Edge defect detection via image analysis

Abstract

The present disclosure includes edge defect detection via image analysis. A method includes identifying an image of an edge of a susceptor pocket formed by a susceptor of a substrate processing system. The method further includes predicting whether an attribute value of an edge of the susceptor satisfies a threshold based on the image. The method further includes causing a corrective action associated with the susceptor to perform in response to the attribute value of the edge satisfying a threshold.

Description

Technical Field

[0001] The present disclosure relates to image analysis, and more particularly to edge defect detection via image analysis. Background Art

[0002] Manufacturing equipment includes different parts used to produce products. For example, substrate processing equipment includes parts used to produce substrates (e.g., process substrates). The quality and cleanliness of the parts will affect the performance data of the product. Summary of the invention

[0003] The following is a simplified summary of the disclosure to provide a basic understanding of some aspects of the disclosure. This summary is not an extensive overview of the disclosure. This summary is neither intended to identify the key or essential elements of the disclosure nor to define any scope of a particular embodiment of the disclosure or any scope of the claims. The sole purpose of this summary is to present some concepts of the disclosure in a simplified form as a prelude to the [embodiment] presented later.

[0004] In an aspect of the present disclosure, a method includes: identifying an image of an edge of a susceptor cavity formed by a susceptor of a substrate processing system. The method further includes: predicting whether a property value of the edge of the susceptor satisfies a threshold based on the image. The method further includes: causing a corrective action associated with the susceptor to be performed in response to the property value of the edge satisfying the threshold.

[0005] In another aspect of the present disclosure, there is a non-transitory computer-readable storage medium storing instructions that, when executed, cause a processing device to perform operations. The operations include: identifying an image of an edge of a susceptor cavity formed by a susceptor of a substrate processing system. The operations further include: predicting, based on the image, whether a property value of the edge of the susceptor satisfies a threshold. The operations further include: causing a corrective action associated with the susceptor to be performed in response to the property value of the edge satisfying the threshold.

[0006] In another aspect of the present disclosure, a system includes a memory and a processing device coupled to the memory. The processing device is used to identify an image of an edge of a susceptor cavity formed by a susceptor of a substrate processing system. The processing device further predicts whether a property value of the edge of the susceptor satisfies a threshold value based on the image. The processing device further causes a corrective action associated with the susceptor to be performed in response to the property value of the edge satisfying the threshold value. BRIEF DESCRIPTION OF THE DRAWINGS

[0007] The present disclosure is illustrated by way of example and not limitation in the accompanying figures.

[0008] Figure 1 is a block diagram illustrating an exemplary system architecture in accordance with certain embodiments.

[0009] Figure 2 A dataset generator for creating a dataset for a machine learning model is shown in accordance with certain embodiments.

[0010] Figure 3 is a block diagram illustrating determining prediction data according to certain embodiments.

[0011] FIG. 4A to FIG. 4C Components of a substrate processing apparatus are shown in accordance with certain embodiments.

[0012] FIG. 5A to FIG. 5E Images showing parts of substrate processing equipment according to certain embodiments.

[0013] FIG. 6A to FIG. 6D Images showing parts of substrate processing equipment according to certain embodiments.

[0014] 7A to 7E is a flow chart of a method associated with edge defect detection according to certain embodiments.

[0015] Figure 8 is a block diagram illustrating a computer system according to some embodiments. DETAILED DESCRIPTION

[0016] Described herein are techniques for edge defect detection via image analysis (eg, automated quality control inspection of susceptor edges, and detection of wafer pocket edge defects in susceptors using image analysis).

[0017] Manufacturing equipment includes different parts used to produce products. For example, substrate processing equipment includes parts used to produce substrates (e.g., process substrates). Some parts, such as susceptors, form edges (e.g., recess edges formed on the upper surface of the susceptor). The quality and cleanliness of the parts can affect the performance data of the product. For example, a susceptor with damaged or accumulated recess edges may cause the performance data of the produced substrates to not meet the threshold (e.g., produce bad wafers).

[0018] In some conventional systems, parts are manually inspected to try to determine if the parts meet the quality and cleanliness to produce a substrate with performance data that will meet the threshold. Depending on the user performing the inspection, manual inspection takes a lot of time and manual inspection may be inaccurate. Inaccuracies in manual inspection may result in the use of defective parts or premature replacement of parts.

[0019] In some conventional systems, intervals are scheduled for the performance of maintenance operations (e.g., cleaning operations, repair operations, and replacement operations) for parts. The intervals are set in an attempt to keep the parts in a condition that produces substrates with performance data that will meet thresholds (e.g., clean, repair, and replace the part before the part produces bad wafers). Over time, parts can become damaged, parts can wear out, parts can accumulate foreign matter, etc., and these conditions may not occur within the set intervals (e.g., they can occur before or after the set intervals). Repairing or replacing parts too early can cause waste, time consumption, reduced yields, and production interruptions, etc. Overdue repairs or replacement of parts can result in produced substrate performance data that does not meet thresholds and equipment damage, etc.

[0020] The devices, systems, and methods disclosed herein provide edge defect detection via image analysis.

[0021] The processing device identifies an image of an edge of a substrate processing equipment part (e.g., a pedestal). An image of a substantially horizontal upper surface of the pedestal may be captured. The pedestal may form one or more recesses (e.g., pedestal recesses). The pedestal may include an upper surface forming one or more recesses (e.g., pedestal recesses). The recess may be further formed by one or more sidewalls (e.g., substantially vertical sidewalls) and a lower surface. The intersection of the upper surface and the sidewall of the recess is an edge (e.g., an upper edge). The pedestal recess may be configured to receive a substrate for performing a substrate processing operation. Over time, the edge may develop defects (e.g., due to placing and removing substrates in the pedestal recess).

[0022] The processing device predicts whether a property value of an edge of the base satisfies a threshold based on the image.

[0023] In some embodiments, the processing device determines at least one of a height of a portion of the image associated with the edge or a number of pixels associated with an edge crack based on the image. An edge with buildup or damage (e.g., cracking) may have a height that satisfies a threshold in the image (e.g., greater than a height in an image of a pedestal that produces good wafers). A damaged (e.g., cracking) edge may have a number of pixels, such as white pixels, associated with an edge crack that satisfies a threshold (e.g., greater than a number of white pixels in an image of a pedestal that produces good wafers).

[0024] In some embodiments, the image is provided to a trained machine learning model, and an output associated with the prediction data is received from the machine learning model.

[0025] In response to the property value of the edge satisfying the threshold (e.g., the edge height satisfying the threshold height, the number of pixels satisfying the threshold amount, and the prediction data indicating that the property value satisfies the threshold, etc.), the processing device causes the execution of a corrective action associated with the substrate processing equipment part. The corrective action may include providing an alarm, causing a cleaning process, causing a repair process, causing a substrate processing equipment part to be replaced, causing further inspection, etc.

[0026] Various aspects of the present disclosure provide technical advantages. The present disclosure avoids the time, inaccuracy and subjectivity of traditional manual inspection. The present disclosure causes corrective actions related to substrate processing equipment parts to produce substrates that meet thresholds, avoid equipment damage, increase yields, and avoid production interruptions, etc.

[0027] Some embodiments of the present disclosure describe performing edge defect detection on a pedestal. In some embodiments, the present invention can be used to perform edge defect detection on other parts (eg, substrate processing equipment parts), such as electrostatic chucks and edge rings.

[0028] Some embodiments of the present disclosure describe performing edge defect detection on a base recess formed by a base (e.g., a groove formed by a part). In some embodiments, the present disclosure can be used to perform edge defect detection on other portions of a part (e.g., a substrate processing equipment part); other portions such as peripheral edges and countertops, etc.

[0029] Some embodiments of the present disclosure describe performing edge defect detection on edges that are substantially vertical and form a circular shape (e.g., a circular recess in a base). In some embodiments, the present disclosure can be used to perform edge defect detection on other types of edges; other types of edges such as edges that are not substantially vertical, and edges that do not form a circular shape (e.g., edges that form a rectangular shape, a triangular shape, and an oval shape, etc.).

[0030] As used herein, the term "production" may refer to producing a final version of a product (e.g., a fully processed substrate), or an intermediate version of a product (e.g., a partially processed substrate). As used herein, producing a substrate may refer to processing a substrate by performing one or more substrate processing operations.

[0031] Figure 1 1 is a block diagram showing an exemplary system 100 (exemplary system architecture) according to some embodiments. The system 100 includes a client device 120, a manufacturing device 124, a sensor 126, a metrology device 128, a prediction server 112, and a data store 140. In some embodiments, the prediction server 112 is part of the prediction system 110. In some embodiments, the prediction system 110 further includes server machines 170 and 180.

[0032] In some embodiments, one or more of the client device 120, the manufacturing equipment 124, the sensor 126, the metrology device 128, the prediction server 112, the data store 140, the server machine 170, and / or the server machine 180 are coupled to each other via the network 130 for generating the prediction data 160 to perform edge defect detection. In some embodiments, the network 130 is a public network that provides the client device 120 with access to the prediction server 112, the data store 140, and other publicly available computing devices. In some embodiments, the network 130 is a private network that provides the client device 120 with access to the manufacturing equipment 124, the sensor 126, the metrology device 128, the data store 140, and other privately available computing devices. In some embodiments, the network 130 includes one or more wide area networks (WANs), local area networks (LANs), wired networks (e.g., Ethernet networks), wireless networks (e.g., 802.11 networks or Wi-Fi networks), cellular networks (e.g., long term evolution (LTE) networks), routers, hubs, switches, server computers, cloud computing networks, and / or combinations thereof.

