Correcting component failure in ion implantation

Abstract

Methods, systems, and non-transitory computer-readable media are provided for correcting component failures in an ion implanted semiconductor manufacturing tool. A method includes the steps of: receiving current sensor data corresponding to a feature from a sensor associated with an ion implantation tool; performing feature analysis to generate additional features for the current sensor data; providing additional features as inputs to the trained machine learning model; obtaining one or more outputs from the trained machine learning model, wherein the one or more outputs indicate a confidence level for the predicted window; predicting whether one or more components of the ion implantation tool are within the pre-failure window based on the confidence level of the predicted window; and in response to predicting that the one or more components are within the pre-failure window, performing a corrective action associated with the ion implantation tool.

Description

[0001] This application is a divisional application of the invention patent application with the application date of January 17, 2020, application number 202080012043.5, and invention name “Correcting component failures in ion implantation semiconductor manufacturing tools”. Technical Field

[0002] The present disclosure relates to correcting component failures, and more particularly, to correcting component failures in semiconductor manufacturing tools. background

[0004] One of the limiting factors for the uninterrupted performance of manufacturing equipment is component failure. For example, failure of a flood gun and a source gun may interrupt the performance of an ion implantation tool. Component failure results in unplanned downtime, which is costly to the user.

[0005] Overview

[0006] The following is a simplified overview of the disclosure to provide a basic understanding of some aspects of the disclosure. This overview is not an extensive overview of the disclosure. The overview is neither intended to identify the key or important elements of the disclosure, nor to describe any scope of a particular embodiment of the disclosure or any scope of the claims. The sole purpose of the overview is to present some concepts of the disclosure in a simplified form as a preface to a more detailed description presented later.

[0007] In aspects of the present disclosure, a method may include receiving current sensor data corresponding to a plurality of features from a plurality of sensors associated with an ion implantation tool. The method may further include performing feature analysis to generate a plurality of additional features for the current sensor data. The method may further include providing a plurality of additional features as input to a trained machine learning model. The method may further include obtaining one or more outputs from the trained machine learning model. The one or more outputs may indicate a confidence level for a predicted window. The method may further include predicting whether one or more components of the ion implantation tool are within a pre-failure window based on the confidence level for the predicted window. The method may further include performing corrective actions associated with the ion implantation tool in response to predicting that one or more components are within the pre-failure window.

[0008] In another aspect of the present disclosure, a method may include receiving historical sensor data corresponding to a plurality of features from a plurality of sensors associated with an ion implantation tool. The method may further include determining a plurality of windows, the plurality of windows including a normal operation window for a first subset of the historical sensor data and a pre-failure window for a second subset of the historical sensor data. The method may further include performing feature analysis to generate a plurality of additional features for the historical sensor data. The method may further include training a machine learning model using training data including the plurality of additional features and a target output including the plurality of windows to generate a trained machine learning model. The trained machine learning model may be capable of generating one or more outputs indicating whether one or more ion implantation tool components are within a pre-failure window.

[0009] In another aspect of the present disclosure, a system may include a memory and a processing device coupled to the memory. The processing device may perform the following operations: receive current sensor data corresponding to a plurality of features from a plurality of sensors associated with an ion implantation tool. The processing device may further perform the following operations: perform feature analysis to generate a plurality of additional features for the current sensor data. The processing device may further perform the following operations: provide a plurality of additional features as input to a trained machine learning model. The processing device may further perform the following operations: obtain one or more outputs from the trained machine learning model. The one or more outputs may indicate a confidence level of a predicted window. The processing device may further perform the following operations: predict whether one or more components of the ion implantation tool are within a pre-failure window based on the confidence level of the predicted window. The processing device may further perform the following operations: in response to predicting that one or more components are within the pre-failure window, perform corrective actions associated with the ion implantation tool. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] The present disclosure is illustrated by way of example and not limitation in the figures of the accompanying drawings.

[0012] Figure 1 is a block diagram illustrating an exemplary system architecture according to certain implementations.

[0013] Figure 2 is an example dataset generator for creating a dataset for a machine learning model according to certain embodiments.

[0014] Figure 3 is a block diagram illustrating a system for predicting component failure according to certain embodiments.

[0015] Figures 4 to 6is a flow chart illustrating an example method for predicting component failure according to certain implementations.

[0016] FIG. 7A to FIG. 7B is a block diagram illustrating a system for predicting component failure according to certain embodiments.

[0017] FIG. 8A to FIG. 8B is a graph illustrating predicted component failure according to certain embodiments.

[0018] Fig. 9 is a block diagram illustrating a computer system according to certain embodiments.

[0019] Specific instructions

[0020] Described herein are techniques for correcting component failures in semiconductor manufacturing tools. Component failures can disrupt the performance of manufacturing equipment. For example, failures of flood guns and source guns disrupt the performance of ion implantation semiconductor manufacturing tools. Component failures can result in unplanned downtime, high user costs, equipment and product damage, and the like. Premature replacement of components can result in high maintenance costs, waste of components, and unnecessary time spent replacing components.

[0021] The devices, systems, and methods disclosed herein provide for correcting component failures in semiconductor manufacturing tools (e.g., life end predictions for plasma source guns in ion implantation semiconductor manufacturing tools). The devices, systems, and methods disclosed herein may provide critical component failure predictions for ion implantation tools using obtained sensor readings (e.g., by using a deep learning model). A processing device may receive current sensor data corresponding to features (e.g., pressure, temperature, flow, power, etc.) from a plurality of sensors associated with manufacturing equipment (e.g., an ion implantation tool). The processing device may further perform feature analysis to generate additional features for the current sensor data. Additional features may be generated based on one or more features. For example, the additional features may include one or more of the following items: ratio, range, increment, maximum value, etc. The processing device may further provide the additional features as input to a trained machine learning model, and then obtain one or more outputs from the trained machine learning model. One or more outputs may indicate a confidence level of the predicted window. The processing device may predict whether one or more components of the ion implantation tool are within the pre-failure window based on the confidence level of the predicted window. The pre-failure window may be a time window (e.g., 24 hours, 48 ​​hours) before a failure of a component is predicted to occur. The processing device may further perform a corrective action associated with the ion implantation tool in response to predicting that one or more components are within the pre-failure window. The corrective action (e.g., correcting and / or preemptively correcting a component failure) may include providing an alarm, interrupting operation of the manufacturing equipment, and / or causing one or more components to be replaced.

