Data processing device, data processing system, data processing method and data processing program

Abstract

This invention provides a data processing apparatus, system, method, and program for compressing and imagerizing multi-wavelength time-series data to suppress the omission of feature data. The data processing apparatus of this invention includes: a preprocessing unit that normalizes multi-wavelength time-series data using predetermined reference data, thereby generating normalized data; an extraction unit that divides the normalized data into a plurality of regions for each predetermined time range and wavelength range, and extracts outliers in each region as representative values; and a generation unit that converts the representative values ​​of each region into color data to generate image data.

Description

Technical Field

[0001] This invention relates to a data processing device, a data processing system, a data processing method, and a data processing program. Prior Technology

[0002] Measurement data obtained during semiconductor manufacturing processes, such as data obtained using emission spectroscopy or mass spectrometry (so-called multi-wavelength time-series data), typically have high resolution and large data volume. This incurs significant management costs. Furthermore, in the case of multi-wavelength time-series data, users often find it difficult to directly determine whether any anomalies have occurred.

[0003] Industry experts believe that, for example, if multi-wavelength time-series data is visualized and displayed, users can easily determine whether any anomalies have occurred. Furthermore, they believe that compressing and managing the visualized data will also reduce management costs.

[0004] However, in the case of multi-wavelength time series data, when displaying or compressing imaged data, characteristic data indicating anomalies may be missing. [Previous Technical Documents] [Patent Literature]

[0005] [Patent Document 1] Japanese Patent Application Publication No. 2020-65013 [Patent Document 2] Japanese Patent Application Publication No. 2005-217448 [Patent Document 3] Japanese Patent No. 5315025 Summary of the Invention

[0006] [The problem the invention aims to solve]

[0007] The present invention provides a data processing apparatus, data processing system, data processing method and data processing program for compressing and image-encoding multi-wavelength time series data to suppress the omission of feature data. [Technical means to solve the problem]

[0008] One embodiment of the data processing apparatus of the present invention has, for example, the following configuration: That is, it has: The preprocessing unit uses prescribed reference data to normalize multi-wavelength time series data, thereby generating normalized data; The extraction unit divides the aforementioned normalized data into multiple regions for each specified time range and wavelength range, and extracts outliers from each region as representative values; and The generation unit converts the representative values ​​of the above regions into color data and generates image data. [Effects of the Invention]

[0009] According to the present invention, a data processing apparatus, data processing system, data processing method, and data processing program are provided for compressing and imagening multi-wavelength time series data to suppress the omission of feature data. Simple Explanation of the Diagram

[0010] Figure 1 is the first figure showing an example of the system structure of a data processing system. Figure 2 is a diagram illustrating one example of a semiconductor manufacturing process. Figure 3 is a diagram showing one example of the hardware configuration of a data processing device. Figure 4 shows an example of OES (Optical Emission Spectroscopy) data. Figure 5 is a diagram showing a specific example of the processing performed in the preprocessing unit. Figure 6 is a diagram showing a specific example of the processing performed by the compression section. Figure 7 shows a specific example of image data. Figure 8 is a flowchart illustrating the process of image processing. Figure 9A is the second figure showing an example of the system configuration of a data processing system. Figure 9B is the third figure showing an example of the system configuration of a data processing system. Figure 10 is an example of learning materials. Figure 11 is a diagram illustrating a specific example of the learning processes performed by the Learning Department. Figure 12 is a diagram showing a specific example of the inference process performed by the inference unit. Figure 13 is a flowchart showing the process of anomaly detection and processing. Figure 14 is the fourth figure showing an example of the system structure of a data processing system. Implementation

[0011] Hereinafter, various embodiments will be described with reference to the accompanying drawings. Furthermore, in this specification and drawings, constituent elements having substantially the same functional configuration are omitted from repeated descriptions by using the same symbols.

[0012] [First Implementation] <System Composition of Data Processing Systems> First, the system configuration of the data processing system according to the first embodiment will be described. Figure 1 is the first figure showing an example of the system configuration of the data processing system. As shown in Figure 1, the data processing system 100 includes a semiconductor manufacturing process, an emission spectrum analysis device 140, and a data processing device 150.

[0013] The semiconductor manufacturing process processes the object (pre-processed wafer 110) in a designated processing space 120 to generate the result (post-processed wafer 130). Furthermore, the pre-processed wafer 110 refers to the wafer (substrate) before being processed in the processing space 120, and the post-processed wafer 130 refers to the wafer (substrate) after being processed in the processing space 120.

[0014] The emission spectroscopy analyzer 140 measures OES data in the processing space 120 in conjunction with the processing of the pre-processing wafer 110. The OES data is "multi-wavelength time series data", which contains luminescence intensity data for each time period corresponding to the number of wavelengths measured.

[0015] The OES data measured by the emission spectroscopy analyzer 140 is stored in the OES data storage unit 155 of the data processing unit 150.

[0016] A data processing program is installed in the data processing device 150. By executing the program, the data processing device 150 functions as a preprocessing unit 151, a compression unit 152, an image processing unit 153, and a display control unit 154.

[0017] The preprocessing unit 151 reads the OES data from the OES data storage unit 155 and performs preprocessing (e.g., normalization) using pre-determined reference data. Furthermore, the preprocessing unit 151 notifies the compression unit 152 of the preprocessed OES data.

[0018] The compression unit 152 compresses the preprocessed OES data and notifies the imaging unit 153 of the compressed OES data. Specifically, the compression unit 152 divides the preprocessed OES data into multiple regions of a predetermined size (predetermined time range and predetermined wavelength range), extracts outliers contained in each region as representative values ​​for each region, and thereby compresses the preprocessed OES data. Furthermore, outliers refer to values ​​in each region that statistically deviate significantly from other values. In this way, by retaining the outliers contained in each region and compressing the preprocessed OES data, the omission of characteristic data (characteristic data indicating the occurrence of anomalies) contained in the OES data can be suppressed.