[0033] In some embodiments, client device 120 includes a computing device such as a personal computer (PC), a laptop computer, a mobile phone, a smart phone, a tablet computer, a small notebook computer, etc. In some embodiments, client device 120 includes a corrective action component 122. In some embodiments, corrective action component 122 may also be included in prediction system 110 (e.g., a machine learning processing system). In some embodiments, corrective action component 122 is instead included in prediction system 110 (e.g., instead of being included in client device 120). Client device 120 includes an operating system that allows a user to merge, generate, view or edit data, provide instructions to prediction system 110 (e.g., a machine learning processing system), etc. One or more.

[0034] In some embodiments, the corrective action component 122 receives user input (e.g., via a graphical user interface (GUI) displayed by the client device 120), receives sensor data 142 from sensors, receives performance data 152 from the metering device 128, etc. In some embodiments, the corrective action component 122 transmits the data (e.g., user input, sensor data 142, performance data 152, etc.) to the prediction system 110, receives prediction data 160 from the prediction system 110, determines a corrective action based on the prediction data 160, and causes the corrective action to be implemented. In some embodiments, the corrective action component 122 stores the data (e.g., user input, sensor data 142, performance data 152, etc.) in the data store 140, and the prediction server 112 retrieves the data from the data store 140. In some embodiments, the prediction server 112 stores the output of the trained machine learning model 190 (e.g., prediction data 160) in the data store 140, and the client device 120 retrieves the output from the data store 140. In some embodiments, corrective action component 122 receives an indication of a corrective action from prediction system 110 (eg, based on prediction data 160 ) and causes performance of the corrective action.

[0035] In some embodiments, the prediction data 160 is associated with a corrective action. In some embodiments, the corrective action is associated with one or more of cleaning a substrate processing equipment part, repairing a substrate processing equipment part, replacing a substrate processing equipment part, computational process control (CPC), statistical process control (SPC) (e.g., SPC is compared to a 3-sigma chart, etc.), advanced process control (APC), model-based process control, preventive operational maintenance, design optimization, manufacturing parameter update, feedback control, machine learning modification, or the like. In some embodiments, the corrective action includes providing an alarm (e.g., if the prediction data 160 indicates a predicted anomaly, such as an anomaly of a substrate processing equipment part or product, an alarm of the substrate processing equipment part or manufacturing equipment 124 is not used). In some embodiments, the corrective action includes providing feedback control (e.g., cleaning, repairing, and / or replacing a substrate processing equipment part in response to the prediction data 160 indicating a predicted anomaly). In some embodiments, the corrective action includes providing machine learning (e.g., causing a modification of a substrate processing equipment part based on the prediction data 160).

[0036] In some embodiments, prediction server 112, server machine 170, and server machine 180 each include one or more computing devices, such as a rack server, a router computer, a server computer, a personal computer, a mainframe computer, a laptop computer, a tablet computer, a desktop computer, a graphics processing unit (GPU), an accelerator application-specific integrated circuit (ASIC) (e.g., a tensor processing unit (TPU)), and the like.

[0037] The prediction server 112 includes a prediction component 114. In some embodiments, the prediction component 114 receives sensor data 142 (e.g., received from the client device 120, retrieved from the data store 140) and generates prediction data 160 associated with edge defect detection. In some embodiments, the prediction component 114 uses one or more trained machine learning models 190 to determine the prediction data 160 for edge defect detection. In some embodiments, the trained machine learning models 190 are trained using historical sensor data 144 and historical performance data 154.

[0038] In some embodiments, the prediction system 110 (e.g., prediction server 112, prediction component 114) uses supervised machine learning (e.g., supervised data sets, historical sensor data 144 labeled with historical performance data 154, etc.) to generate prediction data 160. In some embodiments, the prediction system 110 uses semi-supervised learning (e.g., semi-supervised data sets, performance data 152 is a predicted percentage, etc.) to generate prediction data 160. In some embodiments, the prediction system 110 uses unsupervised machine learning (e.g., unsupervised data sets, clusters, clusters based on historical sensor data 144, etc.) to generate prediction data 160.

[0039] In some embodiments, the manufacturing equipment 124 (e.g., a cluster tool) is part of a substrate processing system (e.g., an integrated processing system). The manufacturing equipment 124 includes one or more of a controller, a housing system (e.g., a substrate carrier, a front opening wafer transfer box (FOUP), an automatic teaching FOUP, a process kit housing system, a substrate housing system and a box, etc.), a side storage box (SSP), an aligner device (e.g., an aligner chamber), a factory interface (e.g., an equipment front end module (EFEM)), a load lock, a transfer chamber, one or more processing chambers, a robot (e.g., disposed in the transfer chamber and disposed in the front interface, etc.), and / or the like. The housing system, the SSP, and the load lock are mounted to the factory interface, and the robot disposed in the factory interface is used to transfer contents (e.g., substrates, process kit rings, carriers, and verification chips, etc.) between the housing system, the SSP, the load lock, and the factory interface. The alignment device is disposed in the factory interface to align the contents. The load lock and the process chamber are mounted to the transfer chamber, and a robot arm disposed in the transfer chamber is used to transfer contents (e.g., substrates, process kit rings, carriers, and verification wafers, etc.) between the load lock, the process chamber, and the transfer chamber. In some embodiments, the manufacturing equipment 124 includes components of a substrate processing system. In some embodiments, the sensor data 142 includes parameters of processes (e.g., etching, heating, cooling, transferring, processing, and flowing, etc.) performed by the components of the manufacturing equipment 124. In some embodiments, the substrate processing equipment parts are components of the process chamber (e.g., showerheads, pedestals, electrostatic chucks, and edge rings, etc.).

[0040] In some embodiments, the sensors 126 provide sensor data 142 (e.g., sensor values, such as historical sensor values ​​and current sensor values) associated with the manufacturing equipment 124. In some embodiments, the sensors 126 include one or more of an imaging sensor (e.g., a camera, an image capture device, etc.), a pressure sensor, a temperature sensor, a flow sensor, a spectral sensor, and / or the like. In some embodiments, the sensor data 142 is used for equipment health and / or product health (e.g., product quality). In some embodiments, the sensor data 142 is received over a period of time.

[0041] In some embodiments, the sensor 126 provides sensor data 142, such as image data, leak rate, temperature, pressure, flow rate (e.g., gas flow), pumping efficiency, spacing (SP), high frequency radio frequency (HFRF), current, power, voltage, and / or values ​​of one or more of the like.

[0042] In some embodiments, sensor data 142 (e.g., historical sensor data 144, current sensor data 146, etc.) is processed by client device 120 and / or by prediction server 112. In some embodiments, processing of sensor data 142 includes generating features. In some embodiments, a feature is a portion of sensor data (e.g., a cropped image), processed image data (e.g., a processed image), a pattern in sensor data 142 (e.g., slope, width, height, peak, etc.), or a combination of values ​​from sensor data 142 (e.g., power derived from voltage and current, etc.). In some embodiments, sensor data 142 includes features used by prediction component 114 to obtain prediction data 160.

[0043] In some embodiments, metrology equipment 128 (e.g., imaging equipment, spectral equipment, elliptical brightness equipment, etc.) is used to determine metrology data (e.g., inspection data, image data, spectral data, elliptical brightness data, material composition, optical or structural data, etc.) corresponding to a substrate produced by manufacturing equipment 124 (e.g., substrate processing equipment). In some examples, metrology equipment 128 is used to inspect a portion (e.g., a layer) of a substrate after manufacturing equipment 124 processes the substrate. In some embodiments, metrology equipment 128 performs scanning acoustic microscopy (SAM), ultrasonic inspection, X-ray inspection, and / or computed tomography (CT) inspection. In some examples, metrology equipment 128 is used to determine the quality of the processed substrate (e.g., the thickness of the layer, the uniformity of the layer, the interlayer spacing of the layer, and / or the like) after manufacturing equipment 124 deposits one or more layers on the substrate. In some embodiments, metrology equipment 128 includes an image capture device (e.g., a SAM device, an ultrasonic device, an X-ray device, a CT device, and / or the like). In some embodiments, performance data 152 includes metrology data from metrology equipment 128.

[0044] In some embodiments, data store 140 is a memory (e.g., random access memory), a disk drive (e.g., a hard drive, a flash drive), a database system, or another type of component or device capable of storing data. In some embodiments, data store 140 includes multiple storage components (e.g., multiple disk drives or multiple databases) across multiple computing devices (e.g., multiple server computers). In some embodiments, data store 140 stores one or more of sensor data 142, performance data 152, and / or prediction data 160.

[0045] Sensor data 142 includes historical sensor data 144 and current sensor data 146. In some embodiments, sensor data 142 may include one or more of image data, pressure data, pressure ranges, temperature data, temperature ranges, flow rate data, power data, comparison parameters for comparing inspection data to threshold data, threshold data, cooling rate data, cooling rate ranges, and / or the like. In some embodiments, at least a portion of sensor data 142 is from sensor 126.

