Information processing apparatus, device difference analysis method, and substrate processing apparatus

Abstract

The invention provides an information processing apparatus, an equipment difference analysis method, and a substrate processing apparatus, which can further improve the precision of equipment difference analysis of the substrate processing apparatus. An information processing apparatus capable of performing equipment difference analysis of a plurality of substrate processing apparatuses includes: a reconstruction error matrix generation unit that checks reconstruction errors of data sets of the plurality of substrate processing apparatuses using a dimensionality reduction model trained with a data set as a reference, and generates a reconstruction error matrix of the plurality of substrate processing apparatuses; a device difference calculation unit that calculates the magnitude of at least some reconstruction errors selected from among the plurality of reconstruction errors included in the reconstruction error matrix as a device difference; and a display control unit that displays the calculated device difference on a display device.

Description

TECHNICAL FIELD

[0001] The present application relates to an information processing apparatus, a device difference analysis method, and a substrate processing apparatus. BACKGROUND

[0002] For example, in a semiconductor manufacturing apparatus that performs a process in accordance with the same recipe, the behavior of a sensor (waveform data of the sensor) is theoretically the same, and therefore a technique is known in which a log data of a sensor of a semiconductor manufacturing apparatus that performs a process in accordance with the same recipe is used to realize an analysis function of a device difference (for example, refer to Patent Literature 1).

[0003] PRIOR ART DOCUMENTS

[0004] PATENT LITERATURE

[0005] Patent Literature 1: Japanese Patent Application Publication No. 2022-3664 SUMMARY

[0006] PROBLEMS TO BE SOLVED BY THE INVENTION

[0007] The present application provides a technique for further improving the accuracy of device difference analysis of a substrate processing apparatus.

[0008] TECHNICAL SOLUTION TO THE PROBLEM

[0009] One embodiment of the present application is an information processing apparatus capable of performing device difference analysis of a plurality of substrate processing apparatuses, including: a reconstruction error matrix generation section that inspects reconstruction errors of data sets of the plurality of substrate processing apparatuses using a dimension reduction model trained using a data set as a reference, and generates a reconstruction error matrix of the plurality of substrate processing apparatuses; a device difference calculation section that calculates a size of at least a part of the reconstruction errors selected from a plurality of the reconstruction errors included in the reconstruction error matrix, as a device difference; and a display control section that causes a display apparatus to display the calculated device difference.

[0010] EFFECT OF THE INVENTION

[0011] According to the present application, a technique for further improving the accuracy of device difference analysis of a substrate processing apparatus can be provided. BRIEF DESCRIPTION OF DRAWINGS

[0012] Figure 1 is a configuration diagram of one example of a substrate processing system of the present embodiment.

[0013] Figure 2 is a hardware configuration diagram of one example of a computer.

[0014] Figure 3 is a functional block diagram of one example of an apparatus controller of the present embodiment.

[0015] Figure 4 FIG. 1 is an explanatory diagram of one example of the process of the training section and the reconstruction error matrix generating section.

[0016] Figure 5 FIG. 2 is an explanatory diagram of one example of the process of the verification stage.

[0017] Figure 6 FIG. 3 is a structural diagram of one example of the reconstruction error matrix.

[0018] Figure 7 FIG. 4 is an explanatory diagram of one example of the reconstruction error matrix in which the range of the reconstruction error is selected.

[0019] Figure 8 FIG. 5 is a diagram of one example of the calculation of the device difference of each level.

[0020] Figure 9 FIG. 6 is a flowchart showing one example of the process of the device difference analysis method performed by the substrate processing system of the present embodiment.

[0021] Explanation of Reference Numerals

[0022] 1 substrate processing system

[0023] 10 substrate processing device

[0024] 11 sensor

[0025] 12 device controller

[0026] 14 server device

[0027] 16 operator terminal

[0028] 30 device difference calculating section

[0029] 32 reconstruction error matrix generating section

[0030] 34 training section

[0031] 50, 52 dimension reduction model

[0032] 54 reference dataset storage section

[0033] 56 analysis dataset storage section

[0034] 502 output device DETAILED DESCRIPTION

[0035] Hereinafter, the present embodiment will be described with reference to the drawings.

