System and method for process chamber health monitoring and diagnosis using a virtual model

Abstract

The method includes acquiring, by a processor, a plurality of sensor values ​​associated with a deposition process performed according to a recipe in a process chamber to deposit a film on a surface of a substrate. The method further includes applying a machine learning model to the plurality of sensor values, the machine learning model being trained based on historical sensor data of a subsystem of the process chamber and task data associated with the recipe for depositing the film. The method further includes generating an output of the machine learning model, the output being indicative of a health state of the subsystem.

Description

Technical Field

[0001] The present disclosure relates to electrical components, and more particularly to monitoring the health of a process chamber and providing a diagnosis using a virtual model.

Background Art

[0002] Products can be created by performing one or more manufacturing processes using manufacturing equipment. For example, semiconductor manufacturing equipment can be used to create semiconductor devices (e.g., substrates, wafers, etc.) by a semiconductor manufacturing process. The manufacturing equipment can deposit a plurality of layers of a film on the surface of a substrate and can perform an etching process to form a complex pattern in the deposited film. For example, the manufacturing equipment can perform a chemical vapor deposition (CVD) process to deposit alternating layers on a substrate. Sensors can be used to determine manufacturing parameters of the manufacturing equipment during the manufacturing process, and measurement equipment can be used to determine property data of the products created by the manufacturing equipment, such as the overall thickness of the layers on the substrate. Generally, the manufacturing equipment can monitor individual sensors to detect problems during the deposition process. However, monitoring individual sensors does not provide an indication of the overall health of different subsystems of the manufacturing equipment, and deteriorating conditions may go undetected and lead to significant downtime and repair times. Therefore, there is a desire for a system that has the ability to generate an indicator of the overall health of the system for each subsystem during the manufacturing process.

Summary of the Invention

[0003] The following is a simplified summary of the present disclosure to provide a basic understanding of some aspects of the present disclosure. This summary is not an extensive overview of the present disclosure. It is not intended to identify key or critical elements of the present disclosure nor to delineate any scope of particular implementations of the present disclosure or any scope of the claims. Its sole purpose is to present some concepts of the present disclosure in a simplified form as a prelude to the more detailed description that is presented later.

[0004] In an aspect of the present disclosure, a method includes obtaining, by a processor, sensor data associated with a deposition process implemented in a process chamber to deposit a film on a surface of a substrate. The sensor data includes sensor values associated with a subsystem of the process chamber. The method further includes obtaining task data associated with a recipe for depositing the film. The method further includes training a machine learning model using a training set based on the sensor data and the task data. The machine learning model is trained to generate prediction data indicative of predicted sensor values of the subsystem.

[0005] In another aspect of the present disclosure, a method includes obtaining, by a processor, a plurality of sensor values associated with a deposition process implemented in a process chamber to deposit a film on a surface of a substrate. The method further includes applying a machine learning model to the plurality of sensor values, the machine learning model being trained based on historical sensor data of a subsystem of the process chamber and task data associated with a recipe for depositing the film. The method further includes generating an output of the machine learning model, the output being indicative of a health state of the subsystem.

[0006] In another aspect of the present disclosure, the system includes a memory and a processing device, the processing device being operably coupled to the memory device to perform operations including obtaining a plurality of sensor values associated with a deposition process performed in a process chamber to deposit a film on a surface of a substrate by a processor. Performing the operations further includes applying a machine learning model to the plurality of sensor values, the machine learning model being trained based on historical sensor data of a subsystem of the process chamber and task data associated with a recipe for depositing the film. Performing the operations further includes generating an output of the machine learning model, the output indicating the health state of the subsystem.

[0007] The present disclosure is illustrated, by way of example and not limitation, in the figures of the accompanying drawings in which like reference numerals indicate similar elements.

Brief Description of the Drawings

[0008]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7

Figure 8

[0009] Described herein are systems and methods for monitoring the health of a process chamber and providing a diagnosis using a virtual model. A film can be deposited on the surface of a substrate during a deposition process (e.g., a chemical vapor deposition (CVD) process, an atomic layer deposition (ALD) process, etc.) implemented in a process chamber of a manufacturing system. For example, in a CVD process, the substrate is exposed to one or more precursors that react on the substrate surface to effect the desired deposition. The film can include one or more layers of material formed during the deposition process, and each layer can include a particular thickness gradient (e.g., a change in thickness along the layer of the deposited film). For example, a first layer can be formed directly on the surface of the substrate (referred to as the proximal layer or proximal end of the film) and can have a first thickness. After the first layer is formed on the surface of the substrate, a second layer having a second thickness can be formed on the first layer. This process continues until the deposition process is complete and a final layer is formed for the film (referred to as the distal layer or distal end of the film). The film can include alternating layers of different materials. For example, the film can include alternating layers of oxides and nitrides (oxide-nitride-oxide-nitride stack or ONON stack), or alternating oxides and polysilicon layers (oxide-polysilicon-oxide-polysilicon stack or OPOP stack), etc. The film can then be subjected to, for example, an etching process to form a pattern on the surface of the substrate, a chemical mechanical polishing (CMP) process to smooth the surface of the film, or any other process necessary to fabricate the finished substrate.

[0010] The process chamber can have a plurality of subsystems that operate during each substrate manufacturing process (e.g., deposition process, etching process, polishing process, etc.). The subsystems can be characterized as a set of sensors related to the operating parameters of the process chamber. The operating parameters can be temperature, flow rate, pressure, etc. In an example, the pressure subsystem can be characterized by one or more sensors that measure gas flow, chamber pressure, control valve angle, foreline (vacuum line between pumps) pressure, pump speed, etc. Thus, the process chamber can include a pressure subsystem, a flow subsystem, a temperature subsystem, etc. Each subsystem can experience degradation and deviate from its optimal performance state. For example, the pressure subsystem can produce a reduced pressure due to one or more of pump problems, control valve problems, etc.

[0011] Existing systems use limit checks of a single sensor value to detect faults within the process chamber. For example, an existing system can monitor whether a sensor value is below or above a predetermined threshold (e.g., a temperature sensor exceeding a predetermined temperature threshold). However, a sensor value below or above a predetermined threshold does not indicate the overall health of the process chamber subsystem as a function of each of the sensors and components of the subsystem. The health of the subsystem can be characterized as the current behavior (current sensor values) of the subsystem compared to the expected behavior (expected sensor values) of the subsystem. As such, existing systems cannot efficiently monitor the health of each subsystem of the process chamber. Furthermore, existing systems cannot efficiently provide a diagnosis of "unsound" subsystems (subsystems where the current behavior exceeds a threshold in relation to the expected behavior, subsystems where multiple sensors have failed, etc.).

[0012] Aspects and implementations of the present disclosure address these and other deficiencies of existing techniques by training a machine learning model that has the ability to monitor and display the health status of each subsystem of a process chamber and provide a diagnosis. In some embodiments, the system of the present disclosure acquires sensor data associated with a previous deposition process performed in a process chamber to deposit a film on a surface of a substrate. The sensor data may include sensor values associated with a subsystem of the process chamber. The system then acquires task data associated with a recipe for depositing the film and maps this task data to the sensor data to generate a training set. The system may train a machine learning model using the training set to generate prediction data indicative of predicted sensor values of the subsystem.