[0046] The performance data 152 includes historical performance data 154 and current performance data 156. The performance data 152 may include, among other things, a property value of a substrate processing equipment part (e.g., a pedestal) and an indication of whether the property value of the substrate processing equipment part (e.g., a pedestal) satisfies a threshold. In some examples, the performance data 152 indicates whether the substrate is properly designed, properly manufactured, and / or properly functioning. In some embodiments, at least a portion of the performance data 152 is associated with the quality of the substrates produced by the manufacturing equipment 124. In some embodiments, at least a portion of the performance data 152 is based on metrology data from the metrology equipment 128 (e.g., the historical performance data 154 includes metrology data indicating properly processed substrates, characteristic data of the substrates, and yield, etc.). In some embodiments, at least a portion of the performance data 152 is based on an inspection of the substrate (e.g., current performance data 156 based on actual inspection). In some embodiments, the performance data 152 includes an absolute value indication (e.g., the inspection data of the bonding interface indicates that the threshold data is missed by the calculated value, the deformation value is missed by the calculated value), or a relative value indication (e.g., the inspection data of the bonding interface indicates that the threshold data is missed by 5%, the deformation value is missed by 5%. In some embodiments, the performance data 152 indicates that a threshold error amount is met (e.g., at least 5% error in production, at least 5% error in flow, at least 5% error in deformation, and specification limit).

[0047] In some embodiments, the client device 120 provides performance data 152 (e.g., product data). In some examples, the client device 120 provides performance data 152 indicating product anomalies (e.g., defective products) (e.g., based on user input). In some embodiments, the performance data 152 includes the amount of normal or abnormal products that have been produced (e.g., 98% normal products). In some embodiments, the performance data 152 indicates the number of products being produced that are predicted to be normal or abnormal. In some embodiments, the performance data 152 includes one or more of the yield of a previous batch of products, the average yield, the predicted yield, the predicted number of defective or non-defective products, or the like. In some examples, in response to the yield of the first batch of products being 98% (e.g., 98% of the products are normal and 2% are abnormal), the client device 120 provides performance data 152 indicating that the upcoming batch of products will have a yield of 98%.

[0048] In some embodiments, historical data includes one or more of historical sensor data 144 and / or historical performance data 154 (e.g., used to train at least a portion of machine learning model 190). Current data includes one or more of current sensor data 146 and / or current performance data 156 (e.g., to be input to at least a portion of trained machine learning model 190 after training model 190 using historical data). In some embodiments, current data is used to retrain trained machine learning model 190.

[0049] In some embodiments, the prediction data 160 is used to cause corrective actions to be performed on substrate processing equipment parts.

[0050] Metrology of products to determine substrate handling equipment parts that do not meet a threshold quality, incorrectly produced components (e.g., bonded sheet metal structures) is expensive in terms of time used to send metrology data and processor overhead to process the metrology data, metrology equipment 128 used, energy consumed, and bandwidth used. By providing sensor data 142 to the model 190 and receiving prediction data 160 from the model 190, the system 100 has the advantage of avoiding the use of metrology equipment 128 and the expensive process of discarding substrates.

[0051] Performing manufacturing processes using substrate processing equipment parts that result in defective products is expensive in terms of time, energy, product, substrate processing equipment parts, manufacturing equipment 124, costs of identifying substrate processing equipment parts that result in defective products, cleaning the substrate processing equipment parts, repairing the substrate processing equipment parts, replacing the substrate processing equipment parts, and discarding the old components, etc. By providing sensor data 142 to the model 190, receiving prediction data 160 from the model 190, and initiating corrective actions based on the prediction data 160, the system 100 has a technical advantage of avoiding the costs of producing, identifying, and discarding defective substrates.

[0052] In some embodiments, the prediction system 110 further includes a server machine 170 and a server machine 180. The server machine 170 includes a data set generator 172 capable of generating a data set (e.g., a set of data inputs and a set of target outputs) to train, validate and / or test (multiple) machine learning models 190. The data set generator 172 has the functions of data collection, compilation, reduction and / or partitioning to put the data in a form for machine learning. In some embodiments (e.g., for small data sets), partitions (e.g., explicit partitions) for post-training validation are not used. Repeated cross-validation (e.g., 5-fold cross-validation, leave-one-out cross-validation) can be used during training, where a given data set is actually repeatedly divided into different training sets and validation sets during training. A model (e.g., the best model and the model with the highest accuracy, etc.) is selected from the model vectors on the automatically separated combined subsets. In some embodiments, the data set generator 172 can explicitly divide historical data (e.g., historical sensor data 144 and corresponding historical performance data 154) into a training set (e.g., sixty percent of the historical data), a validation set (e.g., twenty percent of the historical data), and a test set (e.g., twenty percent of the historical data). In this embodiment, reference will be made to Figure 2 and Fig. 7A Detailed description of some operations of the dataset generator 172. In some embodiments, the prediction system 110 (e.g., via the prediction component 114) generates multiple sets of features (e.g., training features). In some examples, the first set of features corresponds to a first set of sensor data of a type corresponding to each dataset (e.g., training set, validation set, and test set) (e.g., from a first set of sensors, a first combination of values ​​from the first set of sensors, and a first pattern in the values ​​from the first set of sensors), and the second set of features corresponds to a second set of sensor data of a type corresponding to each dataset (e.g., from a second set of sensors different from the first set of sensors, a second combination of values ​​different from the first combination, a second pattern different from the first pattern).

[0053] The server machine 180 includes a training engine 182, a validation engine 184, a selection engine 185, and / or a test engine 186. In some embodiments, an engine (e.g., a training engine 182, a validation engine 184, a selection engine 185, and a test engine 186) refers to hardware (e.g., circuits, dedicated logic, programmable logic, microcode, and processing devices, etc.), software (such as instructions running on a processing device, a general-purpose computer system, or a dedicated computer), firmware, microcode, or a combination of the above. The training engine 182 is capable of training a machine learning model 190 using one or more sets of features associated with a training set from the data set generator 172. In some embodiments, the training engine 182 generates multiple trained machine learning models 190, wherein each trained machine learning model 190 corresponds to a different set of parameters (e.g., sensor data 142) and corresponding responses (e.g., performance data 152) of the training set. In some embodiments, multiple models are trained on the same parameters with different objectives for the purpose of modeling multiple effects. In some examples, a first trained machine learning model is trained using sensor data 142 from all sensors 126 (e.g., sensors 1-5), a second trained machine learning model is trained using a first subset of the sensor data (e.g., from sensors 1, 2, and 4), and a third trained machine learning model is trained using a second subset of sensor data that partially overlaps with the first subset of features (e.g., from sensors 1, 3, 4, and 5).

[0054] The validation engine 184 can validate the trained machine learning models 190 using a corresponding feature set of a validation set from the dataset generator 172. For example, a first set of features of the validation set is used to validate a first trained machine learning model 190 trained using a first set of features of the training set. The validation engine 184 determines the accuracy of each trained machine learning model 190 based on the corresponding feature set of the validation set. The validation engine 184 evaluates and marks (e.g., discards) trained machine learning models 190 having an accuracy that does not meet a threshold accuracy. In some embodiments, the selection engine 185 can select one or more trained machine learning models 190 having an accuracy that meets a threshold accuracy. In some embodiments, the selection engine 185 can select a trained machine learning model 190 with the highest accuracy among the trained machine learning models 190.

[0055] The testing engine 186 can test the trained machine learning model 190 using the corresponding feature set of the test set from the dataset generator 172. For example, the first set of features of the test set is used to test the first trained machine learning model 190 trained using the first set of features of the training set. The testing engine 186 determines the trained machine learning model 190 with the highest accuracy among all the trained machine learning models based on the test set.

[0056] In some embodiments, the machine learning model 190 (e.g., for classification) refers to a model artifact created by the training engine 182 using a training set containing data inputs and corresponding target outputs (e.g., correctly classifying a condition or a sequential level corresponding to the training input). Patterns in the data set can be found, the data inputs can be mapped to the target outputs (correct classifications or levels), and the images that capture these patterns are provided to the machine learning model 190. In some embodiments, the machine learning model 190 uses one or more of Gaussian process regression (GPR), Gaussian process classification (GPC), Bayesian neural network, neural network Gaussian process, deep belief network, Gaussian mixture model, or other probabilistic learning methods. Non-probabilistic methods can also be used, including one or more of support vector machines (SVM), radial basis functions (RBF), clustering, nearest neighbor algorithm (k-NN), linear regression, random forest, neural network (e.g., artificial neural network). In some embodiments, the machine learning model 190 is a multivariate analysis (MVA) regression model.

[0057] The prediction component 114 provides the current sensor data 146 (e.g., as an input) to the trained machine learning model 190 and runs the trained machine learning model 190 (e.g., obtains one or more outputs based on the input). The prediction component 114 can determine (e.g., extract) the prediction data 160 from the trained machine learning model 190 and determine (e.g., extract) uncertainty data indicating a confidence level that the prediction data 160 corresponds to the current performance data 156. In some embodiments, the prediction component 114 or the corrective action component 122 uses the uncertainty data (e.g., an uncertainty function or an extraction function derived from the uncertainty function) to determine whether to use the prediction data 160 to perform a corrective action or whether to further train the model 190.

[0058] For purposes of illustration and not limitation, aspects of the present disclosure describe using historical data (i.e., prior data, historical sensor data 144, and historical performance data 154) to train one or more machine learning models 190, and providing current sensor data 146 to one or more trained probabilistic machine learning models 190 to determine prediction data 160. In other embodiments, (e.g., without using a trained machine learning model), a heuristic model or a rule-based model is used to determine prediction data 160. In other embodiments, a non-probabilistic machine learning model may be used. The prediction component 114 monitors the historical sensor data 144 and the historical performance data 154. In some embodiments, with respect to Figure 2 Any information described by data input 210 is monitored or otherwise used in a heuristic or rule-based model.