[0036] <system structure>

[0037] Figure 1 FIG. 1 is a structural diagram of one example of the substrate processing system 1 of the present embodiment.Figure 1 The illustrated substrate processing system 1 has a substrate processing apparatus 10, sensors 11, an apparatus controller 12, a server apparatus 14, and an operator terminal 16. The substrate processing apparatus 10, the sensors 11, and the apparatus controller 12 are provided in a manufacturing plant 2. In addition, the server apparatus 14 and the operator terminal 16 can be provided in the manufacturing plant 2 or outside the manufacturing plant 2. The operator terminal 16 is an information processing terminal such as a PC (Personal Computer) or a smartphone operated by an operator such as an apparatus manager of the substrate processing apparatus 10 provided in the manufacturing plant 2.

[0038] Figure 1 The substrate processing apparatus 10, the apparatus controller 12, the server apparatus 14, and the operator terminal 16 are communicably connected via networks 18 and 20 such as the Internet or a LAN (Local Area Network).

[0039] The substrate processing apparatus 10 is an apparatus that performs processing such as film formation processing, etching processing, or ashing processing, and for example, processes a semiconductor wafer (hereinafter, simply referred to as a wafer). The substrate processing apparatus 10 is, for example, a semiconductor manufacturing apparatus, a heat treatment apparatus, or a film formation apparatus. The substrate processing apparatus 10 performs a process (processing) by, for example, receiving a recipe from the apparatus controller 12 and executing the recipe. The recipe is, for example, a control command composed of a combination of set values of various kinds such as temperature, gas, pressure, plasma, and machinery. The recipe of the substrate processing apparatus 10 has a plurality of control units called steps. The process of the substrate processing apparatus 10 includes a plurality of steps.

[0040] The substrate processing apparatus 10 is provided with a plurality of sensors 11 such as a temperature sensor that measures temperature and a pressure sensor that measures pressure. In addition, the substrate processing apparatus 10 is provided with an actuator that performs mechanical action by a combination of a power source and a constituent member.

[0041] The apparatus controller 12 has a function of a human-machine interface, can receive an instruction for the substrate processing apparatus 10 from an operator, and provides information on the substrate processing apparatus 10 to the operator. The apparatus controller 12 receives sensor data output from the plurality of sensors 11 provided in the substrate processing apparatus 10. The apparatus controller 12 can perform abnormality detection or abnormality prediction of the substrate processing apparatus 10 or the like.

[0042] Figure 1The device controller 12 is provided for each substrate processing apparatus 10, but may be provided for a plurality of substrate processing apparatuses 10. The device controller 12 may be provided inside the housing of the substrate processing apparatus 10 or outside the housing. In addition, the device controller 12 may have a function of communicating with the device controllers 12 for other substrate processing apparatuses 10. The device controller 12 may have a function of communicating with the device controllers 12 for other substrate processing apparatuses 10 via the server apparatus 14. In this way, the device controller 12 can use information about a plurality of substrate processing apparatuses 10 (such as sensor waveform data for each step when a process is performed according to the same protocol).

[0043] The server device 14 can receive sensor data output from the plurality of sensors 11 provided in the substrate processing apparatus 10 and store the data as a process log for each executed process (each run).

[0044] For example, the server device 14 can save information about multiple substrate processing devices 10 in one or more manufacturing plants 2 (schemes of processes performed in the substrate processing devices 10, sensor data and result data when the processes are performed by executing the schemes, etc.) as process logs for each process execution (run).