[0019] The image processing unit 153 converts the compressed OES data, as instructed by the compression unit 152, into color data, thereby generating image data. Furthermore, the image processing unit 153 stores the generated image data in the image data storage unit 156 and notifies the display control unit 154. By storing image data in the image data storage unit 156, the data volume is significantly reduced compared to OES data, thus reducing management costs.

[0020] The display control unit 154 controls the display of image data generated by the image processing unit 153 on a display device (not shown). As described above, the image data generated by the image processing unit 153 is compressed, so regardless of whether the number of pixels on the display device is limited, it can be displayed on the display device without further processing. That is, when displaying on the display device, situations such as image data reduction and feature data loss can be avoided.

[0021] Furthermore, since the image data generated by the image processing unit 153 suppresses the omission of feature data, the user can visually determine whether there are any abnormalities in the image data displayed on the display device.

[0022] Processing Space in Semiconductor Manufacturing Processes Next, the processing space specified in the semiconductor manufacturing process will be explained. Figure 2 is a diagram showing an example of a semiconductor manufacturing process. As shown in Figure 2, the semiconductor manufacturing process 200 has a plurality of chambers as an example of processing spaces. In the example of Figure 2, the semiconductor manufacturing process 200 has three chambers, denoted by symbols 121 (named "Cavity A") to 123 (named "Cavity C"), in which the wafer 110 before processing is processed.

[0023] Furthermore, in the semiconductor manufacturing process 200, the aforementioned emission spectroscopy analysis device 140 is installed in each chamber, and OES data is measured in each chamber. However, for the sake of simplicity, the following explanation will focus on the case where the OES data measured in one chamber is displayed. Furthermore, this one chamber will be described as chamber A. Moreover, chamber A will be described, for example, as a chamber of an etching apparatus.

[0024] <Hardware Components of Data Processing Devices> Next, the hardware configuration of the data processing device 150 will be explained. Figure 3 is a diagram showing an example of the hardware configuration of the data processing device. As shown in Figure 3, the data processing device 150 includes a CPU (Central Processing Unit) 301, a ROM (Read Only Memory) 302, and a RAM (Random Access Memory) 303. Furthermore, the data processing device 150 includes a GPU (Graphics Processing Unit) 304. Moreover, the processors (processing circuits) such as the CPU 301 and GPU 304, and the memories such as the ROM 302 and RAM 303, constitute what is known as a computer.

[0025] Furthermore, the data processing device 150 includes an auxiliary memory device 305, a display device 306, an operation device 307, an I / F (Interface) device 308, and a drive device 309. Moreover, the hardware components of the data processing device 150 are interconnected via a bus 310.

[0026] CPU301 is a computing device that executes various programs (such as data processing programs) installed in auxiliary memory device 305.

[0027] ROM302 is a non-volatile memory that functions as the main memory device. ROM302 stores various programs and data required for CPU301 to execute various programs installed in auxiliary memory 305. Specifically, ROM302 stores boot programs such as BIOS (Basic Input / Output System) or EFI (Extensible Firmware Interface).

[0028] RAM303 is a volatile memory such as DRAM (Dynamic Random Access Memory) or SRAM (Static Random Access Memory), which functions as the main memory device. RAM303 provides a working area for the CPU301 to execute various programs installed in the auxiliary memory device 305.

[0029] The GPU 304 is a computing device for image processing. In this embodiment, when the CPU 301 executes the data processing program, it performs high-speed calculations on OES data using parallel processing. Furthermore, the GPU 304 is equipped with internal memory (GPU memory) to temporarily store the information required for parallel processing of OES data.

[0030] The auxiliary memory device 305 stores various programs or data used by the CPU 301 when executing various programs. For example, the OES data storage unit 155 and the image data storage unit 156 are implemented in the auxiliary memory device 305.

[0031] Display device 306 is, for example, a display device that displays image data generated by the image processing unit 153. Operation device 307 is an input device used by the user of data processing device 150 to input various instructions to data processing device 150. I / F device 308 is a connection device used to connect to a network (not shown) and to transmit and receive data with other devices (such as emission spectroscopy analyzers).

[0032] The drive unit 309 is a device used to house the recording medium 320. The recording medium 320 mentioned here includes media that record information optically, electrically, or magnetically, such as CD-ROM (Compact Disc-Read Only Memory), floppy disks, and magneto-optical disks. Furthermore, the recording medium 320 may also include semiconductor memory such as ROM and flash memory that record information electrically.

[0033] Furthermore, the various programs installed in the auxiliary memory device 305 can be installed, for example, by placing the distributed recording medium 320 into the drive device 309, and then using the drive device 309 to read the various programs recorded in the recording medium 320 to install them. Alternatively, the various programs installed in the auxiliary memory device 305 can also be downloaded from a network (not shown) for installation.

[0034] <Specific Example of OES Data> Next, a specific example of the OES data stored in the OES data storage unit 155 of the data processing device 150 as measured by the emission spectrum analysis device 140 will be described. FIG. 4 is a diagram showing an example of OES data. As shown in FIG. 4, the OES data 410 includes a group of emission intensity data for each time when measuring each wavelength contained in the visible light wavelength range (200 [nm] to 800 [nm]) every 0.5 [nm]. Furthermore, in the OES data 410, the horizontal axis represents time, and the vertical axis represents the emission intensity of each wavelength.

[0035] In the case of the OES data 410 in FIG. 4, for example, the topmost curve graph represents the emission intensity data at each time with a wavelength = 200 [nm], and the second-layer curve graph represents the emission intensity data at each time with a wavelength = 200.5 [nm]. Also, the third-layer curve graph represents the emission intensity data at each time with a wavelength = 201 [nm].