[0059] In some embodiments, the functionality of client device 120, prediction server 112, server machine 170, and server machine 180 are provided by a smaller number of machines. For example, in some embodiments, server machines 170 and 180 are integrated into a single machine, while in some other embodiments, server machine 170, server machine 180, and prediction server 112 are integrated into a single machine. In some embodiments, client device 120 and prediction server 112 are integrated into a single machine.

[0060] In general, functions described in one embodiment as being performed by client device 120, prediction server 112, server machine 170, and server machine 180 may also be performed on prediction server 112 in other embodiments, as appropriate. In addition, the functionality of a particular component may be performed by different or multiple components operating together. For example, in some embodiments, prediction server 112 determines a corrective action based on prediction data 160. In another example, client device 120 determines prediction data 160 based on data received from a trained machine learning model.

[0061] Additionally, the functionality of a particular component may be performed by different or multiple components operating together.In some embodiments, one or more of prediction server 112, server machine 170, or server machine 180 is accessed as a service provided to other systems or devices through an appropriate application programming interface (API).

[0062] In some embodiments, a "user" is represented as a single individual. However, other embodiments of the present disclosure encompass a "user" that is an entity controlled by multiple users and / or automated sources. In some examples, a group of individual users united as a group of administrators is considered a "user."

[0063] Although embodiments of the present disclosure are discussed in terms of determining predictive data 160 for edge defect detection of substrate processing equipment parts in a manufacturing facility (e.g., a substrate processing facility), in some embodiments, the present disclosure may also be generally applicable to quality detection. Embodiments may generally be applicable to determining the quality of a part based on different types of data.

[0064] Figure 2 A method for creating a machine learning model (e.g., Figure 1 A dataset generator 272 (e.g., Figure 1 In some embodiments, the dataset generator 272 is Figure 1 A portion of the server machine 170. Figure 2 The dataset generated by the dataset generator 272 can be used to train a machine learning model (for example, see Fig.7D ) to cause corrective action to be taken (see, for example, Fig. 7E ).

[0065] Dataset generator 272 (e.g., Figure 1 The dataset generator 172) creates a dataset for use in a machine learning model (e.g., Figure 1 The dataset generator 272 uses the historical sensor data 244 (e.g., Figure 1 Historical sensor data 144) and historical performance data 254 (e.g., Figure 1 The historical performance data 154) is used to create a data set. Figure 2 The system 200 shows a data set generator 272, a data input 210, and a target output 220 (eg, target data).

[0066] In some embodiments, the dataset generator 272 generates a dataset (e.g., a training set, a validation set, and a test set); the dataset includes one or more data inputs 210 (e.g., a training input, a validation input, and a test input) and one or more target outputs 220 corresponding to the data inputs 210. The dataset also includes mapping data that maps the data inputs 210 to the target outputs 220. The data inputs 210 are also referred to as "features," "attributes," or "information." In some embodiments, the dataset generator 272 provides the dataset to the training engine 182, the validation engine 184, or the testing engine 186, where the dataset is used to train, validate, or test the machine learning model 190. Reference Fig. 7A Some embodiments of generating a training set are further described.

[0067] In some embodiments, data set generator 272 generates data input 210 and target output 220. In some embodiments, data input 210 includes one or more sets of historical sensor data 244. In some embodiments, historical sensor data 244 includes one or more sets of sensor data from one or more types of sensors, combinations of sensor data from one or more types of sensors, patterns of sensor data from one or more types of sensors, and / or the like.

[0068] In some embodiments, the dataset generator 272 generates a first data input corresponding to the first set of historical sensor data 244A to train, validate, or test a first machine learning model, and the dataset generator 272 generates a second data input corresponding to the second set of historical sensor data 244B to train, validate, or test a second machine learning model.

[0069] In some embodiments, the data set generator 272 discretizes (e.g., segments) one or more of the data input 210 or the target output 220 (e.g., for use in a classification algorithm for a regression problem). The discretization of the data input 210 or the target output 220 (e.g., segmentation via a sliding window) converts the continuous values ​​of the variable into discrete values. In some embodiments, the discrete values ​​of the data input 210 indicate discrete historical sensor data 244 to obtain the target output 220 (e.g., discrete historical performance data 254).

[0070] The data input 210 and target output 220 for training, validating, or testing a machine learning model include information for a specific facility (e.g., for a specific substrate manufacturing facility). In some examples, the historical sensor data 244 and the historical performance data 254 are for the same manufacturing facility.

[0071] In some embodiments, the information used to train the machine learning model comes from a particular type of manufacturing equipment 124 with particular characteristics at a manufacturing facility, and allows the trained machine learning model to determine results for a particular group of manufacturing equipment 124 based on input of current parameters (e.g., current sensor data 146) associated with one or more components that share the characteristics of the particular group. In some embodiments, the information used to train the machine learning model is for components from two or more manufacturing facilities, and allows the trained machine learning model to determine results for components based on input from one manufacturing facility.

[0072] In some embodiments, after a data set is generated and used to train, validate, or test the machine learning model 190, the machine learning model 190 is further trained, validated, or tested (e.g., Figure 1current performance data 156) or adjustments (e.g., adjusting weights associated with input data to the machine learning model 190, such as connection weights in a neural network).

[0073] Figure 3 is a diagram illustrating a method for generating prediction data 360 (e.g., Figure 1 The frame resistance system 300 of the system 300 of the prediction data 160) is used to predict the frame resistance of the system 300 through a trained machine learning model (e.g., Figure 1 Model 190) determines prediction data 360 for edge defect detection (e.g., for performing corrective action).

[0074] At block 310, the system 300 (e.g., Figure 1 Prediction system 110) (e.g., via Figure 1 The data set generator 172 of the server machine 170 executes the historical data (e.g., Figure 1 The system 300 partitions the data into historical sensor data 344 and historical performance data 354 of the model 190 to generate a training set 302, a validation set 304, and a test set 306. In some examples, the training set is 60% of the historical data, the validation set is 20% of the historical data, and the test set is 20% of the historical data. The system 300 generates multiple sets of features for each of the training set, validation set, and test set. In some examples, if the historical data includes data from 20 sensors (e.g., Figure 1 For example, if the training set is composed of features derived from sensors 126 and 100 products (e.g., each product corresponds to sensor data from 20 sensors), the first set of features is sensors 1-10, the second set of features is sensors 11-20, the training set is products 1-60, the validation set is products 61-80, and the test set is products 81-100. In this example, the first set of features for the training set would be the parameters from sensors 1-10 for products 1-60.

[0075] At block 312, the system 300 uses the training set 302 (e.g., via Figure 1The system 300 may use a training engine 182 of the training set 302 to perform model training. In some embodiments, the system 300 uses multiple sets of features of the training set 302 (e.g., a first set of features of the training set 302 and a second set of features of the training set 302, etc.) to train multiple models. For example, the system 300 trains machine learning models to generate a first trained machine learning model using a first set of features in the training set (e.g., sensor data from sensors 1-10 for products 1-60) and to generate a second trained machine learning model using a second set of features in the training set (e.g., sensor data from sensors 11-20 for products 1-60). In some embodiments, the first trained machine learning model and the second trained machine learning model are combined to generate a third trained machine learning model (e.g., in some embodiments, the third trained machine learning model is a better predictor than the first trained machine learning model or the second trained machine learning model itself). In some embodiments, the feature sets used to compare the models overlap (e.g., the first set of features is sensor data from sensors 1-15, and the second set of features is sensor data from sensors 5-20). In some embodiments, hundreds of models are generated, including models with various feature permutations and model combinations.

[0076] At block 314, the system 300 (eg, via Figure 1 The validation engine 184 of the validation set 304 performs model validation using the validation set 304. The system 300 validates each trained model using the corresponding feature set of the validation set 304. For example, the system 300 validates the first trained machine learning model using the first set of features in the validation set (e.g., parameters from sensors 1-10 for products 61-80), and validates the second trained machine learning model using the second set of features in the validation set (e.g., parameters from sensors 11-20 for products 61-80). In some embodiments, the system 300 validates hundreds of models generated at box 312 (e.g., models with various feature arrangements and combinations of models, etc.). At box 314, the system 300 (e.g., after model validation) determines the accuracy of each of the one or more models, and determines whether the one or more trained models have an accuracy that meets the threshold accuracy. In response to determining that none of the training models have an accuracy that meets the threshold accuracy, the process returns to box 312; in box 312, the system 300 performs model training using a different feature set of the training set. In response to determining that one or more training models have an accuracy that meets a threshold accuracy, flow continues to block 316. The system 300 discards trained machine learning models that have an accuracy below a threshold accuracy (e.g., based on a validation set).

[0077] At block 316, the system 300 (eg, via Figure 1The selection engine 185 of the system 300 performs model selection to determine which of the one or more training models that meet the threshold accuracy has the highest accuracy (e.g., the selected model 308, based on the verification of block 314). In response to determining that two or more trained models that meet the threshold accuracy have the same accuracy, the process returns to block 312; in block 312, the system 300 performs model training using the further refined training set corresponding to the further refined feature set to determine the trained model with the highest accuracy.

