a diagram showing an example of a comparison result of prediction errors before and after optimization. Fig. 9 is a table comparing the root mean square error (RMSE) between sensor data and prediction data for each of zones 1 to 7 before and after optimizing the correction parameters. The process recipe includes five processes with different set temperatures, and two temperature-raising steps and two temperature-lowering steps are executed.

[0100] As shown in Figure 9, after optimizing the correction parameters, the absolute mean squared error improved in all zones 1 to 7. Figure 9 shows that the prediction accuracy improved by 26.4% when averaging the absolute mean squared errors across all zones.

[0101] Fig. 10 is a diagram showing an example of prediction accuracy before optimization. Fig. 10 is a graph showing the transition of the temperature difference between the predicted value predicted using the correction parameters before optimization and the actual measured value. The process recipe is the same as that used when measuring the sensor data shown in Fig. 9, and includes two temperature-raising steps and two temperature-lowering steps. Note that Fig. 10 shows the temperature difference between the predicted value and the actual measured value in one zone.

[0102] As shown in Figure 10, in the time intervals when the furnace temperature is stable, the temperature difference between the predicted value and the actual measured value is small, and the furnace temperature can be predicted accurately even with the pre-optimization correction parameters. On the other hand, in the time intervals when the furnace temperature changes, the temperature difference between the predicted value and the actual measured value becomes large, and it can be seen that the prediction accuracy decreases when the furnace temperature changes.

[0103] Fig. 11 is a diagram showing an example of prediction accuracy after optimization. Fig. 11 is a graph showing the transition of the temperature difference between the predicted value predicted using the correction parameters after optimization and the actual measured value. The process recipe is the same as that used when the sensor data shown in Fig. 10 was measured, and includes two temperature-raising steps and two temperature-lowering steps. Note that Fig. 11 shows the temperature difference between the predicted value and the actual measured value in the same zone as the zone shown in Fig. 10.

[0104] As shown in Figure 11, in the time intervals when the furnace temperature is stable, the temperature difference between the predicted value and the actual measured value is small, and the furnace temperature can be predicted with high accuracy. Also, even in the time intervals when the furnace temperature changes, the temperature difference between the predicted value and the actual measured value is smaller than before optimization, and it can be seen that accurate predictions can be made even when the furnace temperature changes.

[0105] <Effects of the embodiment> The prediction device 140 in this embodiment optimizes a correction parameter that indicates the difference in temperature responsiveness between the basic device and the target device based on a basic model that predicts the furnace temperature of the basic device and sensor data that measures the furnace temperature of the target device, and predicts the furnace temperature of the target device based on the basic model and the correction parameter.

[0106] According to one aspect, the present embodiment predicts the furnace temperature of the target device by taking into account the difference in temperature response between the basic device and the target device, thereby enabling accurate prediction of the furnace temperature of the target device. Furthermore, according to the present embodiment, parameters can be optimized if there is sensor data measured when a process recipe is executed, making it possible to easily track changes over time in the target device.

[0107] The prediction device 140 may correct the predicted value of the furnace temperature predicted based on the basic model using the correction parameter. According to this embodiment, the difference in temperature responsiveness between the basic device and the target device can be reflected in the predicted value based on the basic model, so that the furnace temperature of the target device can be predicted with high accuracy even when the basic model is used.

[0108] The prediction device 140 may update the correction parameters based on the difference between the predicted data and the sensor data. The prediction device 140 may optimize the correction parameters using Bayesian optimization. For example, in a substrate processing apparatus that processes multiple substrates at once, such as a vertical heat treatment apparatus, multiple heating units or cooling units are provided to control the temperature inside the furnace, resulting in a huge number of relationships between setting information and the temperature inside the furnace. According to this embodiment, the correction parameters are optimized using a method based on machine learning, so that the correction parameters can be calculated efficiently.

[0109] The correction parameter may indicate a relationship between setting information related to temperature control of the processing vessel and a temperature difference within the processing vessel. The processing vessel may be divided into multiple zones, and the correction parameter may indicate a relationship between setting information and a temperature difference within each zone. The setting information may include the power of a heating unit that heats the interior of the processing vessel and the output of a cooling unit that cools the interior of the processing vessel. According to this embodiment, even if the processing vessel is divided into multiple zones, the furnace temperature can be accurately predicted by taking into account the temperature responsiveness of each zone.

[0110] The sensor data may be time-series data of the furnace temperature measured when a recipe including multiple processes with different set temperatures is executed. According to this embodiment, since a correction parameter obtained by learning the relationship between the setting information for temperature control and the furnace temperature is used, the furnace temperature can be predicted with high accuracy even when a process recipe involving temperature changes is executed.