[0022] The devices, systems, and methods disclosed herein also provide for training of machine learning models for predicting component failures. In some embodiments, a processing device may receive historical sensor data corresponding to a feature from a sensor associated with a manufacturing equipment (e.g., an ion implantation tool) (e.g., historical values ​​of the sensor, pressure, flow, power, and other data). The processing device may further determine a window corresponding to the historical sensor data. The window may include a normal operation window for a first subset of the historical sensor data and a pre-failure window for a second subset of the historical sensor data. The processing device may further perform feature analysis to generate additional features for the historical sensor data (e.g., ratios, ranges, increments, maximum values, etc.). The processing device may further use the training data including the additional features and the target output including the window to train the machine learning model to generate a trained machine learning model. The trained machine learning model may be able to generate one or more outputs indicating whether one or more components (e.g., ion implantation tool components) are within the pre-failure window (e.g., to perform corrective actions associated with one or more components of the ion implantation tool). The machine learning model may be trained using historical sensor data associated with the first manufacturing equipment, and the machine learning model may be used to predict component failures of other manufacturing equipment.

[0023] Aspects of the present disclosure also provide technical advantages. Conventionally, components are used until failure occurs, or components are replaced prematurely. By predicting a pre-failure window (e.g., end of life, 24 hours, 48 ​​hours) for one or more components through a processing device, the processing device can generate corrective actions to replace one or more components before failure. Replacing components before failure (e.g., rather than after failure) reduces downtime, damage to manufacturing equipment and products, reduces unplanned maintenance, reduces emergency shipments of replacement components, etc. Replacing components within the pre-failure window (e.g., rather than arbitrarily replacing components prematurely) reduces waste of current components that can still be used, reduces costs associated with replacing components too frequently, reduces maintenance, etc. Receiving sensor data, reducing noise, and performing feature analysis (before using a trained machine learning model to predict whether one or more components are within the pre-failure window) provides a significant reduction in energy consumption (e.g., battery consumption), bandwidth, latency, etc., compared to analyzing all sensor data (e.g., including noise and all features).

[0024] Figure 1 1 is a block diagram illustrating an exemplary system architecture 100 according to some embodiments. The system architecture 100 includes a client device 120, a fault prediction server 130, and a data store 140. The fault prediction server 130 may be part of the fault prediction system 110.

[0025] The client device 120, the fault prediction server 130, the data storage 140, the server machine 170, the server machine 180, the manufacturing equipment 124 (e.g., an ion implantation tool, etc.), and the sensor 126 may be coupled to each other via a network 160 for fault prediction. In some embodiments, the network 160 is a public network that provides the client device 120 with access to the fault prediction server 130, the data storage 140, and other publicly available computing devices. In some embodiments, the network 160 is a private network that provides the client device 120 with access to the fault prediction server 130, the data storage 140, and other privately available computing devices. The network 160 may include one or more wide area networks (WANs), local area networks (LANs), wired networks (e.g., Ethernet), 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, and / or combinations thereof.

[0026] The manufacturing equipment 124 may be used for semiconductor processing. The manufacturing equipment 124 may include an ion implantation tool. The ion implantation tool may insert atoms into a semiconductor device to control the flow of electricity through the semiconductor device (e.g., to manufacture transistors, etc.). The manufacturing equipment 124 (e.g., an ion implantation tool) may include components such as a flooding gun 124A, a source gun 124B, etc. The flooding gun 124A may be an electromechanical device that provides a steady flow of low-energy electrons to a target (e.g., a flooded area, an area on an insulator or a semiconductor). The source gun 124B (e.g., a plasma source gun) may be a plasma source for depositing plasma on a semiconductor device (e.g., ejecting plasma at an appreciable flow rate to have high-energy deposition of plasma on a semiconductor device).

[0027] The limiting factor for the uninterrupted performance of the manufacturing equipment 124 (e.g., an ion implantation tool) may be the failure of one or more components (e.g., flood gun 124A, source gun 124B, etc.), which may cause unplanned downtime. Sensors 126 may capture sensor data (e.g., raw sensor data, temperature, pressure, power, flow, etc.) associated with the manufacturing equipment 124. For example, an ion implanter tool may be equipped with hundreds of sensors, with acquisition speeds of several kilohertz. Given the number of sensors, the acquisition speed of the sensor data, and the expected life of the components (e.g., six months, etc.), the amount of captured sensor data (e.g., raw sensor data) may be very large. Sensor data 142 from sensors 126 may be stored in data storage 140.

[0028] As described herein, semiconductor processing may include one or more of the following: semiconductor manufacturing for wafers or display manufacturing (e.g., flat panel display manufacturing). Failure prediction may be associated with one or more components of semiconductor manufacturing equipment (e.g., predicting failures of components used for semiconductor manufacturing for wafers), or associated with one or more components of display manufacturing (e.g., predicting failures of components used for display manufacturing).

[0029] The client device 120 may include a computing device such as a personal computer (PC), a laptop computer, a mobile phone, a smart phone, a tablet computer, a netbook computer, a network-connected television ("smart TV"), a network-connected media player (e.g., a Blu-ray player), a set-top box, an over-the-top (OTT) streaming device, an operator box, etc. The client device 120 may be capable of transmitting information (e.g., a selection of manufacturing equipment 124 for fault prediction) via the network 160 and receiving instructions associated with the predicted fault (e.g., a confidence level of the predicted window, instructions to perform corrective actions, etc.) via the network 160. The instructions associated with the predicted fault may specify that one or more components of the manufacturing equipment 124 are currently associated with a predicted window 156B of time (e.g., a normal operation window, a pre-failure window, a failure window, etc.). The instructions associated with the predicted fault may indicate one or more of the following items: an amount of time until the fault, a component to be replaced, how to replace a component, whether the operation of the manufacturing equipment 124 has been interrupted (e.g., shut down), or whether the operation of the manufacturing equipment 124 should be interrupted. Client devices 120 may display an alert via a graphical user interface (GUI) in response to receiving an indication associated with a predicted failure. Each client device 120 may include an operating system that allows a user to generate, view and edit information and view alerts.

[0030] The client device 120 may include a corrective action component 122. The corrective action component 122 may receive user input (e.g., via a GUI displayed via the client device 120), and may generate an indication to perform fault prediction on the manufacturing equipment 124 based on the user input. The corrective action component 122 may send the indication to the fault prediction server 130. In some embodiments, the corrective action component 122 transmits sensor data 142 (e.g., from a sensor 126 coupled to the manufacturing equipment 124) to the fault prediction server 130. The corrective action component 122 may receive an indication associated with the predicted fault from the fault prediction server 130 (e.g., in response to the fault prediction server 130 determining a pre-failure window). The corrective action component 122 may cause a corrective action to be performed. The corrective action may refer to correcting and / or preemptively correcting a component failure (e.g., based on a predicted pre-failure window). For example, in order to cause a corrective action to be performed, the corrective action component 122 may provide an alert (e.g., via a GUI of the client device 120, via the manufacturing equipment 124, etc.), may interrupt the operation of the manufacturing equipment 124 (e.g., shutting down one or more parts of the manufacturing equipment 124), and / or may cause replacement of one or more components.