[0078] At block 318, system 300 uses test set 306 to (e.g., via Figure 1 The system 300 performs model testing to test the selected model 308 using the test engine 186 of the system 300. The system 300 tests the first trained machine learning model using the first set of features in the test set (e.g., sensor data from sensors 1-10 for products 81-100) to determine that the first trained machine learning model meets the threshold accuracy (e.g., based on the first set of features of the test set 306). In response to the accuracy of the selected model 308 not meeting the threshold accuracy (e.g., the selected model 308 overfits the training set 302 and / or the validation set 304 and is not applicable to other data sets such as the test set 306), the process continues to box 312; in box 312, the system 300 performs model training (e.g., retraining) using a different training set corresponding to a different set of features (e.g., sensor data from a different sensor). In response to determining that the selected model 308 has an accuracy that meets the threshold accuracy based on the test set 306, the process continues to box 320. At least in box 312, the model learns patterns in the historical data to make predictions, and in box 318, the system 300 applies the model to the remaining data (e.g., the test set 306) to test the predictions.

[0079] At block 320, the system 300 receives current sensor data 346 (eg, Figure 1 146 ) and determining (e.g., extracting) predicted data 360 (e.g., Figure 1 In some embodiments, the current sensor data 346 corresponds to features of the same type as in the historical sensor data 344. In some embodiments, the current sensor data 346 corresponds to features of the same type as a subset of the feature types in the historical sensor data 344 used to train the selected model 308.

[0080] In some embodiments, current data is received. In some embodiments, the current data includes current performance data 356 (e.g. Figure 1Current performance data 156) and / or current sensor data 346. In some embodiments, at least a portion of the current data is from a metering device (e.g. Figure 1 In some embodiments, the model 308 is retrained based on current data. In some embodiments, a new model is trained based on current performance data 356 and current sensor data 346.

[0081] In some embodiments, one or more of blocks 310-320 occur in various orders and / or with other operations not presented and described herein. In some embodiments, one or more of blocks 310-320 are not performed. For example, in some embodiments, one or more of the data partitioning of block 310, the model validation of block 314, the model selection of block 316, and / or the model testing of block 318 are not performed.

[0082] FIG. 4A to FIG. 4C A substrate processing equipment component 410 is shown in accordance with certain embodiments. Figure 4A is a top view of a substrate processing equipment part 410, Figure 4B is a perspective view of a substrate processing equipment part 410, and Figure 4C is a perspective view of a substrate processing equipment part 410 and an image capture device 400 .

[0083] The substrate processing equipment part 410 can be a showerhead, a pedestal, an edge ring, an electrostatic chuck, etc. In some embodiments, the substrate processing equipment part 410 is cylindrical (e.g., a circular periphery). In some embodiments, the upper surface 414 of the substrate processing equipment part 410 forms one or more recesses 412 (e.g., a pedestal recess, a wafer recess, and a pedestal recess). In some embodiments, the upper surface 414 of the substrate processing equipment part 410 forms two or more recesses 412 (e.g., two or more pedestal recesses). In some embodiments, the upper surface 414 of the substrate processing equipment part 410 forms three or more recesses 412 (e.g., three or more pedestal recesses). In some embodiments, the upper surface of the substrate processing equipment part 410 forms four or more recesses 412 (e.g., four or more pedestal recesses). In some embodiments, the upper surface of the substrate processing equipment part 410 forms five or more recesses 412 (e.g., five or more pedestal recesses). In some embodiments, six or more recesses 412 (eg, six or more pedestal pockets) are formed on the upper surface of the substrate processing equipment part 410 .

[0084] Each recess 412 (e.g., a pocket) may be further formed by one or more sidewalls 418 and a lower wall. Each recess 412 may be formed by a sidewall 418 having a substantially circular perimeter and a substantially vertical height. The sidewall 418 may be substantially perpendicular (e.g., 90-95 degrees, 90-100 degrees, 95-100 degrees, etc.) to the lower wall and / or upper surface 414. Each recess 412 may be formed by a lower wall; the lower wall is substantially planar, substantially parallel to the upper surface 414 of the substrate processing equipment part 410, and / or substantially perpendicular (e.g., 90-95 degrees, 90-100 degrees, 95-100 degrees, etc.) to the sidewall 418. The sidewall 418 and the upper surface 414 may intersect at an edge 416 (e.g., an upper edge, a circular edge). The sidewall 418 and the lower surface may intersect at a lower edge. Defects in the edge 416 (eg, the upper edge where the sidewall 418 and the upper surface 414 intersect) may result in defects in substrates produced by the substrate processing equipment part 410 .

[0085] The substrate processing equipment part 410 can be configured to receive a substrate (e.g., a substrate having a circular perimeter) in the recess 412 for performing a substrate processing operation. The substrate processing equipment part 410 can be disposed in a processing chamber; the processing chamber is configured to perform one or more substrate processing operations. The substrate processing equipment part 410 and / or one or more portions of the processing chamber can be rotated relative to each other so that different substrate processing operations can be performed on a substrate disposed in the recess 412. The recesses 412 (e.g., and substrates disposed therein) can be separated from each other by barriers (e.g., air curtains, partitions, etc.).

[0086] like Figure 4C As shown, an angle recognition component 420 (e.g., a protector) may be disposed in a recess 412 formed by a substrate processing equipment part 410 (e.g., a protector may be placed on top of a base recess to define an angle across a circumferential edge). An image capture device 400 may be used to capture an image (e.g., sensor data 142) of the angle recognition component 420 and edge 416 (where a sidewall 418 and an upper surface 414 intersect) of a substrate processing equipment part 410 (e.g., a base recess) forming a recess 412 (e.g., a base recess).

[0087] The angle recognition component 420 may have markings (e.g., every degree, every five degrees, etc.). The angle recognition component 420 may be a protector placed on top of a base recess to correctly define the angle across the circumferential edge. The image capture device 400 may include a light (e.g., a light emitting diode (LED)), a diffuser (e.g., a white diffuser), and a camera configured to capture input images at constant angular intervals. In some embodiments, the image capture device 400 captures video and derives images from the video.

[0088] In some embodiments, the image capture device 400 is moved (e.g., manually) to different positions to capture images. In some embodiments, the image capture device 400 includes an actuator (e.g., a motor); the actuator moves the camera (e.g., automatically) to different positions to capture images.

[0089] FIG. 5A to FIG. 5E A substrate processing equipment component 510 (eg, FIG. 4A to FIG. 4C Images 500A-500E of substrate processing equipment parts 410).

[0090] like Figure 5A As shown, an image 500A (eg, an original image, an original input image) may be captured. The angle recognition component 520 (eg, Figure 4C An angle recognition component 420) is placed on or in a recess 512 (e.g., a base recess) to capture images to correctly define angles across the circumferential edge. A light (e.g., an LED) with a diffuser (e.g., a white diffuser) and a camera can be used to capture input images (e.g., still images and / or video) at constant angular intervals.

[0091] Image 500A may include at least a predetermined amount of markings (eg, at least 5 degrees) of angle identification component 520. Image 500A may include forming recess 512 (eg, FIG. 4A to FIG. 4C An edge 516 of a side wall 418 of the recess 412, base recess (eg, where the side wall and the upper surface intersect).

[0092] refer to Figure 5B , image 500B may be image 500A that has been processed. Image 500A may be converted to grayscale to generate image 500B. Vertical and horizontal crop windows may highlight edges in grayscale.

[0093] refer to Figure 5C , image 500C (e.g., grayscale cropped image) may be image 500A and / or image 500B that has been processed. Dynamic bidirectional cropping may be applied to image 500B to obtain a specific region (e.g., the edge of the base) within a given angular interval (e.g., 5 degrees) in the image.

[0094] refer to Figure 5D, image 500D may be processed image 500A, image 500B, and / or image 500C. Image 500D may be a threshold image after preliminary cropping. Lines in image 500D may mark the asymmetry of image 500D relative to the center line. For edge segmentation, adaptive thresholding may be applied based on a predetermined grid size (e.g., an optimized grid size) to convert into a binary image (e.g., having only black pixels and white pixels). To eliminate background distortion, contour detection may be applied. Contours with smaller areas may be eliminated. For a more detailed assessment, the background may be additionally smoothed to highlight the edges of the base. Smoothing may include reducing the size of a shape to be less than a threshold size (e.g., a small dot) to separate the shape from the edge so that the shape can be removed.

[0095] refer to Figure 5E Image 500E may be image 500A, image 500B, image 500C, and / or image 500D that have been processed. Image 500E may be the final cropped image after removing background distortion and smoothing the background. In image 500E, the distribution of pixels highlighting edge cracks may be used to perform dynamic rotation (e.g., to make the image symmetrical), and a final crop may be performed.

[0096] FIG. 6A to FIG. 6D A substrate processing equipment component 610 (eg, FIG. 4A to FIG. 4C The substrate processing equipment part 410, and FIG. 5A to FIG. 5E An image 600 of a substrate processing equipment part 510 ).

[0097] FIG. 6A to FIG. 6C It can be a binary image. In a binary image, white pixels can highlight the outline of edge defects. In some embodiments, linear scaling is applied to convert pixels into metric units (e.g., millimeters (mm)). The total number of white images can be calculated to analyze the edge (e.g., damaged area). The total vertical height of the image can indicate the extended height of the edge due to the crack. The "good" or "bad" classification can be performed by counting (e.g., the number) of white pixels and comparing the extended height of the edge with the standard height. The entire process can be repeated for all possible angular intervals for each recess of the base.

