Parts quality control system for manufacturing equipment

Abstract

The method includes receiving first data indicative of a range of values ​​for a quality parameter of a type of manufacturing chamber component. Each value in the range of values ​​satisfies one or more threshold criteria. The method further includes providing the first data to a physics-based model of the manufacturing chamber. The method further includes receiving second data from the physics-based model indicative of a relationship between the value of the quality parameter and predicted conditions in the manufacturing chamber. The method further includes determining whether a new manufacturing chamber component of the manufacturing chamber component type should be installed in the manufacturing chamber based on the relationship between the value of the quality parameter and the predicted conditions.

Description

[Technical field]

[0001] This disclosure relates to a parts control system, and more particularly to a quality control system for critical parts of components of manufacturing equipment. [Background technology]

[0002] Chambers are used in many types of processing systems. Examples of chambers include etch chambers, deposition chambers, anneal chambers, implantation chambers, etc. Typically, a substrate, such as a semiconductor wafer, is placed on a substrate support within the chamber, and conditions within the chamber are set and maintained to process the substrate. The characteristics of the various components of the processing chamber affect the properties of the finished substrate. Summary of the Invention

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

[0004] In one aspect of the disclosure, a method includes receiving first data indicative of a range of values ​​for a quality parameter of a type of manufacturing chamber component. Each value in the range of values ​​satisfies one or more threshold criteria. The method further includes providing the first data to a physics-based model of the manufacturing chamber. The method further includes receiving second data from the physics-based model indicative of a relationship between the value of the quality parameter and predicted conditions in the manufacturing chamber. The method further includes determining whether a new manufacturing chamber component of the manufacturing chamber component type should be installed in the manufacturing chamber based on the relationship between the value of the quality parameter and the predicted conditions.

[0005] In another aspect of the present disclosure, a non-transitory computer-readable storage medium is disclosed. The non-transitory computer-readable storage medium stores instructions that, when executed by a processing device, cause the processing device to perform operations. The operations include receiving first data indicative of values ​​of one or more quality parameters of one or more manufacturing chamber components installed in the manufacturing chamber. The operations further include receiving second data indicative of values ​​of the quality parameters of a new manufacturing chamber component, the new manufacturing chamber component being considered for installation in the manufacturing chamber. The operations further include providing the first data and the second data to a physics-based model. The operations further include receiving a prediction of performance of the manufacturing chamber with the new manufacturing chamber component installed from the physics-based model. The operations further include determining whether to install the new manufacturing chamber component in the manufacturing chamber based on the predicted performance.

[0006] In another aspect of the disclosure, a method includes receiving predicted performance data for a manufacturing chamber as output from a physics-based model of the manufacturing chamber. The output of the physics-based model is generated considering values ​​of quality parameters of components of the manufacturing chamber. The method further includes receiving measured performance data associated with the manufacturing chamber. The method further includes causing a corrective action to be performed considering the predicted performance data and the measured performance data.

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

[0008] [Figure 1] FIG. 1 is a block diagram illustrating an example system (example system architecture) according to some embodiments. [Figure 2A] FIG. 1 is a block diagram of an example dataset generator used to create a dataset for a model, according to some embodiments. [Figure 2B]FIG. 1 is a block diagram of an example dataset generator used to create a dataset for a model, according to some embodiments. [Diagram 3] FIG. 2 is a block diagram illustrating a system for generating output data (e.g., prediction data 168 of FIG. 1) according to some embodiments. [Figure 4A] 1 is a flow diagram of a method associated with characterizing one or more components of a manufacturing device to trigger corrective action, according to certain embodiments. [Figure 4B] 1 is a flow diagram of a method associated with characterizing one or more components of a manufacturing device to trigger corrective action, according to certain embodiments. [Figure 4C] 1 is a flow diagram of a method associated with characterizing one or more components of a manufacturing device to trigger corrective action, according to certain embodiments. [Figure 4D] 1 is a flow diagram of a method associated with characterizing one or more components of a manufacturing device to trigger corrective action, according to certain embodiments. [Figure 4E] 1 is a flow diagram of a method associated with characterizing one or more components of a manufacturing device to trigger corrective action, according to certain embodiments. [Diagram 5] 1 illustrates a cross-sectional view of a manufacturing chamber (eg, a semiconductor wafer manufacturing chamber) in accordance with some embodiments. [Figure 6] FIG. 1 is a block diagram illustrating a computer system, according to some embodiments. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0009] Described herein is a technique directed to utilizing data regarding characteristics of components of processing equipment. Processing equipment (e.g., manufacturing equipment, manufacturing chambers, etc.) may include multiple components. Processing equipment may be used to produce substrates (e.g., semiconductor wafers). The characteristics of the product (e.g., wafer) produced by the processing equipment are determined by the conditions under which the substrate was processed. Precise knowledge of the property values ​​at the location of the workpiece (e.g., wafer) during operation may be used to predict the properties of the finished product, consistently produce substrates with the same properties (within tolerances), and adjust processing parameters to optimize production of substrates.

[0010] Several components can affect the properties proximate to the wafer in the fabrication chamber. For example, the properties of the showerhead, gas manifold, chamber liner, pumping plate, substrate support, processing ring, etc. can all contribute to properties proximate to the substrate. Typically, components of processing equipment are manufactured within a level of tolerance. In some cases, parts (components) that are within the manufacturing specifications or tolerances may exhibit different properties. In some cases, these differences within the manufacturing specifications may be enough to change the properties at the location of the substrate being processed and affect the properties of the finished substrate. Processing performed using components may result in unacceptable performance (e.g., out-of-spec product properties, product property variations above thresholds, etc.), even for components manufactured within manufacturing specifications.

[0011] In conventional systems, the determination of the quality of components of a manufacturing device may be done empirically. Such an approach may require extensive experimentation, especially when one (or a combination) of multiple components may be responsible for the undesirable characteristics. Experimentation may be costly in terms of time spent, materials used, energy and gas consumed, etc. Additionally, as a manufacturing device processes successive products, the characteristics of the device (and possibly the product) tend to drift. Characteristics may drift as a result of material accumulation, aging of parts, temperature and gas cycling, changes during maintenance procedures, etc. Characteristics of components that affect product quality may drift within or outside of the range of characteristics expected from parts within manufacturing specifications. Acquired knowledge (e.g., experimentally obtained knowledge) that maps the performance of a manufacturing device (e.g., finished product characteristics) to the use of specific components, combinations of components, processing parameters, etc. may become invalid over time due to drift.

[0012] An additional problem with an experimental approach to component quality is that components may have a finite life. Each time any component in the chamber is replaced, a new period of experimentation may be conducted to understand the performance of the new component combination. In some cases, the characteristics of consumable parts (e.g., parts that are frequently replaced) may affect product quality. Empirically characterizing the performance of such parts may be inconvenient, inefficient, or even impossible. If a new set of characteristics is to be generated in a manufacturing chamber (e.g., to process a different type of product), further characterization experiments may be required.

[0013] In one or more embodiments, the disclosed methods and devices address at least some of these deficiencies of conventional approaches. The present disclosure allows for the use of part quality data for various components installed in a manufacturing apparatus to determine which parts to install in the chamber, when to replace the parts, and / or to perform other actions. In some embodiments, part quality data for components that affect substrate processing in the manufacturing chamber is available. In some embodiments, component characteristic values ​​can be measured to determine values ​​of quality parameters of the components. The part (component) quality data may be fed into a physics-based digital twin model of the manufacturing chamber. One or more parameters of the part may be measured. The parameters of the part that fall within the manufacturing specifications may be characterized in more detail than determining whether the part falls within the manufacturing specifications. The data collected by measuring the component characteristics may be correlated to the performance of the component. For example, various characteristics of the electrostatic chuck may be measured, such as surface material properties, surface engineering roughness characterization, heating electrode, RF electrode, or chuck electrode characterization. Characterization of the surface of the heating electrode and / or chucking device may be related to heat transfer from the chuck to the substrate. The characteristics of the chuck electrode can be used to select parameters for optimally chucking the substrate, such as to determine an optimal power for a target chucking strength, to determine an optimal ramp-up of power supplied to the electrode to avoid overshoot or manage ramp-up times, etc.

[0014] In some embodiments, the disclosed method and device further includes providing the generated data considering the measurement data of the component of the manufacturing equipment to a physics-based model, such as a digital twin model. As used herein, a digital twin is a digital replica of a physical asset, such as, for example, a manufactured part, an equipment including a plurality of manufactured parts. The digital twin includes properties of the physical asset, which may include coordinate axis dimensions, weight properties, material properties (e.g., density, surface roughness, emissivity), electrical properties (e.g., conductivity), optical properties (e.g., reflectivity), and the like. The physics-based model is configured to provide predictions of conditions associated with the manufacturing equipment (e.g., conditions in the manufacturing chamber, conditions proximate to the workpiece, and the like). The physics-based model can perform calculations related to various properties, such as heat transfer calculations, gas conductance calculations, and the like. In some embodiments, the disclosed method further includes receiving predicted performance data associated with the component from the physics-based model. The physics-based model can generate the predicted performance data considering measurements of part quality.

[0015] In some embodiments, the disclosed method further includes using the predicted performance data of the manufacturing equipment (or a component of the manufacturing equipment) output by the physics-based model to determine whether the component (or group of components) is suitable for installation in the manufacturing equipment (e.g., a manufacturing or processing chamber). The physics-based model may determine that the value of one or more parameters of the component affects a property value proximate to a workpiece (e.g., a substrate, semiconductor wafer, etc.) associated with the processing equipment. The property value proximate to the workpiece may affect the process performance (e.g., a property of the finished substrate).

[0016] In some embodiments, a process has process specifications, which may include target values ​​for characteristics, allowable tolerances for those characteristics, etc. By providing measurements of quality parameters to the physics-based model, the physics-based model can generate predicted performance data for a manufacturing equipment incorporating components described by the quality parameters. The predicted performance data may be used to determine whether a component that may be installed as part of a manufacturing or processing equipment is suitable for a process having a set of process specifications (e.g., whether the component is capable of generating conditions that meet the target values, whether the component is capable of generating conditions consistently within the tolerances of the process specifications, etc.). The physics-based model may include part quality metrics for components already installed in the chamber. Potential interactions between quality parameters of multiple components may be described by the physics-based model. Comparing conditions associated with the manufacturing equipment predicted by the physics-based model (e.g., characteristic values ​​inside a processing or manufacturing chamber) to measured conditions can highlight potential issues, failures, drift, aging, etc. of the components of the manufacturing equipment.

[0017] In some aspects, the present disclosure enables a method including a processing device receiving first data indicative of a range of values ​​for a quality parameter of a type of manufacturing chamber component. Each value in the range satisfies one or more first threshold criteria. The method further includes providing the first data to a physics-based model of the manufacturing chamber. The method further includes receiving second data from the physics-based model. The second data indicative of a relationship between the value of the quality parameter and predicted conditions in the manufacturing chamber. The method further includes determining whether a manufacturing chamber component of the manufacturing chamber component type should be installed in the manufacturing chamber based on the predicted conditions.

[0018] In another aspect, the present disclosure enables a method that includes receiving data indicative of values ​​of one or more quality parameters of one or more manufacturing chamber components installed in a manufacturing chamber. The method further includes receiving second data indicative of values ​​of the quality parameters of a new manufacturing chamber component. The new manufacturing chamber component is being considered for installation in the manufacturing chamber. The method further includes providing the first data and the second data to a physics-based model. The method further includes receiving a prediction of performance of the manufacturing chamber with the new manufacturing chamber component installed from the physics-based model. The method further includes determining whether to install the new manufacturing chamber component in the manufacturing chamber based on the predicted condition.

[0019] In another aspect, the present disclosure enables a method that includes receiving predicted performance data for a manufacturing chamber. The predicted performance data is output by a physics-based model. The predicted performance data is output considering a value of a first quality parameter of a first component of the manufacturing chamber. The method further includes receiving measured performance data associated with the manufacturing chamber. The method further includes causing a corrective action to be performed considering the predicted performance data and the measured performance data.