[0045] The device controller 12 and the server device 14 can display information about the substrate processing apparatus 10 on the operator terminal 16, or can notify the operator at the operator terminal 16 via e-mail or the like. Furthermore, at least one of the device controller 12, the server device 14, and the operator terminal 16 has a function for editing a recipe to be executed by the substrate processing apparatus 10. Figure 1 The device controller 12, the server device 14, and the operator terminal 16 are examples of information processing devices in this embodiment.

[0046] also, Figure 1 The substrate processing system 1 shown is an example, and there are of course various system configuration examples depending on the application and purpose. Figure 1 The device controller 12 and the server device 14 are divided into an example. For example, various structures such as a structure in which the device controller 12 and the server device 14 are integrated or a structure in which they are further divided can be adopted.

[0047] <Hardware Structure>

[0048] Figure 1 The device controller 12, the server device 14 and the operator terminal 16 shown can be Figure 2 The hardware structure shown is implemented by a computer. Figure 2 This is a hardware configuration diagram of an example of a computer 500 .

[0049] Figure 2 The computer 500 includes an input device 501, an output device 502, an external I / F 503, a RAM 504, a ROM 505, a CPU 506, a communication I / F 507, and an HDD 508, and the like, which are connected to one another via a bus B. The input device 501 and the output device 502 can be used in a manner that they are connected and used as needed.

[0050] The input device 501 is a keyboard, a mouse, a touch panel, or the like, and is used for an operator to input an operation signal. The output device 502 is a display or the like, and displays a processing result of the computer 500. The communication I / F 507 is an interface that connects the computer 500 to the networks 18 and 20 shown in Fig. 1. The HDD 508 is one example of a nonvolatile storage device that stores programs and data. Figure 1

[0051] The external I / F 503 is an interface with an external device. The computer 500 is capable of reading a recording medium 503a such as an SD (Secure Digital) card via the external I / F 503. The external I / F 503 can also write to the recording medium 503a such as an SD card via the external I / F 503.

[0052] The ROM 505 is one example of a nonvolatile semiconductor memory (storage device) that stores programs and data. The RAM 504 is one example of a volatile semiconductor memory (storage device) that temporarily holds programs and data. The CPU 506 is an arithmetic device that realizes control and functions of the entire computer 500 by reading programs and data from the storage device such as the ROM 505 or the HDD 508 onto the RAM 504 and executing processing.

[0053] Figure 1 The device controller 12, the server device 14, and the operator terminal 16 of the substrate processing system 1 shown in Fig. 1 realize various functions described later by executing programs by the computer 500. Figure 2

[0054] <Functional Configuration>

[0055] Hereinafter, an example of the information processing device that performs equipment difference analysis of a plurality of substrate processing devices 10 will be described as the device controller 12. In addition, the information processing device that performs equipment difference analysis of a plurality of substrate processing devices 10 can also be the server device 14 or the operator terminal 16. ​​

[0056] The device controller 12 of the substrate processing system 1 of the present embodiment is implemented by, for example, the functional modules illustrated in FIG. 8. Figure 3 Figure 3 is a functional block diagram of one example of the device controller 12 of the present embodiment. In addition, the functional block diagram of Figure 3 the functional block diagram of the device controller 12 of the present embodiment omits the illustration of structures not needed in the explanation of the present embodiment.

[0057] The device controller 12 implements the equipment difference calculation section 30, the reconstruction error matrix generation section 32, the training section 34, the data storage section 36, the display control section 38, the operation reception section 40, and the data set acquisition section 42 by executing a program for the device controller 12. The reconstruction error matrix generation section 32 has the dimension reduction model 50. The training section 34 has the dimension reduction model 52. The data storage section 36 has the reference data set storage section 54 and the analysis data set storage section 56.

[0058] The data set acquisition section 42 acquires a reference data set and stores it in the reference data set storage section 54. The reference data set contains sensor data output from the sensor 11 in the process in which the substrate processing device 10 serving as a reference executes a process according to a recipe. The substrate processing device 10 serving as a reference is, for example, a golden device that can guarantee perfect work.