[0031] The fault prediction server 130 may include one or more computing devices, such as a rack-mounted server, a router computer, a server computer, a personal computer, a mainframe computer, a laptop computer, a tablet computer, a desktop computer, etc. The fault prediction server 130 may include a fault prediction component 132. In some embodiments, the fault prediction component 132 may receive sensor data 142 (e.g., from a sensor 126 coupled to the manufacturing equipment 124). The sensor data 142 may include recorded values ​​over time and corresponding timestamps for each value (e.g., a first recorded temperature at a first time point, a second recorded temperature at a second time point, etc.). The sensor data 142 may be raw tracking data (e.g., without any feature engineering). The fault prediction component 132 may remove noise from the current sensor data 150, perform feature analysis to generate additional features for the current sensor data 150, predict whether one or more components of the manufacturing equipment 124 are within the pre-failure window, and perform corrective actions associated with the manufacturing equipment 124 in response to predicting that one or more components are within the pre-failure window.

[0032] In order to predict whether a component is within a pre-failure window, the fault prediction component 132 may provide the current sensor data 150 (e.g., the current additional features 154) to a model 190 (e.g., a convolutional long short-term memory (LSTM) (convLSTM) model, a deep learning model, a random forest model, etc.) for fault prediction. The fault prediction component 132 may receive a confidence level 158 for the predicted window 156B from the model 190 based on the current sensor data 150.

[0033] Each feature of the sensor data 150 (e.g., historical feature 146, current feature 152, etc.) may include a sequence (e.g., a first value, a second value, etc.), a timestamp (e.g., at the time of the first value, at the time of the second value, etc.), and an indication of which sensor 126 corresponds to the sequence. Each additional feature (e.g., historical additional feature 148, current additional feature 154) may be generated by performing one or more operations on one or more of the features. The one or more operations may include one or more of a ratio, a range, an increment, or a maximum value of a feature (e.g., corresponding sensor data) from one or more of the plurality of sensors 126. For example, a first feature may be a sequence of pressure measurements received from a pressure sensor of the sensor 126, a second feature may be a sequence of temperature measurements received from a temperature sensor of the sensor 126, and the first additional feature may be a ratio of the sequence of pressure measurements divided by each corresponding temperature measurement (e.g., the first additional feature may be a sequence including a first pressure value at a first point in time divided by a first temperature value at the first point in time, a second pressure value at a second point in time divided by a second temperature value at the second point in time, etc.).

[0034] Data storage 140 may be a memory (e.g., random access memory), a drive (e.g., a hard drive, a flash drive), a database system, or another type of component or device capable of storing data. Data storage 140 may include multiple storage components (e.g., multiple drives or multiple databases) that may span multiple computing devices (e.g., multiple server computers). Data storage 140 may store one or more of the following items: sensor data 142 (e.g., historical sensor data 144, historical features 146, historical additional features 148, current sensor data 150, current features 152, current additional features 154, etc.), windows 156 (e.g., historical windows 156A, predicted windows 156B), confidence levels 158, etc.

[0035] In some embodiments, the fault prediction system 110 further includes a server machine 170 and a server machine 180. The server machine 170 and the server machine 180 may be one or more computing devices (such as rack-mounted servers, router computers, server computers, personal computers, mainframe computers, laptop computers, tablet computers, desktop computers, etc.), data storage (e.g., hard disks, memory databases), networks, software components, or hardware components.

[0036] The server machine 170 includes a data set generator 172 that is capable of generating one or more data sets (e.g., Figure 2 The data input 210 and the target output 220 are used to train, validate or test the machine learning model 190. Figure 2 and Figure 6 Some operations of the data set generator 172 are described in detail. In some embodiments, the data set generator 172 can divide the historical sensor data 144 into a training set (e.g., sixty percent of the historical sensor data 144), a validation set (e.g., twenty percent of the historical sensor data 144), and a test set (e.g., twenty percent of the historical sensor data 144). The server machine 180 includes a training engine 182. In some embodiments, the server machine 180 includes the training engine 182, the validation engine 184, and the testing engine 186. The training engine 182 can be capable of using the training set from the data set generator 172 to train the machine learning model 190. The training engine 182 can generate one or more trained machine learning models 190.

[0037] The validation engine 184 may be able to validate the trained machine learning models 190 using the validation set from the data set generator 172. The validation engine 184 may determine the accuracy of each trained machine learning model 190 based on the validation set. The validation engine 184 may discard trained machine learning models 190 having an accuracy that does not meet a threshold accuracy.

[0038] The testing engine 186 may be able to test the trained machine learning model 190 using the test set from the data set generator 172. The testing engine 186 may determine the most accurate trained machine learning model 190 among all trained machine learning models based on the test set.

[0039] The machine learning model 190 may refer to a model artifact created by the training engine 182 using a training set that includes data inputs and corresponding target outputs (correct answers for the corresponding training inputs). Patterns in the data set may be found that map the data inputs to the target outputs (correct answers), and a mapping that captures these patterns may be provided to the machine learning model 190. In some embodiments, the machine learning model 190 may use one or more LSTM layers and a softmax layer (see FIG. 7A to FIG. 7B ).

[0040] In some embodiments, the fault prediction component 132 may provide the historical sensor data 144 and the historical window 156A to the data set generator 172. The data set generator 172 may provide the historical sensor data 144 as input and the historical window 156 as output to one or more of the training engine 182, the validation engine 184, and / or the testing engine 186 to perform one or more of the following: training, validating, or testing the machine learning model 190.

[0041] In some embodiments, the fault prediction system 110 may generate different models 190 based on one or more of the following: different hyperparameters (e.g., different numbers of LSTM layers), different types of machine learning models, different sets of historical additional features 148, etc. The fault prediction system 110 may perform one or more of the following: train, validate, or test the different models 190, and select the most accurate model 190.

[0042] In some embodiments, the fault prediction component 132 may provide the current sensor data 150 as an input to the trained machine learning model 190 and run the trained machine learning model 190 with the input to obtain one or more outputs. Figure 4 As described in detail, the fault prediction component 132 may be able to determine the predicted window 156B (e.g., based on the output of the trained machine learning model 190, by extracting the confidence level of the predicted window 156B from the output, etc.). The fault prediction component 132 may also determine confidence data based on the output. The confidence data may indicate a confidence level that the predicted window 156B corresponds to the manufacturing equipment 124. The fault prediction component 132 may use the confidence level 158 to select the predicted window 156B.