[0020] Aspects of the present disclosure provide technical advantages over conventional methods. Characterization of potential components to be installed in a manufacturing tool may be accomplished by performing measurements and feeding data indicative of those measurements into a physics-based (e.g., digital twin) model. This characterization avoids the conventional and costly method of testing one or more potential components by fabricating wafers using the components, performing measurements on the wafers, and inferring the quality of the components from the measurements. Aspects of the present disclosure also enable an improved selection process of components to be used in a manufacturing process when multiple nominally equivalent (e.g., manufactured to the same specifications) components are available. Some components may include quality parameters associated with adjustable metrics, e.g., heating elements that may be tuned by providing different amounts of power may perform differently. In some embodiments, characterizing the components may allow the controls to be adjusted to produce more favorable processing conditions. Additionally, chamber components may be re-characterized (e.g., during a planned maintenance event) to account for component drift. If a manufacturing chamber is suspected to have a problem, the components may be re-characterized to confirm or rule out the component as the cause.

[0021] 1 is a block diagram illustrating an example system 100 (example system architecture) according to some embodiments. System 100 includes client devices 120, manufacturing equipment 124, sensors 126, metrology equipment 128, prediction server 112, and data store 140. Prediction server 112 may be part of prediction system 110. Prediction system 110 may further include server machines 170 and 180.

[0022] In some embodiments, the manufacturing tool 124 (e.g., a cluster tool) is part of a substrate processing system (e.g., an integrated processing system). The manufacturing tool 124 includes one or more of a controller, an enclosure system (e.g., a substrate carrier, a front opening unified pod (FOUP), an autoteach FOUP, a process kit enclosure system, a substrate enclosure system, a cassette, etc.), a side storage pod (SSP), an aligner device (e.g., an aligner chamber), a factory interface (e.g., an equipment front end module (EFEM)), a load lock, a transfer chamber, one or more processing chambers, a robot arm (e.g., disposed in the transfer chamber, disposed in the front interface, etc.), etc. The enclosure system, the SSP, and the load lock are attached to the factory interface, and the robot arm disposed in the factory interface is for transferring contents (e.g., a substrate, a process kit ring, a carrier, a validation wafer, etc.) between the enclosure system, the SSP, the load lock, and the factory interface. The aligner device is disposed in the factory interface and aligns the contents. The load locks and processing chambers are mounted in a transfer chamber, and a robot arm located in the transfer chamber is for transferring contents (e.g., substrates, process kit rings, carriers, validation wafers, etc.) between the load locks, processing chambers, and the transfer chamber. In some embodiments, the manufacturing equipment 124 includes components of a substrate processing system. In some embodiments, the manufacturing equipment 124 is used to produce one or more products (e.g., substrates, semiconductors, wafers, etc.). In some embodiments, the manufacturing equipment 124 is used to produce one or more components used in the substrate processing system. The manufacturing equipment 124 can include a variety of components.Examples of chamber components can include a substrate support, a chuck (e.g., an electrostatic chuck, a vacuum chuck, etc.), a ring (e.g., a process kit ring), chamber walls, a base, a showerhead, a gas distribution plate, a liner, a liner kit, a shield, a plasma screen, a flow equalizer, a cooling base, a chamber viewport, a chamber lid, a nozzle, etc.

[0023] The sensors 126 can provide sensor data 142 associated with the manufacturing equipment 124 (e.g., associated with manufacturing a corresponding product, such as a wafer, by the manufacturing equipment 124). The sensors 126 can further include sensors used to measure quality parameters of components of the manufacturing equipment 124. The sensor data 142 can be used, for example, for equipment health and / or product health (e.g., equipment quality and / or product quality). The manufacturing equipment 124 can operate according to a recipe or over a period of time to produce a product. The sensor data 142 can include process data 144 and part quality data 146. The process data can include data associated with processing a product using the manufacturing equipment 124, for example, trace data of a sensor associated with the manufacturing equipment 124. In some embodiments, the process data 144 of the sensor data 142 can include one or more values ​​of temperature (e.g., heater temperature), spacing (SP), pressure, high frequency radio frequency (HFRF), electrostatic chuck (ESC) voltage, current, flow rate (e.g., of one or more gases), power, voltage, etc. The sensor data 142 may include part quality data 146. The part quality data 146 may include data describing characteristics of components associated with the manufacturing equipment 124. The part quality data 146 may include material properties, physical parameters, optical parameters, and the like of the components. The part quality data 146 may include values ​​of one or more of coordinate dimensions (e.g., size and shape of the part including internal subcomponents such as electrodes disposed below the surface of the substrate support), weight properties, material properties (e.g., density, surface roughness, emissivity), electrical properties, optical properties, properties related to heat transfer, and the like. The sensor data 142 may include historical sensor data and current sensor data. The manufacturing equipment 124 may be configured according to manufacturing parameters 150. The manufacturing parameters 150 may be associated with or indicative of parameters such as hardware parameters (e.g., settings or components (e.g., size, type, etc.) of the manufacturing equipment 124) and / or process parameters of the manufacturing equipment.The manufacturing parameters 150 may include historical and / or current manufacturing data. The manufacturing parameters 150 may indicate input settings to the manufacturing devices (e.g., heater power, gas flows, etc.). The sensor data 142 and / or manufacturing parameters 150 may be provided while the manufacturing equipment 124 is performing the manufacturing process (e.g., equipment readings as the product is processed). The sensor data 142 may vary from product to product (e.g., each wafer may be characterized by unique process data 144, each component may be characterized by unique part quality data 146, etc.).

[0024] In some embodiments, the sensor data 142, the metrology data 160, and / or the manufacturing parameters 150 may be processed (e.g., by the client device 120 and / or by the prediction server 112). Processing the data may include generating features. In some embodiments, the features are patterns (e.g., slope, width, height, peaks, etc.) of the sensor data 142, the metrology data 160, and / or the manufacturing parameters 150, or combinations of values ​​(e.g., power derived from voltage and current, etc.) from the sensor data 142, the metrology data 160, and / or the manufacturing parameters 150. The features may be used by the prediction component 114 to perform signal processing and / or to obtain prediction data 168, possibly to perform corrective actions. The prediction data 168 may be any data associated with the prediction system 110, such as predicted performance data of components of the manufacturing equipment 124, predicted performance data of the manufacturing equipment 124, predicted metrology data, etc.

[0025] Each instance (e.g., set) of the process data 144 of the sensor data 142 may correspond to a product (e.g., a wafer), a set of manufacturing equipment, a type of substrate produced by the manufacturing equipment, combinations thereof, etc. Each instance of the metrology data 160 and manufacturing parameters 150 may similarly correspond to a product, a set of manufacturing equipment, a type of substrate produced by the manufacturing equipment, combinations thereof, etc. The data store may further store information relating sets of different data types, e.g., information indicating that a set of sensor data, a set of metrology data, and / or a set of manufacturing data are all associated with the same product, manufacturing equipment, type of substrate, etc.

[0026] Each instance of part quality data 146 of sensor data 142 may correspond to a component of a manufacturing device, a combination of components, a type of chamber or device, etc. The part quality data 146 may include data relating a component of a manufacturing device 124 to the sensor data 144, the manufacturing parameters 150, the metrology data 160, and / or the forecast data 168. For example, a particular component may be installed in a manufacturing chamber. The sensor 126 may be used to measure a quality parameter of the component. The data store 140 may store data relating sensor process data 144 collected during processing with the installed component to the quality data 146, the data store 140 may store data relating metrology data 160 of a product produced with the installed component to the quality data 146, etc.

[0027] In some embodiments, the predictive system 110 can generate the predictive data 168 using physics-based models and / or digital twin models. For example, a predicted change in performance of a manufacturing device when a new component is installed can be generated by feeding part quality data 146 associated with the new component to a physics-based model. The use of physics-based models to generate predictive data is discussed further below, e.g., in connection with FIGS. 4B-4E. In some embodiments, the predictive system 110 can generate the predictive data 168 using machine learning. For example, data indicative of manufacturing conditions may be provided as an input to a trained machine learning model. The machine learning model can provide predicted characteristics of the finished product as an output. The use of machine learning models to generate predictive data is discussed in more detail below, e.g., in connection with FIG. 3.

[0028] Client devices 120, manufacturing equipment 124, sensors 126, metrology equipment 128, prediction server 112, data store 140, server machine 170, and server machine 180 may be coupled together via network 130 to generate sensor data 142, metrology data 160, and / or predictive data 168. This data may be used in implementing corrective actions.

[0029] In some embodiments, the network 130 is a public network that provides the client device 120 with access to the prediction server 112, the data store 140, and / or other publicly available computing devices. In some embodiments, the network 130 is a private network that provides the client device 120 with access to the manufacturing equipment 124, the sensors 126, the metrology equipment 128, the data store 140, and / or other privately available computing devices. The 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.

[0030] The client device 120 may include computing devices such as a personal computer (PC), a laptop, a mobile phone, a smartphone, 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 include a corrective action component 122. The corrective action component 122 may receive user input of instructions associated with the manufacturing equipment 124 (e.g., via a graphical user interface (GUI) displayed via the client device 120). In some embodiments, the corrective action component 122 sends instructions to the prediction system 110, receives output (e.g., prediction data 168) from the prediction system 110, determines a corrective action based on the output, and causes the corrective action to be implemented.

[0031] In some embodiments, the prediction system 110 can further include a prediction component 114. The prediction component 114 can generate prediction data 168 using data obtained from the model 190. In some embodiments, the prediction component 114 provides the prediction data 168 to the client device 120, which takes the prediction data 168 into account and triggers a corrective action via the corrective action component 122. In some embodiments, the corrective action component 122 obtains sensor data 142 associated with one or more components of the manufacturing equipment 124 (e.g., from the data store 140, etc.) and provides the sensor data 142 to the prediction system 110.

[0032] In some embodiments, the corrective action component 122 stores the sensor data 142 (e.g., part quality data 146) in the data store 140, and the prediction server 112 retrieves the sensor data 142 from the data store 140. In some embodiments, the prediction server 112 can store the output of the trained model 190 (e.g., prediction data 168) in the data store 140, and the client device 120 can retrieve the output from the data store 140. In some embodiments, the corrective action component 122 receives corrective action instructions from the prediction system 110 and causes the corrective action to be implemented. Each client device 120 can include an operating system that enables a user to one or more of create, view, or edit data (e.g., instructions associated with a manufacturing equipment 124, corrective actions associated with a manufacturing equipment 124, etc.).

[0033] The part quality data 146 includes data collected by the sensors 126. The part quality data 146 includes measurements of properties of components (parts) associated with the manufacturing equipment 124. The properties and parameters included in the part quality data 146 may affect (or be predicted to affect) conditions proximate to a product (e.g., a semiconductor wafer) during processing. The part quality data 146 may be taken as an input by the prediction system 110, which may generate prediction data 168 as an output. The part quality data 146 may be provided to a physics-based model (e.g., model 190). The physics-based model may be or may include a digital twin model. The physics-based model may output predicted performance data (e.g., prediction data 168) of one or more components, predicted performance data of the manufacturing equipment 124 in which the one or more components are installed, predicted performance data of a produced product (e.g., predicted metrology data), etc. In some embodiments, the components associated with the part quality data 146 may be produced within manufacturing specifications. In some embodiments, part quality data 146 may capture differences between different instances of the same component (eg, same part number) within manufacturing specifications, operational specifications, etc. for that component.

[0034] The predictive system 110 can compare the predicted property data of the corresponding products with the metrology data 160 to output predictions of manufacturing faults, drift in chamber components, and / or corrective actions. The predictive system 110 can input the predicted property data and the metrology data 160 into a model (e.g., model 190) to generate data indicative of corrective actions. The model may be a trained machine learning model in some embodiments. In some embodiments, the corrective actions may be updates to the manufacturing parameters 150 for future substrates produced by the manufacturing tool 124.