[0013] In some embodiments, the system applies the machine learning model to current sensor values to generate an output indicative of the health status of the subsystem. In some embodiments, the output is a scalar value indicative of the difference between the predicted behavior of the process chamber subsystem and the actual behavior of the process chamber subsystem. In some embodiments, the system uses a conversion function to convert the output to a representative value within a predefined range. The representative value provides a user-friendly representation of the subsystem health status. In some embodiments, the machine learning model generates a vectorized version of a scalar value indicative of a fault pattern associated with the process chamber subsystem. The system may then compare the fault pattern to a library of known fault patterns to determine the type of fault the subsystem is experiencing. In some embodiments, the system implements a corrective action to adjust one or more parameters of a deposition process recipe (e.g., a temperature setting for the process chamber, a pressure setting for the process chamber, a flow rate setting for a precursor of a material included in the film deposited on the substrate surface, etc.) based on the fault pattern.

[0014] Aspects of the present disclosure provide significant reduction in the time required to detect a faulty process chamber subsystem during substrate manufacturing, as well as technical advantages such as improved energy consumption. The present disclosure can also result in generating diagnostic data and implementing corrective actions to avoid inconsistent and abnormal products and unplanned user time.

[0015] FIG. 1 depicts an exemplary computer system architecture 100 according to aspects of the present disclosure. In some embodiments, the computer system architecture 100 can be included as part of a manufacturing system for processing substrates, such as the manufacturing system 300 of FIG. 3. The computer system architecture 100 includes a client device 120, a manufacturing apparatus 124, a measurement device 128, a prediction server 112 (e.g., for generating prediction data, providing model adaptation, using a knowledge base, etc.), and a data store 140. The prediction server 112 can be part of a prediction system 110. The prediction system 110 can further include server machines 170 and 180. The manufacturing apparatus 124 can include sensors 125 configured to capture data about substrates being processed in the manufacturing system. In some embodiments, the manufacturing apparatus 124 and sensors 126 can be part of a sensor system that includes a sensor server (e.g., a field service server (FSS) in a manufacturing facility) and a sensor identifier reader (e.g., a front opening unified pod (FOUP) radio frequency identification (RFID) reader for a sensor system). In some embodiments, the measurement device 128 can be part of a measurement system that includes a measurement server (e.g., a measurement database, a measurement folder, etc.) and a measurement identifier reader (e.g., a FOUP RFID reader for a measurement system).

[0016] The manufacturing device 124 can produce products such as electronic devices according to a recipe or by executing operations over a certain period. The manufacturing equipment 124 may include a process chamber such as the process chamber 400 described with respect to FIG. 4. The manufacturing equipment 124 can perform processes for a substrate (e.g., a wafer, etc.) in the process chamber. Examples of substrate processes include a deposition process for depositing one or more layers of a film on the surface of the substrate, an etching process for forming a pattern on the surface of the substrate, and the like. The manufacturing equipment 124 can perform each process according to a process recipe. The process recipe defines a specific set of operations to be performed on the substrate during the process and may include one or more settings associated with each operation. For example, a deposition process recipe may include a temperature setting for the process chamber, a pressure setting for the process chamber, a flow rate setting for the precursor of the material included in the film deposited on the substrate surface, and the like.

[0017] In some embodiments, the manufacturing equipment 124 includes a sensor 126 configured to generate data associated with the substrate processed in the system 100. For example, the process chamber may include one or more sensors configured to generate spectral or non-spectral data associated with the substrate before, during, and / or after a process (e.g., a deposition process) is performed on the substrate. In some embodiments, the spectral data generated by the sensor 126 may indicate the concentration of one or more materials deposited on the surface of the substrate. The sensor 126 configured to generate spectral data associated with the substrate may include a reflectance measurement sensor, an ellipsometry sensor, a thermal spectrum sensor, a capacitance sensor, and the like. The sensor 126 configured to generate non-spectral data associated with the substrate may include a temperature sensor, a pressure sensor, a flow rate sensor, a voltage sensor, and the like. Further details regarding the manufacturing equipment 124 are provided with respect to FIGS. 3 and 4.

[0018] In some embodiments, sensor 126 provides sensor data (e.g., sensor values, features, trace data) associated with manufacturing equipment 124 (e.g., associated with creating a corresponding product such as a wafer by manufacturing equipment 124). Manufacturing equipment 124 can create a product according to a recipe or by executing an operation over a certain period of time. Sensor data received over a period of time (e.g., corresponding to at least a part of a recipe or execution) may be referred to as trace data (e.g., historical trace data, current trace data, etc.) received from different sensors 126 over time. The sensor data may include one or more values such as temperature (e.g., heater temperature), setpoint (SP), pressure, high-frequency radio frequency (HFRF), electrostatic chuck voltage (ESC), current, material flow, power, voltage, etc. The sensor data may be associated with or indicative of manufacturing parameters such as hardware parameters (e.g., settings or components of manufacturing equipment 124 such as size, type, etc.), or process parameters of manufacturing equipment 124. The sensor data may be provided while manufacturing equipment 124 is performing a manufacturing process (e.g., equipment readout while processing a product). The sensor data may vary from substrate to substrate.

[0019] Measurement device 128 can provide measurement data associated with the substrate processed by manufacturing equipment 124. The measurement data may include values such as film property data (e.g., wafer space film properties), dimensions (e.g., thickness, height, etc.), relative permittivity, dopant concentration, density, defects, etc. In some embodiments, the measurement data may further include values of one or more surface profile property data (e.g., etching rate, etching rate uniformity, critical dimensions of one or more features included on the surface of the substrate, critical dimension uniformity across the surface of the substrate, edge placement error, etc.). The measurement data may be of a finished or semi-finished product. The measurement data may vary from substrate to substrate. The measurement data may be generated using, for example, reflectometry techniques, ellipsometry techniques, TEM techniques, etc.

[0020] In some embodiments, the measurement device 128 may be included as part of the manufacturing device 124. For example, the measurement device 128 may be included within or coupled to the inside of a process chamber and configured to generate measurement data about a substrate before, during, and / or after a process (e.g., a deposition process, an etching process, etc.) while the substrate remains within the process chamber. In such a case, the measurement device 128 may be referred to as an in-situ measurement device. In another example, the measurement device 128 may be coupled to another station of the manufacturing device 124. For example, the measurement device may be coupled to a transfer chamber such as transfer chamber 310 of FIG. 3, a load lock such as load lock 320, or a factory interface such as factory interface 306. In such a case, the measurement device 128 may be referred to as an integrated measurement device. In other or similar embodiments, the measurement device 128 is not coupled to a station of the manufacturing device 124. In such a case, the measurement device 128 may be referred to as an in-line measurement device or an external measurement device. In some embodiments, the integrated measurement device and / or the in-line measurement device are configured to generate measurement data about a substrate before and / or after a process.