[0036] Furthermore, the time length of the OES data 410 is set, for example, as the processing time (processing time for one wafer) for processing the pre-process wafer 110 in chamber A. Or, in the case of processing the pre-process wafer 110 in chamber A under a plurality of processing steps, the time length of the OES data 410 can also be the processing time of a part of the processing steps. Or, in the case of performing a part of the processing steps on the pre-process wafer 110 by a plurality of process parameters in chamber A, the time length of the OES data 410 can also be the time for performing the processing by a part of the process parameters.

[0037] On the other hand, in FIG. 4, the OES data 420 represents the situation where the horizontal axis is set as wavelength and the vertical axis is set as time, and the emission intensity data for each point (each time, each wavelength) is arranged. In the case of the OES data 420, 1201 points of emission intensity data are arranged in the horizontal axis direction. Also, in the case of the OES data 420, when the sampling period of the emission intensity data is 0.1 second and the processing time (processing time for one wafer) for processing the pre-process wafer 110 in chamber A is 300 seconds, 3000 points of emission intensity data are arranged in the vertical axis direction.

[0038] Therefore, when it is assumed that the emission intensity data of each point of the OES data 420 is to be converted into color data and the generated image data is to be displayed on the display device 306, a display device with a pixel number of 1201 pixels in the horizontal direction and 3000 pixels in the vertical direction is required.

[0039] On the other hand, when the number of pixels in the display device 306 is limited, the OES data 420 may be reduced and displayed, potentially resulting in missing feature data indicating an anomaly. Therefore, in the data processing apparatus 150 of this embodiment, as described above, the OES data is preprocessed and then compressed, and the compressed OES data is visualized to generate image data. Thus, regardless of whether the number of pixels in the display device 306 is limited, situations such as reduced image data and missing feature data can be avoided when displaying on the display device 306.

[0040] <Specific examples of the processing performed by the preprocessing department> Next, a specific example of the processing of OES data 420 performed by the preprocessing unit 151 of the data processing apparatus 150 will be described. Figure 5 is a diagram showing a specific example of the processing performed by the preprocessing unit. As shown in Figure 5, the preprocessing unit 151 includes a normalization processing unit 510.

[0041] The normalization processing unit 510 reads the OES data 420 stored in the OES data storage unit 155 and uses the reference data to normalize the luminous intensity data of each point contained in the OES data 420.

[0042] Furthermore, reference data can be used, for example, the average value of the luminous intensity data at each wavelength of the OES data measured when processing a reference wafer (a wafer determined to be good). In this case, the difference from the wafer determined to be good can be shown, thus making it easy to determine whether any abnormalities have occurred when generating image data.

[0043] Alternatively, the reference data can be the average of the luminescence intensity data at the wavelength measured in the OES data. In this case, during the pre-processing of the wafer, it can be shown which wavelength has a relatively stronger luminescence intensity, making it easier to determine whether any abnormalities have occurred when generating image data.

[0044] In Figure 5, the preprocessed OES data 500 is an example of preprocessed OES data in which the luminous intensity data of each point has been normalized by the normalization processing unit 510. Also, in Figure 5, the normalized data 501 of each point represents the normalized data of each point contained in a region of a specified size (3 points in the horizontal direction and 3 points in the vertical direction) within the preprocessed OES data 500.

[0045] As shown in the normalized data 501 for each point, by dividing the luminous intensity data of each point by the baseline data, the value of the normalized data becomes approximately close to "1.0".

[0046] <Specific examples of the processing performed by the compression section> Next, a specific example of the processing of the pre-processed OES data 500 performed by the compression unit 152 of the data processing apparatus 150 will be described. Figure 6 is a diagram showing a specific example of the processing performed by the compression unit. As shown in Figure 6, the compression unit 152 has an average value calculation unit 610 and a representative value extraction unit 620.

[0047] The average value calculation unit 610 divides the normalized data of each point contained in the preprocessed OES data 500 into multiple regions of a specified size (e.g., 3 points in the horizontal direction and 3 points in the vertical direction), and calculates the average value of the normalized data of each point in each region.

[0048] The representative value extraction unit 620 compares the difference between the average value calculated in each region and the value of the normalized data (e.g., 9 points of normalized data) contained in each region, and extracts the value of the normalized data with the largest difference (i.e., outlier) as the representative value of each region.

[0049] The example in Figure 6 shows the case where the average calculation unit 610 calculates "0.99" as the average of the 9 points of normalized data 501 contained in a region of a specified size. Also, the example in Figure 6 shows the case where the representative value extraction unit 620 extracts the value of the normalized data 501 containing the 9 points of normalized data 501 contained in a region of a specified size (outlier "1.5"), which has the largest difference from the average ("0.99"), as the representative value 601.

[0050] Thus, representative values ​​are extracted for each region of a specified size (3 points on the horizontal axis and 3 points on the vertical axis). This compresses the number of points in the normalized data contained in the preprocessed OES data 500 to 1 / 3 on the horizontal axis and 1 / 3 on the vertical axis. The result, as shown in Figure 6, is an arrangement of... Regularized data with 1201 points on the horizontal axis and 3000 points on the vertical axis. After preprocessing, the OES data 500 will be compressed into a format with... Compressed data with 400 points on the horizontal axis and 1000 points on the vertical axis. The compressed OES data is 600, which can reduce management costs.

[0051] <Specific Examples of Image Data> Next, a specific example of image data generated by the image processing unit 153 of the data processing device 150 will be described. Figure 7 is a diagram showing a specific example of image data.