[0035] In some embodiments, the metrology data 160 corresponds to historical characteristic data of a product (e.g., produced using historical sensor data and manufacturing parameters associated with historical manufacturing parameters), and the predictive data 168 is associated with predicted characteristic data (e.g., of a product that is produced or has been produced under conditions recorded by current sensor data and / or current manufacturing parameters). In some embodiments, the predictive data 168 is predicted metrology data (e.g., virtual metrology data) of a product that is produced or has been produced according to conditions recorded as current sensor data and / or current manufacturing parameters. In some embodiments, the predictive data 168 is or includes an indication of an anomaly (e.g., an abnormal product, an abnormal component, an abnormal manufacturing equipment, an abnormal energy usage, etc.) and / or one or more causes of the anomaly. In some embodiments, the predictive data 168 includes an indication of aging or drift in some components, such as manufacturing equipment 124, sensors 126, metrology equipment 128, etc. In some embodiments, the predictive data 168 includes an indication of end of life of a component, such as manufacturing equipment 124, sensors 126, metrology equipment 128, etc.

[0036] Running a manufacturing process that results in a defective product can be costly in terms of time, energy, products, components, manufacturing equipment 124, costs of identifying and discarding the defective product, etc. By inputting part quality data 146 (e.g., measurements of properties of one or more components of manufacturing equipment 124) into one or more physics-based digital twin models (e.g., model 190), receiving an output of prediction data 168, and performing corrective actions based on the prediction data 168, system 100 can have the technical advantage of avoiding the costs of manufacturing, identifying, and discarding defective products.

[0037] Running a manufacturing process that results in a failure of a component of the manufacturing equipment 124 can be costly in terms of downtime, product damage, equipment damage, rush ordering of replacement components, etc. By inputting sensor data 142 (e.g., measurements of properties of one or more components of the manufacturing equipment 124) into a physics-based model (e.g., model 190), receiving output of prediction data 168 (e.g., predicted performance of the manufacturing equipment 124), comparing results from the same components over time to diagnose drifting or failing components (recorded as prediction data 168), and taking corrective action (e.g., predicted operational maintenance such as replacing, treating, cleaning, etc. of the component) based on the prediction data 168, the system 100 can have the technical advantage of avoiding one or more of the following costs: unexpected component failure, unscheduled downtime, lost productivity, unexpected equipment failure, product scrap, etc. Monitoring the performance of components, e.g., the manufacturing equipment 124, the sensors 126, the metrology devices 128, etc., over time can provide indications of component degradation. By monitoring the performance of a component over time, the operational life of the component can be extended if, for example, measurements indicate that after a standard replacement interval has elapsed, the component may still perform well (e.g., above a threshold) for some time (e.g., until the next planned maintenance event).

[0038] The manufacturing parameters may not be optimal for producing the product, which may result in costly consequences such as increased consumption of resources (e.g., energy, coolant, gas, etc.), increased time to produce the product, increased component failures, increased amount of defective products, etc. By inputting the part quality data 146 into a trained physics-based model (e.g., model 190), receiving the output of the prediction data 168, and performing corrective actions to update the manufacturing parameters (e.g., set optimal manufacturing parameters) (e.g., based on the part quality data 146), the system 100 may have the technical advantage of using optimal manufacturing parameters (e.g., hardware parameters, process parameters, optimal design) to avoid the costly consequences of non-optimal manufacturing parameters.

[0039] In some embodiments, a component may be considered to be installed as part of a manufacturing tool 124. The manufacturing tool 124 may be used for processing in which certain processing parameters are achieved. Part quality data 146 associated with the component may be measured and fed into the physics-based model. The components may be classified according to the parameters achievable in the manufacturing chamber, the consistency of parameter achievement in the chamber, the resource cost (e.g., energy, gas, etc.) to achieve the target parameters in the chamber, etc. The components may be classified as suitable (or unsuitable) for use in one or more manufacturing processes.

[0040] In some embodiments, the corrective action includes providing a warning (e.g., an alarm that stops or does not run a manufacturing process if the predictive data 168 indicates a predicted anomaly, such as an anomaly in a product, component, or manufacturing equipment 124). In some embodiments, the corrective action includes providing feedback control (e.g., modifying a manufacturing parameter in response to the predictive data 168 indicating an anomaly). In some embodiments, the corrective action includes providing machine learning (e.g., modifying one or more manufacturing parameters based on the predictive data 168). In some embodiments, performing the corrective action includes causing an update of one or more manufacturing parameters.

[0041] In some embodiments, corrective actions include scheduling preventative maintenance. Monitoring part quality and performance can indicate when components should be replaced. Utilizing the output of the physics-based model may indicate, for example, that a manufacturing tool 124 (e.g., a manufacturing chamber) may drift more quickly than other chambers of the same or different type due to the process performed on the tool, the components installed on the tool, adjustments made to the process recipe to achieve target parameters, etc. The manufacturing tool 124 can adjust its maintenance schedule based on the output of the model 190, the predictive data 168, etc.

[0042] The manufacturing parameters may include hardware parameters (e.g., replacing a component, using a specific component, replacing a processing chip, updating firmware, etc.) and / or process parameters (e.g., temperature, pressure, flow rate, speed, current, voltage, gas flow, lift speed, etc.). In some embodiments, the corrective action includes triggering preventative operational maintenance (e.g., replacing, treating, cleaning, etc., of a manufacturing equipment 124 component). In some embodiments, the corrective action includes triggering design optimization (e.g., updating manufacturing parameters, manufacturing process, manufacturing equipment 124, etc., for an optimized product). In some embodiments, the corrective action includes updating a recipe (e.g., putting the manufacturing equipment 124 into an idle mode, sleep mode, warm-up mode, etc.).

[0043] Prediction server 112, server machine 170, and server machine 180 may each include one or more computing devices such as a rack mount server, a router computer, a server computer, a personal computer, a mainframe computer, a laptop computer, a tablet computer, a desktop computer, a graphics processing unit (GPU), an accelerator application specific integrated circuit (ASIC) (e.g., a tensor processing unit (TPU)), etc.

[0044] The prediction server 112 can include a prediction component 114. The prediction component 114 can be used to generate prediction data 168. In some embodiments, the prediction component 114 can receive the sensor data 142 and / or the manufacturing parameters 150 (e.g., received from the client device 120 and retrieved from the data store 140) and generate an output for performing a corrective action associated with the manufacturing equipment 124 based on the current data. In some embodiments, the prediction component 114 can use one or more models 190 to determine an output for performing a corrective action based on the current data. The models 190 can be a single model, an ensemble model, or a collection of models used to process the data. The models 190 can include one or more physics-based digital twin models, supervised machine learning models, unsupervised machine learning models, semi-supervised machine learning models, statistical models, etc.

[0045] In some embodiments, data indicative of characteristics of substrates produced using manufacturing equipment including components characterized by part quality data 146 (e.g., performance associated with predicted data 168 informed by part quality data 146) is provided to a trained machine learning model (e.g., model 190). The machine learning model is trained to output data indicative of corrective actions to produce substrates having target characteristics. In some embodiments, data indicative of predicted characteristics of substrates produced using manufacturing equipment 124 including particular components, and metrology data of substrates produced with those components are provided as inputs to a trained machine learning model (e.g., model 190). The trained machine learning model predicts root causes of differences between predicted and measured data (e.g., manufacturing failures, component aging or drift, etc.).

[0046] Historical sensor data may be used in combination with current sensor data to detect drift, changes, aging, etc., of components of the manufacturing equipment 124. The sensor data 142 (including process data 144 measured during operation of the manufacturing equipment 124 and part quality data 146 associated with measurements or predictions of values ​​of quality parameters of components associated with the manufacturing equipment 124) may be monitored over time. Changes in the sensor data 142 over time may generate information indicative of changes in the manufacturing equipment. The part quality data 146 may also provide information about other components of the manufacturing equipment 124, for example, uneven wear on the surface of a substrate support may indicate a problem with a robot handler that places substrates on the substrate support, rather than a problem with the substrate support itself. The predictive component 114 may use combinations and comparisons of these data types to generate predictive data 168. In some embodiments, the predictive data 168 includes data predicting the life of components of the manufacturing equipment 124, sensors 126, etc.

[0047] Typically, conventional systems perform little or no characterization of the components of the manufacturing equipment other than verifying that the components meet specifications. In some systems, the manufacturer of the components may verify (e.g., measure, provide documentation, etc.) that the components meet specifications. All components that meet the same specifications (e.g., all components with the same part number) may be considered identical. Classifying components as satisfactory or unsatisfactory for use in a particular process, a particular chamber, etc. may be done in response to making measurements of produced substrates and determining whether the product is satisfactory or unsatisfactory. Characterizing the variability between nominally identical parts may be done empirically. For example, two nominally identical components may respond differently to voltages applied, processing parameters applied, etc. Characterizing such differences may be performed based on sensor data, metrology data of the finished product, etc. Characterizing the components allows for selection of components suitable for a particular process, selection of components suitable for installation in a particular chamber, adjustment of the operation of the components, etc., without waste associated with producing and measuring substrates.

[0048] In some embodiments, the predictive component 114 can receive data, such as sensor data 142, manufacturing parameters 150, metrology data 160, and perform pre-processing, such as extracting patterns in the data or combining the data into new composite data. The predictive component 114 can then provide the data as input to the model 190. The model 190 can include a physics-based (e.g., digital twin) model that accepts data indicative of part quality of components of the manufacturing equipment as input (e.g., part quality data 146). The physics-based model can utilize the part quality data 146 to make a prediction of chamber performance. The predictive component 114 can receive the predicted data from the model 190. The predictive component 114 can then cause corrective action to be taken. The corrective action can include sending an alert to the client device 120. The corrective action can include generating and / or storing predictive data 168 indicative of drift, aging, or failure of the chamber component.

[0049] In some embodiments, the model 190 may include a trained machine learning model. The predictive component 114 may provide data as input to the trained machine learning model 190. In some embodiments, the model 190 may receive data indicative of process parameters and quality data of the chamber components. The model 190 may generate data as output indicative of predicted characteristics (e.g., predicted metrology) of products produced using the chamber components described by the input quality data. In some embodiments, the predictive component 114 may provide data indicative of manufacturing parameters, sensor data, maintenance history of the manufacturing equipment, and / or metrology as input data to the model 190. The output of the model 190 may include predicted part quality data for components installed on the manufacturing equipment 124. The predicted part quality data may be used to perform corrective actions such as updating process recipes, scheduling maintenance, and providing alerts to users.

[0050] The data store 140 may be a memory (e.g., random access memory), a drive (e.g., hard drive, flash drive), a database system, or another type of component or device capable of storing data. The data store 140 may include multiple storage components (e.g., multiple drives or multiple databases) that may span multiple computing devices (e.g., multiple server computers). The data store 140 may store sensor data 142, manufacturing parameters 150, metrology data 160, and forecast data 168. The sensor data may include process data 144, which may include time traces of sensor data over the duration of a manufacturing process, associations of data with physical sensors, pre-processed data such as averages and composite data, and data indicative of sensor performance over time (i.e., many manufacturing processes).

[0051] The sensor data 142 may include part quality data 146. The part quality data 146 may include data indicative of measurements of one or more quality parameters of one or more components of a manufacturing equipment. The part quality data 146 may include data associated with components installed on the manufacturing equipment 124 and / or data associated with components not yet installed on the manufacturing equipment. The part quality data 146 may include data indicative of component performance. The data indicative of component performance may be generated as output from a model, e.g., a physics-based (e.g., digital twin) model or a machine learning model. The component performance data may include predicted data of chamber performance in which the component is installed, performance data of a chamber in which multiple components are installed with measured quality parameters, etc.

[0052] The manufacturing parameters 150 and metrology data 160 may include similar characteristics as the sensor data 142, such as pre-processed data, averages, composite data, association of data with equipment or products, etc. The process data 144, part quality data 146, manufacturing parameters 150, and metrology data may include historical data (e.g., at least a portion for training various models represented in FIG. 1 by model 190). The metrology data 160 may be metrology data of produced substrates, as well as sensor data, manufacturing data, and model data corresponding to those products. The metrology data 160 may be utilized to design processes for making additional substrates. The prediction data 168 may include predictions of metrology data resulting from operation of the manufacturing equipment 124, including the manufacturing equipment in which the components associated with the part quality data 146 are installed, predictions of component drift, aging, or failure, predictions of component life, etc. The prediction data 168 may also include data indicative of aging and failure of components of the system 100 over time.