[0059] In addition, the data set acquisition section 42 acquires an analysis data set and stores it in the analysis data set storage section 56. The analysis data set contains sensor data output from the sensor 11 in the process in which the substrate processing device 10 to be subjected to equipment difference analysis executes a process according to a recipe. The data storage section 36 has the reference data set storage section 54 that stores the reference data set and the analysis data set storage section 56 that stores the analysis data set.

[0060] The training section 34 trains the dimension reduction model 52 with the reference data set stored in the reference data set storage section 54. The dimension reduction model 52 projects data of a high-dimensional space into data of a low-dimensional space.

[0061] The dimension reduction model 52 is a PCA (Principal component analysis), a GPLVM (Gaussian Process Latent Variable Model), an MPPCA (Mixture Probabilistic Principal Component Analysis), a Kernel PCA, or a Probabilistic PCA.

[0062] ​The reconstruction error matrix generation section 32 inspects (verifies, checks) the reconstruction error of the analysis data set stored in the analysis data set storage section 56 using the dimension reduction model 50 trained by the training section 34. The reconstruction error refers to the difference between the original data and the data reconstructed from the low-dimensional space to the high-dimensional space after the original data is reduced from the high-dimensional space to the low-dimensional space. Also, the reconstruction error matrix generation section 32 inspects the reconstruction error and generates the reconstruction error matrix described later.

[0063] Referring to Figure 4 The processing of the training section 34 and the reconstruction error matrix generation section 32 will be described. Figure 4 is a diagram for explaining an example of the processing of the training section 34 and the reconstruction error matrix generation section 32.

[0064] In the training phase, the training section 34 trains the dimension reduction model 52 using the reference data set stored in the reference data set storage section 54. The training section 34 trains the dimension reduction model 52 by, for example, unsupervised learning.

[0065] In the inspection phase, the reconstruction error matrix generation section 32 inspects the reconstruction error of the analysis data set stored in the analysis data set storage section 56 using the dimension reduction model 50 trained by the training section 34.

[0066] Using Figure 5 The processing in the inspection phase will be further described. Figure 5 is a diagram for explaining an example of the processing in the inspection phase.

[0067] In step S1, the reconstruction error matrix generation section 32 reduces the original data (actual point) of the analysis data set from the high-dimensional space to the low-dimensional space using the dimension reduction model 50 trained by the training section 34. Figure 5 In step S2, the reconstruction error matrix generation section 32 reconstructs the data reduced in step S1 to the high-dimensional space using the dimension reduction model 50 trained by the training section 34.

[0068] In step S2, the reconstruction error matrix generation section 32 reconstructs the data reduced in step S1 to the high-dimensional space using the dimension reduction model 50 trained by the training section 34.

[0069] As for the position of the data reconstructed to the high-dimensional space, since the data is reconstructed to the original high-dimensional space using the dimension reduction model 50 trained using the reference data set, the position is the position where the data of the wafer processing apparatus 10 serving as a reference should exist. Thus, the distance between the position of the original data in the high-dimensional space and the position of the data obtained by reconstructing the data reduced in step S1 to the high-dimensional space becomes the equipment difference between the wafer processing apparatus 10 serving as a reference and the wafer processing apparatus 10 of the analysis data set.

[0070] In step S3, the reconstruction error matrix generation section 32 inspects the point obtained by reconstructing the data from the low-dimensional space to the high-dimensional space and the original data (actual point) of the analysis data set. Figure 5the reconstruction error between the actual points). In addition, data of a small frequency has a large reconstruction error because it has not learned a proper transformation to a low-dimensional space. Furthermore, in the dimension reduction model 50 trained by the training section 34, the reference data set with a change in the set value has been trained. For example, even in a process in which the same recipe is executed, the substrate processing apparatus 10 sometimes adjusts the set value.

[0071] Therefore, even for an analysis data set with a change in the set value, the dimension reduction model 50 trained by the training section 34 can check the reconstruction error other than the influence of the change in the set value. In this way, the reconstruction error checked by the checking stage indicates the device difference after absorbing the data variation due to the change in the set value.