[0021] The client device 120 can include computing devices such as a personal computer (PC), laptop, mobile phone, smartphone, tablet computer, netbook computer, network-connected television (“smart TV”), network-connected media player (e.g., Blu-ray player), set-top box, over-the-top (OTT) streaming device, operator box, etc. In some embodiments, the measurement data can be received from the client device 120. The client device 120 can display a graphical user interface (GUI), and the GUI enables the user to provide measurement readings about the substrates processed in the manufacturing system as inputs. The client device 120 can include a corrective action component 122. The corrective action component 122 can receive user inputs of metrics associated with the manufacturing equipment 124 (e.g., via a graphical user interface (GUI) displayed by the client device 120). In some embodiments, the corrective action component 122 sends the metric to the prediction system 110, receives an output (e.g., prediction data) from the prediction system 110, determines a corrective action based on the output, and causes the corrective action to be implemented. In some embodiments, the corrective action component 122 receives an indicator of a corrective action from the prediction system 110 and causes the corrective action to be implemented. Each client device 120 can include an operating system that enables the user to perform one or more of generating, displaying, or editing data (e.g., metrics associated with the manufacturing equipment 124, corrective actions associated with the manufacturing equipment 124, etc.).

[0022] The data store 140 can be a memory (e.g., random access memory), a driver (e.g., hard driver, flash driver), a database system, or another type of component or device capable of storing data. The data store 140 can include multiple storage components (e.g., multiple drivers or multiple databases) that can span across multiple computing devices (e.g., multiple server computers). The data store 140 can store data associated with processing substrates in the manufacturing equipment 124. For example, the data store 140 can store data (referred to as process data) collected by the sensors 126 in the manufacturing equipment 124 before, during, or after the substrate process. The process data can refer to historical process data (e.g., process data generated for previous substrates processed in the manufacturing system) and / or current process data (e.g., process data generated for the current substrate being processed in the manufacturing system). The data store can also store spectral or non-spectral data associated with a portion of the substrate being processed in the manufacturing equipment 124. The spectral data can include historical spectral data and / or current spectral data.

[0023] The data store 140 can also store context data associated with one or more substrates being processed in the manufacturing system. The context data can include recipe names, recipe step numbers, preventive maintenance indicators, operators, etc. The context data can refer to historical context data (e.g., context data associated with previous processes performed on previous substrates), and / or current process data (e.g., context data associated with the current process or future process to be performed on the previous substrate). The context data can further include and identify sensors associated with a particular subsystem of the process chamber.

[0024] Data store 140 may also store task data. The task data may include one or more sets of operations to be performed on a substrate during a deposition process, and may also include one or more settings associated with each operation. For example, the task data for a deposition process may include a temperature setting for a process chamber, a pressure setting for the process chamber, a flow rate setting for a precursor of a material of a film to be deposited on a substrate, and the like. In another example, the task data may include a control pressure at a predefined pressure point with respect to a flow rate value. The task data may refer to historical task data (e.g., task data associated with a previous process performed on a previous substrate), and / or current task data (e.g., task data associated with a current process or a future process to be performed on a substrate).

[0025] In some embodiments, data store 140 may be configured to store data that is not accessible to a user of the manufacturing system. For example, process data, spectral data, context data, etc. obtained for a substrate being processed in the manufacturing system may not be accessible to a user (e.g., an operator) of the manufacturing system. In some embodiments, all data stored in data store 140 may be inaccessible to a user of the manufacturing system. In other or similar embodiments, a portion of the data stored in data store 140 may be inaccessible to a user, while another portion of the data stored in data store 140 may be accessible to a user. In some embodiments, one or more portions of the data stored in data store 140 may be encrypted using an encryption mechanism unknown to the user (e.g., the data is encrypted using a private encryption key). In other or similar embodiments, data store 140 may include multiple data stores, where data that is inaccessible to a user is stored in one or more first data stores, and data that is accessible to a user is stored in one or more second data stores.

[0026] In some embodiments, data store 140 may be configured to store data associated with known failure patterns. A failure pattern may be a vector value associated with one or more problems or failures associated with the process chamber subsystem. In some embodiments, a failure pattern may be associated with a corrective action. For example, a failure pattern may include parameter adjustment steps to correct the problem or failure indicated by the failure pattern. Failure patterns are described in more detail in FIG. 7 below.

[0027] In some embodiments, prediction system 110 includes prediction server 112, server machine 170, and server machine 180. Prediction server 112, server machine 170, and server machine 180 each 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, a graphics processing unit (GPU), an application specific integrated circuit (ASIC) accelerator (e.g., a tensor processing unit (TPU)), etc.

[0028] Server machine 170 includes a training set generator 172 capable of generating a training data set (e.g., a set of data inputs and a set of target outputs) for training, validating, and / or testing machine learning model 190. Machine learning model 190 may be any algorithm model capable of learning from data. Some operations of data set generator 172 are described in detail below with respect to FIG. 2. In some embodiments, data set generator 172 may divide the training data into a training set, a validation set, and a test set. In some embodiments, prediction system 110 generates multiple sets of training data.

[0029] Server machine 180 may include a training engine 182, a verification engine 184, a selection engine 185, and / or a test engine 186. The engines may refer to hardware (e.g., circuits, dedicated logic, programmable logic, microcode, processing devices, etc.), software (routines running on a processing device, general-purpose computer system, or dedicated machine, etc.), firmware, microcode, or a combination thereof. The training engine 182 may be capable of training one or more machine learning models 190. The machine learning model 190 may refer to a model artifact created by the training engine 182 using training data (also referred to herein as a training set) that includes training inputs and corresponding target outputs (correct answers for each training input). The training engine 182 may discover patterns in the training data that map the training inputs to the target outputs (expected answers) and provide a machine learning model 190 that captures these patterns. The machine learning model 190 may use one or more of statistical modeling, support vector machine (SVM), radial basis function (RBF), clustering, supervised machine learning, semi-supervised machine learning, unsupervised machine learning, k-nearest neighbor algorithm (k-NN), linear regression, random forest, neural network (e.g., artificial neural network), etc.

[0030] The verification engine 184 may be capable of verifying the machine learning model 190 using a corresponding set of features of the verification set from the training set generator 172. The verification engine 184 may determine the accuracy of the machine learning model 190 based on the corresponding set of features of the verification set. The verification engine 184 may discard a trained machine learning model 190 that has an accuracy that does not meet a threshold accuracy. In some embodiments, the selection engine 185 may be capable of selecting a trained machine learning model 190 that has an accuracy that meets the threshold accuracy. In some embodiments, the selection engine 185 may be capable of selecting the trained machine learning model 190 that has the highest accuracy among the trained machine learning models 190.

[0031] The test engine 186 may be capable of testing the learned machine learning model 190 using a corresponding set of features of the test set from the data set generator 172. For example, a first learned machine learning model 190 trained using a first set of features of the training set may be tested using a first set of features of the test set. The test engine 186 may determine the learned machine learning model 190 having the highest accuracy of the learned machine learning models based on the test set.