[0053] In some embodiments, the prediction system 110 further includes a server machine 170 and a server machine 180. The server machine 170 includes a dataset generator 172 that can generate datasets (e.g., a set of data inputs and a set of target outputs) for training, validating, and / or testing the model 190. Some operations of the dataset generator 172 are described in more detail below with respect to FIG. 2 and FIG. 4A. In some embodiments, the dataset generator 172 can divide the historical data (e.g., historical sensor data, historical metrology data, etc.) and the physical model data (e.g., part quality data 146) into a training set (e.g., 60 percent of the data), a validation set (e.g., 20 percent of the data), and a test set (e.g., 20 percent of the data). In some embodiments, the prediction system 110 (e.g., via the prediction component 114) generates a plurality of feature sets. For example, a first feature set may correspond to a first type of sensor dataset (e.g., from a first sensor set, a first combination of values ​​from the first sensor set, a first pattern of values ​​from the first sensor set) corresponding to each dataset (e.g., training set, validation set, and test set), and a second feature set may correspond to a second type of sensor dataset (e.g., from a second sensor set different from the first sensor set, a second combination of values ​​different from the first combination, a second pattern different from the first pattern) corresponding to each dataset.

[0054] The server machine 180 includes a training engine 182, a validation engine 184, a selection engine 185, and / or a test engine 186. The engines (e.g., training engine 182, validation engine 184, selection engine 185, and test engine 186) may refer to hardware (e.g., circuitry, dedicated logic, programmable logic, microcode, processing device, etc.), software (e.g., instructions executed on a processing device, general-purpose computer system, or dedicated machine), firmware, microcode, or a combination thereof. The training engine 182 may be capable of training a model 190 using one or more feature sets associated with a training set from the dataset generator 172. The training engine 182 may generate multiple trained models 190, each trained model 190 corresponding to a different feature set of the training set (e.g., sensor data from a different sensor set). For example, a first trained machine learning model may have been trained using all features (e.g., X1-X5), a second trained machine learning model may have been trained using a first subset of features (e.g., X1, X2, X4), and a third trained machine learning model may have been trained using a second subset of features (e.g., X1, X3, X4, and X5) that may partially overlap with the first subset of features. The dataset generator 172 may receive the output of the trained models (e.g., 190), collect the data into training, validation, and testing datasets, and use the datasets to train a second model. Some or all of the operations of the server machine 180 may be used to train various types of models, including physics-based models, supervised machine learning models, unsupervised machine learning models, etc.

[0055] The validation engine 184 may be able to validate the trained models 190 using the corresponding feature set of the validation set from the dataset generator 172. For example, a first trained model 190 trained using a first feature set of the training set may be validated using a first feature set of the validation set. The validation engine 184 may determine the accuracy of each of the trained models 190 based on the corresponding feature set of the validation set. The validation engine 184 may discard trained models 190 whose accuracy does not meet a threshold accuracy. In some embodiments, the selection engine 185 may be able to select one or more trained models 190 whose accuracy meets a threshold accuracy. In some embodiments, the selection engine 185 may be able to select the trained model 190 with the highest accuracy among the trained models 190.

[0056] The testing engine 186 may be able to test the trained models 190 using a corresponding feature set of a test set from the dataset generator 172. For example, a first trained model 190 trained using a first feature set of the training set may be tested using a first feature set of the test set. The testing engine 186 may determine the most accurate trained model 190 of all the trained models based on the test set.

[0057] Model 190 may refer to a physics-based digital twin model of a manufacturing tool, e.g., a semiconductor processing chamber. The physics-based model is configured to solve equations that describe the flow of heat, energy, gases, etc. in and around the manufacturing tool. Measurements performed to characterize components of the manufacturing tool 124 can inform parameters of the physics-based model. The physics-based model may be improved by further training, e.g., by applying various process parameters to measure steady-state conditions in the processing chamber and feeding that data to the physics-based model to further refine the operation of the physics-based model.

[0058] Model 190 may refer to a machine learning model, which may be a model artifact created by training engine 182 using a training set that includes data inputs and corresponding target outputs (ground truths for each training input). Patterns may be found in the data set that map the data inputs to the target outputs (ground truths), and machine learning model 190 is provided with a mapping that captures these patterns. In some embodiments, machine learning model 190 may predict properties of a substrate. In some embodiments, machine learning model 190 may predict failure modes of a manufacturing chamber component. In some embodiments, machine learning model 190 may predict quality parameters of a chamber component.

[0059] The prediction component 114 can provide input data to the trained machine learning model 190 and can execute the trained machine learning model 190 on the input to obtain one or more outputs. The prediction component 114 may be able to determine (e.g., extract) prediction data 168 from the output of the trained machine learning model 190 and can determine (e.g., extract) confidence data from the output indicating a confidence level that the prediction data 168 is an accurate predictor of a product produced or a process associated with the input data of the product to be produced or an accurate predictor of a component of the manufacturing equipment 124. The prediction component 114 may be able to determine the prediction data 168 including predictions regarding finished substrate properties and predictions of useful life of components of the manufacturing equipment 124, the sensor 126, or the metrology equipment 128 based on the output of the model 190. The prediction component 114 or the corrective action component 122 can use the confidence data to determine whether to trigger a corrective action associated with the manufacturing equipment 124 based on the prediction data 168.

[0060] The confidence data may include or indicate a confidence level. As an example, the prediction data 168 may indicate characteristics of a finished wafer given a set of manufacturing inputs, including the use of one or more components described by the part quality data 146. The confidence data may indicate that the prediction data 168 is an accurate prediction of a product associated with at least a portion of the input data. In one example, the confidence level is a real number between 0 and 1, inclusive, where 0 indicates no confidence that the prediction data 168 is an accurate prediction of a product processed according to the input data, and 1 indicates absolute confidence that the prediction data 168 will accurately predict characteristics of a product processed according to the input data. In response to confidence data indicating a confidence level below a threshold level for a predetermined number of instances (e.g., a percentage of instances, a frequency of instances, a total number of instances, etc.), the prediction component 116 may cause the model 190 to be retrained (e.g., based on the current sensor data 146, the current manufacturing parameters 150, etc.).

[0061] For purposes of illustration and not limitation, aspects of the disclosure describe using historical data to train one or more models 190 and inputting current data into the one or more trained models 190 to determine predicted data 168. Other implementations use heuristic or rule-based models (e.g., without using trained machine learning models) to determine predicted data. The prediction component 114 can monitor historical data and measurement data 160. Any of the information described with respect to data input 210 of FIG. 2 can be monitored or otherwise used in the heuristic or rule-based models.

[0062] In some embodiments, the functionality of client device 120, prediction server 112, server machine 170, and server machine 180 may be provided by a fewer number of machines. For example, in some embodiments, server machines 170 and 180 may be combined into a single machine, while in some other embodiments, server machine 170, server machine 180, and prediction server 112 may be combined into a single machine. In some embodiments, client device 120 and prediction server 112 may be combined into a single machine.

[0063] In general, functions described in one embodiment as being performed by client device 120, prediction server 112, server machine 170, and server machine 180 may also be performed on prediction server 112 in other embodiments, where appropriate. In addition, functions attributed to particular components may be performed by different components or multiple components operating together. For example, in some embodiments, prediction server 112 may determine corrective actions based on prediction data 168. In another example, client device 120 may determine prediction data 168 based on output from a trained machine learning model or a physics-based (e.g., digital twin) model.

[0064] In addition, the functionality of a particular component may be performed by different components or multiple components working together. One or more of prediction server 112, server machine 170, or server machine 180 may be accessed as a service offered to other systems or devices via an appropriate application programming interface (API).

[0065] In embodiments, a "user" may be represented as a single individual. However, other embodiments of the present disclosure encompass a "user" that is an entity controlled by multiple users and / or automated sources. For example, a set of individual users aggregated as a group of administrators may be considered a "user."

[0066] Embodiments of the present disclosure may be applied to data quality assessment, feature enhancement, model evaluation, virtual metrology (VM), predictive maintenance (PdM), marginal optimization, and the like.

[0067] Although embodiments of the present disclosure are discussed in terms of generating predictive data 168 to implement corrective actions in a manufacturing facility (e.g., a semiconductor manufacturing facility), the embodiments may also be applied generally to improving data processing by utilizing physics-based digital twin models and sensors to characterize the impact of component quality parameters on processing equipment performance.

[0068] 2A-2B are block diagrams of an example dataset generator 272 (e.g., dataset generator 172 of FIG. 1) used to create a dataset for a model (e.g., model 190 of FIG. 1) according to some embodiments. The dataset generator 272 may be part of the server machine 170 of FIG. 1. In some embodiments, the system 100 of FIG. 1 includes multiple models. In such a case, each model may have a separate dataset generator, or the models may share a dataset generator. FIG. 2A illustrates a dataset generator associated with a physics-based model (e.g., a digital twin model) configured to take as inputs manufacturing data (e.g., process set points, hardware parameters, etc.) and part quality data (e.g., data describing measurements of parameters of one or more components of a manufacturing device) and provide as output predictions of conditions associated with performing a manufacturing or processing procedure using the manufacturing device (e.g., measured sensor data of conditions during processing).

[0069] The system 200A of FIG. 2A includes a dataset generator 272A (e.g., dataset generator 172 of FIG. 1). The dataset generator 272A creates a dataset for the physics-based model (e.g., model 190 of FIG. 1). The dataset generator 272A can create the dataset using data retrieved from sensors associated with the processing equipment or manufacturing equipment (e.g., part quality data), data retrieved from a data store, data received from a device acting as a controller for the processing equipment, etc. In some embodiments, the dataset generator 272A creates training inputs (e.g., data inputs 210A) from data associated with generating processing conditions for the manufacture of substrates, e.g., process parameter set points, part quality data associated with components of the processing equipment, etc. The dataset generator 272A also generates target outputs 220A for training the physics-based model. The target outputs include sensor data collected from sensors monitoring conditions proximate to a workpiece being processed by the manufacturing equipment. The training input data 210A and the target output data 220A may be provided to the physics-based model. The physics-based model can use training inputs and target outputs to tune parameters, coefficients, etc. to accurately predict conditions associated with a manufacturing process (e.g., predicting conditions within a manufacturing chamber).

[0070] It is within the scope of this disclosure to represent training inputs, target outputs, model outputs, etc. in various ways: as maps of component properties, vectors of coefficients, functions that replicate physical properties, dimensionality-reduced, remapped, or otherwise preprocessed versions of the data, etc.

[0071] Referring to FIG. 2B, system 200B including dataset generator 272B creates a dataset for a machine learning model. This disclosure enables several different applications of machine learning. System 200B shows a dataset generator used to train a machine learning model to predict part quality data of components installed on a manufacturing device. Other machine learning models may include generating datasets with similar characteristics. Other machine learning models that may be used in conjunction with the present disclosure include machine learning models used to predict the performance of a manufacturing device including one or more components with associated quality data, machine learning models used to predict characteristics of products produced by the manufacturing device, machine learning models used to predict whether an unmonitored quality parameter is significantly affecting product processing, and the like.

[0072] The dataset generator 272B generates data for training the machine learning model. In some embodiments, the dataset generator 272B creates training inputs 210B from sensor data of sensors (e.g., sensors monitoring conditions in a process chamber) associated with the manufacturing equipment and performance data (e.g., metrology data of a finished product). The dataset generator 272B also generates target output data 220B. In some embodiments, the target output data 220B includes data associated with part quality of one or more components associated with the manufacturing equipment. The training input data 210B and the target output data 220B are provided to train the machine learning model.

[0073] In some embodiments, the dataset generator 272 generates a dataset (e.g., a training set, a validation set, a test set) that includes one or more data inputs 210 (e.g., training inputs, validation inputs, test inputs) and may include one or more target outputs 220 corresponding to the data inputs 210. The dataset 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 dataset generator 272 may provide a dataset to the training engine 182, validation engine 184, or test engine 186 of FIG. 1, where the dataset is used to train, validate, or test a model 190 of FIG. 1 (e.g., a physics-based model, a machine learning model, etc.). Some embodiments of generating a training set may be further described with respect to FIG. 4A.