[0043] The confidence data may include or indicate a confidence level 158 of the predicted window 156B corresponding to the future failure of one or more components of the manufacturing equipment 124. In one example, the confidence level is a real number comprised between 0 and 1, where 0 indicates no confidence in the predicted window 156B corresponding to the future failure of one or more components of the manufacturing equipment 124, and 1 indicates absolute confidence in the predicted window 156B corresponding to the future failure of one or more components of the manufacturing equipment 124.

[0044] The fault prediction component 132 can determine a plurality of predicted windows 156B and corresponding confidence levels 158 (e.g., a 10% confidence level for a normal operation window and a 90% confidence level for a pre-fault operation window) based on the output of the model 190. In some embodiments, the fault prediction component 132 selects a prediction window with the highest confidence level. In some embodiments, the fault prediction component 132 selects a prediction window with a confidence level greater than 50%.

[0045] For purposes of illustration and not limitation, aspects of the present disclosure describe training a machine learning model and using the trained learning model to determine the predicted window 156B using the current sensor data 150. In other embodiments, a heuristic model or a rule-based model is used to determine the predicted window 156 based on the sensor data 142 (e.g., historical sensor data 144, current sensor data 150, etc.). The predicted window 156 may be monitored or otherwise used in the heuristic or rule-based model. Figure 2 The data input 210 describes any information.

[0046] In some embodiments, the functionality of client device 120, fault prediction server 130, server machine 170, and server machine 180 may be provided by a smaller number of machines. For example, in some embodiments, server machine 170 and server machine 180 may be integrated into a single machine. In some other embodiments, server machine 170, server machine 180, and fault prediction server 130 may be integrated into a single machine.

[0047] In general, functions described as being performed by the client device 120, the server machine 170, and the server machine 180 in one embodiment may also be performed on the fault prediction server 130 in other embodiments, if appropriate. In addition, functions attributed to a particular component may be performed by different or multiple components operating together. For example, in some embodiments, the fault prediction server 130 may receive user input indicating manufacturing equipment 124 (e.g., semiconductor processing tools) for fault prediction, and the fault prediction server 130 may provide an alarm, shut down the manufacturing equipment 124, etc. based on the confidence level 158 of the predicted window 156B. In another example, the client device 120 may perform one or more of the following: remove noise from the sensor data 142, perform feature analysis on the sensor data 142, determine the confidence level 158 of the predicted window 156B, predict whether one or more components are within the pre-failure window, or perform corrective actions. In another example, the data set generator 172 may remove noise from the historical sensor data 144 and perform feature analysis on the historical sensor data 144.

[0048] Additionally, the functionality of a particular component may be performed by different or multiple components operating together.One or more of the fault prediction server 130, server machine 170, or server machine 180 may be accessed as a service provided to other systems or devices through an appropriate application programming interface (API).

[0049] In an embodiment, a "user" may be represented as a single individual. However, other embodiments of the present disclosure encompass a "user" as an entity controlled by multiple users and / or automated sources. For example, a collection of individual users united as an administrator group may be considered a "user."

[0050] Although embodiments of the present disclosure are discussed with respect to sensor data 150 received from sensors 126 coupled to manufacturing equipment 124, embodiments may also be generally applicable to data received over time (e.g., irregular time series data, etc.). Embodiments may be generally applicable to optimizing processes that generate data over time. Examples of manufacturing equipment 124 for wafer or display manufacturing are physical vapor deposition (PVD) equipment, chemical vapor deposition (CVD) equipment, atomic layer deposition (ALD) equipment, chemical mechanical polishing (CMP) equipment, and etching equipment.

[0051] Figure 2 is an example data set generator 272 (e.g., Figure 1 ) for using historical sensor data 244 (e.g., Figure 1 historical sensor data 144), for machine learning models 290 (e.g., Figure 1 Model 190) creates a data set. Figure 2 The system 200 shows a data set generator 272 , a data input 210 , and a target output 220 .

[0052] In some embodiments, the data set generator 272 generates a data set (e.g., a training set, a validation set, a test set) including one or more data inputs 210 (e.g., training inputs, validation inputs, test inputs) and one or more target outputs 220. The data set may also include mapping data that maps the data inputs 210 to the target outputs 220. The data inputs 210 may also be referred to as "features," "attributes," or "information." In some embodiments, the data set generator 272 may provide the data set to one or more of the training engine 182, the validation engine 184, or the testing engine 186, where the data set is used to train, validate, or test the machine learning model 190. Some embodiments of generating a training set may be with respect to Figure 6 To further describe.

[0053] In some implementations, data input 210 may include one or more feature sets 212A for historical sensor data 244. Each feature set 212 may include at least one of the following: historical features 246 (e.g., Figure 1 of the historical features 146) or the historical additional features 248 (e.g., Figure 1 For example, feature set 212 may include one or more historical additional features 248.

[0054] In some embodiments, the data set generator 272 may generate a first data input 210A corresponding to a first feature set 212A to train, validate, or test a first machine learning model, and the data set generator 272 may generate a second data input 210B corresponding to a second feature set 212B to train, validate, or test a second machine learning model.

[0055] In some embodiments, the data set generator 272 may discretize the target output 220 (e.g., for a classification algorithm for a regression problem). The discretization of the target output 220 may convert the continuous value of the variable into a discrete value. In some embodiments, the discrete value for the target output 220 indicates a history window 256 (e.g., a normal operation window, a pre-failure window, a failure window, etc.). In some embodiments, the discrete value for the target output 220 indicates how much time from installation one or more components fail (e.g., days, hours, etc.).

[0056] The data input 210 and target output 220 used to train, validate, or test a machine learning model may include information from a specific facility (e.g., from a specific semiconductor manufacturing facility). For example, historical sensor data 244 may be associated with Figure 1 The current sensor data 150 is from the same manufacturing facility. In some embodiments, the information used to train the machine learning model may come from a specific group of components of the manufacturing facility with specific characteristics (e.g., components from a specific timeframe, components for a specific type of manufacturing equipment, etc.), and allow the trained machine learning model to predict a pre-failure window for the specific group of components based on historical sensor data associated with one or more components that share the characteristics of the specific group. In some embodiments, the information used to train the machine learning model may be used for components from two or more manufacturing facilities, and may allow the trained machine learning model to determine results for components based on inputs from one manufacturing facility. In some embodiments, the information used to train the machine learning model may be associated with one or more first ion implantation tools, and the trained machine learning model may be used to predict component failures of one or more second ion implantation tools that are different from the one or more first ion implantation tools.

[0057] In some embodiments, after generating data input 210 and using the data set to train, validate, or test the machine learning model 190, additional historical sensor data and corresponding historical windows from one or more manufacturing facilities may be further used to train, validate, or test or adjust the machine learning model 190 (e.g., adjust weights associated with the input data of the machine learning model 190, such as connection weights in a neural network, adjust hyperparameters, etc.).