[0032] As described in more detail below, the prediction server 112 includes a prediction component 114 that is capable of executing the learned machine learning model 190 against current sensor data inputs to obtain one or more outputs, providing data indicative of the expected behavior of each subsystem of the process chamber. The prediction server 112 may further provide data indicative of the health and diagnosis of the process chamber subsystems. This is described in further detail below.

[0033] Client device 120, manufacturing equipment 124, sensor 126, measuring device 128, prediction server 112, data store 140, server machine 170, and server machine 180 may be coupled to each other via network 130. In some embodiments, network 130 is a public network that provides client device 120 with access to prediction server 112, data store 140, and other publicly available computing devices. In some embodiments, network 130 is a private network that provides client device 120 with access to manufacturing equipment 124, measuring device 128, data store 140, and other privately available computing devices. Network 130 may include 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.

[0034] Note that in some other implementations, the functions of server machines 170 and 180, as well as prediction server 112, may be 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 or similar embodiments, server machines 170 and 180, as well as prediction server 112, may be integrated into a single machine.

[0035] Generally, the functions described in one embodiment as being performed by server machine 170, server machine 180, and / or prediction server 112 may also be performed on client device 120. Additionally, the functionality belonging to a particular component may be implemented by different or multiple components operating together.

[0036] In an embodiment, a "user" may be represented as a single individual. However, other embodiments of the present disclosure include the case where a "user" is an entity controlled by multiple users and / or automated sources. For example, a collection of individual users united as a group of administrators may be regarded as a "user".

[0037] FIG. 2 is a flowchart 200 of a method for training a machine learning model according to an aspect of the present disclosure. The method 200 may be implemented by processing logic including hardware (circuits, dedicated logic, etc.), software (such as a run on a general-purpose computer system or a dedicated machine), firmware, or any combination thereof. In one implementation, the method 200 may be executed by a computer system such as the computer system architecture of FIG. 1. In other or similar embodiments, one or more processes of the method 200 may be executed by one or more other machines not depicted in the figure. In some aspects, one or more processes of the method 200 may be executed by the server machine 170, the server machine 180, and / or the prediction server 112.

[0038] For simplicity of explanation, the method is depicted and described as a series of acts. However, acts in accordance with the present disclosure may occur in various orders and / or simultaneously, as well as in conjunction with other acts not presented and described herein. Furthermore, not all of the illustrated acts may be executed to implement the method according to the disclosed subject matter. Additionally, those skilled in the art will understand that the method may alternatively be represented as a series of interrelated states via a state diagram or events. Additionally, it should be understood that the methods disclosed herein can be stored in a product to facilitate transporting and transferring such methods to a computing device. As used herein, the term product is intended to encompass a computer program accessible from any computer-readable device or storage medium.

[0039] In block 210, the processing logic initializes the training set T to an empty set (e.g., {}).

[0040] In block 212, the processing logic obtains sensor data (e.g., sensor values, features, trace data) associated with a previous deposition process performed to deposit one or more layers of a film on the surface of a previous substrate. The sensor data may be further associated with a subsystem of the process chamber. The subsystem may be characterized as a set of sensors associated with the operating parameters of the process chamber. The operating parameters may be temperature, flow rate, pressure, etc. For example, the pressure subsystem may be characterized by one or more sensors that measure gas flow rate, chamber pressure, control valve angle, foreline (vacuum line between pumps) pressure, pump speed, etc. Each process chamber may include a plurality of different subsystems such as a pressure subsystem, a flow rate subsystem, a temperature subsystem, etc.

[0041] In some embodiments, the sensor data associated with the deposition process is historical data associated with one or more previous deposition settings for a previous deposition process performed on a previous substrate in the manufacturing system. For example, the historical data can be historical context data associated with the previous deposition process stored in the data store 140. In some embodiments, the one or more previous deposition settings can include at least one of a previous temperature setting for the previous deposition process, a previous pressure setting for the previous deposition setting, a previous flow rate setting for a precursor of one or more materials of the previous film deposited on the surface of the previous substrate, or any other setting associated with the deposition process. The flow rate setting can refer to a flow rate setting for the precursor (referred to as the initial flow rate setting) in an initial instance of the previous deposition process, a flow rate setting for the precursor (referred to as the final flow rate setting) in a final instance of the previous deposition process, or a ramping rate of the flow rate of the precursor during the deposition process. In one example, the precursor of the previous film can include a boron-containing precursor or a silicon-containing precursor. In some embodiments, the sensor data can also be associated with a previous etching process performed on the previous substrate, or any other process performed in the process chamber.

[0042] In block 214, the processing logic obtains task data associated with a recipe for a film deposited on the surface of the previous substrate. For example, the task data can be a required temperature setting, a pressure setting, a flow rate setting for a precursor of the material of the film to be deposited on the substrate, and the like. The task data can include historical task data for the previous film deposited on the surface of the previous substrate. In some embodiments, the historical task data for the previous film can correspond to historical task values associated with a recipe for the previous film. The processing logic can obtain the task data from the data store 140 according to the embodiments described above.

[0043] In block 216, the processing logic generates first training data based on acquired sensor data associated with a previous deposition process executed on a previous substrate. In block 218, the processing logic generates second training data based on task data associated with a recipe for a film deposited on the surface of the previous substrate.

[0044] In block 220, the processing logic generates a mapping between the first training data and the second training data. The mapping refers to the first training data that includes, or is based on, data for a previous deposition process performed on a previous substrate, and the second training data that includes, or is based on, task data associated with a recipe for a film deposited on the surface of the previous substrate, and the first training data is associated with (or mapped to) the second training data. In block 224, the processing logic adds the mapping to the training set T.

[0045] In block 226, the processing logic determines whether the training set T contains a sufficient amount of training data to train a machine learning model. In some embodiments, the sufficiency of the training set T can be determined simply based on the number of mappings in the training set, while in some other embodiments, it should be noted that the sufficiency of the training set T can be determined based on one or more other criteria (such as a measure of the diversity of training examples) in addition to, or instead of, the number of input / output mappings. In response to determining that the training set does not contain a sufficient amount of training data to train a machine learning model, method 200 returns to block 212. In response to determining that the training set T contains a sufficient amount of training data to train a machine learning model, method 200 proceeds to block 228.

[0046] In block 228, the processing logic provides a training set T to train a machine learning model. In one embodiment, the training set T is provided to the training engine 182 of the server machine 180 to perform the training. In the case of a neural network, for example, the input values of a given input / output mapping are inputs to the neural network, and the output values of the input / output mapping are stored in the output nodes of the neural network. The connection weights within the neural network are then adjusted according to a learning algorithm (e.g., backpropagation, etc.), and this procedure is repeated for other input / output mappings within the training set T.

[0047] In some embodiments, the processing logic may perform an outlier detection method to remove outliers from the training set T before training the machine learning model. The outlier detection method may include techniques for identifying values that are significantly different from the majority of the training data. These values may be generated from errors, noise, etc.