[0074] In some embodiments, the dataset generator 272 can generate a first data input corresponding to a first set of input data (e.g., first set of historical manufacturing data 250A, first set of historical part quality data 246A, first set of process sensor data 244A, first set of metrology data 250A, etc.) for training, validating, or testing a first machine learning model. The dataset generator 272 can generate a second data input corresponding to a second set of input data (e.g., set of historical manufacturing data 250B, etc.) for training, validating, or testing a second machine learning model. The dataset generator 272 can also generate a corresponding set of target output data 220.

[0075] In some embodiments, the dataset generator 272 can perform operations on one or more of the data inputs 210 and the target outputs 220. The dataset generator 272 can extract patterns from the data (gradient, curvature, etc.), can combine the data (average, feature generation, etc.), or can separate the data into groups (e.g., training a model on a subset of the predicted performance data) and use the groups to train separate models.

[0076] The data inputs 210 and target outputs 220 for training, validating, or testing a machine learning model may include information about a particular component of a manufacturing device (e.g., a particular substrate chucking assembly). The data inputs 210 and target outputs 220 may include information about the design of a particular component (e.g., information used for all components of that design). The data inputs 210 and target outputs 220 may include information about the role of a particular component (e.g., information used for the design of all components that play the same role in a process). The data inputs 210 and target outputs 220 may include information about a particular type of process, a design of a target product, characteristics of a target product, or may be grouped together in another manner.

[0077] In some embodiments, the dataset generator 272 can generate a set of target outputs 220. The target outputs 220 may be divided into sets corresponding to sets of input data. Different sets of target outputs 220 may be used in conjunction with similarly defined sets of data inputs 210, including training different models, using different sets for training, validation, and testing, etc.

[0078] The target output 220 may be generated by correlating trends in performance data to appropriate corrective actions using methods other than machine learning. A user may indicate that taking a particular corrective action eliminated differences between past predicted and measured performance, manufacturing faults may be intentionally introduced to generate data useful for training, etc. In some embodiments, a model may be trained without a target output 220 (e.g., an unsupervised model or a semi-supervised model). A model trained without being provided with a target output may, for example, be trained to recognize significant (e.g., outside an error threshold) differences between predicted and measured performance data.

[0079] In some embodiments, the information used to train the machine learning model may be from a particular type of manufacturing equipment (e.g., manufacturing equipment 124 in FIG. 1 ) in a manufacturing facility having particular characteristics, allowing the trained machine learning model to determine outcomes for a particular group of manufacturing equipment 124 based on inputs of predicted and measured performance data associated with one or more components that share the characteristics of the particular group. In some embodiments, the information used to train the machine learning model may be for components from more than one manufacturing facility, allowing the trained machine learning model to determine outcomes for components based on inputs from one manufacturing facility.

[0080] In some embodiments, after generating the dataset and using the dataset to train, validate, or test the machine learning model, the machine learning model may be further trained, validated, or tested, or tuned.

[0081] FIG. 3 is a block diagram illustrating a system 300 for generating output data (e.g., predicted data 168 of FIG. 1 ), according to some embodiments. System 300 can be used to analyze process parameters, sensor data, and metrology data and provide predictive data that takes the data into account and indicates corrective actions. For example, system 300 can indicate that one or more components of a manufacturing device are failing or drifting, that the quality is worse than expected, etc. Systems similar to system 300 may be used for other models, such as machine learning models that predict chamber performance, machine learning models that predict substrate metrology, machine learning models that predict that unmonitored quality parameters contribute significantly to chamber conditions, etc. Some or all of the operations of system 300 can be used to generate data indicative of characterization of one or more manufacturing device components via a physics-based digital twin model. In these cases, data other than those depicted in FIG. 3 may be used as input by system 300 and, if desired, generated as output.

[0082] 3, in block 310, the system 300 (e.g., a component of the forecasting system 110 of FIG. 1) performs data partitioning (e.g., via the data set generator 172 of the server machine 170 of FIG. 1) of the historical data 364 (e.g., historical process sensor data, historical metrology data, and historical part quality data, historical performance data) to generate a training set 302, a validation set 304, and a test set 306. For example, the training set may be 60% of the performance data, the validation set may be 20% of the performance data, and the test set may be 20% of the performance data.

[0083] In block 312, the system 300A performs training of the model using the training set 302 (e.g., via the training engine 182 of FIG. 1). The system 300 may train one model or train multiple models using multiple feature sets of the training set 302 (e.g., a first feature set including a subset of the performance data of the training set 302, a second feature set including a different subset of the performance data of the training set 302, etc.). For example, the system 300 may train machine learning models to generate a first trained machine learning model using the first feature set of the training set and generate a second trained machine learning model using the second feature set of the training set (e.g., data different from the data used to train the first machine learning model). In some embodiments, the first trained machine learning model and the second trained machine learning model may be combined to generate a third trained machine learning model (e.g., which may be a better predictor than the first or second trained machine learning models alone). In some embodiments, the sets of features used to compare the models may overlap (e.g., one model may be trained with performance data indicative of film thickness, another model may be trained with performance data indicative of both film thickness and film stress, different models may be trained with data from different locations on the substrate, different models may be trained with data from different overlapping sets of sensors, etc.). In some embodiments, hundreds of models may be generated, including models with various feature permutations and combinations of models.

[0084] In block 314, the system 300 performs model validation using the validation set 304 (e.g., via the validation engine 184 of FIG. 1). The system 300 can validate each of the trained models using the corresponding feature set of the validation set 304. For example, the validation set 304 can use the same subset of performance data used in the training set 302, but with different input conditions. In some embodiments, the system 300A can validate hundreds of models (e.g., models with various permutations of features, combinations of models, etc.) generated in block 312. In block 314, the system 300 can determine the accuracy of each of the one or more trained models (e.g., via model validation) and can determine whether one or more of the trained models have an accuracy that meets the threshold accuracy. In response to determining that none of the trained models have an accuracy that meets the threshold accuracy, flow returns to block 312 and the system 300 performs training of the models 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 meets the threshold accuracy, flow proceeds to block 316. The system 300 may discard trained machine learning models that have an accuracy below the threshold accuracy (e.g., based on a validation set).

[0085] In block 316, the system 300 may perform model selection (e.g., via the selection engine 185 of FIG. 1) 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 based on the validation of block 314). If only a single model was trained (or in another applicable condition, such as all trained models are used as an ensemble model), the operation of block 316 may be skipped. In response to determining that two or more of the trained models that meet the threshold accuracy have the same accuracy, the flow may return to block 312, where the system 300 performs model training using a further refined training set corresponding to the further refined feature set to determine the trained model with the highest accuracy.

[0086] In block 318, the system 300 performs model testing using the test set 306 (e.g., via the test engine 186 of FIG. 1) to test the selected model 308. The system 300 may test the first trained machine learning model using the first feature set of the test set and determine that the first trained machine learning model meets a threshold accuracy (e.g., based on the first feature set of the test set 306). In response to the accuracy of the selected model 308 not meeting the threshold accuracy (e.g., the selected model 308 is overfitted to the training set 302 and / or the validation set 304 and cannot be applied to other data sets such as the test set 306), the flow proceeds to block 312, where the system 300 performs training (e.g., retraining) of the model using a different training set, possibly corresponding to a different feature set, or a reorganization of the substrates divided into the training set, validation set, and test 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 flow proceeds to block 320. At least in block 312, the model can learn patterns in the simulated sensor data to make predictions, and in block 318, the system 300 can apply the model to the remaining data (e.g., test set 306) to test the predictions.

[0087] In block 320, the system 300 receives current data 354 (e.g., process sensor data, current manufacturing parameter data, current product performance data, etc.) using the trained model (e.g., selected model 308) and determines (e.g., extracts) from the output of the trained model predictive data 368 (e.g., predictive data 168 of FIG. 1) for performing an action (e.g., performing corrective action in connection with manufacturing equipment 124 of FIG. 1, providing an alert to client device 120 of FIG. 1, etc.).

[0088] In some embodiments, retraining of the machine learning model is done by supplying additional data to further train the model. Input / output data 354 of the current model may be provided in block 312. The current input / output data may include the same types of data originally used to train the machine learning model, e.g., inputs including sensor data and manufacturing parameters, target outputs including part quality data, etc. These data may differ from the data originally used to train the model by incorporating combinations of input parameters that were not part of the original training, input parameters outside the parameter space covered by the original training, or may be updated to reflect chamber-specific knowledge (e.g., differences from an ideal chamber due to manufacturing tolerance ranges, aged components, etc.). The selected model 308 may be retrained based on this data.

[0089] In some embodiments, one or more of acts 310-320 may be performed in various orders and / or with other acts not presented and described herein. In some embodiments, one or more of acts 310-320 may not be performed. For example, in some embodiments, one or more of data partitioning of block 310, model validation of block 314, model selection of block 316, or model testing of block 318 may not be performed. A subset of these operations may be performed in training a physics-based digital twin model, for example, to take measurements of part quality parameters of a component as input and generate predicted performance data of a chamber containing the component as output. The training data, input data, output data, and retraining data may be modified as necessary if the machine learning model is trained using different data, for example, trained to make predictions based on different input data, trained to make predictions other than part quality predictions, such as predicting substrate metrology or conditions in a manufacturing chamber, etc.

[0090] 4A-4E are flow diagrams of methods 400A-E associated with characterizing one or more components of a manufacturing device to trigger corrective action, according to certain embodiments. Methods 400A-E may be performed by processing logic, which may include hardware (e.g., circuitry, dedicated logic, programmable logic, microcode, processing device, etc.), software (e.g., instructions executed on a processing device, general-purpose computer system, or dedicated machine), firmware, microcode, or combinations thereof. In some embodiments, methods 400A-E may be performed in part by prediction system 110. Method 400A may be performed in part by prediction system 110 (e.g., server machine 170 and dataset generator 172 of FIG. 1, dataset generator 272 of FIG. 2). Prediction system 110 may use method 400A to generate a dataset for at least one of training, validating, or testing a model, according to an embodiment of the present disclosure. The model may be a physics-based digital twin model (e.g., generating predicted performance data for a manufacturing equipment that includes a component having known values ​​of one or more quality parameters), a machine learning model (e.g., generating predicted performance data for a wafer, generating data indicative of corrective actions associated with a component of the manufacturing equipment, etc.), a statistical model, or another model trained to receive an input and generate an output related to the quality of the component. Methods 400B-E may be performed by prediction server 112, client device 120, etc. In some embodiments, a non-transitory storage medium stores instructions that, when executed by a processing device (e.g., prediction system 110, server machine 180, prediction server 112, etc.), cause the processing device to perform one or more of methods 400A-E.

[0091] For ease of explanation, methods 400A-E are shown and described as a series of operations. However, operations in accordance with the present disclosure may occur in various orders and / or simultaneously, as well as with other operations not shown and described herein. Moreover, not all illustrated operations may be performed to implement methods 400A-E in accordance with the disclosed subject matter. In addition, those skilled in the art will understand and appreciate that methods 400A-E may alternatively be represented as a series of interrelated states via a state diagram or events.

[0092] 4A is a flow diagram of a method 400A for generating a dataset of a model for generating predictive data (e.g., predictive data 168 of FIG. 1 ) according to a particular embodiment. The model may be a machine learning model, a physics-based model, a digital twin model, etc.

[0093] Referring to FIG. 4A, in some embodiments, at block 401, processing logic performing method 400A initializes a training set T to an empty set.

[0094] At block 402, the processing logic generates a first data input (e.g., a first training input, a first validation input) that may include sensor data, metrology data (e.g., film properties such as thickness, material composition, optical properties, roughness, etc.), process parameter data, part quality data, etc. In some embodiments, the first data input may include a first set of features for a type of data (e.g., as described with respect to FIG. 3 ), and the second data input may include a second set of features for a type of data.