[0072] The reconstruction error matrix generating section 32 generates, for example, the reconstruction error matrix 1000 shown in FIG. 10 by repeatedly performing the processing of the checking stage shown in FIG. 9. Figure 5 The reconstruction error matrix 1000 shown in FIG. 10 is a reconstruction error with a minimum unit of the reconstruction error of each step of the sensor 11 that detects the state of the substrate processing apparatus 10. The horizontal axis of the reconstruction error matrix 1000 indicates the run of the process. The vertical axis of the reconstruction error matrix 1000 indicates the step of the sensor 11, the sensor 11, and the group of the sensor 11. The reconstruction error matrix 1000 is generated for each substrate processing apparatus 10. Figure 6 Figure 6 is a structural diagram of one example of the reconstruction error matrix 1000.

[0073] Figure 6 The reconstruction error matrix 1000 shown in FIG. 10 is a reconstruction error with a minimum unit of the reconstruction error of each step of the sensor 11 that detects the state of the substrate processing apparatus 10. The horizontal axis of the reconstruction error matrix 1000 indicates the run of the process. The vertical axis of the reconstruction error matrix 1000 indicates the step of the sensor 11, the sensor 11, and the group of the sensor 11. The reconstruction error matrix 1000 is generated for each substrate processing apparatus 10.

[0074] Returning to Figure 3 , the device difference calculating section 30 calculates the magnitude of the reconstruction error of at least a part selected from the reconstruction error matrix 1000 generated by the reconstruction error matrix generating section 32, as the device difference.

[0075] The device difference calculating section 30 performs the calculation of the device difference in a hierarchical manner according to the range of the minimum unit of the reconstruction error selected from the plurality of minimum units of the reconstruction error included in the reconstruction error matrix 1000. Figure 7 is an explanatory diagram of one example of the reconstruction error matrix 1000 in which the range of the reconstruction error is selected.

[0076] Figure 7 The range 1010 of the reconstruction error of "Sensor Al" is a range of the minimum unit of the reconstruction error selected in the case where the device difference of the process run "6-8" in all steps of the group of the sensor 11 to which "Sensor Al" belongs is calculated. In the range 1010 of the reconstruction error of "Sensor Al", the minimum unit of the reconstruction error of the process run "6-8" in all steps of the group of the sensor 11 to which "Sensor Al" belongs is selected. Figure 7 ​In a case where the range 1010 of the reconstruction error of "Sensor Al" is selected, the device difference calculating section 30 can calculate the device difference of the process execution "6-8" of all steps of the group of the sensor 11 to which "Sensor Al" belongs by calculating the L2 norm as the magnitude of the reconstruction error of the minimum unit included in the range 1010.

[0077] Figure 7 The range 1012 of the reconstruction error of "Sensor Al" is a range of the reconstruction error of the minimum unit selected in a case where the device difference of all process executions of the step "1" of "Sensor Al" is calculated. In the case where the range 1012 of the reconstruction error of "Sensor Al" is selected, Figure 7 In a case where the range 1012 of the reconstruction error of "Sensor Al" is selected, the device difference calculating section 30 can calculate the device difference of all process executions of the step "1" of "Sensor Al" by calculating the L2 norm as the magnitude of the reconstruction error of the minimum unit included in the range 1012.

[0078] Further, in a case where the number of the sensors 11 belonging to the group, the number of the process executions within the data set are different, the magnitude cannot be directly compared due to the influence of the difference in the total number of the sensors 11 belonging to the group and the difference in the number of the process executions within the data set. Therefore, in the present embodiment, the device difference can be adjusted by the following formula (1) on the basis of taking into account the difference in the total number of the sensors 11 belonging to the group.

[0079] Adjusted device difference = L2 norm / number of sensors 11 belonging to the group x average number of sensors 11 of all groups (1)

[0080] In addition, in the present embodiment, the device difference can also be adjusted by the following formula (2) on the basis of taking into account the number of the process executions within the data set.