[0048] As an illustrative example, the machine learning model 190 may use the k-NN algorithm to generate prediction data (e.g., expected behavior under ideal or nearly ideal operating parameters) for the process chamber subsystem using the training set T. In particular, with the k-NN algorithm, the processing logic may determine a decision boundary in the multi-dimensional space defined by the training set. An example of the k-NN algorithm is described in U.S. Patent No. 9,910,430, which is incorporated by reference in its entirety. The decision boundary may then be used to generate prediction data by comparing any current (actual or measured) behavior sensor data of the process chamber subsystem to the expected behavior values.

[0049] In block 230, the processing logic performs a calibration process on the trained machine learning model. In some embodiments, the processing logic may compare the predicted behavior of the process chamber subsystem with the current behavior of the process chamber subsystem based on the difference in the values of the predicted behavior and the current behavior. For example, the processing logic may compare one or more values associated with the predicted data of the pressure subsystem, the flow rate subsystem, or the temperature subsystem with one or more values associated with the currently measured behavior of the pressure subsystem, the flow rate subsystem, or the temperature subsystem, respectively. After block 230, the machine learning model may be used to generate one or more values associated with the predicted behavior of the process chamber subsystem. The one or more values associated with the predicted behavior may be compared with the current (actual) behavior of the process chamber subsystem to generate predicted data. The predicted data may include data indicating the health state of the process chamber subsystem. This is described in more detail in FIG. 5 below.

[0050] In some embodiments, the manufacturing system may include two or more process chambers. For example, the manufacturing system 300 of the example of FIG. 3 illustrates a plurality of process chambers 314, 316, 318. It should be noted that in some embodiments, the data acquired to train the machine learning model and the data collected to be provided as input to the machine learning model may be associated with the same process chamber of the manufacturing system. In other or similar embodiments, the data acquired to train the machine learning model and the data collected to be provided as input to the machine learning model may be associated with different process chambers of the manufacturing system. In other or similar embodiments, the data acquired to train the machine learning model may be associated with the process chamber of a first manufacturing system, and the data collected to be provided as input to the machine learning model may be associated with the process chamber of a second manufacturing system.

[0051] FIG. 3 is a top schematic view of an exemplary manufacturing system 300 according to an aspect of the present disclosure. The manufacturing system 300 may perform one or more processes on a substrate 302. The substrate 302 can be any suitably rigid, fixed-dimension planar article suitable for fabricating an electronic device or circuit components thereon, such as, for example, a silicon-containing disk or wafer, a patterned wafer, a glass plate, or the like.

[0052] The manufacturing system 300 may include a process tool 304 and a factory interface 306 coupled to the process tool 304. The process tool 304 may include a housing 308 having a transfer chamber 310 therein. The transfer chamber 310 may include one or more process chambers (also referred to as processing chambers) 314, 316, 318 disposed around and coupled thereto. The process chambers 314, 316, 318 may be coupled to the transfer chamber 310 through respective ports, such as slit valves or the like. The transfer chamber 310 may also include a transfer chamber robot 312 configured to transfer the substrate 302 between the process chambers 314, 316, 318, load lock 320, and the like. The transfer chamber robot 312 may include one or more arms, and each arm may include one or more end effectors at the end of each arm. The end effector may be configured to handle a particular object, such as a wafer.

[0053] Process chambers 314, 316, 318 may be adapted to perform any number of processes on substrate 302. The same or different substrate processes may occur in each processing chamber 314, 316, 318. Substrate processes may include atomic layer deposition (ALD), physical vapor deposition (PVD), chemical vapor deposition (CVD), etching, annealing, curing, pre-cleaning, metal or metal oxide removal, or the like. Other processes may be performed on the substrate there. Each of process chambers 314, 316, 318 may include one or more sensors configured to capture data about substrate 302 before, after, or during a substrate process. For example, one or more sensors may be configured to capture spectral and / or non-spectral data about a portion of substrate 302 during a substrate process. In other or similar embodiments, one or more sensors may be configured to capture data associated with the environment within process chambers 314, 316, 318 before, after, or during a substrate process. For example, one or more sensors may be configured to capture data associated with the temperature, pressure, gas concentration, etc. of the environment within process chambers 314, 316, 318 during a substrate process.

[0054] The load lock 320 can also be coupled to the housing 308 and the transfer chamber 310. The load lock 320 can be configured to interface with and be coupled to the transfer chamber 310 and the factory interface 306 on one side. The load lock 320 can have an environmentally controlled atmosphere that can be varied in some embodiments from a vacuum environment (where substrates can be transferred to or from the transfer chamber 310) to an inert gas environment near atmospheric pressure (where substrates can be transferred to or from the factory interface 306). The factory interface 306 can be any suitable storage device, such as, for example, an equipment front end module (EFEM). The factory interface 306 can be configured to receive the substrate 302 from a substrate carrier 322 (such as, for example, a front opening unified pod (FOUP)) that docks at various load ports 324 of the factory interface 306. A factory interface robot 326 (shown in dashed lines) can be configured to transfer the substrate 302 between the carrier (also referred to as a container) 322 and the load lock 320. The carrier 322 can be a substrate storage carrier or a replacement part storage carrier.

[0055] The manufacturing system 300 can also be connected to a client device (not shown) configured to provide information regarding the manufacturing system 300 to a user (such as, for example, an operator). In some embodiments, the client device can provide information to a user of the manufacturing system 300 via one or more graphical user interfaces (GUIs). For example, the client device can provide information regarding a target thickness profile for a film to be deposited on the surface of the substrate 302 during a deposition process implemented in the process chambers 314, 316, 318 via the GUI. The client device can also provide information regarding modifications to the process recipe considering each set of deposition settings that are predicted to correspond to the target profile, in accordance with the embodiments described herein.

[0056] The manufacturing system 300 may also include a system controller 328. The system controller 328 can be, for example, a computing device such as a personal computer, a server computer, a programmable logic controller (PLC), a microcontroller, etc., and / or can include the same. The system controller 328 can include one or more processing devices that can be general-purpose processing devices such as a microprocessor, a central processing unit, or the like. More specifically, the processing device can be a complex instruction set computing (CISC) microprocessor, a reduced instruction set computing (RISC) microprocessor, a very long instruction word (VLIW) microprocessor, or a processor implementing another instruction set, or a processor implementing a combination of instruction sets. The processing device can also be one or more special-purpose processing devices such as an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), a digital signal processor (DSP), a network processor, or the like. The system controller 328 can include a data storage device (e.g., one or more disk drivers and / or solid-state drivers), main memory, static memory, a network interface, and / or other components. The system controller 328 can execute instructions to implement any one or more of the methodologies and / or embodiments described herein. In some embodiments, the system controller 328 can execute instructions to perform one or more operations in the manufacturing system 300 according to a process recipe. The instructions can be stored in a computer-readable storage medium that can include main memory, static memory, secondary storage, and / or a processing device (during execution of the instructions).