[0052] In Figure 7, image data 600' represents a specific example of a two-dimensional image generated by converting the compressed data of each point of the compressed OES data 600 notified by the compression unit 152 into color data. In image data 600', the horizontal axis represents wavelength, and the vertical axis represents time. Furthermore, the color difference between each point represents the difference in the value of the compressed data. The example in Figure 7 shows the case where the point with the largest compressed data value is converted to red, the point with the average compressed data value is converted to green, and the point with the smallest compressed data value is converted to blue. It also shows the case where the point with the compressed data value between the largest and average values ​​is converted to a color between red and green on the color wheel, and the point with the compressed data value between the average and the smallest values ​​is converted to a color between green and blue on the color wheel. However, the color assignment for each value of the compressed data is arbitrary, and other assignment methods can also be used to assign each color.

[0053] In the case of image data 600', the omission of feature data is suppressed. Therefore, the user can visually determine whether there are any abnormalities in the image data 600' displayed on the display device 306.

[0054] Furthermore, as shown in Figure 7, in the image processing unit 153, when a specified area 701 of image data 600' is designated, the image data of the specified area 701 can also be magnified to display magnified image data 702. In this way, the user can also observe image data 600' from above, and can also magnify a specific wavelength range and a specific time range for observation.

[0055] Furthermore, although not shown in Figure 7, the compressed OES data 600 can also be displayed as three-dimensional image data. For example, the horizontal axis can be set to wavelength, the depth axis to time, and the height axis to the value of the compressed data, thus displaying it in three dimensions. Moreover, in this case, the differences in the values ​​of the compressed data can also be represented by color differences.

[0056] Furthermore, although not shown in Figure 7, it can also be displayed in image data 600' along with the excited species (molecules in the excited state) corresponding to the wavelength on the horizontal axis. In this way, the user can deduce the cause of the anomaly.

[0057] <The process of image processing performed by the data processing device> Next, the image processing of OES data performed by the data processing device 150 will be explained. Figure 8 is a flowchart showing the image processing flow.

[0058] In step S801, the data processing device 150 obtains OES data from the emission spectroscopy analyzer 140 and stores it in the OES data storage unit 155.

[0059] In step S802, the preprocessing unit 151 of the data processing device 150 reads the OES data from the OES data storage unit 155, and divides the luminous intensity data of each point by the reference data to perform preprocessing.

[0060] In step S803, the compression unit 152 of the data processing device 150 divides the normalized data of each point contained in the preprocessed OES data into a plurality of regions of a predetermined size. Furthermore, the compression unit 152 of the data processing device 150 extracts the value of the largest normalized data (outlier) as the difference between the value of the normalized data and the average value calculated for each region of the predetermined size, thereby compressing the preprocessed OES data.

[0061] In step S804, the image processing unit 153 of the data processing device 150 converts the compressed data of each point of the compressed OES data into color data, thereby generating image data.

[0062] In step S805, the image processing unit 153 of the data processing device 150 displays the image data of the compressed OES data.

[0063] Summary As can be clearly seen from the above description, the data processing device 150 of the first embodiment... • It has a preprocessing unit that uses specified reference data to remove the luminous intensity data of each point in the OES data, and normalizes the luminous intensity data of each point in the OES data, thereby generating normalized data. • It has a compression unit that divides the preprocessed OES data into multiple regions for each specified time range and a specified number of wavelength ranges, and extracts the value of the largest normalized data (outlier) as the representative value of each region, based on the difference between the value of each region and the average value of each region. It has a generation unit that converts the representative values ​​of each region into color data to generate image data.

[0064] Thus, when the OES data is visualized, it is normalized and compressed by extracting outliers for each region of a specified size. This allows the OES data to be compressed without missing any of the feature data contained in the OES data.

[0065] That is, according to the first embodiment, a data processing apparatus, data processing system, data processing method and data processing program can be provided to compress and image OES data by suppressing the omission of feature data.

[0066] [Second Implementation] In the first embodiment described above, a configuration was explained that compresses and visualizes OES data to reduce management costs and enable users to determine whether anomalies have occurred, thereby suppressing the loss of feature data. In contrast, in the second embodiment, a configuration is explained that automatically determines whether anomalies have occurred using visualized OES data. Furthermore, the configuration that automatically determines whether anomalies have occurred using visualized OES data includes, for example: • Determine whether the corresponding wafer is a good product that has undergone normal processing or contains defective components; • Determine whether the corresponding wafer is a good product that has undergone normal processing or a product that is not a good product (at least not a good product that has undergone normal processing); • Determine whether the corresponding wafer is a good product that has been processed normally or a product that contains an anomaly of one of a plurality of modes; Any of the following configurations. In the second embodiment, a configuration that automatically determines whether a corresponding wafer is a good product that has undergone normal processing or an anomaly containing one of a plurality of modes will be described. Hereinafter, the second embodiment will be described focusing on its differences from the first embodiment described above.

[0067] <System Composition of Data Processing Systems> First, the system configuration of the data processing system according to the second embodiment will be described using Figures 9A and 9B. Figure 9A is the second figure showing an example of the system configuration of the data processing system. Specifically, Figure 9A shows an example of the system configuration of the data processing system 900 during the "learning period," which is the period during which the system learns the correspondence between the image data generated by image processing of OES data and the corresponding process parameters, and the processing results of the processed wafer.

[0068] The difference between the data processing system 100 shown in Figure 1 and the data processing system 900 is that the data processing device 920 has a learning unit 921.

[0069] In the data processing system 900, the data processing device 920 obtains processing result information about the processed wafer 130. The processing result information can assume... • This indicates whether the processed wafer 130 is a good product that has undergone normal processing or contains defective wafers; • This indicates whether the processed wafer 130 is a good product that has undergone normal processing or not (at least not a good product that has undergone normal processing); • This indicates whether the processed wafer 130 is a good product that has undergone normal processing or contains information about an anomaly belonging to one of a plurality of modes; Various variations exist, but in this embodiment, we will explain whether the information used indicates a good product that has been processed normally when wafer 130 is processed, or whether it includes information about an abnormal product belonging to one of the plurality of modes. Furthermore, the aforementioned information that can be included in the processing result information may also be generated based on information such as whether the output of the processed wafer is a good product or a defective product when inspected by the inspection device.