[0098] refer to Fig. 6A, image 600A may be a substrate processing equipment part 610 having an edge that satisfies a threshold value for an attribute (e.g., the substrate processing equipment part has a "good" classification). Image 600A may display a standard reference height and a relatively small number of white pixels on top. The height of image 600A may not satisfy the threshold height (e.g., the height is less than the height indicating an edge defect). The number of white pixels of image 600A may not satisfy a threshold amount (e.g., the number of white pixels is less than the number of white pixels indicating an edge defect).

[0099] Referring to figure 613, image 600B may be a substrate processing equipment part 610 having an edge with a property value that does not meet a threshold value (e.g., the substrate processing equipment part has a "good" classification). Image 600B may have an extended height and a greater number of white pixels at the top. The height of image 600B may meet the threshold height (e.g., the height is equal to or greater than the height indicating an edge defect). The number of white pixels of image 600B may meet the threshold amount (e.g., the number of white pixels is equal to or greater than the number of white pixels indicating an edge defect).

[0100] refer to Figure 6C , image 600C may be a substrate processing equipment part 610 having an edge with a property value that does not meet a threshold. A height 604 of image 600C meets a threshold height (e.g., is greater than a height of a pedestal edge that produces substrates having performance data that meets the threshold). A number of pixels 602 of one type (e.g., white pixels) meets a threshold amount (e.g., a number of white pixels representing an area of ​​an edge crack is greater than a number of white pixels of a pedestal edge that produces substrates having performance data that meets the threshold).

[0101] refer to Fig.6D , image 600D may be of a substrate processing equipment part 610. The substrate processing equipment part 610 may include an upper surface 614 forming one or more recesses 612 (e.g., base pockets). The recess 612 may be formed by a sidewall 618 and a lower surface. The intersection of the sidewall 618 and the upper surface 614 may be an edge 616 (e.g., an upper edge).

[0102] The x distance 620, y distance 630, and / or z distance 640 may be measured for the substrate processing equipment part 610. The boundary y bottom 632 may be the lower edge where the lower surface and the sidewall 618 intersect. The boundary x top 622 and the boundary x bottom 624 may form a boundary between the sidewall 618 and the upper surface 614 of the substrate processing equipment part 610. An edge defect (e.g., a cracked edge) at the top of the edge 616 causes the x distance 620 (e.g., the height of the edge crack) to increase, causes the z distance 640 (e.g., the width of the edge crack) to increase, and causes the y distance 630 (e.g., the distance between the boundary y bottom 632 and the boundary x bottom 624) to decrease. The height 604 may be the sum of the x distance 620 and the y distance 630.

[0103] In some embodiments, height 604 is determined via automation (eg, via processing logic). In some embodiments, the automatic height of edge 616 is compared to a manually calculated width of edge 616 (eg, x distance 620, (X+Y)-Y).

[0104] 7A to 7E 700A-700E are flow charts of methods 700A-700E associated with edge defect detection according to certain embodiments. In some embodiments, 700A-700E are performed by processing logic; the processing logic includes hardware (e.g., circuit systems, dedicated logic, programmable logic, microcode, and processing devices, etc.), software (such as instructions running on a processing device, a general-purpose computer system or a dedicated machine), firmware, microcode, or a combination of the above. In some embodiments, methods 700A-700E are performed at least in part by prediction system 110 and / or client device 120. In some embodiments, method 700A is performed at least in part by prediction system 110 (e.g., Figure 1 The server machine 170 and the data set generator 172, and Figure 2 In some embodiments, the prediction system 110 uses method 700A to generate a data set to perform at least one of training, validating, or testing a machine learning model. In some embodiments, methods 700B-700C are performed by a client device 120. In some embodiments, method 700D is performed by a server machine 180 (e.g., training engine 182, etc.). In some embodiments, method 700E is performed by a prediction server 112 (e.g., prediction component 114). In some embodiments, method 700C is performed by a client device 120 (e.g., corrective action component 122). In some embodiments, a non-transitory storage medium stores instructions that, when executed by a processing device (e.g., of the prediction system 110, of the server machine 180, of the prediction server 112, etc.), cause the processing device to perform one or more of methods 700A-700E.

[0105] For ease of explanation, methods 700A-700E are depicted and described as a series of operations. However, operations according to the present disclosure may occur in various orders and / or simultaneously with other operations not presented and described herein. Furthermore, in some embodiments, not all of the operations shown need to be performed to implement methods 700A-700E according to the disclosed objectives. Additionally, one of ordinary skill in the art will understand and appreciate that methods 700A-700E may alternatively be represented as a series of interrelated states via state diagrams or events.

[0106] In some embodiments, one or more of methods 700A-700E is used to distinguish good parts from bad parts, eliminate manual inspection and manual judgment regarding the end of part life, determine the end of life of a base, identify patterns of crack locations and use them to improve coating (e.g., silicon carbide (SiC)) processing, correlate the occurrence of cracks with deposition conditions, and use them to move the processing window to a state that reduces cracks, detect cracks within a threshold accuracy, and / or one or more of the like.

[0107] Fig. 7A 700A is a flow chart of a method for generating a data set for a machine learning model for generating prediction data (e.g., prediction data 160 of FIG. 7).

[0108] refer to Fig. 7A In some embodiments, at block 702, processing logic implementing method 700A initializes a training set T to an empty set.

[0109] At block 704, processing logic generates a message containing sensor data (e.g., Figure 1 Historical sensor data 144, Figure 2 In some embodiments, the first data input includes a first set of features for the sensor data type, and the second data input includes a second set of features for the sensor data type (e.g., such as historical sensor data 244). Figure 2 described).

[0110] At block 706, processing logic generates a first target output for one or more data inputs (e.g., a first data input). In some embodiments, the first target output is historical performance data (e.g., Figure 1 Historical performance data154, Figure 2 Historical performance data 254).

[0111] At block 708, processing logic may optionally generate image data indicating an input / output image. An input / output image (or image data) refers to data inputs (e.g., one or more data inputs described herein), target outputs of the data inputs (e.g., where the target output identifies historical performance data 154), and associations between the data input(s) and the target outputs.

[0112] At block 710 , processing logic adds the image data generated at block 708 to data set T.

[0113] At block 712, the processing logic branches based on whether the dataset T is sufficient to perform at least one of training, validating, and / or testing the machine learning model 190 (e.g., the uncertainty of the trained machine learning model satisfies the threshold uncertainty). If so, execution proceeds to block 714; otherwise, execution continues back to block 704. It should be noted that in some embodiments, the sufficiency of the dataset T is determined simply based on the number of input / output mappings in the dataset; while in some other implementations, the sufficiency of the dataset T is determined based on one or more other criteria (e.g., a measure of diversity and accuracy of data examples, etc.) in addition to or in lieu of the number of input / output mappings.

[0114] At block 714, processing logic provides (e.g., to the server machine 180) a dataset T to train, validate, and / or test the machine learning model 190. In some embodiments, the dataset T is a training set and is provided to the training engine 182 of the server machine 180 to perform training. In some embodiments, the dataset T is a validation set and is provided to the validation engine 184 of the server machine 180 to perform validation. In some embodiments, the dataset T is a test set and is provided to the testing engine 186 of the server machine 180 to perform testing. For example, in the case of a neural network, the input values ​​of a given input / output image (e.g., the numerical values ​​associated with the data input 210) are input to the neural network, and the output values ​​of the input / output image (e.g., the numerical values ​​associated with the target output 220) are stored in the output nodes of the neural network. The connection weights in the neural network are then adjusted according to a learning algorithm (e.g., back-propagation, etc.), and this process is repeated for other input / output mappings in the dataset T.

[0115] After block 714, the machine learning model (e.g., machine learning model 190) may be at least one of trained using the training engine 182 of the server machine 180, validated using the validation engine 184 of the server machine 180, or tested using the testing engine 186 of the server machine 180. The machine learning model is implemented by the prediction component 114 (of the prediction server 112) to generate prediction data (e.g., prediction data 160) for edge defect detection, resulting in corrective action being performed.

[0116] Figure 7B Method 700B is associated with edge defect detection via image analysis in accordance with certain embodiments. In some embodiments, method 700B is performed before and after cleaning of substrate processing equipment parts. Via method 700B, processing logic may dynamically detect the edge of a pedestal cavity from a digital image, and processing logic may evaluate height and edge characteristics to further classify the edge of the pedestal cavity as meeting a threshold condition (e.g., being in good or bad condition).

[0117] At block 720 of method 700B, processing logic identifies a substrate processing equipment part (e.g., FIG. 4A to FIG. 4C An image of an edge of a substrate processing equipment part 410). The substrate processing equipment part may be a base, an edge ring, an electrostatic chuck, etc.

[0118] In some embodiments, processing logic receives information from an image capture device (e.g., Figure 4C The image capture device 400 receives the image. The angle recognition component (e.g., Figure 4C The angle recognition component 420 of the image capture device can be placed in or on a recess formed by the substrate processing equipment part (e.g., a base recess formed by the base). The image capture device can project light (e.g., via an LED light with a white diffuser) to the edge of the substrate processing equipment part and the angle recognition component. The camera of the image capture device can capture images with constant angular intervals of the edge and the angle recognition component. In some embodiments, the image capture device can capture a video containing the image.