[0058] Figure 3 is a diagram for determining the predicted window 356B (eg, Figure 1 The confidence level 358 (e.g., Figure 1 The system 300 may provide a block diagram of a system 300 for predicting failures of semiconductor manufacturing tools (eg, end-of-life prediction of a plasma source gun in an ion implantation semiconductor manufacturing tool).

[0059] At block 310, the system 300 (e.g., Figure 1 The fault prediction system 110 executes the historical sensor data 344 (e.g., Figure 1 The historical sensor data 144) is divided into data (e.g., via Figure 1The system 300 may be configured to generate a data set generator 172 of the server machine 170 to generate a training set 302, a validation set 304, and a test set 306. In some implementations, the system 300 may generate a plurality of feature sets corresponding to each data set.

[0060] At block 312, the system 300 uses the training set 302 to perform model training (e.g., via Figure 1 The system 300 may train multiple models using multiple feature sets of the training set 302 (eg, a first feature set of the training set 302, a second feature set of the training set 302, etc.).

[0061] At block 314, the system 300 uses the validation set 304 to perform model validation (e.g., via Figure 1 314). The system 300 may validate each trained model using the corresponding feature set of the validation set 304. At block 314, the system may determine the accuracy of each of the one or more trained models, and may determine whether one or more of the trained models have an accuracy that satisfies a threshold accuracy. In response to determining that any of the trained models do not have an accuracy that satisfies the threshold accuracy, the process returns to block 312, where the system 300 performs model training using a different feature set of the training set. In response to determining that one or more of the trained models have an accuracy that satisfies the threshold accuracy, the process continues to block 316.

[0062] At block 316, the system 300 performs model selection to determine which of the one or more trained models that meet the threshold accuracy has the highest accuracy (e.g., the selected model 308). In response to determining that two or more of the trained models that meet the threshold accuracy have the same accuracy, the process can return to block 312, where the system 300 performs model training using a further refined training set corresponding to a further refined feature set to determine the trained model with the highest accuracy.

[0063] At block 318, the system 300 performs model testing using the test set 306 (eg, via Figure 1304) to test the selected model 308. At block 318, the system 300 may use the test set 306 to determine whether the accuracy of the selected model 308 meets the threshold accuracy. In response to the accuracy of the selected model 308 not meeting the threshold accuracy (e.g., the selected model 308 overfits to the validation set 304), the process continues to block 312, where the system 300 performs model training using a different training set corresponding to a different feature set. 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 block 320. In at least block 312, the model may learn patterns in the historical sensor data to make predictions, and in block 318, the system 300 may apply the model to the remaining data (e.g., the test set 306) to test the predictions.

[0064] In some embodiments, in addition to using different feature sets (e.g., different combinations of historical additional features 148) for one or more of training, validation, or testing of different models, the system 300 may also include different hyperparameters in different models to determine which features and which hyperparameters provide the highest accuracy. In some embodiments, instead of using different feature sets for one or more of training, validation, or testing of different models, the system 300 uses different hyperparameters in different models (e.g., where each model uses the same set of historical additional features 148) to determine which hyperparameters provide the highest accuracy.

[0065] At block 320, the system 300 uses a trained model (e.g., the selected model 308) to receive current sensor data 350 (e.g., Figure 1 The current sensor data 150) and outputs a confidence level 358 (e.g., Figure 1 Confidence level 158 of the predicted window 156B).

[0066] In response to receiving additional sensor data, the additional sensor data may be input into block 312 to update the trained model via model retraining.

[0067] Figures 4 to 6is a flow chart illustrating example methods 400, 500, and 600 associated with fault prediction according to certain embodiments. The methods 400, 500, and 600 may be performed by processing logic, which may include hardware (e.g., circuits, dedicated logic, programmable logic, microcode, 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 one embodiment, the methods 400, 500, and 600 may be performed in part by the fault prediction system 110. In some embodiments, the methods 400, 500, and 600 may be performed by the fault prediction server 130. In some embodiments, a non-transitory computer-readable storage medium stores instructions that, when executed by a processing device (e.g., a processing device of the fault prediction system 110), cause the processing device to perform the methods 400, 500, and 600.

[0068] For simplicity of description, methods 400, 500, and 600 are depicted and described as a series of actions. However, actions according to the present disclosure may occur in various orders and / or simultaneously and with other actions not presented and described herein. Furthermore, according to the disclosed subject matter, not all illustrated actions may be performed to implement methods 400, 500, and 600. Additionally, those skilled in the art will understand and appreciate that methods 400, 500, and 600 may alternatively be represented as a series of interrelated states via state diagrams or events.

[0069] Figure 4 is a flow chart of a method 400 for predicting component failure according to some embodiments. In some embodiments, the method 400 is performed by processing logic of the fault prediction component 132 of the fault prediction server 130 .

[0070] At block 402, processing logic receives current sensor data (e.g., current sensor data 150) corresponding to a feature from a sensor (e.g., sensor 126) associated with manufacturing equipment (e.g., manufacturing equipment 124, ion implantation tool). A feature may be a sequence of current sensor data, where each sequence of current sensor data is captured by a corresponding sensor. In some embodiments, the current sensor data is streamed to processing logic. Processing logic may receive the sensor data in the form of one or more of a data set, a matrix, and the like. In some embodiments, the sensor data is saved and aggregated in data storage 140.

[0071] In some implementations, processing logic removes noise from the current sensor data at block 404. In some implementations, processing logic removes noise from the current sensor data by averaging the current sensor data over a time interval (e.g., averaging sensor data values ​​over a 10 second time period, etc.). In some implementations, processing logic removes noise by removing outliers from the current sensor data.

[0072] At block 406, processing logic performs feature analysis to generate additional features for the current sensor data (e.g., current additional features 154). The additional features may include one or more of a ratio, range, delta, or maximum value of corresponding sensor data from one or more of the plurality of sensors. In some implementations, the additional features may include statistical features of key sensors (e.g., mean, standard deviation, etc.).

[0073] In some embodiments, processing logic performs feature analysis by receiving user input indicating additional features to be calculated. Figure 5 , based on user input of additional features, based on user input of other parameters, without user input) to generate a model for feature analysis (e.g., feature engineering). The model for feature analysis can be a convolutional neural network (CNN) (e.g., a CNN that performs one-dimensional convolution). The CNN can be good at learning the temporal structure in the sensor data 142, and can determine invariant features for fault and normal data (e.g., for determining a normal operating window, a pre-fault window, etc.).

[0074] Processing logic may perform the analysis by receiving current sensor data in a matrix and processing the matrix via a one-dimensional convolution to output a plurality of additional features.