[0095] For example, method 400A can be used to generate a data set for a machine learning model configured to accept target process parameters (e.g., process set points) and measured performance data (e.g., sensor data, metrology data, etc.) as inputs and generate as output predictions of values ​​of one or more quality parameters of one or more components of manufacturing equipment. At block 402, training inputs may be generated that include sensor data, combinations of sensor data, features of sensor data, etc.

[0096] At block 403, the processing logic generates a first target output for one or more of the data inputs (e.g., a first data input). In some embodiments, the first target output is part quality data. In some embodiments, the first target output is indicative of manufacturing equipment performance. In some embodiments, the first target output is data indicative of a corrective action. In some embodiments, no target output is generated (e.g., for training an unsupervised machine learning model).

[0097] For example, method 400A can be used to generate a data set for a machine learning model configured to accept target process parameters and measured performance data as inputs and generate as output predictions of values ​​of one or more quality parameters of one or more components of the manufacturing equipment. The machine learning model may be further configured to identify potentially problematic quality parameter values, e.g., components that, when replaced, repaired, maintained, etc., can improve processing conditions of the manufacturing equipment. Data indicative of corrective actions (adjusting maintenance schedules, replacing components, updating process recipes, etc.) can be used to generate target outputs for training (validating, testing, etc.) the machine learning model.

[0098] At block 404, processing logic optionally generates mapping data indicative of an input / output mapping. An input / output mapping (or mapping data) may refer to data inputs (e.g., one or more of the data inputs described herein), target outputs for the data inputs, and associations between the data inputs and the target outputs. In some embodiments (e.g., without target output data), these operations may not be performed.

[0099] At block 405, processing logic, in some embodiments, adds the mapping data generated at block 404 to the dataset T.

[0100] At block 406, processing logic branches based on whether the dataset T is sufficient for at least one of training, validating, and / or testing a model, e.g., model 190 of FIG. 1. If so, execution proceeds to block 407; if not, execution continues back to block 402. Note that while in some embodiments the sufficiency of the dataset T may be determined simply based on the number of inputs in the dataset, which in some embodiments are mapped to outputs, in some other implementations the sufficiency of the dataset T may be determined based on one or more other criteria in addition to or instead of the number of inputs (e.g., a measure of diversity of the data examples, accuracy, etc.).

[0101] At block 407, processing logic provides dataset T (e.g., to server machine 180 of FIG. 1 ) to train, validate, and / or test a model, e.g., model 190. In some embodiments, dataset T is a training set and is provided to training engine 182 of server machine 180 to perform training. In some embodiments, dataset T is a validation set and is provided to validation engine 184 of server machine 180 to perform validation. In some embodiments, dataset T is a test set and is provided to test engine 186 of server machine 180 to perform testing.

[0102] FIG. 4B is a method 400B for determining whether a component is suitable for installation in a fabrication chamber, according to some embodiments. Logic similar to that of method 400B can also be used to determine whether a component is suitable for installation in other types of processing equipment. In block 410 of method 400B, processing logic receives data indicating a range of values ​​for a quality parameter of a type of fabrication chamber component. The type of component may be any part included in the fabrication equipment. For example, the type of component may be a part of a particular design (e.g., part number), a part that performs a particular function (e.g., showerhead, gas manifold, pumping plate, substrate support, etc.), a part or group of parts associated with a process, etc. The quality parameter may be any parameter of the component. Physical dimensions of the component (or one or more portions of the component, such as the depth of an electrode below the surface of the substrate support), material properties of the component (emissivity, density, etc.), chemical properties of the component (e.g., composition), electrical properties, optical properties, etc.) may be measured as quality parameters and provided to the physics-based model. The quality parameter may be a parameter expected to affect the performance of the component, for example, to affect conditions proximate to the substrate in a substrate processing chamber. In some embodiments, multiple quality parameter values ​​may be measured and utilized as part of method 400B and other methods of the present disclosure. In some embodiments, the range of values ​​may correspond to measurements of multiple different components of the same component type. In some embodiments, the range of values ​​may correspond to an expected range of the quality parameter exhibited by a set of components. In some embodiments, each value within this range meets one or more threshold criteria and is considered to be within specification. The criteria may be associated with a manufacturing tolerance of the component. The range of values ​​may be associated with threshold criteria and / or manufacturing tolerances of the component.

[0103] At block 412, processing logic provides data indicative of the range of values ​​to a physics-based model of a manufacturing equipment, such as a manufacturing chamber. The physics-based model may be or may include a digital twin model. The physics-based model may be configured to solve equations describing heat transfer, energy balance, gas flow, etc. associated with the manufacturing equipment. The physics-based model may be configured to generate output data indicative of conditions proximate to a workpiece, such as a semiconductor wafer, being processed using the manufacturing equipment.

[0104] At block 414, processing logic receives data from the physics-based model indicating a relationship between values ​​of the quality parameters and predicted conditions in the fabrication chamber. In some embodiments, the relationship may be a relationship between conditions achievable using components having various values ​​of one or more quality parameters. In some embodiments, the relationship may be a probability of achieving some target condition. In some embodiments, the relationship may indicate which input parameters will produce the target condition using components having input quality parameter values.

[0105] At block 416, a determination is made whether a manufacturing chamber component of the manufacturing component type should be installed in the manufacturing equipment based on the relationship between the quality parameter and the predicted conditions. In some embodiments, one or more quality parameters of the chamber component corresponding to the parameters associated with the range of values ​​provided to the physics-based model are measured. The output of the physics-based model may indicate that the chamber component may be less suitable for installation in the chamber. The value of the quality parameter of the component may be such that a target condition in the chamber will not be achieved. The value of the quality parameter of the component may be such that a target condition in the chamber will not be achieved with a target consistency (e.g., percentage of attempts above a target threshold). The value of the quality parameter of the component may be such that a target condition in the chamber is achievable by altering the process parameters in an undesirable manner, such as, for example, using more energy, more gas, more material, longer process time, etc. In some embodiments, the processing logic may receive data indicative of a target condition of the manufacturing process. A determination may be made whether the chamber component is suitable for use in the manufacturing process based on the relationship between the one or more quality parameters of the component and the predicted conditions proximate to the workpiece of the manufacturing equipment.

[0106] In some embodiments, multiple quality parameters of a single part type can be measured, ranges of the multiple quality parameters can be provided to the physics-based model, the multiple quality parameters can be used as a basis for component installation selection, etc. In some embodiments, quality parameter values ​​of chamber components that may be installed in the chamber are provided to the physics-based model. In some embodiments, the quality parameter values ​​of the components are compared to outputs previously generated by the physics-based model, e.g., tables or functions that describe relationships between quality parameter values ​​and chamber conditions.

[0107] 4C illustrates a flow diagram of a method 400C of utilizing part quality data in component classification, according to some embodiments. At block 420, processing logic receives a first process specification associated with a first manufacturing process. In some embodiments, the first manufacturing process is a semiconductor wafer manufacturing process. In some embodiments, the process specification includes target conditions proximate to the substrate for processing. At block 422, processing logic receives a second process specification associated with a second manufacturing process. The second process is different from the first process. At least one of the second process specifications is different from the first process specification. The second process specification can share many characteristics with the first process specification.

[0108] At block 424, processing logic determines a first probability that the manufacturing chamber component promotes conditions in the manufacturing chamber that comply with the first process specification. In some embodiments, determining the probability may include providing the process specification to a physics-based model. In some embodiments, determining the probability may include providing part quality data of the component to the physics-based model. In some embodiments, the probability may be based on a set of process parameter inputs (e.g., ideal process parameters, efficient process parameters, process parameters above an efficiency threshold, etc.). At block 426, processing logic determines a second probability that the component promotes conditions in the manufacturing chamber that comply with a second process specification. In some embodiments, the chamber associated with the first process specification and the chamber associated with the second process specification may be the same chamber. In some embodiments, they may be two different chambers. In some embodiments, they may be two chambers of the same design. In some embodiments, they may be different designs. In some embodiments, a second physics-based model is used to simulate operations associated with the second process specification. The operations of block 426 may share many characteristics with the operations of block 424.

[0109] At block 428, the processing logic assigns a first classification category and a second classification category to the components considering the first and second probabilities, respectively. In some embodiments, components with a probability of achieving the target processing condition above a threshold are categorized into the first category, and components below the threshold are categorized into the second category. In some embodiments, a more fine-grained categorization may be utilized (e.g., components with a 90-100% chance of achieving the target processing specification may be categorized into the first category for that process, components with an 80-90% chance of achieving the target processing specification may be categorized into the second category for that process, etc.). Components may receive different classifications for different sets of process specifications. In some embodiments, components are binned into categories based on the processes they are suitable to perform (e.g., exhibit suitable quality parameters). Thresholds, limits, binning categories, etc. may be adjusted as appropriate for the particular process, component type, target product, target manufacturing facility, etc.

[0110] In some embodiments, multiple component types may be categorized (e.g., binned) based on their performance given a set of process specifications. In some embodiments, components of one type may be categorized (e.g., binned) based on their performance given a set of two or more process specifications. In some embodiments, the physics-based model of the manufacturing equipment includes quality parameters of one or more components installed as part of the manufacturing equipment. In some embodiments, parts may be categorized based on the probability of achieving a target process specification according to a physics-based model that includes quality parameters of components already installed as part of the manufacturing equipment.

[0111] 4D illustrates a flow diagram of a method 400D for determining whether a component should be installed in a set of manufacturing equipment in which other components for which part quality data are available are installed, according to some embodiments. In some embodiments, classification of the components may be performed using a physics-based model of a standard or generic chamber, for example, without considering part quality data of components already installed in the chamber. Parts may be classified by suitability for a particular process (e.g., match vs. not match, degree of suitability, scoring or ranking, etc.). In some embodiments, components may be evaluated for suitability for installation as part of a particular set of manufacturing equipment (e.g., in a particular process chamber).

[0112] At block 430, processing logic receives first data indicative of values ​​of one or more quality parameters of one or more manufacturing chamber components installed in the manufacturing chamber. The values ​​of the one or more quality parameters may have been measured before the components are installed in the chamber. The values ​​of the one or more quality parameters may be used to determine whether the components should be installed in the chamber. The values ​​of the one or more quality parameters may have been predicted by a machine learning model, e.g., a model that takes process parameters, sensor data, etc. as inputs and provides as output a prediction of the one or more quality parameters of the one or more chamber components.

[0113] At block 432, processing logic receives second data indicative of a value of a quality parameter of a new manufacturing chamber component. The new manufacturing chamber component is being considered for installation in the manufacturing chamber. The quality parameter data for the new manufacturing chamber component may have been previously measured. The new component may have been selected for installation in the chamber based on the value of the quality parameter. In some embodiments, the second data may further include data indicative of a value of a second, third, etc. quality parameter of the new manufacturing chamber component.

[0114] At block 434, the processing logic provides the first data and the second data to a physics-based model. The physics-based model may include a digital twin model. The physics-based model predicts the performance of a manufacturing device, such as a processing chamber. The physics-based model may be configured to determine solutions to equations describing conditions in the chamber, such as heat transfer equations, gas flows, energy balances, etc. The physics-based model may be configured for a particular chamber, chamber type, process type, manufacturing facility, etc.

[0115] At block 436, the processing logic receives a prediction of the performance of the manufacturing chamber with the new manufacturing chamber component installed from the physics-based model. The physics-based model performs calculations that describe the process conditions. The physics-based model takes into account one or more quality parameters of one or more components installed on the manufacturing equipment and one or more quality parameters of the new component being considered for installation. The physics-based model may exclude quality parameter data associated with a part installed on the same type of manufacturing equipment as the new part (e.g., the old component may be effectively replaced with the new component). The predicted performance data may include a prediction of the conditions that will be achieved in the manufacturing chamber. The performance data may include a prediction of the resources utilized to reach the target conditions in the chamber, e.g., the efficiency of the combination of chamber components. The performance data may include a prediction of the likelihood of reaching the target conditions.