[0119] At block 722, processing logic predicts whether the attribute value of the edge satisfies a threshold based on the image. Processing logic may perform Figure 7C Method 700C, Fig.7D Method 700D and / or Fig. 7E One or more blocks of method 700E are performed to predict whether a property value of an edge satisfies a threshold.

[0120] In some embodiments, the property value includes one or more of the height of the edge, the number of pixels associated with the deformation of the edge, etc. The threshold value may indicate a substrate processing equipment part that is predicted to produce substrates that do not have performance data that meets the threshold value (e.g., defective). The threshold value may indicate a substrate processing equipment part that is to undergo corrective action (e.g., cleaning, repair, replacement, etc.) to produce substrates with performance data that meets the threshold value (e.g., good wafers).

[0121] Processing logic may determine the location (e.g., angular location of a defect present at the circumference of the cavity), size (e.g., height of a defect present at the circumference of the cavity), and / or number of defects on an edge (e.g., a wafer cavity).

[0122] At block 724, in response to the attribute value satisfying the threshold, flow continues to block 726. At block 724, in response to the attribute value not satisfying the threshold, flow continues to block 720, where a subsequent image of the edge of the substrate processing equipment part is identified and method 700B is repeated.

[0123] At block 726, processing logic causes execution of a corrective action associated with the substrate processing equipment part. The corrective action may include one or more of providing an alarm, causing a cleaning action, causing a repair action, causing a replacement, and determining a predicted end of life of the substrate processing equipment part.

[0124] Blocks 720-726 may be repeated until the property value of each edge meets the threshold. In some embodiments, processing logic may predict the end of life of a substrate processing equipment part based on the number of corrective actions (e.g., cleaning cycles) until the property value of the edge of the substrate processing equipment part meets the threshold.

[0125] Figure 7C is a method associated with edge defect detection via image analysis according to certain embodiments. In some embodiments, method 700B is performed before and after cleaning of substrate processing equipment parts.

[0126] At block 730, processing logic identifies an image of an edge of a substrate processing equipment component (eg, see Figure 5A ). Figure 7C The frame 730 may be similar to Figure 7B Frame 720.

[0127] At block 732, processing logic converts the image to grayscale (see, e.g., Figure 5B ). In some embodiments, the image is captured in grayscale.

[0128] At block 734, processing logic crops the image by bidirectional cropping (e.g., see FIG. 5B to FIG. 5C). Dynamic bidirectional cropping may be applied to obtain an area of ​​the image (eg, the edge of the base) within a predetermined angular interval (eg, five degrees).

[0129] At block 736, processing logic converts the image to a binary image (eg, for edge segmentation) via an adaptive thresholding process based on a predetermined grid size (eg, an optimized grid size). A binary image may include only black pixels and white pixels.

[0130] In some embodiments, processing logic applies a threshold pixel value to an image. In some embodiments, the pixel format of the image is a byte image, where pixel values ​​are numbers stored as 8-bit integers, which give a possible value range from 0 to 255, where 0 is black and 255 is white. Processing logic may convert all pixel values ​​above the threshold pixel value (e.g., 200) to white (e.g., a pixel value of 255), and may convert all pixel values ​​below the threshold pixel value (e.g., 200) to black (e.g., a pixel value of 0). This may remove gray pixel values ​​from the image.

[0131] At block 738, processing logic removes background distortion in the image by applying contour detection. Contours having a small area (e.g., an area less than a threshold) may be eliminated. For viewfinder evaluation, additional smoothing may be performed on the background to highlight the edges of substrate processing equipment parts. Smoothing may include reducing the size of a shape less than a threshold size (e.g., a small dot) to separate the shape from the edge so that the shape can be removed (e.g., via removal of background distortion in block 738).

[0132] In some embodiments, the centroid of the edge in the y direction is determined. If the contour is a threshold distance from the centroid, processing logic may remove the contour.

[0133] At block 740, processing logic dynamically rotates the image so that at least a portion of the image is substantially symmetrical (e.g., Figure 5D ). The image can be rotated so that the lower left black pixels of the image and the lower right black pixels of the image are symmetrical around the center line.

[0134] At block 742, processing logic performs a final crop of the image. The final crop may be performed after removing background distortion and smoothing the background.

[0135] At block 744, processing logic applies linear scaling to the image to convert the units of pixels (eg, convert pixels to metric units such as millimeters). White pixels may highlight the contours occupied by edge defects.

[0136] At block 746, processing logic determines, based on the image, a height of a portion of the image associated with the edge (e.g., Figure 6C604) or pixels associated with edge cracks (e.g., Figure 6C The total number of white pixels may be counted to analyze damaged areas (e.g., edge defects). The total vertical height of the image may indicate the extended height of the edge due to the crack. The “good” or “bad” classification may be performed via counting of white pixels and comparing the extended height of the edge with a standard height.

[0137] At block 748, processing logic determines whether the height of the edge satisfies a threshold height. In response to the height of the edge satisfying the threshold height, flow continues to block 752. In response to the height of the edge not satisfying the threshold height, flow continues to block 750.

[0138] At block 750, processing logic determines whether the number of pixels satisfies a threshold amount. In response to the number of pixels satisfying the threshold height, flow continues to block 752. In response to the height of the edge not satisfying the threshold height, flow continues to block 730. The method 700C may be repeated for all angular intervals (e.g., every 5 degree interval around the edge) for each pocket of a substrate processing equipment part (e.g., a pedestal).

[0139] At block 752, processing logic causes execution of a corrective action associated with a substrate processing equipment part. Figure 7C The frame 752 may be similar to Figure 7B Frame 726.

[0140] Fig.7D is used to train machine learning models (e.g. Figure 1 A method of a machine learning model 190 for determining prediction data for edge defect detection via image analysis (e.g., Figure 1 prediction data 160).

[0141] refer to Fig.7D At block 760 of method 700D, processing logic identifies historical sensor data (e.g., Figure 1 The historical sensor data 144, the historical input sensor data). The historical sensor data may include historical images of the edge of the substrate processing equipment. Figure 7C One or more operations of method 700C process the historical image (eg, the historical image is processed via one or more of blocks 732-744).

[0142] In some embodiments, a network of available images of substrate processing equipment parts is available. In some embodiments, the second portion of the image is of a substrate processing equipment part having an attribute value that satisfies a threshold value (e.g., a bad pedestal image with a broken edge, see FIG. 6B to FIG. 6CIn some embodiments, the first portion of the image is of a substrate processing equipment part having a property value that does not meet a threshold value (eg, a good pedestal image, see Fig. 6A ).

[0143] At block 762, processing logic identifies historical performance data (e.g. Figure 1 The historical sensor data and at least a portion of the historical performance data may be associated with a new substrate processing equipment part (e.g., for benchmarking). The historical performance data may be an indication of whether a property value of the substrate processing equipment part meets a threshold. The historical performance data may indicate whether the substrate processing equipment part is new. The historical performance data may indicate whether the substrate processing equipment part is defective. The historical performance data may be an indication of whether performance data (e.g., property value, defect amount, etc.) of a substrate produced by the substrate processing equipment part meets a threshold (e.g., a good wafer or a bad wafer).

[0144] At box 764, processing logic trains the machine learning model using the data input comprising historical sensor data and the target output comprising historical performance data to generate a trained machine learning model.

[0145] In some embodiments, processing logic (e.g., via a trained machine learning model) correlates chamber wafer performance (e.g., defects, uniformity, etc.) to quantified defects on a susceptor. Defect size and location can be compared and tracked (e.g., via digital logging) between cleaning operations, between susceptors, and / or identify potential problem areas on substrate processing equipment parts related to chamber design and / or process conditions. In some examples, if a substrate has a substrate defect that is close to the same location as an edge defect on a substrate processing equipment part, corrective action will be performed.

[0146] In some embodiments, the historical sensor data of box 760 includes historical images of historical substrate processing equipment parts, and the historical performance data of box 762 corresponds to historical substrate processing equipment parts. The historical performance data may be associated with substrate quality; substrate quality such as metrology data of substrates, substrate throughput, and substrate defects. The historical performance data may be associated with the quality of substrate processing equipment parts; the quality of substrate processing equipment parts such as manual inspection, metrology data of substrate processing equipment parts, and failure time of substrate processing equipment parts. At box 764, a machine learning model may be trained using data inputs comprising historical images and target outputs comprising historical performance data to generate a trained machine learning model configured to predict whether a property value of a substrate processing equipment part meets a threshold. Figure 7B At box 722, the processing logic may use the trained machine learning model trained by method 700D of Figure D to determine whether the attribute value meets the threshold.

[0147] Fig. 7E is used to use a trained machine learning model (e.g. Figure 1 Method 700E of performing edge defect detection on a model 190) to cause execution of corrective action.

[0148] refer to Fig. 7E At block 780 of method 700E, processing logic identifies sensor data. In some embodiments, the sensor data of block 780 includes an image of a substrate processing equipment component. Fig. 7E The frame 780 may be similar to Figure 7B Frame 720.

[0149] At block 782, processing logic provides sensor data as data input to (e.g., via Fig.7D A trained machine learning model trained by box 764).

[0150] At box 784, processing logic receives output associated with the prediction data from the trained machine learning model.

[0151] At block 786, processing logic causes performance of corrective action based on the predicted data. Fig. 7E The frame 786 may be similar to Figure 7B Frame 726.

[0152] Figure 8 800 is a block diagram illustrating a computer system 800 according to some embodiments. In some embodiments, computer system 800 is one or more of client device 120, prediction system 110, server machine 170, server machine 180, or prediction server 112.