[0081] Adjusted device difference = L2 norm / number of process executions within the data set x average number of process executions of all data sets (2)

[0082] In addition, according to the reconstruction error matrix 1000 shown in Figure 6 and Figure 7 the number of steps of each sensor 11 and the summary type numbers such as "max", "mean", "min", "std" are the same, and thus comparison can also be made between different steps or sensors 11.

[0083] Further, according to the reconstruction error matrix 1000 shown in Figure 6 and Figure 7 as shown in Figure 8As shown, the reconstruction error of each step of the sensor 11 is calculated as a minimum unit of the reconstruction error, and the size of the reconstruction error of each step of the sensor 11, the entire steps of the sensor 11, the entire steps of the group of the sensor 11, and the entire steps of the substrate processing apparatus 10 is calculated as the device difference of each level.

[0084] Figure 8 is a diagram illustrating one example of the calculation of the device difference of each level. Figure 8 (A) of indicates the range of the minimum unit of the reconstruction error selected in the case where the size of the reconstruction error of the entire steps of the substrate processing apparatus 10 is calculated. Figure 8 (B) of indicates the range of the minimum unit of the reconstruction error selected in the case where the size of the reconstruction error of the entire steps of the group of the sensor 11 is calculated.

[0085] Figure 8 (C) of indicates the range of the minimum unit of the reconstruction error selected in the case where the size of the reconstruction error of the entire steps of the sensor 11 is calculated. Figure 8 (D) of indicates the range of the minimum unit of the reconstruction error selected in the case where the size of the reconstruction error of each step of the sensor 11 is calculated.

[0086] Returning to Figure 3 , the display control section 38 causes the output device 502 such as a display to display the device difference calculated by the device difference calculation section 30. The operation receiving section 40 receives various operations from the operator. For example, the operation receiving section 40 receives Figure 7 the selection of the range 1010 or 1012 of the reconstruction error as shown.

[0087] <Process>

[0088] Figure 9 is a flowchart indicating one example of the process of the device difference analysis method performed by the substrate processing system 1 of the present embodiment.

[0089] In step S10, the training section 34 of the apparatus controller 12 trains the dimension reduction model 52 with the reference data set stored in the reference data set storage section 54. The dimension reduction model 52 can utilize PCA, GPLVM, MPPCA, Kernel PCA, or Probabilistic PCA, but from the viewpoint of precision, it is preferable to utilize PCA and GPLVM.

[0090] In step S12, the data set acquisition section 42 acquires the analysis data set and stores it in the analysis data set storage section 56.

[0091] In step S14, the reconstruction error matrix generating section 32 uses the reduced dimension model 50 trained by the training section 34 to reduce the original data of the analysis data set from the high-dimensional space to the low-dimensional space, and reconstruct the data from the low-dimensional space to the high-dimensional space.

[0092] In step S16, the reconstruction error matrix generating section 32 inspects the reconstruction error between the original data and the data reconstructed from the low-dimensional space to the high-dimensional space, and generates the reconstruction error matrix 1000 shown in Fig. 10. Figure 6

[0093] In step S18, the device difference calculating section 30 determines whether or not the reconstruction error is selected from the reconstruction error matrix 1000 generated as shown in the range 1010 or 1012 of the reconstruction error. If the reconstruction error is not selected, the device controller 12 ends the processing of the flowchart. Figure 7 Figure 9

[0094] If the reconstruction error is selected, the processing proceeds to step S20, and the device difference calculating section 30 calculates the magnitude of the reconstruction error of the range 1010 or 1012 of the reconstruction error selected from the reconstruction error matrix 1000, for example, as the device difference. For example, as the magnitude of the reconstruction error of the range 1010 or 1012 of the reconstruction error, the device difference calculating section 30 calculates the L2 norm of the plurality of reconstruction errors included in the range 1010 or 1012. The device difference calculating section 30 can reduce the calculation cost as long as the L2 norm of the plurality of reconstruction errors is calculated. Figure 7 Figure 7

[0095] In step S22, the display control section 38 causes the display device or the like output device 502 to display the device difference calculated by the device difference calculating section 30. The operator can confirm the device difference displayed on the display device or the like output device 502.