[0070] In the data processing device 920, the obtained processing result information, along with the corresponding process parameters and the corresponding image data, are stored as learning data in the learning data storage unit 923.

[0071] The learning unit 921 of the data processing device 920 has an anomaly detection model that takes image data and process parameters as input and outputs the processed wafer processing result information.

[0072] Subsequently, the data processing device 920 reads the learning data from the learning data storage unit 923 and performs learning processing on the anomaly detection model. Specifically, the data processing device 920 inputs process parameters and image data into the anomaly detection model, updates the model parameters of the anomaly detection model, and makes the output of the anomaly detection model closer to the corresponding processing result information.

[0073] On the other hand, Figure 9B is the third figure showing an example of the system configuration of the data processing system. Specifically, Figure 9B shows an example of the system configuration of the data processing system 900' in the "inference period", which infers the processing result information of the processed wafer 130 based on the image data generated by image processing of OES data and the corresponding process parameters.

[0074] The difference between the data processing system 100 shown in Figure 1 and the data processing system 900' is that the data processing device 920 has an inference unit 922.

[0075] The inference unit 922 of the data processing apparatus 920 has a learned anomaly detection model generated through learning processing by the learning unit 921. The inference unit 922 inputs image data and process parameters into the learned anomaly detection model, thereby inferring the corresponding processing result information of the processed wafer 130, and outputs the inference result. Furthermore, the inference result output by the inference unit 922 can contain... • This indicates whether the processed wafer 130 is a good product that has undergone normal processing or contains defective wafers; • This indicates whether the processed wafer 130 is a good product that has undergone normal processing or not (at least not a good product that has undergone normal processing); • This indicates whether the processed wafer 130 is a good product that has undergone normal processing or contains information about an anomaly belonging to one of a plurality of modes; Various variations exist, but in this embodiment, we will explain whether the output indicates that the processed wafer 130 is a good product that has been processed normally, or whether it contains information about an anomaly belonging to one of the plurality of modes.

[0076] <Specific Examples of Learning Materials> Next, a specific example of the learning data stored in the learning data storage unit 923 will be explained. As shown in Figure 10, the learning data 1000 includes "device", "process parameters", "wafer", "image data" and "good / includes abnormalities" as information items.

[0077] The "device" stores the name of the chamber in which the wafer 110 was processed before processing. The "process parameters" stores a process parameter identification code, which identifies the process parameters used when the wafer 110 was processed in chamber A before processing.

[0078] The "wafer" stores a wafer identification code that identifies the unprocessed wafer in chamber A. The "image data" stores an image data identification code that identifies the image data generated by the imaging unit 153 by imaging the OES data measured during the processing of the corresponding unprocessed wafer in chamber A.

[0079] Furthermore, the image data used in the learning materials is preprocessed using the average value of the luminous intensity data of each wavelength of the OES data measured when the wafer is judged to be good as the reference data. Also, the image data used in the learning materials is image data of the processing time (processing time of one wafer) of the wafer 110 being processed in chamber A before processing.

[0080] The "Good / Contains Anomalies" field stores processing result information for the processed wafer 130, generated by processing the corresponding pre-processed wafer in chamber A. Specifically, it stores information indicating whether the processed wafer 130 is a good product that has undergone normal processing, or whether it contains an anomaly belonging to one of a plurality of modes.

[0081] Figure 10 illustrates the process of using "Process Parameter 1" in "Cavity A" to process "Wafer 1" to "Wafer 3" respectively, thereby generating "Image Data 1" to "Image Data 3" and outputting the processing result information ("Good").

[0082] Furthermore, the example in Figure 10 illustrates the use of "process parameter 1" in "chamber A" to process "wafer 4", thereby generating "image data 4" and outputting processing result information ("contains anomaly (mode a)").

[0083] Furthermore, the example in Figure 10 shows the process of using "process parameter 2" in "chamber A" to process "wafers 5, 7, and 8", thereby generating "image data 5, 7, and 8" and outputting the processing result information ("good").

[0084] Furthermore, the example in Figure 10 illustrates the use of "process parameter 2" in "chamber A" to process "wafers 6 and 9", thereby generating "image data 6 and 9" and outputting processing result information ("contains anomaly (mode b)" or "contains anomaly (mode c)").

[0085] Furthermore, the abnormal patterns (patterns a, b, c, etc.) are categorized, for example, based on each abnormal cause in the excited state. Abnormal causes include information such as which molecule is in the excited state, or which part of the chamber is abnormal. That is, based on the learned abnormality detection model learned using 1000 training data points, the abnormal causes can be inferred by inferring the abnormal patterns.

[0086] <Specific examples of the processing carried out by the Learning Department> Next, a specific example of the learning process performed by the learning unit 921 of the data processing device 920 will be described. Figure 11 is a diagram showing a specific example of the learning process performed by the learning unit. As shown in Figure 11, the learning unit 921 has an anomaly detection model 1101 and a comparison / modification unit 1102.

[0087] The learning unit 921 reads the process parameters (e.g., process parameters specific to process parameter 1) and image data (e.g., image data specific to image data 1) from the learning data 1000's "process parameters" and "image data".

[0088] Furthermore, the learning unit 921 inputs the process parameters and image data read from the learning data 1000 into the anomaly detection model 1101, executes the anomaly detection model 1101, and outputs the probability distribution of the processing result information.

[0089] The probability distribution of the processing result information output from the anomaly detection model 1101 is input into the comparison / modification unit 1102 and compared with the probability distribution of the processing result information read from the "Good / Contains Anomalies" data in the learning data 1000. For example, when "Good" is read from the "Good / Contains Anomalies" data in the learning data 1000, the probability distribution of the processing result information output from the anomaly detection model 1101 is compared with the probability distribution set to "Good" = 100% and other processing result information = 0%.