[0057] The system controller 328 can receive data from sensors located on or within various parts of the manufacturing system 300, such as the processing chambers 314, 316, 318, the transfer chamber 310, the load lock 320, etc. In some embodiments, the data received by the system controller 328 can include spectral data and / or non-spectral data for a portion of the substrate 302. In other or similar embodiments, the data received by the system controller 328 can include data associated with processing the substrate 302 in the processing chambers 314, 316, 318, as described above. For the purposes of this description, the system controller 328 is described as receiving data from sensors included within the process chambers 314, 316, 318. However, the system controller 328 can receive data from any part of the manufacturing system 300 and can use the data received from that part in accordance with the embodiments described herein. In an illustrative example, the system controller 328 can receive data from one or more sensors for the process chambers 314, 316, 318 before, after, or during the substrate process in the process chambers 314, 316, 318. The data received from the sensors of various parts of the manufacturing system 300 can be stored in the data store 350. The data store 350 can be included as a component within the system controller 328 or can be a component separate from the system controller 328. In some embodiments, the data store 350 can be the data store 140 described with respect to FIG. 1.

[0058] FIG. 4 is a schematic side cross-sectional view of a process chamber 400 according to an aspect of the present disclosure. In some embodiments, the process chamber 400 may correspond to the process chambers 314, 316, 318 described with respect to FIG. 3. The process chamber 400 may be used for processes in which a corrosive plasma environment is provided. For example, the process chamber 400 may be a chamber for a plasma etcher or a plasma etching reactor, etc. In another example, the process chamber may be a chamber for a deposition process, as described previously. In one embodiment, the process chamber 400 includes a chamber body 402 and a showerhead 430 surrounding an internal volume 406. The showerhead 430 may include a showerhead base and a showerhead gas distribution plate. Alternatively, the showerhead 430 may be replaced by a lid and nozzles in some embodiments, or by a plurality of sector showerhead sections and plasma generation devices in other embodiments. The chamber body 402 may be fabricated from other suitable materials such as aluminum, stainless steel, or titanium (Ti). The chamber body 402 generally includes sidewalls 408 and a bottom 410. An exhaust port 426 may be defined within the chamber body 402 and may couple the internal volume 406 to a pump system 428. The pump system 428 may include one or more pumps and throttle valves utilized to evacuate and regulate the pressure of the internal volume 406 of the process chamber 400.

[0059] The showerhead 430 can be supported on the sidewall 408 of the chamber body 402. The showerhead 420 (or lid) can be opened to allow access to the internal volume 406 of the process chamber 400 and can provide sealing of the process chamber 400 while closed. The gas panel 458 can be coupled to the process chamber 400 to provide process and / or purge gases to the internal volume 406 through the showerhead 430 or lid and nozzles (e.g., through the apertures of the showerhead or lid and nozzles). For example, the gas panel 458 can provide a precursor of the material of the film 451 deposited on the surface of the substrate 302. In some embodiments, the precursor can include a silicon-based precursor or a boron-based precursor. The showerhead 430 can include a gas distribution plate (GDP) and can have a plurality of gas delivery holes 432 (also referred to as channels) throughout the GDP. A substrate support assembly 448 is disposed in the internal volume 406 of the process chamber 400 below the showerhead 430. The substrate support assembly 448 holds the substrate 302 during processing (e.g., during a deposition process).

[0060] In some embodiments, the processing chamber 400 may include measurement equipment (not shown) configured to generate in-situ measurement readings during the processes implemented in the process chamber 400. The measurement equipment may be operatively coupled to a system controller (e.g., system controller 328 as described above). In some embodiments, the measurement equipment may be configured to generate measurement readings (e.g., thickness) for the film 451 during a particular instance of the deposition process. The system controller may generate a concentration profile for the film 451 based on the measurement readings received from the measurement equipment. In other or similar embodiments, the processing chamber 400 does not include measurement equipment. In such embodiments, the system controller may receive one or more measurement readings for the film 451 after completion of the deposition process in the process chamber 400. The system controller may determine a deposition rate based on the one or more measurement readings and may associate and generate a thickness profile for the film 451 based on the determined concentration gradient and the determined deposition rate of the deposition process.

[0061] FIG. 5 is a flow diagram of a method 500 for determining a process chamber subsystem health metric using a machine learning model, according to aspects of the present disclosure. The method 500 may be implemented by processing logic that includes hardware (circuits, dedicated logic, etc.), software (such as a run on a general-purpose computer system or a dedicated machine), firmware, or some combination thereof. In one embodiment, the method 500 may be executed by a computer system such as the computer system architecture 100 of FIG. 1. In other or similar embodiments, one or more operations of the method 500 may be executed by one or more other machines not depicted in the figure. In some aspects, one or more operations of the method 500 may be executed by the server machine 170, the server machine 180, and / or the prediction server 112.

[0062] In block 510, the processing logic obtains sensor data associated with operations executed in the process chamber. In some embodiments, the operations may include a deposition process executed in the process chamber to deposit one or more layers of a film on the surface of a substrate, an etching process executed on one or more layers of a film on the surface of the substrate, or any other process executed in the process chamber. The operations may be performed according to a recipe. The sensor data may include one or more values of one or more of temperature (e.g., heater temperature), gap, pressure, high-frequency radio frequency, voltage of an electrostatic chuck, current, material flow, power, voltage, etc. The sensor data may be associated with or indicative of manufacturing parameters such as settings or components (e.g., size, type, etc.) of the manufacturing equipment 124, hardware parameters, or process parameters of the manufacturing equipment 124.

[0063] In block 512, the processing logic applies a machine learning model (e.g., model 190) to the acquired sensor data. The machine learning model may be used to generate one or more values associated with the predicted behavior of the process chamber subsystem. For example, the machine learning model may use the k-NN algorithm to generate the predicted behavior of the process chamber subsystem using the training set T. In some embodiments, the machine learning model is trained using historical sensor data of the subsystem of the process chamber and task data associated with the recipe used to perform the operations.

[0064] In block 514, the processing logic generates an output via the machine learning model based on the sensor data. In some embodiments, the output may be at least one scalar value indicating the difference between the predicted behavior of the process chamber subsystem and the actual behavior of the process chamber subsystem. In particular, the scalar value may indicate the difference between the actual values of a set of sensors associated with the subsystem and the predicted values of the set of sensors.

[0065] In block 516, the processing logic converts the output to a representative value within a default range. The representative value can be used to indicate the health state of the process chamber subsystem (e.g., the current behavior compared to the expected behavior). In some embodiments, the processing logic can generate the representative value by applying a linear or non-linear conversion function to the output value (e.g., a scalar value) to scale the output value within the default range. In some embodiments, the conversion function can include a linear function, a log-odds function, a sigmoid function, an exponential function, etc. A particular conversion function can be used based on the desired sensitivity to the current behavior of the process chamber subsystem. For example, a linear function can be sensitive to each deviation of the current behavior of the process chamber subsystem from the expected behavior, while a sigmoid function is not sensitive to initial changes in the current behavior of the process chamber subsystem. In some embodiments, the user can change the sensitivity (e.g., apply a different conversion function) using the client device 120.