[0075] At box 408, processing logic provides additional features () as input to the trained machine learning model (e.g., after removing noise). The trained machine learning model may include one or more LSTM layers and a softmax layer. The trained machine learning model may have learned spatial features as sequences through one or more LSTM layers. The time series structure may be built into the prediction. The trained machine learning model may be weighted to penalize misclassifications (e.g., to avoid generating false positives). The current prediction generated by method 400 may be based on a previous time step of the current sensor data 150.

[0076] The trained machine learning model may be generated based on historical sensor data from a second plurality of sensors associated with fabrication equipment (eg, a second ion implantation tool) other than the fabrication equipment of block 402 (eg, an ion implantation tool).

[0077] At block 410, processing logic obtains one or more outputs from the trained machine learning model. In some embodiments, the one or more outputs indicate a confidence level (e.g., confidence level 158) for a predicted window (e.g., predicted window 156B). In some embodiments, processing logic extracts the confidence level for the predicted window from the one or more outputs. In some embodiments, processing logic determines a plurality of predicted windows and corresponding confidence levels (e.g., a confidence level of 10% for a normal operating window and a confidence level of 90% for a pre-fault window).

[0078] At block 412, processing logic predicts whether one or more components of manufacturing equipment (e.g., an ion implantation tool) are within a pre-failure window based on a confidence level of the predicted window. Processing logic may predict that the one or more components are within the pre-failure window by determining that the confidence level of the predicted window indicates greater than 50% confidence in the pre-failure window.

[0079] At block 414, processing logic determines whether the confidence level of the predicted window indicates that one or more components of the ion implantation tool are within the pre-failure window. In response to the confidence level of the predicted window indicating that the one or more components are not within the pre-failure window, flow continues to block 402 where additional sensor data is received (e.g., a loop of method 400). In response to the confidence level of the predicted window indicating that the one or more components are within the pre-failure window, flow continues to block 416. The one or more components may be components of the ion implantation tool, such as at least one of a flood gun or a source gun.

[0080] At block 416, processing logic performs corrective actions associated with the ion implantation tool (e.g., in response to predicting that one or more components are within a pre-failure window). The corrective actions may include one or more of causing a graphical user interface to display an alarm, interrupting operation of the manufacturing equipment (e.g., ion implantation tool) (e.g., shutting down, slowing down, stopping a particular process, etc.), or causing one or more components to be replaced.

[0081] Figure 5 is a flow chart of a method 500 for training a machine learning model for predicting component failures according to certain embodiments. In some embodiments, the method 500 comprises Figure 1 In some embodiments, the method 500 is performed by the processing logic of the fault prediction system 110. Figure 1In some embodiments, the method 500 is performed by the processing logic of the server machine 180. Figure 1 The training engine 182 of the server machine 180 is used for execution.

[0082] At block 502, processing logic receives historical sensor data (e.g., historical sensor data 144) corresponding to features (e.g., measurements received from sensor 126 associated with fabrication equipment 124 (e.g., ion implantation tool) from a sensor (e.g., sensor 126).

[0083] In some implementations, processing logic removes noise from the historical sensor data at block 504. Processing logic may remove noise from the historical sensor data by one or more of: averaging the historical sensor data within a time interval, or removing outliers.

[0084] At block 506, processing logic determines a window (e.g., historical window 156A) that includes a normal operating window for a first subset of historical sensor data and a pre-failure window for a second subset of historical sensor data. Processing logic may determine the window by determining a time of failure (e.g., based on a peak in the sensor data value, based on a peak in the health index value, such as at Figure 8B The processing logic may determine sensor data captured greater than a set amount of time (e.g., 24 hours, 48 ​​hours) before the time of the failure corresponds to the normal operation window, sensor data captured between the time of the failure and the set amount of time before the failure corresponds to the pre-failure window, and sensor data captured after the time of the failure corresponds to the failure window.

[0085] At block 508, processing logic performs feature analysis to generate additional features for the historical sensor data (e.g., historical additional features 148). The additional features may include one or more of a ratio, a range, a delta, or a maximum value of corresponding sensor data from one or more of the plurality of sensors. Processing logic may perform feature analysis by receiving the historical sensor data in a matrix and processing the matrix via a one-dimensional convolution to output a plurality of additional features.

[0086] In some implementations, processing logic receives user input corresponding to additional features (e.g., operations and specific sensors associated with the additional features). Processing logic can train a CNN (e.g., based on user input of additional features, using user input of parameters, without user input, etc.), and can use the trained CNN in method 400 to determine additional features (e.g., the current additional features 154) for use with the trained machine learning model.

[0087] At block 510, processing logic trains a machine learning model (e.g., the machine learning model includes one or more LSTM layers and a softmax layer) using the training data including the additional features (e.g., after noise removal) and the target output including the window to generate a trained machine learning model. The trained machine learning model may be capable of generating one or more outputs indicating whether one or more ion implant tool components (e.g., from one or more ion implant tools, from one or more fabrication equipment 124, etc.) are within a pre-failure window (). Figure 4 Method 400).

[0088] In some embodiments, the processing logic uses one or more of different features (e.g., historical features 146, historical additional features 148) or different hyperparameters to train multiple models. The processing logic can perform one or more of the following: training, validating, or testing different models (e.g., evaluating the models) to select the model that gives the highest accuracy.

[0089] In some embodiments, processing logic deploys a trained machine learning model to predict whether one or more components of manufacturing equipment (e.g., a flood gun, a source gun, etc. of an ion implantation tool) are within a pre-failure window () for performing a corrective action (e.g., the corrective action is associated with the one or more ion implantation tools). In some embodiments, the trained machine learning model will receive input based on current sensor data from a second plurality of sensors associated with a second ion implantation tool (e.g., different from the ion implantation tool used to train the machine learning model) for predicting whether the one or more components are within the pre-failure window.

[0090] Figure 6 is a flow chart of a method 600 for generating a data set for a machine learning model for predicting component failures according to certain embodiments. According to embodiments of the present disclosure, the fault prediction system 110 may use the method 600 to perform at least one of the following: training, validating, or testing the machine learning model. In some embodiments, one or more operations of the method 600 may be performed by, for example, Figure 1 and Figure 2 The data set generator 172 of the server machine 170 described above is executed. It can be noted that the data set generator 172 of the server machine 170 described above can be executed by the server machine 170 described above. Figure 1 and Figure 2 Describe the parts to illustrate Figure 6 aspects.

[0091] Reference Figure 6 At block 602, processing logic initializes a data set T to an empty set.

[0092] At block 604, processing logic generates a first data input (eg, a first training input, a first validation input) that includes a first feature set for historical sensor data (eg, relative to Figure 2 The first data input may include one or more features (eg, historical features 146 ) of the historical sensor data (eg, historical sensor data 144 ) and / or one or more additional features (eg, historical additional features 148 ).