[0090] The comparison / modification unit 1102 updates the model parameters of the anomaly detection model 1101 based on the comparison results. In this way, the learning unit 921 can update the model parameters of the anomaly detection model 1101 so that the output when process parameters and image data are used as inputs is close to the processing result information stored in the "Good / Contains Anomalies" of the learning data 1000.

[0091] <Specific examples of inference processing performed by the inference department> Next, a specific example of the inference processing performed by the inference unit 922 of the data processing apparatus 920 will be described. Figure 12 is a diagram showing a specific example of the inference processing performed by the inference unit. As shown in Figure 12, the inference unit 922 has a learned anomaly detection model 1201 (a learned anomaly detection model generated by learning processing on the anomaly detection model 1101) and an output unit 1202.

[0092] The inference unit 922 acquires the process parameters used when processing the pre-processing wafer 110 in chamber A, and the image data generated from the OES data measured during the processing of the pre-processing wafer 110 in chamber A, and inputs them into the learned anomaly detection model 1201.

[0093] When process parameters and image data are input through the inference unit 922, the learned anomaly detection model 1201 outputs the probability distribution of the processing result information.

[0094] When the anomaly detection model 1201 outputs the probability distribution of the processing result information, the output unit 1202 outputs the processing result information corresponding to the largest probability distribution among the probability distributions above a specified threshold. For example, when the probability distribution of "good" is above the specified threshold and is the largest, the output unit 1202 outputs information indicating that the processed wafer 130 is a good product that has been processed normally as an inference result.

[0095] Furthermore, for example, when the probability distribution of "containing an anomaly (mode a)" is above a predetermined threshold and is at its maximum, the output unit 1202 outputs information indicating that the processed wafer 130 contains an anomaly belonging to mode a and information indicating the cause of the anomaly as a deduction result. Furthermore, the output unit 1202 may also be configured to output the most suitable process parameters in addition to outputting the deduction result.

[0096] <Anomaly Detection and Handling Process> Secondly, the flow of anomaly detection processing performed by data processing systems 900 and 900' will be explained. Figure 13 is a flowchart showing the anomaly detection processing flow. Furthermore, the steps S801 to S804 in each process shown in Figure 13 are the same as the steps S801 to S804 in Figure 8, therefore, the explanation is omitted here.

[0097] In step S1301, the data processing device 920 determines whether the current period is a learning period or an inference period. In step S1301, if it is determined to be a learning period (if it is in step S1301), proceed to step S1302.

[0098] In step S1302, the data processing device 920 obtains the processing result information and the corresponding process parameters.

[0099] In step S1303, the data processing device 920 establishes a correspondence between the obtained processing result information, process parameters, and image data, generates learning data, and stores it in the learning data storage unit 923.

[0100] In step S1304, the learning unit 921 of the data processing device 920 uses learning data to perform learning processing on the anomaly detection model, and ends the anomaly detection processing after generating the learned anomaly detection model.

[0101] On the other hand, in step S1301, if it is determined that the inference period is in progress (if it is not in step S1301), proceed to step S1305.

[0102] In step S1305, the inference unit 922 of the data processing device 920 inputs the image data and process parameters into the learned anomaly detection model 1201 and outputs the probability distribution of the processing result information.

[0103] In step S1305, the inference unit 922 of the data processing device 920 sends the inference result output from the output unit 1202 to the semiconductor manufacturing process based on the probability distribution of the processing result information output from the learned anomaly detection model 1201.

[0104] Summary As can be clearly seen from the above description, the data processing device 920 of the second embodiment... It features an anomaly detection model that learns the correspondence between image data and process parameters generated by visualizing OES data, and the processing results of the wafer after processing. • It possesses a learned anomaly detection model, which is generated by learning from the anomaly detection model using training data; • The image data generated by visualizing OES data and process parameters are input into the learned anomaly detection model to infer the processing result information of the wafer after processing.

[0105] Thus, by using image data to infer the composition of the processed wafer processing result information, according to the second embodiment, it is possible to automatically determine whether any abnormalities have occurred.

[0106] [Third Implementation] In the second embodiment described above, a case was described in which a data processing device is set up in each processing space, and a learned anomaly detection model is generated in each data processing device using the learning data generated in each data processing device. In contrast, in the third embodiment, the server device collects the learning data generated in each data processing device and performs learning processing.

[0107] Therefore, according to the third embodiment, it is possible to provide a learned anomaly detection model generated by learning processing using more learning data.

[0108] Hereinafter, the third embodiment will be described focusing on the differences from the first and second embodiments described above.

[0109] <System Composition of Data Processing Systems> First, the system configuration of the data processing system according to the third embodiment will be described. Figure 14 is the fourth figure showing an example of the system configuration of the data processing system.

[0110] As shown in Figure 14, the data processing system 1400 includes: a plurality of semiconductor manufacturing processes; an emission spectrum analysis device 140, a data processing device 1401, a cooperation unit 1411 corresponding to each semiconductor manufacturing process; and a server device 1420.

[0111] The emission spectroscopy analyzer 140 has already been described, so the description is omitted here.

[0112] The data processing apparatus 1401 includes a preprocessing unit 151, a compression unit 152, an image processing unit 153, and an inference unit 922. Furthermore, the preprocessing unit 151, compression unit 152, image processing unit 153, and inference unit 922 of the data processing apparatus 1401 have been described using Figures 9A and 9B, therefore, their description is omitted here.

[0113] Furthermore, the data processing apparatus 1401 includes: an OES data storage unit 155, which stores OES data measured by the emission spectroscopy analyzer 140; and an image data storage unit 156, which stores image data generated by the imaging unit 153. Moreover, the data processing apparatus 1401 includes a learning data storage unit 923, which establishes a correspondence between image data, processing result information, and process parameters and stores them as learning data. Furthermore, the OES data storage unit 155, the image data storage unit 156, and the learning data storage unit 923 have already been described, therefore, their description is omitted here.