[0116] At block 438, the processing logic determines whether a new manufacturing chamber component should be installed in the manufacturing chamber based on the predicted performance. This determination is made by considering the output from the physics-based model. In some embodiments, the new component may have been pre-screened for inclusion in the chamber. One or more quality parameters of the component may have been measured and fed into the physics-based model and compared to a chart or table of characteristics, or the like. In some embodiments, the suitability of the component for installation in the chamber may be initially determined using data associated with the generic chamber. Method 400D may be used to refine the selection process of components to be installed in the manufacturing chamber by considering characteristics of components already installed in the chamber. In some embodiments, a component that was pre-screened and found to be suitable for installation in the chamber may be found to be unsuitable for installation when data indicating the quality of the components found in the chamber is provided to the physics-based model. In some embodiments, the new component may be installed in the chamber. In some embodiments, the process recipe may be updated by considering the output of the physics-based model to, for example, increase the likelihood of reaching the target process parameters with the new component installed in the manufacturing chamber.

[0117] In some embodiments, the output of the physics-based model may indicate that the chamber with the new components installed (or any chamber with any combination of components) will perform well enough for the application, but below a higher performance target. For example, the physics-based model may predict that the combination of components will perform well enough to produce the target conditions, but with a smaller margin of error than other chambers, other combinations of components, etc. A chamber may be predicted to achieve the target parameters, but not a more stringent set of parameters. Drift or aging may be expected to render such a chamber inoperable in a shorter time than a chamber that can achieve a more stringent set of parameters. The maintenance schedule for the first chamber may be adjusted to allow for more frequent maintenance events, reducing unplanned downtime for the chamber. The schedule for sending products from the first chamber to quality control may be changed to increase the number of products that are qualified. The schedule for checking the quality of the chamber (e.g., qualification events) may be adjusted.

[0118] FIG. 4E is a flow diagram of a method for utilizing part quality data for performing corrective actions, according to some embodiments. At block 440, processing logic receives predicted performance data of the manufacturing chamber from a physics-based model. The physics-based model may be a digital twin of the manufacturing chamber. The physics-based model may receive as input data indicative of values ​​of a quality parameter of a component of the manufacturing chamber. An output from the physics-based model may be generated considering the input. In some embodiments, the physics-based model may receive data indicative of values ​​of a second quality parameter of the chamber component. In some embodiments, the physics-based model may receive data indicative of values ​​of one or more quality parameters of the second chamber component. The physics-based model may generate the predicted performance data based on values ​​of one or more quality parameters of one or more chamber components installed in the manufacturing chamber. In some embodiments, one or more of the quality parameters may be within a manufacturing specification for the associated chamber component.

[0119] In some embodiments, the first chamber component was pre-screened for installation in the chamber prior to installation. In some embodiments, the first chamber component may have been selected for installation based on a pre-screening process. The chamber component may be assigned a classification as a result of the pre-screening process. One or more quality parameters of the chamber component may be used to assign the classification. One or more target process specifications (e.g., target conditions in the processing chamber) may be used to assign the classification. The classification may be assigned based on output from a physics-based model.

[0120] In some embodiments, the first chamber component may be installed as part of the manufacturing equipment based on output from a physics-based model of the manufacturing equipment. The physics-based model of the manufacturing equipment may utilize quality parameter data associated with the first chamber component and quality parameter data associated with one or more chamber components installed in the chamber. The physics-based model may have determined that the first chamber component is suitable for installation in the chamber based on the quality parameters of the components and the target process parameters (e.g., a combination of components including the first component is determined to be capable of achieving the target condition, is determined to have an acceptable probability of reaching the target condition, is determined to be capable of achieving the target condition while consuming an acceptable amount of resources, etc.).

[0121] In some embodiments, the predicted performance data may include predicted conditions in the processing chamber, such as predictions of temperature, pressure, etc. In some embodiments, the predicted performance data may include predictions of resources expended to achieve target conditions. In some embodiments, the predicted performance data includes predicted metrology data of the manufactured product. In some embodiments, further processing may be performed on the output of the physics-based model. For example, the predicted process conditions (or data indicative of the process conditions) output by the physics-based model may be provided to another model, which may predict substrate metrology based on the process conditions. In some embodiments, the output of the physics-based model may be fed to a second physics-based model, a digital twin model, a rule-based model, a machine learning model, etc., to receive further predicted performance data.

[0122] At block 442, processing logic receives measured performance data associated with the fabrication chamber. The measured performance data may be data obtained by a sensor (e.g., sensor data 142 of FIG. 1), a metrology tool (e.g., metrology tool 128), etc. The measured performance data may include many of the same characteristics as the predicted performance data associated with block 440. The measured performance data may include a description of the conditions within the fabrication chamber, an indication of the resources expended to obtain the conditions, measurements made on the finished product (e.g., substrate thickness, chemical properties, optical properties, etc.), etc.

[0123] In block 444, processing logic causes corrective actions to be taken considering the predicted and measured performance data. Processing logic may compare the predicted performance data to the measured performance data. Differences between the predicted and measured performance may be used in determining corrective actions to be taken. Differences between the measured and predicted conditions in the chamber may be used to update the process recipe to achieve target conditions, differences between measured and predicted resource usage may be used to schedule preventive maintenance of one or more components, differences between measured and predicted metrology data may be used to schedule replacement of one or more components, etc.

[0124] The corrective actions performed may be selected to accomplish one or more of reducing future differences between measured and predicted data (e.g., the difference between future predicted data of a future process and measured data of the process) or improving future measured data (e.g., achieving a target performance metric, achieving a target performance metric while consuming fewer resources, etc.). The corrective actions may include one or more of providing an alert to a user, updating a process recipe, scheduling preventive or corrective maintenance, updating a preventive maintenance schedule, scheduling replacement of a component, correcting sensor drift, or updating a physics-based model associated with the manufacturing equipment.

[0125] In some embodiments, the corrective action is selected in response to receiving an output from the trained machine learning model. The machine learning model may be trained by providing the machine learning model with training input data and target output data. The machine learning model may be trained to take predicted and measured performance data of the manufacturing equipment as inputs and provide as outputs data indicative of the quality of the chamber components, recommended corrective actions, etc. In some embodiments, the machine learning model is provided with the predicted and measured performance data as inputs. Additional data may also be provided to the machine learning model, such as data regarding drift or aging of the components, maintenance data, chamber history data, etc. In some embodiments, such data may be utilized in the calculated predicted performance data. The trained machine learning model may be trained (e.g., by providing part quality data of one or more components of the manufacturing chamber as target outputs) to generate as outputs estimates of values ​​of one or more quality parameters of one or more components of the manufacturing equipment. In some embodiments, the estimates of the quality parameters may be used to determine corrective actions, for example, may indicate that some change has occurred in the quality of the part since the quality parameter was measured.

[0126] In some embodiments, a machine learning model may be used to determine whether a chamber component contributes to chamber drift or failure. The predicted and measured performance data may be provided as inputs to a trained machine learning model. The trained machine learning model may provide as output a prediction of which component is responsible for the discrepancy between the predicted and measured performance data (e.g., the model may have been trained using predicted and measured performance data from a chamber with a known faulty component). In some embodiments, the discrepancy between the predicted and measured performance data may indicate a failed or drifting component (e.g., as identified by the machine learning model). In some embodiments, the discrepancy between the predicted and measured performance data may indicate that an uncharacterized component's quality parameters are having a measurable impact on conditions in the chamber. Identification of components with further critical quality parameters may be used to refine or improve part quality measurements, part classifications, physics-based models, digital twin models, and the like.

[0127] In some embodiments, predicted performance data of a manufacturing equipment (e.g., a processing chamber) may be used to determine the predicted performance of a finished product (e.g., a semiconductor wafer) using a machine learning model. The machine learning model may be trained by providing training inputs and target outputs. The training inputs and target outputs may include historical data. The training inputs may include conditions (or predicted conditions) within the processing environment (such as a manufacturing chamber). The target outputs may include measured metrology data of one or more products produced at the conditions provided as training inputs. The trained machine learning model may be provided with predicted conditions within the chamber (which may be based on the output of a physics-based model) during processing operations. The trained machine learning model may then predict metrology data of products processed at these conditions.

[0128] FIG. 5 illustrates a cross-sectional view of a fabrication chamber 500 (e.g., a semiconductor wafer fabrication chamber) according to some embodiments. The fabrication chamber 500 may be one or more of an etch chamber, a deposition chamber (including atomic layer deposition, chemical vapor deposition, physical vapor deposition, or plasma enhanced versions thereof), an anneal chamber, and the like. For example, the fabrication chamber 500 may be a chamber for a plasma etcher, a plasma cleaner, and the like. Examples of chamber components may include a substrate support 504, a chuck (e.g., electrostatic chuck, vacuum chuck, and the like), a ring (e.g., process kit ring), chamber walls, a base, a showerhead 506, a gas distribution plate, a liner, a liner kit, a shield, a plasma screen, a flow equalizer, a cooling base, a chamber viewport, a chamber lid, a nozzle, and the like. Quality parameters of any of these or other parts included in the chamber may be measured and used to classify the parts, provided to physics-based models to predict chamber performance, provided to machine learning models to predict substrate performance, and the like.

[0129] In one embodiment, the fabrication chamber 500 includes a chamber body 508 and a showerhead 506 that enclose an interior volume 510. In some chambers, the showerhead 506 may be replaced by a lid and nozzle. The chamber body 508 may be constructed from aluminum, stainless steel, or other suitable materials. The chamber body 508 generally includes sidewalls 512 and a bottom 514.

[0130] An exhaust port 516 may be defined in the chamber body 508 and may couple the interior volume 510 to a pumping system 518. The pumping system 518 may include one or more pumps and valves utilized to evacuate and regulate the pressure of the interior volume 510 of the fabrication chamber 500.

[0131] The showerhead 506 may be supported on a sidewall 512 of the chamber body 508 or on top of the chamber body. The showerhead 506 (or, in some embodiments, a lid) may be opened to allow access to the interior volume 510 of the fabrication chamber 500 and may provide a seal for the fabrication chamber 500 when closed. A gas panel 520 may be coupled to the fabrication chamber 500 to supply process or cleaning gases to the interior volume 510 through the showerhead 506 (or lid and nozzles). The showerhead 506 may include multiple gas supply holes throughout. Examples of process gases that may be used to process substrates in the fabrication chamber 500 include, among others, C 2 F 6 ,SCIENCE FICTION 6 , SiCl 4 , HBr, NF 3 , C.F. 4 , CHF 3 , F 2 , Cl 2 , CCl 4 , BCl 3 , and SiF 4 Halogen-containing gases such as O 2 or N 2 Other gases include O. Examples of carrier gases include N 2 , He, Ar, and other gases that are inert to the process gas (eg, non-reactive gases).

[0132] The substrate support 504 is disposed in an interior volume 510 of the fabrication chamber 500 below a showerhead 506. In some embodiments, the substrate support 504 includes a susceptor 522 and a shaft 524. The substrate support 504 supports the substrate 502 during processing. In some embodiments, one or more heaters 526 and reflectors 528 are also disposed within the fabrication chamber 500.

[0133] Part quality data for any parts contained in the fabrication chamber may be provided to the physics-based model, utilized to make predictions of chamber conditions, utilized to predict substrate performance data, etc. As an illustrative example, part quality data for the substrate support 504 may be measured and provided to the physics-based model, etc. The substrate support 504 may include electronics that provide power and control to electrodes disposed within the substrate support 504. The substrate support 504 shown in FIG. 5 includes an electrostatic chuck assembly, however substrate supports including other components are possible and within the scope of the present disclosure. The substrate support 504 may include one or more clamping electrodes (not shown). The clamping electrodes may be controlled by a chuck power supply 530. The chuck power supply 530 may include separate outputs to each clamping electrode such that the clamping electrodes may be individually controlled. For clarity, only one output of the chuck power supply 530 (and other similarly disposed components) is shown in FIG. 5.