[0096] <Summary>

[0097] In the existing device difference analysis, it is assumed that the behavior of the sensor data in the process is the same if the set values of the recipes are the same, and it is determined that there is a device difference in a case where the behavior of the sensor data in the process is different. However, even in the same recipe, the set values are sometimes adjusted. Therefore, in the existing device difference analysis, the difference in the set values is sometimes detected as a device difference, and the true device difference cannot be analyzed.

[0098] ​​​​​In addition, in the existing device difference analysis, if it is desired to hierarchically calculate the device difference of each step of the sensor 11, the entire steps of the sensor 11, the entire steps of the group of the sensor 11, and the entire steps of the substrate processing apparatus 10, a respective model is required for each hierarchy. The existing device difference analysis cannot compare the device difference of each step of the sensor 11, the entire steps of the sensor 11, the entire steps of the group of the sensor 11, and the entire steps of the substrate processing apparatus 10 within or between hierarchies.

[0099] The comparison of the device difference within the hierarchy is, for example, a comparison of the device difference of "Sensor Al" belonging to a group of different sensors 11 and the device difference of "Sensor Bl". The comparison of the device difference between the hierarchies is, for example, a comparison of the device difference of the substrate processing apparatus 10 and the device difference of "Sensor Al".

[0100] In the present embodiment, one dimension reduction model 52 is trained with the reference data set as a whole, and, for example, the device difference of each hierarchy is calculated by selecting the range of the reconstruction error of the minimum unit of the reconstruction error matrix 1000 shown in FIG. 10. Figure 6 The L2 norm is calculated for the range of the reconstruction error of the minimum unit of the reconstruction error matrix 1000 shown in FIG. 10 to obtain the device difference of each hierarchy, and thus the device difference between or within the hierarchies can be directly compared.

[0101] Thus, in the present embodiment, even if a log such as a tracking log in which there is a change in the set value, the device difference analysis can be performed, and thus the range of the log in which the device difference analysis can be performed is expanded. For example, the change in the set value is a change in the set value that does not affect the process. For example, in the present embodiment, in the device difference analysis of the log in which there is a change in the set value, in a case where the temperature is finely adjusted in order to optimize the process result, the difference in the temperature is not analyzed as the device difference.

[0102] In addition, in the present embodiment, the device difference between or within the hierarchies can be examined using one dimension reduction model 50, and thus the device difference from the entire substrate processing apparatus 10 to the details can be directly compared. Furthermore, in the present embodiment, the device difference between or within the hierarchies can be examined using one dimension reduction model 50, and thus the calculation cost can be reduced.

[0103] In the above, the present embodiment trains the dimension reduction model 52 using the reference data set of the substrate processing apparatus 10 as a reference (normal data as a reference), and examines the reconstruction error of the analysis data set (other data) using the trained dimension reduction model 50. The reconstruction error in this case becomes an index for evaluating whether the other data is normal.

[0104] The reconstruction error is the difference between the original data and the reconstructed data. The reconstruction error of other data having similar characteristics to normal data tends to become smaller. On the other hand, the reconstruction error of abnormal data containing a device difference or noise is likely to become larger.

[0105] Therefore, in the present embodiment, the dimension reduction model 52 is trained using the reference data set of the reference substrate processing apparatus 10 as a reference, and the trained dimension reduction model 50 is used to check the reconstruction error of the check data set, which is used as the device difference.

[0106] According to the substrate processing system 1 of the present embodiment, it is possible to provide a technique for further improving the accuracy of the device difference analysis of the substrate processing apparatus 10.