[0114] Collaboration unit 1411 reads the learning data (e.g., learning data 1) stored in learning data storage unit 923 and sends it to server device 1420. Furthermore, the image data contained in the learning data exchanged between collaboration unit 1411 and server device 1420 is compressed, thus reducing the amount of communication compared to the case of exchanging OES data itself.

[0115] Furthermore, the collaboration unit 1411 sends learning data to the server device 1420 and applies the learned anomaly detection model obtained from the server device 1420 to the inference unit 922. In this way, the inference unit 922 can perform inferences using the learned anomaly detection model, which is common to various semiconductor manufacturing processes and is generated by the server device 1420.

[0116] When the server device 1420 sends learning data from each cooperating unit 1411 corresponding to each semiconductor manufacturing process, it stores the learning data in the learning data storage unit 1422. Furthermore, the server device 1420 has a learning unit 921.

[0117] The learning unit 921 uses the learning data stored in the learning data storage unit 1422 to perform learning processing on the anomaly detection model possessed by the learning unit 921. In this way, the learning unit 921 can generate a common learned anomaly detection model applicable to various semiconductor manufacturing processes.

[0118] Furthermore, in the server device 1420, the learned anomaly detection model generated by the learning unit 921 is sent to each cooperating unit 1411 corresponding to each semiconductor manufacturing process.

[0119] Summary As can be clearly seen from the above description, in the third embodiment, the learning unit 921 and the inference unit 922 are separate entities, and a common learned anomaly detection model is generated by collecting learning data obtained from each semiconductor manufacturing process.

[0120] Therefore, according to the third embodiment, it is possible to generate an anomaly detection model that is learned from individual differences and does not depend on the semiconductor manufacturing process.

[0121] [Fourth Implementation] In the first to third embodiments described above, OES data is exemplified as multi-wavelength time-series data measured during the processing of the pre-processing wafer 110 in the semiconductor manufacturing process. However, the multi-wavelength time-series data measured during the processing of the pre-processing wafer 110 in the semiconductor manufacturing process is not limited to OES data. For example, it could also be quality analysis data measured by a gas quality analysis device in the analysis chamber. Alternatively, it could be reflected light data obtained by measuring the reflected light when light from an external light source in the chamber is projected onto the wafer surface.

[0122] Furthermore, in the first embodiment described above, a region with 3 points in the horizontal direction and 3 points in the vertical direction is exemplified as a region of a predetermined size, but the region of a predetermined size is not limited to this. Also, the shape of the region of a predetermined size is not limited to a square, and the number of points in the horizontal direction and the number of points in the vertical direction may be different.

[0123] Furthermore, in the first embodiment described above, the image processing unit 153 was described as image processing the compressed OES data compressed by the compression unit 152 and storing it in the image data storage unit 156. However, in the data processing apparatus 150, it is sometimes necessary to manage the OES data before compression (pre-processed OES data) or the OES data before pre-processing (original OES data) for a certain period of time for application purposes. Therefore, in addition to the functions described in the first embodiment, the image processing unit 153 of the data processing apparatus 150 may also have the function of image processing the original OES data or the function of image processing the pre-processed OES data. Furthermore, the image processing unit 153 may also have the function of performing image compression processing such as JPEG (Joint Photographic Experts Group) compression on the image data and storing it in the image data storage unit 156.

[0124] Furthermore, in the second and third embodiments described above, the input of process parameters and image data into the anomaly detection model was explained. However, it is also possible to input only image data into the anomaly detection model. Alternatively, in addition to process parameters and image data, other data besides process parameters and image data may also be input into the anomaly detection model.

[0125] Furthermore, in the third embodiment described above, the case where the server device 1420 collects learning data generated by the data processing device 1401 corresponding to a plurality of semiconductor manufacturing processes was explained. However, learning data can also be exchanged between data processing devices via the server device 1420 (or directly). In this case, the collaboration unit 1411 adds learning data (e.g., learning data 2) received from the collaboration unit 1411 of other semiconductor manufacturing processes to the learning data storage unit 923. In this way, the learning unit 921 can use learning data (e.g., learning data 1) obtained from the corresponding semiconductor manufacturing process, as well as learning data (e.g., learning data 2) obtained from other semiconductor manufacturing processes, to perform learning processing for anomaly detection models. That is, the learning unit 921 can also perform learning processing for events that have not occurred in the corresponding semiconductor manufacturing process. Furthermore, the learned anomaly detection model generated in this way can also be sent and received between data processing devices via server device 1420 (or directly), thereby being applied to other semiconductor manufacturing processes. That is, the learned anomaly detection model generated in any semiconductor manufacturing process can also be horizontally expanded to other semiconductor manufacturing processes via server device 1420 (or directly).

[0126] Furthermore, in the third embodiment described above, the server device 1420 is described as having a learning unit 921 that provides the generated, learned anomaly detection model to the data processing device 1401. However, an inference unit that applies the generated, learned anomaly detection model may also be configured in the server device 1420. In this case, when image data and process parameters are received from the cooperation unit 1411 corresponding to any semiconductor manufacturing process in the server device 1420, the inference result based on the received image data and process parameters is sent to the corresponding cooperation unit 1411. That is, the generated, learned anomaly detection model can also be executablely configured in the server device 1420, thereby sharing the learned anomaly detection model among each semiconductor manufacturing process.

[0127] Furthermore, the present invention is not limited to the configurations described in the above embodiments, or the combinations thereof with other elements. Modifications can be made in these respects without departing from the spirit of the invention, and can be appropriately determined according to its application.