[0153] In some embodiments, the computer system 800 is connected (e.g., via a network, such as a local area network (LAN), an intranet, an extranet, or the Internet) to other computer systems. In some embodiments, the computer system 800 operates with the capabilities of a server or client computer in a client-server environment, or operates as a peer computer in a peer-peer or distributed network environment. In some embodiments, the computer system 800 is composed of a personal computer (PC), a tablet PC, a set-top box (STB), a personal digital assistant (PDA), a cellular phone, a network device, a server, a network router, a switch or a bridge, or any device capable of (sequentially or otherwise) executing a set of instructions that specify the actions to be taken by the device. In addition, the term "computer" should include any collection of computers that execute a set (or multiple sets) of instructions individually or jointly to perform any one or more methods described herein.

[0154] In a further aspect, the computer system 800 includes a processing device 802, a volatile memory 804 (e.g., a random access memory (RAM)), a non-volatile memory 806 (e.g., a read-only memory (ROM) or an electronically erasable programmable ROM (EEPROM)), and a data storage device 816, which communicate with each other via a bus 808.

[0155] In some embodiments, the processing device 802 is provided by one or more processors; the processor is such as a general-purpose processor (e.g., a complex instruction set computing (CISC) microprocessor, a reduced instruction set computing (RISC) microprocessor, a very long instruction word (VLIW) microprocessor, a microprocessor that implements other types of instruction sets, or a microprocessor that implements a combination of multiple types of instruction sets) or a special-purpose processor (e.g., an application-specific integrated circuit (ASIC), a field programmable gate array (FPGA), a digital signal processor (DSP), or a network processor).

[0156] In some embodiments, the computer system 800 further includes a network interface device 822 (e.g., coupled to the network 874). In some embodiments, the computer system 800 also includes a video display 810 (e.g., LCD), an alphanumeric input device 812 (e.g., keyboard), a cursor control device 814 (e.g., mouse), and a signal generating device 820.

[0157] In some embodiments, the data storage device 816 includes a non-transitory computer-readable storage medium 824 on which are stored instructions 826 encoding any one or more of the methods or functions described herein, including instructions for Figure 1 The components (eg, corrective action component 122, prediction component 114, etc.) of the system may be encoded with instructions that may be used to implement the methods described herein (eg, one or more of methods 700A-700E).

[0158] In some embodiments, instructions 826 also reside, completely or partially, within volatile memory 804 and / or within processing device 802 during execution by computer system 800; thus, in some embodiments, volatile memory 804 and processing device 802 also constitute machine-readable storage media.

[0159] Although the computer-readable storage medium 824 is shown as a single medium in the illustrative example, the term "computer-readable storage medium" shall include a single medium or multiple media (e.g., a centralized or distributed database and / or associated caches and servers) that store one or more sets of executable instructions. The term "computer-readable storage medium" shall also include any tangible medium that can store or encode a set of instructions for execution by a computer, which set of instructions causes the computer to perform any one or more of the methods described herein. The term "computer-readable storage medium" shall include, but is not limited to, solid-state memories, optical media, and magnetic media.

[0160] In some embodiments, the methods, components, and features described herein are implemented by separate hardware components or integrated into the functionality of other hardware components such as ASICs, FPGAs, DSPs, or similar devices. In some embodiments, the methods, components, and features are implemented by firmware modules or functional circuit systems within hardware devices. In some embodiments, the methods, components, and features are implemented in any combination of hardware devices and computer program components or in a computer program.

[0161] Unless otherwise specifically stated, terms such as "identify," "predict," "cause," "capture," "process," "convert," "crop," "threshold," "remove," "rotate," "provide," "obtain," "train," "further train," "retrain," "receive," "determine," "update," or the like refer to actions and processes performed or implemented by a computer system; a computer system manipulates and converts data represented as physical (electronic) quantities within computer system buffers and memories into other data similarly represented as physical quantities within computer system memories or buffers or other such information storage, transmission, or display devices. In some embodiments, the terms "first," "second," "third," and "fourth," etc., as used herein, are intended as labels to distinguish different components and do not have an ordinal meaning according to their numerical names.

[0162] The examples described herein also relate to devices for performing the methods described herein. In some embodiments, the device is specially configured to perform the methods described herein, or includes a general-purpose computer system selectively programmed by a computer program stored in the computer system. Such a computer program is stored in a computer-readable tangible storage medium.

[0163] The methods and illustrative examples described herein are not inherently related to any particular computer or other device. In some embodiments, various general purpose systems are used according to the teachings described herein. In some embodiments, more specialized devices are created to perform the methods described herein and / or each of their respective functions, routines, subroutines, or operations. Structural examples of various these systems are set forth in the description above.

[0164] The above description is intended to be illustrative rather than restrictive. Although the present disclosure has been described with reference to specific illustrative examples and embodiments, it should be appreciated that the present disclosure is not limited to the described examples and embodiments. The scope of the present disclosure should be determined with reference to the attached claims and the full scope of equivalents granted by the claims.

Claims

1. A method, comprising: an image identifying an edge of a susceptor cavity formed by a susceptor of a substrate processing system; predicting, based on the image, whether a property value of the edge of the base satisfies a threshold; and A corrective action associated with the base is caused to be performed in response to the property value of the edge satisfying a threshold. 2 . The method of claim 1 , wherein the property value comprises at least one of a height of the edge or pixels associated with a deformation of the edge.

3. The method of claim 1, wherein: The image is captured in response to an angle recognition component disposed in a recess of the base and light projected onto the edge and the angle recognition component; and At least a portion of the angle recognition component is in the image.

4. The method of claim 1, further comprising: capturing video of the edge by moving an image capture device along the edge; as well as The video is processed to identify a plurality of images at predetermined angular intervals, the plurality of images including the image.

5. The method of claim 1, wherein the image: is converted to grayscale; is clipped via dynamic bidirectional clipping; and The binary image is converted into a binary image via adaptive thresholding based on a predetermined grid size, the binary image comprising pixels of a first type and pixels of a second type, wherein the pixels of the first type show edge defects of the edge.

6. The method of claim 1, wherein at least a portion of background distortion in the image is removed via applying contour detection to the image.

7. The method of claim 1, wherein the image is dynamically rotated so that at least a portion of the image is substantially symmetrical.

8. The method of claim 1, wherein the predicting whether the attribute value satisfies the threshold comprises: providing the image as input to a trained machine learning model; obtaining an output associated with prediction data from the trained machine learning model; as well as A determination is made based on the prediction data whether the attribute value of the edge satisfies the threshold.

9. The method of claim 8, wherein the trained machine learning model is trained using an input comprising historical images of a historical base and a target output comprising historical performance data of the historical base.

10. A non-transitory computer-readable storage medium storing instructions that, when executed, cause a processing device to perform operations comprising: an image identifying an edge of a susceptor cavity formed by a susceptor of a substrate processing system; predicting, based on the image, whether a property value of the edge of the base satisfies a threshold; and A corrective action associated with the base is caused to be performed in response to the property value of the edge satisfying the threshold. 11 . The non-transitory computer-readable storage medium of claim 10 , wherein the property value comprises at least one of a height of the edge or pixels associated with a deformation of the edge.

12. The non-transitory computer-readable storage medium of claim 10, wherein: The image is captured in response to an angle recognition component disposed in a recess of the base and light projected onto the edge and the angle recognition component; and At least a portion of the angle recognition component is in the image.

13. The non-transitory computer-readable storage medium of claim 10, wherein the operations further comprise: capturing video of the edge by moving an image capture device along the edge; as well as The video is processed to identify a plurality of images at predetermined angular intervals, the plurality of images including the image.

14. The non-transitory computer-readable storage medium of claim 10, wherein the image: is converted to grayscale; is clipped via dynamic bidirectional clipping; and The binary image is converted into a binary image via adaptive thresholding based on a predetermined grid size, the binary image comprising pixels of a first type and pixels of a second type, wherein the pixels of the first type show edge defects of the edge.

15. The non-transitory computer-readable storage medium of claim 10, wherein: At least a portion of background distortion in the image is removed via applying contour detection to the image; and The image is dynamically rotated such that at least a portion of the image is substantially symmetrical.

16. The non-transitory computer-readable storage medium of claim 10, wherein the predicting whether the attribute value satisfies the threshold comprises: providing the image as input to a trained machine learning model; obtaining an output associated with prediction data from the trained machine learning model; as well as A determination is made based on the prediction data whether the attribute value of the edge satisfies the threshold.

17. A system, comprising: Memory; as well as a processing device, the processing device being coupled to the memory, the processing device: an image identifying an edge of a susceptor cavity formed by a susceptor of a substrate processing system; predicting, based on the image, whether a property value of the edge of the base satisfies a threshold; and A corrective action associated with the base is caused to be performed in response to the property value of the edge satisfying the threshold.

18. The system of claim 17, wherein the property value comprises at least one of a height of the edge or pixels associated with a deformation of the edge.

19. The system of claim 17, wherein: The image is captured in response to an angle recognition component disposed in a recess of the base and light projected onto the edge and the angle recognition component; and At least a portion of the angle recognition component is in the image.

20. The system of claim 17, wherein the processing device further: capturing video of the edge by moving an image capture device along the edge; and The video is processed to identify a plurality of images at predetermined angular intervals, the plurality of images including the image.

Images