[0093] At block 606, processing logic generates a first target output for one or more of the data inputs (eg, a first data input).The first target output provides an indication of a history window (eg, history window 156A).

[0094] At block 608, processing logic optionally generates mapping data indicating an input / output mapping. The input / output mapping (or mapping data) may refer to a data input (e.g., one or more of the data inputs described herein), a target output for the data input (e.g., where the target output identifies a predicted window), and an association between the data input and the target output.

[0095] At block 610 , processing logic adds the mapping data generated at block 610 to the data set T.

[0096] At block 612, processing logic branches based on whether the data set T is sufficient to at least one of: train, validate, or test the machine learning model 190. If so, execution proceeds to block 614, otherwise, execution continues back to block 604. It should be noted that in some embodiments, the sufficiency of the data set T may be determined based solely on the number of input / output mappings in the data set, while in some other embodiments, the sufficiency of the data set T may be determined based on one or more other criteria (e.g., a measure of diversity of data examples, accuracy, etc.) in addition to or in lieu of the number of input / output mappings.

[0097] At box 614, the processing logic provides a data set T to train, validate, or test the machine learning model 190. In some embodiments, the data set 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 data set 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 data set T is a test set and is provided to the test engine 186 of the server machine 180 to perform testing. In some embodiments, the data set T can be divided into a training set, a validation set, and a test set (for example, the training set can be 60%, the validation set can be 20%, and the test set can be 20%). In response to the machine learning model being trained (for example, and validated, tested, and meeting a threshold accuracy), the trained machine learning model (for example, by the fault prediction component 132) can be used for fault prediction (see Figure 3 to Figure 4 ).

[0098] In the case of a neural network, for example, the input values ​​of a given input / output mapping (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 mapping (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 the process is repeated for other input / output mappings in the data set T. The trained machine learning model can be implemented by the fault prediction component 132 (the fault prediction component of the fault prediction server 130) to predict the fault window for one or more components.

[0099] FIG. 7A to FIG. 7B is a block diagram illustrating a system 700A and a system 700B for fault prediction according to certain embodiments.

[0100] Reference Fig. 7A , the system 700A may receive input data 710. The input data 710 may be sensor data in a matrix. Noise may be removed from the sensor data (eg, by averaging raw data within a time interval to generate the sensor data, by removing outliers from the sensor data).

[0101] The system 700A may perform a one-dimensional convolution 720 (e.g., a one-dimensional convolution of a trained CNN) on the input data 710. In some implementations, the CNN is trained (e.g., the CNN performs the one-dimensional convolution) based on user input associated with the additional features (e.g., indicating an operation for generating the additional features). The system 700A may perform a one-dimensional convolution 720 (e.g., after noise removal) on the input data 710 to perform feature analysis to generate additional features for the input data. The additional features may include one or more of a ratio, a range, an increment, a maximum value, etc. of the corresponding sensor data.

[0102] The system 700A may input additional features into the LSTM layer 730 of the machine learning model. The number of LSTM layers may be a hyperparameter that is adjusted by training and retraining the machine learning model based on sensor data.

[0103] The system 700A may transmit the output of the LSTM layer 730 to the softmax layer 740 , and the softmax layer 740 may generate corresponding confidence levels for one or more predicted windows (eg, categories).

[0104] Reference Figure 7B , system 700B includes an LSTM layer 730 that can receive additional features based on input data 710. The output of the LSTM layer 730 can be transmitted to a softmax layer 740. The softmax layer can generate one or more outputs. The one or more outputs can include corresponding confidence levels for one or more predicted windows. For example, the softmax layer can generate a first confidence level for a normal operation window, a second confidence level for a pre-fault window, and a third confidence level for a fault window. The confidence levels can total 100%. Windows corresponding to confidence levels greater than 50% can be used.

[0105] FIG. 8A to FIG. 8B are graphs 800A and 800B illustrating failure prediction according to certain embodiments.

[0106] Reference Fig. 8A , chart 800A displays feature values ​​(e.g., historical additional features 148, current additional features 154, etc.) over time. The first time window may correspond to category 0 (e.g., normal operation window). The second time window may correspond to category 1 (e.g., pre-failure window). The third time window may correspond to category 2 (e.g., failure window). Category 0 may end and category 1 may begin at a set amount of time (e.g., 24 hours, 48 ​​hours, etc.) before a failure date (e.g., historical failure date, predicted failure date). Category 1 may end and category 2 may begin at the time when one or more components fail. The historical sensor data may be labeled according to the corresponding window (e.g., category 0, 1, or 2).

[0107] Reference Figure 8B , chart 800B shows a health index plotted over time (eg, the health index has a corresponding time stamp). The health index may be based on one or more of the results of a convolutional LSTM, sensor data, additional features, etc.

[0108] The health index may be substantially stable within a normal operation window. The health index may reach a peak during a pre-fault window, and the health index may drop substantially at the time of the fault. A first subset of sensor data may correspond to timestamps in a normal operation window, a second subset of sensor data may correspond to timestamps in a pre-fault window, and a third subset of sensor data may correspond to timestamps in a fault window. Each subset of sensor data may be labeled according to a corresponding window (e.g., a category).

[0109] Fig. 9 900 is a block diagram showing a computer system 900 according to certain embodiments. In some embodiments, the computer system 900 may be connected to other computer systems (e.g., via a network, such as a local area network (LAN), an intranet, an extranet, or the Internet). The computer system 900 may operate as a server or client computer in a client-server environment, or as a peer computer in a peer-to-peer or distributed network environment. The computer system 900 may be provided by a personal computer (PC), a tablet PC, a set-top box (STB), a personal digital assistant (PDA), a mobile phone, a network device, a server, a network router, a switch or a bridge, or a device capable of executing an instruction set (sequential or other) of an action to be performed by a specified device. In addition, the term "computer" shall include any computer set that executes an instruction set (or multiple instruction sets) individually or together to perform any one or more of the methods described herein.

[0110] In a further aspect, the computer system 900 may include a processing device 902, a volatile memory 904 (e.g., a random access memory (RAM)), a non-volatile memory 906 (e.g., a read-only memory (ROM) or an electrically erasable programmable ROM (EEPROM)), and a data storage device 916, which may communicate with each other via a bus 908.

[0111] The processing device 902 may be provided by one or more processors, 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 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).

[0112] The computer system 900 may further include a network interface device 922. The computer system 900 may also include a video display unit 910 (eg, LCD), an alphanumeric input device 912 (eg, keyboard), a cursor control device 914 (eg, mouse), and a signal generating device 920.

[0113] In some embodiments, the data storage device 916 may include a non-transitory computer-readable storage medium 924 on which instructions 926 encoding any one or more of the methods or functions described herein may be stored, including instructions for Figure 1 The fault prediction component 132 or the corrective action component 122 may be encoded with instructions and instructions for implementing the methods described herein.