[0111] The prediction device 140 may optimize the correction parameters, which have been optimized based on sensor data measured at a first time point, based on sensor data measured at a second time point. The prediction device 140 can optimize the correction parameters if the most recent sensor data is available, and therefore can easily follow changes over time in the target device.

[0112] [Other embodiments] The substrate processing apparatus for performing processes including the substrate processing method of the present disclosure is not limited to thermal processing apparatuses, and any type of apparatus such as atomic layer deposition (ALD) apparatus, capacitively coupled plasma (CCP), inductively coupled plasma (ICP), radial line slot antenna (RLSA), electron cyclotron resonance plasma (ECR), or helicon wave plasma (HWP) can be used as the substrate processing apparatus.

[0113] Furthermore, the substrate processing apparatus of the present disclosure can be applied to both plasma-using and non-plasma-using apparatuses that perform predetermined processing on substrates (e.g., film formation processing, etching processing, etc.). The substrate processing apparatus of the present disclosure can be applied to any of single-wafer apparatuses that process substrates one by one, batch apparatuses that process multiple substrates at once, and semi-batch apparatuses that process a smaller number of substrates at once than the number processed at once by a batch apparatus.

[0114] The information processing apparatus and substrate processing apparatus according to the presently disclosed embodiments are illustrative in all respects and are not limiting. The embodiments may be modified and improved in various ways without departing from the spirit and scope of the appended claims. The matters described in the above-described embodiments may be configured differently within a consistent range, and may be combined within a consistent range. [Explanation of symbols]

[0115] 100: Substrate processing system 110: Host device 120: Substrate processing equipment 121: Control device 140: Prediction device 150: Server device 201: Model memory unit 210: Data acquisition unit 220: Model acquisition section 230: Optimization section 240: Recipe Reception 250: Prediction Department

Claims

1. a model acquisition unit configured to acquire a trained model for predicting a temperature inside a processing vessel of the first substrate processing apparatus; a data acquiring unit configured to acquire sensor data indicating a sensor value measuring a temperature inside a processing vessel of the second substrate processing apparatus; an optimization unit configured to optimize a parameter indicating a difference in temperature responsiveness between the first substrate processing apparatus and the second substrate processing apparatus based on the trained model and the sensor data; a prediction unit configured to predict a temperature in a processing vessel of the second substrate processing apparatus based on the trained model and the parameters; and An information processing device comprising:

2. 2. The information processing device according to claim 1, The prediction unit corrects the predicted value of the temperature predicted based on the trained model using the parameters. Information processing device.

3. 3. The information processing device according to claim 2, the optimization unit is configured to update the parameters based on a difference between prediction data indicating the prediction value and the sensor data. Information processing device.

4. 4. The information processing device according to claim 3, the optimization unit is configured to optimize the parameters by Bayesian optimization. Information processing device.

5. 5. The information processing device according to claim 1, the parameter indicates a relationship between setting information regarding temperature control of the processing vessel and a temperature difference within the processing vessel; Information processing device.

6. 6. The information processing device according to claim 5, The processing vessel is divided into a plurality of zones, the parameter indicates a relationship between the setting information and the temperature difference in each of the zones; Information processing device.

7. 6. The information processing device according to claim 5, the setting information includes power of a heating unit that heats the inside of the processing vessel and power of a cooling unit that cools the inside of the processing vessel. Information processing device.

8. 5. The information processing device according to claim 1, the sensor data is time-series data of the temperature measured when a recipe including a plurality of processes with different set temperatures is executed; Information processing device.

9. 5. The information processing device according to claim 1, the optimization unit is configured to optimize the parameter, which has been optimized based on the sensor data measured at a first time point, based on the sensor data measured at a second time point. Information processing device.

10. The information processing device a step of acquiring a trained model for predicting a temperature inside a processing vessel of the first substrate processing apparatus; acquiring sensor data measuring a temperature inside a processing vessel of the second substrate processing apparatus; optimizing a parameter indicating a difference in temperature responsiveness between the first substrate processing apparatus and the second substrate processing apparatus based on the trained model and the sensor data; predicting a temperature in a processing vessel of the second substrate processing apparatus based on the trained model and the parameters; An information processing method that performs the above.

11. In the information processing device, a step of acquiring a trained model for predicting a temperature inside a processing vessel of the first substrate processing apparatus; acquiring sensor data measuring a temperature inside a processing vessel of the second substrate processing apparatus; optimizing a parameter indicating a difference in temperature responsiveness between the first substrate processing apparatus and the second substrate processing apparatus based on the trained model and the sensor data; predicting a temperature in a processing vessel of the second substrate processing apparatus based on the trained model and the parameters; A program to execute.

Images