[0066] FIG. 6 is a graph showing an example sigmoid conversion according to an aspect of the present disclosure. As illustrated, the x-axis can represent a scalar value (e.g., an output) from a machine learning model. The y-axis can represent a representative value within a default range, e.g., within 0-1, where the representative value “0” indicates that the actual behavior of the process chamber subsystem is the same or similar to the expected behavior of the process chamber subsystem, and the representative value “1” indicates that the current behavior of the process chamber subsystem is significantly deviated (e.g., greater than a predefined threshold) from the expected behavior of the process chamber subsystem. As shown in FIG. 6, increasing x values initially result in relatively small changes to the y value (before x = -4), then accelerate as the x value increases beyond x = -4, and then decelerate as the x value increases beyond x = 4. As such, using a sigmoid conversion, the prediction component 114 is not sensitive to initial deviations of the current behavior from the expected behavior, or deviations when the actual behavior is already significantly deviated from the expected behavior.

[0067] Returning to FIG. 5, in block 518, the processing logic displays the representative value on a client device (e.g., client device 120). In some embodiments, different representative values may be associated with different health status indicators (e.g., healthy, deteriorating, critical, failed, etc.), and the health status indicators may be displayed on the client device. The representative value may be associated with different health status indicators based on whether it exceeds a threshold or is lower than the threshold. For example, using the graph in FIG. 6 as an illustrative example, a representative value of 0 to 0.01 may indicate a healthy process chamber subsystem, a representative value of 0.01 to 0.5 may indicate a deteriorating process chamber subsystem, a representative value of 0.5 to 0.99 may indicate a critical process chamber subsystem, and a representative value of 0.99 to 1.0 may indicate that a failure has occurred. The illustrated set of health status indicators is merely exemplary, and any indicator may be used.

[0068] FIG. 7 is a flowchart of a method 700 for determining a fault classification of a process chamber subsystem using a machine learning model, according to an aspect of the present disclosure. Method 700 may be implemented by processing logic that includes hardware (circuits, dedicated logic, etc.), software (such as a run on a general-purpose computer system or a dedicated machine), firmware, or some combination thereof. In one implementation, method 700 may be implemented by a computer system such as computer system architecture 100 of FIG. 1. In other or similar implementations, one or more operations of method 700 may be performed by one or more other machines not depicted in the figure. In some aspects, one or more operations of method 600 may be performed by server machine 170, server machine 180, and / or prediction server 112.

[0069] In block 710, the processing logic obtains sensor data associated with operations executed in the process chamber. In some embodiments, the operations may include a deposition process executed in the process chamber to deposit one or more layers of a film on the surface of a substrate, an etching process executed on one or more layers of a film on the surface of the substrate, and the like. The operations may be executed according to a recipe. The sensor data may include one or more values such as temperature (e.g., heater temperature), interval, pressure, high-frequency radio frequency, voltage of an electrostatic chuck, current, material flow, power, voltage, and the like. The sensor data may be associated with or indicative of manufacturing parameters such as hardware parameters (e.g., size, type, etc.) of the manufacturing equipment 124, or process parameters of the manufacturing equipment 124.

[0070] In block 712, the processing logic applies a machine learning model (e.g., model 190) to the acquired sensor data. The machine learning model may be used to generate one or more values associated with the predicted behavior of the process chamber subsystem. For example, the machine learning model may use the k-NN algorithm to generate the predicted behavior of the process chamber subsystem using the training set T. In some embodiments, the machine learning model is trained using historical sensor data of the subsystem of the process chamber and task data associated with the recipe used to perform the operations.

[0071] In block 714, the processing logic generates an output via the machine learning model based on the sensor data. In some embodiments, the output may include a vectorized version of one or more scalar values generated by the machine learning model. For example, the output may be at least one vector value indicating a pattern (e.g., a fault pattern) that explains the contribution of each sensor to one or more scalar values generated by the machine learning model.

[0072] In block 716, the processing logic determines whether the process chamber subsystem has suffered a failure. In some embodiments, the failure may include mechanical failure, high or low pressure, high or low gas flow, high or low temperature, etc. In some embodiments, the processing logic may determine whether the process chamber subsystem has suffered a failure by comparing the output to a predetermined threshold. In some embodiments, the processing logic may determine whether the process chamber subsystem has suffered a failure by determining that the output does not match the expected behavior. In response to the processing logic determining that the process chamber subsystem has not suffered a failure (e.g., the scalar value of the output does not exceed a predetermined threshold), the processing logic may proceed to block 710. In response to the processing logic determining that the process chamber subsystem has suffered a failure (e.g., the scalar value of the output exceeds a predetermined threshold), the processing logic may proceed to block 718.

[0073] In block 718, the processing logic may identify the type of failure based on the output. In some embodiments, the processing logic may use a classification algorithm, such as a radial basis function (RBF) network, a neural network, or any other statistical or machine learning-based model. An example of an RBF network is described in U.S. Patent No. 9,852,371, which is incorporated herein by reference in its entirety. In particular, the processing logic may compare the fault pattern to a library of known fault patterns and determine the type of failure based on the similarity of the fault pattern when compared to the known fault patterns. In some embodiments, the prediction system 110 may generate a classification algorithm. In some embodiments, the classification algorithm may learn new fault patterns, for example, using a feedback mechanism. In some embodiments, the classification algorithm may also generate a confidence value. The confidence value may indicate the confidence level of the prediction. The confidence value may be generated by the processing logic based on the similarity of the fault pattern to the known fault patterns.

[0074] In block 720, the processing logic may execute a corrective action based on the identified failure. In some embodiments, the corrective action may include generating an alert or an indicator for the determined problem for the client device 120. In some embodiments, the corrective action may include the processing logic adjusting one or more parameters of a deposition process recipe (e.g., a temperature setting for the process chamber, a pressure setting for the process chamber, a flow rate setting for a precursor of a material included in the film deposited on the substrate surface, etc.) based on a desired property for the film. In some embodiments, the deposition process recipe may be adjusted before, during (e.g., in real time), or after the deposition process.

[0075] FIG. 8 is a block diagram illustrating a computer system 800 according to a particular embodiment. In some embodiments, the computer system 800 may be connected to other computer systems (e.g., via a network such as a local area network (LAN), intranet, extranet, or the Internet). The computer system 800 may operate in the role of 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 800 may be provided by a personal computer (PC), tablet PC, set-top box (STB), personal digital assistant (PDA), cellular phone, web appliance, server, network router, switch or bridge, or any device capable of executing a set (serial or otherwise) of instructions to specify actions to be taken by that device. Further, the term "computer" shall be taken to include any collection of computers that individually or collectively execute a set (or multiple sets) of instructions to perform any one or more of the methods described herein.

[0076] In a further aspect, computer system 800 may include a processing device 802, volatile memory 804 (e.g., random access memory (RAM)), non-volatile memory 806 (e.g., read only memory (ROM) or electrically erasable programmable ROM (EEPROM)), and a data storage device 816, which may communicate with each other via a bus 808.