[0114] The instructions 926 may also remain, in whole or in part, within the volatile memory 904 and / or within the processing device 902 during execution of the instructions 926 by the computer system 900, and thus the volatile memory 904 and the processing device 902 may also constitute machine-readable storage media.

[0115] Although computer-readable storage medium 924 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, the set of instructions causing 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.

[0116] The methods, components, and features described herein may be implemented by discrete hardware components, or may be integrated into the functionality of other hardware components (e.g., ASICS, FPGAs, DSPs, or similar devices). In addition, the methods, components, and features may be implemented by firmware modules or functional circuits within a hardware device. Furthermore, the methods, components, and features may be implemented in any combination of hardware devices and computer program components or in a computer program.

[0117] Unless expressly stated otherwise, terms such as "receive," "execute," "provide," "obtain," "extract," "predict," "remove," "make," "interrupt," "determine," "train," "deploy," or similar terms refer to actions and processes performed or implemented by a computer system that manipulates and transforms data represented as physical (electronic) quantities within computer system registers and memories into other data similarly represented as physical quantities within computer system memories or registers or other such information storage, transmission, or display devices. Additionally, the terms "first," "second," "third," "fourth," etc., as used herein, are meant to be labels used to distinguish between different elements and may not have an ordinal meaning according to the numerical representations of these terms.

[0118] The examples described herein also relate to an apparatus for performing the methods described herein. The apparatus may be specially constructed for performing the methods described herein, or the apparatus may comprise a general purpose computer system selectively programmed by a computer program stored in the computer system. Such a computer program may be stored in a computer readable tangible storage medium.

[0119] The methods and illustrative examples described herein are not inherently related to any particular computer or other device. Various general purpose systems may be used in accordance with the teachings described herein, or it may prove convenient to construct more specialized equipment to perform each of the methods described herein and / or the individual functions, routines, subroutines, or operations of the methods. Examples of the structures of various such systems are set forth in the description above.

[0120] The above description is intended to be illustrative, not restrictive. Although the present disclosure has been described with reference to specific illustrative examples and implementations, it will be appreciated that the present disclosure is not limited to the described examples and implementations. The scope of the present disclosure should be determined with reference to the appended claims together with the full scope of equivalents to which the claims are entitled.

Claims

1. A method comprising the following steps: determining, based on the sensor data, that one or more components of wafer or display manufacturing equipment are within a pre-failure window following a normal operating window, wherein corresponding data points in the normal operating window are substantially stable along a first health index value, wherein the corresponding data points in the pre-failure window increase from the first health index value to a peak at a second health index value; and Responsive to determining that the one or more components are within the pre-failure window, causing corrective actions associated with the one or more components of the wafer or display manufacturing equipment to be performed.

2. The method of claim 1, wherein the step of determining that the one or more components are within the pre-failure window comprises the steps of: providing the sensor data as input to a trained machine learning model; and One or more outputs are obtained from the trained machine learning model indicating that the one or more components are within the pre-failure window.

3. The method of claim 2, wherein the trained machine learning model is trained based on historical sensor data and based on a historical window associated with the historical sensor data, the historical window comprising the normal operation window and the pre-failure window.

4. The method of claim 1, wherein the sensor data is associated with the one or more components of the wafer or display manufacturing equipment.

5. The method of claim 1, wherein the corresponding data points of the pre-failure window decrease from the peak at the second health index value to a third health index value, the third health index value corresponding to a failure time at the end of the pre-failure window. The method of claim 5 , wherein the failure time corresponds to a failure of the one or more components.

7. The method of claim 1 , wherein causing the corrective action to be performed comprises one or more of: causing the graphical user interface to display an alert; interrupting the operation of the wafer or display manufacturing equipment; or causing the one or more components to be replaced.

8. A method comprising the following steps: identifying historical sensor data associated with wafer or display manufacturing equipment; and A historical window associated with the historical sensor data is identified, wherein: corresponding data points of the historical sensor data are substantially stable along a first health index value during a normal operating window of the historical window; The corresponding data point of the historical sensor data in a pre-fault window of the historical window increases from the first health index value to a peak value at a second health index value; and The historical sensor data and the historical window are used to determine whether one or more components are within the pre-failure window based on current sensor data so that corrective actions associated with the one or more components are performed.

9. The method of claim 8, further comprising the steps of: A machine learning model is trained using training data including the historical sensor data and a target output including the historical window to generate a trained machine learning model, wherein the trained machine learning model is capable of generating one or more outputs indicating whether the one or more components are within the pre-failure window so that the corrective action associated with the one or more components is performed.

10. The method of claim 8, wherein the current sensor data is associated with the one or more components of current wafer or display manufacturing equipment.

11. The method of claim 8, wherein the corresponding data points of the pre-failure window decrease from the peak at the second health index value to a third health index value corresponding to a failure time at an end of the pre-failure window.

12. The method of claim 11, wherein the failure time corresponds to a failure of the one or more components.

13. The method of claim 8, wherein causing the corrective action to be performed comprises one or more of: causing the graphical user interface to display an alert; interrupting the operation of the wafer or display manufacturing equipment; or causing the one or more components to be replaced.

14. A system, comprising: Memory; and A processing device is coupled to the memory, and the processing device performs the following operations: determining based on the sensor data that one or more components of wafer or display manufacturing equipment are within a pre-failure window following a normal operating window, wherein corresponding data points in the normal operating window are substantially stable along a first health index value, wherein the corresponding data points in the pre-failure window increase from the first health index value to a peak at a second health index value; and Responsive to determining that the one or more components are within the pre-failure window, causing corrective actions associated with the one or more components of the wafer or display manufacturing equipment to be performed.

15. The system of claim 14, wherein to determine that the one or more components are within the pre-failure window, the processing device performs the following operations: providing the sensor data as input to a trained machine learning model; and One or more outputs are obtained from the trained machine learning model indicating that the one or more components are within the pre-failure window.

16. The system of claim 15, wherein the trained machine learning model is trained based on historical sensor data and based on a historical window associated with the historical sensor data, the historical window comprising the normal operation window and the pre-fault window.

17. The system of claim 14, wherein the sensor data is associated with the one or more components of the wafer or display manufacturing equipment.

18. The system of claim 14, wherein the corresponding data points of the pre-failure window decrease from the peak at the second health index value to a third health index value, the third health index value corresponding to a failure time at an end of the pre-failure window.

19. The system of claim 18, wherein the failure time corresponds to a failure of the one or more components.

20. The system of claim 14, wherein to cause the corrective action to be performed, the processing device performs one or more of the following operations: causing the graphical user interface to display an alert; interrupting the operation of the wafer or display manufacturing equipment; or causing the one or more components to be replaced.

Images