[0128] 100: Data Processing System 110: Wafer before processing 120: Processing space 121: Chamber A 122: Chamber B 123: Chamber C 130: Processed wafer 140: Emission Spectroscopy Analyzer 150: Data processing device 151: Pre-processing Department 152: Compression section 153: Graphics Department 154: Display Control Unit 155: OES Data Storage Department 156: Image Data Storage Department 200: Semiconductor Manufacturing Process 301: CPU 302:ROM 303: RAM 304: GPU 305: Auxiliary Memory Device 306: Display device 307: Operating device 308: I / F device 309: Drive unit 310: Busbar 320: Recording medium 410: OES Data 420: OES Data 500: Preprocessed OES data 500': Image data 501: Standardized Documentation 510: Standardization Processing Department 600: Compressed OES data 600': Image data 601: Representative value 610: Average Calculation Section 620: Representative Value Extraction Department 701: Specified Area 702: Enlarged Image Data 900: Data Processing System 900': Data Processing System 910: Inspection device 920: Data Processing Device 921: Study Department 922: Inference Department 923: Study Materials Storage Department 1000: Study Materials 1101: Anomaly Detection Model 1102: Comparison / Change Department 1201: Anomaly Detection Models Learned 1202: Output Section 1400: Data Processing System 1401: Data Processing Device 1411: Collaboration Department 1420: Server device 1422: Study Materials Storage Department

Claims

1. A data processing apparatus comprising: a preprocessing unit that normalizes multi-wavelength time series data using predetermined reference data to generate normalized data; an extraction unit that divides the normalized data into a plurality of regions for each predetermined time range and predetermined wavelength range, and extracts outliers in each region as representative values; and a generation unit that converts the representative values ​​of each region into color data to generate image data, wherein the extraction unit calculates the average value of the normalized data contained in each region, and extracts the outlier with the largest difference from each average value from the normalized data contained in each region as the representative value of each region.

2. The data processing apparatus of claim 1, wherein the preprocessing unit uses the aforementioned multi-wavelength time series data when processing a reference substrate in a semiconductor manufacturing process, or the time series data of the reference wavelength in the aforementioned multi-wavelength time series data as the aforementioned specified reference data, to normalize the aforementioned multi-wavelength time series data.

3. The data processing apparatus of claim 1, wherein the aforementioned multi-wavelength time series data is data measured by an emission spectroscopy analyzer or a quality analyzer during substrate processing in a semiconductor manufacturing process.

4. The data processing apparatus of claim 3 further includes an inference unit that re-inputs the image data generated by the generation unit and based on the multi-wavelength time-series data measured during the processing of the substrate into a model that has learned the correspondence between the image data generated by the generation unit and the processing result of the substrate after processing. This inference unit infers the processing result of the substrate after processing.

5. The data processing apparatus of claim 4, wherein the inference unit infers the situation that the processed substrate contains an abnormality and the cause of the abnormality of the processed substrate as the processing result.

6. The data processing apparatus of claim 4 further includes a learning unit that performs learning processing on the model described above, such that when image data is input into the model, the output is close to the processing result of the substrate after processing, wherein the image data is generated by the generation unit and based on the multi-wavelength time series data measured during the processing of the substrate.

7. A data processing system comprising: a plurality of data processing devices as claimed in claim 4; and a server device connected to the plurality of data processing devices, receiving from the plurality of data processing devices at least one of the following: image data generated based on the multi-wavelength time-series data measured during the processing of the substrate, or a model that has learned the correspondence between the image data and the processing result of the processed substrate.

8. A data processing system comprising: a data processing apparatus as claimed in claim 4; and a server apparatus having a learning unit that performs learning processing on the model such that the output when image data is input to the model is close to the processing result of the substrate after processing, wherein the image data is generated by the generation unit and based on the multi-wavelength time series data measured during the processing of the substrate.

9. A data processing method comprising: a preprocessing step, which normalizes multi-wavelength time series data using prescribed reference data to generate normalized data; an extraction step, which divides the normalized data into a plurality of regions for each prescribed time range and prescribed wavelength range, and extracts outliers in each region as representative values; and a generation step, which converts the representative values ​​of each region into color data to generate image data, wherein the extraction step calculates the average value of the normalized data contained in each region, and extracts the outlier with the largest difference from each average value from the normalized data contained in each region as the representative value of each region.

10. The data processing method of claim 9, wherein the above-mentioned preprocessing step is used in the semiconductor manufacturing process to process the above-mentioned multi-wavelength time series data as a reference substrate, or the time series data of the reference wavelength in the above-mentioned multi-wavelength time series data as the reference data specified above, to normalize the above-mentioned multi-wavelength time series data.

11. The data processing method of claim 9, wherein the aforementioned multi-wavelength time series data is data measured by an emission spectroscopy analyzer or a quality analyzer during the substrate processing in the semiconductor manufacturing process.

12. A data processing program for causing a computer to perform the following steps: a preprocessing step, which normalizes multi-wavelength time series data using specified reference data to generate normalized data; an extraction step, which divides the normalized data into a plurality of regions for each specified time range and specified wavelength range, and extracts outliers in each region as representative values; and a generation step, which converts the representative values ​​of each region into color data to generate image data, wherein the extraction step calculates the average value of the normalized data contained in each region, and extracts the outlier with the largest difference from each average value from the normalized data contained in each region as the representative value of each region.

13. The data processing program of claim 12, wherein the above-mentioned preprocessing step is used in the semiconductor manufacturing process to process the above-mentioned multi-wavelength time series data as a reference substrate, or the time series data of the reference wavelength in the above-mentioned multi-wavelength time series data as the reference data specified above, to normalize the above-mentioned multi-wavelength time series data.

14. The data processing program of claim 12, wherein the aforementioned multi-wavelength time series data is data measured by an emission spectroscopy analyzer or a quality analyzer during substrate processing in a semiconductor manufacturing process.

Images