[0107] The preferred embodiments of the present embodiment have been described in detail above, but the present embodiment is not limited to the above-described embodiments, and various modifications and substitutions can be made to the above-described embodiments without departing from the scope of the present embodiment.

[0108] The substrate processing apparatus 10 of the present embodiment can also be applied to any type of apparatus among an atomic layer deposition (ALD) apparatus, a capacitively coupled plasma (CCP), an inductively coupled plasma (ICP), a radial line slot antenna (RLSA), an electron cyclotron resonance plasma (ECR), or a helicon wave plasma (HWP). The substrate processing apparatus 10 of the present embodiment can also be applied to a CVD (chemical vapor deposition) apparatus, an oxidation annealing apparatus.

[0109] The substrate processing system 1 of the present embodiment is not limited to Figure 1 the structure shown, and various system structure examples certainly exist depending on the use, purpose. The substrate processing apparatus 10 of the present embodiment can be applied to any apparatus among a single-wafer apparatus that processes a substrate one by one, a batch processing apparatus that processes a plurality of substrates at once, and a semi-batch processing apparatus. The process performed by the substrate processing apparatus 10 of the present embodiment can be exemplified by a film formation process or an etching process, and the like.

Claims

1.An information processing apparatus, characterized by comprising: an apparatus difference analysis unit that analyzes apparatus differences of a plurality of substrate processing apparatuses, a reconstruction error matrix generation unit that inspects reconstruction errors of data sets of the plurality of substrate processing apparatuses using a reduced dimension model trained using a data set as a reference, and generates a reconstruction error matrix of the plurality of substrate processing apparatuses, an apparatus difference calculation unit that calculates magnitudes of at least a part of the reconstruction errors selected from a plurality of the reconstruction errors included in the reconstruction error matrix as apparatus differences, and a display control unit that causes a display apparatus to display the calculated apparatus differences. 2.The information processing apparatus according to claim 1, characterized in that: the reconstruction error matrix generation unit inspects reconstruction errors of a minimum unit, and generates the reconstruction error matrix for each of the substrate processing apparatuses, the apparatus difference calculation unit performs calculation of the apparatus differences in a hierarchical manner according to a range of the reconstruction errors of the minimum unit selected from a plurality of the reconstruction errors of the minimum unit included in the reconstruction error matrix. 3.The information processing apparatus according to claim 2, characterized in that: the apparatus difference calculation unit calculates magnitudes of the reconstruction errors of each step of a sensor that detects a state of the plurality of substrate processing apparatuses, all steps of the sensor, all steps of a group of the sensor, and all steps of the substrate processing apparatuses, as the apparatus differences of each of the hierarchical levels, as the reconstruction errors of the minimum unit. 4.The information processing apparatus according to claim 3, characterized in that: the data set is held for each substrate processing performed, and a set value and log data of each sensor of the substrate processing apparatus during a process of performing substrate processing according to the set value are associated with each other in the data set, the log data includes a case where the set value is changed for each of the substrate processing performed. 5.The information processing apparatus according to any one of claims 1 to 4, characterized in that: the apparatus difference calculation unit calculates an L2 norm as the magnitude of the reconstruction error. 6.The information processing apparatus according to any one of claims 1 to 4, characterized in that: the reduced dimension model is PCA, GPLVM, MPPCA, kernel PCA, or probabilistic PCA. 7.An apparatus difference analysis method, characterized by comprising: using a reduced dimension model trained using a data set as a reference, inspecting reconstruction errors of data sets of a plurality of substrate processing apparatuses, and generating a reconstruction error matrix of the plurality of substrate processing apparatuses, calculating magnitudes of at least a part of the reconstruction errors selected from a plurality of the reconstruction errors included in the reconstruction error matrix as apparatus differences, and causing a display apparatus to display the calculated apparatus differences. 8.A substrate processing apparatus, characterized by comprising: ​ An information processing apparatus having any one of claims 1 to 4.

Images