[0077] The processing device 802 may be provided by one or more processors such as general-purpose processors (e.g., complex instruction set computing (CISC) microprocessors, reduced instruction set computing (RISC) microprocessors, very long instruction word (VLIW) microprocessors, microprocessors implementing other types of instruction sets, or microprocessors implementing a combination of types of instruction sets), or dedicated processors (e.g., application specific integrated circuits (ASICs), field programmable gate arrays (FPGAs), digital signal processors (DSPs), or network processors).

[0078] The computer system 800 may further include a network interface device 822 (e.g., coupled to network 874). The computing device 800 may also include a video display device 810 (e.g., LCD), an alphanumeric input device 812 (e.g., keyboard), a cursor control device 814 (e.g., mouse), and a signal generation device 820.

[0079] In some embodiments, the data storage device 816 may include a non-transitory computer-readable storage medium 824 that may store instructions 826 encoding one or more of the methods or functions described herein, including instructions encoding components of FIG. 1 (e.g., corrective action component 122, prediction component 114, etc.) and instructions for performing the methods described herein.

[0080] Instruction 826 may also exist, in whole or in part, within volatile memory 804 and / or within processing device 802 during its execution by computer system 800, and thus, volatile memory 804 and processing device 802 may also constitute a machine-readable storage medium.

[0081] Computer-readable storage medium 824 is shown as a single medium in the illustrative example, but the term "computer-readable storage medium" shall include a single medium or multiple media (e.g., a centralized or distributed database, and / or associated cache and server) that store one or more sets of executable instructions. The term "computer-readable storage medium" shall also include any tangible medium that has the ability to store and encode a set of instructions for causing a computer to perform any one or more of the methods described herein. The term "computer-readable storage medium" shall include, but not be limited to, solid state memory, optical media, and magnetic media.

[0082] The methods, components, and features described herein may be implemented by separate hardware components or integrated into the functions of other hardware components such as ASICs, FPGAs, DSPs, or similar devices. Additionally, the methods, components, and features may be implemented by firmware modules or functional circuits within the hardware device. Further, the methods, components, and features may be implemented in any combination of hardware devices and computer program components or in a computer program.

[0083] Unless otherwise specifically stated, terms such as "receive", "perform", "provide", "acquire", "cause", "access", "determine", "add", "use", "train", or the like refer to acts and processes implemented or carried out by a computer system that manipulates data represented as physical (electronic) quantities within computer system registers and memories and converts them into other data similarly represented as physical quantities within a computer system memory or register or other such information storage, transmission, or display device. Also, terms such as "first", "second", "third", "fourth", etc., when used in this specification, are intended as labels to distinguish different elements and may not have an ordering meaning according to their numerical representation.

[0084] The examples described in this specification also relate to an apparatus for performing the methods described herein. This apparatus can be specially constructed to perform the methods described herein, or it can include a general-purpose computer system selectively programmed by a computer program stored in the computer system. Such a computer program can be stored on a computer-readable tangible storage medium.

[0085] The methods and illustrative examples described in this specification are not inherently related to any particular computer or other device. Various general-purpose systems can be used in accordance with the techniques described herein, or it can be shown to be convenient to construct more specialized devices for implementing each of the methods described herein and / or their individual functions, routines, subroutines, or operations. Examples of structures for various such systems are specified in the above description.

[0086] The foregoing description is illustrative and not restrictive. Although the present disclosure has been described with reference to specific illustrative examples and implementations, it should be recognized that the present disclosure is not limited to the examples and implementations described. The scope of the present disclosure should be determined with reference to the following claims, along with the full scope of equivalents to which the claims are entitled.

Claims

1. A method, comprising: obtaining, by a processor, a plurality of sensor values associated with a deposition process executed in a process chamber according to a recipe to deposit a film on a surface of a substrate; applying a machine learning model to the plurality of sensor values, wherein the machine learning model is trained based on historical sensor data of a subsystem of the process chamber and task data associated with the recipe for depositing the film; generating an output of the machine learning model, wherein the output indicates a health state of the subsystem and the output includes a vector value indicating a failure pattern; using a classification algorithm to determine a type of failure that the subsystem is subject to based on the failure pattern; and the method includes the above steps.

2. The method according to claim 1, wherein the output includes a scalar value indicating a difference between measured values of a set of sensors associated with the subsystem and predicted values of the set of sensors.

3. The method according to claim 1, further comprising using a conversion function to convert the output into a representative value within a predefined range.

4. The method according to claim 3, wherein the conversion function includes at least one of a linear function, a logit function, a sigmoid function, or an exponential function.

5. The method according to claim 1, further comprising using a classification algorithm to learn a new failure pattern or generate a confidence value indicating a confidence level of a prediction.

6. The method according to claim 1, wherein the classification algorithm compares the failure pattern with a library of known failure patterns.

7. The classification algorithm is the method according to claim 1, including a radial basis function (RBF) network or a neural network.

8. The method according to claim 1, further including performing a corrective measure based on the failure pattern.

9. A system, including a memory, and a processing device, wherein the processing device acquires a plurality of sensor values associated with a deposition process executed according to a recipe in a process chamber to deposit a film on a surface of a substrate; applies a machine learning model to the plurality of sensor values, where the machine learning model is trained based on historical sensor data of a subsystem of the process chamber and task data associated with the recipe for depositing the film; generates an output of the machine learning model, where the output indicates a health state of the subsystem and the output includes a vector value indicating a failure pattern; uses a classification algorithm to determine a type of failure that the subsystem undergoes based on the failure pattern; and is operably coupled to the memory to perform operations including the above.

10. The system according to claim 9, wherein the output includes a scalar value indicating a difference between measured values of a set of sensors associated with the subsystem and predicted values of the set of sensors.

11. The system according to claim 9, wherein the processing device is further configured to perform an additional operation of using a conversion function to convert the output into a representative value within a predefined range. The system according to claim 9, further performing an additional operation including using a classification algorithm to learn a new fault pattern or to generate a confidence value indicating a confidence level of a prediction.

13. The system according to claim 9, wherein the processing device further performs an additional operation including executing a corrective measure based on the fault pattern.

14. A method comprising: obtaining, by a processor, sensor data associated with a deposition process executed in a process chamber to deposit a film on a surface of a substrate, the sensor data including sensor values associated with a subsystem of the process chamber; obtaining task data associated with a recipe for depositing the film; determining whether a training set includes a sufficient amount of training data to train a machine learning model based on a number of mappings in the training set or a measure of diversity of the training set; training the machine learning model using the training set based on the sensor data and the task data, the machine learning model being trained to generate prediction data indicating predicted sensor values of the subsystem; including.

15. The method according to claim 14, further comprising performing an outlier detection technique to remove one or more anomalies from the training set.

16. The method according to claim 14, wherein the machine learning model includes a k-nearest neighbor (k-NN) algorithm.

17. The method according to claim 14, wherein the subsystem comprises a set of sensors for monitoring operating parameters of the process chamber.

18. The method according to claim 17, wherein the operation parameter includes pressure associated with the process chamber, flow rate associated with the process chamber, or temperature associated with the process chamber.

Images