[0134] The substrate support 504 may include one or more heating elements 536 disposed within the support (e.g., the substrate support 504 may include an electrostatic chuck heater assembly). The embedded heating elements may be regulated by a heater power supply 532. The heater power supply 532 may have many of the same characteristics as the chuck power supply 530. As shown in FIG. 5, the heating elements 536 may be disposed at different depths within the substrate support (e.g., different distances from the surface of the substrate support). The depth of the heating elements below the surface of the substrate support is one example of a possible part quality parameter of the substrate support. Other elements (chuck electrodes, RF elements, etc.) may also be disposed at various depths below the surface of the substrate support. In some embodiments, the heating elements 536 may be individually controlled. The physics-based digital twin model can provide as output predicted performance data of the manufacturing chamber including the substrate support 504, which can determine whether the substrate support 504 should be installed in the chamber, can inform set points of various electrodes housed in the substrate support 504 to reach one or more target characteristic value profiles (e.g., target temperature profile on the substrate) during substrate processing, etc. The substrate support 504 can further include one or more radio frequency (RF) elements controlled by an RF power generator 534. The RF power generator 534 and the RF elements can have some features in common with the chucking and heating system of the substrate support 504, as described above. Other types of electrodes can also be present in the substrate support and are within the scope of this disclosure.

[0135] Measurements of other components of the fabrication chamber can be used similarly to inform the physics-based model of the chamber. Dimensions, material properties, chemical properties, electrical properties, optical properties, etc. of any component of the chamber can be measured and provided to the physics-based model. The output of the physics-based model may be used to classify components, determine whether a component should be installed in the chamber, update the process recipe for a process running in the chamber, etc.

[0136] FIG. 6 is a block diagram illustrating a computer system 600 according to some embodiments. In some embodiments, the computer system 600 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 600 may operate in the capacity of a server or a client computer in a client-server environment, or as a peer computer in a peer-to-peer or distributed network environment. The computer system 600 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 web appliance, a server, a network router, a switch or bridge, or any device capable of executing a set of instructions (sequential or otherwise) that specify actions to be taken by that device. Furthermore, the term "computer" is intended to include any collection of computers that individually or jointly execute a set (or sets) of instructions to perform any one or more of the methods described herein.

[0137] In a further aspect, the computer system 600 may include a processing device 602, a volatile memory 604 (e.g., random access memory (RAM)), a non-volatile memory 606 (e.g., read only memory (ROM) or electrically erasable programmable ROM (EEPROM)), and a data storage device 618, which may communicate with each other via a bus 608.

[0138] The processing device 602 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 implementing other types of instruction sets, or a combination of multiple types of instruction sets, etc.) 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, etc.).

[0139] Computer system 600 may further include a network interface device 622 (e.g., coupled to a network 674). Computer system 600 may also include a video display unit 610 (e.g., LCD), an alphanumeric input device 612 (e.g., keyboard), a cursor control device 614 (e.g., mouse), and a signal generating device 620.

[0140] In some implementations, the data storage device 618 may include a non-transitory computer-readable storage medium 624 (e.g., a non-transitory machine-readable storage medium) capable of storing instructions 626 encoding any one or more of the methods or functions described herein, including instructions for encoding the components of FIG. 1 (e.g., the prediction component 114, the model 190, etc.) and performing the methods described herein.

[0141] The instructions 626 may also reside, completely or partially, within the volatile memory 604 and / or within the processing device 602 during execution thereof by the computer system 600, and thus the volatile memory 604 and the processing device 602 may also constitute machine-readable storage media.

[0142] Although the computer-readable storage medium 624 is shown as a single medium in the illustrative example, the term "computer-readable storage medium" is intended to include a single medium or multiple media (e.g., centralized or distributed databases and / or associated caches and servers) that store one or more sets of executable instructions. The term "computer-readable storage medium" is also intended to include any tangible medium capable of storing or encoding a set of instructions for execution by a computer that causes the computer to perform any one or more of the methods described herein. The term "computer-readable storage medium" is intended to include, but is not limited to, solid-state memory, optical media, and magnetic media.

[0143] The methods, components, and features described herein may be implemented by discrete hardware components or integrated into the functionality 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 a hardware device. Additionally, the methods, components, and features may be implemented in any combination of hardware devices and computer program components, or in a computer program.

[0144] Unless otherwise specified, terms such as "receive," "perform," "provide," "obtain," "cause," "access," "determine," "add," "use," "train," "generate," "identify," "allocate," "update," "schedule," "modify," and the like may refer to actions and processes performed or implemented by a computer system that manipulate and transform data represented as physical (electronic) quantities in computer system registers and memory into other data similarly represented as physical quantities in computer system memory or registers or other such information storage, transmission or display devices. Also, terms such as "first," "second," "third," "fourth," and the like, as used herein, are intended as labels to distinguish different elements and may not have any significance indicative of an order by their numerical designation.

[0145] The examples described herein also relate to apparatus for carrying out the methods described herein. The apparatus may be specially constructed to carry out the methods described herein, or 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.

[0146] The methods and illustrative embodiments described herein are not inherently related to any particular computer or other apparatus. Various general-purpose systems may be used in accordance with the teachings described herein, or it may prove convenient to construct more specialized apparatus to perform the methods described herein and / or each of their individual functions, routines, subroutines, or operations. Examples of structures for a variety of these systems are set forth in the description above.

[0147] The above description is intended to be illustrative and not limiting. Although the present disclosure has been described with reference to certain illustrative examples and implementations, it will be recognized 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 following claims, along with the full scope of equivalents to which such claims are entitled.

Claims

1. receiving first data indicative of a first range of values ​​for a first quality parameter of a first type of manufacturing chamber component, each value in the first range of values ​​satisfying one or more first threshold criteria; providing the first data to a physics-based model of a manufacturing chamber; receiving second data from the physics-based model indicative of a relationship between a value of the first quality parameter and a first predicted condition within the manufacturing chamber; determining whether to install a first manufacturing chamber component of the first manufacturing chamber component type in the manufacturing chamber based on the relationship between the value of the first quality parameter and a first predicted condition; A method comprising:

2. The method of claim 1 , wherein the one or more first threshold criteria include one or more manufacturing specifications for the first manufacturing chamber component type.

3. 2. The method of claim 1, wherein determining whether the first manufacturing chamber component of the first manufacturing chamber component type should be installed in the manufacturing chamber comprises determining whether the first manufacturing chamber component is capable of enabling the manufacturing chamber to achieve a target processing condition by utilizing the second data indicative of a relationship between a value of the first quality parameter and a first predicted condition in the manufacturing chamber and third data indicative of a value of the first quality parameter associated with the first manufacturing chamber component.

4. Receiving a first process specification associated with a first manufacturing process; determining a first probability that the first manufacturing chamber component will promote conditions in the manufacturing chamber that comply with the first process specification; assigning a first classification category to the first manufacturing chamber component taking into account the first probability; The method of claim 1 further comprising:

5. receiving a second process specification associated with a second manufacturing process; determining a second probability that the first manufacturing chamber component promotes conditions within the manufacturing chamber that comply with the second process specification; assigning a second classification category to the first manufacturing chamber component taking into account the second probability; and The method of claim 4 further comprising:

6. receiving third data indicative of a range of values ​​for a second quality parameter of the first manufacturing chamber component type, where each value in the second range of values ​​satisfies one or more threshold criteria; providing the third data to the physics-based model of the manufacturing chamber; and receiving fourth data from the physics-based model indicative of a relationship between a value of the second quality parameter and a second predicted condition within the manufacturing chamber; determining whether to install the first manufacturing chamber component of the manufacturing chamber component type in the manufacturing chamber based on the second predicted condition; and The method of claim 1 further comprising:

7. receiving third data indicative of a second range of values ​​for a second quality parameter of a second type of manufacturing chamber component, each value in the second range of values ​​satisfying one or more second threshold criteria; providing the third data to the physics-based model of the manufacturing chamber; and receiving fourth data from the physics-based model indicative of a relationship between a value of the second quality parameter and a second predicted condition within the manufacturing chamber; determining whether to install a second manufacturing chamber component of the second manufacturing chamber component type in the manufacturing chamber based on the second predicted condition; The method of claim 1 further comprising:

8. determining whether the first manufacturing chamber component should be installed in the manufacturing chamber; receiving third data indicative of values ​​of one or more quality parameters of one or more components installed in the manufacturing chamber; receiving fourth data indicative of a value of the first quality parameter of the first manufacturing chamber component; providing the third data and the fourth data to the physics-based model of the manufacturing chamber; receiving fifth data from the physics-based model that predicts a performance of the manufacturing chamber given the values ​​of a quality parameter of the one or more components installed in the manufacturing chamber and the value of the quality parameter of the first manufacturing chamber component; The method of claim 1 further comprising:

9. A non-transitory computer-readable storage medium storing instructions that, when executed by a processing device, cause the processing device to: receiving first data indicative of values ​​of one or more quality parameters of one or more manufacturing chamber components installed in the manufacturing chamber; receiving second data indicative of a first value of a first quality parameter of a new manufacturing chamber component, the new manufacturing chamber component being considered for installation in the manufacturing chamber; providing the first data and the second data to a physics-based model; receiving from the physics-based model a prediction of performance of the fabrication chamber with the new fabrication chamber component installed; determining whether the new fabrication chamber component should be installed in the fabrication chamber based on the predicted performance; and A non-transitory computer-readable storage medium for causing a computer to perform operations including:

10. The operation, updating a process recipe for a process associated with the manufacturing chamber taking into account the prediction of performance of the manufacturing chamber; 10. The non-transitory computer-readable storage medium of claim 9, further comprising:

11. 10. The non-transitory computer-readable storage medium of claim 9, wherein the new manufacturing chamber component has been pre-screened for use in a manufacturing process associated with the manufacturing chamber taking into account the first value of the first quality parameter of the new manufacturing chamber component.

12. The operation, receiving third data indicative of a second value of a second quality parameter of the new manufacturing chamber component; providing the third data to the physics-based model, wherein the prediction of performance from the physics-based model is generated taking into account the first data, the second data, and the third data; 10. The non-transitory computer-readable storage medium of claim 9, further comprising:

13. receiving predicted performance data for a first manufacturing chamber as output from a first physics-based model of the first manufacturing chamber, the output of the first physics-based model being generated taking into account a first value of a first quality parameter of a first component of the first manufacturing chamber; receiving measured performance data associated with the first manufacturing chamber; performing corrective action in consideration of the predicted and measured performance data; and A method comprising:

14. 14. The method of claim 13, wherein the output of the first physics-based model further accounts for a second value of a second quality parameter of the first component of the first manufacturing chamber.

15. 14. The method of claim 13, wherein the output of the first physics-based model further accounts for a second value of a second quality parameter of a second component of the first manufacturing chamber.

16. 14. The method of claim 13, wherein the first value of the first quality parameter of the first component in the first manufacturing chamber is within manufacturing specifications.

17. the first component of the first manufacturing chamber comprising: the first value of the first quality parameter of the first component; and a process specification for a manufacturing process associated with the first manufacturing chamber; an output of a second physics-based model, the second physics-based model configured to predict conditions that will be achieved in the second manufacturing chamber during the manufacturing process with the first component installed in the second manufacturing chamber; and and placing the first component in the first manufacturing chamber based on a classification assigned to the first component taking into account the above. The method of claim 13.

18. 14. The method of claim 13, wherein the first component of the first manufacturing chamber is installed in the first manufacturing chamber based on an output of a second physics-based model, the output of the second physics-based model being generated taking into account the first value of the first quality parameter of the first component and one or more values ​​of one or more quality parameters of one or more components installed in the manufacturing chamber.

19. providing historical performance data of the first manufacturing chamber as a training input for training a machine learning model; providing one or more quality parameters of one or more components installed in the first manufacturing chamber as target outputs for training the machine learning model, wherein the corrective action is taken taking into account an output of the trained machine learning model; The method of claim 13 further comprising:

20. To implement corrective measures, providing performance data of the first manufacturing chamber to a trained machine learning model; receiving as output from the trained machine learning model a predicted value of one or more quality parameters of one or more components of the first manufacturing chamber; identifying corrective actions to take given the output from the trained machine learning model; and The method of claim 13, comprising:

Images